NX Nastran

Basic Dynamic Analysis

User’s Guide
 Proprietary & Restricted Rights Notice

 2004 UGS PLM Solutions, Inc. All Rights Reserved. This software and related documentation are
proprietary to UGS PLM Solutions. LIMITATIONS TO U.S. GOVERNMENT RIGHTS. UNPUBLISHED -
RIGHTS RESERVED UNDER THE COPYRIGHT LAWS OF THE UNITED STATES. This computer software
and related computer software documentation have been developed exclusively at private expense and are
provided subject to the following rights: If this computer software and computer software documentation
qualify as “commercial items” (as that term is defined in FAR 2.101), their use, duplication, or disclosure by
the U.S. Government is subject to the protections and restrictions as set forth in the UGS PLM Solutions
commercial license for the software and/or documentation as prescribed in FAR 12.212 and FAR
27.405(b)(2)(i) (for civilian agencies) and in DFARS 227.7202-1(a) and DFARS 227.7202-3(a) (for the
Department of Defense).,or any successor or similar regulation, as applicable or as amended from time to
time. If this computer software and computer documentation do not qualify as “commercial items,” then they
are “restricted computer software,” and are provided with “restrictive rights,” and their use, duplication or
disclosure by the U.S. Government is subject to the protections and restrictions as set forth in FAR 27.404(b)
and FAR 52-227-14 (for civilian agencies), and DFARS 227.7203-5(c) and DFARS 252.227-7014 (for the
Department of Defense), or any successor or similar regulation as applicable or as amended from time to time.
UGS PLM Solutions, Inc, 5400 Legacy Drive, Plano, Texas 75024.
NASTRAN is a registered trademark of the National Aeronautics and Space Administration. NX Nastran is
an enhanced proprietary version developed and maintained by UGS PLM Solutions, Inc.
MSC is a registered trademark of MSC.Software Corporation. MSC.Nastran and MSC.Patran are trademarks
of MSC.Software Corporation.
All other trademarks are the property of their respective owners.
 C O N T E N T S
NX Nastran Basic Dynamic Analysis User’s Guide

NX Nastran Basic Dynamic Analy-

sis User’s Guide

Preface ■ About this Book, xiv

1
Fundamentals of ■ Overview, 2
Dynamic Analysis
■ Equations of Motion, 3
■ Dynamic Analysis Process, 12
■ Dynamic Analysis Types, 14

2
Finite Element ■ Overview, 16
Input Data
■ Mass Input, 17
■ Damping Input, 24
■ Units in Dynamic Analysis, 28
■ Direct Matrix Input, 30
❑ Direct Matrix Input, 31
❑ DMIG Bulk Data User Interface, 32
❑ DMIG Case Control User Interface, 34

3
Real Eigenvalue ■ Overview, 36
Analysis
■ Reasons to Compute Normal Modes, 38
■ Overview of Normal Modes Analysis, 39
■ Methods of Computation, 45
❑ Lanczos Method, 45
❑ Givens and Householder Methods, 46
❑ Modified Givens and Modified Householder Methods, 46
❑ Automatic Givens and Automatic Householder Methods, 46
❑ Inverse Power Method, 47
❑ Sturm Modified Inverse Power Method, 47

■ Comparison of Methods, 48
■ User Interface for Real Eigenvalue Analysis, 50
❑ User Interface for the Lanczos Method, 50
❑ User Interface for the Other Methods, 52

■ Solution Control for Normal Modes Analysis, 56

 ❑ Executive Control Section, 56
❑ Case Control Section, 56
❑ Bulk Data Section, 57
■ Examples, 58
❑ Two-DOF Model, 58
❑ Cantilever Beam Model, 62
❑ Bracket Model, 67
❑ Car Frame Model, 72
❑ Test Fixture Model, 77
❑ Quarter Plate Model, 79
❑ DMIG Example, 83

4
Rigid-body Modes ■ Overview, 88
■ SUPORT Entry, 90
❑ Treatment of SUPORT by Eigenvalue Analysis Methods, 90
- EIGR Methods, 90
- Lanczos Method, 91
❑ Theoretical Considerations, 91
❑ Modeling Considerations, 94

■ Examples, 96
❑ Unconstrained Beam Model, 96
❑ Unconstrained Bracket Example, 101

5
Frequency ■ Overview, 104
Response
■ Direct Frequency Response Analysis, 106
Analysis
❑ Damping in Direct Frequency Response, 106

■ Modal Frequency Response Analysis, 108

❑ Damping in Modal Frequency Response, 109
❑ Mode Truncation in Modal Frequency Response Analysis, 112
❑ Dynamic Data Recovery in Modal Frequency Response Analysis, 112

■ Modal Versus Direct Frequency Response, 114

■ Frequency-Dependent Excitation Definition, 115
❑ Frequency-Dependent Loads – RLOAD1 Entry, 116
❑ Frequency-Dependent Loads – RLOAD2 Entry, 117
❑ Spatial Distribution of Loading – DAREA Entry, 118
❑ Time Delay – DELAY Entry, 118
❑ Phase Lead – DPHASE Entry, 119
❑ Dynamic Load Tabular Function – TABLEDi Entries, 120
❑ DAREA Example, 123
❑ Static Load Sets – LSEQ Entry, 124
❑ LSEQ Example, 125
❑ Dynamic Load Set Combination – DLOAD, 126

■ Solution Frequencies, 127

❑ FREQ, 128
 ❑ FREQ1, 128
❑ FREQ2, 129
❑ FREQ3, 130
❑ FREQ4, 131
❑ FREQ5, 131
■ Frequency Response Considerations, 132
■ Solution Control for Frequency Response Analysis, 133
■ Examples, 136
❑ Two-DOF Model, 136
❑ Cantilever Beam Model, 142
❑ Bracket Model, 148

6
Transient ■ Overview, 152
Response
■ Direct Transient Response Analysis, 153
Analysis
❑ Damping in Direct Transient Response, 154
❑ Initial Conditions in Direct Transient Response, 156

■ Modal Transient Response Analysis, 158

❑ Damping in Modal Transient Response Analysis, 159
❑ Mode Truncation in Modal Transient Response Analysis, 162
❑ Dynamic Data Recovery in Modal Transient Response Analysis, 163

■ Modal Versus Direct Transient Response, 164

■ Transient Excitation Definition, 165
❑ Time-Dependent Loads – TLOAD1 Entry, 166
❑ Time-Dependent Loads – TLOAD2 Entry, 167
❑ Spatial Distribution of Loading – DAREA Entry, 168
❑ Time Delay – DELAY Entry, 168
❑ Dynamic Load Tabular Function – TABLEDi Entries, 169
❑ DAREA Example, 172
❑ Static Load Sets – LSEQ Entry, 173
❑ LSEQ Example, 174
❑ Dynamic Load Set Combination – DLOAD, 175

■ Integration Time Step, 177

■ Transient Excitation Considerations, 178
■ Solution Control for Transient Response Analysis, 179
■ Examples, 182
❑ Two-DOF Model, 182
❑ Cantilever Beam Model, 185
❑ Bracket Model, 193

7
Enforced Motion ■ Overview, 200
■ The Large Mass Method in Direct Transient and Direct Frequency
Response, 201
 ■ The Large Mass Method in Modal Transient and Modal Frequency
Response, 203
■ User Interface for the Large Mass Method, 205
❑ Frequency Response, 206
- Enforced Acceleration, 207
- Enforced Velocity, 207
- Enforced Displacement, 207
❑ Transient Response, 207
- TLOAD1 Format, 208
- TLOAD2 Format, 208
■ Examples, 209
❑ Two-DOF Model, 209
❑ Cantilever Beam Model, 216

8
Restarts In ■ Overview, 224
Dynamic Analysis
■ Automatic Restarts, 225
■ Structure of the Input File, 226
■ User Interface, 227
❑ Cold Start Run, 227
❑ Restart Run, 227
- NASTRAN Statement Section, 227
- File Management Section (FMS), 227
- Executive Control Section, 228
- Case Control Section, 228
- Bulk Data Section, 229
■ Determining the Version for a Restart, 232
■ Examples, 234
- Remark, 236

9
Plotted Output ■ Overview, 248
■ Structure Plotting, 249
■ X-Y Plotting, 255

10
Guidelines for ■ Overview, 270
Effective Dynamic
■ Overall Analysis Strategy, 271
Analysis
■ Units, 274
■ Mass, 275
■ Damping, 276
■ Boundary Conditions, 279
 ■ Loads, 280
■ Meshing, 281
■ Eigenvalue Analysis, 283
■ Frequency Response Analysis, 284
❑ Number of Retained Modes, 284
❑ Size of the Frequency Increment, 284
❑ Relationship of Damping to the Frequency Increment, 284
❑ Verification of the Applied Load, 285

■ Transient Response Analysis, 286

❑ Number of Retained Modes, 286
❑ Size of the Integration Time Step, 286
❑ Duration of the Computed Response, 287
❑ Value of Damping, 287
❑ Verification of the Applied Load, 287

■ Results Interpretation and Verification, 288

■ Computer Resource Requirements, 290

11
Advanced ■ Overview, 294
Dynamic Analysis
■ Dynamic Reduction, 295
Capabilities
■ Complex Eigenvalue Analysis, 296
■ Response Spectrum Analysis, 297
■ Random Vibration Analysis, 299
■ Mode Acceleration Method, 300
■ Fluid Structure Interaction, 301
❑ Hydroelastic Analysis, 301
❑ Virtual Fluid Mass, 301
❑ Coupled Acoustics, 301
❑ Uncoupled Acoustics, 302

■ Nonlinear Transient Response Analysis, 303

❑ Geometric Nonlinearity, 303
❑ Material Nonlinearity, 303
❑ Contact, 304
❑ Nonlinear-Elastic Transient Response Analysis, 304
❑ Nonlinear Normal Modes Analysis, 306

■ Superelement Analysis, 307

❑ Component Mode Synthesis, 307

■ Design Optimization and Sensitivity, 309

■ Control System Analysis, 311
■ Aeroelastic Analysis, 312
❑ Aerodynamic Flutter, 312
- Dynamic Aeroelastic Response, 313
■ DMAP, 315
 A
Glossary of Terms ■ Glossary of Terms, 318

B
Nomenclature for ■ Nomenclature for Dynamic Analysis, 322
Dynamic Analysis ❑ General , 322
❑ Structural Properties, 324
- Geometry, 324
- Stiffness, 324
- Mass, 324
- Damping, 324
❑ Multiple Degree-of-Freedom System, 325

C
The Set Notation ■ Overview, 328
System Used in ❑ Displacement Vector Sets, 328
Dynamic Analysis

D
Solution ■ Overview, 336
Sequences for ❑ Structured Solution Sequences for Basic Dynamic Analysis, 336
Dynamic Analysis ❑ Rigid Formats for Basic Dynamic Analysis, 336

E
Case Control ■ Overview, 338
Commands for ❑ Input Specification, 338
Dynamic Analysis ❑ Analysis Specification, 338
❑ Output Specification, 338

<E>
Case Control
Commands for
Dynamic Analysis

F
Bulk Data Entries ■ Overview, 368
for Dynamic
Analysis
 F
Bulk Data Entries
for Dynamic
Analysis

G
Parameters for ■ Overview, 434
Dynamic Analysis

H
File Management ■ Overview, 442
Section
■ Definitions, 443
■ MSC.Nastran Database, 444
■ File Management Commands, 445
❑ INIT, 445
- Purpose, 445
- Format (Simplified) for All DBsets Except SCRATCH, 445
- Format (Simplified) for the SCRATCH DBset, 446
❑ ASSIGN, 446
- Purpose, 446
- Format (Simplified) to Assign Logical Files, 446
- Format (Simplified) to Assign FORTRAN Files, 447
❑ EXPAND, 448
- Purpose, 448
- Format, 448
- RESTART, 449
- Purpose, 449
- Format, 449
❑ INCLUDE, 450
- Purpose, 450
- Format, 450
❑ Summary, 450

I
Grid Point Weight ■ Overview, 452
Generator
■ Commonly Used Features, 453
■ Example with Direction Dependent Masses, 456

J
Numerical ■ Overview, 468
Accuracy ❑ Linear Equation Solution, 468
Considerations ❑ Eigenvalue Analysis, 468
❑ Matrix Conditioning, 469
❑ Definiteness of Matrices, 470
 ❑ Numerical Accuracy Issues, 470
❑ Sources of Mechanisms, 471
❑ Sources of Nonpositive Definite Matrices, 472
❑ Detection and Avoidance of Numerical Problems, 472
- Static Analysis, 472
- Eigensolutions Using the Inverse Iteration and Lanczos Methods, 474
- Eigensolutions Using Transformation Methods, 474
- Frequency Response Analysis, 475
- Transient Response Analysis, 475
- The Large Mass Method, 475
- The Lagrange Multiplier Method, 476

K
Diagnostic ■ Overview, 478
Messages for
Dynamic Analysis

L
References and ■ Overview, 496
Bibliography
■ General References, 497
■ Bibliography, 498
❑ DYNAMICS – GENERAL, 498
❑ DYNAMICS – ANALYSIS / TEST CORRELATION, 509
❑ DYNAMICS – COMPONENT MODE SYNTHESIS, 513
❑ DYNAMICS – DAMPING, 517
❑ DYNAMICS – FREQUENCY RESPONSE, 518
❑ DYNAMICS – MODES, FREQUENCIES, AND VIBRATIONS, 519
❑ DYNAMICS – RANDOM RESPONSE, 530
❑ DYNAMICS – REDUCTION METHODS, 531
❑ DYNAMICS – RESPONSE SPECTRUM, 532
❑ DYNAMICS – SEISMIC, 533
❑ DYNAMICS – TRANSIENT ANALYSIS, 534

INDEX NX Nastran Basic Dynamic Analysis User’s Guide

 Preface

■ About this Book

xiv

About this Book

The NX Nastran Basic Dynamic Analysis User’s Guide is a guide to the proper use of
MSC.Nastran® (MSC.Nastran is a registered trademark of MSC.Software Corporation) for
solving various dynamic analysis problems. This guide serves as both an introduction to
dynamic analysis for the new user and a reference for the experienced user. The major
emphasis focuses on understanding the physical processes in dynamics and properly
applying MSC.Nastran to model dynamic processes while restricting mathematical
derivations to a minimum.
The basic types of dynamic analysis capabilities available in MSC.Nastran are described in
this guide. These common dynamic analysis capabilities include normal modes analysis,
transient response analysis, frequency response analysis, and enforced motion. These
capabilities are described and illustrative examples are presented. Theoretical derivations
of the mathematics used in dynamic analysis are presented only as they pertain to the
proper understanding of the use of each capability.

To effectively use this guide, it is important for you to be familiar with MSC.Nastran’s
static analysis capability and the principles of dynamic analysis. Basic finite element
modeling and analysis techniques are covered only as they pertain to MSC.Nastran
dynamic analysis. For more information on static analysis and modeling, refer to the NX
Nastran Linear Static Analysis User’s Guide and to the Getting Started with NX Nastran User’s
Guide.
This guide is an update to the NX Nastran Basic Dynamic Analysis User’s Guide for
Version 68, which borrowed much material from the MSC.Nastran Handbook for Dynamic
Analysis. However, not all topics covered in that handbook are covered here. Dynamic
reduction, response spectrum analysis, random response analysis, complex eigenvalue
analysis, nonlinear analysis, control systems, fluid-structure coupling and the Lagrange
Multiplyer Method will be covered in the NX Nastran Advanced Dynamic Analysis User’s
Guide.

Acknowledgements
Much of the basis of this guide was established by Michael Gockel with the MSC.Nastran
Handbook for Dynamic Analysis. This guide has used that information as a starting point.
John Muskivitch began the guide and was the primary contributor. Other major
contributors included David Bella, Franz Brandhuber, Michael Gockel, and John Lee.
Other technical contributors included Dean Bellinger, John Caffrey, Louis Komzsik, Ted
Rose, and Candace Hoecker. Ken Ranger and Sue Rice provided the bracket and test
fixture models, respectively, and John Furno helped to run those models.
This guide benefitted from intense technical scrutiny by Brandon Eby, Douglas Ferg,
John Halcomb, David Herting, Wai Ho, Erwin Johnson, Kevin Kilroy, Mark Miller, and
William Rodden. Gert Lundgren of LAPCAD Engineering provided the car model.
Customer feedback was received from Mohan Barbela of Martin-Marietta Astro-Space,
 CHAPTER xv
Preface

Dr. Robert Norton of Jet Propulsion Laboratory, Alwar Parthasarathy of SPAR Aerospace
Limited, Canada, and Manfred Wamsler of Mercedes-Benz AG. The efforts of all have
made this guide possible, and their contributions are gratefully acknowledged.

Grant Sitton
June 1997
xvi
 NX Nastran Basic Dynamic Analysis User’s Guide

1
CHAPTER
Fundamentals of Dynamic Analysis

■ Overview

■ Equations of Motion

■ Dynamic Analysis Process

■ Dynamic Analysis Types

 2

1.1 Overview
®
In static structural analysis, it is possible to describe the operation of MSC.Nastran
(MSC.Nastran is a registered trademark of MSC.Software Corporation) without a detailed
discussion of the fundamental equations. Due to the several types of dynamic analyses and
the different mathematical form of each, some knowledge of both the physics of dynamics
and the manner in which the physics is represented is important to using MSC.Nastran
effectively and efficiently for dynamic analysis.

You should become familiar with the notation and terminology covered in this chapter.
This knowledge will be valuable to understand the meaning of the symbols and the reasons
for the procedures employed in later chapters. “References and Bibliography” on
page 495 provides a list of references for structural dynamic analysis.

Dynamic Analysis Versus Static Analysis. Two basic aspects of dynamic analysis
differ from static analysis. First, dynamic loads are applied as a function of time. Second,
this time-varying load application induces time-varying response (displacements,
velocities, accelerations, forces, and stresses). These time-varying characteristics make
dynamic analysis more complicated and more realistic than static analysis.
This chapter introduces the equations of motion for a single degree-of-freedom dynamic
system (see “Equations of Motion” on page 3), illustrates the dynamic analysis process
(see “Dynamic Analysis Process” on page 12), and characterizes the types of dynamic
analyses described in this guide (see “Dynamic Analysis Types” on page 14). Those who
are familiar with these topics may want to skip to subsequent chapters.
 CHAPTER 1 3
Fundamentals of Dynamic Analysis

1.2 Equations of Motion

The basic types of motion in a dynamic system are displacement u and the first and second
derivatives of displacement with respect to time. These derivatives are velocity and
acceleration, respectively, given below:

du
u· = ------ = v = velocity
dt
2 Eq. 1-1
d u dv
u·· = --------- = ------ = a = acceleration
2 dt
dt

Velocity and Acceleration. Velocity is the rate of change in the displacement with
respect to time. Velocity can also be described as the slope of the displacement curve.
Similarly, acceleration is the rate of change of the velocity with respect to time, or the slope
of the velocity curve.

Single Degree-of-Freedom System. The most simple representation of a dynamic

system is a single degree-of-freedom (SDOF) system (see Figure 1-1). In an SDOF system,
the time-varying displacement of the structure is defined by one component of motion u ( t ) .
Velocity u· ( t ) and acceleration u·· ( t ) are derived from the displacement.

p(t)
m = mass (inertia)
u(t )
b = damping (energy dissipation
k = stiffness (restoring force) m
p = applied force
u = displacement of mass k b
·
u = velocity of mass
··
u = acceleration of mass

Figure 1-1 Single Degree-of-Freedom (SDOF) System

Dynamic and Static Degrees-of-Freedom. Mass and damping are associated with the
motion of a dynamic system. Degrees-of-freedom with mass or damping are often called
dynamic degrees-of-freedom; degrees-of-freedom with stiffness are called static
degrees-of-freedom. It is possible (and often desirable) in models of complex systems to
have fewer dynamic degrees-of-freedom than static degrees-of-freedom.
The four basic components of a dynamic system are mass, energy dissipation (damper),
resistance (spring), and applied load. As the structure moves in response to an applied
load, forces are induced that are a function of both the applied load and the motion in the
individual components. The equilibrium equation representing the dynamic motion of the
system is known as the equation of motion.
4

Equation of Motion. This equation, which defines the equilibrium condition of the
system at each point in time, is represented as

mu·· ( t ) + bu· ( t ) + ku ( t ) = p ( t ) Eq. 1-2

The equation of motion accounts for the forces acting on the structure at each instant in
time. Typically, these forces are separated into internal forces and external forces. Internal
forces are found on the left-hand side of the equation, and external forces are specified on
the right-hand side. The resulting equation is a second-order linear differential equation
representing the motion of the system as a function of displacement and higher-order
derivatives of the displacement.

Inertia Force. An accelerated mass induces a force that is proportional to the mass and the
acceleration. This force is called the inertia force mu·· ( t ) .

Viscous Damping. The energy dissipation mechanism induces a force that is a function
of a dissipation constant and the velocity. This force is known as the viscous damping force
bu· ( t ) . The damping force transforms the kinetic energy into another form of energy,
typically heat, which tends to reduce the vibration.

Elastic Force. The final induced force in the dynamic system is due to the elastic
resistance in the system and is a function of the displacement and stiffness of the system.
This force is called the elastic force or occasionally the spring force ku ( t ) .

Applied Load. The applied load p ( t ) on the right-hand side of Eq. 1-2 is defined as a
function of time. This load is independent of the structure to which it is applied (e.g., an
earthquake is the same earthquake whether it is applied to a house, office building, or
bridge), yet its effect on different structures can be very different.

Solution of the Equation of Motion. The solution of the equation of motion for
quantities such as displacements, velocities, accelerations, and/or stresses—all as a
function of time—is the objective of a dynamic analysis. The primary task for the dynamic
analyst is to determine the type of analysis to be performed. The nature of the dynamic
analysis in many cases governs the choice of the appropriate mathematical approach. The
extent of the information required from a dynamic analysis also dictates the necessary
solution approach and steps.

Dynamic analysis can be divided into two basic classifications: free vibrations and forced
vibrations. Free vibration analysis is used to determine the basic dynamic characteristics
of the system with the right-hand side of Eq. 1-2 set to zero (i.e., no applied load). If
damping is neglected, the solution is known as undamped free vibration analysis.

Free Vibration Analysis. In undamped free vibration analysis, the SDOF equation of
motion reduces to

mu·· ( t ) + ku ( t ) = 0 Eq. 1-3

 CHAPTER 1 5
Fundamentals of Dynamic Analysis

Eq. 1-3 has a solution of the form

u ( t ) = A sin ω n t + B cos ω n t Eq. 1-4

The quantity u ( t ) is the solution for the displacement as a function of time t . As shown in
Eq. 1-4, the response is cyclic in nature.

Circular Natural Frequency. One property of the system is termed the circular natural
frequency of the structure ω n . The subscript n indicates the “natural” for the SDOF system.
In systems having more than one mass degree of freedom and more than one natural
frequency, the subscript may indicate a frequency number. For an SDOF system, the
circular natural frequency is given by

k
ωn = ---- Eq. 1-5
m

The circular natural frequency is specified in units of radians per unit time.

Natural Frequency. The natural frequency f n is defined by

ωn
f n = ------ Eq. 1-6
2π

The natural frequency is often specified in terms of cycles per unit time, commonly cycles
per second (cps), which is more commonly known as Hertz (Hz). This characteristic
indicates the number of sine or cosine response waves that occur in a given time period
(typically one second).
The reciprocal of the natural frequency is termed the period of response T n given by

1 2π
T n = ---- = ------ Eq. 1-7
fn ωn

The period of the response defines the length of time needed to complete one full cycle of
response.
In the solution of Eq. 1-4, A and B are the integration constants. These constants are
determined by considering the initial conditions in the system. Since the initial
displacement of the system u ( t = 0 ) and the initial velocity of the system u· ( t = 0 ) are known,
A and B are evaluated by substituting their values into the solution of the equation for
displacement and its first derivative (velocity), resulting in

u· ( t = 0 )
B = u(t = 0) and A = ----------------------
ωn
Eq. 1-8

These initial value constants are substituted into the solution, resulting in

u· ( 0 )
u ( t ) = ----------- sin ω n t + u ( 0 ) cos ω n t Eq. 1-9
ωn
6

Eq. 1-9 is the solution for the free vibration of an undamped SDOF system as a function of
its initial displacement and velocity. Graphically, the response of an undamped SDOF
system is a sinusoidal wave whose position in time is determined by its initial
displacement and velocity as shown in Figure 1-2.

Amplitude u(t)

Time t
Figure 1-2 SDOF System – Undamped Free Vibrations

If damping is included, the damped free vibration problem is solved. If viscous damping
is assumed, the equation of motion becomes

mu·· ( t ) + bu· ( t ) + ku ( t ) = 0 Eq. 1-10

Damping Types. The solution form in this case is more involved because the amount of
damping determines the form of the solution. The three possible cases for positive values
of b are
• Critically damped
• Overdamped
• Underdamped
Critical damping occurs when the value of damping is equal to a term called critical
damping b cr . The critical damping is defined as

b cr = 2 km = 2mω n Eq. 1-11

For the critically damped case, the solution becomes

– bt ⁄ 2m
u ( t ) = ( A + Bt )e Eq. 1-12

Under this condition, the system returns to rest following an exponential decay curve with
no oscillation.

A system is overdamped when b > b cr and no oscillatory motion occurs as the structure
returns to its undisplaced position.
 CHAPTER 1 7
Fundamentals of Dynamic Analysis

Underdamped System. The most common damping case is the underdamped case
where b < b cr . In this case, the solution has the form

– bt ⁄ 2m
u(t) = e ( A sin ω d t + B cos ω d t ) Eq. 1-13

Again, A and B are the constants of integration based on the initial conditions of the
system. The new term ω d represents the damped circular natural frequency of the system.
This term is related to the undamped circular natural frequency by the following
expression:

2
ωd = ωn 1 – ζ Eq. 1-14

The term ζ is called the damping ratio and is defined by

b
ζ = ------- Eq. 1-15
b cr

The damping ratio is commonly used to specify the amount of damping as a percentage of
the critical damping.
In the underdamped case, the amplitude of the vibration reduces from one cycle to the next
following an exponentially decaying envelope. This behavior is shown in Figure 1-3. The
amplitude change from one cycle to the next is a direct function of the damping. Vibration
is more quickly dissipated in systems with more damping.
Amplitude u(t)

Time t
Figure 1-3 Damped Oscillation, Free Vibration

The damping discussion may indicate that all structures with damping require damped
free vibration analysis. In fact, most structures have critical damping values in the 0 to 10%
range, with values of 1 to 5% as the typical range. If you assume 10% critical damping,
Eq. 1-4 indicates that the damped and undamped natural frequencies are nearly identical.
This result is significant because it avoids the computation of damped natural frequencies,
which can involve a considerable computational effort for most practical problems.
Therefore, solutions for undamped natural frequencies are most commonly used to
determine the dynamic characteristics of the system (see “Real Eigenvalue Analysis” on
page 35). However, this does not imply that damping is neglected in dynamic response
8

analysis. Damping can be included in other phases of the analysis as presented later for
frequency and transient response (see “Frequency Response Analysis” on page 103 and
“Transient Response Analysis” on page 151).

Forced Vibration Analysis. Forced vibration analysis considers the effect of an applied
load on the response of the system. Forced vibrations analyses can be damped or
undamped. Since most structures exhibit damping, damped forced vibration problems are
the most common analysis types.
The type of dynamic loading determines the mathematical solution approach. From a
numerical viewpoint, the simplest loading is simple harmonic (sinusoidal) loading. In the
undamped form, the equation of motion becomes

mu·· ( t ) + ku ( t ) = p sin ωt Eq. 1-16

In this equation the circular frequency of the applied loading is denoted by ω . This loading
frequency is entirely independent of the structural natural frequency ω n , although similar
notation is used.
This equation of motion is solved to obtain

p⁄k
u ( t ) = A sin ω n t + B cos ω n t + ----------------------------- sin ωt Eq. 1-17
2 2
1 – ω ⁄ ωn

Initial Condition Steady-State

Solution Solution

where:

u· ( t = 0 ) ωp ⁄ k
A = ----------------------
ωn
– ------------------------------------------
2 2
( 1 – ω ⁄ ω n )ω n

B = u(t = 0)

Again, A and B are the constants of integration based on the initial conditions. The third
term in Eq. 1-17 is the steady-state solution. This portion of the solution is a function of the
applied loading and the ratio of the frequency of the applied loading to the natural
frequency of the structure.
The numerator and denominator of the third term demonstrate the importance of the
relationship of the structural characteristics to the response. The numerator p ⁄ k is the
static displacement of the system. In other words, if the amplitude of the sinusoidal
loading is applied as a static load, the resulting static displacement u is p ⁄ k . In addition,
to obtain the steady state solution, the static displacement is scaled by the denominator.

The denominator of the steady-state solution contains the ratio between the applied
loading frequency and the natural frequency of the structure.
 CHAPTER 1 9
Fundamentals of Dynamic Analysis

Dynamic Amplification Factor for No Damping. The term

1
-----------------------------
2 2
1 – ω ⁄ ωn

is called the dynamic amplification (load) factor. This term scales the static response to
create an amplitude for the steady state component of response. The response occurs at the
same frequency as the loading and in phase with the load (i.e., the peak displacement
occurs at the time of peak loading). As the applied loading frequency becomes
approximately equal to the structural natural frequency, the ratio ω ⁄ ω n approaches unity
and the denominator goes to zero. Numerically, this condition results in an infinite (or
undefined) dynamic amplification factor. Physically, as this condition is reached, the
dynamic response is strongly amplified relative to the static response. This condition is
known as resonance. The resonant buildup of response is shown in Figure 1-4.
Amplitude u(t)

Time t
Figure 1-4 Harmonic Forced Response with No Damping

It is important to remember that resonant response is a function of the natural frequency

and the loading frequency. Resonant response can damage and even destroy structures.
The dynamic analyst is typically assigned the responsibility to ensure that a resonance
condition is controlled or does not occur.
10

Solving the same basic harmonically loaded system with damping makes the numerical
solution more complicated but limits resonant behavior. With damping, the equation of
motion becomes

mu·· ( t ) + bu· ( t ) + ku ( t ) = p sin ωt Eq. 1-18

In this case, the effect of the initial conditions decays rapidly and may be ignored in the
solution. The solution for the steady-state response is

sin ( ωt + θ )
u ( t ) = p ⁄ k ---------------------------------------------------------------------------------- Eq. 1-19
2 2 2 2
(1 – ω ⁄ ωn ) + ( 2ζω ⁄ ω n )

The numerator of the above solution contains a term that represents the phasing of the
displacement response with respect to the applied loading. In the presence of damping,
the peak loading and peak response do not occur at the same time. Instead, the loading
and response are separated by an interval of time measured in terms of a phase angle θ as
shown below:

–1 2ζω ⁄ ω n
θ = – tan ----------------------------------- Eq. 1-20
2 2
( 1 – ω ⁄ ωn )

The phase angle θ is called the phase lead, which describes the amount that the response
leads the applied force.

Note: Some texts define θ as the phase lag, or the amount that the response lags the
applied force. To convert from phase lag to phase lead, change the sign of θ in
Eq. 1-19 and Eq. 1-20.

Dynamic Amplification Factor with Damping. The dynamic amplification factor for the
damped case is

1
---------------------------------------------------------------------------------- Eq. 1-21
2 2 2 2
(1 – ω ⁄ ωn ) + ( 2ζω ⁄ ω n )

The interrelationship among the natural frequency, the applied load frequency, and the
phase angle can be used to identify important dynamic characteristics. If ω ⁄ ω n is much less
than 1, the dynamic amplification factor approaches 1 and a static solution is represented
with the displacement response in phase with the loading. If ω ⁄ ω n is much greater than 1,
the dynamic amplification factor approaches zero, yielding very little displacement
response. In this case, the structure does not respond to the loading because the loading is
changing too fast for the structure to respond. In addition, any measurable displacement
response will be 180 degrees out of phase with the loading (i.e., the displacement response
will have the opposite sign from the force). Finally if ω ⁄ ω n = 1 , resonance occurs. In this
case, the magnification factor is 1 ⁄ ( 2ζ ) , and the phase angle is 270 degrees. The dynamic
amplification factor and phase lead are shown in Figure 1-5 and are plotted as functions of
forcing frequency.
 CHAPTER 1 11
Fundamentals of Dynamic Analysis

Amplification Factor
1

360°
Phase Lead θ (Degrees)

180°

ωn
Forcing Frequency ω

Figure 1-5 Harmonic Forced Response with Damping

In contrast to harmonic loadings, the more general forms of loading (impulses and general
transient loading) require a numerical approach to solving the equations of motion. This
technique, known as numerical integration, is applied to dynamic solutions either with or
without damping. Numerical integration is described in “Transient Response Analysis”
on page 151.
12

1.3 Dynamic Analysis Process

Before conducting a dynamic analysis, it is important to define the goal of the analysis prior
to the formulation of the finite element model. Consider the dynamic analysis process to
be represented by the stepsLK in Figure 1-6. The analyst must evaluate the finite element
model in terms of the type of dynamic loading to be applied to the structure. This dynamic
load is known as the dynamic environment. The dynamic environment governs the
solution approach (i.e., normal modes, transient response, frequency response, etc.). This
environment also indicates the dominant behavior that must be included in the analysis
(i.e., contact, large displacements, etc.). Proper assessment of the dynamic environment
leads to the creation of a more refined finite element model and more meaningful results.

Dynamic
Environment

Finite Element
Model

Modal Yes Results No

Analysis? Satisfactory?
No
Yes

No Results Forced-Response
Satisfactory? Analysis

Yes

End

Figure 1-6 Overview of Dynamic Analysis Process

An overall system design is formulated by considering the dynamic environment. As part

of the evaluation process, a finite element model is created. This model should take into
account the characteristics of the system design and, just as importantly, the nature of the
dynamic loading (type and frequency) and any interacting media (fluids, adjacent
structures, etc.). At this point, the first step in many dynamic analyses is a modal analysis
to determine the structure’s natural frequencies and mode shapes (see “Real Eigenvalue
Analysis” on page 35).

In many cases the natural frequencies and mode shapes of a structure provide enough
information to make design decisions. For example, in designing the supporting structure
for a rotating fan, the design requirements may require that the natural frequency of the
 CHAPTER 1 13
Fundamentals of Dynamic Analysis

supporting structure have a natural frequency either less than 85% or greater than 110% of
the operating speed of the fan. Specific knowledge of quantities such as displacements and
stresses are not required to evaluate the design.

Forced response is the next step in the dynamic evaluation process. The solution process
reflects the nature of the applied dynamic loading. A structure can be subjected to a
number of different dynamic loads with each dictating a particular solution approach. The
results of a forced-response analysis are evaluated in terms of the system design.
Necessary modifications are made to the system design. These changes are then applied
to the model and analysis parameters to perform another iteration on the design. The
process is repeated until an acceptable design is determined, which completes the design
process.

The primary steps in performing a dynamic analysis are summarized as follows:

1. Define the dynamic environment (loading).
2. Formulate the proper finite element model.
3. Select and apply the appropriate analysis approach(es) to determine the behavior
of the structure.
4. Evaluate the results.
14

1.4 Dynamic Analysis Types

This guide describes the following types of dynamic analysis that can be performed with
MSC.Nastran:
• Real eigenvalue analysis (undamped free vibrations).
• Linear frequency response analysis (steady-state response of linear structures to
loads that vary as a function of frequency).
• Linear transient response analysis (response of linear structures to loads that vary
as a function of time).
The above list comprises the basic types of dynamic analysis. Many MSC.Nastran
advanced dynamic analysis capabilities, such as shock/response spectrum analysis,
random response analysis, design sensitivity, design optimization, aeroelasticity, and
component mode synthesis, can be used in conjunction with the above analyses.

In practice, very few engineers use all of the dynamic analysis types in their work.
Therefore, it may not be important for you to become familiar with all of the types. Each
type can be considered independently, although there may be many aspects common to
many of the analyses.
Real eigenvalue analysis is used to determine the basic dynamic characteristics of a
structure. The results of an eigenvalue analysis indicate the frequencies and shapes at
which a structure naturally tends to vibrate. Although the results of an eigenvalue analysis
are not based on a specific loading, they can be used to predict the effects of applying
various dynamic loads. Real eigenvalue analysis is described in “Real Eigenvalue
Analysis” on page 35.
Frequency response analysis is an efficient method for finding the steady-state response to
sinusoidal excitation. In frequency response analysis, the loading is a sine wave for which
the frequency, amplitude, and phase are specified. Frequency response analysis is limited
to linear elastic structures. Frequency response analysis is described in “Frequency
Response Analysis” on page 103.
Transient response analysis is the most general method of computing the response to
time-varying loads. The loading in a transient analysis can be of an arbitrary nature, but is
explicitly defined (i.e., known) at every point in time. The time-varying (transient) loading
can also include nonlinear effects that are a function of displacement or velocity. Transient
response analysis is most commonly applied to structures with linear elastic behavior.
Transient response analysis is described in “Transient Response Analysis” on page 151.
Additional types of dynamic analysis are available with MSC.Nastran. These types are
described briefly in “Advanced Dynamic Analysis Capabilities” on page 293 and will be
described fully in the NX Nastran Advanced Dynamic Analysis User’s Guide.
 NX Nastran Basic Dynamic Analysis User’s Guide

2
CHAPTER
Finite Element Input Data

■ Overview

■ Mass Input

■ Damping Input

■ Units in Dynamic Analysis

■ Direct Matrix Input

16

2.1 Overview
When conducting an MSC.Nastran dynamic analysis, the formulation of the model
depends upon the desired results and the type of dynamic loading. For example, if stresses
are of interest, a finer finite element mesh is required than if only the system level
displacement patterns are needed. Many of the modeling considerations involved in a
static analysis are similarly applied in dynamic analysis. A dynamic analysis, however,
requires additional input data which is used to define the dynamic character of the
structure.
In static analysis the stiffness properties are defined by element and material properties.
These same properties are also required for dynamic analysis along with the addition of
mass and damping. “Mass Input” on page 17 describes mass input, and “Damping
Input” on page 24 describes damping input.
Correct specification of units is very important for dynamic analysis. Incorrect
specification of units is probably more difficult to diagnose in dynamic analysis than in
static analysis. Because MSC.Nastran has no built-in set of units, you must ensure their
consistency (and accuracy). “Units in Dynamic Analysis” on page 28 describes the
common variables and units for dynamic analysis.
“Direct Matrix Input” on page 30 concludes this chapter with a discussion of direct matrix
input.
 CHAPTER 2 17
Finite Element Input Data

2.2 Mass Input

Mass input is one of the major entries in a dynamic analysis. Mass can be represented in a
number of ways in MSC.Nastran. The mass matrix is automatically computed when mass
density or nonstructural mass is specified for any of the standard finite elements (CBAR,
CQUAD4, etc.) in MSC.Nastran, when concentrated mass elements are entered, and/or
when full or partial mass matrices are entered.

Lumped and Coupled Mass. Mass is formulated as either lumped mass or coupled
mass. Lumped mass matrices contain uncoupled, translational components of mass.
Coupled mass matrices contain translational components of mass with coupling between
the components. The CBAR, CBEAM, and CBEND elements contain rotational masses in
their coupled formulations, although torsional inertias are not considered for the CBAR
element. Coupled mass can be more accurate than lumped mass. However, lumped mass
is more efficient and is preferred for its computational speed in dynamic analysis.
The mass matrix formulation is a user-selectable option in MSC.Nastran. The default mass
formulation is lumped mass for most MSC.Nastran finite elements. The coupled mass
matrix formulation is selected using PARAM,COUPMASS,1 in the Bulk Data. Table
Table 2-1 shows the mass options available for each element type.

Table 2-1 Element Mass Types

Element Type Lumped Mass Coupled Mass*

CBAR X X
CBEAM X X
CBEND X
CONEAX X
CONMi X X
CONROD X X
CRAC2D X X
CRAC3D X X
CHEXA X X
CMASSi X
CPENTA X X
CQUAD4 X X
CQUAD8 X X
CQUADR X X
CROD X X
CSHEAR X
18

Table 2-1 Element Mass Types (continued)

Element Type Lumped Mass Coupled Mass*

CTETRA X X
CTRIA3 X X
CTRIA6 X X
CTRIAR X X
CTRIAX6 X X
CTUBE X X

* Couple mass is selected by PARAM,COUPMASS,1.

The MSC.Nastran coupled mass formulation is a modified approach to the classical

consistent mass formulation found in most finite element texts. The MSC.Nastran lumped
mass is identical to the classical lumped mass approach. The various formulations of mass
matrices can be compared using the CROD element. Assume the following properties:

2 (Torsion) 4 (Torsion) Length L

1 (Translation) 3 (Translation) Area A
Torsional Constant J
Young’s Modulus E
L Shear Modulus G
Mass Density ρ
Polar Moment of Inertia I ρ
1-4 are
Degrees-of-Freedom

CROD Element Stiffness Matrix. The CROD element’s stiffness matrix [ K ] is given by:

1 2 3 4
AE – AE
------- 0 ----------- 0 1
L L
GJ – GJ
0 ------- 0 ---------- 2
[K] = L L Eq. 2-1
– AE AE
----------- 0 ------- 0 3
L L
– GJ GJ
0 ---------- 0 ------- 4
L 0

The zero entries in the matrix create independent (uncoupled) translational and rotational
behavior for the CROD element, although for most other elements these
degrees-of-freedom are coupled.
 CHAPTER 2 19
Finite Element Input Data

CROD Lumped Mass Matrix. The CROD element classical lumped mass matrix is the
same as the MSC.Nastran lumped mass matrix. This lumped mass matrix is

1
--- 0 0 0
2
[ M ] = ρAL 0 0 0 0 Eq. 2-2
1
0 0 --- 0
2
0 0 0 0

The lumped mass matrix is formed by distributing one-half of the total rod mass to each of
the translational degrees-of-freedom. These degrees-of-freedom are uncoupled and there
are no torsional mass terms calculated.

The CROD element classical consistent mass matrix is

1 1
--- 0 --- 0
3 6
Iρ Iρ
0 ------- 0 -------
[ M ] = ρAL 3A 6A Eq. 2-3
1 1
--- 0 --- 0
6 3
Iρ Iρ
0 ------- 0 -------
6A 3A

This classical mass matrix is similar in form to the stiffness matrix because it has both
translational and rotational masses. Translational masses may be coupled to other
translational masses, and rotational masses may be coupled to other rotational masses.
However, translational masses may not be coupled to rotational masses.

CROD Coupled Mass Matrix. The CROD element MSC.Nastran coupled mass matrix is

5 1
------ 0 ------ 0
12 12
[ M ] = ρAL 0 0 0 0 Eq. 2-4
1 5
------ 0 ------ 0
12 12
0 0 0 0

The axial terms in the CROD element coupled mass matrix represent the average of
lumped mass and classical consistent mass. This average is found to yield the best results
for the CROD element as described below. The mass matrix terms in the directions
transverse to the element axes are lumped mass, even when the coupled mass option is
selected. Note that the torsional inertia is not included in the CROD element mass matrix.
20

Lumped Mass Versus Coupled Mass Example. The difference in the axial mass
formulations can be demonstrated by considering a fixed-free rod modeled with a single
CROD element (Figure 2-1). The exact quarter-wave natural frequency for the first axial
mode is

E⁄ρ
1.5708 ------------
l

Using the lumped mass formulation for the CROD element, the first frequency is predicted
to be

E⁄ρ
1.414 ------------
l

which underestimates the frequency by 10%. Using a classical consistent mass approach,
the predicted frequency

E⁄ρ
1.732 ------------
l

is overestimated by 10%. Using the coupled mass formulation in MSC.Nastran, the

frequency

E⁄ρ
1.549 ------------
l

is underestimated by 1.4%. The purpose of this example is to demonstrate the possible

effects of the different mass formulations on the results of a simple problem. Remember
that not all dynamics problems have such a dramatic difference. Also, as the model’s mesh
becomes finer, the difference in mass formulations becomes negligible.
 CHAPTER 2 21
Finite Element Input Data

u(t)
1

l Single Element Model

π E⁄ρ E ⁄ ρ
Theoretical Natural Frequency: ω n = --- ---------------- = 1.5708 ----------------
2 l l

E ⁄ ρ E ⁄ ρ
MSC.Nastran Lumped Mass: ωn = 2 ---------------- = 1.414 ----------------
l l

E ⁄ ρ E ⁄ ρ
Classical Consistent Mass: ωn = 3 ---------------- = 1.732 ----------------
l l

E ⁄ ρ E ⁄ ρ
MSC.Nastran Coupled Mass: ωn = 12 ⁄ 5 ---------------- = 1.549 ----------------
l l
Figure 2-1 Comparison of Mass Formulations for a ROD

CBAR, CBEAM Lumped Mass. The CBAR element lumped mass matrix is identical to
the CROD element lumped mass matrix. The CBEAM element lumped mass matrix is
identical to that of the CROD and CBAR mass matrices with the exception that torsional
inertia is included.

CBAR, CBEAM Coupled Mass. The CBAR element coupled mass matrix is identical to
the classical consistent mass formulation except for two terms: (1) the mass in the axial
direction is the average of the lumped and classical consistent masses, as explained for the
CROD element; and (2) there is no torsional inertia. The CBEAM element coupled mass
matrix is also identical to the classical consistent mass formulation except for two terms:
(1) the mass in the axial direction is the lumped mass; and (2) the torsional inertia is the
lumped inertia.
Another important aspect of defining mass is the units of measure associated with the mass
definition. MSC.Nastran assumes that consistent units are used in all contexts. You must
be careful to specify structural dimensions, loads, material properties, and physical
properties in a consistent set of units.
22

All mass entries should be entered in mass consistent units. Weight units may be input
instead of mass units, if this is more convenient. However, you must convert the weight to
mass by dividing the weight by the acceleration of gravity defined in consistent units:

ρ m = ( 1 ⁄ g )ρ w Eq. 2-5

where:

ρ m = mass or mass density

g = acceleration of gravity
ρ w = weight or weight density

The parameter

PARAM,WTMASS,factor
performs this conversion. The value of the factor should be entered as 1 ⁄ g . The default
value for the factor is 1.0. Hence, the default value for WTMASS assumes that mass (and
mass density) is entered, instead of weight (and weight density).
When using English units if the weight density of steel is entered as RHO = 0.3 lb/in 3 , using
PARAM,WTMASS,0.002588 converts the weight density to mass density for the
acceleration of gravity g = 386.4 in/sec 2 . The mass density, therefore, becomes
2 4 3
7.765E-4 lb f -sec /in . If the weight density of steel is entered as RHO = 80000 N/m when
using metric units, then using PARAM,WTMASS,0.102 converts the weight density to
mass density for the acceleration of gravity g = 9.8 m/sec 2 . The mass density, therefore,
becomes 8160 kg/m 3 .
PARAM,WTMASS is used once per run, and it multiplies all weight/mass input (including
CMASSi, CONMi, and nonstructural mass input). Therefore, do not mix input type; use
all mass (and mass density) input or all weight (or weight density) input.
PARAM,WTMASS does not affect direct input matrices M2GG or M2PP (see “Direct
Matrix Input” on page 30). PARAM,CM2 can be used to scale M2GG; there is no
parameter scaling for M2PP. PARAM,CM1 is similar to PARAM,WTMASS since CM1
scales all weight/mass input (except for M2GG and M2PP), but it is active only when
M2GG is also used. In other words, PARAM,CM1 is used in addition to
PARAM,WTMASS if M2GG is used.

MSC.Nastran Mass Input. Mass is input to MSC.Nastran via a number of different

entries. The most common method to enter mass is using the RHO field on the MATi entry.
This field is assumed to be defined in terms of mass density (mass/unit volume). To
determine the total mass of the element, the mass density is multiplied by the element
volume (determined from the geometry and physical properties). For a MAT1 entry, a
mass density for steel of 7.76E-4 lb f -sec 2 /in4 is entered as follows:

1 2 3 4 5 6 7 8 9 10
$MAT1 MID E G NU RHO A TREF GE

MAT1 2 30.0E6 0.3 7.76E-4

 CHAPTER 2 23
Finite Element Input Data

Grid point masses can be entered using the CONM1, CONM2, and CMASSi entries. The
CONM1 entry allows input of a fully coupled 6 × 6 mass matrix. You define half of the
terms, and symmetry is assumed. The CONM2 entry defines mass and mass moments of
inertia for a rigid body. The CMASSi entries define scalar masses.

Nonstructural Mass. An additional way to input mass is to use nonstructural mass,

which is mass not associated with the geometric cross-sectional properties of an element.
Examples of nonstructural mass are insulation, roofing material, and special coating
materials. Nonstructural mass is input as mass/length for line elements and mass/area for
elements with two-dimensional geometry. Nonstructural mass is defined on the element
property entry (PBAR, for example).
24

2.3 Damping Input

Damping is a mathematical approximation used to represent the energy dissipation
observed in structures. Damping is difficult to model accurately since it is caused by many
mechanisms including
• Viscous effects (dashpot, shock absorber)
• External friction (slippage in structural joints)
• Internal friction (characteristic of the material type)
• Structural nonlinearities (plasticity, gaps)
Because these effects are difficult to quantify, damping values are often computed based
on the results of a dynamic test. Simple approximations are often justified because the
damping values are low.

Viscous and Structural Damping. Two types of damping are generally used for
linear-elastic materials: viscous and structural. The viscous damping force is proportional
to velocity, and the structural damping force is proportional to displacement. Which type
to use depends on the physics of the energy dissipation mechanism(s) and is sometimes
dictated by regulatory standards.
The viscous damping force f v is proportional to velocity and is given by

f v = bu· Eq. 2-6

where:

b = viscous damping coefficient

u· = velocity

The structural damping force f s is proportional to displacement and is given by

fs = i ⋅ G ⋅ k ⋅ u Eq. 2-7

where:

G = structural damping coefficient

k = stiffness
u = displacement
i = –1 (phase change of 90 degrees)

For a sinusoidal displacement response of constant amplitude, the structural damping

force is constant, and the viscous damping force is proportional to the forcing frequency.
Figure 2-2 depicts this and also shows that for constant amplitude sinusoidal motion the
two damping forces are equal at a single frequency.
 CHAPTER 2 25
Finite Element Input Data

At this frequency,

Gk
G k = b ω∗ or b = ------- Eq. 2-8
ω∗

where ω∗ is the frequency at which the structural and viscous damping forces are equal for
a constant amplitude of sinusoidal motion.

Viscous Damping
·
f v = bu = i b ω u
Structural Damping
fs = i G k u
Damping Force f

ω∗
Forcing Frequency ω
Figure 2-2 Structural Damping and Viscous Damping Forces for Constant Amplitude
Sinusoidal Displacement

If the frequency ω∗ is the circular natural frequency ω n , Eq. 2-8 becomes

G k
b = --------- = G ω n m Eq. 2-9
ωn

Recall the definition of critical damping from Eq. 1-11

b cr = 2 km = 2mω n Eq. 2-10

Some equalities that are true at resonance ( ω n ) for constant amplitude sinusoidal
displacement are

b G
------- = ζ = ---- Eq. 2-11
b cr 2

1 1
and Q = ------ = ---- Eq. 2-12
2ζ G

where Q is the quality or dynamic magnification factor, which is inversely proportional to

the energy dissipated per cycle of vibration.

The Effect of Damping. Damping is the result of many complicated mechanisms. The
effect of damping on computed response depends on the type and loading duration of the
dynamic analysis. Damping can often be ignored for short duration loadings, such as those
26

resulting from a crash impulse or a shock blast, because the structure reaches its peak
response before significant energy has had time to dissipate. Damping is important for
long duration loadings (such as earthquakes), and is critical for loadings (such as rotating
machinery) that continually add energy to the structure. The proper specification of the
damping coefficients can be obtained from structural tests or from published literature that
provides damping values for structures similar to yours.

As is discussed in detail in “Frequency Response Analysis” on page 103 and “Transient

Response Analysis” on page 151, certain solution methods allow specific forms of
damping to be defined. The type of damping used in the analysis is controlled by both the
solution being performed and the MSC.Nastran data entries. In transient response
analysis, for example, structural damping must be converted to equivalent viscous
damping.

Structural Damping Specification. Structural damping is specified on the MATi and

PARAM,G Bulk Data entries. The GE field on the MATi entry is used to specify overall
structural damping for the elements that reference this material entry. This definition is via
the structural damping coefficient GE.
For example, the MAT1 entry:

1 2 3 4 5 6 7 8 9 10
$MAT1 MID E G NU RHO A TREF GE

MAT1 2 30.0E6 0.3 7.764E-4 0.10

specifies a structural damping coefficient of 0.1.

An alternate method for defining structural damping is through PARAM,G,r where r is the
structural damping coefficient. This parameter multiplies the stiffness matrix to obtain the
structural damping matrix. The default value for PARAM,G is 0.0. The default value
causes this source of structural damping to be ignored. Two additional parameters are
used in transient response analysis to convert structural damping to equivalent viscous
damping: PARAM,W3 and PARAM,W4.
PARAM,G and GE can both be specified in the same analysis.

Viscous Damping Specification. Viscous damping is defined by the following elements:

CDAMP1 entry Scalar damper between two degrees-of-freedom (DOFs) with

reference to a PDAMP property entry.
CDAMP2 entry Scalar damper between two DOFs without reference to a property
entry.
CDAMP3 entry Scalar damper between two scalar points (SPOINTs) with
reference to a PDAMP property entry.
CDAMP4 entry Scalar damper between two scalar points (SPOINTs) without
reference to a property entry.
 CHAPTER 2 27
Finite Element Input Data

CVISC entry Element damper between two grid points with reference to a
PVISC property entry.
CBUSH entry A generalized spring-and-damper structural element that may be
nonlinear or frequency dependent. It references a PBUSH entry.
Viscous damping for modal transient response and modal frequency response is specified
with the TABDMP1 entry.
Note that GE and G by themselves are dimensionless; they are multipliers of the stiffness.
The CDAMPi and CVISC entries, however, have damping units.

Damping is further described in “Frequency Response Analysis” on page 103 and

“Transient Response Analysis” on page 151 as it pertains to frequency and transient
response analyses.
28

2.4 Units in Dynamic Analysis

Because MSC.Nastran does not assume a particular set of units, you must ensure that the
units in your MSC.Nastran model are consistent. Because there is more input in dynamic
analysis than in static analysis, it is easier to make a mistake in units when performing a
dynamic analysis. The most frequent source of error in dynamic analysis is incorrect
specification of the units, especially for mass and damping.

Table 2-2 shows typical dynamic analysis variables, fundamental and derived units, and
common English and metric units. Note that for English units all “lb” designations are lb f .
The use of “lb” for mass (i.e., lbm ) is avoided.

Table 2-2 Engineering Units for Common Variables

Common Common
Variable Dimensions*
English Units Metric Units
Length L in m

Mass M lb-sec /in

2 kg

Time T sec sec

Area 2 2 2
L in m

Volume 3 3 3
L in m

Velocity LT
–1 in/sec m/sec

Acceleration –2 2 2
LT in/sec m/sec
Rotation – rad rad

Rotational Velocity T
–1 rad/sec rad/sec

Rotational Acceleration –2 2 2
T rad/sec rad/sec

Circular Frequency T
–1 rad/sec rad/sec

Frequency T
–1 cps; Hz cps; Hz

Eigenvalue –2 2 2 2 2
T rad /sec rad /sec
Phase Angle – deg deg

Force MLT
–2 lb N

Weight MLT
–2 lb N

Moment ML T
2 –2 in-lb N-m

Mass Density –3 2 4 3
ML lb-sec /in kg/m
 CHAPTER 2 29
Finite Element Input Data

Table 2-2 Engineering Units for Common Variables (continued)

Common Common
Variable Dimensions*
English Units Metric Units

Young’s Modulus –1 –2 2 2
ML T lb/in Pa; N/m
Poisson’s Ratio – – –

Shear Modulus –1 –2 2 2
ML T lb/in Pa; N/m

Area Moment of Inertia 4 4 4

L in m

Torsional Constant 4 4 4
L in m

Mass Moment of Inertia 2 2 2

ML in-lb-sec kg-m

Stiffness MT
–2 lb/in N/m

Viscous Damping Coefficient MT

–1 lb-sec/in N-sec/m

Stress –1 –2 2 2
ML T lb/in Pa; N/m
Strain – – –

*L denotes length
M denotes mass
T denotes time
– denotes dimensionless
30

2.5 Direct Matrix Input

The finite element approach simulates the structural properties with mathematical
equations written in matrix format. The structural behavior is then obtained by solving
these equations. Usually, all of the structural matrices are generated internally based on
the information that you provide in your MSC.Nastran model. Once you provide the grid
point locations, element connectivities, element properties, and material properties,
MSC.Nastran generates the appropriate structural matrices.

External Matrices. If structural matrices are available externally, you can input the
matrices directly into MSC.Nastran without providing all the modeling information.
Normally this is not a recommended procedure since it requires additional work on your
part. However, there are occasions where the availability of this feature is very useful and
in some cases crucial. Some possible applications are listed below:
• Suppose you are a subcontractor on a classified project. The substructure that you
are analyzing is attached to the main structure built by the main contractor. The
stiffness and mass effects of this main structure are crucial to the response of your
component, but geometry of the main structure is classified. The main contractor,
however, can provide you with the stiffness and mass matrices of the classified
structure. By reading these stiffness and mass matrices and adding them to your
MSC.Nastran model, you can account for the effect of the attached structure
without compromising security.
• Perhaps you are investigating a series of design options on a component attached
to an aircraft bulkhead. Your component consists of 500 DOFs and the aircraft
model consists of 100,000 DOFs. The flexibility of the backup structure is
somewhat important. You can certainly analyze your component by including
the full aircraft model (100,500 DOFs). However, as an approximation, you can
reduce the matrices for the entire aircraft down to a manageable size using
dynamic reduction (see “Advanced Dynamic Analysis Capabilities” on
page 293). These reduced mass and stiffness matrices can then be read and added
to your various component models. In this case, you may be analyzing a
2000-DOF system, instead of a 100,500-DOF system.
• The same concept can be extended to a component attached to a test fixture. If the
finite element model of the fixture is available, then the reduced mass and
stiffness matrices of the fixture can be input. Furthermore, there are times
whereby the flexibility of the test fixture at the attachment points can be measured
experimentally. The experimental stiffness matrix is the inverse of the measured
flexibility matrix. In this instance, this experimental stiffness matrix can be input
to your model.

One way of reading these external matrices is through the use of the direct matrix input
feature in MSC.Nastran.
 CHAPTER 2 31
Finite Element Input Data

Direct Matrix Input

The direct matrix input feature can be used to input stiffness, mass, damping, and load
matrices attached to the grid and/or scalar points in dynamic analysis. These matrices are
referenced in terms of their external grid IDs and are input via DMIG Bulk Data entries. As
shown in Table 2-3, there are seven standard kinds of DMIG matrices available in dynamic
analysis.

Table 2-3 Types of DMIG Matrices in Dynamics

Matrix G Type P Type

Stiffness K2GG K2PP
Mass M2GG M2PP
Damping B2GG B2PP
Load P2G –

The symbols for g-type matrices in mathematical format are [ K 2gg ] , [ M 2gg ] , [ B2gg ] , and { P 2g }.
The three matrices K2GG, M2GG, and B2GG must be real and symmetric. These matrices
are implemented at the g-set level (see “The Set Notation System Used in Dynamic
Analysis” on page 327 for a description of the set notation for dynamic analysis). In other
words, these terms are added to the corresponding structural matrices at the specified
DOFs prior to the application of constraints (MPCs, SPCs, etc.).

The symbols for p-type matrices in standard mathematical format are [ K 2pp ] , [ M 2pp ] , and
2
[ B pp ] . The p-set is a union of the g-set and extra points. These matrices need not be real or
symmetric. The p-type matrices are used in applications such as control systems. Only the
g-type DMIG input matrices are covered in this guide.
32

DMIG Bulk Data User Interface

In the Bulk Data Section, the DMIG matrix is defined by a single DMIG header entry
followed by a series of DMIG data entries. Each of these DMIG data entries contains a
column of nonzero terms for the matrix.

Header Entry Format:

1 2 3 4 5 6 7 8 9 10
DMIG NAME “0" IFO TIN TOUT POLAR NCOL

Column Entry Format:

DMIG NAME GJ CJ G1 C1 A1 B1

G2 C2 A2 B2 -etc.-

Example:

DMIG STIF 0 6 1

DMIG STIF 5 3 5 3 250.

5 5 -125. 6 3 -150.

Field Contents
NAME Name of the matrix.
IFO Form of matrix input:
1 = Square
9 or 2 = Rectangular
6 = Symmetric (input only the upper or lower half)
TIN Type of matrix being input:
1 = Real, single precision (one field is used per element)
2 = Real, double precision (one field per element)
3 = Complex, single precision (two fields are used per element)
4 = Complex, double precision (two fields per element)
TOUT Type of matrix to be created:
0 = Set by precision system cell (default)
1 = Real, single precision
2 = Real, double precision
3 = Complex, single precision
4 = Complex, double precision
 CHAPTER 2 33
Finite Element Input Data

Field Contents
POLAR Input format of Ai, Bi. (Integer = blank or 0 indicates real, imaginary format;
integer > 0 indicates amplitude, phase format.)
NCOL Number of columns in a rectangular matrix. Used only for IFO = 9.
GJ Grid, scalar, or extra point identification number for the column index or
column number for IFO = 9.
CJ Component number for GJ for a grid point.
Gi Grid, scalar, or extra point identification number for the row index.
Ci Component number for Gi for a grid point.
Ai, Bi Real and imaginary (or amplitude and phase) parts of a matrix element. If
the matrix is real (TIN = 1 or 2), then Bi must be blank.
34

DMIG Case Control User Interface

In order to include these matrices, the Case Control must contain the appropriate K2GG,
M2GG, or B2GG command. (Once again, only the g-type DMIG input matrices are
included in this guide.)

Examples
1. K2GG = mystiff
The above Case Control command adds terms that are defined by the DMIG Bulk
Data entries with the name “mystiff” to the g-set stiffness matrix.
2. M2GG = yourmass
The above Case Control command adds terms that are defined by the DMIG Bulk
Data entries with the name “yourmass” to the g-set mass matrix.
3. B2GG = ourdamp
The above Case Control command adds terms that are defined by the DMIG Bulk
Data entries with the name “ourdamp” to the g-set damping matrix.

Use of the DMIG entry for inputting mass and stiffness is illustrated in one of the examples
in “Real Eigenvalue Analysis” on page 35.
 NX Nastran Basic Dynamic Analysis User’s Guide

3
CHAPTER
Real Eigenvalue Analysis

■ Overview

■ Reasons to Compute Normal Modes

■ Overview of Normal Modes Analysis

■ Methods of Computation

■ Comparison of Methods

■ User Interface for Real Eigenvalue Analysis

■ Solution Control for Normal Modes Analysis

■ Examples
36

3.1 Overview
The usual first step in performing a dynamic analysis is determining the natural
frequencies and mode shapes of the structure with damping neglected. These results
characterize the basic dynamic behavior of the structure and are an indication of how the
structure will respond to dynamic loading.

Natural Frequencies. The natural frequencies of a structure are the frequencies at which
the structure naturally tends to vibrate if it is subjected to a disturbance. For example, the
strings of a piano are each tuned to vibrate at a specific frequency. Some alternate terms
for the natural frequency are characteristic frequency, fundamental frequency, resonance
frequency, and normal frequency.

Mode Shapes. The deformed shape of the structure at a specific natural frequency of
vibration is termed its normal mode of vibration. Some other terms used to describe the
normal mode are mode shape, characteristic shape, and fundamental shape. Each mode
shape is associated with a specific natural frequency.
Natural frequencies and mode shapes are functions of the structural properties and
boundary conditions. A cantilever beam has a set of natural frequencies and associated
mode shapes (Figure 3-1). If the structural properties change, the natural frequencies
change, but the mode shapes may not necessarily change. For example, if the elastic
modulus of the cantilever beam is changed, the natural frequencies change but the mode
shapes remain the same. If the boundary conditions change, then the natural frequencies
and mode shapes both change. For example, if the cantilever beam is changed so that it is
pinned at both ends, the natural frequencies and mode shapes change (see Figure 3-2).

z x
1

z x
2

z x
3

z x
4

Figure 3-1 The First Four Mode Shapes of a Cantilever Beam

 CHAPTER 3 37
Real Eigenvalue Analysis

z x
1

z x
2

z x
3

z x
4

Figure 3-2 The First Four Mode Shapes of a Simply Supported Beam

Computation of the natural frequencies and mode shapes is performed by solving an

eigenvalue problem as described in “Rigid-Body Mode of a Simple Structure” on page 42.
Next, we solve for the eigenvalues (natural frequencies) and eigenvectors (mode shapes).
Because damping is neglected in the analysis, the eigenvalues are real numbers. (The
inclusion of damping makes the eigenvalues complex numbers; see “Advanced Dynamic
Analysis Capabilities” on page 293.) The solution for undamped natural frequencies and
mode shapes is called real eigenvalue analysis or normal modes analysis.
The remainder of this chapter describes the various eigensolution methods for computing
natural frequencies and mode shapes, and it concludes with several examples.
38

3.2 Reasons to Compute Normal Modes

There are many reasons to compute the natural frequencies and mode shapes of a
structure. One reason is to assess the dynamic interaction between a component and its
supporting structure. For example, if a rotating machine, such as an air conditioner fan, is
to be installed on the roof of a building, it is necessary to determine if the operating
frequency of the rotating fan is close to one of the natural frequencies of the building. If the
frequencies are close, the operation of the fan may lead to structural damage or failure.

Decisions regarding subsequent dynamic analyses (i.e., transient response, frequency

response, response spectrum analysis, etc.) can be based on the results of a natural
frequency analysis. The important modes can be evaluated and used to select the
appropriate time or frequency step for integrating the equations of motion. Similarly, the
results of the eigenvalue analysis-the natural frequencies and mode shapes-can be used in
modal frequency and modal transient response analyses (see “Frequency Response
Analysis” on page 103 and “Transient Response Analysis” on page 151 ).

The results of the dynamic analyses are sometimes compared to the physical test results.
A normal modes analysis can be used to guide the experiment. In the pretest planning
stages, a normal modes analysis can be used to indicate the best location for the
accelerometers. After the test, a normal modes analysis can be used as a means to correlate
the test results to the analysis results.
Design changes can also be evaluated by using natural frequencies and normal modes.
Does a particular design modification cause an increase in dynamic response? Normal
modes analysis can often provide an indication.
In summary, there are many reasons to compute the natural frequencies and mode shapes
of a structure. All of these reasons are based on the fact that real eigenvalue analysis is the
basis for many types of dynamic response analyses. Therefore, an overall understanding
of normal modes analysis as well as knowledge of the natural frequencies and mode
shapes for your particular structure is important for all types of dynamic analysis.
 CHAPTER 3 39
Real Eigenvalue Analysis

3.3 Overview of Normal Modes Analysis

The solution of the equation of motion for natural frequencies and normal modes requires
a special reduced form of the equation of motion. If there is no damping and no applied
loading, the equation of motion in matrix form reduces to

[ M ] { u·· } + [ K ] [ u ] = 0 Eq. 3-1

where:

[M]
= mass matrix

[ K ] = stiffness matrix

This is the equation of motion for undamped free vibration. To solve Eq. 3-1 assume a
harmonic solution of the form

{ u } = { φ } sin ωt Eq. 3-2

where:

{ φ } = the eigenvector or mode shape

ω = is the circular natural frequency

Aside from this harmonic form being the key to the numerical solution of the problem, this
form also has a physical importance. The harmonic form of the solution means that all the
degrees-of-freedom of the vibrating structure move in a synchronous manner. The
structural configuration does not change its basic shape during motion; only its amplitude
changes.
If differentiation of the assumed harmonic solution is performed and substituted into the
equation of motion, the following is obtained:

2
– ω [ M ] { φ } sin ωt + [ K ] { φ } sin ωt = 0 Eq. 3-3

which after simplifying becomes

2
([K] – ω [M]){φ} = 0 Eq. 3-4

This equation is called the eigenequation, which is a set of homogeneous algebraic

equations for the components of the eigenvector and forms the basis for the eigenvalue
problem. An eigenvalue problem is a specific equation form that has many applications in
linear matrix algebra. The basic form of an eigenvalue problem is

[ A – λI ]x = 0 Eq. 3-5

where:
A = square matrix
λ = eigenvalues
40

I = identity matrix
x = eigenvector

In structural analysis, the representations of stiffness and mass in the eigenequation result
in the physical representations of natural frequencies and mode shapes. Therefore, the
eigenequation is written in terms of K , ω , and M as shown in Eq. 3-4 with ω 2 = λ .

There are two possible solution forms for Eq. 3-4:

1. If det ( [ K ] – ω 2 [ M ] ) ≠ 0 , the only possible solution is

{φ} = 0 Eq. 3-6

This is the trivial solution, which does not provide any valuable information from
a physical point of view, since it represents the case of no motion. (“det” denotes
the determinant of a matrix.)
2. If det ( [ K ] – ω 2 [ M ] ) = 0 , then a non-trivial solution ( { φ } ≠ 0 ) is obtained for

2
([K] – ω [M]){φ} = 0 Eq. 3-7
From a structural engineering point of view, the general mathematical eigenvalue
problem reduces to one of solving the equation of the form

2
det ( [ K ] – ω [ M ] ) = 0 Eq. 3-8
or

det ( [ K ] – λ [ M ] ) = 0 Eq. 3-9

2
where λ = ω
The determinant is zero only at a set of discrete eigenvalues λ i or ω 2i . There is an
eigenvector { φ i } which satisfies Eq. 3-7 and corresponds to each eigenvalue. Therefore,
Eq. 3-7 can be rewritten as

2
[ K – ωi M ] { φi } = 0 i = 1, 2, 3… Eq. 3-10

Each eigenvalue and eigenvector define a free vibration mode of the structure. The i-th
eigenvalue λ i is related to the i-th natural frequency as follows:

ωi
f i = ------ Eq. 3-11
2π

where:

f i = i-th natural frequency

ωi = λi

The number of eigenvalues and eigenvectors is equal to the number of degrees-of-freedom

that have mass or the number of dynamic degrees-of-freedom.
 CHAPTER 3 41
Real Eigenvalue Analysis

There are a number of characteristics of natural frequencies and mode shapes that make
them useful in various dynamic analyses. First, when a linear elastic structure is vibrating
in free or forced vibration, its deflected shape at any given time is a linear combination of
all of its normal modes

{u} = ∑ ( φ i )ξ i Eq. 3-12

i

where:

{ u } = vector of physical displacements

{ φ i } = i-th mode shape

ξ i = i-th modal displacement

Second, if [ K ] and [ M ] are symmetric and real (as is the case for all the common structural
finite elements), the following mathematical properties hold:

T
{ φi } [ M ] { φj } = 0 if i ≠ j Eq. 3-13

T
{ φ j } [ M ] { φ j } = m j = j-th generalized mass Eq. 3-14

and

T
{ φj } [ K ] { φj } = 0 if i ≠ j Eq. 3-15

T 2
{ φ j } [ K ] { φ j } = k j = j-th generalized stiffness = ω m j Eq. 3-16

Also, from Eq. 3-14 and Eq. 3-16 Rayleigh’s equation is obtained

T
2 { φj } [ K ] { φj }
ωj = -------------------------------------- Eq. 3-17
T
{ φj } [ M ] { φj }

Eq. 3-13 and Eq. 3-15 are known as the orthogonality property of normal modes, which
ensures that each normal mode is distinct from all others. Physically, orthogonality of
modes means that each mode shape is unique and one mode shape cannot be obtained
through a linear combination of any other mode shapes.
In addition, a natural mode of the structure can be represented by using its generalized
mass and generalized stiffness. This is very useful in formulating equivalent dynamic
models and in component mode synthesis (see “Advanced Dynamic Analysis
Capabilities” on page 293).

If a structure is not totally constrained in space, it is possible for the structure to displace
(move) as a rigid body or as a partial or complete mechanism. For each possible
component of rigid-body motion or mechanism, there exists one natural frequency which
is equal to zero. The zero-frequency modes are called rigid-body modes. Rigid-body
42

motion of all or part of a structure represents the motion of the structure in a stress-free
condition. Stress-free, rigid-body modes are useful in conducting dynamic analyses of
unconstrained structures, such as aircraft and satellites. Also, rigid-body modes can be
indicative of modeling errors or an inadequate constraint set.
For example, the simple unconstrained structure in Figure 3-3 has a rigid-body mode.

u1 u2

m1 m2
k

 u1  1 
ω1 = 0 { φ1 } =   =  
u
 2 1 

Figure 3-3 Rigid-Body Mode of a Simple Structure

When both masses move the same amount (as a rigid body), there is no force induced in
the connecting spring. A detailed discussion of rigid-body modes is presented in
“Rigid-body Modes” on page 87.
An important characteristic of normal modes is that the scaling or magnitude of the
eigenvectors is arbitrary. Mode shapes are fundamental characteristic shapes of the
structure and are therefore relative quantities. In the solution of the equation of motion,
the form of the solution is represented as a shape with a time-varying amplitude.
Therefore, the basic mode shape of the structure does not change while it is vibrating; only
its amplitude changes.
For example, three different ways to represent the two modes of a two-DOF structure are
shown in Figure 3-4. The graphical representation of the eigenvectors in the figure shows
the modal displacements rotated by 90 degrees in order to view the deformation better.

u1 u1 u1

m1

u2
u2 u2
m2

k2

Mode 1 Mode 2

 u1  3   300   – 0.6   u1   – 1.6   160   1.0 

φ1 =   =   =   =  { φ 2 } =   =   =   =  
 2
u 1
   100   – 0.2   2
u  0.8   – 80   – 0.5 

Figure 3-4 Representations of Mode Shapes for a Two-DOF System

 CHAPTER 3 43
Real Eigenvalue Analysis

A common misconception about mode shapes is that they define the structural response.
Again, mode shapes are relative quantities. They cannot be used alone to evaluate
dynamic behavior. As described earlier, it is the relation between the structural loading
and the natural frequencies that determines the absolute magnitude of dynamic response.
The relation of a specific loading to a set of natural frequencies provides explicit scale
factors that are used to determine the extent to which each particular mode is excited by
the loading. After the individual modal responses to a particular loading are determined,
only then can the various engineering design decisions be made with the actual (absolute)
values of stress and/or displacement. Methods that use the modal results to determine
forced response are termed modal methods or modal superposition methods. Modal
frequency response analysis and modal transient response analysis are described in
“Frequency Response Analysis” on page 103 and “Transient Response Analysis” on
page 151, respectively.

Mode Shape Normalization. Although the scaling of normal modes is arbitrary, for
practical considerations mode shapes should be scaled (i.e., normalized) by a chosen
convention. In MSC.Nastran there are three normalization choices, MASS, MAX, and
POINT normalization.
MASS normalization is the default method of eigenvector normalization. This method
scales each eigenvector to result in a unit value of generalized mass

T
{ φ j } [ M ] { φ j } = 1.0 Eq. 3-18

Numerically this method results in a modal mass matrix that is an identity matrix. This
normalization approach is appropriate for modal dynamic response calculations because
it simplifies both computational and data storage requirements. When mass normalization
is used with a model of a heavy, massive structure, the magnitude of each of the terms of
the eigenvectors is very small.
In MAX normalization, each eigenvector is normalized with respect to the largest a-set
component. (“Advanced Dynamic Analysis Capabilities” on page 293 and “The Set
Notation System Used in Dynamic Analysis” on page 327 provide discussions of the
a-set.) This normalization results in the largest a-set displacement value being set to a unit
(1.0) value. This normalization approach can be very useful in the determination of the
relative participation of an individual mode. A small generalized mass obtained using
MAX normalization may indicate such things as local modes or isolated mechanisms.
POINT normalization of eigenvectors allows you to chose a specific displacement
component at which the modal displacement is set to 1 or -1. This method is not
recommended because for complex structures the chosen component in the
non-normalized eigenvector may have a very small value of displacement (especially in
higher modes). This small value can cause larger numbers to be normalized by a small
number, resulting in possible numerical roundoff errors in mode shapes.
44

Although mode shapes are relative quantities, a number of modal quantities can be helpful
in predicting qualitative responses or in isolating troublesome modal frequencies. Since
relative strains, internal loads, and stresses develop when a structure deforms in a mode
shape, you may recover these quantities during a normal modes analyses. Basically, any
quantity that you can recover for static analysis is also available for normal modes analysis.

It is important to remember that these output quantities are based on the relative
displacements of a mode shape. The output quantities can be compared for a given mode,
but not necessarily from one mode to another. However, they can still be effectively used
in the analysis/design process.

Modal quantities can be used to identify problem areas by indicating the more highly
stressed elements. Elements that are consistently highly stressed across many or all modes
will probably be highly stressed when dynamic loads are applied.

Modal strain energy is a useful quantity in identifying candidate elements for design
changes to eliminate problem frequencies. Elements with large values of strain energy in
a mode indicate the location of large elastic deformation (energy). These elements are
those which most directly affect the deformation in a mode. Therefore, changing the
properties of these elements with large strain energy should have more effect on the
natural frequencies and mode shapes than if elements with low strain energy were
changed.
Structures with two or more identical eigenvalues are said to have repeated roots.
Repeated roots occur for structures that have a plane of symmetry or that have multiple,
identical pieces (such as appendages). The eigenvectors for the repeated roots are not
unique because many sets of eigenvectors can be found that are orthogonal to each other.
An eigenvector that is a linear combination of the repeated eigenvectors is also a valid
eigenvector. Consequently, small changes in the model can make large changes in the
eigenvectors for the repeated roots. Different computers can also find different
eigenvectors for the repeated roots. Rigid-body modes (see “Rigid-body Modes” on
page 87) represent a special case of repeated roots.
 CHAPTER 3 45
Real Eigenvalue Analysis

3.4 Methods of Computation

Seven methods of real eigenvalue extraction are provided in MSC.Nastran. These methods
are numerical approaches to solving for natural frequencies and modes shapes. The reason
for seven different numerical techniques is because no one method is the best for all
problems. While most of the methods can be applied to all problems, the choice is often
based on the efficiency of the solution process.

The methods of eigenvalue extraction belong to one or both of the following two groups:
• Transformation methods
• Tracking methods
In the transformation method, the eigenvalue equation is first transformed into a special
form from which eigenvalues may easily be extracted. In the tracking method, the
eigenvalues are extracted one at a time using an iterative procedure.
The recommended real eigenvalue extraction method in MSC.Nastran is the Lanczos
method. The Lanczos method combines the best characteristics of both the tracking and
transformation methods. For most models the Lanczos method is the best method to use.
Four of the real eigenvalue extraction methods available in MSC.Nastran are
transformation methods. These methods are
• Givens method
• Householder method
• Modified Givens method
• Modified Householder method
Two of the real eigenvalue extraction methods available in MSC.Nastran are classified as
tracking methods. These methods are
• Inverse power method
• Sturm modified inverse power method
The remainder of this section briefly describes the various methods. The theory and
algorithms behind each method can be found in the MSC.Nastran Numerical Methods User's
Guide.

Lanczos Method
The Lanczos method overcomes the limitations and combines the best features of the other
methods. It requires that the mass matrix be positive semidefinite and the stiffness be
symmetric. Like the transformation methods, it does not miss roots, but has the efficiency
of the tracking methods, because it only makes the calculations necessary to find the roots
requested by the user. This method computes accurate eigenvalues and eigenvectors.
Unlike the other methods, its performance has been continually enhanced since its
46

introduction giving it an advantage. The Lanczos method is the preferred method for most
medium- to large-sized problems, since it has a performance advantage over other
methods.

Givens and Householder Methods

The Givens and Householder modal extraction methods require a positive definite mass
matrix (all degrees-of-freedom must have mass). There is no restriction on the stiffness
matrix except that it must be symmetric. These matrices always result in real (positive)
eigenvalues. The Givens and Householder methods are the most efficient methods for
small problems and problems with dense matrices when a large portion of the eigenvectors
are needed. These methods find all of the eigenvalues and as many eigenvectors as
requested. While these methods do not take advantage of sparse matrices, they are
efficient with the dense matrices sometimes created using dynamic reduction (see
“Advanced Dynamic Analysis Capabilities” on page 293).
The Givens and Householder methods fail if the mass matrix is not positive definite. To
minimize this problem, degrees-of-freedom with null columns are removed by the
automatic application of static condensation (see “Advanced Dynamic Analysis
Capabilities” on page 293) called auto-omit. Applying the auto-omit process is a
precaution and may not remove all possible causes of mass matrix singularity, such as a
point mass offset from a grid point, but it greatly improves the reliability and convenience
of the Givens and Householder methods.
Givens and Householder methods use different transformation schemes to obtain the
eigenvalues. For problems in which no spill occurs (i.e., all of the matrices fit in your
computer's main memory), the Householder method costs about half as much as the
Givens method for vector processing computers. In addition, the Householder method can
take advantage of parallel processing computers.

Modified Givens and Modified Householder Methods

The modified Givens and modified Householder methods are similar to their standard
methods with the exception that the mass matrix can be singular. Although the mass
matrix is not required to be nonsingular in the modified methods, a singular mass matrix
can produce one or more infinite eigenvalues. Due to roundoff error, these infinite
eigenvalues appear in the output as very large positive or negative eigenvalues. To reduce
the incidence of such meaningless results, degrees-of-freedom with null masses are
eliminated by automatic static condensation as in the case of the unmodified methods.
The modified methods require more computer time than the standard methods.

Automatic Givens and Automatic Householder Methods

Many times you may not know whether the mass matrix is singular. To assist you in
choosing the appropriate method, two options–automatic Givens and automatic
Householder–are available. Initially the automatic methods use the standard methods. In
the first step of the method, if the mass matrix is not well-conditioned for decomposition,
the method shifts to the corresponding modified method. The modified methods are more
 CHAPTER 3 47
Real Eigenvalue Analysis

expensive and may introduce numerical noise due to the shift, but they resolve most of the
numerical problems of the ill-conditioned mass matrix. The automatic methods, therefore,
use the modified methods when necessary for numerical stability but use the standard
methods when the numerical stability is accurate.

Inverse Power Method

The inverse power method is a tracking method since the lowest eigenvalue and
eigenvector in the desired range are found first. Then their effects are "swept" out of the
dynamic matrix, the next higher mode is found, and its effects are "swept" out, and so on.
Hence, the term "tracking," which means that one root at a time is found. In addition, each
root is found via an iterative procedure. (The classical literature often refers to this method
as the inverse iteration method with sweeping.)

However, the inverse power method can miss modes, making it unreliable. The Sturm
modified inverse power method is a more reliable tracking method.

Sturm Modified Inverse Power Method

This method is similar to the inverse power method except that it uses Sturm sequence
logic to ensure that all modes are found. The Sturm sequence check determines the
number of eigenvalues below a trial eigenvalue, then finds all of the eigenvalues below this
trial eigenvalue until all modes in the designed range are computed. This process helps to
ensure that modes are not missed.
The Sturm modified inverse power method is useful for models in which only the lowest
few modes are needed. This method is also useful as a backup method to verify the
accuracy of other methods.
48

3.5 Comparison of Methods

Since MSC.Nastran provides a variety of real eigensolution methods, you must decide
which is best for your application. The best method for a particular model depends on four
factors: the size of the model (the total number of degrees-of-freedom as well as the
number of dynamic degrees-of-freedom), the number of eigenvalues desired, the available
real memory of your computer, and the conditioning of the mass matrix (whether there are
massless degrees-of-freedom). In general, the Lanczos method is the most reliable and
efficient, and is the recommended choice.
For small, dense models whose matrices fit into memory, we recommend using one of the
automatic methods (automatic Householder or automatic Givens). Both automatic
Householder and automatic Givens run modified methods if the mass matrix is singular;
however, they run the unmodified methods, which are faster, if the mass matrix is not
singular. Of the two automatic methods, the automatic Householder method runs faster
on computers with vector processing and also supports parallel processing computers.
Note that most real world problems are not small and dense, unless you use reductive
methods such as General Dynamic Reduction (GDR) or superelements.

The Sturm modified inverse power method can be the best choice when the model is too
large to fit into memory, only a few modes are needed, and a reasonable eigenvalue search
range is specified. This method is also a backup method for the other methods and is used
when a check of the other methods' results is needed.

For medium to large models the Lanczos method is the recommended method. In addition
to its reliability and efficiency, the Lanczos method supports sparse matrix methods that
substantially increase its speed and reduce disk space requirements.

Table 3-1 Comparison of Eigenvalue Methods

Method

Modified Sturm
Givens,
Givens, Inverse Power Modified Lanczos
Householder
Householder Inverse Power
Reliability High High Poor (can miss High High
modes)

Relative Cost:

Few Modes Medium Medium Low Low Medium

Many Modes High High High High Medium

Limitations Cannot analyze Expensive for Can miss modes Expensive for Difficulty with
singular [M] many modes many modes massless
mechanisms

Expensive for Expensive for Expensive for

problems that do problems that do many modes
not fit in memory not fit in memory
 CHAPTER 3 49
Real Eigenvalue Analysis

Table 3-1 Comparison of Eigenvalue Methods (continued)

Method

Modified Sturm
Givens,
Givens, Inverse Power Modified Lanczos
Householder
Householder Inverse Power
Best Small, dense Small, dense To detemine a To detemine a Medium to large
Application matrices that fit matrices that fit few modes few modes models
in memory in memory

Use with dynamic Use with dynamic Backup method

reduction reduction
(Chapter 11) (Chapter 11)
50

3.6 User Interface for Real Eigenvalue Analysis

The EIGR and EIGRL Bulk Data entries define the method and select the parameters that
control the eigenvalue extraction procedure. The EIGRL entry is used for the Lanczos
method, and the EIGR entry is used for all of the other methods.

User Interface for the Lanczos Method

The EIGRL entry has the following format:

Format:

1 2 3 4 5 6 7 8 9 10
EIGRL SID V1 V2 ND MSGLVL MAXSET SHFSCL NORM

option_1=value_1, option_2=value_2, etc.

Example:

EIGRL 1 0.1 3.2

Field Contents
SID Set identification number. (Unique Integer > 0)
V1, V2 The V1 field defines the lower frequency bound; the V2 field defines the
upper frequency field. For vibration analysis: frequency range of
interest. For buckling analysis: eigenvalue range of interest. (Real or
blank, – 5 × 10 16 ≤ V1 < V2 ≤ 5. × 10 16 )
ND Number of eigenvalues and eigenvectors desired. (Integer > 0 or blank)
MSGLVL Diagnostic level. (0 ≤ Integer ≤ 4; Default = 0)
MAXSET Number of vectors in block or set. (1 ≤ Integer ≤ 15; Default = 7)
SHFSCL Estimate of the first flexible mode natural frequency. (Real or blank)
NORM Method for normalizing eigenvectors (Character: "MASS" or "MAX")
MASS Normalize to unit value of the generalized mass. Not
available for buckling analysis. (Default for normal modes
analysis.)
MAX Normalize to unit value of the largest displacement in the
analysis set. Displacements not in the analysis set may be
larger than unity. (Default for buckling analysis.)

Examples of the results of using explicit or default values for the V1, V2, and ND fields are
shown in Table 3-2. The defaults on the EIGRL entry are designed to provide the
minimum number of roots in cases where the input is ambiguous. Alternatively, you can
 CHAPTER 3 51
Real Eigenvalue Analysis

write the entry above using the new free field format. You must specify certain new
parameters such as ALPH,NUMS, and Fi when using the free field format. See the NX
Nastran Quick Reference Guide for details.

1 2 3 4 5 6 7 8 9 10
EIGRL 1

V1 = 0.1, V2 = 3.2

Table 3-2 Number and Type of Roots Found with the EIGRL Entry

Case V1 V2 ND Number and Type of Roots Found

1 V1 V2 ND Lowest ND in range or all in range, whichever is smaller
2 V1 V2 All in range
3 V1 ND Lowest ND in range [V1, ∞]
4 V1 Lowest root in range [V1, ∞]
5 ND Lowest ND roots in [-∞, ∞]
6 Lowest root
7 V2 ND Lowest ND or all in range [-∞, V2], whichever is smaller
8 V2 All below V2

The MSGLVL field of the EIGRL entry is used to control the amount of diagnostic output.
The value of 0 produces no diagnostic output. The values 1, 2, or 3 provide more output
with the higher values providing increasingly more output. In some cases, higher
diagnostic levels may help to resolve difficulties with special modeling problems.
The MAXSET field is used to control the block size. The default value of 7 is recommended
for most applications. There may be special cases where a larger value may result in
quicker convergence of many multiple roots or a lower value may result in more efficiency
when the structure is lightly coupled. However, the default value has been chosen after
reviewing the results from a wide range of problems on several different computer types
with the goal of minimizing the computer time. A common occurrence is for the block size
to be reset by MSC.Nastran during the run because there is insufficient memory for a block
size of 7. Computational efficiency tends to degrade as the block size decreases. Therefore,
it is, important to examine the eigenvalue analysis summary output to determine that
MSC.Nastran has sufficient memory to use an efficient block size. A smaller block size may
be more efficient when only a few roots are requested. The minimum recommended block
size is 2.
The SHFSCL field allows a user-designated shift to be used to improve performance
especially when large mass techniques are used in enforced motion analysis (see “Enforced
Motion” on page 199). Large mass techniques can cause a large gap between the rigid
52

body (see “Rigid-body Modes” on page 87) and flexible frequencies, which can degrade
performance of the Lanczos method or cause System Fatal Message 5299. When SHFSCL
is used, its value should be set close to the expected first nonzero natural frequency.

The Lanczos method normalizes (i.e., scales) the computed eigenvectors using the MASS
or MAX method. These methods are specified using the NORM field (new for Version 68).
The MASS method normalizes to a unit value of the generalized mass (i.e., m j = 1.0 ). The
MAX method normalizes to a unit value of the largest component in the a-set (see
“Advanced Dynamic Analysis Capabilities” on page 293). The default is MASS.

You can use the continuation entry to specify V1, V2, ND, MSGLVL, MAXSET, SHFSCL
and NORM if you have not specified them on the parent entry. To apply the continuation
entry use the following format: 'option_i=value_i', e.g., ND=6. Using the continuation
entry is the only way to specify the three new options, ALPH, NUMS and Fi.

NUMS The number of segments that a frequency range will be broken into for
parallel processing. You must define a value greater than 1 to take
advantage of parallel processing. You may also specify NUMS using
the NUMSEG keyword on the NASTRAN statement. If you specify
both, then NUMS takes precedence.
Fi Directly specifies the upper frequencies of each segment, such that
V1 < F1 < F2 < ... F15 < V2.
ALPH Automatically generates the Fi values based on the following formula:

i NUMS
Fi = ( V2 – V1 ) ⋅ ( 1.0 – ALPH ) ⁄ ( 1.0 – ALPH )

If you specify both ALPH and Fi, then Fi takes precedence over ALPH as long as they are
consistent. If ALPH is multiplied by 100, it may be specified on the FRQSEQ keyword of
the NASTRAN statement.
For a detailed description of the EIGRL input, see the NX Nastran Numerical Methods
User’s Guide.

User Interface for the Other Methods

The data entered on the EIGR entry selects the eigenvalue method and the frequency range
or number of required roots. The basic format of the Bulk Data entry is as follows:

1 2 3 4 5 6 7 8 9 10
EIGR SID METHOD F1 F2 NE ND

NORM G C
 CHAPTER 3 53
Real Eigenvalue Analysis

The METHOD field selects the eigenvalue method from the following list:

AGIV Automatic Givens method

AHOU Automatic Householder method
GIV Givens method
HOU Householder method
INV Inverse power
MGIV Modified Givens method
MHOU Modified Householder method
SINV Sturm modified inverse power method

The F1 field specifies the lowest frequency of interest in the eigenvalue extraction. The F2
field specifies the highest frequency of interest in the eigenvalue extraction. The units are
cycles per unit time. The value for F1 must be greater than or equal to 0.0.
The NE field is used by the INV method only. It defines the estimated number of roots in
the range. A good estimate results in a more efficient solution. A high estimate helps to
ensure that all modes are computed within the range.

The ND field is used to specify the desired number of roots (for tracking methods) or
eigenvectors (for transformation methods), beginning with F1.
The NORM field on the continuation entry is used to specify the method of eigenvector
normalization. The choices are

MASS Mass normalization (default–if used, the continuation entry is not required).
MAX Normalization to a unit value of the largest component in the analysis set.
POINT Normalization to a unit value at a user-specified a-set grid pointG and
component C.
54

There is an interrelationship among the F1, F2, and ND fields on the EIGR entry as defined
in Table 3-3.

Table 3-3 Relationship Between the METHOD Field and Other Fields

METHOD Field
Field GIV, MGIV, AGIV, HOU, MHOU, or
INV or SINV
AHOU
F1, F2 Frequency range of interest. F1 must Frequency range of interest. If ND is
be input. If METHOD = “SINV” and not blank, F1 and F2 are ignored. If
ND is blank, then F2 must be input. ND is blank, eigenvectors are found
whose natural frequencies lie in the
(Real ≥ 0.0)
range between F1 and F2.

(Real ≥ 0.0; F1 < F2)

NE Estimate of number of roots in range Not used
(Required for METHOD = “INV”).
Not used by “SINV” method.
(Integer > 0)
ND Desired number of roots. If this field Desired number of eigenvectors. If
is blank and METHOD = “SINV”, ND is zero, the number of
then all roots between F1 and F2 are eigenvectors is determined from F1
searched, and the limit is 600 roots. and F2. If all three are blank, then ND
is automatically set to one more than
(Integer > 0, Default is 3 ∗ NE for
the number of degrees-of-freedom
METHOD = “INV” only)
listed on SUPORT entries.
(Integer ≥ 0; Default = 0)

The rules for METHOD = GIV, HOU, MGIV, MHOU, AGIV, and AHOU are identical. If
any of these methods are selected, MSC.Nastran finds all of the eigenvalues but only
computes the eigenvectors specified by F1 and F2 or those specified by ND (the desired
number). F1 and F2 specify the lower and upper bounds of the frequency range in which
eigenvectors are computed, and ND specifies the number of eigenvectors, beginning with
the lowest (or the first rigid-body mode, if present). If F1, F2, and ND entries are present,
ND takes precedence.

If METHOD = SINV, the values of F1, F2, and ND determine the number of eigenvalues
and eigenvectors that are computed. These entries also provide hints to help MSC.Nastran
find the eigenvalues. F1 and F2 specify the frequency range of interest within which
MSC.Nastran searches for modes. MSC.Nastran attempts to find all of the modes in the
range between F1 and F2 or the number specified by ND, whichever is less. If searching
stops because ND modes are found, there is no guarantee that they are the lowest
eigenvalues. If ND modes are not found in the range of interest, SINV usually finds one
mode (or possibly more) outside the range F1 and F2 before stopping the search.
 CHAPTER 3 55
Real Eigenvalue Analysis

The SINV method is particularly efficient when only a small number of eigenvalues and
eigenvectors are to be computed. Often only the lowest mode is of interest. The following
example illustrates an EIGR entry which extracts only the lowest nonzero eigenvalue.

1 2 3 4 5 6 7 8 9 10
EIGR 13 SINV 0.0 0.01 1

It is assumed in the example above that the frequency of the lowest mode is greater than
0.01 cycles per unit time. MSC.Nastran finds one eigenvalue outside the range F1, F2, and
then stops the search. The eigenvalue found is the lowest nonzero eigenvalue (or a
member of the lowest closely spaced cluster of eigenvalues in cases with close roots)
provided that there are no negative eigenvalues and that the SUPORT entry has been used
to specify the correct number of zero eigenvalues (see “Rigid-body Modes” on page 87).

The following examples demonstrate the use of the EIGR data entry.

1 2 3 4 5 6 7 8 9 10
EIGR 1 AHOU 10

In this example, the automatic Householder method is selected, and the lowest 10 modes
are requested. Since the default MASS eigenvector normalization is requested, no
continuation entry is needed.

EIGR 2 AHOU 100.

MAX

In this example, the same method is requested, but all the modes below 100 cycles per unit
time are requested with MAX vector normalization.

EIGR 3 SINV 0.1 100.0 6

POINT 32 3

In this example, the Sturm modified inverse power method is requested for the first six
modes found in the range specified (0.1 to 100 Hz). The POINT normalization method is
requested with each eigenvector scaled such that grid point 32 in the T3 direction has a
magnitude of 1.0. Note that this degree-of-freedom must be in the a-set.
56

3.7 Solution Control for Normal Modes Analysis

This section describes input required for the selection and control of normal modes
analysis.

Executive Control Section

You can run a normal modes analysis, as an independent solution, in SOL 103 of the
Structured Solution Sequences or in SOL 3 of the Rigid Format Solution Sequences. We
recommend using SOL 103 since it contains our most recent enhancements. The Executive
Control Section can also contain diagnostic DIAG16, which prints the iteration information
used in the INV or SINV method. You may also run a normal modes analysis as part of the
other solution sequences (such as modal transient response, modal frequency response,
design optimization, and aeroelasticity).

Case Control Section

The most important eigenanalysis command in the Case Control is the METHOD
command. This command is required. The set identification number specified by the
METHOD Case Control command refers to the set identification number of an EIGR or
EIGRL entry in the Bulk Data.

When a modal analysis is performed, the MSC.Nastran output file contains various
diagnostic messages and an eigenvalue analysis summary. Optional grid and element
output are available using standard Case Control output requests. Eigenvectors are
printed only if a DISPLACEMENT or VECTOR command is included. These requests are
summarized in Table 3-4.

Table 3-4 Eigenvalue Extraction Output Requests

Grid Output
DISPLACEMENT Requests the eigenvector (mode shape) for a set of grid points.
(or VECTOR)
GPFORCE Requests the grid point force balance table to be computed for
each mode for a set of grid points.
GPSTRESS Requests grid point stresses to be computed for a set of grid
points. This request must be accompanied by the ELSTRESS
Case Control re quest and the definition of stress surfaces
and/or stress volumes in the OUTPUT(POST) section of the
Case Control. This request also requires the use of Rigid
Format Alter RF3D81 when used in SOL 3.
SPCFORCE Requests forces of single-point constraint to be computed for
a set of grid points for each mode.
 CHAPTER 3 57
Real Eigenvalue Analysis

Table 3-4 Eigenvalue Extraction Output Requests (continued)

Element Output
ELSTRESS (or Requests the computation of modal stresses for a set of
STRESS) elements for each mode.
ESE Requests the computation of modal element strain energies
for a set of elements for each mode.
ELFORCE (or Requests the computation of modal element forces for a set of
FORCE) elements for each mode.
STRAIN Requests the computation of modal element strains for a set of
ele ments.
Miscellaneous
MODES A special Case Control request that permits selective output
requests to be processed on selective modes.

Bulk Data Section

In addition to Bulk Data entries required to define the structural model, the only other
required Bulk Data entry is the eigenvalue selection entry EIGR or EIGRL. The EIGR entry
is used to select the modal extraction parameters for the inverse power, Sturm modified
inverse power, Givens, Householder, modified Givens, modified Householder, automatic
Givens, and automatic Householder methods. The EIGRL entry is used to select the modal
extraction parameters for the Lanczos method.
58

3.8 Examples
This section provides several normal-modes analysis examples showing the input and
output. These examples are as follows:

Number of Element Output Analysis

Model Units
Grid Points Types Requests Method
bd03two 3 CELAS2 DISPLACEMENT AHOU Metric
CONM2 SPCFORCE
ELFORCE
bd03bar1, 11 CBAR DISPLACEMENT SINV Metric
bd03bar2
bd03bkt 236 CQUAD4 DISPLACEMENT Lanczos English
CONM2 STRESS
RBE2 ESE
MODES
bd03car 972 CQUAD4 DISPLACEMENT Lanczos English
CTRIA3 ESE
CELAS2
bd03fix 8157 CHEXA DISPLACEMENT Lanczos English
CPENTA
bd03plt1, 81 CQUAD4 DISPLACEMENT Lanczos Metric
bd03plt2
bd03dmi 7 CBAR None Lanczos Metric
DMIG

These examples are described in the sections that follow.

Two-DOF Model
This example is a restrained two-DOF model with two springs and two masses as
illustrated in Figure 3-5.
 CHAPTER 3 59
Real Eigenvalue Analysis

Grid Point 1 m 1 = 0.1 kg

k 1 = 100 N/m

Grid Point 2 m 2 = 10 kg

k 2 = 1.0E4 N/m
z

Figure 3-5 Two-DOF Model

The masses are constrained to deflect in only the y-direction. The example illustrates
normal modes analysis (SOL 103) using automatic selection of the Householder or
modified Householder method (METHOD = AHOU on the EIGR entry). The eigenvectors
are normalized to the unit value of the largest displacement component (NORM = MAX on
the EIGR entry). The input file is shown in Listing 3-1.
60

Listing 3-1 Input File for the Two-DOF Model

$ FILE bd03two.dat
$
$ TWO DOF SYSTEM
$ CHAPTER 3, NORMAL MODES
$
TIME 5
SOL 103 $ NORMAL MODES ANALYSIS
CEND
$
TITLE = TWO DOF SYSTEM
SUBTITLE = NORMAL MODES ANALYSIS
$
$ SELECT SPC
SPC = 10
$
$ SELECT EIGR ENTRY
METHOD = 99
$
$ SELECT OUTPUT
SET 1 = 1,2
DISPLACEMENT = 1
SET 2 = 3
SPCFORCE = 2
SET 3 = 11,12
ELFORCE = 3
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$EIGR SID METHOD F1 F2 NE ND +EIG
$+EIG NORM G C
EIGR 99 AHOU 0. 200. +EIG1
+EIG1 MAX
$
GRID 1 0. 2. 0.
GRID 2 0. 1. 0.
GRID 3 0. 0. 0.
GRDSET 13456
CONM2 1 1 0.1
CONM2 2 2 10.0
CELAS2 11 100.0 1 2 2 2
CELAS2 12 1.0E4 2 2 3 2
SPC 10 3 2
$
ENDDATA

The printed output is shown in Figure 3-6. The eigenvalue summary lists the eigenvalue
2
ω n , circular frequency ω n (radians per second), natural frequency f n (cycles per second),
generalized mass (see Eq. 3-14), and generalized stiffness (see Eq. 3-16) for each mode. The
eigenvectors, SPC forces, and spring forces are shown for each mode.
 CHAPTER 3 61
Real Eigenvalue Analysis

R E A L E I G E N V A L U E S

MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED

NO. ORDER MASS STIFFNESS
1 1 9.048751E+02 3.008114E+01 4.787562E+00 1.904875E-01 1.723674E+02
2 2 1.105125E+03 3.324342E+01 5.290854E+00 2.105125E-01 2.326426E+02

EIGENVALUE = 9.048751E+02
CYCLES = 4.787562E+00 R E A L E I G E N V E C T O R N O . 1
POINT ID. TYPE T1 T2 T3 R1 R2 R3
1 G 0.0 1.000000E+00 0.0 0.0 0.0 0.0
2 G 0.0 9.512492E-02 0.0 0.0 0.0 0.0

EIGENVALUE = 1.105125E+03
CYCLES = 5.290854E+00 R E A L E I G E N V E C T O R N O . 2

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 1.000000E+00 0.0 0.0 0.0 0.0
2 G 0.0 -1.051249E-01 0.0 0.0 0.0 0.0

EIGENVALUE = 9.048751E+02
F O R C E S O F S I N G L E - P O I N T C O N S T R A I N T

POINT ID. TYPE T1 T2 T3 R1 R2 R3

3 G 0.0 -9.512491E+02 0.0 0.0 0.0 0.0

EIGENVALUE = 1.105125E+03
F O R C E S O F S I N G L E - P O I N T C O N S T R A I N T

POINT ID. TYPE T1 T2 T3 R1 R2 R3

3 G 0.0 1.051249E+03 0.0 0.0 0.0 0.0

EIGENVALUE = 9.048751E+02
F O R C E S I N S C A L A R S P R I N G S ( C E L A S 2 )
ELEMENT FORCE ELEMENT FORCE ELEMENT FORCE ELEMENT FORCE
ID. ID. ID. ID.
11 9.048751E+01 12 9.512491E+02

EIGENVALUE = 1.105125E+03
F O R C E S I N S C A L A R S P R I N G S ( C E L A S 2 )
ELEMENT FORCE ELEMENT FORCE ELEMENT FORCE ELEMENT FORCE
ID. ID. ID. ID.
11 1.105125E+02 12 -1.051249E+03

Figure 3-6 Output from the Two-DOF System

62

Cantilever Beam Model

This example is a fixed-free aluminum cantilever beam with properties as shown in
Figure 3-7.

y
y

x z
L
r

4
L = 3.0 m r = 0.014 m J = 6.0E-8 m
2 4 3
A = 6.158E-4 m I1 = I2 = 3.0E-8 m ρ w = 2.65E4 N/m
2
E = 7.1E10 N/m ν = 0.33

Nonstructural = 2.414 N/m

Weight

Figure 3-7 Cantilever Beam Model

The ρ w term is the weight density and must be converted to mass density ρ m for
consistency of units. PARAM,WTMASS is used to convert this weight density to mass
density. WTMASS = 1 ⁄ g = 1 ⁄ 9.81 = 0.102 sec 2 ⁄ m where g is the acceleration of gravity in
2 3
m/sec . Therefore, ρ m = ρ w ⋅ WTMASS = 2.65E4 ⋅ ( 0.102 ) = 2703 kg/m . The nonstructural
weight of 2.414 N/m is added to the beam. This nonstructural weight per length is also
scaled by PARAM,WTMASS.
The example illustrates normal modes analysis (SOL 103) using the Sturm modified
inverse power method (METHOD = SINV on the EIGR entry). Mass normalization (the
default) is chosen for the eigenvectors. All frequencies between 0 and 50 Hz are requested.
Two models are run. In the first model, manufacturing tolerances make the cross section
slightly out of round, making I1 and I2 slightly different. In the second model, the cross
section is perfectly round, making I1 and I2 identical.

Consider the first model. Due to the manufacturing tolerances, I1 = 2.9E-8 m 4 and
4
I2 = 3.1E-8 m . The input file is shown in Listing 3-2.
 CHAPTER 3 63
Real Eigenvalue Analysis

Listing 3-2 Input File for the First Beam Model

$ FILE bd03bar1.dat
$
$ CANTILEVER BEAM MODEL
$ CHAPTER 3, NORMAL MODES
$
SOL 103 $ NORMAL MODES ANALYSIS
TIME 10
CEND
$
TITLE = CANTILEVER BEAM
SUBTITLE = NORMAL MODES
LABEL = MODEL 1 (I1 NE I2)
$
SPC = 1
$
$ OUTPUT REQUEST
DISPLACEMENT = ALL
$
$ SELECT EIGR ENTRY
METHOD = 10
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$EIGR SID METHOD F1 F2 NE ND +EIG
$+EIG NORM G C
EIGR 10 SINV 0. 50.
$
$ ALUMINUM PROPERTIES:
$ E = 7.1E10 N/m**2, NU = 0.33, RHO = 2.65E4 N/m**3 (W E I G H T DENSITY)
$MAT1 MID E G NU RHO
MAT1 1 7.1+10 0.33 2.65+4
$
$ CONVERT WEIGHT TO MASS: MASS = (1/G)*WEIGHT
$ G = 9.81 m/sec**2 --> WTMASS = 1/G = 0.102
PARAM WTMASS 0.102
$
$ I1 AND I2 SLIGHTLY DIFFERENT DUE TO MANUFACTURING TOLERANCE
$ ADD NONSTRUCTURAL WEIGHT OF 2.414 N/M
PBAR 1 1 6.158-4 2.9-8 3.1-8 6.-8 2.414
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
CBAR 1 1 1 2 0. 1. 0.
CBAR 2 1 2 3 0. 1. 0.
CBAR 3 1 3 4 0. 1. 0.
CBAR 4 1 4 5 0. 1. 0.
CBAR 5 1 5 6 0. 1. 0.
CBAR 6 1 6 7 0. 1. 0.
CBAR 7 1 7 8 0. 1. 0.
CBAR 8 1 8 9 0. 1. 0.
CBAR 9 1 9 10 0. 1. 0.
CBAR 10 1 10 11 0. 1. 0.
GRID 1 0.0 0. 0.
GRID 2 0.3 0. 0.
GRID 3 0.6 0. 0.
GRID 4 0.9 0. 0.
GRID 5 1.2 0. 0.
GRID 6 1.5 0. 0.
GRID 7 1.8 0. 0.
GRID 8 2.1 0. 0.
GRID 9 2.4 0. 0.
GRID 10 2.7 0. 0.
GRID 11 3.0 0. 0.
SPC1 1 123456 1
$
ENDDATA
64

The first two resulting y-direction modes are illustrated in Figure 3-8. Displacements in
the y-direction displacements are controlled by the I1 term. Because the structure is also
free to displace in the z-direction, similar modes occur in that direction and are controlled
by the I2 term.

Figure 3-8 First Two Mode Shapes in the Y-Direction

Printed output is shown in Figure 3-9. Note that modes 1 and 3 are y-direction (T2) modes
and modes 2 and 4 are z-direction (T3) modes.
 CHAPTER 3 65
Real Eigenvalue Analysis

E I G E N V A L U E A N A L Y S I S S U M M A R Y (STURM INVERSE POWER)

NUMBER OF EIGENVALUES EXTRACTED . . . . . . 6
NUMBER OF TRIANGULAR DECOMPOSITIONS . . . . 9
TOTAL NUMBER OF VECTOR ITERATIONS . . . . . 59
REASON FOR TERMINATION: ALL EIGENVALUES FOUND IN RANGE.

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 1.629657E+02 1.276580E+01 2.031740E+00 1.000000E+00 1.629657E+02
2 2 1.742047E+02 1.319866E+01 2.100632E+00 1.000000E+00 1.742047E+02
3 3 6.258656E+03 7.911166E+01 1.259101E+01 1.000000E+00 6.258656E+03
4 4 6.690287E+03 8.179417E+01 1.301795E+01 1.000000E+00 6.690287E+03
5 6 4.809111E+04 2.192968E+02 3.490218E+01 1.000000E+00 4.809111E+04
6 5 5.140773E+04 2.267327E+02 3.608563E+01 1.000000E+00 5.140773E+04

EIGENVALUE = 1.629657E+02
CYCLES = 2.031740E+00 R E A L E I G E N V E C T O R N O . 1

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 0.0 1.391872E-02 2.259793E-10 0.0 -1.470464E-09 9.057010E-02
3 G 0.0 5.301210E-02 8.606855E-10 0.0 -2.725000E-09 1.678406E-01
4 G 0.0 1.133067E-01 1.839608E-09 0.0 -3.765826E-09 2.319480E-01
5 G 0.0 1.908986E-01 3.099362E-09 0.0 -4.598456E-09 2.832321E-01
6 G 0.0 2.820258E-01 4.578870E-09 0.0 -5.232973E-09 3.223138E-01
7 G 0.0 3.831632E-01 6.220901E-09 0.0 -5.685052E-09 3.501587E-01
8 G 0.0 4.911331E-01 7.973862E-09 0.0 -5.976744E-09 3.681249E-01
9 G 0.0 6.032288E-01 9.793808E-09 0.0 -6.137029E-09 3.779972E-01
10 G 0.0 7.173455E-01 1.164657E-08 0.0 -6.202180E-09 3.820101E-01
11 G 0.0 8.321185E-01 1.350998E-08 0.0 -6.215970E-09 3.828595E-01

EIGENVALUE = 1.742047E+02
CYCLES = 2.100632E+00 R E A L E I G E N V E C T O R N O . 2

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G -1.363453E-33 2.249974E-10 -1.391872E-02 0.0 9.057010E-02 1.464532E-09
3 G -2.693333E-33 8.574880E-10 -5.301210E-02 0.0 1.678406E-01 2.716750E-09
4 G -3.956893E-33 1.834005E-09 -1.133067E-01 0.0 2.319480E-01 3.758543E-09
5 G -5.123021E-33 3.092054E-09 -1.908986E-01 0.0 2.832321E-01 4.594735E-09
6 G -6.163003E-33 4.571193E-09 -2.820258E-01 0.0 3.223138E-01 5.234397E-09
7 G -7.051232E-33 6.214489E-09 -3.831632E-01 0.0 3.501587E-01 5.692031E-09
8 G -7.765834E-33 7.970303E-09 -4.911331E-01 0.0 3.681249E-01 5.988579E-09
9 G -8.289217E-33 9.794348E-09 -6.032288E-01 0.0 3.779972E-01 6.152242E-09
10 G -8.608492E-33 1.165196E-08 -7.173455E-01 0.0 3.820101E-01 6.219055E-09
11 G -8.715798E-33 1.352052E-08 -8.321185E-01 0.0 3.828595E-01 6.233260E-09

EIGENVALUE = 6.258656E+03
CYCLES = 1.259101E+01 R E A L E I G E N V E C T O R N O . 3

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 0.0 -7.568120E-02 -9.828895E-08 0.0 5.937006E-07 -4.571418E-01
3 G 0.0 -2.464387E-01 -3.200556E-07 0.0 8.257535E-07 -6.358200E-01
4 G 0.0 -4.318525E-01 -5.608564E-07 0.0 7.289605E-07 -5.612913E-01
5 G 0.0 -5.632111E-01 -7.314543E-07 0.0 3.723918E-07 -2.867391E-01
6 G 0.0 -5.916957E-01 -7.684467E-07 0.0 -1.426249E-07 1.098159E-01
7 G 0.0 -4.941767E-01 -6.417950E-07 0.0 -6.984853E-07 5.378201E-01
8 G 0.0 -2.743728E-01 -3.563287E-07 0.0 -1.184615E-06 9.121327E-01
9 G 0.0 4.170797E-02 5.417504E-08 0.0 -1.522753E-06 1.172494E+00
10 G 0.0 4.159041E-01 5.401546E-07 0.0 -1.689206E-06 1.300660E+00
11 G 0.0 8.124724E-01 1.055190E-06 0.0 -1.730573E-06 1.332512E+00

EIGENVALUE = 6.690287E+03
CYCLES = 1.301795E+01 R E A L E I G E N V E C T O R N O . 4

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 3.698388E-22 1.207686E-08 7.568131E-02 0.0 -4.571424E-01 1.427123E-08
3 G 7.305710E-22 -2.529794E-08 2.464390E-01 0.0 -6.358204E-01 -3.100643E-07
4 G 1.073314E-21 -1.866785E-07 4.318528E-01 0.0 -5.612913E-01 -7.615683E-07
5 G 1.389628E-21 -4.615435E-07 5.632114E-01 0.0 -2.867385E-01 -1.006117E-06
6 G 1.671725E-21 -7.409781E-07 5.916957E-01 0.0 1.098168E-01 -7.569575E-07
7 G 1.912658E-21 -8.602044E-07 4.941765E-01 0.0 5.378208E-01 4.733732E-08
8 G 2.106496E-21 -6.796384E-07 2.743725E-01 0.0 9.121327E-01 1.181005E-06
9 G 2.248464E-21 -1.591279E-07 -4.170818E-02 0.0 1.172493E+00 2.239405E-06
10 G 2.335068E-21 6.217935E-07 -4.159040E-01 0.0 1.300659E+00 2.875692E-06
11 G 2.364175E-21 1.520816E-06 -8.124720E-01 0.0 1.332511E+00 3.057266E-06

Figure 3-9 Printed Results from the First Model

66

Now, consider the second model for which I1 and I2 are identical. Printed output is shown
in Figure 3-10. Note that modes 1 and 2; 3 and 4; 5 and 6; etc., have identical frequencies;
this is a case of repeated roots. Also note that the eigenvectors are not pure y- or pure
z-translation (as they were in the first model); the eigenvectors are linear combinations of
the y and z modes since this model has repeated roots.

E I G E N V A L U E A N A L Y S I S S U M M A R Y (STURM INVERSE POWER)

NUMBER OF EIGENVALUES EXTRACTED . . . . . . 8
NUMBER OF TRIANGULAR DECOMPOSITIONS . . . . 6
TOTAL NUMBER OF VECTOR ITERATIONS . . . . . 68
REASON FOR TERMINATION: ALL EIGENVALUES FOUND IN RANGE.

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 1.685851E+02 1.298403E+01 2.066473E+00 1.000000E+00 1.685851E+02
2 3 1.685851E+02 1.298403E+01 2.066473E+00 1.000000E+00 1.685851E+02
3 2 6.474471E+03 8.046410E+01 1.280626E+01 1.000000E+00 6.474471E+03
4 5 6.474471E+03 8.046410E+01 1.280626E+01 1.000000E+00 6.474471E+03
5 4 4.974941E+04 2.230458E+02 3.549883E+01 1.000000E+00 4.974941E+04
6 6 4.974941E+04 2.230458E+02 3.549883E+01 1.000000E+00 4.974941E+04
7 7 1.870792E+05 4.325266E+02 6.883875E+01 1.000000E+00 1.870792E+05
8 8 1.870792E+05 4.325266E+02 6.883875E+01 1.000000E+00 1.870792E+05
EIGENVALUE = 1.685851E+02
CYCLES = 2.066473E+00 R E A L E I G E N V E C T O R N O . 1

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 4.198658E-20 9.109908E-03 1.052335E-02 0.0 -6.847622E-02 5.927882E-02
3 G 8.293447E-20 3.469681E-02 4.008020E-02 0.0 -1.268971E-01 1.098529E-01
4 G 1.218317E-19 7.416008E-02 8.566642E-02 0.0 -1.753660E-01 1.518116E-01
5 G 1.577188E-19 1.249444E-01 1.443303E-01 0.0 -2.141397E-01 1.853772E-01
6 G 1.897125E-19 1.845877E-01 2.132276E-01 0.0 -2.436876E-01 2.109563E-01
7 G 2.170278E-19 2.507827E-01 2.896933E-01 0.0 -2.647398E-01 2.291808E-01
8 G 2.389961E-19 3.214496E-01 3.713246E-01 0.0 -2.783232E-01 2.409396E-01
9 G 2.550813E-19 3.948168E-01 4.560753E-01 0.0 -2.857873E-01 2.474010E-01
10 G 2.648916E-19 4.695067E-01 5.423540E-01 0.0 -2.888212E-01 2.500274E-01
11 G 2.681884E-19 5.446261E-01 6.291288E-01 0.0 -2.894634E-01 2.505833E-01
EIGENVALUE = 1.685851E+02
CYCLES = 2.066473E+00 R E A L E I G E N V E C T O R N O . 2

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G 1.219015E-33 -1.052334E-02 9.109882E-03 0.0 -5.927866E-02 -6.847615E-02
3 G 2.407892E-33 -4.008016E-02 3.469673E-02 0.0 -1.098526E-01 -1.268970E-01
4 G 3.537263E-33 -8.566635E-02 7.415994E-02 0.0 -1.518114E-01 -1.753659E-01
5 G 4.579275E-33 -1.443302E-01 1.249442E-01 0.0 -1.853771E-01 -2.141396E-01
6 G 5.508284E-33 -2.132276E-01 1.845875E-01 0.0 -2.109563E-01 -2.436876E-01
7 G 6.301480E-33 -2.896932E-01 2.507825E-01 0.0 -2.291809E-01 -2.647399E-01
8 G 6.939436E-33 -3.713246E-01 3.214495E-01 0.0 -2.409399E-01 -2.783234E-01
9 G 7.406566E-33 -4.560753E-01 3.948168E-01 0.0 -2.474014E-01 -2.857874E-01
10 G 7.691474E-33 -5.423541E-01 4.695069E-01 0.0 -2.500278E-01 -2.888214E-01
11 G 7.787220E-33 -6.291289E-01 5.446264E-01 0.0 -2.505838E-01 -2.894636E-01
EIGENVALUE = 6.474471E+03
CYCLES = 1.280626E+01 R E A L E I G E N V E C T O R N O . 3

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G -1.606174E-26 6.893317E-02 3.123881E-02 0.0 -1.886937E-01 4.163812E-01
3 G -3.172614E-26 2.244652E-01 1.017221E-01 0.0 -2.624463E-01 5.791277E-01
4 G -4.660607E-26 3.933468E-01 1.782550E-01 0.0 -2.316833E-01 5.112443E-01
5 G -6.033448E-26 5.129929E-01 2.324757E-01 0.0 -1.183568E-01 2.611722E-01
6 G -7.257351E-26 5.389377E-01 2.442332E-01 0.0 4.532863E-02 -1.000244E-01
7 G -8.302282E-26 4.501138E-01 2.039804E-01 0.0 2.219953E-01 -4.898660E-01
8 G -9.142668E-26 2.499085E-01 1.132523E-01 0.0 3.764997E-01 -8.308035E-01
9 G -9.757999E-26 -3.798927E-02 -1.721591E-02 0.0 4.839685E-01 -1.067950E+00
10 G -1.013329E-25 -3.788206E-01 -1.716723E-01 0.0 5.368715E-01 -1.184688E+00
11 G -1.025940E-25 -7.400293E-01 -3.353632E-01 0.0 5.500190E-01 -1.213700E+00
EIGENVALUE = 6.474471E+03
CYCLES = 1.280626E+01 R E A L E I G E N V E C T O R N O . 4

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 0.0 0.0 0.0 0.0 0.0
2 G -6.181717E-19 3.123822E-02 -6.893279E-02 0.0 4.163791E-01 1.886905E-01
3 G -1.221121E-18 1.017205E-01 -2.244643E-01 0.0 5.791260E-01 2.624437E-01
4 G -1.794002E-18 1.782531E-01 -3.933455E-01 0.0 5.112445E-01 2.316835E-01
5 G -2.322707E-18 2.324743E-01 -5.129921E-01 0.0 2.611744E-01 1.183602E-01
6 G -2.794219E-18 2.442331E-01 -5.389376E-01 0.0 -1.000213E-01 -4.532376E-02
7 G -3.196926E-18 2.039817E-01 -4.501146E-01 0.0 -4.898637E-01 -2.219917E-01
8 G -3.520915E-18 1.132542E-01 -2.499097E-01 0.0 -8.308033E-01 -3.764996E-01
9 G -3.758207E-18 -1.721460E-02 3.798841E-02 0.0 -1.067952E+00 -4.839722E-01
10 G -3.902959E-18 -1.716725E-01 3.788207E-01 0.0 -1.184692E+00 -5.368777E-01
11 G -3.951610E-18 -3.353654E-01 7.400307E-01 0.0 -1.213704E+00 -5.500259E-01

Figure 3-10 Printed Results from the Second Model

 CHAPTER 3 67
Real Eigenvalue Analysis

This second model was rerun, changing the mass from lumped (the default) to coupled by
adding PARAM,COUPMASS,1 to the Bulk Data. The resulting frequencies are shown
below and are compared to those of the lumped mass model and the theoretical results.
Note that the frequency difference is greater at higher frequencies. For most production-
type models (i.e., complex three-dimensional structures), this difference is negligible.

Frequencies (Hz)
Lumped Mass Coupled Mass
Theory
Model Model
2.076 2.066 2.076
13.010 12.806 13.010
36.428 35.499 36.437
71.384 68.838 71.451

Bracket Model
This example is a steel bracket as shown in Figure 3-11.

Figure 3-11 Bracket Model

A concentrated mass is suspended from the center of the hole in bracket. This mass has the
following properties:

2
m = 0.0906 lb-sec /in
2
I11 = 0.35 in-lb-sec
2
I22 = 0.56 in-lb-sec
2
I33 = 0.07 in-lb-sec

The concentrated mass (grid point 999) is connected to the bracket by an RBE2 element
connecting 24 grid points, as shown in Figure 3-12.
68

Figure 3-12 Bracket Model Showing RBE2 Element (Dashed Lines)

The bracket is clamped by constraining six degrees-of-freedom for each of 12 grid points
near the base.
This example illustrates a normal modes analysis (SOL 103) using the Lanczos method
(EIGRL entry). All frequencies below 100 Hz are requested. The MODES Case Control
command is used to specify the number of times a subcase is repeated and therefore
enables different output requests for each mode. The output requests for this problem are
eigenvectors for all modes (DISPLACEMENT = ALL above the subcase level), corner
stresses for the first two modes (STRESS(CORNER) = ALL and MODES = 2 in Subcase 1),
and element strain energies for the third mode (ESE = ALL). An abridged version of the
input file is shown in Listing 3-3.
 CHAPTER 3 69
Real Eigenvalue Analysis

Listing 3-3 Abridged Input File for the Bracket Model

$ FILE bd03bkt.dat
$
$ BRACKET MODEL
$ CHAPTER 3, NORMAL MODES
$
TIME 10
SOL 103 $ NORMAL MODES ANALYSIS
CEND
$
TITLE = BRACKET MODEL
SUBTITLE = NORMAL MODES ANALYSIS
$
SPC = 1
$
$ SELECT EIGRL
METHOD = 777
$
$ OUTPUT REQUESTS
DISPLACEMENT = ALL
SUBCASE 1
MODES = 2 $ USE FOR FIRST TWO MODES
STRESS(CORNER) = ALL
SUBCASE 3
ESE = ALL
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$EIGRL SID V1 V2 ND MSGLVL MAXSET SHFSCL NORM
EIGRL 777 100.
$
$ CONCENTRATED MASS, SUPPORTED WITH AN RBE2 ELEMENT
$
GRID 999 3.0 3.0 3.8
$
$CONM2 EID G CID M X1 X2 X3 +CON1
$+CON1 I11 I21 I22 I31 I32 I33
CONM2 999 999 0.0906 +CM1
+CM1 0.35 0.56 0.07
$
RBE2 999 999 123456 126 127 91 81 80 +RB1
+RB1 90 95 129 128 96 86 85 228 +RB2
+RB2 229 199 189 190 200 195 227 226 +RB3
+RB3 194 184 185
$
$ STEEL, M A S S DENSITY FOR RHO
MAT1 1 3.+7 1.153+7 7.76-4
$
... basic model ...
$
ENDDATA

Figure 3-13 shows an abridged version of the resulting MSC.Nastran output. The circular
total element strain energy (ESE) for each mode of the entire model is ESE = ω 2i ⁄ 2 for the i-
th mode when φ Ti Mφ i = 1 . The frequency of the third mode is 234.49 radians per second
squared; therefore, the total strain energy of the model is ESEtotal = ( 234.49 ) 2 ⁄ 2 = 27493 for
the third mode. The printed ESE total = 27492 . The eigenvectors are printed for each mode,
70

the element corner stresses are printed for the first and second modes, and the element
strain energies are printed for the third mode. (Only the headers are shown in the figure
in order to save space.)

E I G E N V A L U E A N A L Y S I S S U M M A R Y (LANCZOS ITERATION)
BLOCK SIZE USED ...................... 6
NUMBER OF DECOMPOSITIONS ............. 2
NUMBER OF ROOTS FOUND ................ 3
NUMBER OF SOLVES REQUIRED ............ 3
TERMINATION MESSAGE : REQUIRED NUMBER OF EIGENVALUES FOUND.

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 3.930304E+03 6.269214E+01 9.977763E+00 1.000000E+00 3.930304E+03
2 2 2.878402E+04 1.696586E+02 2.700200E+01 1.000000E+00 2.878402E+04
3 3 5.498442E+04 2.344876E+02 3.731985E+01 1.000000E+00 5.498442E+04

EIGENVALUE = 3.930304E+03
CYCLES = 9.977763E+00 R E A L E I G E N V E C T O R N O . 1

EIGENVALUE = 2.878402E+04
CYCLES = 2.700200E+01 R E A L E I G E N V E C T O R N O . 2

EIGENVALUE = 5.498442E+04
CYCLES = 3.731985E+01 R E A L E I G E N V E C T O R N O . 3

EIGENVALUE = 3.930304E+03
S T R E S S E S I N Q U A D R I L A T E R A L E L E M E N T S ( Q U A D 4 ) OPTION = CUBIC

EIGENVALUE = 2.878402E+04
S T R E S S E S I N Q U A D R I L A T E R A L E L E M E N T S ( Q U A D 4 ) OPTION = CUBIC

E L E M E N T S T R A I N E N E R G I E S

ELEMENT-TYPE = QUAD4 * TOTAL ENERGY OF ALL ELEMENTS IN PROBLEM = 2.749221E+04

MODE 3 TOTAL ENERGY OF ALL ELEMENTS IN SET -1 = 2.749221E+04
*
TYPE = QUAD4 SUBTOTAL 2.749221E+04 100.0000

Figure 3-13 Abridged Output from the Bracket Model

The deformed shape resulting from the first mode is illustrated in Figure 3-14 and is
overlaid on the undeformed shape. Figure 3-15 illustrates the stress contours plotted on
the deformed shape of the second mode. The element strain energy contour plot for the
third mode is shown in Figure 3-16.
 CHAPTER 3 71
Real Eigenvalue Analysis

Figure 3-14 Deformed Shape of the First Mode

Figure 3-15 Second Mode Deformation with Element Stress Contours

72

Figure 3-16 Element Strain Energy Contours for the Third Mode

Car Frame Model

Figure 3-17 shows a model of an aluminum car frame. The frame model is comprised of
plate elements (CQUAD4 and CTRIA3), with springs (CELAS2) representing the
suspension. Spring stiffnesses are input in the three translational directions; a stiffness of
500 lb/in is used in the vertical direction (T2), and stiffnesses of 1000 lb/in are used in the
other translational directions (T1 and T3). When using CELASi elements to connect two
grid points, it is recommended that the coordinates of the two grid points be identical in
order to represent coaxial springs (noncoincident coordinates can lead to errors). The goal
of the analysis is to compute resonant frequencies up to 50 Hz using the Lanczos method.
Element strain energies are computed for the springs in order to help characterize the
resulting modes.
Listing 3-4 shows the input file. Modal displacements are written to the plot file and are
not printed. The rigid body mass matrix (see “Parameters for Dynamic Analysis” on
page 433) is computed via the PARAM,GRDPNT,0 entry. "Include" files are used to
partition the input file into several smaller files. The INCLUDE statement inserts an
external file into the input file. The basic file is bd03car.dat. The springs are contained in
file bd03cars.dat (Listing 3-5), and the rest of the input file is contained in file bd03carb.dat
(not shown).
 CHAPTER 3 73
Real Eigenvalue Analysis

Figure 3-17 Car Frame Model

Listing 3-4 Basic Input File for the Car Model

$ FILE bd03car.dat
$
$ CAR FRAME MODEL
$ CHAPTER 3, NORMAL MODES
$
$ MODEL COURTESY LAPCAD ENGINEERING
$ CHULA VISTA, CALIFORNIA
$
SOL 103 $ NORMAL MODES ANALYSIS
TIME 30
CEND
$
TITLE = CAR MODEL WITH SUSPENSION SPRINGS
SUBTITLE = MODAL ANALYSIS CASE CONTROL
ECHO = UNSORT
$
METHOD = 1
$
DISPLACEMENT(PLOT) = ALL
SET 99 = 1001,1002,1003,1011,1012,1013,
1021,1022,1023,1031,1032,1033
ESE = 99
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
INCLUDE 'bd03cars.dat' $ Car springs
$
$EIGRL SID V1 V2 ND MSGLVL MAXSET SHFSCL NORM
EIGRL 1 -1. 50.
$
$MAT1 MID E G NU RHO
MAT1 1 1.0E7 0.33 2.59E-4
$
$ PRINT RIGID-BODY MASS
PARAM,GRDPNT,0
$
ECHOOFF
INCLUDE 'bd03carb.dat' $ Rest of Bulk Data
$
ENDDATA
74

Listing 3-5 Input File for the Springs

$ FILE bd03cars.dat
$
$ SPRINGS CONNECTED TO GROUND
$ GRIDS 1059,1562,1428,1895 HAVE THE SAME COORDS. AS 59,562,428,895
$
GRID 1059 152.012 32.7964 -2.90000 123456
GRID 1562 152.012 32.7964 -67.1000 123456
GRID 1428 35.6119 30.8257 -0.66667 123456
GRID 1895 35.6119 30.8257 -69.3333 123456
$
CELAS2 1001 1000. 59 1 1059 1
CELAS2 1002 500. 59 2 1059 2
CELAS2 1003 1000. 59 3 1059 3
CELAS2 1011 1000. 562 1 1562 1
CELAS2 1012 500. 562 2 1562 2
CELAS2 1013 1000. 562 3 1562 3
CELAS2 1021 1000. 428 1 1428 1
CELAS2 1022 500. 428 2 1428 2
CELAS2 1023 1000. 428 3 1428 3
CELAS2 1031 1000. 895 1 1895 1
CELAS2 1032 500. 895 2 1895 2
CELAS2 1033 1000. 895 3 1895 3

Figure 3-18 shows the grid point weight generator output. The grid point weight
generator indicates that the translational mass is 0.8027 lb-sec 2 ⁄ in. . This type of information
is useful in verifying your model. Figure 3-19 shows some of the rest of the output. The
eigenvalue analysis summary indicates that there are 12 modes below 50 Hz. The large
element strain energies in the first six modes indicate that these are primarily suspension
modes, comprised of the car frame acting as a rigid body supported by the flexible springs.
Approximately 95% of the total strain energy is represented by spring deformation as
opposed to frame deformation. Modes 7 and above show insignificant spring strain
energy, indicating that these are primarily frame modes.

O U T P U T F R O M G R I D P O I N T W E I G H T G E N E R A T O R
REFERENCE POINT = 0
M O
* 8.027376E-01 0.000000E+00 0.000000E+00 0.000000E+00 -2.808118E+01 -2.179473E+01 *
* 0.000000E+00 8.027376E-01 0.000000E+00 2.808118E+01 0.000000E+00 7.610537E+01 *
* 0.000000E+00 0.000000E+00 8.027376E-01 2.179473E+01 -7.610537E+01 0.000000E+00 *
* 0.000000E+00 2.808118E+01 2.179473E+01 2.325008E+03 -2.153940E+03 2.662697E+03 *
* -2.808118E+01 0.000000E+00 -7.610537E+01 -2.153940E+03 1.020870E+04 7.626398E+02 *
* -2.179473E+01 7.610537E+01 0.000000E+00 2.662697E+03 7.626398E+02 9.437676E+03 *
S
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
DIRECTION
MASS AXIS SYSTEM (S) MASS X-C.G. Y-C.G. Z-C.G.
X 8.027376E-01 0.000000E+00 2.715050E+01 -3.498177E+01
Y 8.027376E-01 9.480728E+01 0.000000E+00 -3.498177E+01
Z 8.027376E-01 9.480728E+01 2.715050E+01 0.000000E+00
I(S)
* 7.509408E+02 8.764180E+01 -3.962963E-01 *
* 8.764180E+01 2.011031E+03 -2.217236E-01 *
* -3.962963E-01 -2.217236E-01 1.630595E+03 *
I(Q)
* 2.017097E+03 *
* 7.448741E+02 *
* 1.630595E+03 *
Q
* 6.905332E-02 9.976128E-01 5.012719E-04 *
* -9.976128E-01 6.905355E-02 -4.555999E-04 *
* -4.891269E-04 -4.686146E-04 9.999998E-01 *

Figure 3-18 Output from the Grid Point Weight Generator

 CHAPTER 3 75
Real Eigenvalue Analysis

E I G E N V A L U E A N A L Y S I S S U M M A R Y (LANCZOS ITERATION)
BLOCK SIZE USED ...................... 7
NUMBER OF DECOMPOSITIONS ............. 3
NUMBER OF ROOTS FOUND ................ 12
NUMBER OF SOLVES REQUIRED ............ 11
TERMINATION MESSAGE : REQUIRED NUMBER OF EIGENVALUES FOUND.

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 2.346479E+03 4.844047E+01 7.709540E+00 1.000000E+00 2.346479E+03
2 2 2.654886E+03 5.152559E+01 8.200551E+00 1.000000E+00 2.654886E+03
3 3 3.769821E+03 6.139887E+01 9.771934E+00 1.000000E+00 3.769821E+03
4 4 4.633242E+03 6.806792E+01 1.083335E+01 1.000000E+00 4.633242E+03
5 5 5.078395E+03 7.126286E+01 1.134184E+01 1.000000E+00 5.078395E+03
6 6 8.485758E+03 9.211817E+01 1.466106E+01 1.000000E+00 8.485758E+03
7 7 2.805541E+04 1.674975E+02 2.665805E+01 1.000000E+00 2.805541E+04
8 8 5.350976E+04 2.313218E+02 3.681600E+01 1.000000E+00 5.350976E+04
9 9 5.940912E+04 2.437399E+02 3.879240E+01 1.000000E+00 5.940912E+04
10 10 8.476198E+04 2.911391E+02 4.633622E+01 1.000000E+00 8.476198E+04
11 11 9.134271E+04 3.022296E+02 4.810133E+01 1.000000E+00 9.134271E+04
12 12 9.726959E+04 3.118807E+02 4.963736E+01 1.000000E+00 9.726959E+04

E L E M E N T S T R A I N E N E R G I E S

ELEMENT-TYPE = ELAS2 * TOTAL ENERGY OF ALL ELEMENTS IN PROBLEM = 1.173240E+03

MODE 1 TOTAL ENERGY OF ALL ELEMENTS IN SET 99 = 1.104569E+03
*
ELEMENT-ID STRAIN-ENERGY PERCENT OF TOTAL STRAIN-ENERGY-DENSITY
1002 2.735009E+02 23.3116
1003 4.059090E-02 .0035
1012 2.685884E+02 22.8929
1013 3.240471E-02 .0028
1021 8.017746E-02 .0068
1022 2.833448E+02 24.1506
1023 3.963123E-01 .0338
1031 7.903841E-02 .0067
1032 2.781467E+02 23.7076
1033 3.573737E-01 .0305

TYPE = ELAS2 SUBTOTAL 1.104569E+03 94.1469

E L E M E N T S T R A I N E N E R G I E S

ELEMENT-TYPE = ELAS2 * TOTAL ENERGY OF ALL ELEMENTS IN PROBLEM = 1.402770E+04

MODE 7 TOTAL ENERGY OF ALL ELEMENTS IN SET 99 = 3.172818E+03
*
ELEMENT-ID STRAIN-ENERGY PERCENT OF TOTAL STRAIN-ENERGY-DENSITY
1001 1.278503E+02 .9114
1002 5.884620E+02 4.1950
1003 8.020268E+01 .5717
1011 1.278420E+02 .9114
1012 5.883779E+02 4.1944
1013 8.022697E+01 .5719
1021 7.525866E+01 .5365
1022 4.885996E+02 3.4831
1023 2.255234E+02 1.6077
1031 7.528390E+01 .5367
1032 4.896509E+02 3.4906
1033 2.255398E+02 1.6078

TYPE = ELAS2 SUBTOTAL 3.172818E+03 22.6182

Figure 3-19 Abridged Output from the Car Model

Mode shapes for modes 7, 8, 9, and 10 are shown in Figure 3-20. Mode 7 is an overall
twisting mode; mode 8 is a "roof collapse" mode; mode 9 is a local (front) roof mode; and
mode 10 is a local rear mode. Plots such as these, in conjunction with element strain
energies, help to illustrate each of the mode shapes.
76

Figure 3-20 Mode Shapes for Modes 7, 8, 9, and 10

 CHAPTER 3 77
Real Eigenvalue Analysis

Test Fixture Model

This example is an aluminum test fixture, which is shown in Figure 3-21. The model is
comprised of 8157 grid points, 5070 CHEXA elements, and 122 CPENTA elements. The
primary plates are 1 inch thick, and the gusset plates are 0.5 inch thick. The base of the
fixture is constrained to have no vertical (y) motion, and the bolt holes at the base are
constrained to also have no horizontal (x and z) motion.

Figure 3-21 Test Fixture Model

A portion of the input file is shown in Listing 3-6. The Lanczos method is used to compute
the modes. The first six modes are requested (ND is 6, with V1 and V2 blank). The
GRDSET Bulk Data entry removes the rotational DOFs (456) from the analysis since the
solid elements have no rotational stiffness. The Bulk Data is in free format.
78

Listing 3-6 Abridged Input File for Test Fixture Model

$ FILE bd03fix.dat
$
$ TEST FIXTURE
$ CHAPTER 3, NORMAL MODES
$
TIME 240
SOL 103
CEND
ECHO = NONE
DISPLACEMENT(PLOT) = ALL
SPC = 1
$
METHOD = 1
$
BEGIN BULK
$
$GRDSET, ,CP, , , , CD, PS
GRDSET , , , , , , , 456
$
$EIGRL, SID, V1, V2, ND, MSGLVL, MAXSET, SHFSCL, NORM
EIGRL , 1, , , 6
$
$MAT1, MID, E, G, NU, RHO
MAT1 , 3, 1.00E7, , 0.334, 2.5383-4
$
... basic model ...
$
ENDDATA

Figure 3-22 shows the first four mode shapes. The first mode is a bending mode, the
second and third modes are twist modes, and the fourth mode is a bending mode.
 CHAPTER 3 79
Real Eigenvalue Analysis

Figure 3-22 Test Fixture Mode Shapes

Quarter Plate Model

This example is a quarter model of a simply supported flat plate, shown in Figure 3-23.
This example illustrates the use of multiple boundary conditions (new in Version 68) for
modeling symmetric structures. In this case the plate is doubly symmetric.
80

Lines of Symmetry

ss ss

ss ss
ss ss

ss ss
Full Model
1
z
y ss
2
x
ss
Quarter Model
Figure 3-23 Derivation of Quarter Plate Model

A portion of the input file is shown in Listing 3-7. Four subcases are used—one for each
of the following sets of boundary conditions for the quarter model:

• Symmetric-antisymmetric
• Antisymmetric-symmetric
• Symmetric-symmetric
• Antisymmetric-antisymmetric
The BC Case Control command identifies multiple boundary conditions. The SPCADD
Bulk Data entry defines a union of SPC sets.
 CHAPTER 3 81
Real Eigenvalue Analysis

Listing 3-7 Input File (Abridged) for the Quarter Plate Model

$ FILE bd03plt1.dat
$
$ QUARTER PLATE MODEL
$ CHAPTER 3, NORMAL MODES
$
SOL 103 $ NORMAL MODES ANALYSIS
TIME 10
CEND
TITLE = SIMPLY SUPPORTED PLATE USING SYMMETRY
SUBTITLE = NORMAL MODES CASE CONTROL
LABEL = QUARTER PLATE MODEL
$
DISPLACEMENT = ALL
$
SUBCASE 1
LABEL = SYM-ASYM
BC = 1
METHOD = 1
SPC = 101
SUBCASE 2
LABEL = ASYM-SYM
BC = 2
SPC = 102
METHOD = 1
SUBCASE 3
LABEL = SYM-SYM
BC = 3
SPC = 103
METHOD = 1
SUBCASE 4
LABEL = ASYM-ASYM
BC = 4
SPC = 104
METHOD = 1
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10......
$
$ SYM-ASYM
SPCADD 101 11 1 4
$ ASYM-SYM
SPCADD 102 11 2 3
$ SYM-SYM
SPCADD 103 11 1 3
$ ASYM-ASYM
SPCADD 104 11 2 4
$
$EIGRL SID V1 V2
EIGRL 1 -0.1 100.
$
SPC 1 1 246 0.00
SPC 1 2 246 0.00
... etc. ...
SPC 11 80 12356 0.00
SPC 11 81 123456 0.00
$
... basic model ...
$
ENDDATA
82

Figure 3-24 shows the quarter plate mode shapes and the corresponding mode shapes for
a full model of the same structure. Note that the quarter plate modes match the full plate
modes.

Figure 3-24 Mode Shapes for the Quarter Model (Left) and Full Model (Right)
 CHAPTER 3 83
Real Eigenvalue Analysis

DMIG Example
This example illustrates the use of a DMIG entry to input external mass and stiffness. The
cantilever beam model shown in Figure 3-25 is used for this purpose. The model consists
of CBAR elements 1 through 4. Element 5 is a model from another subcontractor that is
input via DMIG entries. The model contains two DOFs (R2 and T3) per grid point.

X
1 2
2
3 4 5
5
6
1 3 4
= Grid Point ID
L = Element ID

Figure 3-25 Planar Cantilever Beam

The stiffness and mass matrices from the contractor for element 5 are as follows:

5, 3 5, 5 6, 3 6, 5
5, 3  500038.8 
 
K5 = 5, 5  – 250019.4 166679.6 
6, 3  – 500038.8 250019.4 500038.8 
 
6, 5  – 250019.4 83339.8 250019.4 166679.6 

5, 3 5, 5 6, 3 6, 5
5, 3  3.5829 0. 0. 0. 
 
M5 = 5, 5  0. 0. 0. 0. 
6, 3  0. 0. 3.5829 0. 

6, 5  0. 0. 0. 0. 

Since the matrices are symmetric, only the lower or upper triangular portion of the
matrices need to be provided via the DMIG entries.

The corresponding input file is shown in Listing 3-8.

84

Listing 3-8 Input File for the DMIG Example

$ FILE bd03dmi.dat
$
$ DMIG EXAMPLE
$ CHAPTER 3, NORMAL MODES
$
SOL 103 $ NORMAL MODES ANALYSIS
TIME 10
CEND
TITLE = DMIG TO READ STIFFNESS AND MASS FOR ELEM 5
SUBTITLE = PLANAR PROBLEM
$
SPC = 10
$
$ SPECIFY K2GG AND M2GG
K2GG = EXSTIF
M2GG = EXMASS
$
METHOD = 10
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$EIGRL SID V1 V2 ND
EIGRL 10 2
$
CBAR 1 1 1 2 10
CBAR 2 1 2 3 10
CBAR 3 1 3 4 10
CBAR 4 1 4 5 10
$
$ HEADER ENTRY FOR STIFFNESS
DMIG EXSTIF 0 6 1
$
DMIG EXSTIF 5 3 5 3 500039. +000001
++0000015 5 -250019. 6 3 -500039. +000002
++0000026 5 -250019. +000003
$
DMIG EXSTIF 5 5 5 5 166680. +000004
++0000046 3 250019. 6 5 83340. +000005
$
DMIG EXSTIF 6 3 6 3 500039. +000006
++0000066 5 250019. +000007
$
DMIG EXSTIF 6 5 6 5 166680.
$
$ HEADER ENTRY FOR MASS
DMIG EXMASS 0 6 1
$
$ DATA ENTRIES FOR MASS
$
DMIG EXMASS 5 3 5 3 3.5829
DMIG EXMASS 6 3 6 3 3.5829
$
GRID 1 0. 0. 0. 1246
GRID 2 1. 0. 0. 1246
GRID 3 2. 0. 0. 1246
GRID 4 3. 0. 0. 1246
GRID 5 4. 0. 0. 1246
GRID 6 5. 0. 0. 1246
GRID 10 0. 0. 10. 123456
MAT1 1 7.1+10 .33 2700.
PBAR 1 1 2.654-3 5.869-7
SPC1 10 123456 1
$
ENDDATA
 CHAPTER 3 85
Real Eigenvalue Analysis

In this example, EXSTIF is chosen as the name of the input stiffness matrix. Therefore, in
order to bring in this stiffness matrix and add it to the global stiffness matrix, the Case
Control command K2GG = EXSTIF is required. Note that these stiffness terms are
additions to any existing terms in the global stiffness matrix at that location and not
replacements of the stiffness terms at that location. In the Bulk Data Section, five DMIG
entries are required—one for the header entry and four for the data column entries since
there are four non-null columns in the above matrix.
For the header entry, the same name EXSTIF must be used to match the name called out in
the Case Control Section. The third field is "0", which must be the value used for the header
entry. The fourth field (IFO) is set to "6" to denote a symmetric matrix input. The fifth field
(TIN) is set to "1" to denote that the matrix is real, single precision.
The terms in the matrix are referenced in terms of their external grid IDs when using the
DMIG entries. Physically, each term in a particular column of the stiffness matrix ( K ij )
represents the induced reactive load in the i-th degree-of-freedom due to a unit
displacement in the j-th direction with all other displacement degrees-of-freedom held to
zero. Since the matrix is symmetric, only the lower triangular portion of the matrix is input.
The first DMIG data column entry defines the first column of the above matrix. Field 2 of
this DMIG entry must have the same name EXSTIF as referenced by the Case Control
K2GG = EXSTIF command. Fields 3 and 4 of this entry identify this column in terms of its
external grid ID and corresponding degree-of-freedom, respectively. In this case, it is grid
point5, degree-of-freedom 3 (z-translation at grid point 5). Once this column is defined,
follow the format description as described in the section for column data entry format, and
then you can input the four terms in this column row-by-row. These four terms are defined
by sets of three fields. They are the external grid ID number, corresponding degree-of-
freedom, and the actual matrix term, respectively. The first term of column one is defined
by external grid ID 5, degree-of-freedom 3 (z-translation at grid point 5) with a stiffness
value of 500039. The second term of column one is defined by external grid ID 5, degree-
of-freedom 5 (y-rotation at grid point 5) with a stiffness value of -250019. The third term of
column one is defined by external grid ID 6, degree-of-freedom 3 (z-translation at grid
point 6) with a stiffness value of -500039. The fourth term of column one is defined by
external grid ID 6, degree-of-freedom 5 (y-rotation at grid point 6) with a stiffness value of
-250019.

The next DMIG entry defines the second column of the above matrix. Fields 3 and 4 of this
entry identify this column in terms of its external grid ID and corresponding degree-of-
freedom, respectively. In this case, it is grid point 5, degree-of-freedom 5 (y-rotation at grid
point 5). The rest of the procedure is similar to that of column one with the exception that
only three terms need to be input due to symmetry.
The next two DMIG entries defines columns three and four of the stiffness matrix,
respectively. Note that due to symmetry, one less row needs to be defined for each
additional column.
86

The mass matrix is input in a similar manner as the stiffness matrix with the following
exceptions:
• The command M2GG = EXMASS instead of K2GG = EXSTIF is used in the Case
Control Section. In this case, EXMASS is the name of the mass matrix referenced
in field two of the DMIG Bulk Data entries.
• The matrix defined in the DMIG entries is expressed in the mass matrix terms
rather than in stiffness matrix terms.
• Since there are only two non-null columns for the mass matrix, only two DMIG
data entries are required instead of the four entries needed for the stiffness matrix.
Mass matrices input using DMIG are not scaled by PARAM,WTMASS.

In this example the small-field input format is used, and the maximum number of
characters that can be input are eight (including sign and decimal point). Greater input
precision can be achieved by using the large-field format and by changing the TIN field to
2 for the DMIG entries.
The first two computed natural frequencies for this example are 1.676 Hz and 10.066 Hz.
 NX Nastran Basic Dynamic Analysis User’s Guide

4
CHAPTER
Rigid-body Modes

■ Overview
■ SUPORT Entry

■ Examples
88

4.1 Overview
A structure or a portion of a structure can displace without developing internal loads or
stresses if it is not sufficiently tied to ground (constrained). These stress-free displacements
are categorized as rigid-body modes or mechanism modes.

Rigid-body Modes. Rigid-body modes occur in unconstrained structures, such as

satellites and aircraft in flight. For a general, unconstrained 3-D structure without
mechanisms, there are six rigid-body modes often described as T1, T2, T3, R1, R2, and R3,
or combinations thereof. Rigid-body modes can also be approximated for certain kinds of
dynamic or modal tests in which the test specimen is supported by very flexible supports,
such as bungee cords or inflatable bags. In this case the test specimen itself does not distort
for the lowest mode(s) but instead displaces as a rigid body with all of the deformation
occurring in the flexible support. Rigid-body modes can improperly occur if a structure
that should be constrained is not fully constrained (for example, in a building model for
which the boundary conditions (SPCs) were forgotten).

Mechanism Modes. A mechanism mode occurs when a portion of the structure can
displace as a rigid body, which can occur when there is an internal hinge in the structure.
An example of a mechanism is a ball and socket joint or a rudder in an airplane. A
mechanism mode can also occur when two parts of a structure are improperly joined. A
common modeling error resulting in a mechanism is when a bar is cantilevered from a
solid element; the bar has rotational stiffness and the solid has no rotational stiffness,
resulting in a pinned connection when the two are joined.
The presence of rigid-body and/or mechanism modes is indicated by zero frequency
eigenvalues. Due to computer roundoff, the zero frequency eigenvalues are numerical
zeroes on the order of 1.0E-4 Hz or less for typical structures. The same unconstrained
model may give different values of the rigid-body frequencies when run on different
computer types.

Rigid-body modes generated by MSC.Nastran are orthogonal with respect to each other
and with respect to the flexible (or elastic) modes. Each rigid-body mode has the following
property:

T
{ φ rig } [ M ] { φ rig } > 0
Eq. 4-1
T
{ φ rig } [ K ] { φ rig } = 0

where { φ rig } denotes the rigid-body mode shapes, [ M ] denotes the mass matrix, and [ K ]
denotes the stiffness matrix. The rigid-body modes contain no strain energy.

Figure 4-1 depicts the rigid-body modes for a two-dimensional model of a straight beam.
Because the two-dimensional model has three DOFs per grid point (x-translation,
y-translation, and z-rotation), there are three rigid-body modes. The figure shows classical
rigid-body modes, whereby one mode is purely x-translation, another is purely
 CHAPTER 4 89
Rigid-body Modes

y-translation, and another is purely z-rotation about the center of the beam. Because
rigid-body modes are a special case of repeated roots, any linear combination of these
displacement shapes also comprises a valid set of rigid-body modes.

Figure 4-1 Rigid-Body Modes of a 2-D Beam

90

4.2 SUPORT Entry

Rigid-body modes are computed in MSC.Nastran without requiring special user
intervention, although the use of a SUPORT Bulk Data entry makes the mode shapes look
cleaner. The SUPORT entry also makes the rigid-body mode shapes repeatable when the
mass or stiffness of the model changes and the model is reanalyzed. The SUPORT (note
spelling!) entry does not constrain the model; it simply provides a frame of reference for
the rigid-body shape calculations by defining the r-set components of motion. The
SUPORT entry is not required for any of the dynamic analysis methods except for response
spectrum analysis (see “Advanced Dynamic Analysis Capabilities” on page 293).

If the SUPORT is used, one DOF should be specified for each rigid-body mode or
mechanism. The form of the SUPORT entry is shown below:

1 2 3 4 5 6 7 8 9 10
SUPORT ID1 C1 ID2 C2 ID3 C3 ID4 C4

Field Contents
IDi Grid or scalar point identification number.
Ci Component numbers. (0 or blank for scalar points and any unique
combination of the integers 1 through 6 for grid points).

Treatment of SUPORT by Eigenvalue Analysis Methods

The eigenvalue extraction methods treat the SUPORT entry differently as described below.

EIGR Methods
Each of the eigenvalue extraction methods selected on the EIGR Bulk Data entry (AGIV,
AHOU, GIV, HOU, INV, MGIV, MHOU, and SINV) treats the SUPORT in the same
manner. Eigenvalues are first computed using the information on the EIGR Bulk Data
entry. The frequencies of the first N r modes (where N r is the number of SUPORT DOFs)
are replaced with a value of 0.0 Hz. The first N r eigenvectors are replaced by modes that
are calculated by moving each SUPORT DOF a unit distance and then mass
orthogonalizing them with respect to the other modes. The fact that the eigenvectors are
calculated via kinematics is the reason that the SUPORT entry produces cleaner rigid-body
modes (as opposed to the rigid-body modes computed without the use of the SUPORT
entry).
Note that MSC.Nastran has no built-in checks to warn if any of the N r eigenvalues are not
rigid-body modes. This replacement can hide potential modeling problems. The results
from UIM 3035 should be carefully checked (see “Examples” on page 96). The strain
energy term for each SUPORT DOF should be zero.
 CHAPTER 4 91
Rigid-body Modes

A poor choice of DOFs on the SUPORT entry can adversely affect the computation of the
flexible modes for the INV and SINV methods. Flexible modes computed with the other
methods (AGIV, AHOU, GIV, HOU, MGIV, and MHOU) are not adversely affected by a
poor choice of SUPORT DOFs. Again, the results of UIM 3035 should be carefully checked
to ensure that a proper choice of SUPORT DOFs is made.

Lanczos Method
In the Lanczos method, the SUPORT entry attempts to give 0.0 Hz modes. The Lanczos
routine compares the values of the calculated modes (without the SUPORT entry) and
determines if the calculated frequencies for the N r modes are near 0.0 Hz. If the computed
frequencies are near 0.0 Hz, the computed values are replaced with 0.0 Hz. If a computed
frequency is not near 0.0 Hz, then it is retained. (Note that this may cause problems in
response spectrum analysis where 0.0 Hz is required for the SUPORT frequencies.) The
Lanczos-computed, rigid-body eigenvectors are used, i.e., the rigid-body eigenvectors
computed by the SUPORT entry are not used.

In the above paragraph, “near” means that the eigenvalues are less than 0.01 times the shift
scale (the SHFSCL field on the EIGRL entry). All computed eigenvalues less than this
threshold are candidate rigid-body modes. There is some logic to determine the
relationship of these candidate rigid-body modes to the number of DOFs on the SUPORT
entry. Suppose that there are three eigenvalues less than the threshold of 0.01 times the
shift scale. If your SUPORT entry defines two DOFs, then the first two frequencies become
0.0 Hz, and the third becomes something that is nonzero but small (on the order of 1.0E-6,
for example). On the other hand, if your SUPORT entry defines four DOFs, then only the
first three are treated as rigid-body modes, and the fourth is not replaced. Furthermore,
the use of a SUPORT entry forces an extra decomposition, which increases the computer
run time.
Because the SUPORT entry is not used to compute the rigid-body eigenvectors, there is
probably little to be gained by using the SUPORT entry with the Lanczos method unless
response spectrum analysis is being performed.

Theoretical Considerations
Degrees-of-freedom defined on the SUPORT entry are placed in the r-set. When an r-set is
present, static rigid-body vectors are calculated in MSC.Nastran by first partitioning the
a-set into the r- and l-sets

 ul 
{ ua } =   Eq. 4-2
 ur 

Introducing this partitioning in the stiffness matrix results in

K l l K lr  u l  0 
  = P  Eq. 4-3
K rl K rr  u r   r
92

for the rigid-body modes defined in the r-set.

There is no load on l-set DOFs. The load { Pr } on the r-set is not needed in subsequent
equations. Then solve for u l in terms of u r

{ ul } = [ D ] { ur } Eq. 4-4

–1
where: [ D ] = – [ K ll ] [ K lr ]

The matrix [ D ] is used to construct a set of rigid-body vectors [ ψ rig ]

D
[ ψ rig ] = Eq. 4-5
Ir

where [ ψ rig ] represents the motion of the a-set for a unit motion of each SUPORT DOF with
all other SUPORT DOFs constrained and [ Ir ] is an r × r identity matrix.
The rigid-body vectors can be used to create a rigid-body mass matrix [ M r ]

T D T D
[ M r ] = [ ψ rig ] [ M aa ] [ ψ ri g ] = [ M aa ] Eq. 4-6
Ir Ir

To improve the quality of the rigid-body mode shapes, orthogonalization is applied to

create a diagonal mass matrix [ M o ] by

T
[ M o ] = [ φ ro ] [ M r ] = [ φ ro ] Eq. 4-7

where: [ φ ro ] is a transformation matrix.

This transformation matrix is used to construct the final set of rigid-body mode shape
vectors by

Dφ ro
[ φ rig ] = [ ψ rig ] [ φ ro ] = Eq. 4-8
φ ro

such that

T
[ φ rig ] [ K aa ] [ φ rig ] ≈ 0 Eq. 4-9

T
[ φ rig ] [ M aa ] [ φ rig ] = Mo Eq. 4-10

where: M o is a diagonal matrix.

Care must be taken when selecting SUPORT DOFs. Each SUPORT DOF must be able to
displace independently without developing internal stresses. In other words, the SUPORT
DOFs must be statically determinate. The SUPORT is used only to facilitate the calculation
of rigid-body vectors. If you do not specify the r-set DOFs, the rigid-body modes are
 CHAPTER 4 93
Rigid-body Modes

calculated directly by the method selected for the flexible frequency modes. If an
insufficient number of r-set DOFs are specified, the calculation of rigid-body modes is
unreliable.

As a modeling aid, MSC.Nastran calculates equivalent internal strain energy (work) for
each rigid-body vector as follows:

K ll K lr D
[X] = D
T
Ir Eq. 4-11
K rl K rr Ir

which can be simplified as

T
[ X ] = [ D ] [ K l l ] [ D ] + [ K rr ] Eq. 4-12

When r-set DOFs exist, the printed strain energies are the diagonal elements of [ X ] divided
by 2 and should be approximately zero.
Note that [ X ] is the transformation of the stiffness matrix [ K aa ] to r-set coordinates, which
by definition of rigid-body (i.e., zero frequency) vector properties should be null. If this is
not the case, the equilibrium may be violated by the r-set choice or other modeling errors
may exist. The matrix [ X ] is also called the rigid-body check matrix.

MSC.Nastran also calculates a rigid-body error ratio

T
[ K rr ] + [ K lr ] [ D ]
ε = ------------------------------------------------------- Eq. 4-13
K rr

2
where: = Euclidean norm of the matrix = ∑ ∑ x ij
i j

One value of ε is calculated using Eq. 4-13 based on all SUPORT DOFs. Therefore, in
UIM 3035 the same ε is printed for every supported DOF.
The rigid-body error ratio and the strain energy should be zero if a set of statically
determinate SUPORT DOFs is chosen. Roundoff error may lead to computational zero
values for these quantities. ("Computational zero" is a small number ( 10 –5 , for example)
that normally is 0.0 except for numerical roundoff.) The rigid-body error ratio and strain
energy may be significantly nonzero for any of the following reasons:
• Roundoff error accumulation.
• The u r set is overdetermined, leading to redundant supports. The condition gives
high strain energy.
• The ur set is underspecified, leading to a singular reduced-stiffness matrix and a
MAXRATIO error. This condition gives a high rigid-body error ratio.
• The multipoint constraints are statically indeterminate. This condition gives high
strain energy and a high rigid-body error ratio.
94

• There are too many single-point constraints. This condition gives high strain
energy and a high rigid-body error ratio.
• is null. This condition gives a unit value for the rigid-body error but low
[ K rr ]
strain energy. This is an acceptable condition and may occur when generalized
dynamic reduction is used (see “Advanced Dynamic Analysis Capabilities” on
page 293).

Modeling Considerations
When using a SUPORT you must select a set of DOFs that is capable of constraining all the
rigid-body modes. Another way to state this requirement is that the r-set must be able to
constrain the structure in a statically determinate manner. There are usually many choices
of DOFs that satisfy this requirement. Two choices that work for simple three-dimensional
structures are:
• Six DOFs on one grid point when all its degrees-of-freedom have stiffness.
• Three translation DOFs normal to one plane, two translation DOFs normal to an
orthogonal plane, and then one translational DOF normal to the last orthogonal
plane. Such a system can be used, for instance, on a model composed entirely of
solid elements that have no inherent stiffness for grid point rotation. See
Figure 4-2.

Plane 1

2
4
Plane 3
Plane 2 5 6
1, 2, 3 Normal to Plane 1
4, 5 Normal to Plane 2
6 Normal to Plane 3
Figure 4-2 Statically Determinate r-set

There are special cases where a model need not have six rigid-body modes. A planar
model has only three rigid-body modes, while an airplane with a free rudder has seven, for
example. If you use the SUPORT, it is your responsibility to determine all the modes of
rigid-body motion, then provide r-set DOFs that define these rigid-body modes. Another
special case is the application of enforced motion by the large mass technique (see
“Enforced Motion” on page 199). If the input points describe redundant load paths,
diagnostics are produced that indicate overconstraint. For this case, these diagnostics may
be safely ignored.
 CHAPTER 4 95
Rigid-body Modes

Poorly-constrained rigid-body modes result from either constraining DOFs with relatively
little stiffness or from constraining a set of DOFs that are almost linearly dependent on one
another. An example of the former is a model of a very thin cylindrical shell. The degrees-
of-freedom normal to the shell and their associated bending degrees-of-freedom may all be
too soft to avoid numerical conditioning problems. A modeling cure for this condition is
to connect many grid points to a new reference grid point with an RBE3 element and then
to place the reference grid point in the r-set. The RBE3 element does not affect the flexible
modes when applied in this manner.

An example of a structure whose r-set shows poor linear independence is a slightly curved
bar modeled using coordinate systems that follow the curve of the bar such that the x-axis
is always tangent to the bar. The x DOFs at each end of the bar describe linearly
independent DOFs in a mathematical sense. However, numerical truncation produces
poor conditioning if the angle between the ends is less than a few degrees. This condition
is detected by the automatic diagnostics discussed earlier. This problem can be corrected
(or better yet, avoided) by making a careful sketch of all r-set DOFs, including their
locations in space and the orientation of their global coordinates. Then apply the three-
plane test described earlier.
Using a physical analogy, a good r-set can be chosen by finding one grid point that sustains
all possible loadings well if it is tied to ground in an actual hardware test. If there is no such
grid point, the ties to ground should be spread over enough grid points to sustain the loads
without damaging the structure. An RBE3 element used for this purpose can provide good
rigid-body modes without affecting the flexible modes.
96

4.3 Examples
This section provides several rigid-body modes examples showing input and output.
These examples are as follows:

Table 4-1 Unconstrained Beam Model Summary

Analysis SUPORT/ Redundancy of

Model
Method No SUPORT SUPORT
bd04bar1 Lanczos No SUPORT –
bd04bar2 Lanczos SUPORT Statically determinate
bd04bar3 Lanczos SUPORT Underdetermined
bd04bar4 Lanczos SUPORT Overdetermined
bd04bar5 SINV No SUPORT –
bd04bar6 SINV SUPORT Statically determinate
bd04bar7 SINV SUPORT Underdetermined
bd04bar8 SINV SUPORT Overdetermined
bd04bkt Lanczos No SUPORT –

These examples are described in the sections that follow.

Unconstrained Beam Model

The constraints (SPCs) on the example cantilever beam model from “Cantilever Beam
Model” on page 142 are removed to create an unconstrained structure as shown in
Figure 4-3. A GRDSET entry is added with the z-translation, x-rotation, and y-rotation
directions constrained to make the problem two-dimensional. Therefore, there are three
DOFs per grid point (x-translation, y-translation, and z-rotation) and three rigid-body
modes.

X
Z
Figure 4-3 Unconstrained Beam Model

Modes are computed using two methods (Lanczos and SINV), with and without a
SUPORT entry. The SUPORT entry is used in three ways:
• Statically determinate (grid point 1, components 1, 2, and 6)
• Underdetermined (grid point 1, components 1 and 2)
• Overdetermined (grid point 1, components 1, 2, and 6, plus grid point 11,
component1)
 CHAPTER 4 97
Rigid-body Modes

Listing 4-1 shows a portion of the input file for the statically determinate SUPORT and the
Lanczos method.

Listing 4-1 Input File for Cantilever Beam Model

$ FILE bd04bar2.dat
$
$ CANTILEVER BEAM MODEL
$ CHAPTER 4, RIGID-BODY MODES
$
SOL 103
TIME 10
CEND
TITLE = CANTILEVER BEAM
SUBTITLE = NORMAL MODES
LABEL = USE SUPORT, STATICALLY DETERMINATE
$
$ OUTPUT REQUEST
DISPLACEMENT = ALL
$
$ SELECT EIGRL ENTRY
METHOD = 10
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ STATICALLY DETERMINATE SUPORT
$SUPORT G C
SUPORT 1 126
$
$ MAKE 2D MODEL
GRDSET 345
$
$EIGRL SID V1 V2
EIGRL 10 -0.1 50.
$
... basic model ...
$
ENDDATA

Table 4-2 lists the computed frequencies. The overdetermined run for the Lanczos method
works well; the same run for the SINV method gives an extra zero frequency mode that
does not really exist. The overdetermined runs have redundant SUPORTs in the
x-direction; therefore, two rigid-body modes are computed in this direction when using the
SINV method. In all cases the flexible (greater than 0) frequencies are correct.
98

Table 4-2 Frequencies for the Unconstrained Beam Models

Lanczos Method Frequencies (Hz) SINV Method Frequencies (Hz)

Mode Stat. Over- Stat. Over

No Under No Under
Det. Det. Det. Det.
SUPORT Det. SUPORT Det.
SUPORT SUPORT SUPORT SUPORT

1 2.15E-7 0.0 0.0 0.0 6.88E-7 0.0 0.0 0.0

2 4.02E-7 0.0 0.0 0.0 6.79E-6 0.0 0.0 0.0
3 8.07E-6 3.07E-5 0.0 0.0 7.12E-6 1.45E-6 0.0 0.0
4 12.82 12.82 12.82 12.82 12.82 12.82 12.82 0.0
5 34.62 34.62 34.62 34.62 34.62 34.62 34.62 12.82
6 66.60 66.60 66.60 34.62
7 66.60

Figure 4-4 shows the output for the rigid-body modes computed for the SINV method
when using no SUPORT. The rigid-body frequencies are denoted by computational zeroes
on the order of 10–5 Hz or less. Note that the magnitude may be different when the same
problem is run on a different computer type.

Figure shows the output for the rigid-body modes computed for the SINV method when
using the statically determinate SUPORT. The accuracy of the statically determinate
SUPORT DOFs is verified by the computational zeroes for epsilon and the strain energy
printed in UIM 3035. Note that the three rigid-body modes have frequencies of 0.0 Hz. The
SUPORT entry provides cleaner mode shapes than those shown in Figure 4-4 as illustrated
by the purely x-translation, y-translation, and z-rotation eigenvectors shown in Figure 4-5.
 CHAPTER 4 99
Rigid-body Modes

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 1.866121E-11 4.319862E-06 6.875275E-07 1.000000E+00 1.866121E-11
2 2 1.819082E-09 4.265070E-05 6.788069E-06 1.000000E+00 1.819082E-09
3 3 2.000299E-09 4.472470E-05 7.118156E-06 1.000000E+00 2.000299E-09
4 4 6.483918E+03 8.052278E+01 1.281560E+01 1.000000E+00 6.483918E+03
5 5 4.732272E+04 2.175379E+02 3.462222E+01 1.000000E+00 4.732272E+04
6 6 1.751285E+05 4.184836E+02 6.660372E+01 1.000000E+00 1.751285E+05

EIGENVALUE = 1.866121E-11
CYCLES = 6.875275E-07 R E A L E I G E N V E C T O R N O . 1

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 4.273806E-34 -2.470432E-01 0.0 0.0 0.0 -1.067134E-01
2 G 4.273806E-34 -2.790573E-01 0.0 0.0 0.0 -1.067134E-01
3 G 4.273806E-34 -3.110713E-01 0.0 0.0 0.0 -1.067134E-01
4 G 4.273806E-34 -3.430853E-01 0.0 0.0 0.0 -1.067134E-01
5 G 4.273806E-34 -3.750993E-01 0.0 0.0 0.0 -1.067134E-01
6 G 4.273806E-34 -4.071134E-01 0.0 0.0 0.0 -1.067134E-01
7 G 4.273806E-34 -4.391274E-01 0.0 0.0 0.0 -1.067134E-01
8 G 4.273806E-34 -4.711414E-01 0.0 0.0 0.0 -1.067134E-01
9 G 4.273806E-34 -5.031555E-01 0.0 0.0 0.0 -1.067134E-01
10 G 4.273806E-34 -5.351695E-01 0.0 0.0 0.0 -1.067134E-01
11 G 4.273806E-34 -5.671835E-01 0.0 0.0 0.0 -1.067134E-01

EIGENVALUE = 1.819082E-09
CYCLES = 6.788069E-06 R E A L E I G E N V E C T O R N O . 2

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G -2.953214E-01 -5.597336E-01 0.0 0.0 0.0 3.291537E-01
2 G -2.953214E-01 -4.609875E-01 0.0 0.0 0.0 3.291537E-01
3 G -2.953214E-01 -3.622414E-01 0.0 0.0 0.0 3.291537E-01
4 G -2.953214E-01 -2.634953E-01 0.0 0.0 0.0 3.291537E-01
5 G -2.953214E-01 -1.647492E-01 0.0 0.0 0.0 3.291537E-01
6 G -2.953214E-01 -6.600305E-02 0.0 0.0 0.0 3.291537E-01
7 G -2.953214E-01 3.274305E-02 0.0 0.0 0.0 3.291537E-01
8 G -2.953214E-01 1.314892E-01 0.0 0.0 0.0 3.291538E-01
9 G -2.953214E-01 2.302354E-01 0.0 0.0 0.0 3.291538E-01
10 G -2.953214E-01 3.289815E-01 0.0 0.0 0.0 3.291538E-01
11 G -2.953214E-01 4.277276E-01 0.0 0.0 0.0 3.291538E-01

EIGENVALUE = 2.000299E-09
CYCLES = 7.118156E-06 R E A L E I G E N V E C T O R N O . 3

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G -2.953611E-01 5.596586E-01 0.0 0.0 0.0 -3.291096E-01
2 G -2.953611E-01 4.609257E-01 0.0 0.0 0.0 -3.291096E-01
3 G -2.953611E-01 3.621928E-01 0.0 0.0 0.0 -3.291096E-01
4 G -2.953611E-01 2.634599E-01 0.0 0.0 0.0 -3.291096E-01
5 G -2.953611E-01 1.647270E-01 0.0 0.0 0.0 -3.291096E-01
6 G -2.953611E-01 6.599414E-02 0.0 0.0 0.0 -3.291095E-01
7 G -2.953611E-01 -3.273870E-02 0.0 0.0 0.0 -3.291095E-01
8 G -2.953611E-01 -1.314715E-01 0.0 0.0 0.0 -3.291095E-01
9 G -2.953611E-01 -2.302044E-01 0.0 0.0 0.0 -3.291095E-01
10 G -2.953611E-01 -3.289373E-01 0.0 0.0 0.0 -3.291095E-01
11 G -2.953611E-01 -4.276701E-01 0.0 0.0 0.0 -3.291095E-01

Figure 4-4 Unconstrained Beam Modes Without SUPORT (SINV Method)

100

*** USER INFORMATION MESSAGE 3035 FOR DATA BLOCK KLR

SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 7.7496606E-17 5.5879354E-09
2 7.7496606E-17 0.0000000E+00
3 7.7496606E-17 7.1622708E-11

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 0.0 0.0 0.0 1.000000E+00 0.0
2 2 0.0 0.0 0.0 1.000000E+00 0.0
3 3 0.0 0.0 0.0 1.000000E+00 0.0
4 4 6.483918E+03 8.052278E+01 1.281560E+01 1.000000E+00 6.483918E+03
5 5 4.732272E+04 2.175379E+02 3.462222E+01 1.000000E+00 4.732272E+04
6 6 1.751285E+05 4.184836E+02 6.660372E+01 1.000000E+00 1.751285E+05
7 7 4.616299E+05 6.794335E+02 1.081352E+02 1.000000E+00 4.616299E+05

EIGENVALUE = 0.000000E+00
CYCLES = 0.000000E+00 R E A L E I G E N V E C T O R N O . 1

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
2 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
3 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
4 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
5 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
6 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
7 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
8 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
9 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
10 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0
11 G 4.176756E-01 0.0 0.0 0.0 0.0 0.0

EIGENVALUE = 0.000000E+00
CYCLES = 0.000000E+00 R E A L E I G E N V E C T O R N O . 2

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 4.176756E-01 0.0 0.0 0.0 0.0
2 G 0.0 4.176756E-01 0.0 0.0 0.0 -3.593777E-15
3 G 0.0 4.176756E-01 0.0 0.0 0.0 -6.723840E-15
4 G 0.0 4.176756E-01 0.0 0.0 0.0 -9.575675E-15
5 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.220145E-14
6 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.439249E-14
7 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.596912E-14
8 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.698929E-14
9 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.745155E-14
10 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.770949E-14
11 G 0.0 4.176756E-01 0.0 0.0 0.0 -1.758923E-14

EIGENVALUE = 0.000000E+00
CYCLES = 0.000000E+00 R E A L E I G E N V E C T O R N O . 3

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 -7.163078E-01 0.0 0.0 0.0 4.775385E-01
2 G 0.0 -5.730462E-01 0.0 0.0 0.0 4.775385E-01
3 G 0.0 -4.297847E-01 0.0 0.0 0.0 4.775385E-01
4 G 0.0 -2.865231E-01 0.0 0.0 0.0 4.775385E-01
5 G 0.0 -1.432615E-01 0.0 0.0 0.0 4.775385E-01
6 G 0.0 1.508516E-14 0.0 0.0 0.0 4.775385E-01
7 G 0.0 1.432615E-01 0.0 0.0 0.0 4.775385E-01
8 G 0.0 2.865231E-01 0.0 0.0 0.0 4.775385E-01
9 G 0.0 4.297847E-01 0.0 0.0 0.0 4.775385E-01
10 G 0.0 5.730463E-01 0.0 0.0 0.0 4.775385E-01
11 G 0.0 7.163078E-01 0.0 0.0 0.0 4.775385E-01

Figure 4-5 Unconstrained Beam Modes With Statically Determinate SUPORT

(SINV Method)

Figure 4-6 shows the epsilon and strain energy printed in UIM 3035 for the three SUPORT
cases (statically determinate, overdetermined, and underdetermined).

It can be seen from this example that aside from clean rigid-body vectors there is no
advantage to using a SUPORT entry to compute rigid-body modes.
 CHAPTER 4 101
Rigid-body Modes

Statically determinate SUPORT:

*** USER INFORMATION MESSAGE 3035 FOR DATA BLOCK KLR

SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 7.7496606E-17 5.5879354E-09
2 7.7496606E-17 0.0000000E+00
3 7.7496606E-17 7.1622708E-11

Overdetermined SUPORT:

*** USER INFORMATION MESSAGE 3035 FOR DATA BLOCK KLR

SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 1.9913979E-01 7.2869660E+06 ****
2 1.9913979E-01 0.0000000E+00 ****
3 1.9913979E-01 7.1850081E-11 ****
4 1.9913979E-01 7.2869660E+06 ****

Underdetermined SUPORT:

*** USER INFORMATION MESSAGE 4158---STATISTICS FOR SYMMETRIC DECOMPOSITION OF DATA BLOCK KLL FOLLOW

MAXIMUM RATIO OF MATRIX DIAGONAL TO FACTOR DIAGONAL = 9.1E+13 AT ROW NUMBER 31

*** USER WARNING MESSAGE 4698. STATISTICS FOR DECOMPOSITION OF MATRIX KLL .
THE FOLLOWING DEGREES OF FREEDOM HAVE FACTOR DIAGONAL RATIOS GREATER THAN
1.00000E+05 OR HAVE NEGATIVE TERMS ON THE FACTOR DIAGONAL.

GRID POINT ID DEGREE OF FREEDOM MATRIX/FACTOR DIAGONAL RATIO MATRIX DIAGONAL

11 R3 9.49483E+13 2.84000E+04

*** USER INFORMATION MESSAGE 3035 FOR DATA BLOCK KLR

SUPPORT PT.NO. EPSILON STRAIN ENERGY EPSILONS LARGER THAN 0.001 ARE FLAGGED WITH ASTERISKS
1 7.6288287E-17 5.5879354E-09
2 7.6288287E-17 -7.2759576E-12

Figure 4-6 UIM 3035 Results

Unconstrained Bracket Example

The constraints (SPCs) on the example bracket model from “Real Eigenvalue Analysis” on
page 35 are removed to create an unconstrained model (see “Examples” on page 58 for a
description of the model). Figure 4-7 shows the bracket model. The model is a
three-dimensional model and therefore produces six rigid-body modes. The MSC.Nastran
results are shown in Figure 4-8 for the first eight modes. The Lanczos method is used.

Figure 4-7 Unconstrained Bracket Model

102

R E A L E I G E N V A L U E S
MODE EXTRACTION EIGENVALUE RADIANS CYCLES GENERALIZED GENERALIZED
NO. ORDER MASS STIFFNESS
1 1 -1.690642E-07 4.111741E-04 6.544039E-05 1.000000E+00 -1.690642E-07
2 2 -9.807991E-09 9.903530E-05 1.576196E-05 1.000000E+00 -9.807991E-09
3 3 -5.515176E-09 7.426423E-05 1.181952E-05 1.000000E+00 -5.515176E-09
4 4 -3.390596E-09 5.822883E-05 9.267405E-06 1.000000E+00 -3.390596E-09
5 5 1.266017E-08 1.125174E-04 1.790770E-05 1.000000E+00 1.266017E-08
6 6 1.726585E-08 1.313996E-04 2.091289E-05 1.000000E+00 1.726585E-08
7 7 2.649932E+06 1.627861E+03 2.590821E+02 1.000000E+00 2.649932E+06
8 8 4.279463E+06 2.068686E+03 3.292416E+02 1.000000E+00 4.279463E+06

Figure 4-8 Unconstrained Bracket Frequencies

The six rigid-body modes have computational zero frequencies on the order of 10–5 Hz.
Note that the magnitudes of the rigid-body modes may be different when the same
problem is run on a different computer type. Also note that the output is sorted by the
value of the eigenvalue in ascending order.
 NX Nastran Basic Dynamic Analysis User’s Guide

5
CHAPTER
Frequency Response Analysis

■ Overview
■ Direct Frequency Response Analysis

■ Modal Frequency Response Analysis

■ Modal Versus Direct Frequency Response
■ Frequency-Dependent Excitation Definition
■ Solution Frequencies
■ Frequency Response Considerations

■ Solution Control for Frequency Response Analysis

■ Examples
104

5.1 Overview
Frequency response analysis is a method used to compute structural response to
steady-state oscillatory excitation. Examples of oscillatory excitation include rotating
machinery, unbalanced tires, and helicopter blades. In frequency response analysis the
excitation is explicitly defined in the frequency domain. All of the applied forces are
known at each forcing frequency. Forces can be in the form of applied forces and/or
enforced motions (displacements, velocities, or accelerations).

Phase Shift. Oscillatory loading is sinusoidal in nature. In its simplest case, this loading
is defined as having an amplitude at a specific frequency. The steady-state oscillatory
response occurs at the same frequency as the loading. The response may be shifted in time
due to damping in the system. The shift in response is called a phase shift because the peak
loading and peak response no longer occur at the same time. An example of phase shift is
shown in Figure 5-1.

Phase
Shift
Loading Response

Time

Figure 5-1 Phase Shift

Complex Numbers. The important results obtained from a frequency response analysis
usually include the displacements, velocities, and accelerations of grid points as well as the
forces and stresses of elements. The computed responses are complex numbers defined as
magnitude and phase (with respect to the applied force) or as real and imaginary
components, which are vector components of the response in the real/imaginary plane.
These quantities are graphically presented in Figure 5-2.
 CHAPTER 5 105
Frequency Response Analysis

i (Imaginary)

ui u

θ
ur r (Real)

Figure 5-2 Complex Plane

where:

u = magnitude = u 2r + u 2i

θ = phase angle = tan–1 ( u i ⁄ u r )

u r = real component = u cos θ

u i = imaginary component = u sin θ

Two different numerical methods can be used in frequency response analysis. The direct
method solves the coupled equations of motion in terms of forcing frequency. The modal
method utilizes the mode shapes of the structure to reduce and uncouple the equations of
motion (when modal or no damping is used); the solution for a particular forcing
frequency is obtained through the summation of the individual modal responses. The
choice of the method depends on the problem. The two methods are described in “Direct
Frequency Response Analysis” on page 106 and “Modal Frequency Response Analysis”
on page 108.
106

5.2 Direct Frequency Response Analysis

In direct frequency response analysis, structural response is computed at discrete
excitation frequencies by solving a set of coupled matrix equations using complex algebra.
Begin with the damped forced vibration equation of motion with harmonic excitation

[ M ] { x·· ( t ) } + [ B ] { x· ( t ) } + [ K ] { x ( t ) } = { P ( ω ) }e
iω t
Eq. 5-1

The load in Eq. 5-1 is introduced as a complex vector, which is convenient for the
mathematical solution of the problem. From a physical point of view, the load can be real
or imaginary, or both. The same interpretation is used for response quantities.

For harmonic motion (which is the basis of a frequency response analysis), assume a
harmonic solution of the form:

iω t
{ x } = { u ( ω ) }e Eq. 5-2

where { u ( ω ) } is a complex displacement vector. Taking the first and second derivatives of
Eq. 5-2, the following is obtained:

{ x· } = iω { u ( ω ) }e
iωt

Eq. 5-3

2 i ωt
{ x·· } = – ω { u ( ω ) }e

When the above expressions are substituted into Eq. 5-1, the following is obtained:

2 iωt iω t iω t iω t
– ω [ M ] { u ( ω ) }e + iω [ B ] { u ( ω ) }e + [ K ] { u ( ω ) }e = { P ( ω ) }e Eq. 5-4

which after dividing by e iωt simplifies to

2
[ – ω M + iωB + K ] { u ( ω ) } = { P ( ω ) } Eq. 5-5

The equation of motion is solved by inserting the forcing frequency ω into the equation of
motion. This expression represents a system of equations with complex coefficients if
damping is included or the applied loads have phase angles. The equations of motion at
each input frequency are then solved in a manner similar to a statics problem using
complex arithmetic.

Damping in Direct Frequency Response

Damping simulates the energy dissipation characteristics of a structure. Damping in direct
frequency response is represented by the damping matrix [ B ] and additions to the stiffness
matrix [ K ] .
 CHAPTER 5 107
Frequency Response Analysis

The damping matrix is comprised of several matrices

1 2
[B] = [B ] + [B ] Eq. 5-6

where:

1
[ B ] = damping elements (CVISC, CDAMPi) and B2GG
2
[ B ] = B2PP direct input matrix and transfer functions

In frequency response, PARAM,G and GE on the MATi entry do not form a damping
matrix. Instead, they form the following complex stiffness matrix:

[ K ] = ( 1 + iG ) [ K ] + i ∑ G E [ K E ] Eq. 5-7

where:

[ K ] = global stiffness matrix

G = overall structural damping coefficient (PARAM,G)
[ K E ] = element stiffness matrix

G E = element structural damping coefficient (GE on the MATi entry)

When the above parameters and/or coefficients are specified, they are automatically
incorporated into the stiffness matrix and therefore into the equation of motion for the
solution. All of the forms of damping can be used in the same analysis, and their effects
are added together.
In frequency response analysis, it is not necessary to assume an equivalent viscous form for
structural damping since the solution is complex. Therefore, a complex stiffness matrix is
allowed.
108

5.3 Modal Frequency Response Analysis

Modal frequency response analysis is an alternate approach to computing the frequency
response of a structure. This method uses the mode shapes of the structure to reduce the
size, uncouple the equations of motion (when modal or no damping is used), and make the
numerical solution more efficient. Since the mode shapes are typically computed as part
of the characterization of the structure, modal frequency response is a natural extension of
a normal modes analysis.

As a first step in the formulation, transform the variables from physical coordinates { u ( ω ) }
to modal coordinates { ξ ( ω ) } by assuming

i ωt
{ x } = [ φ ] { ξ ( ω ) }e Eq. 5-8

The mode shapes [ φ ] are used to transform the problem in terms of the behavior of the
modes as opposed to the behavior of the grid points. Eq. 5-8 represents an equality if all
modes are used; however, because all modes are rarely used, the equation usually
represents an approximation.
To proceed, temporarily ignore all damping, which results in the undamped equation for
harmonic motion

2
–ω [ M ] { x } + [ K ] { x } = { P ( ω ) } Eq. 5-9

at forcing frequency ω .
Substituting the modal coordinates in Eq. 5-8 for the physical coordinates in Eq. 5-9 and
dividing by e iωt , the following is obtained:

2
–ω [ M ] [ φ ] { ξ ( ω ) } + [ K ] [ φ ] { ξ ( ω ) } = { P ( ω ) } Eq. 5-10

Now this is the equation of motion in terms of the modal coordinates. At this point,
however, the equations remain coupled.
To uncouple the equations, premultiply by [ φT ] to obtain

2 T T T
– ω [φ ] [ M][φ ]{ξ( ω)} + [φ ] [K][φ ]{ξ( ω)} = [φ ] {P(ω)} Eq. 5-11

where:

[ φ ] [ M ] [ φ ] = modal (generalized) mass matrix

T

[ φ ] [ K ] [ φ ] = modal (generalized) stiffness matrix

T

[ φ ] { P } = modal force vector

T

The final step uses the orthogonality property of the mode shapes to formulate the
equation of motion in terms of the generalized mass and stiffness matrices, which are
diagonal matrices. These diagonal matrices do not have the off-diagonal terms that couple
 CHAPTER 5 109
Frequency Response Analysis

the equations of motion. Therefore, in this form the modal equations of motion are
uncoupled. In this uncoupled form, the equations of motion are written as a set of
uncoupled single degree-of-freedom systems as

2
–ω mi ξi ( ω ) + ki ξ ( ω ) = pi ( ω ) Eq. 5-12

where:

m i = i-th modal mass

k i = i-th modal stiffness

p i = i-th modal force

The modal form of the frequency response equation of motion is much faster to solve than
the direct method because it is a series of uncoupled single degree-of-freedom systems.
Once the individual modal responses ξi ( ω ) are computed, physical responses are
recovered as the summation of the modal responses using

iωt
{ x } = [ φ ] { ξ ( ω ) }e Eq. 5-13

These responses are in complex form (magnitude/phase or real/imaginary) and are used
to recover additional output quantities requested in the Case Control Section.

Damping in Modal Frequency Response

If a damping matrix [ B ] exists, the orthogonality property (see “Overview of Normal
Modes Analysis” on page 39) of the modes does not, in general, diagonalize the
generalized damping matrix

T
[ φ ] [ B ] [ φ ] ≠ diagonal Eq. 5-14

If structural damping is used, the orthogonality property does not, in general, diagonalize
the generalized stiffness matrix

T
[ φ ] [ K ] [ φ ] ≠ diagonal Eq. 5-15

where: K = ( 1 + iG ) [ K ] + i ∑ G E [ K E ]

In the presence of a [ B ] matrix or a complex stiffness matrix, the modal frequency approach
solves the coupled problem in terms of modal coordinates using the direct frequency
approach described in “Direct Frequency Response Analysis” on page 106

2 T T T T
[ – ω [ φ ] [ M ] [ φ ] + iω [ φ ] [ B ] [ φ ] + [ φ ] [ K ] [ φ ] ] { ξ ( ω ) } = [ φ ] { P ( ω ) } Eq. 5-16

Eq. 5-16 is similar to Eq. 5-5 for the direct frequency response analysis method except that
Eq. 5-16 is expressed in terms of modal coordinates ξ . Since the number of modes used in
a solution is typically much less than the number of physical variables, using the coupled
solution of the modal equations is less costly than using physical variables.
110

If damping is applied to each mode separately, the uncoupled equations of motion can be
maintained. When modal damping is used, each mode has damping b i where b i = 2m i ω i ζi .
The equations of motion remain uncoupled and have the form

2
– ω m i ξ i ( ω ) + iωb i ξ i ( ω ) + k i ξ i ( ω ) = p i ( ω ) Eq. 5-17

for each mode.

Each of the modal responses is computed using

pi ( ω )
ξ i ( ω ) = ------------------------------------------------ Eq. 5-18
2
– m i ω + ib i ω + k i

The TABDMP1 Bulk Data entry defines the modal damping ratios. A table is created by
the frequency/damping pairs specified on the TABDMP1 entry. The solution refers to this
table for the damping value to be applied at a particular frequency. The TABDMP1 Bulk
Data entry has a Table ID. A particular TABDMP1 table is activated by selecting the Table
ID with the SDAMPING Case Control command.

1 2 3 4 5 6 7 8 9 10
TABDMP1 TID TYPE

f1 g1 f2 g2 f3 g3 -etc.- ENDT

Field Contents
TID Table identification number.
TYPE Type of damping units:
G (default)
CRIT
Q
fi Frequency value (cycles per unit time).
gi Damping value in the units specified.

At resonance, the three types of damping are related by the following equations:

bi Gi
ζ i = ------- = ------
b cr 2

b cr = 2m i ω i Eq. 5-19

1 1
Q i = ------------- = ------
( 2ζ i ) Gi
 CHAPTER 5 111
Frequency Response Analysis

Note that the i subscript is for the i-th mode, and not the i-th excitation frequency.

The values of fi and gi define pairs of frequencies and dampings. Note that gi can be
entered as one of the following: structural damping (default), critical damping, or quality
factor. The entered damping is converted to structural damping internally using Eq. 5-19.
Straight-line interpolation is used for modal frequencies between consecutive fi values.
Linear extrapolation is used at the ends of the table. ENDT ends the table input.

For example, if modal damping is entered using Table 5-1 and modes exist at 1.0, 2.5, 3.6,
and 5.5 Hz, MSC.Nastran interpolates and extrapolates as shown in Figure 5-3 and in the
table. Note that there is no table entry at 1.0 Hz; MSC.Nastran uses the first two table
entries at f = 2.0 and f = 3.0 to extrapolate the value for f = 1.0 .
0.20
+
ζ +
+ +
0.10
+ = entered value
= computed value

0.0
2.0 4.0 6.0
f (Hz)
Figure 5-3 Example TABDMP1

Table 5-1 Example TABDMP1 Interpolation/Extrapolation

Entered Computed
f ζ f ζ
2.0 0.16 1.0 0.14
3.0 0.18 2.5 0.17
4.0 0.13 3.6 0.15
6.0 0.13 5.5 0.13

1 2 3 4 5 6 7 8 9 10
TABDMP1 10 CRIT +TAB1

+TAB1 2.0 0.16 3.0 0.18 4.0 0.13 6.0 0.13 +TAB2

+TAB2 ENDT

Modal damping is processed as a complex stiffness when PARAM,KDAMP is entered as

-1. The uncoupled equation of motion becomes

2
– ω m i ξ i ( ω ) + ( 1 + iG ( ω ) )k i ξ i ( ω ) = p i ( ω ) Eq. 5-20
112

The default for PARAM,KDAMP is 1, which processes modal damping as a damping

matrix as shown in Eq. 5-17.
The decoupled solution procedure used in modal frequency response can be used only if
either no damping is present or modal damping alone (via TABDMP1) is used. Otherwise,
the modal method uses the coupled solution method on the smaller modal coordinate
matrices if nonmodal damping (i.e., CVISC, CDAMPi, GE on the MATi entry, or
PARAM,G) is present.

Mode Truncation in Modal Frequency Response Analysis

It is possible that not all of the computed modes are required in the frequency response
solution. You need to retain, at a minimum, all the modes whose resonant frequencies lie
within the range of forcing frequencies. For example, if the frequency response analysis
must be between 200 and 2000 Hz, all modes whose resonant frequencies are in this range
should be retained. This guideline is only a minimum requirement, however. For better
accuracy, all modes up to at least two to three times the highest forcing frequency should
be retained. In the example where a structure is excited to between 200 and 2000 Hz, all
modes from 0 to at least 4000 Hz should be retained.

The frequency range selected on the eigenvalue entry (EIGRL or EIGR) is one means to
control the modes used in the modal frequency response solution. Also, three parameters
are available to limit the number of modes included in the solution. PARAM,LFREQ gives
the lower limit on the frequency range of retained modes, and PARAM,HFREQ gives the
upper limit on the frequency range of retained modes. PARAM,LMODES gives the
number of lowest modes to be retained. These parameters can be used to include the
proper set of modes. Note that the default is for all computed modes to be retained.

Dynamic Data Recovery in Modal Frequency Response

Analysis
In modal frequency response analysis, two options are available for recovering
displacements and stresses: the mode displacement method and the matrix method. Both
methods give the same answers, although with differences in cost.

The mode displacement method computes the total physical displacements for each
excitation frequency from the modal displacements, and then computes element stresses
from the total physical displacements. The number of operations is proportional to the
number of excitation frequencies.

The matrix method computes displacements per mode and element stresses per mode, and
then computes physical displacements and element stresses as the summation of modal
displacements and element stresses. Costly operations are proportional to the number of
modes.
 CHAPTER 5 113
Frequency Response Analysis

Since the number of modes is usually much less that the number of excitation frequencies,
the matrix method is usually more efficient and is the default. The mode displacement
method can be selected by using PARAM,DDRMM,-1 in the Bulk Data. The mode
displacement method is required when “frequency-frozen” structural plots are requested
(see “Plotted Output” on page 247).

The mode acceleration method (“Advanced Dynamic Analysis Capabilities” on page 293)
is another data recovery method for modal frequency response analysis. This method can
provide better accuracy since detailed local stresses and forces are subject to mode
truncation and may not be as accurate as the results computed with the direct method.
114

5.4 Modal Versus Direct Frequency Response

Some general guidelines can be used when selecting modal frequency response analysis
versus direct frequency response analysis. These guidelines are summarized in Table 5-2.

Table 5-2 Modal Versus Direct Frequency Response

Modal Direct
Small Model X
Large Model X
Few Excitation Frequencies X
Many Excitation Frequencies X
High Frequency Excitation X
Nonmodal Damping X
Higher Accuracy X

In general, larger models may be solved more efficiently in modal frequency response
because the numerical solution is a solution of a smaller system of uncoupled equations.
The modal method is particularly advantageous if the natural frequencies and mode
shapes were computed during a previous stage of the analysis. In that case, you simply
perform a restart (see “Restarts In Dynamic Analysis” on page 223). Using the modal
approach to solve the uncoupled equations is very efficient, even for very large numbers
of excitation frequencies. On the other hand, the major portion of the effort in a modal
frequency response analysis is the calculation of the modes. For large systems with a large
number of modes, this operation can be as costly as a direct solution. This result is
especially true for high-frequency excitation. To capture high frequency response in a
modal solution, less accurate, high-frequency modes must be computed. For small models
with a few excitation frequencies, the direct method may be the most efficient because it
solves the equations without first computing the modes. The direct method is more
accurate than the modal method because the direct method is not concerned with mode
truncation.
Table 5-2 provides an overview of which method to use. Many additional factors may be
involved in the choice of a method, such as contractual obligations or local standards of
practice.
 CHAPTER 5 115
Frequency Response Analysis

5.5 Frequency-Dependent Excitation Definition

An important aspect of a frequency response analysis is the definition of the loading
function. In a frequency response analysis, the force must be defined as a function of
frequency. Forces are defined in the same manner regardless of whether the direct or
modal method is used.
The following Bulk Data entries are used for the frequency-dependent load definition:

RLOAD1 Tabular input-real and imaginary.

RLOAD2 Tabular input-magnitude and phase.
DAREA Spatial distribution of dynamic load.
LSEQ Generates the spatial distribution of dynamic loads from static load
entries.
DLOAD Combines dynamic load sets.
TABLEDi Tabular values versus frequency.
DELAY Time delay.
DPHASE Phase lead.

The particular entry chosen for defining the dynamic loading is largely a function of user
convenience for concentrated loads. Pressure and distributed loads, however, require a
more complicated format.

There are two important aspects of dynamic load definition. First, the location of the
loading on the structure must be defined. Since this characteristic locates the loading in
space, it is called the spatial distribution of the dynamic loading. Secondly, the frequency
variation in the loading is the characteristic that differentiates a dynamic load from a static
load. This frequency variation is called the temporal distribution of the load. A complete
dynamic loading is a product of spatial and temporal distributions.

Using Table IDs and Set IDs in MSC.Nastran makes it possible to apply many complicated
and temporally similar loadings with a minimum of input. Combining simple loadings to
create complicated loading distributions that vary in position as well as frequency is also a
straightforward task.
The remainder of this section describes the Bulk Data entries for frequency-dependent
excitation. The description is given in terms of the coefficients that define the dynamic
load. See “Bulk Data Entries for Dynamic Analysis” on page 367 for more complete Bulk
Data descriptions.
116

Frequency-Dependent Loads – RLOAD1 Entry

The RLOAD1 entry is a general form in which to define a frequency-dependent load. It
defines a dynamic loading of the form

i { θ – 2πfτ }
{ P ( f ) } = { A [ C ( f ) + iD ( f ) ]e } Eq. 5-21

The values of the coefficients are defined in tabular format on a TABLEDi entry. You need
not explicitly define a force at every excitation frequency. Only those values that describe
the character of the loading are required. MSC.Nastran will interpolate for intermediate
values.

1 2 3 4 5 6 7 8 9 10
RLOAD1 SID DAREA DELAY DPHASE TC TD

Field Contents
SID Set ID defined by a DLOAD Case Control command or a DLOAD Bulk
Data entry.
DAREA Identification number of the DAREA entry set that defines A.
(Integer > 0)
DELAY Identification number of the DELAY entry set that defines τ .
DPHASE Identification number of the DPHASE entry set that defines θ .
TC TABLEDi entry that defines C(f).
TD TABLEDi entry that defines D(f).

Note that f is the frequency in cycles per unit time and that e iwt = ( cot wt + i sin wt ) .
 CHAPTER 5 117
Frequency Response Analysis

Frequency-Dependent Loads – RLOAD2 Entry

The RLOAD2 entry is a variation of the RLOAD1 entry used for defining a
frequency-dependent load. Whereas the RLOAD1 entry defines the real and imaginary
parts of the complex load, the RLOAD2 entry defines the magnitude and phase.

The RLOAD2 entry defines dynamic excitation in the form

i { φ ( f ) + θ – 2πfτ }
{ P ( f ) } = { AB ( f )e } Eq. 5-22

The RLOAD2 definition may be related to the RLOAD1 definition by

iφ ( f )
C ( f ) + iD ( f ) = B ( f )e Eq. 5-23

RLOAD1 RLOAD2
Definition Definition

1 2 3 4 5 6 7 8 9 10
RLOAD2 SID DAREA DELAY DPHASE TB TP

Field Contents
SID Set ID defined by a DLOAD Case Control command.
DAREA Identification number of the DAREA entry set that defines A.
(Integer > 0)
DELAY Identification number of the DELAY entry set that defines τ .
(Integer > 0)
DPHASE Identification number of the DPHASE entry set that defines θ in
degrees. (Integer > 0)
TB TABLEDi entry defining amplitude versus frequency pairs for B ( f ) .
(Integer > 0)
TP TABLEDi entry defining phase angle versus frequency pairs for φ ( f ) in
degrees. (Integer > 0)

Note that f is the frequency in cycles per unit time.

118

Spatial Distribution of Loading – DAREA Entry

The DAREA entry defines the degrees-of-freedom where the dynamic load is to be applied
and the scale factor to be applied to the loading. The DAREA entry provides the basic
spatial distribution of the dynamic loading.

1 2 3 4 5 6 7 8 9 10
DAREA SID P1 C1 A1 P2 C2 A2

Field Contents
SID Set ID specified by RLOADi entires.
Pi Grid, extra, or scalar point ID.
Ci Component number.
Ai Scale factor.

A DAREA entry is selected by RLOAD1 or RLOAD2 entries. Any number of DAREA

entries may be used; all those with the same SID are combined.

Time Delay – DELAY Entry

The DELAY entry defines the time delay τ in an applied load.

1 2 3 4 5 6 7 8 9 10
DELAY SID P1 C1 P2 C2 τ2

Field Contents
SID Set ID specified by an RLOADi entry.
Pi Grid, extra, or scalar point ID.
Ci Component number.
τi Time delay for Pi, Ci. (Default = 0.0)

A DAREA entry must be defined for the same point and component. Any number of
DELAY entries may be used; all those with the same SID are combined.
 CHAPTER 5 119
Frequency Response Analysis

Phase Lead – DPHASE Entry

The DPHASE entry defines the phase lead θ .

1 2 3 4 5 6 7 8 9 10
DPHASE SID P1 C1 θ1 P2 C2 θ2

Field Contents
SID Set ID specified by an RLOADi entry.
Pi Grid, extra, or scalar point ID.
Ci Component number.
θi Phase lead (in degrees) for Pi, Ci. (Default = 0.0)

A DAREA entry must be defined for the same point and component. Any number of
DPHASE entires may be used; all those with the same SID are combined.
120

Dynamic Load Tabular Function – TABLEDi Entries

The TABLEDi entries (i = 1 through 4) each define a tabular function for use in generating
frequency-dependent dynamic loads. The form of each TABLEDi entry varies slightly,
depending on the value of i, as does the algorithm for y(x). The x values need not be evenly
spaced.
The TABLED1, TABLED2, and TABLED3 entries linearly interpolate between the end
points and linearly extrapolate outside of the endpoints, as shown in Figure 5-4. The
TABLED1 entry also performs logrithmic interpolation between points. The TABLED4
entry assigns the endpoint values to any value beyond the endpoints.

Linear Interpolation Between Endpoints

y

Linear Extrapolation of Segment x6 - x7

Linear Extrapolation of Segment x2 - x1

x
x1 x2 x3 x4 x5 x6 x7

Figure 5-4 Interpolation and Extrapolation for TABLED1, TABLED2,

and TABLED3 Entries

The TABLED1 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED1 TID XAXIS YAXIS

x1 y1 x2 y2 x3 y3 -etc.- ENDT

Field Contents
TID Table identification number.
XAXIS Specifies a linear or logarithmic interpolation for the x-axis. (Character:
“LINEAR” or “LOG”; default = “LINEAR”)
YAXIS Specifies a linear or logarithmic interpolation for the y-axis. (Character:
“LINEAR” or “LOG”; default = “LINEAR”)
xi, yi Tabular values. Values of x are frequency in cycles per unit time.
ENDT Ends the table input.
 CHAPTER 5 121
Frequency Response Analysis

The TABLED1 entry uses the algorithm

y = yT ( x ) Eq. 5-24

The algorithms used for interpolation and extrapolation are as follows:

XAXIS YAXIS yT ( x )

xj – x x – xi
LINEAR LINEAR ----------------- yi + ----------------- yj
xj – xi xj – xi

ln ( xj ⁄ x ) ln ( x ⁄ x i )
LOG LINEAR --------------------------- yi + --------------------------- yj
ln ( xj ⁄ xi ) ln ( xj ⁄ xi )

xj – x x – xi
LINEAR LOG exp ----------------- ln yi + ----------------- ln yj
xj – xi xj – xi

ln ( xj ⁄ x ) ln ( x ⁄ x i )
LOG LOG exp --------------------------- ln y i + --------------------------- yj
ln ( xj ⁄ xi ) ln ( xj ⁄ xi )

The TABLED2 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED2 TID X1

x1 y1 x2 y2 x3 y3 -etc.- ENDT

Field Contents
TID Table identification number.
X1 Table parameter.
xi, yi Tabular values. Values of x are frequency in cycles per unit time.

The TABLED2 entry uses the algorithm

y = y T ( x – X1 ) Eq. 5-25

ENDT ends the table input.

122

The TABLED3 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED3 TID X1 X2

x1 y1 x2 y2 x3 y3 -etc.- ENDT

Field Contents
TID Table identification number.
X1, X2 Table parameters (X2 ≠ 0.0).
xi, yi Tabular values. Values of x are frequency in cycles per unit time.

The TABLED3 entry uses the algorithm

y = y T  -----------------
x – X1
Eq. 5-26
X2

ENDT ends the table input.

The TABLED4 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED4 TID X1 X2 X3 X4

A0 A1 A2 A3 A4 A5 -etc.- ENDT

Field Contents
TID Table identification number.
Xi Table parameters (X2 ≠ 0.0; X3 < X4).
Ai Coefficients.

The TABLED4 entry uses the algorithm

N
x – X1 i
y = ∑ Ai  ----------------- Eq. 5-27
X2
i = 0

N is the degree of the power series. When x < X3, X3 is used for x; when x > X4, X4 is used
for x. This condition has the effect of placing bounds on the table; there is no extrapolation
outside of the table boundaries.

ENDT ends the table input.

 CHAPTER 5 123
Frequency Response Analysis

DAREA Example
Suppose the following command is in the Case Control Section:

DLOAD = 35
in addition to the following entries in the Bulk Data Section:

1 2 3 4 5 6 7 8 9 10
$RLOAD1 SID DAREA DELAY DPHASE TC TD

RLOAD1 35 29 31 40

$DAREA SID POINT COMPONENT SCALE

DAREA 29 30 1 5.2

$DELAY SID POINT COMPONENT LAG

DELAY 31 30 1 0.2

$TABLED1 ID XAXIS YAXIS

$ x1 y1 x2 y2 x3 y3 x4 y4

TABLED1 40 LINEAR LINEAR

0.0 4.0 2.0 8.0 6.0 8.0 ENDT

The DLOAD Set ID 35 in Case Control selects the RLOAD1 entry in the Bulk Data having
a Set ID 35. On the RLOAD1 entry is a reference to DAREA Set ID 29, DELAY Set ID 31,
and TABLED1 Set ID 40. The DAREA entry with Set ID 29 positions the loading on grid
point 30 in the 1 direction with a scale factor of 5.2 applied to the load. The DELAY entry
with Set ID 31 delays the loading on grid point 30 in the 1 direction by 0.2 units of time.
The TABLED1 entry with Set ID 40 defines the load in tabular form. This table is shown
graphically in Figure 5-5. The result of these entries is a dynamic load applied to grid point
30, component T1, scaled by 5.2 and delayed by 0.2 units of time.

10

Amplitude
on 5
TABLED1

0 2 4 6
Frequency (Hz)
Figure 5-5 TABLED1 - Amplitude Versus Frequency
124

Static Load Sets – LSEQ Entry

MSC.Nastran does not have specific data entries for many types of dynamic loads. Only
concentrated forces and moments can be specified directly using DAREA entries. To
accommodate more complicated loadings conveniently, the LSEQ entry is used to define
static load entries that define the spatial distribution of dynamic loads.

1 2 3 4 5 6 7 8 9 10
LSEQ SID DAREA ID LOAD ID TEMP ID

The LSEQ Bulk Data entry contains a reference to a DAREA Set ID and a static Load Set ID.
The static loads are combined with any DAREA entries in the referenced set. The DAREA
Set ID does not need to be defined with a DAREA Bulk Data entry. The DAREA Set ID is
referenced by an RLOADi entry. This reference defines the temporal distribution of the
dynamic loading. The Load Set ID may refer to one or more static load entries (FORCE,
PLOADi, GRAV, etc.). All static loads with the Set ID referenced on the LSEQ entry define
the spatial distribution of the dynamic loading. MSC.Nastran converts this information to
equivalent dynamic loading.
Figure 5-6 demonstrates the relationships of these entries. To activate a load set defined in
this manner, the DLOAD Case Control command refers to the Set ID of the selected
DLOAD or RLOADi entry, and the LOADSET Case Control command refers to the Set ID
of the selected LSEQ entries. The LSEQ entries point to the static loading entries that are
used to define dynamic loadings with DAREA Set ID references. Together this relationship
defines a complete dynamic loading. To apply dynamic loadings in this manner, the
DLOAD and LOADSET Case Control commands and the RLOADi and LSEQ Bulk Data
entries must be defined. A DAREA Bulk Data entry does not need to be defined since the
RLOADi and LSEQ entries reference a common DAREA ID. The LSEQ entry can also be
interpreted as an internal DAREA entry generator for static load entries.

DLOAD LOADSET

Case Control
Bulk Data
RLOADi LSEQ

Dynamic Load DAREA Static Load Entries

(Temporal (ReferenceLink) (Spatial
Distribution) Distribution)
Figure 5-6 Relationship of Dynamic and Static Load Entries
 CHAPTER 5 125
Frequency Response Analysis

LSEQ Example
Suppose the following commands are in the Case Control Section:

LOADSET = 27
DLOAD = 25

and the following entries are in the Bulk Data Section:

1 2 3 4 5 6 7 8 9 10
$LSEQ SID DAREA LID

LSEQ 27 28 26

$RLOAD1 SID DAREA DELAY DPHASE TC TD

RLOAD1 25 28 29

DAREA
Reference
$STATIC LOAD SID
Link
PLOAD1 26 etc.

FORCE 26 etc.

$TABLED1 TID

TABLED1 29 etc.

In the above, the LOADSET request in Case Control selects the LSEQ Set ID 27 entry. The
DLOAD request in Case Control selects the RLOAD1 Set ID 25 entry. This RLOAD1 entry
refers to a TABLED1 ID 29, which is used to define the frequency-dependent variation in
the loading. DAREA Set ID 28 links the LSEQ and RLOAD1 entries. In addition, the LSEQ
entry refers to static Load Set ID 26, which is defined by FORCE and PLOAD1 entries. The
FORCE and PLOAD1 entries define the spatial distribution of the dynamic loading and
through the DAREA link refer to the RLOAD1/TABLED1 combination for the
frequency-varying characteristics of the load. Note that there is no DAREA entry.
126

Dynamic Load Set Combination – DLOAD

One of the requirements of frequency-dependent loads is that RLOAD1s and RLOAD2s
must have unique SIDs. If they are to be applied in the same analysis, they must be
combined using the DLOAD Bulk Data entry. The total applied load is constructed from a
linear combination of component load sets as follows:

{ P } = S ∑ Si { Pi } Eq. 5-28
i

where:

S = overall scale factor

S i = scale factor for the i-th load set

P i = i-th load set

P = total applied load

The DLOAD Bulk Data entry has the following format:

1 2 3 4 5 6 7 8 9 10
DLOAD SID S S1 L1 S2 L2

Field Contents
SID Load Set ID.
S Overall scale factor.
Si Individual scale factors.
Li Load Set ID numbers for RLOAD1 and RLOAD2 entries.

As an example, in the following DLOAD entry

$DLOAD SID S S1 L1 S2 L2 -etc.-

DLOAD 33 3.25 0.5 14 2.0 27

a dynamic Load Set ID of 33 is created by taking 0.5 times the loads in the Load Set ID of
14, adding to it 2.0 times the loads in the Load Set ID of 27, and multiplying that sum by an
overall scale factor of 3.25.
As with other frequency-dependent loads, a dynamic load combination defined by the
DLOAD Bulk Data entry is selected by the DLOAD Case Control command.
 CHAPTER 5 127
Frequency Response Analysis

5.6 Solution Frequencies

A major consideration when you conduct a frequency response analysis is selecting the
frequency at which the solution is to be performed. There are six Bulk Data entries that you
can use to select the solution frequencies. It is important to remember that each specified
frequency results in an independent solution at the specified excitation frequency.
To select the loading frequencies, use the FREQ, FREQ1, FREQ2, FREQ3, FREQ4 and
FREQ5 Bulk Data entries.

FREQ Defines descrete excitation frequencies.

FREQ1 Defines a starting frequency F start , a frequency incitement ∆f , and
the number of frequency increments to solve NDF.
FREQ2 Defines a starting frequency Fstart , and ending frequency Fend , and
the number of logarithmic intervals, NF, to be used in the
frequency range.
FREQ3* Defines the number of excitation frequencies used between modal
pairs in a given range.
FREQ4* Defines excitation frequencies using a spread about each normal
mode within a range.
FREQ5* Defines excitation frequencies as all frequencies in a given range as
a defined fraction of the normal modes.

* Used for modal solution only.

The FREQUENCY Case Control command selects FREQi Bulk Data entries. All FREQi
entries with the same selected Set ID are applied in the analysis; therefore, you can use any
combination of FREQ, FREQ1, FREQ2, FREQ3, FREQ4 and FREQ5 entries.

The examples that follow show the formats of the FREQi entries. Notice that the six sets of
excitation frequencies shown in the examples will be combined in a single analysis if the
Set IDs are identical.
128

FREQ
This FREQ entry specifies ten specific (unequally spaced loading frequencies to be
analyzed.

1 2 3 4 5 6 7 8 9 10
$FREQ SID F F F F F F F

$ F F F F F F F F

FREQ 3 2.98 3.05 17.9 21.3 25.6 28.8 31.2

29.2 22.4 19.3

Field Contents
SID Set ID specified by a FREQUENCY Case Control command.
F Frequency value (cycles per unit time).

FREQ1
The FREQ1 example specifies 14 frequencies between 2.9 Hz and 9.4 Hz in increments of
0.5 Hz.

$FREQ1 SID F start ∆f NDF

FREQ1 6 2.9 0.5 13

Field Contents
SID Set ID specified by a FREQUENCY Case Control command.
F start Starting frequency in set (cycles per unit time).

∆f Frequency increment (cycles per unit time).

NDF Number of frequency increments.

 CHAPTER 5 129
Frequency Response Analysis

FREQ2
The FREQ2 example specifies six logarithmic frequency intervals between 1.0 and 8.0 Hz,
resulting in frequencies at 1.0, 1.4142, 2.0, 2.8284, 4.0, 5.6569, and 8.0 Hz being used for the
analysis.

1 2 3 4 5 6 7 8 9 10
$FREQ2 SID F start F end NF

FREQ2 9 1.0 8.0 6

Field Contents
SID Set ID specified by a FREQUENCY Case Control command.
F start Starting frequency (cycles per unit time).

F end Ending frequency (cycles per unit time).

NF Number of logarithmic intervals.

130

FREQ3
The FREQ3 example requests 10 frequencies between each set of modes within the range
20 and 2000, plus ten frequencies between 20 and the lowest mode in the range, plus 10
frequencies between the highest mode in the range and 2000.

1 2 3 4 5 6 7 8 9 10
$FREQ3 SID F1 F2 TYPE NEF CLUSTER

FREQ3 6 20.0 2000.0 LINEAR 10 2.0

Field Contents
SID Set ID specified by a FREQUENCY Case Control command.
F1 Lower bound of modal frequency range in cycles per unit time.
(Real > 0.0)
F2 Upper bound of modal frequency range in cycles per unit time.
(Real > 0.0, F2 ≥ F1, Default = F1)
TYPE LINEAR or LOG. Specifies linear or logarithmic interpolation between
frequencies. (Character; Default = “LINEAR”)
NEF Number of excitation frequencies within each subrange including the
end points. (Integer > 1, Default = 10)
CLUSTER A CLUSTER value greater than 1 provides closer spacing of excitation
frequencies near the modal frequencies, where greater resolution is
needed. (Real > 0.0; Default = 1.0)
 CHAPTER 5 131
Frequency Response Analysis

FREQ4
The example FREQ4 chooses 21 equally spaced frequencies across a frequency band of
0.7 ⋅ f N to 1.3 ⋅ f N for each natural frequency between 20 and 2000.

1 2 3 4 5 6 7 8 9 10
$FREQ4 SID F1 F2 FSPD NFM

FREQ4 6 20.0 2000.0 0.30 21

Field Contents
SID Set ID specified by a FREQUENCY Case Control command.
F1 Lower bound of modal frequency range in cycles per unit time.
(Real > 0.0)
F2 Upper bound of modal frequency range in cycles per unit time.
(Real > 0.0, F2 ≥ F1, Default = F1)
FSPD Frequency spread, +/– the fractional amount specified for each mode
which occurs in the frequency range F1 to F2. (1.0 > Real > 0.0,
Default = 0.10)
NFM Number of evenly spaced frequencies per “spread” mode. (Integer > 0;
Default = 3; If NFM is even, NFM + 1 will be used.)

FREQ5
The example FREQ5 will compute excitation frequencies which are 0.6, 0.8, 0.9, 0.95, 1.0,
1.05, 1.1, and 1.2 times the natural frequencies for all natural frequencies, but use only the
computed frequencies that fall within the range 20 and 2000.

1 2 3 4 5 6 7 8 9 10
$FREQ5 SID F1 F2 FR1 FR2 FR3 FR4 FR5

FR6 FR7 -etc.-

FREQ5 6 20.0 2000.0 1.0 0.6 0.8 0.9 0.95

1.05 1.1 1.2

Field Contents
SID Set ID specified by a FREQUENCY Case Control command.
F1 Lower bound of modal frequency range in cycles per unit time.
(Real > 0.0)
F2 Upper bound of modal frequency range in cycles per unit time.
(Real > 0.0, F2 ≥ F1, Default = F1)
FRi Fractions of the natural frequencies in the range F1 to F2. (Real > 0.0)
132

5.7 Frequency Response Considerations

Exciting an undamped (or modal or viscous damped) system at 0.0 Hz using direct
frequency response analysis gives the same results as a static analysis and also gives almost
the same results when using modal frequency response (depending on the number of
retained modes). Therefore, if the maximum excitation frequency is much less than the
lowest resonant frequency of the system, a static analysis is probably sufficient.

Undamped or very lightly damped structures exhibit large dynamic responses for
excitation frequencies near resonant frequencies. A small change in the model (or running
it on another computer) may result in large changes in such responses.

Use a fine enough frequency step size ( ∆f ) to adequately predict peak response. Use at
least five points across the half-power bandwidth (which is approximately 2ζf n for an
SDOF system) as shown in Figure 5-7.

Peak Response

Peak
------------ = Half-Power Point
Response 2

Frequency

Half-Power Bandwidth
= Excitation Frequency
Figure 5-7 Half-Power Bandwidth

For maximum efficiency, an uneven frequency step size should be used. Smaller frequency
spacing should be used in regions near resonant frequencies, and larger frequency step
sizes should be used in regions away from resonant frequencies.
 CHAPTER 5 133
Frequency Response Analysis

5.8 Solution Control for Frequency Response Analysis

The following tables summarize the data entries that can be used to control a frequency
response analysis. Certain data entries are required, some data entries are optional, and
others are user selectable.

In the Executive Control Section of the MSC.Nastran input file, a solution must be selected
using the SOL i statement where i is an integer value chosen from Table 5-3.

Table 5-3 Frequency Response Solutions in MSC.Nastran

Structured
Rigid
Solution
Formats
Sequences
Direct 8 108
Modal 11 111

Solutions 108 and 111 are the preferred SOLs; these are the ones used in the examples that
follow.
In the Case Control Section of the MSC.Nastran input file, you must select the solution
parameters associated with the current analysis (i.e., frequencies, loads, and boundary
conditions), and also the output quantities required from the analysis. The Case Control
commands directly related to frequency response analysis are listed in Table 5-4. They can
be combined in the standard fashion with the more generic entries, such as SPC, MPC, etc.

Table 5-4 Case Control Commands for Frequency Response Solution Control

Case
Direct or
Control Description Required/Optional
Modal
Command
DLOAD Both Select the dynamic load set from Required
Bulk Data
FREQUENCY Both Select FREQi entries from Bulk Data Required
METHOD Modal Select the eigenvalue extraction Required
parameters
LOADSET Both Select the LSEQ set from Bulk Data Optional
SDAMPING Modal Select the modal damping table Optional
from Bulk Data
OFREQUENCY Both Select the frequencies for output Optional
(default = all)
134

The types of results available from a frequency response analysis are similar to those for a
typical static analysis except that the results are a complex function of the applied loading
frequency. Additional quantities (characteristic of dynamic problems) are also available.
The output quantities are summarized in Table 5-5 and Table 5-6.

Table 5-5 Grid Output from a Frequency Response Analysis

Case Control
Description
Command
ACCELERATION Grid point acceleration results for a set of grid points.
DISPLACEMENT Grid point displacement results for a set of grid points.
(or VECTOR)
OLOAD Requests the applied load table to be output for a set of grid
points.
SACCELERATION Requests the solution set acceleration output: d-set in direct
solutions and modal variables in modal solutions.
SDISPLACEMENT Requests the solution set displacement output: d-set in direct
solutions and modal variables in modal solutions.
SVECTOR Requests the real eigenvector output for the a-set in modal
solutions.
SVELOCITY Requests the solution set velocity output: d-set in direct solutions
and modal variables in modal solutions.
SPCFORCES Requests the forces of a single-point constraint for a set of grid
points.
VELOCITY Grid point velocity results for a set of grid points.

Frequency response output is in real/imaginary format (the default) or magnitude/phase

format (the phase angle is in degrees). Frequency response output is also in SORT1 or
SORT2 format. In SORT1 format, the results are listed by frequency; in SORT2 format, the
results are listed by grid point or element number. SORT1 is the default for direct
frequency response analysis (SOL 108), and SORT2 is the default for modal frequency
response analysis (SOL 111). PARAM,CURVPLOT,1 and PARAM,DDRMM,-1 are
necessary to obtain SORT1 output in SOL 111. These output formats are specified with the
Case Control commands. The command
DISPLACEMENT(PHASE, SORT2) = ALL

prints displacements in magnitude/phase and SORT2 formats. The output formats are
illustrated in the first example in “Examples” on page 136.
 CHAPTER 5 135
Frequency Response Analysis

Table 5-6 Element Output from a Frequency Response Analysis

Case Control Command Description

ELSTRESS (or STRESS) Element stress results for a set of elements.
ELFORCE (or FORCE) Element force results for a set of elements.
STRAIN Element strain results for a set of elements.

A number of Bulk Data entries are unique to frequency response analysis. These entries
can be combined with other generic entries in the Bulk Data. Bulk Data entries directly
related to frequency response analysis are summarized in Table 5-7.

Table 5-7 Bulk Data Entries for Frequency Response Analysis

Bulk Data Direct Required/

Description
Entry or Modal Optional
FREQ, FREQi Both Excitation frequencies Required
RLOADi Both Dynamic loading Required
EIGR or EIGRL Modal Eigenvalue analysis parameters Required
LSEQ Both Dynamic loading from static loads Optional
TABLEDi Both Frequency-dependent tables Both*
DAREA Both Load component and scale factor Required†
DELAY Both Time delay on dynamic load Optional
DPHASE Both Phase angle on dynamic load Optional
DLOAD Both Dynamic load combination, Optional
required if RLOAD1 and RLOAD2
are used
TABDMP1 Modal Modal damping table Optional

* Required for RLOAD1; optional for RLOAD2.

†The DAREA ID is required; the DAREA Bulk Data entry is not required if an
LSEQ entry is used.
136

5.9 Examples
This section provides several examples showing the input and output. These examples are:

Frequency Response Bulk

Model Output
Data Entries
bd05two EIGRL, FREQ1, TABDMP1, X-Y plots (linear), printed
RLOAD1, DAREA, TABLED1 results (SORT1, SORT2)
bd05bar EIGRL, FREQ1, TABDMP1, X-Y plots (log)
DLOAD, RLOAD2, DAREA,
DPHASE, TABLED1
bd05bkt EIGRL, FREQ1, TABDMP1, X-Y plot (log)
RLOAD1, LSEQ, TABLED1,
PLOAD4

These examples are described in the sections that follow.

Two-DOF Model
Consider the two-DOF system shown in Figure 5-8. Modal frequency response (SOL 111)
is run with a 20 N load applied to the primary mass (grid point 2) across a frequency range
of 2 to 10 Hz with an excitation frequency increment of 0.05 Hz. Uniform modal damping
of 5% critical damping is used. Listing 5-1 shows part of the input file.

Auxiliary Structure,
Grid Point 1
P = 20 N sin ωt

Primary Structure,
Grid Point 2

x
Figure 5-8 Two-DOF Model
 CHAPTER 5 137
Frequency Response Analysis

Listing 5-1 Input File (Abridged) for the Two-DOF Example

$ FILE bd05two.dat
$
$ TWO-DOF SYSTEM
$ CHAPTER 5, FREQUENCY RESPONSE
$
TIME 5
SOL 111 $ MODAL FREQUENCY RESPONSE
CEND
TITLE = TWO-DOF SYSTEM
SUBTITLE = MODAL FREQUENCY RESPONSE
LABEL = 20 N FORCE APPLIED TO PRIMARY MASS
$
$ SPECIFY SPC
SPC = 996
$
$ SPECIFY MODAL EXTRACTION
METHOD = 10
$
$ SPECIFY DYNAMIC INPUT
DLOAD = 999
FREQ = 888
SDAMPING = 777
$
$ SELECT OUTPUT
SET 11 = 1,2
DISPLACEMENT(PHASE,PLOT) = 11
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ ENTRIES FOR FREQUENCY RESPONSE
$
$ LOAD DEFINITION
$RLOAD1 SID DAREA TC
RLOAD1 999 997 901
$DAREA SID P1 C1 A1
DAREA 997 2 2 20.0
$TABLED1 TID +TABL1
$+TABL1 X1 Y1 X2 Y3 ETC.
TABLED1 901 +TAB901
+TAB901 0.0 1.0 10.0 1.0 ENDT
$
$ ALTERNATE LOAD DEFINITION USING DLOAD
$DLOAD SID S S1 RLOAD1
$DLOAD 999 1.0 1.0 998
$RLOAD1 SID DAREA TC
$RLOAD1 998 997 901
$
$ FREQUENCY RANGE 2-10 HZ
$FREQ1 SID F1 DF NDF
FREQ1 888 2. 0.05 160
$
$ MODAL DAMPING OF 5% CRITICAL
$TABDMP1 TID TYPE +TABD1
$+TABD1 F1 G1 F2 G2 ETC.
TABDMP1 777 CRIT +TABD7
+TABD7 0. 0.05 100. 0.05 ENDT
$
$ MODAL EXTRACTION
$EIGRL SID V1 V2 ND MSGLVL
EIGRL 10 -0.1 20. 0
$
... basic model ...
$
ENDDATA
138

Table 5-8 shows the relationship between the Case Control commands and the Bulk Data
entries. Note that the RLOAD1 entry references the DAREA and TABLED1 entries. The
input file also shows an alternate way to specify the dynamic load, by using a DLOAD Bulk
Data entry. Because there is only a single RLOAD1 entry, the DLOAD Bulk Data entry is
not required.

Table 5-8 Relationship Between the Case Control Commands and Bulk Data Entries
for the Two-DOF Model

Case Control Bulk Data

METHOD EIGRL
FREQUENCY FREQ1
SDAMPING TABDMP1
DLOAD RLOAD1

DAREA
TABLED1

The RLOAD1 entry describes a sinusoidal load in the form

i { θ – 2πfτ }
{ P ( f ) } = { A [ C ( f ) + iD ( f ) ]e } Eq. 5-29

where:
A = 20.0 (entered on the DAREA entry)
C = 1.0 for all frequencies entered on the TABLED1 entry
D = 0.0 (field 7 of the RLOAD1 entry is blank)
θ = 0.0 (field 5 of the RLOAD1 entry is blank)
τ = 0.0 (field 4 of the RLOAD1 entry is blank)
 CHAPTER 5 139
Frequency Response Analysis

Output can be printed in either real/imaginary or magnitude/phase format and in either

SORT1 or SORT2 format. These formats are illustrated in Figure 5-9, Figure 5-10, and
Figure 5-11 showing a portion of their printed output.

POINT-ID = 1
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

FREQUENCY TYPE T1 T2 T3 R1 R2 R3
2.000000E+00 G 0.0 2.813052E-03 0.0 0.0 0.0 0.0
0.0 -2.107985E-04 0.0 0.0 0.0 0.0
2.050000E+00 G 0.0 2.866642E-03 0.0 0.0 0.0 0.0
0.0 -2.229164E-04 0.0 0.0 0.0 0.0
2.100000E+00 G 0.0 2.923141E-03 0.0 0.0 0.0 0.0
0.0 -2.358382E-04 0.0 0.0 0.0 0.0
2.150000E+00 G 0.0 2.982732E-03 0.0 0.0 0.0 0.0
0.0 -2.496362E-04 0.0 0.0 0.0 0.0
2.200000E+00 G 0.0 3.045609E-03 0.0 0.0 0.0 0.0
0.0 -2.643908E-04 0.0 0.0 0.0 0.0

POINT-ID = 2
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

FREQUENCY TYPE T1 T2 T3 R1 R2 R3
2.000000E+00 G 0.0 2.374954E-03 0.0 0.0 0.0 0.0
0.0 -1.129933E-04 0.0 0.0 0.0 0.0
2.050000E+00 G 0.0 2.397706E-03 0.0 0.0 0.0 0.0
0.0 -1.180853E-04 0.0 0.0 0.0 0.0
2.100000E+00 G 0.0 2.421475E-03 0.0 0.0 0.0 0.0
0.0 -1.234173E-04 0.0 0.0 0.0 0.0
2.150000E+00 G 0.0 2.446311E-03 0.0 0.0 0.0 0.0
0.0 -1.290072E-04 0.0 0.0 0.0 0.0
2.200000E+00 G 0.0 2.472262E-03 0.0 0.0 0.0 0.0
0.0 -1.348744E-04 0.0 0.0 0.0 0.0

Figure 5-9 Real/Imaginary Output in SORT2 Format

140

POINT-ID = 1
C O M P L E X D I S P L A C E M E N T V E C T O R
(MAGNITUDE/PHASE)

FREQUENCY TYPE T1 T2 T3 R1 R2 R3
2.000000E+00 G 0.0 2.820939E-03 0.0 0.0 0.0 0.0
0.0 355.7145 0.0 0.0 0.0 0.0
2.050000E+00 G 0.0 2.875296E-03 0.0 0.0 0.0 0.0
0.0 355.5535 0.0 0.0 0.0 0.0
2.100000E+00 G 0.0 2.932640E-03 0.0 0.0 0.0 0.0
0.0 355.3874 0.0 0.0 0.0 0.0
2.150000E+00 G 0.0 2.993161E-03 0.0 0.0 0.0 0.0
0.0 355.2159 0.0 0.0 0.0 0.0
2.200000E+00 G 0.0 3.057064E-03 0.0 0.0 0.0 0.0
0.0 355.0386 0.0 0.0 0.0 0.0

POINT-ID = 2
C O M P L E X D I S P L A C E M E N T V E C T O R
(MAGNITUDE/PHASE)

FREQUENCY TYPE T1 T2 T3 R1 R2 R3
2.000000E+00 G 0.0 2.377640E-03 0.0 0.0 0.0 0.0
0.0 357.2761 0.0 0.0 0.0 0.0
2.050000E+00 G 0.0 2.400612E-03 0.0 0.0 0.0 0.0
0.0 357.1805 0.0 0.0 0.0 0.0
2.100000E+00 G 0.0 2.424619E-03 0.0 0.0 0.0 0.0
0.0 357.0823 0.0 0.0 0.0 0.0
2.150000E+00 G 0.0 2.449710E-03 0.0 0.0 0.0 0.0
0.0 356.9813 0.0 0.0 0.0 0.0
2.200000E+00 G 0.0 2.475939E-03 0.0 0.0 0.0 0.0
0.0 356.8773 0.0 0.0 0.0 0.0

Figure 5-10 Magnitude/Phase Output in SORT2 Format

 CHAPTER 5 141
Frequency Response Analysis

FREQUENCY = 2.000000E+00
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 2.813051E-03 0.0 0.0 0.0 0.0
0.0 -2.107985E-04 0.0 0.0 0.0 0.0
2 G 0.0 2.374954E-03 0.0 0.0 0.0 0.0
0.0 -1.129933E-04 0.0 0.0 0.0 0.00

FREQUENCY = 2.050000E+00
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 2.866640E-03 0.0 0.0 0.0 0.0
0.0 -2.229163E-04 0.0 0.0 0.0 0.0
2 G 0.0 2.397706E-03 0.0 0.0 0.0 0.0
0.0 -1.180853E-04 0.0 0.0 0.0 0.00

FREQUENCY = 2.100000E+00
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 2.923141E-03 0.0 0.0 0.0 0.0
0.0 -2.358381E-04 0.0 0.0 0.0 0.0
2 G 0.0 2.421475E-03 0.0 0.0 0.0 0.0
0.0 -1.234173E-04 0.0 0.0 0.0 0.00

FREQUENCY = 2.150000E+00
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 2.982731E-03 0.0 0.0 0.0 0.0
0.0 -2.496362E-04 0.0 0.0 0.0 0.0
2 G 0.0 2.446311E-03 0.0 0.0 0.0 0.0
0.0 -1.290072E-04 0.0 0.0 0.0 0.00

FREQUENCY = 2.200000E+00
C O M P L E X D I S P L A C E M E N T V E C T O R
(REAL/IMAGINARY)

POINT ID. TYPE T1 T2 T3 R1 R2 R3

1 G 0.0 3.045608E-03 0.0 0.0 0.0 0.0
0.0 -2.643907E-04 0.0 0.0 0.0 0.0
2 G 0.0 2.472263E-03 0.0 0.0 0.0 0.0
0.0 -1.348744E-04 0.0 0.0 0.0 0.0

Figure 5-11 Real/Imaginary Output in SORT1 Format

Figure 5-12 shows the plots of the the resulting displacement magnitudes for grid points 1
and 2. Note that the response for grid point 1 is nearly an order of magnitude larger than
that of grid point 2. This large difference in response magnitudes is characteristic of
dynamic absorbers (also called tuned mass dampers), in which an auxiliary structure (i.e.,
the small mass and stiffness) is attached to the primary structure in order to decrease the
dynamic response of the primary structure. If this same model is rerun without the
auxiliary structure, the response of the primary structure (grid point 2) at 5.03 Hz is twice
what it was with the auxiliary structure attached, as shown in Figure 5-13.
142

0.12

Displacement 1 (m)
0

0.014
Displacement 2 (m)

0
2 Frequency (Hz) 10
Figure 5-12 Displacement Response Magnitudes With the Auxiliary Structure

0.02
Displacement 2 (m)

0
2 Frequency (Hz) 10
Figure 5-13 Displacement Response Magnitude Without the Auxiliary Structure

Cantilever Beam Model

Consider the cantilever beam shown in Figure 5-14. This model is a planar model of the
cantilever beam introduced in “Real Eigenvalue Analysis” on page 35 with unrestrained
DOFs in the T2 and R3 directions. Two loads are applied: one at grid point 6 and the other
at grid point 11. The loads have the frequency variation shown in Figure 5-15. The loads
in the figure are indicated with a heavy line in order to emphasize their values. The load
at grid point 6 has a 45-degree phase lead, and the load at grid point 11 is scaled to be twice
that of the load at grid point 6. Modal frequency response is run across a frequency range
 CHAPTER 5 143
Frequency Response Analysis

of 0 to 20 Hz. Modal damping is used with 2% critical damping between 0 and 10 Hz and
5% critical damping above 10 Hz. Modes to 500 Hz are computed using the Lanczos
method.

6 11
x

p(f) 2p(f)
Figure 5-14 Cantilever Beam Model with Applied Loads
144

4.0

Load Magnitude 6 (N)

3.0

0
100
Load Phase 6 (deg)

45

0
Load Magnitude 11 (N)

8.0

6.0

0
100
Load Phase 11 (deg)

0
0 Frequency (Hz) 20
Figure 5-15 Applied Loads

Listing 5-2 shows the abridged input file. The output quantities, as defined in the Case
Control Section, are the applied loads (OLOAD) for grid points 6 and 11, physical
displacements (DISPLACEMENT) for grid points 6 and 11, solution set displacements
(SDISPLACEMENT) for modes 1 and 2, and element forces (ELFORCE) for element 6.
These output quantities are plotted rather than printed.
 CHAPTER 5 145
Frequency Response Analysis

Listing 5-2 Input File (Abridged) for the Beam Example

$ FILE bd05bar.dat
$
$ CANTILEVER BEAM MODEL
$ CHAPTER 5, FREQUENCY RESPONSE
$
SOL 111 $ MODAL FREQUENCY RESPONSE
TIME 10
CEND
TITLE = CANTILEVER BEAM
SUBTITLE = MODAL FREQUENCY RESPONSE
$
SPC = 21
$
DLOAD = 22
FREQ = 27
SDAMPING = 20
$
METHOD = 10
$
SET 15 = 6,11
OLOAD(PHASE,PLOT) = 15
$
$ PHYSICAL OUTPUT REQUEST
SET 11 = 6,11
DISPLACEMENT(PHASE,PLOT) = 11
$
$ MODAL SOLUTION SET OUTPUT
SET 12 = 1,2
SDISP(PHASE,PLOT) = 12
$
$ ELEMENT FORCE OUTPUT
SET 13 = 6
ELFORCE(PHASE,PLOT) = 13
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
EIGRL 10 -0.1 500. 0
$
FREQ1 27 0.0 0.05 400
$
TABDMP1 20 CRIT +TABD1
+TABD1 0.0 0.02 10.0 0.02 10.01 0.05 25.0 0.05 +TABD2
+TABD2 ENDT
$
$ DYNAMIC LOADING
$DLOAD SID S S1 L1 S2 L2
DLOAD 22 1.0 1.0 231 1.0 232
$RLOAD2 SID DAREA DELAY DPHASE TB TP
RLOAD2 231 241 261 25
RLOAD2 232 242 25
$DAREA SID P1 C1 A1
DAREA 241 6 2 1.0
DAREA 242 11 2 2.0
$DPHASE SID P1 C1 TH1
DPHASE 261 6 2 45.
$TABLED1 TID +TABL1
$+TABL1 X1 Y1 X2 Y2 ETC.
TABLED1 25 +TABL1
+TABL1 0. 1. 5.0 3. 15.0 3.0 20.0 1. +TABL2
+TABL2 25.0 1. ENDT
$
... basic model ...
$
ENDDATA
146

Table 5-9 shows the relationship between the Case Control commands and the Bulk Data
entries. Note that the DLOAD Bulk Data entry references two RLOAD2 entries, each of
which references a separate DAREA entry and a common TABLED1 entry. The RLOAD2
entry for grid point 6 also references a DPHASE entry that defines the 45-degree phase
lead.

Table 5-9 Relationship Between Case Control Commands and Bulk Data Entries for
the Beam Model

Case Control Bulk Data

METHOD EIGRL
FREQUENCY FREQ1
SDAMPING TABDMP1
DLOAD DLOAD

RLOAD2 231

DAREA
DPHASE
TABLED1

RLOAD2 232
DAREA

The RLOAD2 entry describes a sinusoidal load in the form

i { φ ( f ) + θ – 2πfτ }
{ P ( f ) } = { AB ( f )e } Eq. 5-30

where:

A = 1.0 for grid point 6 and 2.0 for grid point 11 (entered on the DAREA entry)
B = function defined on the TABLED1 entry
φ = 0.0 (field 7 of the RLOAD2 entry is blank)
θ = phase lead of 45 degrees for grid point 6 (entered on the DPHASE entry)
τ = 0.0 (field 4 of the RLOAD2 entry is blank)

Logarithmic plots of the output are shown in the following figures. Figure 5-16 shows the
magnitude of the displacements for grid points 6 and 11. Figure 5-17 shows the magnitude
of the modal displacements for modes 1 and 2. Figure 5-18 shows the magnitude of the
bending moment at end A in plane 1 for element 6. Logarithmic plots are especially useful
for displaying frequency response results since there can be several orders of magnitude
between the maximum and minimum response values.
 CHAPTER 5 147
Frequency Response Analysis

1.0

Displacement 6 (m)
1.0E-5

1.0
Displacement 11 (m)

1.0E-4
0 Frequency (Hz) 20

Figure 5-16 Displacement Magnitude (Log)

1.0
SDisplacement 1

1.0E-4
1.0E-2
SDisplacement 2

1.0E-4

0 Frequency (Hz) 20
Figure 5-17 Modal Displacement Magnitude (Log)
148

1000

Moment A1 (N-m)
0.1
0 Frequency (Hz) 20

Figure 5-18 Bending Moment Magnitude at End A, Plane 1 (Log)

Bracket Model
Consider the bracket model shown in Figure 5-19. An oscillating pressure load of 3 psi is
applied to the elements on the top face in the z-direction. The model is constrained at its
base. Modal frequency response is run from 0 to 100 Hz with a frequency step size of 0.2
Hz. Eigenvalues to 1000 Hz are computed using the Lanczos method. Modal damping is
applied as 2% critical damping for all modes.

Figure 5-19 Bracket Model

Listing 5-3 shows the abridged input file. The LSEQ entry is used to apply the pressure
loads (PLOAD4 entries). Note that the LSEQ and RLOAD1 entries reference a common
DAREA ID (999) and that there is no explicit DAREA entry. Table 5-10 shows the
relationship between the Case Control commands and the Bulk Data entries.
 CHAPTER 5 149
Frequency Response Analysis

Listing 5-3 Input File (Abridged) for the Bracket Model

$ FILE bd05bkt.dat
$
$ BRACKET MODEL
$ CHAPTER 5, FREQUENCY RESPONSE
$
SOL 111 $ MODAL FREQUENCY RESPONSE
TIME 100
CEND
TITLE = BRACKET MODEL
SUBTITLE = MODAL FREQUENCY RESPONSE ANALYSIS
$
SPC = 1
$
METHOD = 777
DLOAD = 2
LOADSET = 3
SDAMPING = 4
FREQUENCY = 5
$
$ OUTPUT REQUEST
SET 123 = 999
DISPLACEMENT(PHASE,PLOT)=123
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10......
$
$ NORMAL MODES TO 1000 HZ
$EIGRL SID V1 V2
EIGRL 777 -0.1 1000.

$
$ EXCITATION FREQUENCY DEFINITION 0 TO 100 HZ
$FREQ1 SID F1 DF NDF
FREQ1 5 0.0 0.2 500
$
$ MODAL DAMPING OF 2% CRITICAL FOR ALL MODES
$TABDMP1 TID TYPE +TABD1
$+TABD1 F1 G1 F2 G2 ETC.
TABDMP1 4 CRIT +TABD1
+TABD1 0.0 0.02 1000.0 0.02 ENDT
$
$ LOAD DEFINITION
$
$RLOAD1 SID DAREA DELAY DPHASE TC TD
RLOAD1 2 999 22
$
$LSEQ SID DAREA LID TID
LSEQ 3 999 1
$
$TABLED1 TID +TABL1
$+TABL1 X1 Y1 X2 Y2 ETC.
TABLED1 22 +TABL1
+TABL1 0.0 1.0 1000.0 1.0 ENDT
$
$ PRESURE LOAD OF 3 PSI PER ELEMENT
$PLOAD4 SID EID P1
PLOAD4 1 171 -3.
PLOAD4 1 172 -3.
PLOAD4 1 160 -3.
etc.
$
... basic model ...
$
ENDDATA
150

Table 5-10 Relationship Between Case Control Commands and

Bulk Data Entries for the Bracket Model

Case Control Bulk Data

METHOD EIGRL
FREQUENCY FREQ1
SDAMPING TABDMP1
PLOAD4
LOADSET LSEQ

(DAREA ID)

DLOAD RLOAD1
TABLED1

Figure 5-20 shows a logarithmic plot of the z-displacement magnitude of grid point 999,
which is the concentrated mass at the center of the cutout.

10

1.000
Displacement (m)

0.100

0.010

0.001

0.0001
0 Frequency (Hz) 100
Figure 5-20 Displacement Magnitude (Log)
 NX Nastran Basic Dynamic Analysis User’s Guide

6
CHAPTER
Transient Response Analysis

■ Overview
■ Direct Transient Response Analysis

■ Modal Transient Response Analysis

■ Modal Versus Direct Transient Response
■ Transient Excitation Definition
■ Integration Time Step
■ Transient Excitation Considerations

■ Solution Control for Transient Response Analysis

■ Examples
152

6.1 Overview
Transient response analysis is the most general method for computing forced dynamic
response. The purpose of a transient response analysis is to compute the behavior of a
structure subjected to time-varying excitation. The transient excitation is explicitly defined
in the time domain. All of the forces applied to the structure are known at each instant in
time. Forces can be in the form of applied forces and/or enforced motions (see “Enforced
Motion” on page 199).

The important results obtained from a transient analysis are typically displacements,
velocities, and accelerations of grid points, and forces and stresses in elements.

Depending upon the structure and the nature of the loading, two different numerical
methods can be used for a transient response analysis: direct and modal. The direct
method performs a numerical integration on the complete coupled equations of motion.
The modal method utilizes the mode shapes of the structure to reduce and uncouple the
equations of motion (when modal or no damping is used); the solution is then obtained
through the summation of the individual modal responses. The choice of the approach is
problem dependent. The two methods are described in “Direct Transient Response
Analysis” on page 153 and “Modal Transient Response Analysis” on page 158.
 CHAPTER 6 153
Transient Response Analysis

6.2 Direct Transient Response Analysis

In direct transient response, structural response is computed by solving a set of coupled
equations using direct numerical integration. Begin with the dynamic equation of motion
in matrix form

[ M ] { u·· ( t ) } + [ B ] { u· ( t ) } + [ K ] { u ( t ) } = { P ( t ) } Eq. 6-1

The fundamental structural response (displacement) is solved at discrete times, typically

with a fixed integration time step ∆t .

By using a central finite difference representation for the velocity { u· ( t ) } and the
acceleration { u·· ( t ) } at discrete times,

1
{ u· n } = --------- { u n + 1 – u n – 1 }
2∆t

1
{ u·· n } = -------- { u n + 1 – 2u n + u }
2 n–1
∆t Eq. 6-2

and averaging the applied force over three adjacent time points, the equation of motion can
be rewritten as:

M B
-------- ( u n + 1 – 2u n + u n – 1) + --------- ( u n + 1 – u n – 1 ) Eq. 6-3
∆t
2 2∆t
K 1
+ ---- ( u n + 1 + u n + u n – 1) = --- ( P n + 1 + P n + P n – 1 )
3 3

Collecting terms, the equation of motion can be rewritten as:

[ A1 ] { un + 1 } = [ A2 ] + [ A3 ] { un } + [ A4 ] { un – 1 } Eq. 6-4

where:

M B K
A1 = -------2- + --------- + ----
2∆t 3
∆t

1
[ A 2 ] = --3- { Pn + 1 + Pn + P n – 1 }

2M K
[ A3 ] = -------- – ----
2 3
∆t

M B K
[ A4 ] = – -------- + --------- – ----
2 2∆t 3
∆t

Matrix [ A 1 ] is termed the dynamic matrix, and [ A 2 ] is the applied force (averaged over
three adjacent time points). This approach is similar to the classical Newmark-Beta direct
integration method except that { P ( t ) } is averaged over three time points and [ K ] is
modified such that the dynamic equation of motion reduces to a static solution
[ K ] { u n } = { P n } if no [ M ] or [ B ] exists.
154

The transient solution is obtained by decomposing [ A1 ] and applying it to the right-hand

side of the above equation. In this form, the solution behaves like a succession of static
solutions with each time step performing a forward-backward substitution (FBS) on a new
load vector. Note that the transient nature of the solution is carried through by modifying
the applied force matrix [ A 2 ] with the [ A3 ] and [ A 4 ] terms.

In its simplest form, the [ M ] , [ B ] , and [ K ] matrices are assumed to be constant throughout
the analysis and do not change with time. Special solution methods are available in
MSC.Nastran for variations in these matrices (see the NX Nastran Advanced Dynamic
Analysis User’s Guide).

A significant benefit presents itself if ∆t remains constant during the analysis. With a
constant ∆t , the [ A 1 ] matrix needs to be decomposed only once. Each progressive step in
the analysis is only an FBS of a new load vector. If ∆t is changed, [ A 1 ] must be
redecomposed, which can be a costly operation in large problems.

Another efficiency in the direct transient solution is that the output time interval may be
greater than the solution time interval. In many cases it is not necessary to sample output
response at each solution time. For example, if the solution is performed every 0.001
second the results can be output every fifth time step or every 0.005 second. This efficiency
reduces the amount of output.

Damping in Direct Transient Response

The damping matrix [ B ] is used to represent the energy dissipation characteristics of a
structure. In the general case, the damping matrix is comprised of several matrices

1 2 G 1
[ B ] = [ B ] + [ B ] + ------- [ K ] + ------- ∑ G E [ K E ] Eq. 6-5
W3 W4

where:

1
[ B ] = damping elements (CVISC, CDAMPi) + B2GG
2
[ B ] = B2PP direct input matrix + transfer functions
G = overall structural damping coefficient (PARAM,G)
W 3 = frequency of interest in radians per unit time (PARAM,W3) for the conversion of
overall structural damping into equivalent viscous damping
[ K ] = global stiffness matrix
G E = element structural damping coefficient (GE on the MATi entry)

W 4 = frequency of interest in radians per unit time (PARAM,W4) for conversion of

element structural damping into equivalent viscous damping
[ K E ] = element stiffness matrix
 CHAPTER 6 155
Transient Response Analysis

Transient response analysis does not permit the use of complex coefficients. Therefore,
structural damping is included by means of equivalent viscous damping. To appreciate
the impact of this on the solution, a relation between structural damping and equivalent
viscous damping must be defined.
The viscous damping force is a damping force that is a function of a damping coefficient b
and the velocity. It is an induced force that is represented in the equation of motion using
the [B] matrix and velocity vector.

[ M ] { u·· ( t ) } + [ B ] { u· ( t ) } + [ K ] { u ( t ) } = { P ( t ) } Eq. 6-6

The structural damping force is a displacement-dependent damping. The structural

damping force is a function of a damping coefficient G and a complex component of the
structural stiffness matrix.

[ M ] { u·· ( t ) } + ( 1 + iG ) [ K ] { u ( t ) } = { P ( t ) } Eq. 6-7

Assuming constant amplitude oscillatory response for an SDOF system, the two damping
forces are identical if

Gk = bω Eq. 6-8

or

Gk
b = ------- Eq. 6-9
ω

Therefore, if structural damping G is to be modeled using equivalent viscous damping b ,

then the equality Eq. 6-9 holds at only one frequency (see Figure 6-1).

Two parameters are used to convert structural damping to equivalent viscous damping.
An overall structural damping coefficient can be applied to the entire system stiffness
matrix using PARAM,W3,r where r is the circular frequency at which damping is to be
made equivalent. This parameter is used in conjunction with PARAM,G. The default
value for W3 is 0.0, which causes the damping related to this source to be ignored in
transient analysis.
PARAM,W4 is an alternate parameter used to convert element structural damping to
equivalent viscous damping. PARAM,W4,r is used where r is the circular frequency at
which damping is to be made equivalent. PARAM,W4 is used in conjunction with the GE
field on the MATi entry. The default value for W4 is 0.0 which causes the related damping
terms to be ignored in transient analysis.

Units for PARAM,W3 and PARAM,W4 are radians per unit time. The choice of W3 or W4
is typically the dominant frequency at which the damping is active. Often, the first natural
frequency is chosen, but isolated individual element damping can occur at different
frequencies and can be handled by the appropriate data entries.
156

Damping Structural Damping fs = iGku

Force
Equivalent b = G k ⁄ ω 3 (or ω 4 )
Viscous
Damping

ω 3 (or ω 4 ) ω
·
f v = bu = ib ω u

Figure 6-1 Structural Damping Versus Viscous Damping

(Constant Oscillatory Displacement)

Initial Conditions in Direct Transient Response

You may impose initial displacements and/or velocities in direct transient response. The
TIC Bulk Data entry is used to define initial conditions on the components of grid points.
The IC Case Control command is used to select TIC entries from the Bulk Data.
If initial conditions are used, initial conditions should be specified for all DOFs having
nonzero values. Initial conditions for any unspecified DOFs are set to zero.
Initial conditions { u 0 } and { u· 0 } are used to determine the values of { u –1 } , { P 0 } , and { P –1 }
used in Eq. 6-4 to calculate { u 1 } .

{ u – 1 } = { u 0 } – { u· 0 }∆t Eq. 6-10

{ P – 1 } = [ K ] { u – 1 } + [ B ] { u· 0 } Eq. 6-11

In the presence of initial conditions, the applied load specified at t = 0 is replaced by

{ P 0 } = [ K ] { u 0 } + [ B ] { u· 0 } Eq. 6-12

Regardless of the initial conditions specified, the initial acceleration for all points in the
structure is assumed to be zero (constant initial velocity).
The format for the TIC entry is

1 2 3 4 5 6 7 8 9 10
TIC SID G C U0 V0

Field Contents
SID Set ID specified by the IC Case Control command.
G Grid, scalar, or extra point.
C Component number.
 CHAPTER 6 157
Transient Response Analysis

Field Contents
U0 Initial displacement.
V0 Initial velocity.

Initial conditions may be specified only in direct transient response. In modal transient
response all initial conditions are set to zero. Initial conditions may be specified only in the
a-set (see “Advanced Dynamic Analysis Capabilities” on page 293).
158

6.3 Modal Transient Response Analysis

Modal transient response is an alternate approach to computing the transient response of
a structure. This method uses the mode shapes of the structure to reduce the size, uncouple
the equations of motion (when modal or no damping is used), and make the numerical
integration more efficient. Since the mode shapes are typically computed as part of the
characterization of the structure, modal transient response is a natural extension of a
normal modes analysis.

As a first step in the formulation, transform the variables from physical coordinates { u } to
modal coordinates { ξ } by

{ u( t)} = [ φ ]{ ξ(t )} Eq. 6-13

The mode shapes [ φ ] are used to transform the problem in terms of the behavior of the
modes as opposed to the behavior of the grid points. Eq. 6-13 represents an equality if all
modes are used; however, because all modes are rarely used, the equation usually
represents an approximation.

To proceed, temporarily ignore the damping, resulting in the equation of motion

[ M ] { u·· ( t ) } + [ K ] { u ( t ) } = { P ( t ) } Eq. 6-14

If the physical coordinates in terms of the modal coordinates (Eq. 6-13 is substituted into
Eq. 6-14), the following equation is obtained:

··
[M ][ φ]{ ξ(t )} + [K ]{ φ}{ ξ( t)} = {P(t) } Eq. 6-15

This is now the equation of motion in terms of the modal coordinates. At this point,
however, the equations remain coupled.
To uncouple the equations, premultiply by [ φ ]T to obtain

T ·· T T
[φ ] [M ][ φ]{ ξ } + [φ ] [K ][ φ]{ ξ } = [φ ] { P( t) } Eq. 6-16

where:

T
[ φ ] [ M ] [ φ ] = modal (generalized) mass matrix
T
[ φ ] [ K ] [ φ ] = modal (generalized) stiffness matrix
T
[ φ ] { P } = modal force vector

The final step uses the orthogonality property of the mode shapes to formulate the
equation of motion in terms of the generalized mass and stiffness matrices that are
diagonal matrices. These matrices do not have off-diagonal terms that couple the
equations of motion. Therefore, in this form, the modal equations of motion are
uncoupled. In this uncoupled form, the equations of motion are written as a set on
uncoupled SDOF systems as
 CHAPTER 6 159
Transient Response Analysis

··
mi ξi ( t ) + ki ξi ( t ) = pi ( t ) Eq. 6-17

where:

m i = i-th modal mass

k i = i-th modal stiffness

p i = i-th modal force

Note that there is no damping in the resulting equation. The next subsection describes how
to include damping in modal transient response.

Once the individual modal responses ξ i ( t ) are computed, physical responses are
recovered as the summation of the modal responses

{ u( t)} = [ φ ]{ ξ(t )} Eq. 6-18

Since numerical integration is applied to the relatively small number of uncoupled

equations, there is not as large a computational penalty for changing ∆t as there is in direct
transient response analysis. However, a constant ∆t is still recommended.
Another efficiency option in the modal transient solution is that the output time interval
may be greater than the solution time interval. In many cases, it is not necessary to sample
output response at each solution time. For example, if the solution is performed every
0.001 second, the results can be output every fifth time step or every 0.005 second. This
efficiency reduces the amount of output.

Damping in Modal Transient Response Analysis

If the damping matrix [ B ] exists, the orthogonality property (see “Overview of Normal
Modes Analysis” on page 39) of the modes does not, in general, diagonalize the
generalized damping matrix

T
[ φ ] [ B ] [ φ ] ≠ diagonal Eq. 6-19

In the presence of a [ B ] matrix, the modal transient approach solves the coupled problem
in terms of modal coordinates using the direct transient numerical integration approach
described in Section 4.2 as follows:

[ A1 ] { ξn + 1 } = [ A2 ] + [ A3 ] { ξn } + [ A4 ] { ξn – 1 } Eq. 6-20

where:

[ A 1 ] = [ φ T ] -------
M B K
- + --------- + ---- [ φ ]
2
∆t 2∆t 3

[ A 2 ] = 1--- [ φ ] T { P
3 n + 1 + Pn + Pn – 1 }
160

[ A 3 ] = [ φ T ] 2M K
-------- – ---- [ φ ]
2
∆t 3

M
[ A 4 ] = [ φ ]T – ------- B K
- + --------- – ---- [ φ ]
2
∆t 2∆t 3

These equations are similar to the direct transient method except that they are in terms of
modal coordinates. Since the number of modes used in a solution is typically much less
than the number of physical variables, the direct integration of the modal equations is not
as costly as with physical variables.
If damping is applied to each mode separately, the decoupled equations of motion can be
maintained. When modal damping is used, each mode has damping b i . The equations of
motion remain uncoupled and have the following form for each mode:

·· ·
mi ξi ( t ) + bi ξi ( t ) + ki ξi ( t ) = pi ( t ) Eq. 6-21

or

·· · 2 1
ξ i ( t ) + 2ζ i ω i ξ i ( t ) + ω i ξ i ( t ) = ------ p i ( t ) Eq. 6-22
mi

where:

ζ i = b i ⁄ ( 2m i ω i ) ≡ modal damping ratio

2
ω i = k i ⁄ m i ≡ modal frequency (eigenvalue)

The TABDMP1 Bulk Data entry defines the modal damping ratios. A table is created by
the frequency-damping pairs specified on a TABDMP1 entry. The solution refers to this
table for the damping value to be applied at a particular frequency. The TABDMP1 Bulk
Data entry has a Set ID. A particular TABDMP1 table is activated by selecting the Set ID
with SDAMPING = Set ID Case Control command.

1 2 3 4 5 6 7 8 9 10
TABDMP1 ID TYPE

f1 g1 f2 g2 f3 g3 -etc.- ENDT

Field Contents
TID Table identification number.
TYPE Type of damping units:
G (default)
CRIT
Q
 CHAPTER 6 161
Transient Response Analysis

Field Contents
fi Frequency value (cycles per unit time).
gi Damping value in the units specified.

At resonance, the three types of damping are related by the following equations:

bi Gi
ζ i = ------- = ------
b cr 2

b cr = 2m i ω i Eq. 6-23

1 1
Q i = -------- = ------
2ζ i Gi

The values of fi (units = cycles per unit time) and gi define pairs of frequencies and
dampings. Note that gi can be entered as structural damping (default), critical damping,
or quality factor. The entered damping is internally converted to structural damping using
Eq. 6-23. Straight-line interpolation is used for modal frequencies between consecutive fi
values. Linear extrapolation is used at the ends of the table. ENDT ends the table input.

For example, if modal damping is entered using Table 6-1 and if modes exist at 1.0, 2.5, 3.6,
and 5.5 Hz, MSC.Nastran interpolates and extrapolates as shown in Figure 6-2 and the
table. Note that there is no table entry at 1.0 Hz; MSC.Nastran uses the first two table
entries at f = 2.0 and f = 3.0 to extrapolate the value for f = 1.0 .

0.20
+
+
+ +
0.10
+ = entered
value
= computed value

0.0
2.0 4.0 6.0
f(Hz)

Figure 6-2 Example TABDMP1

162

Table 6-1 Example TABDMP1 Interpolation/Extrapolation

Entered Computed
f ζ f ζ
2.0 0.16 1.0 0.14
3.0 0.18 2.5 0.17
4.0 0.13 3.6 0.15
6.0 0.13 5.5 0.13

1 2 3 4 5 6 7 8 9 10
TABDMP1 10 CRIT +TAB1

+TAB1 2.0 0.16 3.0 0.18 4.0 0.13 6.0 0.13 +TAB2

+TAB2 ENDT

With the modal equations in the form of Eq. 6-22, an efficient uncoupled analytical
integration algorithm is used to solve for modal response as decoupled SDOF systems.
Each of the modal responses is computed using

0, No initial conditions for

modal transient response.

·
– bt ⁄ 2m  ξ o + b ⁄ ( 2m ) ξ o 
ξ( t) = e  ξ o cos ω d t + ----------------------------------------- sin ω d t Eq. 6-24
 ωd 
t
– bt ⁄ 2m 1 bτ ⁄ 2m
+e ----------- ∫ e p ( τ ) sin ω d ( t – τ )dτ
mω d
0

In a modal transient analysis, you may add nonmodal damping (CVISC, CDAMPi, GE on
the MATi entry, or PARAM,G). With nonmodal damping, there is a computational penalty
due to the coupled [ B ] matrix, causing the coupled solution algorithm to be used. In modal
transient response analysis, it is recommended that you use only modal damping
(TABDMP1). If discrete damping is desired, direct transient response analysis is
recommended.
Note that there are no nonzero initial conditions for modal transient response analysis.

Mode Truncation in Modal Transient Response Analysis

It is possible that not all of the computed modes are required in the transient response
solution. Often, only the lowest few suffice for dynamic response calculation. It is quite
common to evaluate the frequency content of transient loads and determine a frequency
above which no modes are noticeably excited. This frequency is called the cutoff
frequency. The act of specifically not using all of the modes of a system in the solution is
 CHAPTER 6 163
Transient Response Analysis

termed mode truncation. Mode truncation assumes that an accurate solution can be
obtained using a reduced set of modes. The number of modes used in a solution is
controlled in a modal transient response analysis through a number of methods.

The frequency range selected on the eigenvalue entry (EIGRL or EIGR) is one means to
control the frequency range used in the transient response solution. Also, three parameters
are available to limit the number of modes included in the solution. PARAM,LFREQ gives
the lower limit on the frequency range of retained modes, and PARAM,HFREQ gives the
upper limit on the frequency range of retained modes. PARAM,LMODES gives the
number of the lowest modes to be retained. These parameters can be used to include the
desired set of modes. Note that the default is for all computed modes to be retained.
It is very important to remember that truncating modes in a particular frequency range
may truncate a significant portion of the behavior in that frequency range. Typically,
high-frequency modes are truncated because they are more costly to compute. So,
truncating high-frequency modes truncates high frequency response. In most cases,
high-frequency mode truncation is not of concern. You should evaluate the truncation in
terms of the loading frequency and the important characteristic frequencies of the
structure.

Dynamic Data Recovery in Modal Transient Response

Analysis
In modal transient response analysis, two options are available for recovering
displacements and stresses: mode displacement method and matrix method. Both
methods give the same answers, although with cost differences.
The mode displacement method computes the total physical displacements for each time
step from the modal displacements and then computes element stresses from the total
physical displacements. The number of operations is proportional to the number of time
steps.
The matrix method computes displacements per mode and element stresses per mode, and
then computes physical displacements and element stresses as the summation of modal
displacements and element stresses. Costly operations are proportional to the number of
modes.
Since the number of modes is usually much less that the number of time steps, the matrix
method is usually more efficient and is the default. The mode displacement method can
be selected by using PARAM,DDRMM,-1 in the Bulk Data. The mode displacement
method is required when “time frozen” deformed structure plots are requested (see
“Plotted Output” on page 247).

The mode acceleration method (“Advanced Dynamic Analysis Capabilities” on page 293)
is another form of data recovery for modal transient response analysis. This method can
provide better accuracy since detailed local stresses and forces are subject to mode
truncation and may not be as accurate as the results computed with the direct method.
164

6.4 Modal Versus Direct Transient Response

Some general guidelines can be used in selecting modal transient response analysis versus
direct transient response analysis. These guidelines are summarized in Table 6-2.

Table 6-2 Modal Versus Direct Transient Response

Modal Direct
Small Model X
Large Model X
Few Time Steps X
Many Time Steps X
High Frequency Excitation X
Normal Damping X
Higher Accuracy X
Initial Conditions X

In general, larger models may be solved more efficiently in modal transient response
because the numerical solution is a solution of a smaller system of uncoupled equations.
This result is certainly true if the natural frequencies and mode shape were computed
during a previous stage of the analysis. Using Duhamel’s integral to solve the uncoupled
equations is very efficient even for very long duration transients. On the other hand, the
major portion of the effort in a modal transient response analysis is the calculation of the
modes. For large systems with a large number of modes, this operation can be as costly as
direct integration. This is especially true for high-frequency excitation. To capture high
frequency response in a modal solution, less accurate high-frequency modes must be
computed. For small models with a few time steps, the direct method may be the most
efficient because it solves the equations without first computing the modes. The direct
method is more accurate than the modal method because the direct method is not
concerned with mode truncation. For systems with initial conditions, direct transient
response is the only choice.

Table 6-2 provides a starting place for evaluating which method to use. Many additional
factors may be involved in the choice of a method, such as contractual obligations or local
standards of practice.
 CHAPTER 6 165
Transient Response Analysis

6.5 Transient Excitation Definition

An important aspect of a transient response analysis is the definition of the loading
function. In a transient response analysis, the force must be defined as a function of time.
Forces are defined in the same manner whether the direct or modal method is used.

The following Bulk Data entries are used for the transient load definition:

TLOAD1 Tabular input

TLOAD2 Analytical function
DAREA Spatial distribution of dynamic load
TABLEDi Tabular values versus time
LSEQ Generates the spatial distribution of dynamic loads from static
load entries
DLOAD Combines dynamic load sets
DELAY Time delay

The particular entry chosen for defining the dynamic loading is largely a function of user
convenience for concentrated loads. Pressure and distributed loads, however, require a
more complicated format.
There are two important aspects of dynamic load definition. First, the location of the
loading on the structure must be defined. Since this characteristic locates the loading in
space, it is called the spatial distribution of the dynamic loading. Secondly, the time
variation in the loading is the characteristic that differentiates a dynamic load from a static
load. This time variation is called the temporal distribution of the load. A complete
dynamic loading is a product of spatial and temporal distributions.
Using Table IDs and Set IDs in MSC.Nastran makes it possible to apply many complicated
and temporally similar loadings with a minimum of input. Combining simple loadings to
create complicated loading distributions that vary in position as well as time is also a
straightforward task.

The remainder of this section describes the Bulk Data entries for transient excitation. The
description is given in terms of the coefficients that define the dynamic load. See
Appendix F for more complete Bulk Data descriptions.
166

Time-Dependent Loads – TLOAD1 Entry

The TLOAD1 entry is the most general form in which to define a time-dependent load. It
defines a dynamic loading of the form

{P(t )} = {A ⋅ F(t – τ) } Eq. 6-25

The coefficients of the force are defined in tabular format. You need not explicitly define a
force at every instant in time for which the transient solution is evaluated. Only those
values which describe the character of the loading are required. MSC.Nastran interpolates
linearly for intermediate values.

1 2 3 4 5 6 7 8 9 10
TLOAD1 SID DAREA DELAY TYPE TID

Field Contents
SID Set ID defined by a DLOAD Case Control command or a DLOAD Bulk
Data entry.
DAREA Identification number of DAREA entry that defines A. (Integer > 0)
DELAY Identification number of DAREA entry that defines τ . (Integer > 0)
TYPE Excitation function as defined below. Additional information is in
“Enforced Motion” on page 199.

Integer Excitation Function

0 or blank Force or Moment
1 Enforced Displacement
2 Enforced Velocity
3 Enforced Acceleration

Values 1, 2, and 3 apply only to the large mass method for enforced
motion.
TID TABLEDi entry that defines F(t).
 CHAPTER 6 167
Transient Response Analysis

Time-Dependent Loads – TLOAD2 Entry

The TLOAD2 entry is a general analytical form with which to define a time-dependent
load. The value of the force at a particular instant in time is determined by evaluating the
analytic function at the specific time. You enter the appropriate constants in the function.

The TLOAD2 entry defines dynamic excitation in the form:


 0 , t < ( T1 + τ ) or t > ( T2 + τ )
{P( t) } =  Eq. 6-26
 A t̃ B e C t̃ cos ( 2πF t̃ + P ) , ( T1 + τ ) ≤ t ≤ ( T2 + τ )


where: t̃ = t – ( T1 – τ )

1 2 3 4 5 6 7 8 9 10
TLOAD2 SID DAREA DELAY TYPE T1 T2 F P

C B

Field Contents
SID Set ID defined by a DLOAD Case Control command.
DAREA Identification number of DAREA entry that defines A. (Integer > 0)
DELAY Identification number of DAREA entry that defines τ . (Integer > 0)
TYPE Defined as on the TLOAD1 entry.
T1, T2 Time constants.
F Frequency (cycles per unit time).
P Phase angle (degrees).
C Exponential coefficient.
B Growth coefficient.
168

Spatial Distribution of Loading – DAREA Entry

The DAREA entry defines the degrees-of-freedom where the dynamic load is to be applied
and a scale factor to be applied to the loading. The DAREA entry provides the basic spatial
distribution of the dynamic loading.

1 2 3 4 5 6 7 8 9 10
DAREA SID P1 C1 A1 P2 C2 A2

Field Contents
SID Set ID specified by TLOADi entries.
Pi Grid, extra, or scalar point ID.
Ci Component number.
Ai Scale factor.

A DAREA entry is selected by the TLOAD1 or TLOAD2 entry. Any number of DAREA
entries may be used; all those with the same SID are combined.

Time Delay – DELAY Entry

The DELAY entry defines the time delay τ in an applied load.

1 2 3 4 5 6 7 8 9 10
DELAY SID P1 C1 τ1 P2 C2 τ2

Field Contents
SID Set ID specified by TLOADi entry.
Pi Grid, extra, or scalar point ID.
Ci Component number.
τi Time delay for Pi, Ci.

A DAREA entry must be defined for the same point and component.
Any number of DELAY entries may be used; all those with the same SID are combined.
 CHAPTER 6 169
Transient Response Analysis

Dynamic Load Tabular Function – TABLEDi Entries

The TABLEDi entries (i = 1 through 4) each define a tabular function for use in generating
frequency-dependent dynamic loads. The form of each TABLEDi entry varies slightly,
depending on the value of i, as does the algorithm for y(x). The x values need not be evenly
spaced.
The TABLED1, TABLED2, and TABLED3 entries linearly interpolate between the end
points and linearly extrapolate outside of the endpoints as shown in Figure 6-3. In
Version 69 the TABLED1 added the option to perform logrithmic interpolation between
points, also. The TABLED4 entry uses the endpoint values for values beyond the
endpoints.

Linear Interpolation Between Endpoints

y

Linear Extrapolation of Segment x6 – x7

Linear Extrapolation of Segment x2 – x1

x
x1 x2 x3 x4 x5 x6 x7

Figure 6-3 Interpolation and Extrapolation for TABLED1, TABLED2,

and TABLED3 Entries

The TABLED1 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED1 TID XAXIS YAXIS

x1 y1 x2 y2 x3 y3 -etc.- ENDT

Field Contents
TID Table identification number.
XAXIS Specifies a linear or logarithmic interpolation for the x-axis.
(Character: “LINEAR” or “LOG”; Default = “LINEAR”)
YAXIS Specifies a linear or logarithmic interpolation for the y-axis.
(Character: “LINEAR” or “LOG”; Sefault = “LINEAR”)
xi, yi Tabular values. Values of x are frequency in cycles per unit time.
ENDT Ends the table input.

The TABLED1 entry uses the algorithm

y = yT ( x ) Eq. 6-27
170

The algorithms used for interpolation and extrapolation are as follows:

XAXIS YAXIS yT ( x )
LINEAR LINEAR xj – x x – xi
----------------- yi + ----------------- yj
xj – xi xj – xi
LOG LINEAR ln ( xj ⁄ x ) ln ( x ⁄ xi )
--------------------------- yi + --------------------------- yj
ln ( xj ⁄ xi ) ln ( xj ⁄ xi )
LINEAR LOG xj – x x – xi
exp ----------------- ln yi + ----------------- ln yj
xj – xi xj – xi

LOG LOG ln ( xj ⁄ x ) ln ( x ⁄ xi )
exp --------------------------- ln yi + --------------------------- ln yj
ln ( xj ⁄ xi ) ln ( xj ⁄ xi )

The TABLED2 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED2 TID X1

x1 y1 x2 y2 x3 y3 -etc.- ENDT

Field Contents
TID Table identification number.
X1 Table parameter.
xi, yi Tabular values.
ENDT Ends the table input.

The TABLED2 entry uses the algorithm

y = y T ( x – X1 ) Eq. 6-28
 CHAPTER 6 171
Transient Response Analysis

The TABLED3 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED3 TID X1 X2

x1 y1 x2 y2 x3 y3 -etc.- ENDT

Field Contents
TID Table identification number.
X1, X2 Table parameters.
xi, yi Tabular values.
ENDT Ends the table input.

The TABLED3 entry uses the algorithm

y = y T  -----------------
x – X1
Eq. 6-29
X2

The TABLED4 entry has the following format:

1 2 3 4 5 6 7 8 9 10
TABLED4 TID X1 X2 X3 X4

A0 A1 A2 A3 A4 A5 -etc.- ENDT

Field Contents
TID Table identification number.
Xi Table parameters. (X2 ≠ 0.0; X3 < X4).
Ai Coefficients.

The TABLED4 entry uses the algorithm

N
i
Ai  -----------------
x – X1
y = ∑ X2
Eq. 6-30
i = 0

N is the degree of the power series. When x < X3 , X3 is used for x ; when x > X4 , X4 is used
for x . This condition has the effect of placing bounds on the table; there is no extrapolation
outside of the table boundaries.
ENDT ends the table input.
172

DAREA Example
Suppose the following command is in the Case Control Section:

DLOAD = 35
in addition to the following entries in the Bulk Data Section:

1 2 3 4 5 6 7 8 9 10
$TLOAD1 SID DAREA DELAY TYPE TID

TLOAD1 35 29 31 40

$DAREA SID POINT COMPONENT SCALE

DAREA 29 30 1 4.0

$DELAY SID POINT COMPONENT LAG

DELAY 31 30 1 0.2

$TABLED1 ID

$ X1 y1 x2 y2 x3 y3 x4 y4

TABLED1 40 LINEAR LINEAR

0.0 0.0 0.3 1.0 2.0 1.0 ENDT

The DLOAD Set ID 35 in the Case Control selects the TLOAD1 entry in the Bulk Data
having a Set ID 35. On the TLOAD1 entry is a reference to DAREA Set ID 29, DELAY Set
ID 31, and TABLED1 Set ID 40. The DAREA entry with Set ID 29 positions the loading on
grid point 30 in the 1 direction with a scale factor of 4.0 applied to the load. The DELAY
entry with Set ID 31 delays the loading on grid point 30 in the 1 direction by 0.2 units of
time. The TABLED1 entry with Set ID 40 defines the load time history in tabular form. The
result of these entries is a dynamic load applied to grid point 30, component T1, scaled by
4.0 and delayed by 0.2 units of time.
Figure 6-4 shows the TABLED1 time history and the applied load (scaled by the DAREA
entry and time shifted by the DELAY entry).
 CHAPTER 6 173
Transient Response Analysis

1.0

Amplitude
on
TABLED1

0
0.5 1.0 1.5 2.0 2.5
Time (sec)

4.0
Force
Amplitude

0
0.5 1.0 1.5 2.0 2.5
Time (sec)

Figure 6-4 Time History from the TABLED1 Entry (Top) and Applied Load (Bottom)

Static Load Sets – LSEQ Entry

MSC.Nastran does not have specific data entries for many types of dynamic loads. Only
concentrated forces and moments can be specified directly using DAREA entries. To
accommodate more complicated loadings conveniently, the LSEQ entry is used to define
static load entries that define the spatial distribution of dynamic loads.

1 2 3 4 5 6 7 8 9 10
LSEQ SID DAREA ID LOAD ID TEMP ID

The LSEQ Bulk Data entry contains a reference to a DAREA Set ID and a static Load Set ID.
The static loads are combined with any DAREA entry in the referenced set. The DAREA
Set ID does not need to be defined with a DAREA Bulk Data entry. The DAREA Set ID is
referenced by a TLOADi entry. This reference defines the temporal distribution of the
dynamic loading. The Load Set ID may refer to one or more static load entries (FORCE,
PLOADi, GRAV, etc.). All static loads with the Set ID referenced on the LSEQ entry define
the spatial distribution of the dynamic loading. MSC.Nastran converts this information to
equivalent dynamic loading.

Figure 6-5 demonstrates the relationships of these entries. To activate a load set defined in
this manner, the DLOAD Case Control command refers to the Set ID of the selected
DLOAD or TLOADi entry, and the LOADSET Case Control command refers to the Set ID
174

of the selected LSEQ entries. The LSEQ entries point to the static loading entries that are
used to define dynamic loadings with DAREA Set ID references. Together this relationship
defines a complete dynamic loading. To apply dynamic loadings in this manner, the
DLOAD and LOADSET Case Control commands and the TLOADi and LSEQ Bulk Data
entries must be defined. A DAREA Bulk Data entry does not need to be defined since the
TLOADi and LSEQ entries reference a common DAREA ID. The LSEQ entry can also be
interpreted as an internal DAREA entry generator for static load entries.

DLOAD LOADSET

Case Control
Bulk Data
TLOADi LSEQ

Dynamic Load DAREA Static Load Entries

(Temporal (Reference (Spatial
Distribution) Link) Distribution)
Figure 6-5 Relationship of Dynamic and Static Load Entries

LSEQ Example
Suppose the following commands are in the Case Control Section:
LOADSET = 27
DLOAD = 25
in addition to the following entries in the Bulk Data Section:

1 2 3 4 5 6 7 8 9 10
$LSEQ SID DAREA LID

LSEQ 27 28 26

$TLOAD1 SID DAREA DELAY TYPE TID

TLOAD1 25 28 29
DAREA
Reference
Link
$STATIC LOAD SID

PLOAD1 26 -etc.-

FORCE 26 -etc.-

$TABLED1 TID

TABLED1 29 -etc.-
 CHAPTER 6 175
Transient Response Analysis

In the above, the LOADSET request in Case Control selects the LSEQ Set ID 27 entry. The
DLOAD request in Case Control selects the TLOAD1 Set ID 25 entry. This TLOAD1 entry
refers to a TABLED1 ID 29, which is used to define the temporal variation in the loading.
DAREA Set ID 28 links the LSEQ and TLOAD1 entries. In addition, the LSEQ entry refers
to static Load Set ID 26, which is defined by FORCE and PLOAD1 entries. The FORCE and
PLOAD1 entries define the spatial distribution of the dynamic loading and through the
DAREA link refer to the TLOAD1/TABLED1 combination for the time-varying
characteristics of the load.

Dynamic Load Set Combination – DLOAD

One of the requirements of transient loads is that all TLOAD1s and TLOAD2s must have
unique SIDs. If they are to be applied in the same analysis, they must be combined using
the DLOAD Bulk Data entry. The total applied load is constructed from a combination of
component load sets as follows:

{ P } = S ∑ Si { Pi } Eq. 6-31
i

where:

S = overall scale factor

S i = scale factor for i-th load set

P i = i-th set of loads

P = total applied load

The DLOAD Bulk Data entry has the following format:

1 2 3 4 5 6 7 8 9 10
DLOAD SID S S1 L1 S2 L2 -etc.-

Field Contents
SID Load set ID.
S Overall scale factor.
Si Individual scale factors.
Li Load set ID number for TLOAD1 and TLOAD2 entries.
176

As an example, in the following DLOAD entry:

1 2 3 4 5 6 7 8 9 10
$DLOAD SID S S1 L1 S2 L2 -etc.-

DLOAD 33 3.25 0.5 14 2.0 27

a dynamic Load Set ID of 33 is created by taking 0.5 times the loads in Load Set ID of 14,
adding to it 2.0 times the loads in Load Set ID of 27, and multiplying that sum by an overall
scale factor of 3.25.

As with other transient loads, a dynamic load combination defined by the DLOAD Bulk
Data entry is selected by the DLOAD Case Control command.
 CHAPTER 6 177
Transient Response Analysis

6.6 Integration Time Step

The TSTEP Bulk Data entry is used to select the integration time step for direct and modal
transient response analysis. This entry also controls the duration of the solution and which
time steps are to be output. The TSTEP Bulk Data entry is selected by the Set ID referenced
on the TSTEP Case Control command.
The integration time step must be small enough to represent accurately the variation in the
loading. The integration time step must also be small enough to represent the maximum
frequency of interest. The maximum frequency of interest is often called the cut-off
frequency. It is recommended to use at least ten solution time steps per period of response
for the cutoff frequency. For a given integration time step, integration errors increase with
increasing natural frequency because there is an upper limit to the frequency that can be
represented by a given time step. Also, integration errors accumulate with total time.
In both direct and modal transient analysis, the cost of integration is directly proportional
to the number of time steps. For example, doubling the load duration doubles the
integration effort.
In specifying the duration of the analysis on the TSTEP entry, it is important to use an
adequate length of time to properly capture long period (low frequency) response. In
many cases, the peak dynamic response does not occur at the peak value of load nor
necessarily during the duration of the loading function. A good rule is: always solve for
at least one cycle of response for the lowest frequency mode after the peak excitation.
You may change ∆t during a run but doing so causes the dynamic matrix to be
redecomposed, which can be costly in direct transient response analysis.
The TSTEP Bulk Data entry has the following format:

1 2 3 4 5 6 7 8 9 10
TSTEP SID N1 ∆t1 NO1

N2 ∆t2 NO2

-etc.-

Field Contents
SID Set ID specified by a TSTEP Case Control command.
Ni Number of time steps of value ∆t i .

∆t i Integration time step.

NOi Output every NOi-th time step.

178

6.7 Transient Excitation Considerations

A number of important considerations must be remembered when applying transient
loads. The averaging of applied loads (Eq. 6-3) in the integration smooths the force and
decreases the apparent frequency content. Very sharp spikes in a loading function induce
a high-frequency transient response. If the high-frequency transient response is of primary
importance in an analysis, a very small integration time step must be used.

It is also important to avoid defining discontinuous forcing functions when describing

applied loads. The numerical integration of discontinuous forcing functions may cause
different results for the same analysis run on different computers because of slight
numerical differences on different computer types. If the analysis calls for loadings with
sharp impulses, it is best to smooth the impulse over at least one integration time
increment.
The loading function must accurately describe the spatial and temporal distribution of the
dynamic load. Simplifying assumptions must not change the character of the load in
magnitude, location, or frequency content.
 CHAPTER 6 179
Transient Response Analysis

6.8 Solution Control for Transient Response Analysis

The following tables summarize the data entries that can be used to control a transient
response analysis. Certain data entries are required, some data entries are optional, while
others are user selectable.

In the Executive Control Section of the MSC.Nastran input file, a solution must be selected
using the SOL i statement where i is an integer value chosen from Table 6-3.

Table 6-3 Transient Response Solutions in MSC.Nastran

Structured
Rigid
Solution
Formats
Sequences
Direct 9 109
Modal 12 112

We have applied these solutions in the examples that follow.

In the Case Control Section of the MSC.Nastran input file, you must select the solution
parameters associated with the current analysis (i.e., time steps, loads, and boundary
conditions) and also the output quantities required from the analysis. The Case Control
commands directly related to transient response analysis are listed in Table 6-4. They can
be combined in the standard fashion with the more generic commands, such as SPC, MPC,
etc.

Table 6-4 Transient Response Case Control Commands

Case Direct
Required/
Control or Description
Optional
Commands Modal
DLOAD Both Select the dynamic load set from the Bulk Data Required*
TSTEP Both Select the TSTEP entry from the Bulk Data Required
METHOD Modal Select the eigenvalue extraction parameters Required
LOADSET Both Select the LSEQ set from the Bulk Data Optional
SDAMPING Modal Select the modal damping table from the Bulk Data Optional
IC Direct Select TIC entries for initial conditions from the Optional
Bulk Data
OTIME Both Select the times for output (default = all) Optional

* Not required when using initial conditions.

180

The types of results available from a transient response analysis are similar to those for a
static analysis except that the results vary with time. Additional quantities are also
available, which is characteristic of dynamics problems. The output quantities are
summarized in Table 6-5 and Table 6-6.

Table 6-5 Grid Point Output from a Transient Response Analysis

Case Control
Description
Command
ACCELERATION Grid point acceleration time history for a set of grid points
DISPLACEMENT Grid point displacement time history for a set of grid points
(or VECTOR)
GPSTRESS Grid point stress time history (requires SURFACE/VOLUME
definition in the OUTPUT(POST) section of the Case Control)
OLOAD Requests applied load table to be output for a set of grid points
SACCELERATION Requests solution set acceleration output: d-set in direct
solutions and modal variables in modal solutions
SDISPLACEMENT Requests solution set displacement output: d-set in direct
solutions and modal variables in modal solutions
SVECTOR Requests real eigenvector output for the a-set in modal
solutions
SVELOCITY Requests solution set velocity output: d-set in direct solutions
and modal variables in modal solutions
SPCFORCES Requests forces of single-point constraint for a set of grid points
VELOCITY Grid point velocity time history for a set of grid points

Table 6-6 Element Output from a Transient Analysis

Case Control
Description
Command
ELSTRESS (or STRESS) Element stress time history for a set of elements
ELFORCE (or FORCE) Element force time history for a set of elements
STRAIN Element strain time history for a set of elements

Because the results may be output for many time steps, the volume of output can be very
large. Prudent selection of the output quantities is recommended.
 CHAPTER 6 181
Transient Response Analysis

A number of Bulk Data entries are unique to transient response analysis. They can be
combined with other generic entries in the Bulk Data. Bulk Data entries directly related to
transient analysis are summarized in Table 6-7.

Table 6-7 Bulk Data Entries for Transient Response Analysis

Bulk Data Direct or Required/

Description
Entry Modal Optional
TSTEP Both Integration time step and solution control Required
TLOADi Both Dynamic loading Required*
EIGR or EIGRL Modal Eigenvalue analysis parameters Required
LSEQ Both Dynamic loading from static loads Optional
TABLEDi Both Time-dependent tables for TLOADi Optional*
TIC Direct Initial conditions on grid, scalar, and Optional
extra points
DAREA Both Load component and scale factor Required†
DELAY Both Time delay on dynamic load Optional
DLOAD Both Dynamic load combination, required if Optional
TLOAD1 and TLOAD2 are used
TABDMP1 Modal Modal damping table Optional

* Not required for initial conditions.

†The DAREA ID is required; the DAREA Bulk Data entry is not required if an LSEQ
entry is used.
182

6.9 Examples
This section provides several examples showing the input and output. These examples are

Transient Response
Model Output
Bulk Data Entries
bd06two TSTEP, TIC X-Y plots
bd06bar EIGRL, TSTEP, X-Y plots
TABDMP1, DLOAD,
TLOAD2, DAREA,
DELAY
bd06bkt EIGRL, TSTEP, X-Y plot
TABDMP1, TLOAD1,
LSEQ, TABLED1,
PLOAD4

These examples are described in the sections that follow.

Two-DOF Model
Consider the two-DOF system shown in Figure 6-6. Direct transient response (SOL 109) is
run with an initial displacement of 0.1 meter at grid point 2. The analysis is run for a
duration of 10 seconds with a ∆t of 0.01 second. Damping is neglected in the analysis.
Listing 6-1 shows part of the input file.

Grid Point 1 m1

k1

Grid Point 2 m2

k2

x
Figure 6-6 Two-DOF Model
 CHAPTER 6 183
Transient Response Analysis

Listing 6-1 Input File (Abridged) for the Two-DOF Example

$ FILE bd06two.dat
$
$ TWO-DOF SYSTEM
$ CHAPTER 6, TRANSIENT RESPONSE
$
TIME 5
SOL 109 $ DIRECT TRANSIENT RESPONSE
CEND
TITLE = TWO-DOF SYSTEM
SUBTITLE = DIRECT FREQUENCY RESPONSE
LABEL = INITIAL DISPL. AT GRID 2
$
$ SPECIFY SPC
SPC = 996
$
$ SPECIFY DYNAMIC INPUT
TSTEP = 888
IC = 777
$
$ SELECT OUTPUT
SET 11 = 1,2
DISPLACEMENT(PLOT) = 11
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ ENTRIES FOR TRANSIENT RESPONSE
$
$ INITIAL CONDITION
$TIC SID G C U0 V0
TIC 777 2 2 0.1
$
$ TIME STEP
$TSTEP SID N1 DT1 NO1
TSTEP 888 1000 0.01 1
$
... basic model ...
$
ENDDATA

Table 6-8 shows the relationship between the Case Control commands and the Bulk Data
entries. This example represents the simplest form of dynamic response input. The only
required entries are those that define the time step and the initial conditions. Note that the
unspecified initial conditions are assumed to be zero. Note, too, that the initial conditions
are available only for direct transient response analysis.
184

Table 6-8 Relationship Between Case Control Commands

and Bulk Data Entries for the Two-DOF Model

Case Control Bulk Data

TSTEP TSTEP
IC TIC

Figure 6-7 shows the plots of the resulting displacements for grid points 1 and 2. Note that
there are two frequencies of response: a higher frequency of about 5 Hz, and a lower
frequency of about 0.25 Hz. The energy (and hence response) appears to be transferred
repetitively between grid points 1 and 2 as represented by the lower frequency response.
This energy transfer is called beating. Beating occurs when there are closely-spaced modes
(in this case, 4.79 Hz and 5.29 Hz) in which energy transfer can readily occur. The response
is comprised of two frequencies as given below:

f1 + f2
f higher = ------------------
2
Eq. 6-32
f2 – f1
f lower = -----------------
2

where:
f1 = lower of the closely-spaced mode frequencies
f2 = higher of the closely-spaced mode frequencies

In this example, f higher is 5.04 Hz and f lower is 0.25 Hz. The lower frequency is called the
beat frequency and is the frequency at which energy transfer occurs.
 CHAPTER 6 185
Transient Response Analysis

Figure 6-7 Displacements of Grid Points 1 and 2

Cantilever Beam Model

Consider the cantilever beam shown in Figure 6-8. This beam model is the same as in
“Examples” in Chapter 5. Modal transient response (SOL 112) is run with loads applied to
grid points 6 and 11 as shown in Figure 6-9. The analysis is run for a duration of 2 seconds
with a ∆t of 0.001 second. Modal damping of 5% critical damping is used for all modes.
Modes up to 3000 Hz are computed using the Lanczos method. Listing 6-2 shows part of
the input file.
186

6 11
x

p6(t) p11(t)

Figure 6-8 Cantilever Beam Model with Applied Loads

Force 6 (N)
Time
(sec)
-3
0.1 0.5

Force 11 (N)
Time
(sec)
-6
0.5
Figure 6-9 Applied Loads for the Beam Model
 CHAPTER 6 187
Transient Response Analysis

Listing 6-2 Input File (Abridged) for the Beam Example

$ FILE bd06bar.dat
$
$ CANTILEVER BEAM MODEL
$ CHAPTER 6, TRANSIENT RESPONSE
$
SOL 112 $ MODAL TRANSIENT RESPONSE
TIME 10
CEND
TITLE = CANTILEVER BEAM
SUBTITLE = MODAL TRANSIENT RESPONSE
$
SPC = 21
DLOAD = 22
TSTEP = 27
SDAMPING = 25
$
METHOD = 10
$
$ PHYSICAL OUTPUT REQUEST
SET 11 = 6,11
DISPLACEMENT(PLOT) = 11
ACCELERATION(PLOT) = 11
$
$ MODAL SOLUTION SET OUTPUT
SET 12 = 1,2
SDISP(PLOT) = 12
$
$ ELEMENT FORCE OUTPUT
SET 13 = 6
ELFORCE(PLOT) = 13
$
$ APPLIED LOAD OUTPUT
SET 15 = 6,11
OLOAD(PLOT) = 15
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$EIGRL SID V1 V2 MSGLVL
EIGRL 10 -0.1 3000. 0
$
$TSTEP SID N1 DT1 NO1
TSTEP 27 2000 0.001 1
$
$ MODAL DAMPING OF 5% IN ALL MODES
$TABDMP1 TID TYPE +TABD
$+TABD F1 G1 F2 G2 ETC.
TABDMP1 25 CRIT +TABD
+TABD 0. 0.05 1000. 0.05 ENDT
$
$ DYNAMIC LOADING
$DLOAD SID S S1 L1 S2 L2
DLOAD 22 1.0 1.0 231 1.0 232
$TLOAD2 SID DAREA DELAY TYPE T1 T2 F P +TL1
$+TL1 C B
TLOAD2 231 241 0 0.0 0.5 2.0 90.
TLOAD2 232 242 262 0 0.0 0.5 4.0 90.
$DAREA SID P1 C1 A1
DAREA 241 11 2 6.0
DAREA 242 6 2 3.0
$DELAY SID P1 C1 T1
DELAY 262 6 2 0.1
$
... basic model ...
$
ENDDATA
188

Table 6-9 shows the relationship between the Case Control commands and the Bulk Data
entries. The DLOAD Bulk Data entry references two TLOAD2 entries, each of which
references separate DAREA entries. A TLOAD2 entry also references a DELAY entry to
apply the time delay to the load at grid point 6.

Table 6-9 Relationship Between Case Control Commands and Bulk Data Entries
for the Bar Model

Case Control Bulk Data

METHOD EIGRL
TSTEP TSTEP
SDAMPING TABDMP1
DLOAD DLOAD
TLOAD2 231

DAREA
TLOAD2 232

DAREA
DELAY

Plotted output is shown in the following figures. Figure 6-10 shows the applied loads at
grid points 6 and 11. Figure 6-11 shows the plots of the displacements for grid points 6 and
11. Figure 6-12 shows the accelerations for grid points 6 and 11. Figure 6-13 shows the
bending moment at end A in plane 1 for element 6. Figure 6-14 shows the modal
displacements for modes 1 and 2.
 CHAPTER 6 189
Transient Response Analysis

3.0

Force 6 (N)
0.0

-3.0

6.0
Force 11 (N)

0.0

-6.0
0.0 2.0
Time (sec)
Figure 6-10 Applied Loads at Grid Points 6 and 11
190

0.03

Displacement 6 (m) 0.0

-0.02

0.08
Displacement 11 (m)

0.0

-0.06
0.0 2.0
Time (sec)
Figure 6-11 Displacements at Grid Points 6 and 11
 CHAPTER 6 191
Transient Response Analysis

4.0

Acceleration 6 (m/sec)
2
0.0

-6.0

10.0
Acceleration 11 (m/sec)
2

0.0

-15.0
0.0 2.0
Time (sec)
Figure 6-12 Accelerations at Grid Points 6 and 11
192

25.0

Bending Moment A1 (Nm)

0.0

-20.0
0.0 2.0
Time (sec)
Figure 6-13 Bending Moment A1 for Element 6
 CHAPTER 6 193
Transient Response Analysis

0.08

SDisplacement1
0.0

-0.10

0.0015
SDisplacement2

0.0

-0.0010
0.0 2.0
Time (sec)
Figure 6-14 Modal Displacements for Modes 1 and 2

Bracket Model
Consider the bracket model shown in Figure 6-15. A pressure load of 3 psi is applied to the
elements in the top face in the z-direction with the time history shown in Figure 6-16. The
modal transient analysis is run for 4 seconds with a time step size of 0.005 second. Modal
damping of 2% critical damping is used for all modes. Modes up to 3000 Hz are computed
with the Lanczos method. The model is constrained near the base.
194

Figure 6-15 Bracket Model

Pressure

3 psi

0 0.10 0.15 5
Time (sec)
Figure 6-16 Time Variation for Applied Load

Listing 6-3 shows the abridged input file. The LSEQ entry is used to apply the pressure
loads (PLOAD4 entries). Note that the LSEQ and TLOAD1 entries reference a common
DAREA ID (999) and that there is no explicit DAREA entry. Table 6-10 shows the
relationship between the Case Control commands and the Bulk Data entries.
 CHAPTER 6 195
Transient Response Analysis

Listing 6-3 Input File (Abridged) for the Bracket Model

$ FILE bd06bkt.dat
$
$ BRACKET MODEL
$ CHAPTER 6, TRANSIENT RESPONSE
$
SOL 112 $ MODAL TRANSIENT RESPONSE
TIME 100
CEND
TITLE = BRACKET MODEL
SUBTITLE = MODAL TRANSIENT RESPONSE ANALYSIS
$
SPC = 1
$
METHOD = 777
$
DLOAD = 2
LOADSET = 3
SDAMPING = 4
TSTEP = 5
$
$ OUTPUT REQUEST
SET 123 = 999
DISPLACEMENT(PLOT)=123
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10......
$
$ NORMAL MODES TO 3000 HZ
$EIGRL SID V1 V2
EIGRL 777 -0.1 3000.
$
$ 4 SECONDS OF RESPONSE
$TSTEP SID N1 DT1 NO1
TSTEP 5 800 0.005 1
$
$ MODAL DAMPING OF 2% CRITICAL
$TABDMP1 TID TYPE +TABD1
$+TABD1 F1 G1 F2 G2 ETC.
TABDMP1 4 CRIT +TABD1
+TABD1 0.0 0.02 3000.0 0.02 ENDT
$
$ LOAD DEFINITION
$
$TLOAD1 SID DAREA DELAY TYPE TID
TLOAD1 2 999 22
$
$LSEQ SID DAREA LID TID
LSEQ 3 999 1
$
$ TIME HISTORY
$TABLED1 TID +TABL1
$+TABL1 X1 Y1 X2 Y2 ETC.
TABLED1 22 +TABL1
+TABL1 0.0 0.0 0.1 0.0 0.15 1.0 5.0 1.0 +TABL2
+TABL2 ENDT
$
$ PRESSURE LOAD OF 3 PSI PER ELEMENT
$PLOAD4 SID EID P1
PLOAD4 1 171 -3.
PLOAD4 1 172 -3.
PLOAD4 1 160 -3.
etc.
$
... basic model ...
$
ENDDATA
196

Table 6-10 Relationship Between Case Control Commands and

Bulk Data Entries for the Bracket Model

Case Control Bulk Data

METHOD EIGRL
FREQUENCY FREQ1
SDAMPING TABDMP1
PLOAD4
LOADSET LSEQ

(DAREA ID)

DLOAD RLOAD1
TABLED1

Figure 6-17 shows a plot of the z-displacement of grid point 999, which is the concentrated
mass at the center of the cutout.

0
Displacement (m)

-0.04

-0.08
0 Time (sec) 4
Figure 6-17 Displacement Time History for Grid Point 999
 CHAPTER 6 197
Transient Response Analysis
198
 NX Nastran Basic Dynamic Analysis User’s Guide

7
CHAPTER
Enforced Motion

■ Overview
■ The Large Mass Method in Direct Transient and Direct Frequency
Response
■ The Large Mass Method in Modal Transient and Modal Frequency
Response

■ User Interface for the Large Mass Method

■ Examples
200

7.1 Overview
Enforced motion specifies the displacement, velocity, and/or acceleration at a set of grid
points for frequency and transient response. Enforced motion is used when base motion is
specified instead of or in conjunction with applied loads. A common application is an
earthquake excitation applied to a building. In this case there are no applied loads, instead
the base of the building undergoes an enforced displacement or acceleration time history.

MSC.Nastran does not include a completely automatic method for prescribing enforced
motion in dynamics. Instead, the procedures described in “Frequency Response
Analysis” on page 103 and “Transient Response Analysis” on page 151 for specifying
applied forces are used in conjunction with techniques that convert applied forces into
enforced motion. One such method that is applicable to both transient response and
frequency response is described in this chapter—the large mass method. Another method
for enforced motion, described in the NX Nastran Advanced Dynamic Analysis User’s Guide,
is the Lagrange multiplier technique.
 CHAPTER 7 201
Enforced Motion

7.2 The Large Mass Method in Direct Transient and Direct

Frequency Response
If a very large mass m o , which is several orders of magnitude larger than the mass of the
entire structure, is connected to a degree-of-freedom and a dynamic load p is applied to the
same degree-of-freedom, then the acceleration of the degree-of-freedom, to a close
approximation, is as follows:

1
u·· = ------- p Eq. 7-1
mo

In other words, the load that produces a desired acceleration u·· is approximately

p = m o u·· Eq. 7-2

The accuracy of this approximation increases as m o is made larger in comparison to the

mass of the structure. The only limit for the size of m o is numeric overflow in the computer.
MSC recommends that the value of m o be approximately 10 6 times the mass of the entire
structure for an enforced translational degree-of-freedom and 106 times the mass moment
of inertia of the entire structure for a rotational DOF. The factor 10 6 is a safe limit that
should produce approximately six digits of numerical accuracy.
The large mass method is implemented in direct transient and frequency response analysis
by placing large masses m o on all enforced degrees-of-freedom and supplying applied
dynamic loads specified by Eq. 7-2; that is, the function u·· is input on entries normally used
for the input of loads, and the scale factor m o can be input on DAREA or DLOAD Bulk Data
entries, whichever is more convenient. CMASSi or CONMi entries should be used to input
the large masses.

Eq. 7-2 is not directly helpful if enforced displacement or enforced velocity is specified
rather than enforced acceleration. However, Eq. 7-2 can be made serviceable in frequency
response analysis by noting that

2
u·· = iωu· = – ω u Eq. 7-3

so that

2
p = m o u·· = iωm o u· = – ω m o u Eq. 7-4

The added factor ( iω or – ω 2 ) can be carried by the function tabulated on the TABLEDi entry
used to specify the frequency dependence of the dynamic load.
In the case of transient analysis, provision is made on the TLOAD1 and TLOAD2 entries
for you to indicate whether an enforced displacement, velocity, or acceleration is supplied
(TYPE = 1, 2, or 3). MSC.Nastran then automatically differentiates a specified velocity once
or a specified displacement twice to obtain an acceleration. The remaining required user
actions are the same as for enforced acceleration.
202

In summary, the user actions for direct frequency and direct transient response are
• Remove any constraints from the enforced degrees-of-freedom.
• Apply large masses m o with CMASSi or CONMi Bulk Data entries to the DOFs
where the motion is enforced. The magnitude of m o should be approximately 10 6
times the entire mass of the structure (or approximately 10 6 times the entire mass
moment of inertia of the structure if the component of enforced motion is a
rotation).
• In the case of direct frequency response, apply a dynamic load computed
according to Eq. 7-4 to each enforced degree-of-freedom.
• In the case of direct transient response,
• Indicate in field 5 of the TLOAD1 and TLOAD2 entries whether the
enforced motion is a displacement, velocity, or acceleration.
• Apply a dynamic load to each enforced degree-of-freedom equal to mo u ,
m o u· , or m o u·· , depending on whether the enforced motion is a displacement,
velocity, or acceleration.
Be careful when using PARAM,WTMASS. The WTMASS parameter multiplies the large
mass value, which changes the effective enforced acceleration to

p
u·· = -------------------------------------- Eq. 7-5
m o ⋅ WTMASS

Enforced velocity and enforced displacement are changed likewise.

You may well ask whether a stiff spring may be used instead of a large mass. In that case
the applied load is

p = ko u Eq. 7-6

where k o is the stiffness of the stiff spring and u is the enforced displacement. The large
stiffness method certainly works, but the large mass method is preferred because it is easier
to estimate a good value for the large mass than to estimate a good value for the stiff spring.
In addition and more importantly, the large mass method is far superior when modal
methods are used. If very stiff springs are used for modal analysis rather than very large
masses, the vibration modes corresponding to the very stiff springs have very high
frequencies and in all likelihood, are not included among the modes used in the response
analysis. This is the main reason that large masses should be used instead of stiff springs.
The stiff spring method is advantageous in the case of enforced displacement because it
avoids the roundoff error that occurs while differentiating the displacement to obtain
acceleration in the large mass method. The stiff spring method also avoids the problem of
rigid-body drift when applying enforced motion on statically determinate support points.
(Rigid-body drift means that the displacement increases continuously with time, which is
often caused by the accumulation of small numerical errors when integrating the equations
of motion.)
 CHAPTER 7 203
Enforced Motion

7.3 The Large Mass Method in Modal Transient and Modal

Frequency Response
The steps described in the previous section must also be followed when a modal method
of response analysis is used. Also, if the enforced degrees of freedom are not sufficient to
suppress all rigid-body motions, which may be the case for an airplane in flight, additional
DOFs that describe the remaining rigid-body motions can also be entered on the SUPORT
entry. Use of the SUPORT entry is discussed in “Rigid-body Modes” on page 87. Note
that the use of the SUPORT entry is optional.
The rigid-body mode(s) can be removed from consideration either by not computing them
or by using PARAM,LFREQ,r where r is a small positive number (0.001 Hz, for example).
If this is done, the displacements, velocities, and accelerations obtained are relative to the
overall motion of the structure and are not absolute response quantities. Stresses and
element forces are the same as when the rigid-body modes are included because the
rigid-body modes do not contribute to them. Rigid-body modes can be discarded to
remove rigid-body drift.
Rigid-body modes occur when the structure is unconstrained and large masses are applied
at the DOFs which, if constrained, result in a statically determinate structure. Redundant
constrained DOFs, which result in a statically indeterminate structure, present a different
situation when the constraints are removed and large masses are applied at those
redundant DOFs. In that case, very low-frequency modes occur but they are not all
rigid-body modes; some are modes that represent the motion of one large mass relative to
the others. These very low-frequency, nonrigid-body modes do contribute to element
forces and stresses, and they must be retained in the solution. In some cases, their
frequencies are not necessarily small; they may be only an order or two in magnitude less
than the frequency of the first flexible mode. If PARAM,LFREQ,r is used to remove the
true rigid-body modes, then r must be set below the frequency of the first relative motion
mode.
Consider the two-dimensional clamped bar in Figure 7-1. Assume that each end of the bar
is subjected to the same enforced acceleration time history in the y-direction. One way to
model the bar is to use two large masses (one at each end), which are unconstrained in the
y-direction. This model provides two very low-frequency modes: one that is a rigid-body
mode and one that is not. The second mode contributes to element forces and stresses, and
removing its contribution leads to an error because with two such large masses, one mass
can drift over time relative to the other mass. A better way to model the case of identical
inputs at multiple locations is to use one large mass connected to the end points by an RBE2
element. This model provides only one rigid-body mode, which can be safely discarded if
only the answers relative to the structure are desired.
204

·· ··
u(t) u(t)

x
Basic Model

p( t) p( t)

Two Large Masses at Each End

RB E2
E2 RB

p( t)

One Large Mass and One RBE2 Element

Figure 7-1 Clamped-Clamped Bar Undergoing Enforced Acceleration
 CHAPTER 7 205
Enforced Motion

7.4 User Interface for the Large Mass Method

There is no special user interface for the large mass method other than to specify a large
mass at excitation DOFs and to specify the large force. For transient response the type of
enforced motion (displacement, velocity, or acceleration) must be specified on the TLOADi
entries. The remainder of the input is identical to that of frequency response (“Frequency
Response Analysis” on page 103) or transient response (“Transient Response Analysis”
on page 151) analysis.

The force applied at a point is the product of terms from the DLOAD, DAREA, and
TABLEDi entries. The scaling of the large force is arbitrary; it can be on any one or more
of these entries. These entries follow:
206

DLOAD Format

1 2 3 4 5 6 7 8 9 10
DLOAD SID S S1 L1 S2 L2 -etc.-

DAREA Format

DAREA SID P1 C1 A1 P2 C2 A2

TABLED1 Format

TABLED1 TID

x1 y1 x2 y2 x3 y3 -etc.- ENDT

TABLED2 Format

TABLED2 TID X1

x1 y1 x2 y2 x3 y3 -etc.-

TABLED3 Format

TABLED3 TID X1 X2

x1 y1 x2 y2 x3 y3 -etc.-

TABLED4 Format

TABLED4 TID X1 X2 X3 X4

A0 A1 A2 A3 A4 A5 -etc.-

The TABLED4 entry defines a power series and is convenient in frequency response for
enforced constant velocity or displacement.

Frequency Response
If a DLOAD entry is used to scale RLOAD1 input, the applied force magnitude in terms of
MSC.Nastran input is

Pi ( f ) = S ⋅ ∑ ( Si ⋅ Ai [ Ci ( f ) + i ⋅ D i ( f ) ] ) Eq. 7-7
i

where S and Si are input on the DLOAD Bulk Data entry, A i is input on the DAREA entry,
and C i ( f ) and D i ( f ) are input on the TABLEDi entries. Note that the “i” non-subscript term
in the expression i ⋅ D i ( f ) is – 1 .
 CHAPTER 7 207
Enforced Motion

If a DLOAD entry is used to scale RLOAD2 input, the applied force magnitude in terms of
MSC.Nastran input is

Pi ( f ) = S ⋅ ∑ ( Si ⋅ Ai ⋅ Bi ( f ) ) Eq. 7-8
i

where S and S i are input on the DLOAD Bulk Data entry, A i is input on the DAREA entry,
and Bi ( f ) is input on the TABLEDi entry.
Specification of the large force value depends upon whether acceleration, velocity, or
displacement is enforced.

Enforced Acceleration
Enforced acceleration is the easiest to apply since the required force is directly proportional
to the desired acceleration times the large mass:

p ( ω ) = m o u·· ( ω ) Eq. 7-9

Enforced Velocity
Enforced velocity requires a conversion factor

p ( ω ) = iωm o u· = i ( 2πf )m o u· ( ω ) Eq. 7-10

For constant velocity u· ( ω ) , it may be easiest to use the RLOAD1 and TABLED4 entries
because the imaginary term i ⋅ D ( f ) of Eq. 7-7 and the frequency-dependent term 2πf can
be specified directly.

Enforced Displacement
Enforced displacement also requires a conversion factor

2 2
p ( ω ) = – ω m o u = – ( 2πf ) m o u ( ω ) Eq. 7-11

For constant displacement u ( ω ) , it may be easiest to use the TABLED4 entry because the
frequency-dependent term – ( 2πf )2 can be specified directly.

Transient Response
For transient response, the type of enforced motion (displacement, velocity, or
acceleration) is specified with the TYPE field (field 5) on the TLOAD1 and TLOAD2 Bulk
Data entries.
208

TLOAD1 Format
1 2 3 4 5 6 7 8 9 10
TLOAD1 SID DAREA DELAY TYPE TID

TLOAD2 Format
TLOAD2 SID DAREA DELAY TYPE T1 T2 F P

C B

TYPE 0 (or blank) = applied force (default)

TYPE 1 = enforced displacement
TYPE 2 = enforced velocity
TYPE 3 = enforced acceleration
MSC.Nastran converts enforced displacements and velocities into accelerations by
differentiating once for velocity and twice for displacement. Note that for enforced
acceleration, you can specify either force (TYPE = 0 or blank) or acceleration (TYPE = 3);
they are the same for the large mass method.

You still need to use the large mass when specifying any type of enforced motion in
transient response analysis.
 CHAPTER 7 209
Enforced Motion

7.5 Examples
This section provides several examples showing the input and output. These examples are:

Model Analysis Type Enforced Motion

bd07two Frequency Response Constant Acceleration
bd07bar1 Transient Response Ramp, Acceleration
bd07bar2 Transient Response Ramp, Displacement
bd07bar3 Transient Response Ramp, Displacement,
Discard Rigid-Body Mode

These examples are described in the sections that follow.

Two-DOF Model
Consider the two-DOF model first introduced in “Real Eigenvalue Analysis” on page 35
and shown below in Figure 7-2. For this example, apply a constant magnitude base
acceleration of 1.0 m/sec 2 over the frequency range of 2 to 10 Hz and run modal frequency
response with 5% critical damping in all modes. The acceleration input is applied to the
large mass (grid point 3). The input file for this model is shown in Listing 7-1.

Grid Point 1 m 1 = 0.1 kg

Grid Point 2 m 2 = 10.0 kg

Grid Point 3 Large 7

m 3 = 10 kg
Mass
x

Figure 7-2 Two-DOF Model with Large Mass

210

Listing 7-1 Input File for Enforced Constant Acceleration

$ FILE bd07two.dat
$
$ TWO-DOF SYSTEM
$ CHAPTER 7, ENFORCED MOTION
$
TIME 5
SOL 111
CEND
TITLE = TWO-DOF SYSTEM
SUBTITLE = MODAL FREQUENCY RESPONSE
LABEL = ENFORCED CONSTANT ACCELERATION MAGNITUDE
$
$ SPECIFY MODAL EXTRACTION
METHOD = 10
$
$ SPECIFY DYNAMIC INPUT
DLOAD = 999
FREQ = 888
SDAMPING = 777
$
$ SELECT OUTPUT
DISPLACEMENT(PHASE,PLOT) = ALL
ACCELERATION(PHASE,PLOT) = ALL
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ LARGE MASS AT BASE GRID POINT
CONM2 999 3 1.0E7
$
$ LOAD DEFINITION (INCLUDES SCALE FACTORS FOR ENFORCED ACCELERATION)
$DLOAD SID S S1 RLOAD1
DLOAD 999 1.0E7 1.0 998
$RLOAD1 SID DAREA TC
RLOAD1 998 997 901
$DAREA SID P1 C1 A1
DAREA 997 3 2 1.0
$TABLED4 TID X1 X2 X3 X4 +TAB4
TABLED4 901 0. 1. 0. 100. +TAB901
$+TAB4 A0 A1 A2 A3 A4 A5
+TAB901 1.0 ENDT
$
$ MODAL EXTRACTION
$EIGRL SID V1 V2 ND MSGLVL
EIGRL 10 -1. 30. 0
$
$ FREQUENCY RANGE 2-10 HZ
$FREQ1 SID F1 DF NDF
FREQ1 888 2. 0.05 160
$
$ MODAL DAMPING OF 5% CRITICAL
$TABDMP1 TID TYPE
$+TAB1 F1 G1 F2 G2 ETC
TABDMP1 777 CRIT +TABD7
+TABD7 0. 0.05 100. 0.05 ENDT
$
... basic model ...
$
ENDDATA
 CHAPTER 7 211
Enforced Motion

The large mass value is chosen as 1.0E7 kilograms and is input via the CONM2 entry. The
scale factor for the load (1.0E7) is input on the DLOAD Bulk Data entry. The factor of 1.0E7
is approximately six orders of magnitude greater than the overall structural mass (10.1 kg).
The TABLED4 entry defines the constant acceleration input. (One of the other TABLEDi
entries can also be used, but the TABLED4 entry is chosen to show how to use it for
enforced constant velocity and displacement later in this example.)

Figure 7-3 shows the X-Y plots resulting from the input point (grid point 3) and an output
point (grid point 1). The plots show acceleration and displacement magnitudes. Note that
the acceleration input is not precisely 1.0 m/sec 2 ; there is a very slight variation between
0.9999 and 1.0000 due to the large mass approximation.

Figure 7-3 Displacements and Accelerations for the Two-DOF Model

212

Figure 7-3 Displacements and Accelerations for the Two-DOF Model (continued)
This model was analyzed with several values of large mass. Table 7-1 shows the results.
Note that the model with the 106 mass ratio is the model discussed earlier. Peak frequency
response results are compared for each model, and the natural frequencies are compared
to those of the constrained model in “Real Eigenvalue Analysis” on page 35. The table
shows that a mass ratio of 10 6 is a good value to use for this model.
 CHAPTER 7 213
Enforced Motion

Table 7-1 Models with Different Large Mass Ratios

2
Ratio of Natural Response Peaks (m/sec )
Large Mass Frequencies* ·· ·· ·· ··
to Structure (Hz) X1 X2 X3 X3
max max max min

-9
10
2 52.0552 6.5531 1.0335 0.9524
10 4.8011
5.3025
-10
10
4 52.2823 6.7641 1.0003 0.9995
10 4.7877
5.2910
0.0
6 4.7876 52.2836 6.7661 1.0000 0.9999
10
5.2909
0.0
8 4.7876 52.2836 6.7662 1.0000 1.0000
10
5.2909

* Resonant frequencies for the constrained model are 4.7876 and 5.2909 Hz.

This model can also be changed to apply constant velocity or constant displacement at its
base. Listing 7-2 is an abridged input file for the model, showing the Bulk Data entries
required for enforced constant acceleration, enforced constant velocity, and enforced
constant displacement. Note that only one of these is usually applied to any model, but all
three are shown here for comparison purposes.
214

Listing 7-2 Bulk Data Entries for Enforced Constant Motion

$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ ENTRIES FOR ENFORCED MOTION
$
$ LARGE MASS AT BASE GRID POINT
CONM2 999 3 1.0E7
$
$ LOAD DEFINITION
$DLOAD SID S S1 RLOAD1
DLOAD 999 1.0E7 1.0 998
$DAREA SID P1 C1 A1
DAREA 997 3 2 1.0
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ UNIQUE ENTRIES FOR ENFORCED CONSTANT ACCELERATION MAGNITUDE
$
$RLOAD1 SID DAREA TC
RLOAD1 998 997 901
$TABLED4 TID X1 X2 X3 X4 +TAB4
TABLED4 901 0. 1. 0. 100. +TAB901
$+TAB4 A0 A1 A2 A3 A4 A5
+TAB901 1.0 ENDT
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ UNIQUE ENTRIES FOR ENFORCED CONSTANT VELOCITY MAGNITUDE
$
$RLOAD1 SID DAREA TD
RLOAD1 998 997 902
$TABLED4 TID X1 X2 X3 X4 +TAB4
TABLED4 902 0. 1. 0. 100. +TAB902
$+TAB4 A0 A1 A2 A3 A4 A5
+TAB902 0.0 6.283185 ENDT
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$ $
$
$ UNIQUE ENTRIES FOR ENFORCED CONSTANT DISPLACEMENT MAGNITUDE
$
$RLOAD1 SID DAREA TC
RLOAD1 998 997 903
$TABLED4 TID X1 X2 X3 X4 +TAB4
TABLED4 903 0. 1. 0. 100. +TAB903
$+TAB4 A0 A1 A2 A3 A4 A5
+TAB903 0.0 0.0 -39.4784 ENDT

Each input utilizes the TABLED4 entry. The TABLED4 entry uses the algorithm
N
i
A i  -----------------
x – X1
Y = ∑  X2  Eq. 7-12
i = 0

where x is input to the table, Y is returned, and N is the degree of the power series. When
x < X3 , X3 is used for x ; when x > X4 , X4 is used for x . This condition has the effect of
placing bounds on the TABLED4 entry; note that there is no extrapolation outside of the
table boundaries. There are N+1 entries to this table.
 CHAPTER 7 215
Enforced Motion

Constant acceleration is the easiest to apply since the force is proportional to the mass for
all frequencies. The power series for this case becomes

A0 + A1  -----------------
x – X1
Eq. 7-13
X2

where:

A0 = 1.0
X1 = 0.0
X2 = 1.0

Therefore, these terms define a constant (1.0 in this case).

Constant velocity involves a scale factor that is directly proportional to circular frequency
( 2πf ). The power series for this case becomes

A0 + A1  -----------------
x – X1
Eq. 7-14
X2

where:

A0 = 0.0
A1 = 2π = 6.283185
X1 = 0.0
X2 = 1.0

Note that a phase change of 90 degrees is also required; this change is input using the TD
field (field 7) of the RLOAD1 entry.
Constant displacement involves a scale factor that is proportional to the circular frequency
squared ( 2πf ) 2 with a sign change. The power series for this case becomes

x – X1 2
A0 + A1  ----------------- + A2  -----------------
x – X1
Eq. 7-15
 X2   X2 

where:

A0 = 0.0
A1 = 0.0

A2 = – ( 2π ) 2 = – 39.4784

X1 = 0.0
X2 = 1.0
216

Table 7-2 summarizes the coefficients for the power series.

Table 7-2 Coefficients for the Power Series

Type of
A0 A1 A2
Excitation
2
Enforced u 0.0 0.0 – ( 2π )
·
Enforced u 0.0 2π –
··
Enforced u 1.0 – –

Cantilever Beam Model

Consider the cantilever beam first introduced in “Frequency Response Analysis” on
page 103 and shown in Figure 7-4. In this case the planar model is analyzed for bending;
therefore, only three DOFs per grid point are considered: T1 (x-translation), T2 (y-
translation), and R3 (z-rotation). An acceleration ramp function in the y-direction is
enforced at the base (grid point 1) by applying a large mass and a force. T1 and R3 are
constrained at grid point 1 since the enforced motion is in only the T2-direction. Modal
transient response analysis (SOL 112) is run with 5% critical damping used for all modes.
Modes up to 3000 Hz are computed with the Lanczos method. Figure 7-5 shows the
idealized ramp function and the MSC.Nastran implementation. The excitation is not cut
off abruptly; instead it is cut off over two time steps. A time step of 0.001 second is used,
and the analysis is run for 1.0 second. Listing 7-3 shows the abridged input file.

Large Mass x
1.0E9 kg

Grid Point 1
Figure 7-4 Beam Model with Large Mass
 CHAPTER 7 217
Enforced Motion

0.15

Idealized Input
·· 2
Y (m/sec )

0 0.05 Time (sec)

0.15

MSC.Nastran Input
·· 2
Y (m/sec )

Time (sec)
0.05
0.052
Figure 7-5 Idealized Ramp Function Versus MSC.Nastran Ramp Function
218

Listing 7-3 Abridged Input File for Enforced Acceleration

$ FILE bd07bar1.dat
$
$ CANTILEVER BEAM MODEL
$ CHAPTER 7, ENFORCED MOTION
$
SOL 112
TIME 10
CEND
TITLE = CANTILEVER BEAM
SUBTITLE = MODAL TRANSIENT RESPONSE
LABEL = ENFORCED ACCELERATION
$
SPC = 21
DLOAD = 22
TSTEP = 27
SDAMPING = 25
$
METHOD = 10
$
$ PHYSICAL OUTPUT REQUEST
SET 11 = 1,11
DISPLACEMENT(PLOT) = 11
ACCELERATION(PLOT) = 11
$
$ XYPLOTS
$
... X-Y plot commands ...
$
BEGIN BULK
$
$.......2.......3.......4.......5.......6.......7.......8.......9.......10.....$
$
$ LARGE MASS OF 1.0E9
$CONM2 EID G CID M
CONM2 15 1 1.0E9
$
$ CONSTRAIN MASS IN 1,6 DIRECTIONS
SPC 21 1 16
$
$ DYNAMIC LOADING
$DLOAD SID S S1 L1
DLOAD 22 1.0E9 0.102 23
$TLOAD1 SID DAREA DELAY TYPE TID
TLOAD1 23 24 0 25
$DAREA SID P1 C1 A1
DAREA 24 1 2 0.15
$TABLED1 TID +TABL1
$+TABL1 X1 Y1 X2 Y2 ETC.
TABLED1 25 +TABL1
+TABL1 0.0 0.0 0.05 1.0 0.052 0.0 0.1 0.0 +TABL2
+TABL2 ENDT
$
$ CONVERT WEIGHT TO MASS: MASS = (1/G)*WEIGHT
$ G = 9.81 m/sec**2 --> WTMASS = 1/G = 0.102
PARAM WTMASS 0.102
$
$EIGRL SID V1 V2 MSGLVL
EIGRL 10 -1. 3000. 0
$
$TSTEP SID N1 DT1 NO1
TSTEP 27 1000 0.001 1
$
$ MODAL DAMPING OF 5% IN ALL MODES
$TABDMP1 TID TYPE +TABD
$+TABD F1 G1 F2 G2 ETC.
TABDMP1 25 CRIT +TABD
+TABD 0. 0.05 1000. 0.05 ENDT
$
... basic model ...
$
ENDDATA
 CHAPTER 7 219
Enforced Motion

A large mass of 1.0E9 kg is placed at grid point 1. This grid point is constrained in the T1-
and R3-directions but is free in the T2-direction. The load is scaled to give a peak input
acceleration of 0.15 m/sec 2 . This scaling is performed by applying a scale factor of 1.0E9 in
the S field (field 3) of the DLOAD entry, a scale factor of 0.102 in the S1 field (field 4) of the
DLOAD entry, and a factor of 0.15 in the A1 field (field 5) of the DAREA entry. The applied
load is scaled by 0.102 because the large mass is also scaled by 0.102 due to the
PARAM,WTMASS entry (see Eq. 7-5). The time variation is specified with the TABLED1
entry. The TLOAD1 entry specifies the type of loading (field 5) as 0 (applied force); this
gives the same answers if the type is specified as 3 (enforced acceleration).

Figure 7-6 shows the displacement and acceleration response at grid points 1 (base) and
11 (tip). Note that at the end of the acceleration pulse the base has a constant velocity, and
therefore, a linearly increasing displacement.

Figure 7-6 Response for Enforced Acceleration

Next, consider the same model with a 0.015 meter displacement imposed instead of an
acceleration. The same ramp time history function is used (with a peak enforced
displacement of 0.015 meter) so that the only change to the input file is to change the
excitation type from 0 (applied force) to 1 (enforced displacement on field 5 of the TLOAD1
220

entry) and the amplitude in the DAREA entry from 0.15 to 0.015. Figure 7-7 shows the
idealized input displacement time history. Figure 7-8 shows the displacement and
acceleration response at grid points 1 and 11.

0.15

Idealized Input
Y(m)

0 0.05 Time (sec)

Figure 7-7 Response for Enforced Displacement (With the Rigid-Body Mode)

Figure 7-8 Response for Enforced Displacement (Without the Rigid-Body Mode)

Now, consider a change to the enforced displacement run. In this case, remove the rigid-
body mode's contribution either by not computing the rigid-body mode (by setting V1 to
a small positive value, such as 0.01 Hz) or by neglecting the rigid-body mode in the
transient response (by setting PARAM,LFREQ to a small positive number, such as
0.01 Hz). Figure 7-9 shows the resulting displacement and acceleration responses at grid
points 1 and 11. Note that the responses are relative to the structure and are not absolute.
The relative displacement of grid point 1 should be zero, and it is very close to zero (i.e.,
– 10
10 ) as a result of the sufficiently large mass.
 CHAPTER 7 221
Enforced Motion

Figure 7-9 Response for Enforced Displacement (Without the Rigid-Body Mode)
222
 NX Nastran Basic Dynamic Analysis User’s Guide

8
CHAPTER
Restarts In Dynamic Analysis

■ Overview
■ Automatic Restarts

■ Structure of the Input File

■ User Interface
■ Determining the Version for a Restart
■ Examples
224

8.1 Overview
A restart is a logical way of continuing from a previous run without having to start from
the beginning. The savings can be in terms of both time and money. Some forms of restarts
are used practically every day perhaps without you realizing it. An example of a restart
can be as simple as reading a book. Normally, you probably do not finish reading a book
in one continuous stretch. You may read a hundred pages today and another fifty pages
tomorrow, and so on. Each time that you continue from where you left off previously is a
restart. It is much more time consuming and impractical to start from page one every time
that you pick up the book.

This analogy can be applied to MSC.Nastran. In the case of a static analysis, the most
expensive and time-consuming part of the run is the decomposition of the stiffness matrix.
This fact is especially true for large models. Now suppose after completing the original
run, you want to obtain additional output (e.g., stresses, displacements, etc.) or add more
load cases. You can always start from the beginning and redo the whole problem, or you
can perform a restart at a fraction of the time and cost. In the case of additional output
requests and additional load conditions, the decomposition of the stiffness matrix, which
was performed in the previous run, is not redone if restart is used.

In dynamic analysis, the calculation of normal modes is, in general, the most expensive
operation. Therefore, a common application of restart is the performance of a transient or
frequency response analysis by restarting from the normal modes calculation, which was
saved in the database from a previous run. This restart process avoids the recalculation of
the normal modes.

The cost of restarting is measured in the disk space required to store data blocks for
subsequent use. Judging whether to not save data blocks and simply rerun the analysis or
to save data blocks for restarting is determined by several factors, such as the amount of
available disk space and your computer’s solution speed. Note that the database can be
copied to tape which then provides you with more free disk space. When you are ready to
perform a restart, this database can then be copied from the tape back to your disk.
 CHAPTER 8 225
Restarts In Dynamic Analysis

8.2 Automatic Restarts

In Version 66, we introduced a new Executive System that more effectively uses modern
database technology. Among its many features, the new Executive System implements the
NASTRAN Data Definition Language (NDDL). With NDDL, MSC.Nastran is able to use
automatic restart logic with the superelement solution sequences. For more information
on the NDDL, refer to the NX Nastran DMAP Programmer’s Guide.

In order to maintain upward compatibility, we retained the old solution sequences and
added a new series of superelement solution sequences in Version 66. These solution
sequences are also known as the Structured Solution Sequences (solution sequences
numbers greater than 100). The automatic restart logic is available only for these solution
sequences. Restart is no longer available for the rigid format solution sequences. Use these
new Structured Solution Sequences rather than the old ones since all the new features are
automatically implemented in these new solution sequences. Improvements made to the
automatic restart logic for Version 67 and later, have made it even more efficient and
robust. In this chapter, we will address the restart logic for the Structured Solution
Sequences. Note that you are not required to use superelements in order to use the
superelement solution sequences. If you are familiar with the rigid format solutions,
converting to the new Structured Solution Sequences only requires that you replace the
solution command. For example, a normal modes run only requires replacing the SOL 3
command with the SOL 103 command in the Executive Control Section.
226

8.3 Structure of the Input File

Before presenting details on how restarts work, the following simple flow diagram of the
MSC.Nastran input file structure may be beneficial:

NASTRAN Statements Optional

File Management Optional

Statements

Executive Control
Statements

CEND Required Delimiter

Case Control
Commands

BEGIN BULK Required Delimiter

Bulk Data Entries

ENDDATA Required Delimiter

This order must be followed as shown for all MSC.Nastran input files. For details
regarding the statements, commands, and entries, see the NX Nastran Quick Reference
Guide.
 CHAPTER 8 227
Restarts In Dynamic Analysis

8.4 User Interface

There are two types of runs—cold start and restart runs—which are described below.

Cold Start Run

The first run, which is called the cold start run, is identical to what you usually do for
submitting an MSC.Nastran job with the exception that the database must be saved. In
addition to your normal output files (e.g., F06 file), four database files are created as a result
of this run, assuming that the default database initialization values are used. The
convention for the filenames is machine dependent; therefore, you should refer to the
machine-dependent documentation for your specific computer for the exact syntax. For a
typical UNIX machine with an MSC.Nastran input file called “run1.dat”, the following
sample submittal command can be used:
nastran run1
where “nastran” is the name of the shell script for executing MSC.Nastran. Note that the
above submittal command assumes that you have not modified the “scr=no” option as
your default nastran submittal option. The “scr=no” option saves the database at the end
of the run, which is required if you intend to perform restarts. The following four database
files are created as a result of the above command.
run1.DBALL
run1.MASTER
run1.USROBJ
run1.USRSOU
The last two files are not needed and can be deleted if you do not want to store your own
DMAP, which is usually the case.

Restart Run
NASTRAN Statement Section
This section is normally the same as your cold start run. The BUFFSIZE must not be
changed in a restart run.

File Management Section (FMS)

This section tells MSC.Nastran that you are performing a restart run. The RESTART
statement is required in any restart run. The general format for the RESTART statement is
as follows:

RESTART VERSION=a,b
where “a” is the version from which you restart (default value for “a” is LAST) and “b”
indicates whether version “a” is to be kept (KEEP) or deleted (NOKEEP) at the end of the
run. The default value for “b” is NOKEEP. Due to the default values, the following two
restart statements are identical:
228

RESTART VERSION=LAST,NOKEEP
RESTART
Whenever a restart is performed, a new version number is automatically created in the
database. For each restart, the current version number is incremented by one regardless of
whether the job ran successfully or not. There are two exceptions to this rule and they are
discussed later on. You also need to tell MSC.Nastran which database you want to attach
to your current run. There are several ways to accomplish this; one way is to use the
ASSIGN statement. For example, if you are restarting from the database created by
run1.dat, the following FMS statements can be used in your current run:

RESTART
ASSIGN MASTER=’run1.MASTER’
For UNIX machines, the filenames are case sensitive. They should be entered exactly as
they were created in the cold start run and enclosed with single quotes as shown above.
An alternate way is to use the DBS statement on the submittal line instead of the ASSIGN
statement. Assuming the current run is called run2.dat, then the equivalent submittal
statement is as follows:

nastran run2 dbs=run1

The ASSIGN statement is not needed in this case as long as the run1 database files are
contained in the submittal directory. Again the syntax is for a typical UNIX machine. You
should refer to the machine-dependent documentation for your particular computer for
specific details.

Executive Control Section

This section is the same as your normal run with the exception of perhaps the “SOL x”
command. For example, if you are performing a normal modes analysis in run1.dat, then
the “SOL x” command in run1.dat should reference “SOL 103". In run2.dat, if you are
performing a modal transient restart from run1.dat, then the “SOL x” statement in this case
should then reference “SOL 112". DIAGs can be turned on or off.

Case Control Section

The automatic restart logic compares the modifications made to the Case Control and Bulk
Data Sections in determining which operations need to be processed upon restart.
Therefore, you must be very careful with the changes that you make in your restart run.
Adhering to the following rules will avoid unnecessary reprocessing of previously
completed operations.
• You must include all “solution-type” related Case Control commands, which are
unchanged as compared to the cold start run, in your restart run. In other words,
do not make unnecessary LOAD, SPC, MPC, or METHOD command changes or
remove them from the Case Control Section unless these are actual changes. This
process is clarified later with the example problems.
• Output requests can be modified. A typical example can be a request of the
eigenvector printout which was not requested in the cold start run.
 CHAPTER 8 229
Restarts In Dynamic Analysis

Bulk Data Section

As mentioned in the previous section, the automatic restart logic compares the changes
made in the Bulk Data Section and determines the path that it follows. A copy of the Bulk
Data is stored for each version. The restart run must not contain any Bulk Data entry that
was included in the previous runs and saved in the database. The Bulk Data Section in the
current restart run should contain only new entries, changed entries, and/or the deletion
of old entries from the database. This philosophy is slightly different than the one used in
the Case Control Section.
Table 8-1 shows an example where a normal modes run is performed in the cold start run.
The eigenvalue table and the grid point weight generator output are requested in this run.
For the restart run, the eigenvector output is desired; furthermore, reprinting the grid point
weight generator output is not desired. One way to accomplish this is to delete the existing
“PARAM,GRDPNT,0" entry from the database.
230

Table 8-1 Listing of the Cold Start and Restart Input Files

Cold Start Run Restart Run

$ $
$ FILENAME - run1.dat $ FILENAME - run2.dat
$ $
$ TYPICAL UNIX SUBMITTAL $ TYPICAL UNIX SUBMITTAL
$ COMMAND $ COMMAND
$ $
$ nastran run1 $ nastran run2
$ $
SOL 103 RESTART
TIME 5 ASSIGN MASTER=’run1.MASTER’
CEND SOL 103
TITLE = COLDSTART RUN TIME 5
METHOD = 10 CEND
SPC = 1 TITLE = RESTART RUN
BEGIN BULK METHOD = 10
$ SPC = 1
GRID,1,,0.,0.,0. $
GRID,2,,10.,0.,0. $ REQUEST FOR EIGENVECTOR
CROD,1,10,1,2 $ PRINTOUT
PROD,10,1,1.0 $
MAT1,1,1.+7,,.32 DISP = ALL
EIGRL,10,,,1 BEGIN BULK
CONM2,100,2,,10. $
PARAM,WTMASS,.00259 $ SKIP GRID POINT WEIGHT
$ $ GENERATOR OUTPUT
$ REQUEST FOR GRID POINT $
$ WEIGHT GENERATOR /,7
$ OUTPUT $
$ ENDDATA
PARAM,GRDPNT,0
$
SPC1,1,123456,1
SPC1,1,23456,2
$
ENDDATA

The format for the delete entry is “/,K1,K2" where K1 and K2 are the sorted Bulk Data
sequence number of the first and last entries in the sequence to be removed, respectively.
In the event that K2 = K1, the following two entries are identical:

/,K1,K2
/,K1
The values of K1 and K2 can be obtained in the F06 file of your cold start run as long as the
Bulk Data is echoed with the “ECHO = SORT” Case Control command, which is the default
option. From Figure 8-1, the sorted Bulk Data count for “PARAM,GRDPNT,0" is “7";
therefore, the delete entry in this case can be either “/,7,7" or “/,7". No additional Bulk
 CHAPTER 8 231
Restarts In Dynamic Analysis

Data entry is required for this restart run since there are no other changes involved for this
model. Note that the same SPC and METHOD commands are used in both the cold start
and restart runs since neither the actual boundary condition nor the desired eigenvalue
calculation has changed.

S O R T E D B U L K D A T A E C H O

CARD
COUNT . 1 .. 2 .. 3 .. 4 .. 5 .. 6 .. 7 .. 8 .. 9 .. 10 .
1- CONM2 100 2 10.
2- CROD 1 10 1 2
3- EIGRL 10 1
4- GRID 1 0. 0. 0.
5- GRID 2 10. 0. 0.
6- MAT1 1 1.+7 .32 .097
7- PARAM GRDPNT 0
8- PARAM WTMASS .00259
9- PROD 10 1 1.0
10- SPC1 1 23456 2
11- SPC1 1 123456 1
ENDDATA
TOTAL COUNT= 12

Figure 8-1 Echo of the Sorted Bulk Data Input for the Cold Start Run

You can include as many of these delete entries as necessary. However, if the case requires
many changes, it is probably more convenient to delete the entire old Bulk Data Section
from the database and replace it with the fully revised Bulk Data Section, including the
new and modified entries. This operation can be accomplished by including the full new
Bulk Data plus the following entry in the Bulk Data Section of the restart run:
/1,i
where i is any positive integer that is greater than or equal to the number of Bulk Data
entries from your cold start run.
For conventional dynamic analysis (i.e., non-superelement), restarts involving model
changes (e.g., changing the thickness of a plate) are not very efficient. Therefore, the
savings is probably minimal, if any. However, in the case of additional output requests or
a restart from a modes run to a response run, the savings can be substantial. This type of
restart is covered extensively in “Examples” on page 234. For superelement analysis, even
restarts involving model changes can be beneficial as long as these changes are localized.
232

8.5 Determining the Version for a Restart

Not all versions in the database are restartable. For each run, a message is printed near the
top of the F04 file indicating the version number of the current job. In general, if a job
completes without any fatal messages, then that particular version is restartable. It is a
good idea to keep a brief log of all the restartable versions since this is probably the most
reliable source of information regarding whether a version is restartable. If a restart job
failed, e.g., due to Bulk Data error, then this newly created version is not restartable, which
is indicated by the following error message at the bottom of the F06 file.

*** USER WARNING MESSAGE 784 (XCLNUP)

VERSION = yyy PROJECT = "zzz" OF THIS DATA BASE IS NOT VALID FOR RESTART PURPOSES.

USER ACTION:
SUBSEQUENT RESTARTS SHOULD REFERENCE VERSION = xxx OR A PRIOR VALID VERSION

If the version is not restartable, you must restart from a previous valid version. The “xxx”
and “yyy” above denote version numbers. The “zzz” denotes a project description
provided by you. This project description is alphanumeric and can contain up to 40
characters (the default is blank). The project description is often not used and is an optional
statement.

If, for some reason, the records for the old runs are no longer available, then the DBDIR
FMS statement can be used to query the database contents to find out which versions are
being stored in the database. The following simple setup is all that is required for this
purpose.
ASSIGN MASTER=’ddddd.MASTER’
DBDIR VERSION=*,PROJECT=*
ENDJOB
where ddddd.MASTER is the name of the database being used.
The Executive Control, Case Control, or Bulk Data Section is not required in this case.
Furthermore, a new version is not created in this case. Near the top of the F06 output, a
PROJECT VERSION DIRECTORY TABLE is printed listing all the versions in the database.
A “**” next to a version number indicates that this particular version was deleted from the
database. This deletion may be due to the NOKEEP option or the use of the DBCLEAN
command when performing a restart run. A version number with a “**” next to it is not
restartable. A version number without a “**” next to it is restartable if the run that created
the version did not terminate with UWM 784. For more advanced users, this DBDIR FMS
statement can also be used to check the database directory for the existence of data blocks,
e.g., UG. If the job fails very early in the run (e.g., error in the FMS section), then a new
version may not be created.

It is always good practice to back up the database on a regular basis. If the system aborts
the run (e.g., disk space exhausted or time quota exceeded on a system level), then there is
a chance that the database is corrupted and will not allow restarts. Another good practice
to ensure that only good models are retained in the database is to perform the following:
 CHAPTER 8 233
Restarts In Dynamic Analysis

• Use RESTART VERSION = a,KEEP

where “a” is a version number.
• If a version contains errors or is no longer of interest, then use the FMS statement
DBCLEAN to remove obsolete or incorrect versions of the model from the
database. Removing these versions allows the Executive System to reuse some of
this space for new versions.
234

8.6 Examples
The examples perform a typical series of runs starting from a normal modes run and
restarting into transient and frequency response analyses. Table 8-2 summarizes this
series of nine runs along with a brief description. Listings of the nine runs are also included
(Listing 8-1 through Listing 8-9).

Table 8-2 Typical Series of Restart Runs

Run Solution
Name of Version Version
Sequence Sequence Description of Runs
Input File Created Deleted
Number Number

1 bd08bar1.dat 103 Perform a normal modes cold start 1 None

analysis and save the database.
2 bd08bar2.dat 103 Restart from run number 1 and 2 None
request eigenvector output.
3 bd08bar3.dat 103 The first two modes of the struc- 3 2
ture are very close to one of the
forcing frequencies. The structure
is modified in order to stay away
from resonance. This run restarts
from run number 2 to delete the
old PBAR entry and replace it with
the modified PBAR entry. The
modes are recalculated, and the
eigenvector output is requested in
this run.
4 bd08bar4.dat N/A This run deletes the data blocks in None 1
the database previously occupied
by Version 1. This space can then
be reused by future restarts. Note
that the statement DBCLEAN does
not reduce the database size. It
merely removes some of its con-
tents so that this space can be
reused. A new version is not cre-
ated as a result of this run. This is
an optional run especially if disk
space is of no concern to you.
 CHAPTER 8 235
Restarts In Dynamic Analysis

Table 8-2 Typical Series of Restart Runs (continued)

Run Solution
Name of Version Version
Sequence Sequence Description of Runs
Input File Created Deleted
Number Number

5 bd08bar5.dat 112 This is a transient restart run from 4 None

the modes saved in Version 3. The
applied load is a unit step func-
tion. The modes calculated in run
number 3 are also saved at the end
of this run. Since the calculation of
the modes is the most expensive
operation in a dynamic analysis, it
is probably a good idea to save
Version 3 once you have confi-
dence in the results. This way you
can always restart from this ver-
sion. Partial output is shown at
the top of Figure 8-2. A 1% critical
damping value is applied to the
structure.
6 bd08bar6.dat 112 This is another transient restart 5 4
run using the solution from run
number 5. The purpose of this run
is to request additional output.
Partial output is shown at the bot-
tom of Figure 8-2. Note that the
maximum displacement at grid
point 11 is the same as in run num-
ber 5 as expected since you are
asking for data recovery on the
same point due to the same load-
ing condition.
7 bd08bar7.dat 112 This is another restart run from 6 5
Version 6 with a different load con-
dition (triangular pulse). Partial
output is shown in Figure 8-3. In
this case, you can just as easily
restart from Version 3.
236

Table 8-2 Typical Series of Restart Runs (continued)

Run Solution
Name of Version Version
Sequence Sequence Description of Runs
Input File Created Deleted
Number Number

8 bd08bar8.dat 111 This is a frequency response 7 6

restart run. Note that this can also
be restarted from Version 3 since it
was saved in the database. Partial
output is shown in Figure 8-4. A
2% critical damping value is
applied to the structure.
9 bd08bar9.dat N/A This is a database directory print- None None
out run. As shown in Figure 8-5,
there are seven versions in the
database. Run numbers 4 and 9
did not create any new versions.
Only Versions 3 and 7 are restart-
able. This is an optional run.

Remark
If the results for run number 1 are not going to be used for any future purposes, then you
may consider making run number 3 as a cold start run instead of a restart run. Model
changes do not save you much time, if any, in a non-superelement analysis. By making
run 3 a cold start run, you reduce the total amount of disk space required. In this case, run
number 4 is not necessary since you are starting with a new database. However, if you
want to keep both physical models in the database, then run number 3 should be a restart
run as shown in this example. An application of this can be a parametric study of two
different configurations. This type of restart allows you to make efficient data recovery or
response analysis from two different physical models. However, this type of restart is not
used often in a non-superelement analysis since, in general, it is not very efficient.
However, in a superelement analysis (see “Advanced Dynamic Analysis Capabilities” on
page 293), this type of restart can still be very efficient since the changes can be localized to
a small region.
 CHAPTER 8 237
Restarts In Dynamic Analysis

Listing 8-1 Input File for Normal Modes Run

$ FILE - bd08bar1.dat
$
$ NORMAL MODES RUN
$
ID CANT BEAM
SOL 103
TIME 10
CEND
TITLE = CANTILEVER BEAM - NORMAL MODES - COLD START RUN
SPC = 1
METHOD = 10
$
BEGIN BULK
$
CBAR 1 1 1 2 0. 1. 0.
CBAR 2 1 2 3 0. 1. 0.
CBAR 3 1 3 4 0. 1. 0.
CBAR 4 1 4 5 0. 1. 0.
CBAR 5 1 5 6 0. 1. 0.
CBAR 6 1 6 7 0. 1. 0.
CBAR 7 1 7 8 0. 1. 0.
CBAR 8 1 8 9 0. 1. 0.
CBAR 9 1 9 10 0. 1. 0.
CBAR 10 1 10 11 0. 1. 0.
EIGRL 10 -0.1 50.
GRID 1 0.0 0. 0.
GRID 2 0.3 0. 0.
GRID 3 0.6 0. 0.
GRID 4 0.9 0. 0.
GRID 5 1.2 0. 0.
GRID 6 1.5 0. 0.
GRID 7 1.8 0. 0.
GRID 8 2.1 0. 0.
GRID 9 2.4 0. 0.
GRID 10 2.7 0. 0.
GRID 11 3.0 0. 0.
MAT1 1 7.1+10 .33 2.65+4
PARAM AUTOSPC YES
PARAM WTMASS .102
PBAR 1 1 6.158-4 3.-8 3.-8 6.-8 2.414
SPC1 1 123456 1
$
ENDDATA
238

Listing 8-2 Input File for Requesting Eigenvectors

$ FILE - bd08bar2.dat
$
$ NORMAL MODES RUN
$ REQUEST EIGENVECTOR PRINTOUTS FROM PREVIOUS RUN
$
RESTART VERSION=1,KEEP
ASSIGN MASTER=’bd08bar1.MASTER’
$
ID CANT BEAM
SOL 103
TIME 10
CEND
TITLE = EIGENVECTORS DATA RECOVERY RESTART RUN
SPC = 1
METHOD = 10
DISP = ALL $ PRINT EIGENVECTORS
$
BEGIN BULK
$
ENDDATA

Listing 8-3 Input File for Modifying a Bar Element

$ FILE - bd08bar3.dat
$
$ NORMAL MODES RUN
$ MODIFY PBAR
$
RESTART
ASSIGN MASTER=’bd08bar1.MASTER’
$
ID CANT BEAM
SOL 103
TIME 10
CEND
TITLE = CANTILEVER BEAM - NORMAL MODES - RESTART RUN
SPC = 1
METHOD = 10
DISP = ALL
$
BEGIN BULK
$
$ DELETE OLD PBAR ENTRY, LINE 26 OF SORTED BULK DATA COUNT
$
/,26
$
$ ADD NEW PBAR ENTRY
$
PBAR,1,1,6.158-4,2.9-8,3.1-8,6.-8,2.414
$
ENDDATA
 CHAPTER 8 239
Restarts In Dynamic Analysis

Listing 8-4 Input File for Cleaning a Database

$ FILE - bd08bar4.dat
$
assign master=’bd08bar1.MASTER’
dbclean version=1
endjob

Listing 8-5 Input File for Transient Response

$ FILE - bd08bar5.dat
$
$ THIS IS A TRANSIENT RESTART RUN FROM THE MODES
$ CALCULATED BY THE RUN “bd08bar3.dat”
$
RESTART VERSION=3,KEEP
ASSIGN MASTER=’bd08bar1.MASTER’
ID CANT BEAM
SOL 112
TIME 10
CEND
TITLE = TRANSIENT RESTART - UNIT STEP FUNCTION INPUT
SUBTITLE = REQUEST DISPLACEMENT TIME HISTORY AT GRID POINT 11
SPC = 1
METHOD = 10
SET 1 = 11
DISP = 1
SUBCASE 1
SDAMP = 100
TSTEP = 100
DLOAD = 100
$
BEGIN BULK
$
$ ADDITIONAL ENTRIES FOR DYNAMIC LOADS
$ FOR UNIT STEP FUNCTION
$
$ SID DAREA DELAY TYPE TID
TLOAD1 100 101 102
$
DAREA,101,11,3,1.0
$
TABLED1,102,,,,,,,,+TBL1
+TBL1,0.0,0.0,.001,1.0,10.0,1.0,ENDT
$
$ TRANSIENT TIME STEPS
$
$ SID N(1) DT(1) NO(1)
TSTEP 100 600 .001 5
$
$ MODAL DAMPING TABLE
$
TABDMP1,100,CRIT,,,,,,,+TDAMP
+TDAMP,0.,.01,200.,.01,ENDT
$
ENDDATA
240

Listing 8-6 Input File for an Additional Output Request

$ FILE - bd08bar6.dat
$
$ THIS IS ANOTHER TRANSIENT RESTART RUN. THE PURPOSE
$ OF THIS RUN IS TO REQUEST ADDITIONAL OUTPUT.
$
RESTART
ASSIGN MASTER=’bd08bar1.MASTER’
ID CANT BEAM
SOL 112
TIME 10
CEND
TITLE = T R A N S I E N T R E S T A R T
SUBTITLE = U N I T S T E P F U N C T I O N I N P U T
SPC = 1
METHOD = 10
SET 1 = 11
SET 2 = 10
ACCE = 2
SUBCASE 1
SDAMP = 100
TSTEP = 100
DLOAD = 100
$
$
$ PLOT RESULTS
$
$...X-Y plot commands ...
$
BEGIN BULK
$
$
ENDDATA
 CHAPTER 8 241
Restarts In Dynamic Analysis

Listing 8-7 Input File for an Additional Transient Load

$ FILE - bd08bar7.dat
$
$ THIS IS ANOTHER TRANSIENT RESTART RUN USING
$ A DIFFERENT LOAD CONDITION. NOTE THAT SINCE
$ THERE ARE NO MODEL CHANGES, THE SAME MODES
$ WERE USED FROM THE DATABASE FOR THE RESPONSE
$ CALCULATIONS.
$
RESTART
ASSIGN MASTER=’bd08bar1.MASTER’
ID CANT BEAM
SOL 112
TIME 10
CEND
$
$ NOTE THAT TITLE CHANGES HAVE NO EFFECT
$ ON SOLUTION PROCESS, THEY ONLY CHANGE THE
$ PRINTOUT TITLE
$
TITLE = T R A N S I E N T R E S T A R T
SUBTITLE = TRIANGLE PULSE - 1.0 AT T=0 AND 0.0 AFTER .2 SEC
SPC = 1
METHOD = 10
SET 1 = 11
DISP = 1
SUBCASE 1
SDAMP = 100
TSTEP = 100
DLOAD = 300
$
$ PLOT RESULTS
$
$...X-Y plot commands ...
$
BEGIN BULK
$
$ SID DAREA DELAY TYPE TID
TLOAD1 300 301 302
$
DAREA,301,11,3,1.0
$
TABLED1,302,,,,,,,,+TBL3
+TBL3,0.0,0.0,.001,1.0,.20,0.0,10.0,0.0,+TBL4
+TBL4,ENDT
$
ENDDATA
242

Listing 8-8 Input File for Frequency Response Analysis

$ FILE - bd08bar8.dat
$
$ THIS IS RESTART RUN TO PERFORM FREQUENCY RESPONSE
$
RESTART
ASSIGN MASTER=’bd08bar1.MASTER’
ID CANT BEAM
SOL 111
TIME 10
CEND
TITLE = CANTILEVER BEAM - FREQUENCY RESPONSE RESTART
SPC = 1
METHOD = 10
SET 1 = 11
DISP(PHASE) = 1
SUBCASE 1
$
$ A TWO-PERCENT CRITICAL DAMPING IS
$ APPLIED TO THIS RUN AS OPPOSED TO
$ ONE-PERCENT CRITICAL DAMPING IN
$ THE TRANSIENT ANALYSIS
$
SDAMP = 1000
DLOAD = 1000
FREQ = 1000
$
$ PLOT RESULTS
$
$...X-Y plot commands ...
$
BEGIN BULK
$
$ ADDITIONAL ENTRIES FOR FREQUENCY RESPONSE
$
$ SID DAREA M N TC TD
$
RLOAD1 1000 1001 1002
$
DAREA,1001,11,3,0.1
$
TABLED1,1002
,0.,1.,200.,1.,ENDT
$
$ FORCING FREQUENCIES
$
$ RESONANT FREQUENCIES
$
FREQ,1000,2.03174,2.100632,12.59101,13.01795
FREQ,1000,34.90217,36.08563
$
$ SPREAD THROUGHOUT
$ FREQUENCY RANGE OF INTEREST
$ WITH BIAS BETWEEN
$ HALF-POWER POINTS
$
FREQ,1000,1.437,1.556,1.675,1.794,1.913
FREQ,1000,2.046,2.059,2.073,2.087
FREQ,1000,2.224,2.347,2.47,2.593,2.716
FREQ,1000,8.903,9.641,10.378,11.116,11.853
FREQ,1000,12.676,12.762,12.847,12.933
FREQ,1000,13.781,14.543,15.306,16.068,16.831
FREQ,1000,24.680,26.724,28.769,30.813,32.858
FREQ,1000,35.139,35.376,35.612,35.849
FREQ,1000,41.189,46.292,51.395,56.499,61.602
$
FREQ1,1000,0.,.5,200
$
$ DAMPING
$
TABDMP1,1000,CRIT,,,,,,,+DAMP
+DAMP,0.,.02,200.,.02,ENDT
$
ENDDATA
 CHAPTER 8 243
Restarts In Dynamic Analysis

Listing 8-9 Input File to Print the Database Dictionary

$ FILE - bd08bar9.dat
$
assign master=’bd08bar1.MASTER’
dbdir
endjob

***** PARTIAL OUTPUT FROM bd08bar5.f06 ***** POINT-ID = 11

D I S P L A C E M E N T V E C T O R

TIME TYPE T1 T2 T3 R1 R2 R3
.0 G .0 .0 .0 .0 .0 .0
5.000000E-03 G 9.024590E-22 1.614475E-21 1.923969E-05 .0 -3.001072E-05 3.518218E-20
1.000000E-02 G 3.367912E-21 8.536304E-21 7.739825E-05 .0 -1.118816E-04 1.271978E-19
.
.
2.300003E-01 G 3.941339E-19 8.273585E-18 7.832402E-03 .0 -3.655237E-03 5.515358E-18
2.350003E-01 G 3.956722E-19 8.583470E-18 7.862186E-03 .0 -3.686721E-03 5.464303E-18
2.400003E-01 G 3.957035E-19 9.029934E-18 7.880123E-03 .0 -3.709195E-03 5.756178E-18
2.450003E-01 G 3.944803E-19 9.498899E-18 7.886210E-03 .0 -3.726098E-03 6.049827E-18
2.500003E-01 G 3.927639E-19 9.878397E-18 7.884440E-03 .0 -3.754001E-03 6.040225E-18
2.550003E-01 G 3.909298E-19 1.020556E-17 7.873885E-03 .0 -3.796297E-03 5.837433E-18
2.600002E-01 G 3.885339E-19 1.063947E-17 7.844862E-03 .0 -3.831216E-03 5.871013E-18
2.650001E-01 G 3.848278E-19 1.128651E-17 7.785291E-03 .0 -3.828911E-03 6.419444E-18
2.700001E-01 G 3.796167E-19 1.207232E-17 7.690622E-03 .0 -3.778875E-03 7.287524E-18
.
.
5.949959E-01 G 1.986670E-19 -1.304284E-17 4.027668E-03 .0 -2.010369E-03 -4.506177E-18
5.999959E-01 G 2.106813E-19 -1.285776E-17 4.244778E-03 .0 -2.079488E-03 -4.219749E-18

----------------------------------------------------------------------------------------------------------------------------
***** PARTIAL OUTPUT FROM bd08bar6.f06 *****
X Y - O U T P U T S U M M A R Y ( R E S P O N S E )

SUBCASE CURVE FRAME XMIN-FRAME/ XMAX-FRAME/ YMIN-FRAME/ X FOR YMAX-FRAME/ X FOR

ID TYPE NO. CURVE ID. ALL DATA ALL DATA ALL DATA YMIN ALL DATA YMAX
1 DISP 1 11( 5) 0.000000E+00 5.999959E-01 0.000000E+00 0.000000E+00 7.886210E-03 2.450003E-01
0.000000E+00 5.999959E-01 0.000000E+00 0.000000E+00 7.886210E-03 2.450003E-01
1 ACCE 2 10( 5) 0.000000E+00 5.999959E-01 -8.511892E-01 2.650001E-01 9.618800E-01 5.000000E-03
0.000000E+00 5.999959E-01 -8.511892E-01 2.650001E-01 9.618800E-01 5.000000E-03

Figure 8-2 Partial Output from Transient Analysis with Unit Step Function Input
244

***** PARTIAL OUTPUT FROM bd08bar7.f06 *****

POINT-ID = 11
D I S P L A C E M E N T V E C T O R

TIME TYPE T1 T2 T3 R1 R2 R3
.0 G .0 .0 .0 .0 .0 .0
5.000000E-03 G 8.975580E-22 1.606407E-21 1.913647E-05 .0 -2.984776E-05 3.499062E-20
1.000000E-02 G 3.317352E-21 8.430443E-21 7.628838E-05 .0 -1.102005E-04 1.252464E-19
.
.
1.650002E-01 G 2.077055E-19 2.232167E-18 4.099007E-03 .0 -1.833254E-03 1.845365E-18
1.700002E-01 G 2.104969E-19 2.275425E-18 4.180374E-03 .0 -1.908738E-03 1.502202E-18
1.750002E-01 G 2.131340E-19 2.401124E-18 4.255632E-03 .0 -2.000815E-03 1.338691E-18
1.800002E-01 G 2.146708E-19 2.724639E-18 4.308840E-03 .0 -2.072395E-03 1.631772E-18
1.850002E-01 G 2.144943E-19 3.204978E-18 4.328179E-03 .0 -2.097317E-03 2.240218E-18
1.900002E-01 G 2.129652E-19 3.684523E-18 4.313660E-03 .0 -2.081293E-03 2.749993E-18
1.950002E-01 G 2.109444E-19 4.053196E-18 4.272623E-03 .0 -2.048310E-03 2.916553E-18
.
.
5.949959E-01 G 1.315062E-19 -1.196586E-17 2.607416E-03 .0 -1.203858E-03 -4.331415E-18
5.999959E-01 G 1.410901E-19 -1.111419E-17 2.775622E-03 .0 -1.249621E-03 -3.777846E-18

X Y - O U T P U T S U M M A R Y ( R E S P O N S E )

SUBCASE CURVE FRAME XMIN-FRAME/ XMAX-FRAME/ YMIN-FRAME/ X FOR YMAX-FRAME/ X FOR

ID TYPE NO. CURVE ID. ALL DATA ALL DATA ALL DATA YMIN ALL DATA YMAX
1 DISP 1 11( 5) 0.000000E+00 6.000000E-01 -4.039956E-03 4.199981E-01 4.328179E-03 1.850002E-01
0.000000E+00 5.999959E-01 -4.039956E-03 4.199981E-01 4.328179E-03 1.850002E-01

Figure 8-3 Partial Output from Transient Analysis with a Triangular Pulse

***** PARTIAL OUTPUT FROM bd08bar8.f06 *****

POINT-ID = 11
C O M P L E X D I S P L A C E M E N T V E C T O R
(MAGNITUDE/PHASE)

FREQUENCY TYPE T1 T2 T3 R1 R2 R3
.0 G 2.016639E-20 4.838324E-19 4.085192E-04 .0 2.024146E-04 3.208819E-19
.0 .0 .0 .0 180.0000 .0
5.000000E-01 G 2.136733E-20 5.467976E-19 4.323836E-04 .0 2.134119E-04 3.549737E-19
359.4255 359.0918 359.4344 .0 179.4676 359.2066
.
.
2.073000E+00 G 4.228285E-19 1.831657E-16 8.402211E-03 .0 3.874130E-03 8.302084E-17
303.5255 169.4034 303.5720 .0 123.7533 170.2924
2.087000E+00 G 4.789151E-19 1.747242E-16 9.514135E-03 .0 4.382166E-03 7.877973E-17
288.0493 145.7774 288.0963 .0 108.2796 146.6783
2.100632E+00 G 5.003335E-19 1.561973E-16 9.936970E-03 .0 4.572148E-03 7.006467E-17
270.0177 122.0074 270.0652 .0 90.2504 122.9200
2.224000E+00 G 1.560747E-19 1.943312E-17 3.091950E-03 .0 1.408677E-03 8.308275E-18
199.3224 32.8670 199.3742 .0 19.5776 33.8900
2.347000E+00 G 7.916779E-20 5.971262E-18 1.564160E-03 .0 7.050781E-04 2.427647E-18
190.2228 19.2229 190.2792 .0 10.5024 20.3631
.
.
9.950000E+01 G .0 .0 5.272466E-07 .0 8.710860E-07 1.024487E-21
.0 .0 180.4391 .0 .6610 180.7005
1.000000E+02 G .0 .0 5.216914E-07 .0 8.615909E-07 1.013485E-21
.0 .0 180.4361 .0 .6566 180.6966

X Y - O U T P U T S U M M A R Y ( R E S P O N S E )

SUBCASE CURVE FRAME XMIN-FRAME/ XMAX-FRAME/ YMIN-FRAME/ X FOR YMAX-FRAME/ X FOR

ID TYPE NO. CURVE ID. ALL DATA ALL DATA ALL DATA YMIN ALL DATA YMAX
1 DISP 1 11( 5,--) 1.000000E-01 1.000000E+02 5.216914E-07 1.000000E+02 9.936970E-03 2.100632E+00
0.000000E+00 1.000000E+02 5.216914E-07 1.000000E+02 9.936970E-03 2.100632E+00
1 DISP 1 11(--, 11) 1.000000E-01 1.000000E+02 1.804361E+02 1.000000E+02 3.594344E+02 5.000000E-01
0.000000E+00 1.000000E+02 0.000000E+00 0.000000E+00 3.594344E+02 5.000000E-01

Figure 8-4 Partial Output from Frequency Response Analysis

 CHAPTER 8 245
Restarts In Dynamic Analysis

***** PARTIAL OUTPUT FROM bd08bar9.f06 *****

P R O J E C T V E R S I O N D I R E C T O R Y P R I N T
PROJECT_ID ASSIGNED INT. VALUE VERSION_ID CREATION TIME
---------- ------------------- ---------- -------------
" B L A N K " 1 ** 1 9/ 2/93 13:22.20
** 2 9/ 2/93 13:23. 0
3 9/ 2/93 13:23.25
** 4 9/ 2/93 13:23.59
** 5 9/ 2/93 13:24.28
** 6 9/ 2/93 13:24.57
7 9/ 2/93 13:25.26

Figure 8-5 Partial Output from a Database Directory Run

246
 NX Nastran Basic Dynamic Analysis User’s Guide

9
CHAPTER
Plotted Output

■ Overview
■ Structure Plotting

■ X-Y Plotting
248

9.1 Overview
Plotted output is important in verifying your model and understanding its results. Plots
show information in a format that is much easier to interpret than printed output. Plots are
especially important for dynamic analysis because the analysis can produce voluminous
output. For example, consider a transient response analysis for which there are 1000
output time steps, 100 grid points of interest, and 10 elements of interest. Printed output
is too large to interpret efficiently and effectively, and it does not easily show the time
variation. Plotted output overcomes these problems.
There are two kinds of plotted output: structure plots and X-Y plots. Structure plots can
depict the entire structure or a portion of it. Structure plots are useful for verifying proper
geometry and connectivity. They also can be used to show the deformed shape or stress
contours at a specified time or frequency. X-Y plots, on the other hand, show how a
single-response quantity, such as a grid point displacement or element stress, varies across
a portion or all of the time or frequency range.

There are numerous commercial and in-house plotting programs that interface to
MSC.Nastran for structure and/or X-Y plotting. Popular commercial programs include
MSC/ARIES, PDA PATRAN, and SDRC I-DEAS, among others. Details about these
programs can be obtained from their companies.
The commercial programs are similar because they operate interactively. The
MSC.Nastran plot capabilities, on the other hand, are performed as a batch operation,
which means that you predefine your plots when you make your MSC.Nastran run.
This chapter briefly describes the kinds of plots available in MSC.Nastran. Detailed
information is located in the MSC.Nastran Reference Manual.
 CHAPTER 9 249
Plotted Output

9.2 Structure Plotting

Structure plotting is performed to verify the model’s geometry and element connectivity
prior to performing a dynamic analysis.
After (or during) the analysis, structure plotting is performed to view deformed shapes
and contours. For dynamic response, deformed shape and contour plots can be made for
normal modes analysis (for which there is a plot, or set of plots, per mode), frequency
response analysis (for which there is a plot, or set of plots, per output frequency), and
transient response analysis (for which there is a plot, or set of plots, per output time).
Structure plot commands are described in the NX Nastran Reference Manual. In the
MSC.Nastran input file, structure plotting commands are listed in the OUTPUT(PLOT)
section, which immediately precedes the Bulk Data Section. The structure plotting
commands define the set of elements to be plotted (SET), the viewing axes (AXES), the
viewing angles (VIEW) as well as the plot type and parameters (PLOT). Optionally, the
scale of the plotted deformation (MAXIMUM DEFORMATION) can be specified; if not
specified, the plotted deformation is scaled such that the maximum deformation is 5% of
the maximum dimension of the structure.

Listing 9-1 shows the structure plotting commands applied to a normal modes analysis of
the bracket model. Figure 9-1 shows the resulting structure plots. The first plot shows the
undeformed shape, and the next two plots show the undeformed shape overlaid on the
deformed shapes for modes 1 and 2. The default is chosen such that the maximum plotted
deformation is 5% of the maximum dimension of the bracket; the actual maximum
deformation is printed at the top of the plot. The plots shown in this chapter were
converted to PostScript format for printing on a PostScript-compatible printer.
Listing 9-1 Normal Modes Structure Plot Commands for the Bracket Model

$ PLOT COMMANDS FOR BRACKET NORMAL MODES ANALYSIS

... Executive, Case Control ...
$
$ OUTPUT REQUESTS
DISPLACEMENT(PLOT) = ALL
$
$ STRUCTURE PLOTS
OUTPUT(PLOT)
CSCALE = 3.0
SET 333 = ALL
AXES MX,MY,Z
VIEW 20.,20.,20.
FIND SCALE, ORIGIN 5, SET 333
$ PLOT UNDEFORMED SHAPE
PLOT SET 333, ORIGIN 5
$ PLOT DEFORMED, UNDEFORMED SHAPES
PLOT MODAL DEFORMATION 0,1
PLOT MODAL DEFORMATION 0,2
$
BEGIN BULK
$
... Bulk Data ...
$
ENDDATA
250

Undeformed Shape

Max. Def. = 3.82

Mode 1

Max. Def. = 5.06

Mode 2

Figure 9-1 Normal Modes Structure Plots for the Bracket Model

Listing 9-2 shows the structure plotting commands applied to a modal frequency response
analysis of the cantilever beam model. The displacements are computed in
magnitude/phase form. PARAM,DDRMM,-1 and PARAM,CURVPLOT,1 are required in
the Bulk Data to create structure plots at specified frequencies. (Note that
PARAM,DDRMM,-1 generally increases the amount of computer time and is not
recommended unless otherwise required.) Figure 9-2 shows the resulting plots. The first
 CHAPTER 9 251
Plotted Output

plot shows the undeformed shape, and the next plot shows the magnitude of response at
2.05 Hz, which is overlaid on the undeformed shape. The default is chosen such that the
maximum plotted deformation is 5% of the length of the bar; the actual maximum
deformation is printed at the top of the plot.

Listing 9-2 Frequency Response Structure Plot Commands for the Bar Model – Magnitude/Phase

$ PLOT COMMANDS FOR BAR MODAL FREQUENCY RESPONSE

$ --- MAGNITUDE, PHASE ---
... Executive, Case Control ...
$
$ OUTPUT REQUESTS
$ MAGNITUDE, PHASE
DISPLACEMENT(PHASE,PLOT) = ALL
$
$ STRUCTURE PLOTS
OUTPUT(PLOT)
$ DEFINE ELEMENTS IN PLOT SET
SET 333 = ALL
$ PLOT AXES: R=-Z, S=X, T=Y
AXES MZ,X,Y
VIEW 0.,0.,0.
FIND SCALE, ORIGIN 5, SET 333
$ PLOT UNDEFORMED SHAPE
PLOT SET 333, ORIGIN 5
$ PLOT DEFORMED SHAPE AT 2.05 HZ
PLOT FREQUENCY DEFORMATION 0 RANGE 2.05,2.051 MAGNITUDE SET 333
$
BEGIN BULK
$
$ REQUIRED FOR "FREQUENCY FROZEN" STRUCTURE PLOTS
PARAM,DDRMM,-1
PARAM,CURVPLOT,1
$
... rest of Bulk Data ...
$
ENDDATA

Undeformed Shape

Max. Def. = 0.394

Frequency 2.05
Figure 9-2 Frequency Response Structure Plots for the Bar Model –
Magnitude/Phase

The same plots are regenerated except that now the displacements are in real/imaginary
format (the default). Listing 9-3 shows the plot commands. Note that the imaginary
component is selected by PHASE LAG -90. Figure 9-3 shows the resulting plots. The
default is chosen such that the maximum plotted deformation is 5% of the length of the bar;
the actual maximum deformation is printed at the top of the plot.
252

Listing 9-3 Frequency Response Structure Plot Commands for the Bar Model - Real/Imaginary

$ PLOT COMMANDS FOR BAR MODAL FREQUENCY RESPONSE

$ --- REAL, IMAGINARY ---
... Executive, Case Control ...
$
$ OUTPUT REQUESTS
$ REAL, IMAGINARY
DISPLACEMENT(PLOT) = ALL
$
$ STRUCTURE PLOTS
OUTPUT(PLOT)
$ DEFINE ELEMENTS IN PLOT SET
CSCALE = 1.8
SET 333 = ALL
$ PLOT AXES: R=-Z, S=X, T=Y
AXES MZ,X,Y
VIEW 0.,0.,0.
FIND SCALE, ORIGIN 5, SET 333
$ PLOT UNDEFORMED SHAPE
PLOT SET 333, ORIGIN 5
$ PLOT DEFORMED SHAPE--REAL
PLOT FREQ DEFORM 0 RANGE 2.05,2.051 SET 333
$ PLOT DEFORMED SHAPE--IMAGINARY
PLOT FREQ DEFORM 0 RANGE 2.05,2.051 PHASE LAG -90. SET 333
$
BEGIN BULK
$
$ REQUIRED FOR "FREQUENCY FROZEN" STRUCTURE PLOTS
PARAM,DDRMM,-1
PARAM,CURVPLOT,1
$
... rest of Bulk Data ...
$
ENDDATA

Undeformed Shape

Max. Def. = 0.185

Frequency 2.05 Phase 0.

Max. Def. = 0.348

Frequency 2.05 Phase -90.00

Figure 9-3 Frequency Response Structure Plots for the Bar Model – Real/Imaginary
 CHAPTER 9 253
Plotted Output

Listing 9-4 shows the structure plotting commands applied to a modal transient response
analysis of the cantilever beam model. PARAM,DDRMM,-1 is required in the Bulk Data
in order to create structure plots at various times and/or frequencies. (Note that
PARAM,DDRMM,-1 generally increases the amount of computer time and is not
recommended unless otherwise required.) Figure 9-4 shows the resulting plots. The plots
show the displacements at the following times: 0.25, 0.50, and 0.75 seconds, which are
overlaid on the undeformed shape. The default is chosen such that the maximum plotted
deformation is 5% of the length of the bar; the actual maximum deformation is printed at
the top of the plot. For this case it may be better to specify a maximum deformation so that
the plots will show relative amplitudes.

Listing 9-4 Transient Response Structure Plot Commands for the Bar Model

$ PLOT COMMANDS FOR BAR MODAL TRANSIENT RESPONSE

... Executive, Case Control ...
$
$ OUTPUT REQUESTS
DISPLACEMENT(PLOT) = ALL
$
$ STRUCTURE PLOTS
OUTPUT(PLOT)
$ DEFINE ELEMENTS IN PLOT SET
SET 333 = ALL
$ PLOT AXES: R=-Z, S=X, T=Y
AXES MZ,X,Y
VIEW O.,O.,O.
FIND SCALE, ORIGIN 5, SET 333
$ PLOT DEFORMED SHAPE AT TIMES 0.25,0.5,0.75
PLOT TRANSIENT DEFORMATION 0 RANGE 0.25,0.251 SET 333
PLOT TRANSIENT DEFORMATION 0 RANGE 0.50,0.501 SET 333
PLOT TRANSIENT DEFORMATION 0 RANGE 0.75,0.751 SET 333
$
BEGIN BULK
$
$ REQUIRED FOR "TIME FROZEN" STRUCTURE PLOTS
PARAM,DDRMM,-1
$
... rest of Bulk Data ...
$
ENDDATA
254

Max. Def. = 0.040

Time 0.25

Max. Def. = 0.069

Time 0.50

Max. Def. = 0.057

Time 0.75
Figure 9-4 Transient Response Structure Plots for the Bar Model

Although structure plotting is best performed in an interactive environment outside of

MSC.Nastran, the batch structure plotting capability in MSC.Nastran is nevertheless a
useful tool for model verification and results processing. The batch plotting capability can
save time and effort when many plots are required for a model that is run repeatedly.
 CHAPTER 9 255
Plotted Output

9.3 X-Y Plotting

X-Y plots are used to display frequency and transient response results where the x-axis is
frequency or time and the y-axis is any output quantity. Unlike structure plotting, which
is often performed in an interactive environment, X-Y plotting is ideal for the MSC.Nastran
batch environment due to the large volume of data.
X-Y plot commands are contained in the OUTPUT(XYPLOT) section that immediately
precedes the Bulk Data Section. You define the titles (XTITLE and YTITLE) and plots
(XYPLOT). You can specify the plots to be generated in log format (XLOG and YLOG), and
you can specify different line styles. You can also specify that the plots are to be made in
pairs (with a top and a bottom plot), which is particularly useful for frequency response
results when you want to display magnitude/phase or real/imaginary pairs. Details
about X-Y plotting and its commands are located in the MSC.Nastran Reference Manual.
Element force and stress component numbers are also described in the manual. The X-Y
plot commands use numbers to identify a single component of grid point and element data.
Note that frequency response requires the use of complex force and stress components.
Once a good set of X-Y plot commands is established, it is wise to use this set repeatedly.
The examples that follow provide a good starting point.
Listing 9-5 shows X-Y plot commands for a modal frequency response analysis of the
cantilever beam model. Plots are made in pairs in magnitude/phase format. The “t” as the
second letter in XTGRID, YTGRID, YTLOG, and YTTITLE corresponds to the top plot of
each pair; the letter “b” corresponds to the bottom plot (XBGRID, YBGRID, etc.) Plots are
made for the applied loads (OLOAD) at grid points 6 and 11, displacements (DISP) of grid
points 6 and 11, bending moment (ELFORCE) at end A in plane 1 for element 6, and modal
displacements (SDISP) for modes 1 and 2. For CBAR elements, force component 2 is the
bending moment component at end A in plane 1 for real or magnitude output, and force
component 10 is the similar component for imaginary or phase output. Figure 9-5 shows
the resulting plots. The plots in this chapter were converted to PostScript format for
printing on a PostScript-compatible printer.
The X-Y plotter makes reasonable choices of upper and lower bounds for the axes for both
the x- and y-axes on most plots. If it does not, the bounds can be fixed with the XMIN,
XMAX, YMIN, YMAX, and their variations for half-frame curves. Instances where setting
bounds explicitly results in better plots include the following situations:
• If you expect a variable to be constant or vary only slightly but want to plot it to
confirm that it indeed does not vary, set the YMIN and YMAX to include the
expected value, but separate them by at least 10 percent of their average value.
For this case the automatic bound selection chooses bounds very close to each
other to make the data fill up the plot. This selection causes the bounds to be
nearly equal and magnifies the scale of the plot orders of magnitude larger than
other plots made of varying functions. The response appears to be erratic when,
in fact, it is smooth within engineering criteria. The extreme cases occur when the
256

function varies only in its last digit. Then the function appears to be oscillating
between the upper and lower limits or it can even cause a fatal error due to
numerical overflow when it attempts to divide numbers by the difference
YMAX-YMIN.
• The automatic bound selector tends to round up the bounds to integer multiples
of 100. When plotting phase angles, bounds that cause grid lines at 90 degrees are
more readable. For example, for a plot that traverses the range of 0 to 360 degrees,
the usual selected bounds are 0 to 400 degrees. If you prefer to have grid lines
drawn at integer multiples of 90 degrees, set YMIN to 0.0 and YMAX at 360.0
degrees.
• When plotting log plots, any bounds you input may be rounded up or down to a
value that the plotter considers more reasonable. In general, changing bounds on
log plots sometimes requires experimentation before a reasonable set can be
found.
• Producing good quality plots is an interactive process whether the plot is
produced by an interactive or batch plotter. The restart feature discussed in
“Restarts In Dynamic Analysis” in Chapter 8 can reduce the computer costs for
this iteration since restarts performed to change only plot requests are made
efficiently.
 CHAPTER 9 257
Plotted Output

Listing 9-5 X-Y Plot Commands for the Bar Frequency Response Analysis

$ X-Y PLOT COMMANDS FOR BAR MODAL FREQUENCY RESPONSE

$ --- MAGNITUDE, PHASE ---
...Executive, Case Control ...
$
$ APPLIED LOAD OUTPUT
SET 15 = 6,11
OLOAD(PHASE,PLOT) = 15
$
$ PHYSICAL OUTPUT REQUEST
SET 11 = 6,11
DISPLACEMENT(PHASE,PLOT) = 11
$
$ MODAL SOLUTION SET OUTPUT
SET 12 = 1,2
SDISP(PHASE,PLOT) = 12
$
$ ELEMENT FORCE OUTPUT
SET 13 = 6
ELFORCE(PHASE,PLOT) = 13
$
OUTPUT(XYPLOT)
XTGRID = YES
YTGRID = YES
XBGRID = YES
YBGRID = YES
$
$ PLOT RESULTS
XTITLE = FREQUENCY
$
YTLOG = YES
YTTITLE = DISPL. MAG. 6
YBTITLE = DISPL. PHASE 6
XYPLOT DISP /6(T2RM,T2IP)
YTTITLE = DISPL. MAG. 11
YBTITLE = DISPL. PHASE 11
XYPLOT DISP /11(T2RM,T2IP)
$
YTTITLE = SDISP. MAG. MODE 1
YBTITLE = SDISP. PHASE MODE 1
$ XYPLOT SDISP /mode(T1)
XYPLOT SDISP /1(T1RM,T1IP)
YTTITLE = SDISP. MAG. MODE 2
YBTITLE = SDISP. PHASE MODE 2
XYPLOT SDISP /2(T1RM,T1IP)
$
YTTITLE = BEND. MOMENT A1 EL. 6 MAG.
YBTITLE = BEND. MOMENT A1 EL. 6 PHASE
XYPLOT ELFORCE /6(2,10)
$
YTLOG = NO
YBMAX = 90.0
YBMIN = 0.0
CURVELINESYMBOL = -2
YTMAX = 4.0
YTMIN = 0.0
YTTITLE = LOAD MAG. 6
YBTITLE = LOAD PHASE 6
XYPLOT OLOAD /6(T2RM,T2IP)

YTMAX = 8.0
YTMIN = 0.0
YTTITLE = LOAD MAG. 11
YBTITLE = LOAD PHASE 11
XYPLOT OLOAD /11(T2RM,T2IP)
$
BEGIN BULK
$
... Bulk Data ...
$
ENDDATA
258

1.0

Displ. Mag. 6 1.0E-5

400
Displ. Phase 6

1.0
Displ. Mag. 11

1.0E-4
400
Displ. Phase 11

0
0 Frequency (Hz) 20
Figure 9-5 X-Y Plots for the Bar Frequency Response Analysis
 CHAPTER 9 259
Plotted Output

1.0

Sdisp. Mag. Mode1

1.0E-4
200
Sdisp. Phase Mode1

1.0E-2
Sdisp. Mag. Mode 2

1.0E-4
360
Sdisp. Phase Mode 2

160
0 Frequency (Hz) 20

Figure 9-5 X-Y Plots for the Bar Frequency Response Analysis (continued)*
260

1.0E3

Bend. Moment A1 El. 6 Mag.

1.0E-1

400
Bend. Moment A1 El. 6 Phase

0
0 Frequency (Hz) 20
Figure 9-5 X-Y Plots for the Bar Frequency Response Analysis (continued)
 CHAPTER 9 261
Plotted Output

4.0

Load Mag. 6
0
100
Load Phase 6

0
8.0
Load Mag. 11

0
100
LoadP hase 11

0
0 20
Frequency (Hz)
Figure 9-5 X-Y Plots for the Bar Frequency Response Analysis (continued)
Listing 9-6 shows X-Y plot commands for a modal transient response analysis of the
cantilever beam model. Plots are made for the applied loads (OLOAD) at grid points 6 and
11, displacements (DISP) of grid points 6 and 11, accelerations (ACCE) for grid points 6 and
11, bending moment (ELFORCE) at end A in plane 1 for element 6, and modal
displacements (SDISP) for modes 1 and 2. Figure 9-6 shows the resulting plots.
262

Listing 9-6 X-Y Plot Commands for the Bar Transient Response Analysis

$ X-Y PLOT COMMANDS FOR BAR MODAL TRANSIENT RESPONSE

... Executive, Case Control ...
$
$ PHYSICAL OUTPUT REQUEST
SET 11 = 6,11
DISPLACEMENT(PLOT) = 11
ACCELERATION(PLOT) = 11
$
$ MODAL SOLUTION SET OUTPUT
SET 12 = 1,2
SDISP(PLOT) = 12
$
$ ELEMENT FORCE OUTPUT
SET 13 = 6
ELFORCE(PLOT) = 13
$
$ APPLIED LOAD OUTPUT
SET 15 = 6,11
OLOAD(PLOT) = 15
$
OUTPUT(XYPLOT)
XGRID = YES
YGRID = YES
$
$ PLOT RESULTS
XTITLE = TIME
$
YTITLE = DISPL. 6
XYPLOT DISP /6(T2)
YTITLE = DISPL. 11
XYPLOT DISP /11(T2)
YTITLE = ACCEL. 6
XYPLOT ACCE /6(T2)
YTITLE = ACCEL. 11
XYPLOT ACCE /11(T2)
$
YTITLE = SDISP. 1
XYPLOT SDISP /1(T1)
YTITLE = SDISP. 2
XYPLOT SDISP /2(T1)
$
YTITLE = BEND. MOMENT A1, EL. 6
XYPLOT ELFORCE /6(2)
$
YTITLE = LOAD 6
XYPLOT OLOAD /6(T2)
YTITLE = LOAD 11
XYPLOT OLOAD /11(T2)
$
BEGIN BULK
$
... Bulk Data ...
$
ENDDATA
 CHAPTER 9 263
Plotted Output

0.03

Displ. 6
0.0

-0.02

0.08
Displ. 11

0.0

-0.06
0.0 2.0
Time (sec)
Figure 9-6 X-Y Plots for the Bar Transient Response Analysis
264

4.0

0.0

Accel. 6

-6.0

10.0
Accel. 11

0.0

-15.0
0.0 2.0
Time (sec)
Figure 9-6 X-Y Plots for the Bar Transient Response Analysis (continued)
 CHAPTER 9 265
Plotted Output

25.0

Bend. Moment A1, El. 6

0.0

-20.0
0.0 2.0
Time (sec)
Figure 9-6 X-Y Plots for the Bar Transient Response Analysis (continued)
266

0.08

Sdisp. 1
0.0

-0.10

0.0015
Sdisp. 2

0.0

-0.0010
0.0 2.0
Time (sec)
Figure 9-6 X-Y Plots for the Bar Transient Response Analysis (continued)
 CHAPTER 9 267
Plotted Output

3.0

Load 6
0.0

-3.0

6.0
Load 11

0.0

-6.0
0.0 2.0
Time (sec)
Figure 9-6 X-Y Plots for the Bar Transient Response Analysis (continued)
268
 NX Nastran Basic Dynamic Analysis User’s Guide

10
CHAPTER
Guidelines for Effective Dynamic
Analysis

■ Overview
■ Overall Analysis Strategy
■ Units

■ Mass
■ Damping
■ Boundary Conditions

■ Loads
■ Meshing
■ Eigenvalue Analysis

■ Frequency Response Analysis

■ Transient Response Analysis

■ Results Interpretation and Verification

■ Computer Resource Requirements

270

10.1 Overview
Dynamic analysis is more complicated than static analysis because of more input (mass,
damping, and time- and frequency-varying loads) and more output (time- and frequency-
varying results). Results from static analysis are usually easier to interpret, and there are
numerous textbook solutions for static analysis that make it relatively easy to verify certain
static analyses. Nevertheless, the guidelines in this chapter help you to perform dynamic
analysis in a manner that will give you the same level of confidence in the dynamic results
that you would have with static results.
This chapter covers the following topics:
• Overall analysis strategy
• Units
• Mass
• Damping
• Boundary conditions
• Loads
• Meshing
• Eigenvalue analysis
• Frequency response analysis
• Transient response analysis
• Results interpretation
• Computer resource estimation
 CHAPTER 10 271
Guidelines for Effective Dynamic Analysis

10.2 Overall Analysis Strategy

Part of any analysis strategy, whether it be for dynamic analysis or static analysis, is to gain
confidence with the modeling procedures first. The best way to accomplish this is to run
small, simple models, preferably models that have textbook solutions. The references (see
“References and Bibliography” on page 495) provide numerous textbook solutions. Start
with a simple model first and then gradually add complexity, verifying the results at each
stage. Follow the steps outlined below and in Figure 10-1.

Once you have confidence in a small model and are ready to analyze your actual model,
again do the analysis in steps. The following is a suggested order for performing dynamic
analysis on any structure:

1. Create the initial model only; do not apply any loads. Verify the model's
connectivity, element and material properties, and boundary conditions. Use a
graphical tool such as MSC/PATRAN to assist in this. Make sure that mass is
specified for this model.
2. Perform a static analysis (SOL 101) first in order to verify proper load paths and
overall model integrity. (Note that you have to constrain the structure for static
analysis even if you were not planning to do so for dynamic analysis.) For a three-
dimensional model, you should run three load cases, each with a 1g gravity load
applied in a different direction. Compute displacements and SPC forces, and
verify the results. Check for unusually large grid point displacements and
unreasonable SPC forces. The use of a graphical postprocessor can aid you at this
step.
Next, apply static loads that have the same spatial distribution that your
subsequent dynamic loads have. Verify the results for reasonableness. Do not go
to dynamic analysis until you are satisfied with the results from your static
analysis.
It is recommended at this stage that the model contain PARAM,GRDPNT,n
(where n is a reference grid point or 0, the origin of the basic coordinate system).
Verify the results from the grid point weight generator in order to ensure that the
model's rigid-body mass and inertia look reasonable. This step, in conjunction
with the static analysis results, helps to ensure that the proper mass units are
specified.
3. Perform an eigenvalue analysis (SOL 103) next. Compute only a few modes first,
verify their frequencies, and view their mode shapes for reasonableness. If your
graphical postprocessor can animate the mode shapes, do so because that helps
you to visualize them. Things to check at this step are local mode shapes, in
which one or a few grid points are moving a very large amount relative to the rest
of the model (this can indicate poor stiffness modeling in that region), and
unwanted rigid-body modes (which can arise due to improper specification of the
boundary conditions or a mechanism).
Once you are satisfied with these results, perform the full eigenvalue analysis (for
as many modes as you need).
272

4. If you have frequency-dependent loads, perform frequency response analysis

(SOL 108 or SOL 111) using the dynamic load spatial distribution. If your
structure is constrained, then apply the dynamic load at only one frequency,
which should be 0.0 Hz. Compare the 0.0 Hz displacement results to the static
analysis displacement results. The results should be the same if direct frequency
response (without structural damping) is used. If the results are not equal, then
there is probably an error in the specification of the dynamic load, and you should
check the LSEQ and DAREA entries. If modal frequency response (without
structural damping) is used, then the 0.0 Hz results should be close to the static
results; the difference is due to modal truncation.
Next apply the load across the entire frequency range of interest. If you are
running modal frequency response, then make sure that you have enough modes
to ensure accurate results for even the highest forcing frequency. Also be sure to
have a small enough ∆f in order to accurately capture the peak response. Verify
these results for reasonableness (it may be easier to look at magnitude and phase
results instead of real and imaginary results, which are the default values).
If your ultimate goal is a transient response analysis for which damping is to be
neglected, then the frequency response analysis can also omit damping.
However, if damping is to be included, then use the correct damping in your
frequency response analysis. The proper specification of damping can be verified
by looking at the half-power bandwidth.
Plots are important at this stage to assist in results interpretation. X-Y plots are
necessary in order to see the variation in response versus frequency. Deformed
structure plots at a frequency near a resonant frequency can also help to interpret
the results. If structure plots are made, look at the imaginary component because
the single degree-of-freedom (SDOF) displacement response at resonance is
purely imaginary when damping is present (this response does not occur in
practice because the response is usually due to several modes).
5. If you have time-dependent loads, perform transient response (SOL 109 or SOL
112) analysis. If your structure is constrained, apply the load "very quickly" (over
one or two time steps) as a step function and look at the displacement results. The
duration of the analysis needs to be as long as the period of the lowest frequency
mode. For an SDOF system, a quickly applied load results in a peak displacement
response that is twice the response resulting from the same load applied
statically. This peak response does not occur in your actual model because of
multiple modes and damping, but the results should be close. If your structure is
unconstrained, the displacements will grow with time (unless the rigid-body
modes are excluded in a modal transient response analysis). However, the
stresses should be roughly twice those from the static analysis. In any event,
examine the results via X-Y plots to ensure reasonableness.
Once you are satisfied, apply the correct time variation to the load and compute
the results. Again, use X-Y plots to verify the accuracy of the results.
 CHAPTER 10 273
Guidelines for Effective Dynamic Analysis

6. Finally, perform any other dynamic analyses, such as response spectrum analysis,
random response, nonlinear transient response, or dynamic response
optimization. The confidence gained by using the previous steps first helps to
ensure that you have an accurate model at this stage.
These and other guidelines are described further in the remainder of this chapter.

Verify Model Connectivity,

Properties, Boundary Conditions

Verify Static Analysis

(with gravity load, with actual load)

Verify Normal Modes Analysis

(few modes, many modes)

Verify Frequency Response Analysis

(0.0 Hz, entire frequency range)

Verify Transient Response Analysis

(step load, actual load)

Run Other Dynamic Response Analyses

(response spectrum, random response, etc.)

Figure 10-1 Simplified Flow Chart of the Overall Analysis Strategy

274

10.3 Units
Mistakes in units and boundary conditions (Section 10.6) are among the most common
mistakes made when performing finite element analysis. MSC.Nastran does not assume a
particular set of units, but it does assume that they are consistent.

It is up to you to ensure that the units that you use are both consistent and correct.
Table 2-2 lists consistent units for common variables.

There are several ways to verify units. For mass, you can print the results from the grid
point weight generator and verify that the mass is correct. For stiffness, you can apply a
simple load and verify that the resulting static displacements seem reasonable. For both,
you can verify that the natural frequencies are reasonable.
These checks assume that you have enough knowledge about your structure to know when
the results are reasonable and when a mistake has been made. In other words, running
small models and/or proceeding through dynamic analysis via the steps outlined in the
previous section are necessary in order to be confident that the correct units are specified.
 CHAPTER 10 275
Guidelines for Effective Dynamic Analysis

10.4 Mass
Mistakes with mass primarily involve mistakes in mass units as described in the previous
section. A common mistake is to mix mass and weight units. Using PARAM,WTMASS
does not solve this problem because it scales all mass and weight input (except certain
types of direct input matrices) and still leaves the mixture of units. Also, the use of
PARAM,WTMASS can have unwanted effects as well because it also scales the large mass
used in enforced motion, thereby scaling the value of the enforced motion input. In order
to reduce the chances for error, it is recommended that only mass units be used
everywhere. Therefore, avoid the use of weight input wherever possible as well as the use
of PARAM,WTMASS.

One way to verify mass input is to apply gravity in several directions and print the SPC
forces. This verification ensures that the mass is correctly applied and that the units are
correct.
276

10.5 Damping
Proper specification of damping is probably the most difficult modeling input to verify
because its verification can only be done via dynamic response analysis. In addition, its
units are not necessarily familiar because damping is normally not a modeling input that
you use frequently, unless dynamic response analysis is all you run. Also, there are several
ways to specify damping in MSC.Nastran, which increase the chances of making a mistake.
Finally, even though the damping units are correct and the damping is input correctly in
MSC.Nastran, it is difficult to know that the damping specification itself is correct from a
physical standpoint. While there are relatively easy checks for correctness in mass and
stiffness input, including comparison to static and modal test data, there are no such easy
checks for damping input.

The easiest way to specify damping is to use modal damping, which is often specified as
the percentage of critical damping. The TABDMP1 Bulk Data entry is used to specify
modal damping. The following are several potential mistakes associated with using modal
damping.
• Forgetting to select the TABDMP1 entry using the SDAMPING Case Control
command.
• Forgetting to specify the damping TYPE (field 3). The default damping type is
structural damping, which is twice the critical damping percentage.
• Not making the table cover an adequate range of frequencies. Like almost all of
the MSC.Nastran tables, the TABDMP1 entry extrapolates beyond the endpoints
of the table by using the first two or the last two entered values. As a rule, you
should try to provide enough input points so that the table lookup point is always
an interpolated value.
Modal damping can be used only in modal frequency response and modal transient
response. Other forms of damping have to be used for the direct methods of response. For
frequency response analysis, GE (field 9 of the MAT1 entry) and PARAM,G define
structural damping. These variables are also used to specify structural damping for
transient response analysis but are not activated unless PARAM,W3 and PARAM,W4 are
set to nonzero values. A common mistake is to forget to set these values.

Note that damping is additive, that is, the damping matrix is comprised of all of the input
damping sources. For example, damping due to CDAMPi elements adds to damping due
to PARAM,G and GE. Mixing of damping types can increase the chances for error, and you
are cautioned against the mixing of damping types unless it is necessary.
In many cases damping is not an important consideration. For example, a structure's peak
response due to an impulsive load is relatively unaffected by damping since the peak
response occurs during the first cycle of response. Damping in a long duration transient
excitation, such as an earthquake, can make a difference in the peak response on the order
of 10 to 20% or so, but this difference is small when compared to the other modeling
uncertainties. Therefore, it is often conservative to ignore damping in a transient response
analysis.
 CHAPTER 10 277
Guidelines for Effective Dynamic Analysis

For frequency response analysis, however, the value of damping is critical to the value of
the computed response at resonance since the dynamic magnification factor is inversely
proportional to the damping value. One way to verify the accuracy of the modal damping
ratio input is to run the modal frequency response across the half-power bandwidth of the
modal frequencies of interest as shown in Figure 10-2.

Peak Response

Peak
----------- = Half-Power Point
Response 2

f1 fn f2 Frequency

Half-Power
Bandwidth
Figure 10-2 Half-Power Bandwidth

For lightly damped structures ( ζ < 0.1 ), the approximate relationship between the critical
damping ratio , the half-power bandwidth ( f 2 – f 1 ), and the resonant frequency f n is

f2 – f1
ζ ≈ ---------------- Eq. 10-1
2f n

Another approximate way to verify damping is to run transient response analysis and look
at the successive peak values of the free vibration response as shown in Figure 10-3. In
direct transient response analysis, this plot can be generated by using a displacement initial
condition; in modal transient response analysis, it can be generated by applying an
impulsive force that ramps up and down quickly with respect to the dominant period of
response. The logarithmic decrement ∆ is the natural log of the amplitude ratio of two
successive cycles of free vibration response given by

x( n – 1 )
∆ = ln ------------------ Eq. 10-2
xn

The logarithmic decrement provides an approximate damping relationship for lightly

damped structures as given by

∆
ζ ≈ ------ Eq. 10-3
2π

In transient response analysis, remember to use PARAM,W3 or PARAM,W4 to include

structural damping if GE or PARAM,G is used.
278

Figure 10-3 Damped Free Vibration Response

Both the half-power bandwidth method and the logarithmic decrement method assume an
SDOF response. These approximations are less accurate when there are multiple modes of
response; however, they are useful for verifying that the damping input is within a factor
of two or three of the desired damping.
 CHAPTER 10 279
Guidelines for Effective Dynamic Analysis

10.6 Boundary Conditions

The proper specification of boundary conditions is just as important for dynamic analysis
as it is for static analysis. The improper specification of the boundary conditions leads to
incorrect answers. One such improper specification of boundary conditions is forgetting
to fully constrain the structure. Unlike static analysis, for which an under-constrained
model does not run, an under-constrained model does run in dynamic analysis. You
should perform an eigenvalue analysis first and verify that there are no unwanted rigid-
body modes.
The large mass used for enforced motion simulates a constrained condition as well as adds
a rigid-body mode. The value of the large mass is important for obtaining accurate
answers. The large mass value must be large enough to properly simulate the constrained
condition, but it must not be so large as to create numerical difficulties when solving the
equations. A range of 10 3 to 10 6 times the overall structural mass is adequate to meet both
conditions. One way to verify that a proper value is chosen is to run a normal modes
analysis with the enforced DOF constrained via SPCs. Then, run a normal modes analysis
with your choice of the large mass(es) and compare the frequencies of the flexible modes.
If the frequencies compare favorably (i.e., to within four or five significant digits), then the
large mass value is accurate. If the frequencies do not compare, then increase the value of
the large mass(es). A common mistake is to use too low a value (or omit it entirely) for the
rotational components. Using the grid point weight generator is very important to obtain
the overall structural mass and inertias so that you can specify a good value for the large
mass(es).
280

10.7 Loads
Because of their time- or frequency-varying nature, it is more complicated to apply
dynamic loads than it is to apply static loads. Therefore, it is important to verify that the
dynamic loads are correctly specified and that there are no discontinuous loads.

The best way to verify proper dynamic load specification is to plot the loads as a function
of time or frequency. Another way to verify proper dynamic load specification is to print
the loads as a function of time or frequency. Use the OLOAD Case Control command to
plot or print the loads.
 CHAPTER 10 281
Guidelines for Effective Dynamic Analysis

10.8 Meshing
An adequate element mesh is required to create an accurate model. For static analysis the
mesh density is primarily controlled by the load paths; the element mesh must be fine
enough so that there is a smooth transition of stress from one element to another in the
region of interest.
Load paths are also important for dynamic analysis, but there is an additional
consideration: the mesh must be fine enough to accurately represent the deformed shape
of the modes of interest. If many modes are to be considered in the analysis, then the model
must be fine enough to accurately represent the mode shapes of the highest modes of
interest. Table 10-1 shows the frequencies resulting from several cantilever beam models;
each column represents the same structure but with a different number of elements. Note
that the frequencies are closer to the theoretical results as the model fineness in increased.
In addition, note that the error is greatest in the higher modes. This table shows the
computed frequencies compared to theory; it does not show the computed mode shapes
compared to theory. However, the comparison for mode shapes shows even more error
than is shown for the frequencies.

A general rule is to use at least five to ten grid points per half-cycle of response amplitude.
Figure 10-4 shows the theoretical response shape for the fifth mode. Note that there are
four half-cycles in the mode shape, which means that 20 to 40 grid points at a minimum are
recommended to accurately represent that mode shape. This modeling guideline is also
reflected in Table 10-1, which shows that the 40-element model is much more accurate
than the 10-element model.

Table 10-1 Frequencies for a Cantilever Beam Model (Lumped Mass)

Frequencies (Hz)
Mode 10-Element 40-Element 70-Element 100-Element
Theory
Model Model Model Model
1 2.08 2.07 2.08 2.08 2.08
2 13.24 12.81 13.00 13.01 13.01
3 36.47 35.50 36.37 36.41 36.42
4 71.52 68.84 71.22 71.34 71.36
5 118.21 112.53 117.65 117.90 117.95

Four Half-Cycles
Figure 10-4 Fifth Mode Shape of a Cantilever Beam
282

Another way to verify the accuracy of the mesh density is to apply static loads that give a
deformed shape the same as the mode of interest and perform stress discontinuity
calculations. This process can be laborious and is not recommended as a general checkout
procedure.
 CHAPTER 10 283
Guidelines for Effective Dynamic Analysis

10.9 Eigenvalue Analysis

In addition to the meshing guidelines described earlier, the other primary factor in
eigenvalue analysis is the proper selection of the eigenvalue solution method. The Lanczos
method is the best overall method and is the recommended method to use. The automatic
Householder method is useful for small, dense matrices. The SINV method is useful when
only a few modes are required. The other methods should be regarded as backup methods.

Carefully examine the computed frequencies and mode shapes. Viewing only one or the
other is usually not enough to verify accuracy of your model. Modes with 0.0 Hz (or
computational zero) frequencies indicate rigid-body or mechanism modes. If these modes
are unintended, then there is a mistake in boundary conditions or connectivity. The
existence and cause of unintended zero-frequency modes can also be ascertained from the
mode shapes. In addition, mode shape plots are useful for assessing local modes, in which
a group of one or a few grid points displaces and the rest of the structure does not. Local
modes may also be unintended and are often the result of incorrect connectivity or element
properties.
284

10.10 Frequency Response Analysis

Several factors are important for computing accurate frequency response results. These
factors include the number of retained modes (for modal frequency response analysis), the
frequency increment ∆f , and damping. These guidelines are only approximate but are
nevertheless useful. Running a normal modes analysis first helps to compute accurate
frequency response results.

Number of Retained Modes

Use enough modes to cover the range of excitation. For example, if the model is to be
excited from 1 to 100 Hz, you must use all of the modes with frequencies up to at least 100
Hz. This is only a minimum requirement, however. A better guideline is to use enough
modes to cover two to three times the range of excitation in order to provide accurate
answers at the high end of the frequency range. For example, when excitation is applied
to 100 Hz, modes with frequencies up to 200 to 300 Hz should all be used.

Size of the Frequency Increment

The size of the frequency increment ∆f must be small enough to ensure that the magnitude
of the peak response is accurately computed. To ensure this, you need to choose a
frequency increment small enough so that there are at least five to ten increments within
the half-power bandwidth frequencies (illustrated in Figure 10-2).

The frequency increment is defined by ∆f on the FREQ1 Bulk Data entry. Note that FREQ
and FREQ2 entries can be used in conjunction with FREQ1 to define more solutions in the
areas of resonance (the frequencies of these solutions should have been determined by a
prior normal modes analysis). A nonuniform ∆f imposes no cost increase relative to a
uniform ∆f .

Relationship of Damping to the Frequency Increment

The response at resonance is inversely proportional to the amount of damping, but the
half-power bandwidth is directly proportional to the amount of damping. For lightly-
damped structures ( ζ < 0.1 ), an approximate relationship between the half-power
bandwidth ( f 2 – f 1 ), resonant frequency f n , and critical damping ratio ζ is given by Eq. 10-1.
This equation can be rewritten to define the appropriate value of ∆f :

f 2 – f 1 = ( m – 1 )∆f Eq. 10-4

f2 – f1 ( m – 1 )∆f
ζ ≈ ---------------- = -------------------------- Eq. 10-5
2f n 2f n

2ζf n
∆f = -------------------- Eq. 10-6
(m – 1)
 CHAPTER 10 285
Guidelines for Effective Dynamic Analysis

where m is the number of frequency points within the half-power bandwidth. For
example, with 1% critical damping and six points used to define the half-power
bandwidth, the maximum frequency increment is 0.004 ⋅ f n . The frequency increment is
smaller for lighter damped structures.
Another good check is looking at the X-Y plots. If the response curves are not smooth, there
is a good chance that ∆f is too large.

Verification of the Applied Load

The applied load can be verified by exciting your model at 0.0 Hz and comparing the
results to a static solution with the same spatial load distribution. The 0.0 Hz results should
match the static results if direct frequency response analysis is used without structural
damping. If modal frequency response analysis is used without structural damping, those
results should be close to the static results; any difference is due to mode truncation. If the
0.0 Hz results do not match the static results, check the LSEQ and DAREA entries. Also,
use OLOAD to print the applied force in order to compare the loads.
286

10.11 Transient Response Analysis

Several factors are important in computing accurate transient response. These factors
include: the number of retained modes (for modal transient response), the integration time
step ∆t , the time duration of the computed response, and damping. The guidelines are
only approximate but are nevertheless useful. Running a normal modes analysis first helps
to compute transient response.

Number of Retained Modes

In modal transient response analysis, a larger number of modes produces a more accurate
response (although at the expense of increased run times). The number of modes must be
large enough to cover the frequency range of interest. (The term "range of interest" means
the range of frequencies whose response is to be computed as well as the range of
frequencies for which the loading is applied.) As a general rule, there should be enough
modes to cover a range up to about two times the highest frequency of interest. For
example, if response needs to be computed to 100 Hz, then enough modes should be used
in the transient response to cover up to at least 200 Hz. As another general rule, you should
use enough modes to cover a range up to two to ten times the dominant frequency of
response.

A sufficient number of modes must be retained to cover the time and spatial distribution
of loading. For example, if the applied transient load has a spatial distribution resembling
the fifth mode shape, then that mode should be included in the transient response.

Size of the Integration Time Step

The value of the integration time step ∆t , denoted by DT on the TSTEP Bulk Data entry, is
important for the accurate integration of the equations of motion. Rough guidelines for the
selection of ∆t are as follows:
• ∆t must be small enough to accurately capture the magnitude of the peak
response, which means that at least ten time steps per cycle of the highest mode
be used. For example, if the highest frequency of interest is 100 Hz, then ∆t
should be 0.001 second or smaller.
• ∆t must be small enough to accurately represent the frequency content of the
applied loading. If the applied loading has a frequency content of 1000 Hz, then
∆t must be 0.001 second or less (preferably much less in order to represent the
applied loading with more than one point per cycle).

The integration time step can be changed in a transient response analysis, but it is not
recommended. Much of the cost of direct transient response occurs with the
decomposition of the dynamic matrix, which occurs only once if there is a constant ∆t . A
new decomposition is performed every time ∆t changes, thereby increasing the cost of the
analysis if the time increment is changed. Therefore, the recommendation is to determine
the smallest ∆t required based on the loading and frequency range and then use that ∆t
throughout the entire transient response analysis.
 CHAPTER 10 287
Guidelines for Effective Dynamic Analysis

Duration of the Computed Response

The length of the time duration is important so that the lowest flexible (e.g., non-rigid
body) mode oscillates through at least one cycle. For example, if the lowest flexible mode
has a frequency of 0.2 Hz, then the time duration of the computed response should be at
least 5.0second. A better guideline is to make the duration the longer of the following:
twice the period of the lowest flexible mode or one period of the lowest flexible mode after
the load has been removed or reached a constant value. The time duration is set on the
TSTEP entry by multiplying (the integration time step) by the number of time steps (N).

Value of Damping
The proper selection of the damping value is relatively unimportant for analyses that are
of very short duration, such as a crash impulse or a shock blast. The specification of the
damping value is most important for long duration loadings (such as earthquakes) and is
critical for loadings (such as sine dwells) that continually add energy into the system.

Verification of the Applied Load

The applied load can be verified by applying the load suddenly (over one or two time
increments) and comparing the results to a static solution with the same spatial load
distribution. The transient results should have a peak value of twice the static results. If
the transient results are not twice the static results, check the LSEQ and DAREA entries.
Another way to verify the applied load is to inspect it visually via the X-Y plots.
288

10.12 Results Interpretation and Verification

Because of the time- and frequency-varying nature of dynamic analysis, results can be
more difficult to interpret than for static analysis. The key to proper results interpretation
is plotting. Structure plotting is necessary for a proper understanding of the deformed
shapes of the modes, and X-Y plotting is necessary for a proper understanding of the
frequency and transient response.

Comparing results to hand calculations, to known results from similar models, or to test
data is also very useful. Do not accept any analysis results without first performing some
type of verification.

Consider the cantilever beam model shown in Figure 10-5. This is a planar model of the
cantilever beam used in the examples in Chapter 3 through Chapter 7. Figure 10-5 shows
the loads applied to the beam.

6 11

1N 2N

Figure 10-5 Cantilever Beam Model with Static Loads

A static analysis (SOL 101) is run first. Then, modal frequency response (SOL 111) is run
from 0.0 to 20.0 Hz with a damping ratio of 2% critical damping used for all modes. Modes
are computed up to 500 Hz using the Lanczos method. Finally, modal transient response
(SOL112) is run with the time variation shown in Figure 10-6. Modes are computed up to
3000Hz (using the Lanczos method), and a damping ratio of 2% critical damping is used
for all modes.

1.0

Load
Factor

0.003 0.006
Time (sec)
Figure 10-6 Time Variation of Transient Loads

Table 10-2 shows the y-displacements for grid points 6 and 11. As expected, the frequency
response results at 0.0 Hz are the same as the static analysis results (see “Frequency
Response Analysis” on page 284). The frequency response results at 2.05 Hz (very near
the first natural frequency) are approximately 25 times the static analysis results. The
factor of 25 is the dynamic amplification factor at resonance for a damping ratio of 2%
critical damping.
 CHAPTER 10 289
Guidelines for Effective Dynamic Analysis

1 1
Amplification Factor at Resonance = ------ = ---------------- = 25
2ζ 2 ( .02 )

The transient response results are approximately twice the static analysis results. The
factor of two is the amplification of response for a transient load applied suddenly (see
“Transient Response Analysis” on page 286).

Table 10-2 Comparison of Results for the Cantilever Beam Model

Frequency Response
Static Analysis Analysis Results Transient Response
Results Peak Displacement
0.0 Hz 2.05 Hz
Displacement 6 Y 3.17E-3 3.17E-3 7.61E-2 6.29E-3
Displacement 11 Y 9.77E-3 9.77E-3 2.25E-1 1.87E-2
290

10.13 Computer Resource Requirements

The efficiency of a dynamic analysis can be measured based on computer resource
requirements, such as computer runtime, disk storage, and memory. These requirements
increase as the problem size increases, but they can also vary greatly depending on other
factors.
In general, a dynamic analysis uses more computer resources than a static analysis. In
normal modes analysis, the requirements vary depending on the density of the model, the
eigenvalue extraction method used, the number of modes requested, and the type of
decomposition (symmetric or sparse).

If your model has already been built, estimate its resource requirements by using the new
ESTIMATE utility. For instructions on the use of the ESTIMATE utility, see Section 6.1 of
the MSC.Nastran Configuration and Operations Guide for Version 69. If your model has not
been created, but you wish to estimate resource usage, then read the following.
In the past, we have established benchmark runs to provide guidelines on performance for
normal modes analyses. We used a cylindrical plate model scaled to various sizes and
solved for ten modes using different solution methods. Testing has shown that this model
provides a very good guideline for most industry models.
Since the previous edition of this book established that the Sparse Lanczos method is far
superior in performance than other methods, we now only use this one method for the
benchmark runs. Using MSC.Nastran Version 69 on an SGI R10000 workstation, we
applied the Sparse Lanczos method to models of various sizes and extracted data on
memory and disk space requirement.
The memory requirements increase with problem size. Empirical formulas were
developed for these benchmark problems to estimate the memory requirement given the
number of degrees-of-freedom (DOF) contained in the model.
For the equation

Log(mem) = (.7934) ⋅ Log(DOF) – 2.3671 Eq. 10-7

mem The memory required in megabytes.

DOF The number of degrees-of-freedom contained in the model.

Note that these memory requirements are for no spill conditions. The symmetric solver can
run with less memory, but because of spill conditions, the performance is degraded in both
computer runtime and disk space usage.

Empirical formulas were developed for these benchmark problems to estimate the disk
space requirement given the number of degrees-of-freedom (DOF) contained in the model.
 CHAPTER 10 291
Guidelines for Effective Dynamic Analysis

For the equation

Log(space) = (1.0815) ⋅ Log(DOF) – 2.5411 Eq. 10-8

space The disk space usage in megabytes.

DOF Represents the number of degrees-of-freedom contained in the model.

These values were obtained from a 32-bit computer and therefore should be doubled for a
64-bit computer. If you wish to calculate more than ten modes, then multiply the space
calculated above by 1.18^(#modes/10 - 1). For example, for 30 modes you would multiply
the space value calculated above by 1.18^2.
292
 NX Nastran Basic Dynamic Analysis User’s Guide

11
CHAPTER
Advanced Dynamic Analysis
Capabilities

■ Overview
■ Dynamic Reduction
■ Complex Eigenvalue Analysis

■ Response Spectrum Analysis

■ Random Vibration Analysis
■ Mode Acceleration Method

■ Fluid Structure Interaction

■ Nonlinear Transient Response Analysis
■ Superelement Analysis

■ Design Optimization and Sensitivity

■ Control System Analysis

■ Aeroelastic Analysis

■ DMAP
294

11.1 Overview
The previous chapters describe the most common types of dynamic analysis: normal
modes analysis, frequency response analysis, and transient response analysis. However,
MSC.Nastran contains numerous additional types of dynamic analyses; many of these
types are documented in the NX Nastran Advanced Dynamic Analysis User’s Guide.
The advanced dynamic analysis capabilities include
• Dynamic reduction
• Complex eigenanalysis
• Response spectrum analysis
• Random response analysis
• Mode acceleration method
• Fluid-structure interaction
• Nonlinear transient response analysis
• Superelement analysis
• Design sensitivity and optimization
• Control systems
• Aeroelastic analysis
• DMAP
These capabilities are described briefly in the remainder of this chapter.
 CHAPTER 11 295
Advanced Dynamic Analysis Capabilities

11.2 Dynamic Reduction

Dynamic reduction is an optional numerical approach that can be used in MSC.Nastran to
reduce a dynamic model to one with fewer degrees-of-freedom. Typically, the intent of
dynamic reduction is to increase the efficiency of a dynamic solution by working with
smaller matrices while maintaining the dynamic characteristics of the system.
Dynamic reduction is used for a number of reasons. One possible reason may be that a
particular model may be too large from a computer resource standpoint (computer
runtime and/or disk space) to be solved without using reduction. A related issue is that
the model may have more detail than required. Many times dynamic analyses are
performed on models that were initially created for detailed static stress analyses, which
typically require refined meshes to accurately predict stresses. When these static models
are used in a dynamic analysis, the detailed meshes result in significantly more detail than
is reasonably required to predict the natural frequencies and mode shapes accurately.
Two methods are available in MSC.Nastran for the dynamic reduction of the
eigenequation prior to modal extraction: static condensation (also called Guyan reduction)
and generalized dynamic reduction (GDR). Once the natural frequencies and mode shapes
are calculated for the reduced model, these modes can be used in the transient or frequency
response solution process.

Static Condensation. In the static condensation process you select a set of dynamic DOFs
called the a-set; these are the retained DOFs that form the analysis set. The complementary
set is called the o-set and is the set of DOFs that are omitted from the dynamic analysis
through a reduction process. The process distributes the o-set mass, stiffness, and
damping to the a-set DOFs by using a transformation that is based on a partition of the
stiffness matrix (hence the term static condensation). This reduction process is exact only
for static analysis, which leads to approximations in the dynamic response. The a-set DOFs
are defined by the ASET or ASET1 Bulk Data entires, and the o-set DOFs are defined by the
OMIT or OMIT1 Bulk Data entries.

Generalized Dynamic Reduction (GDR). Generalized dynamic reduction (GDR) offers a

more accurate solution by including an approximation of the dynamic effects when
forming the transformation. Bulk Data entries for GDR are DYNRED, QSET, and QSET1,
and SPOINT. In addition, the Case Control command DYNRED is also required.
It is emphasized that dynamic reduction is an optional technique and is best left to the
dynamic analysis specialist. Whereas dynamic reduction was required in the days of small
computer memory and disk space, now it is no longer required due to increased computer
resources and better eigenanalysis methods (in particular, the Lanczos method).
296

11.3 Complex Eigenvalue Analysis

Complex eigenvalue analysis is used to compute the damped modes of structures and
assess the stability of systems modeled with transfer functions (including
servomechanisms and rotating systems).

Complex eigenvalue analysis solves for the eigenvalues and mode shapes based on the
following equation in operator notation:

2
[ p M + pB + K ] { u } = 0 Eq. 11-1

where: p = α + iω
This equation is similar to that for normal modes analysis (see “Real Eigenvalue Analysis”
on page 35) except that damping is added and the eigenvalue is now complex. In addition,
the mass, damping, and stiffness matrices may be unsymmetric, and they may contain
complex coefficients.
Complex eigenvalue analysis is controlled with the EIGC Bulk Data entry (similar to the
EIGRL or EIGR Bulk Data entries for normal modes analysis). There are four methods of
solution: upper Hessenberg, complex Lanczos, determinant search, and inverse power.
Complex eigenvalue analysis is available as a direct method (SOLs 107, 67, or 28), in which
the equations are of the same size as the number of physical variables. Complex eigenvalue
analysis is also available as a modal method (SOLs 110, 70, and 29), in which undamped
modes are first computed and then are used to transform the matrices from physical to
modal variables.
 CHAPTER 11 297
Advanced Dynamic Analysis Capabilities

11.4 Response Spectrum Analysis

Response spectrum analysis is an approximate method of computing the peak response of
a transient excitation applied to a simple structure or component. This method is used in
civil engineering to predict the peak response of a component on a building that is
subjected to earthquake excitation; it is also used in aerospace engineering to predict the
peak response of equipment in a spacecraft that is subjected to an impulsive load due to
stage separation. Because it is an approximate method, response spectrum analysis is often
used as a design tool. Response spectrum analysis is also called shock spectrum analysis.
There are two parts to response spectrum analysis: (1) generation of the spectrum and
(2) use of the spectrum for dynamic response such as stress analysis. Both are available in
MSC.Nastran.
Figure 11-1 depicts the generation of a response spectrum.

(d)
f1 f2 f3 fmax (e)
(c)
... Peak
(b)
Response

Series of f1 f2 f3 . . . fmax
(a) Transient
Response Oscillators Resonant Frequency

Base
Transient Structure
Excitation
Transient excitation (a) is applied to a base structure (b), from which
transient response (c) is computed for each floor. This response is
applied to a series of oscillators (d), for which the peak response is
plotted (e). Steps (d) and (e) are repeated for different damping values
to form response spectra as shown below.
= 5% critical damping
= 2% critical damping
= 0% critical damping
Peak
Response

Resonant Frequency (f)

Figure 11-1 Response Spectrum Generation

Note that the peak response for one oscillator does not necessarily occur at the same time
as the peak response for another oscillator. Also note that there is no phase information
since only the magnitude of peak response is computed. It is assumed in this process that
each oscillator mass is very small relative to the base structural mass so that the oscillator
does not influence the dynamic behavior of the base structure.
298

Once a spectrum is computed, it can be used for the dynamic response analysis of an
MSC.Nastran model of the component. For example, the spectrum generated for a floor in
a building that is subjected to an earthquake can then be applied to a complex model of a
piece of equipment attached to that floor. The peak response of each mode of the
equipment model is obtained from the spectrum, and these peak modal responses are
combined to create the overall response.

Because the peak responses do not all occur at the same time and only the magnitude of
peak responses are computed, various methods are used to combine the peak responses
into the overall response. The combination methods implemented in MSC.Nastran are
SRSS (square root of the sum of the squares), ABS (absolute values), and NRL (U.S. Navy
shock design modal summation). The typical response quantities computed are grid point
displacements and element stresses.
 CHAPTER 11 299
Advanced Dynamic Analysis Capabilities

11.5 Random Vibration Analysis

Random vibration is vibration that can be described only in a statistical sense. The
instantaneous magnitude is not known at any given time; rather, the magnitude is
expressed in terms of its statistical properties (such as mean value, standard deviation, and
probability of exceeding a certain value).
Examples of random vibration include earthquake ground motion, ocean wave heights
and frequencies, wind pressure fluctuations on aircraft and tall buildings, and acoustic
excitation due to rocket and jet engine noise. These random excitations are usually
described in terms of a power spectral density (PSD) function.

MSC.Nastran performs random response analysis as a postprocessing step after frequency

response analysis. The frequency response analysis is used to generate the transfer
function, which is the ratio of the output to the input. The input PSD multiplies the transfer
function to form a response PSD. The input PSD can be in the form of auto- or
cross-spectral densities. Random response output consists of the response PSD,
autocorrelation functions, number of zero crossings with positive slope per unit time, and
RMS (root-mean-square) values of response.
300

11.6 Mode Acceleration Method

The mode acceleration method is an alternate form of data recovery for modal frequency
response and modal transient response. The mode acceleration method accounts for the
higher truncated modes to give more accurate answers than either the mode displacement
or matrix methods (see “Modal Frequency Response Analysis” on page 108 and “Modal
Transient Response Analysis” on page 158), particularly if the number of retained modes
is small compared to the number of physical degrees-of-freedom.

The higher modes respond in a quasi-static manner to lower frequency excitation.

Therefore, the inertia and damping forces contain little contribution from these higher
modes. In the mode acceleration method the inertia and damping forces are first computed
from the modal solution. These forces are then combined with the applied forces and are
used to obtain more accurate displacements by performing a static analysis. The mode
acceleration method provides more accurate answers but with increased cost.
The mode acceleration method is selected by specifying PARAM,DDRMM,-1 and
PARAM,MODACC,0.
 CHAPTER 11 301
Advanced Dynamic Analysis Capabilities

11.7 Fluid Structure Interaction

Four major methods are available in MSC.Nastran to model fluid effects. These methods
are described below.

Hydroelastic Analysis
Small motions of compressible or incompressible fluids coupled to a structure may be
analyzed with this option. The fluid is modeled with axisymmetric hydroelastic elements
(CFLUIDi), which may be connected to an arbitrary structure modeled with an
axisymmetric wetted surface. Each fluid point (RINGFL) on a cross section defines the
scalar pressure, which is expanded to a Fourier series around the circumference. Complex
modes and frequency response solutions are available for the coupled fluid-structure
problems. Normal modes solutions are available for fluid-only problems. All solutions
may include gravity effects (i.e., sloshing) on a free surface. This capability was developed
specifically to analyze liquid-fueled booster rockets but may also be useful for problems
involving other types of axisymmetric storage tanks.

Virtual Fluid Mass

Small motions of incompressible fluids may be coupled directly to the structure with this
option. Fluids may be coupled to the interior and exterior surfaces (with infinite fluid
boundaries). There is no explicit fluid model; only the wetted structural elements (ELIST)
have to be defined. Although free surfaces are allowed, no gravity effects are included
directly. Since the fluid is represented by a coupled mass matrix attached directly to the
structural points, this capability is allowed in all dynamic solution sequences. This
capability may be used to model a wide variety of fluid-structure interaction problems.
Some examples are fuel tanks, nuclear fluid containers, drilling platforms, and underwater
devices.

Coupled Acoustics
You may analyze the dynamics of compressible fluids contained within arbitrarily shaped
structures with the coupled fluid-structure method. You would model a
three-dimensional fluid with conventional solid elements (CHEXA, etc.) using acoustic
property and material data. Each grid point in the fluid defines the pressure at its location
as its degree of freedom. The fluid is automatically connected to the structure via the
geometry and ACMODL Bulk Data inputs. You can connect acoustic absorber elements
(CHACAB) to the structural surfaces to simulate soundproofing material. In addition, the
CAABSF element is now available. This element acts as a thin layer of acoustic absorbing
material along the fluid-structure interface. Version 69 introduced several features, which
are parallel to those available for structural analysis, such as direct damping, modal
damping and the ability to control the modes in a response analysis through the use of
parameter. You can define panels to provide integrated response data. Effects of gravity,
large motions, and static pressures are ignored. Complex eigenvalues, frequency response,
and transient response are the available solution sequences. Design sensitivity and
optimization processes may reference the acoustic outputs as responses with appropriate
302

design constraints. Applications for the coupled fluid-structure option are automotive and
truck interiors, aircraft cabins, and acoustic devices, such as loudspeakers and
microphones.

Uncoupled Acoustics
Several methods are available in MSC.Nastran for the analysis of normal modes of
compressible fluids bounded by rigid containers and/or free surfaces. One method is the
“acoustic cavity” capability, which uses two-dimensional slot elements and axisymmetric
ring elements to define the fluid region. This method was specifically developed for the
acoustic analysis of solid rocket motor cavities. A better method is to use the
three-dimensional fluid elements for the “coupled acoustics” described above and provide
the appropriate boundary conditions.
 CHAPTER 11 303
Advanced Dynamic Analysis Capabilities

11.8 Nonlinear Transient Response Analysis

The analysis techniques described thus far are applicable for linear-elastic analysis for
which the mass, stiffness, and damping matrices are constant over time and constant for all
values of applied force. However, there are many cases for which the matrices are not
constant, and these cases must be solved with nonlinear analysis techniques.
Nonlinear analysis requires iterative solution methods, thereby making it far more
computationally intensive than a corresponding linear analysis. Nonlinear transient
response analysis is available in MSC.Nastran in SOLs 29 and 129 (preferred). Nonlinear
problems are classified into three broad categories: geometric nonlinearity, material
nonlinearity, and contact.

Geometric Nonlinearity
Geometrically nonlinear problems involve large displacements; “large” means that the
displacements invalidate the small displacement assumptions inherent in the equations of
linear analysis. For example, consider a thin plate subject to an out-of-plane load. If the
deflection of the plate’s midplane is approximately equal to the thickness of the plate, then
the displacement is considered large, and a linear analysis is not applicable.
Another aspect of geometric nonlinear analysis involves follower forces. Consider a
clamped plate subject to a pressure load. As shown in Figure 11-2, the load has followed
the plate to its deformed position. Capturing this behavior requires the iterative update
techniques of nonlinear analysis.

Figure 11-2 Follower Forces on a Cantilever Beam

Material Nonlinearity
Material nonlinear analysis can be used to analyze problems where the stress-strain
relationship of the material is nonlinear. In addition, moderately large strain values can be
analyzed. Examples of material nonlinearities include metal plasticity, materials (such as
soils and concrete), and rubbery materials (where the stress-strain relationship is nonlinear
elastic). Various yield criteria, such as von Mises or Tresca (for metals) and Mohr-Coulomb
or Drucker-Prager (for frictional materials, such as soils or concrete), can be selected. Three
hardening rules are available in MSC.Nastran: isotropic hardening, kinematic hardening,
or combined isotropic and kinematic hardening. With such generality, most plastic
material behavior with or without the Bauschinger effect can be modeled. In addition,
gaps can be used to model the effects due to structural separation.
304

Contact
Contact occurs when two or more structures (or portions of structures) collide. Contact can
be modeled as point-to-point contact (CGAP) or as contact along a line (BLSEG).

Nonlinear-Elastic Transient Response Analysis

There are numerous structures that contain nonlinear elastic elements. These elements
possess nonlinear force-deflection relationships, yet they are elastic since they load and
unload along the same force-deflection curve. Examples of the nonlinear elastic
force-deflection curves are shown in Figure 11-3. These types of elements are common in
many mechanical and structural systems including piping systems supported by nonlinear
springs, chains (stiffness in tension only), or constant force devices; base-mounted
equipment and structure supported by nonlinear shock isolation systems; structures with
gaps (piping systems and buildings); and soil or concrete structures (which only exhibit
stiffness when in compression). Other systems exhibit nonlinearities that are proportional
to the velocity, such as seat belts and piping supports.

Force
Force

Displacement Displacement

(a) Bilinear Spring (b) Preload Constant Force

Device with Hard Stop
Force

Force

Displacement Displacement

(c) Gap and Spring in (d) Stiffness in Compression

Series Only
Force

Force

Displacement Displacement

(e) Softening (f) Hardening

System System

Figure 11-3 Examples of Nonlinear Elastic Systems

 CHAPTER 11 305
Advanced Dynamic Analysis Capabilities

There are several methods in MSC.Nastran for solving nonlinear elastic problems. A
general nonlinear solution scheme can be used, in which the material properties are
specified as nonlinear elastic (NLELAST on the MATS1 entry). Nonlinear element stiffness
matrices are generated, and equilibrium iterations and convergence checks are required at
each load step. The new CBUSH element, introduced in Version 69 adds some capabilities
in this area, also. In addition, gap elements (CGAP) can be used in conjunction with elastic
elements to produce systems with piece-wise linear force-deflection curves. This process
also requires the additional computations of nonlinear element stiffness matrix generation,
equilibrium iteration, and convergence testing. Nonlinear analyses requiring these
additional computations can be substantially more costly than an equivalent linear
analysis even if the nonlinearities are relatively few in number.
An efficient technique called the pseudoforce method exists in MSC.Nastran, in which the
localized nonlinearities are treated so that they require no additional computer time when
compared to a purely linear analysis. In this method, which is available only for transient
response solutions (SOLs 109 and 112, for example), deviations from linearity are treated
as additional applied loads. The dynamic equations of motion are written as

[ M ] { u·· ( t ) } + [ B ] { u· ( t ) } + [ K ] { u ( t ) } = { P ( t ) } + { N ( t ) } Eq. 11-2

where [ M ] , [ B ] , and [ K ] denote the system mass, damping, and stiffness matrices,
respectively. The vectors { P ( t ) } and { u ( t ) } denote applied nodal loads and system
displacements, respectively, as functions of time. The vector { N ( t ) } denotes the nonlinear
forces, which are added to the right-hand side of Eq. 11-2 (and hence are treated as
additional applied loads).
The nonlinear forces are evaluated at the end of one time step for use in the successive time
step. The equations of motion therefore become the following:

[ M ] { u·· ( t ) } + [ C ] { u· ( t ) } + [ K ] { u ( t ) } = { P ( t ) } + { N ( t – ∆t ) } Eq. 11-3

Note that the nonlinear force lags the true solution by a time step, which may require using
small integration time steps (i.e., smaller than those required for a purely linear analysis).
Eq. 11-3 can be solved in physical or modal coordinates (the nonlinearity itself must be
expressed in terms of physical coordinates).

A nonlinear force can be used in conjunction with a linear elastic element to produce the
desired force-deflection curve as illustrated in Figure 11-4. The nonlinear dynamic force is
formulated using a NOLINi entry and a TABLEDi entry, which contains a
force-versus-deflection table describing the nonlinear force. For desired force-deflection
curves more complicated than the bilinear stiffness shown in the figure, the nonlinear force
is made correspondingly more complex.
306

Force

Force

Force
= +
Displacement Displacement Displacement

Nonlinear = Elastic + Nonlinear

Element Element Force

Figure 11-4 Formulation of a Nonlinear Element

Nonlinear Normal Modes Analysis

There are times when normal modes need to be calculated for a nonlinear structure. For
inflatable, rotating, or tension structures, the stiffness—and therefore, the modes—
depends upon the applied load and deformation. A nonlinear analysis can be restarted
into SOL 103 to compute the normal modes of the deformed structure.
 CHAPTER 11 307
Advanced Dynamic Analysis Capabilities

11.9 Superelement Analysis

Superelement analysis is a form of substructuring wherein a model is divided into pieces
with each piece represented by a reduced stiffness, mass, and damping matrix. Each
superelement is processed individually, and the solutions are then combined to solve the
entire model. The final analysis (in which all of the individual superelement solutions are
combined) involves much smaller matrices than would be required if the entire model
were solved in a one-step process. An example of a superelement model is shown in
Figure 11-5.
Superelement analysis has the advantage of reducing computer resource requirements,
especially if changes are made to only one portion (superelement) of the model; in this case,
only the affected superelement needs to be reanalyzed and the final analysis repeated.
Superelement analysis presents procedural advantages as well, particularly when multiple
engineering contractors are involved in an analysis. Imagine a model of a rocket and
payload: one contractor models the booster, another contractor models the engines, and
another contractor models the satellite. Each contractor can reduce his model to its
boundary degrees-of-freedom, which is suitable for superelement analysis. The systems
integrator then combines these reduced models into one model for a liftoff analysis.
Superelement analysis has the advantage that matrices can be passed from one
organization to another without revealing proprietary modeling details or concern about
whether the same superelement interior grid point and element numbers are used by every
participant.

Figure 11-5 Superelements Used to Model a Car Door

Component Mode Synthesis

Component mode synthesis (CMS) is a form of superelement dynamic reduction wherein
matrices are defined in terms of modal coordinates (corresponding to the superelement
modes) and physical coordinates (corresponding to the grid points on the superelement
308

boundaries). CMS is advantageous because there are fewer modal coordinates than
physical coordinates—perhaps only one percent as many. In addition, CMS can utilize
modal test data, thereby increasing the accuracy of the overall analysis.
 CHAPTER 11 309
Advanced Dynamic Analysis Capabilities

11.10 Design Optimization and Sensitivity

Design optimization is used to produce a design that possesses some optimal
characteristics, such as minimum weight, maximum first natural frequency, or minimum
noise levels. Design optimization is available in MSC.Nastran in SOL 200, in which a
structure can be optimized considering simultaneous static, normal modes, buckling,
transient response, frequency response, and aeroelastic analyses.

Possible analysis response types include:

• Weight
• Volume
• Eigenvalues
• Buckling load factor
• Static displacement, stress, strain, and element force
• Composite stress, strain, and failure criterion
• Frequency response displacement, velocity, acceleration, stress, and force
• Transient response displacement, velocity, acceleration, stress, and force
• Damping level in a flutter analysis
• Trim and stability derivative responses for static aeroelastic analysis
A frequent task is to minimize the total structural weight while ensuring that the design
still satisfies all of the design-related and performance-related constraints. The structural
weight is termed the design objective, which the optimizer attempts to minimize. The
design variables are the properties that can be changed to minimize the weight. The design
variables shown in Figure 11-6 are the height and width of the flange and web. Constraints
on the design variables may be a minimum allowable flange width w and a maximum
allowable beam height h . With the tip load shown in the figure, one performance
constraint on the design may be the maximum allowable transverse displacement at the
tip, while another constraint may be the maximum allowable bending stress. The
design-related and performance-related constraints place limits on the optimizer’s ability
to reduce the weight.

P
tw

h
tf
l
w

Figure 11-6 Cantilever I-Beam Example

310

The optimizer in MSC.Nastran uses sensitivity analysis. The sensitivity matrix [ S ] defines
the change in response for a perturbation in design variable. A term in [ S ] is given by

∂R i
S ij = -------- Eq. 11-4
∂X j

where:

S ij = i, j term in [ S ]

∂ = partial derivative
R i = i-th response

X j = j-th design variable

The sensitivity is depicted in Figure 11-7, in which the sensitivity is the slope of the curve
of response versus the design variable value. Design optimization and design sensitivity
are described in greater depth in the NX Nastran Design Sensitivity and Optimization User’s
Guide.

Response R

Design Variable X

Figure 11-7 Response Sensitivity

 CHAPTER 11 311
Advanced Dynamic Analysis Capabilities

11.11 Control System Analysis

A control system provides feedback (output) to an input. In aircraft, sample control
systems are flap settings and actuator positions.
A control system involves a general input-output relationship called a transfer function.
Transfer functions are represented in MSC.Nastran by the TF Bulk Data entry. Transfer
functions can utilize grid points, extra points (EPOINT), and scalar points (SPOINT).
Nonlinearities can be simulated via the NOLINi entries with which nonlinear transient
loads are expressed as functions of displacements or velocities. Complex eigenvalue
analysis can be used to determine stability when control systems include damping and
unsymmetric matrices.
312

11.12 Aeroelastic Analysis

MSC.Nastran provides efficient solutions of the problems of aeroelasticity, which is a
branch of applied mechanics that deals with the interaction of aerodynamic, inertial, and
structural forces. It is important in the design of airplanes, helicopters, missiles,
suspension bridges, tall chimneys, and power lines. Aeroservoelasticity is a variation in
which the interaction of automatic controls requires additional consideration.

The primary concerns of aeroelasticity include flying qualities (that is, stability and
control), flutter, and structural loads arising from maneuvers and atmospheric turbulence.
Methods of aeroelastic analysis differ according to the time dependence of the inertial and
aerodynamic forces that are involved. For the analysis of flying qualities and maneuvering
loads wherein the aerodynamic loads vary relatively slowly, quasi-static methods are
applicable. The remaining problems are dynamic, and the methods of analysis differ
depending on whether the time dependence is arbitrary (that is, transient or random) or
simply oscillatory in the steady state.

MSC.Nastran considers two classes of problems in dynamic aeroelasticity: aerodynamic

flutter and dynamic aeroelastic response.

Aerodynamic Flutter
Flutter is the oscillatory aeroelastic instability that occurs at some airspeed at which the
energy extracted from the airstream during a period of oscillation is exactly dissipated by
the hysteretic damping of the structure. The motion is divergent in a range of speeds above
the flutter speed. Flutter analysis utilizes complex eigenvalue analysis to determine the
combination of airspeed and frequency for which the neutrally damped motion is
sustained (see Figure 11-8).
 CHAPTER 11 313
Advanced Dynamic Analysis Capabilities

0.2
1st Mode
2nd Mode
0.1

Damping, g
0.0

Flutter Divergence

-0.1

-0.2
0 500 1000 1500 2000
Velocity (ft/sec)
Figure 11-8 Flutter Stability Curve

Three methods of flutter analysis are provided in MSC.Nastran: the American flutter
method (called the K method in MSC.Nastran), an efficient K method (called the KE
method) for rapid flutter evaluations, and the British flutter method (called the PK method)
for more realistic representation of the unsteady aerodyamic influence as
frequency-dependent stiffness and damping terms. Complex eigenvalue analysis is used
with the K method, and the QR transformation method is used with the KE and PK
methods.

Dynamic Aeroelastic Response

The aeroelastic dynamic response problem determines the response of the aircraft to time-
or frequency-varying excitations. Atmospheric turbulence is the primary example of this
type of excitation, but wind shear and control surface motion can also have an aeroelastic
component. Methods of generalized harmonic (Fourier) analysis are applied to the linear
system to obtain the response to the excitation in the time domain. The gust response
analysis may be regarded either as a stationary random loading or as a discrete gust.
The gust analysis capability computes the response to random atmospheric turbulence,
discrete one-dimensional gust fields, and control surface motion and other dynamic
loading. The random response parameters calculated are the power spectral density, root
mean square response, and mean frequency of zero-crossings. The response to the discrete
gust and control surface motion is calculated by direct and inverse Fourier transform
methods since the oscillatory aerodynamics are known only in the frequency domain. The
time histories of response quantities are the output in the discrete case (see Figure 11-9).
314

3.0E8

Root Bending Moment (in-lbs)

-5.0E8
0 Time (sec) 8
Figure 11-9 Transient Response Resulting from a Gust
 CHAPTER 11 315
Advanced Dynamic Analysis Capabilities

11.13 DMAP
MSC.Nastran provides a series of solution sequences (SOL 103 for normal modes analysis,
for example) written in their own language tailored to matrix manipulation. These
solution sequences consist of a series of DMAP (direct matrix abstraction program)
statements. MSC.Nastran’s DMAP capability enables you to modify these solution
sequences and write your own solution sequences.

DMAP is a high-level language with its own compiler and grammatical rules. DMAP
statements contain data blocks and parameters and operate on them in a specified manner.
For example, the DMAP statement

ADD U1,U2/U3 $
adds matrices U1 and U2 together and calls the output U3. The DMAP statement
MATPRN U3// $
prints the matrix U3.

Numerous DMAP Alters for dynamic analysis are provided (see the NX Nastran Release
Guide for further information). Many of these DMAP Alters are for dynamic analysis.
There are alters for model checkout, modal initial conditions, the addition of static results
to transient response results, frequency-dependent impedance, and modal test-analysis
correlation, among others. The NX Nastran Advanced Dynamic Analysis User’s Guide
describes a DMAP Alter for enforced motion using the Lagrange multiplier approach.
In summary, DMAP is a powerful capability that allows you to modify MSC.Nastran to
satisfy your needs.
316
 NX Nastran Basic Dynamic Analysis User’s Guide

A
APPENDIX
Glossary of Terms

■ Glossary of Terms
318

1.1 Glossary of Terms

CMS Component mode synthesis.
Critical Damping The lowest value of damping for which oscillation does not occur
when the structure is displaced from its rest position. Values of
damping less than the critical damping value create an
underdamped system for which oscillatory motion occurs.
Damping Energy dissipation in vibrating structures.
DOF Degree-of-freedom.
Dynamic Amplification Ratio of dynamic response to static response, which is a function
Factor of the forcing frequency, natural frequency, and damping.
Forced Vibration Vibration response due to applied time-varying forces.
Analysis
Free Vibration AnalysisVibration response when there is no applied force. Normal
modes analysis and transient response to initial conditions are
examples of free vibration analysis.
Frequency Computation of the steady-state response to simple harmonic
Response excitation.
Analysis
GDR Generalized dynamic reduction.
LMT Lagrange multiplier technique.
Mechanism Mode Stress-free, zero-frequency motions of a portion of the structure.
A mechanism can be caused by an internal hinge.
Mode Shape The deformed shape of a structure when vibrating at one of its
natural frequencies.
Natural Frequency The frequency with which a structure vibrates during free
vibration response. Structures have multiple natural frequencies.
Phase Angle The angle between the applied force and the response. In
MSC.Nastran this angle is a phase lead, whereby the response
leads the force.
Positive Definite A matrix whose eigenvalues are all greater than zero.
Positive A matrix whose eigenvalues are greater than or equal to zero.
Semi-Definite
Repeated Roots Two or more identical natural frequencies.
Resonance Large-amplitude vibrations that can grow without bound. At
resonance, energy is added to the system. This occurs, for
example, when the harmonic excitation frequency is equal to one
of the natural frequencies. The response at resonance is
controlled entirely by damping.
 CHAPTER A 319
Glossary of Terms

Rigid-Body Drift Transient displacements of an unconstrained model that grow

continuously with time. This is often caused by the accumulation
of small numerical errors when integrating the equations of
motion.
Rigid-Body Mode Stress-free, zero-frequency motions of the entire structure.
SDOF Single degree-of-freedom.
Structural Damping that is proportional to displacement.
Damping
Transient Computation of the response to general time-varying excitation.
Response
Analysis
Viscous Damping that is proportional to velocity.
Damping
320
 NX Nastran Basic Dynamic Analysis User’s Guide

B
APPENDIX
Nomenclature for Dynamic Analysis

■ Nomenclature for Dynamic Analysis

322

B.1 Nomenclature for Dynamic Analysis

The appendix provides nomenclature for terms commonly used in dynamic analysis.

General
⋅ Multiplication
≈ Approximately
[] Matrix
{} Vector

i –1
g Acceleration of Gravity
∆t Time Step

∆f Frequency Step

i, j, k Subscripts (Indices)

∞ Infinity

u Displacement

u(0) Initial Displacement

u· Velocity

u· ( 0 ) Initial Velocity

u·· Acceleration

m Mass

m0 Large Mass

b Damping

b cr Critical Damping

k Stiffness

p Applied Force

ω Circular Frequency

ωn Circular Natural Frequency

2 Eigenvalue
ω

λ Eigenvalue
 CHAPTER B 323
Nomenclature for Dynamic Analysis

f Frequency

fn Natural Frequency

Tn Period

ωd Damped Circular Natural Frequency

ζ Damping Ratio

Q Quality Factor

θ Phase Angle

∆ Logarithmic Decrement
324

Structural Properties
Geometry

L Length

A Area

Stiffness

E Young’s Modulus

G Shear Modulus

J Torsional Constant

ν Poisson’s Ratio

I Area Moment of Inertia

Mass

ρ Mass Density

ρm Mass Density

ρw Weight Density

Ip Polar Moment of Inertia

Damping

G Overall Structural Damping Coefficient

GE Element Structural Damping Coefficient

 CHAPTER B 325
Nomenclature for Dynamic Analysis

Multiple Degree-of-Freedom System

{u} Displacement Vector

{ u· } Velocity Vector

{ u·· } Acceleration Vector

{φ} Mode Shape Vector

{ φ rig } Rigid-Body Modes

ξi i-th Modal Displacement

· i-th Modal Velocity

ξi

·· i-th Modal Acceleration

ξi

mj j-th Generalized Mass

kj j-th Generalized Stiffness

λk k-th Eigenvalue

[M] Matrix

[ Mr ] Rigid-Body Mass Matrix

[K] Stiffness Matrix

[ KE ] Element Stiffness Matrix

[B] Damping Matrix

[P] Force Matrix

326
 NX Nastran Basic Dynamic Analysis User’s Guidek

C
APPENDIX
The Set Notation System Used in
Dynamic Analysis

■ Overview
328

C.1 Overview
The set notation system used for dynamic analysis in MSC.Nastran continues and expands
the set notation system for static analysis. Because of the great variety of physical
quantities and displacement sets used in dynamic analysis, becoming familiar with the set
notation system in MSC.Nastran aids its efficient use for dynamic analysis.

Displacement Vector Sets

When you create a model in MSC.Nastran, equations are allocated for the purpose of
assembling the necessary structural equations. These equations are rows and columns in
the matrix equations that describe the structural behavior. Six equations are created per
grid point, one equation is created per SPOINT or EPOINT. The model details (elements,
properties, loads, etc.) are used to create the appropriate row and column entries in the
matrices (e.g., stiffness coefficients). Certain data entries (i.e., SPCs, MPCs, ASETs, etc.)
cause matrix operations to be performed in the various stages of the solution process. To
organize the matrix operations, MSC.Nastran assigns each DOF to displacement sets.
Most matrix operations used in a structural analysis involve partitioning, merging, and/or
transforming matrix arrays from one subset to another. All components of motion of a
given type form a vector set which is distinguished from other vector sets. A given
component of motion can belong to many sets.

In MSC.Nastran, there are two basic types of sets:

• Combined sets
• Mutually exclusive sets
Combined sets are formed through the union (i.e., combination) of two or more sets.
Mutually exclusive sets are important in the solution process because if a DOF is a member
of a mutually exclusive set, it cannot be a member of any of the other mutually exclusive
sets. The importance of this property is demonstrated as the sets are described below.
When MSC.Nastran starts to assemble the structural equations, it allocates six equations
for each grid point (GRID) and 1 equation for each scalar point (SPOINT). These equations
are associated with a displacement set defined as the g-set. Fundamentally, the g-set
represents an unconstrained set of structural equations.
The next step in the solution process is to partition the g-set into two subsets: the m-set and
the n-set. The dependent DOFs of all multipoint constraint relations (MPCs, RBEs, etc.)
that define the m-set are condensed into a set of independent DOFs (the n-set). The n-set
represents all of the independent DOFs that remain after the dependent DOFs were
removed from the active set of equations.

Using the n-set, SPCs are applied to the independent equations to further partition the
equations. The degrees-of-freedom defined on SPC entries define the DOFs in the s-set.
When you reduce the n-set by applying the s-set constraints, the f-set remains. The f-set is
the "free" DOF of the structure. At this stage of the solution, the f-set is comprised of the
remaining equations that represent a constrained structure. If the applied constraints are
applied properly, the f-set equations represent a statically stable solution.
 CHAPTER C 329
The Set Notation System Used in Dynamic Analysis

If static condensation is to be performed, the f-set is partitioned into the o-set and the a-set.
The o-set degrees-of-freedom are those that are to be eliminated from the active solution
through a static condensation. The remaining DOFs reside in the a-set. The a-set is termed
the "analysis" set. The a-set is often the partition at which the solution is performed.
The a-set can be divided into physical degrees-of-freedom (the t-set) and generalized
coordinates (the q-set). This partitioning is used in dynamic analysis if generalized
dynamic reduction (GDR) is performed.
If the SUPORT entry is used, the degrees-of-freedom defined on the SUPORT entry are
placed in the r-set. When the r-set is partitioned from the t-set, the -set remains. This final
set is termed the "leftover" set and is the lowest level of partitioning performed in
MSC.Nastran static analysis. The l-set partition is the matrix on which the final solution is
performed. Under special circumstances, the l-set is divided into two types of DOFs. The
DOFs that are held fixed in component mode synthesis (CMS) or GDR are called b-set
points, and those DOFs that are free to move in CMS and GDR are called c-set points.
Table C-1 demonstrates the basic partitioning operations. When a particular set above has
no DOFs associated with it, it is a null set. Its partition is then applied, and the DOFs are
moved to the subsequent partition.
330

Table C-1 Basic Partitioning Operations

g-set – m-set = n-set

n-set – s-set = f-set
f-set – o-set = a-set
a-set – r-set = l-set

In the above table, the m-, s-, o-, and r-sets form the mutually exclusive sets. Physically, the
mutually exclusive set partitioning ensures that operations cannot be performed on DOFs
that are no longer active. For example, if you apply an SPC to a DOF which is a dependent
degree-of-freedom on an RBAR, a FATAL error is issued. Using an SPC entry moves a
DOF to the s-set, but this cannot occur if the same DOF is already a member of the m-set.
Both the m-set and s-set are mutually exclusive.

Table C-1 is a representation of all sets, set partitions, and set combinations used in
MSC.Nastran.
A number of additional, mutually independent sets of physical displacements (namely, q-
set, c-set, b-set and e-set) are used in dynamic analysis to supplement the sets used in static
analysis. The q-, c-, and b-sets facilitate generalized dynamic reduction and component
mode synthesis. The e-set is used to represent control systems and other nonstructural
physical variables. The p-set is created by combining the g-set with the e-set variables.
In addition to the combined sets described above, the v-set is a combined set created by
combining the c-, r- and o-sets. The DOFs in these sets are the DOFs free to vibrate during
component mode synthesis or generalized dynamic reduction.

Some additional sets (sa-, k-, ps-, and pa-sets) represented in Figure C-1 are defined and
used in aeroelastic analysis.

The modal coordinate set ξ is separated into zero frequency modes ξ o and elastic (finite
frequency) modes ξ f . For dynamic analysis by the modal method, the extra points u e are
added to the modal coordinate set ξ i to form the h-set as shown in Figure C-2.
The parameter PARAM,USETPRT can be used to print of lists of degrees-of-freedom and
the sets to which they belong. Different values of the PARAM create various tables in the
printed output.

The supersets formed by the union of other sets have the following definitions.
 CHAPTER C 331
The Set Notation System Used in Dynamic Analysis

Mutually
Exclusive Sets Combined Sets
m
sb
s
sg
o
g
q n p
ne ps
r f pa
a fe
c t d
l
b
e
sa
k

Figure C-1 Combined Sets Formed from Mutually Exclusive Sets

The set names have the definitions described in Table C-2.

Table C-2 Sets in MSC.Nastran

Mutually Exclusive Sets

Set Name Meaning
m Points eliminated by multipoint constraints.
sb* Points eliminated by single-point constraints that are included in
boundary condition changes and by the automatic SPC feature.
sg* Points eliminated by single-point constraints that are listed on
grid point Bulk Data entries.
o Points omitted by structural matrix partitioning.
q Generalized coordinates for dynamic reduction or component
mode synthesis.
r Reference points used to determine free body motion.
c The free boundary set for component mode synthesis or dynamic
reduction.
b Coordinates fixed for component mode analysis or dynamic
reduction.
e Extra degrees-of-freedom introduced in dynamic analysis.
sa Permanently constrained aerodynamic points.
k Aerodynamic points.

* Strictly speaking, sb and sg are not exclusive with respect to one another.
Degrees-of-freedom may exist in both sets simultaneously. These sets are
exclusive, however, from the other mutually exclusive sets.
332

Table C-3 Sets in MSC.Nastran

Combined Sets (+ Indicates Union of Two Sets)

Set Name Meaning
s = sb + sg All points eliminated by single-point constraints.

l = b+c The structural coordinates remaining after the reference

coordinates are removed (points left over).
t = l+r The total set of physical boundary points for superelements.

a = t+q The set assembled in superelement analysis.

d = a+e The set used in dynamic analysis by the direct method.

f = a+o Unconstrained (free) structural points.

fe = f + e Free structural points plus extra points.

n = f+s All structural points not constrained by multipoint

constraints.
ne = n + e All structural points not constrained by multipoint
constraints plus extra points.
g = n+m All structural (grid) points including scalar points.

p = g+e All physical points.

ps = p + sa Physical and constrained aerodynamic points.

pa = ps + k Physical set for aerodynamics.

fr = o + l Statically independent set minus the statically determinate

supports ( o + l = f – q – r ).

v = o+c+r The set free to vibrate in dynamic reduction and component

mode synthesis.
 CHAPTER C 333
The Set Notation System Used in Dynamic Analysis

Physical Set
e

Modal Sets h

ξo
ξi
ξf

where:

ξ o = rigid-body (zero frequency) modal coordinates

ξ f = finite frequency modal coordinates

ξ i = ξ o + ξ f = the set of all modal coordinates

u h = ξ i + u e = the set used in dynamic analysis by the modal method

Figure C-2 Set Notation for Physical and Modal Sets

334
 NX Nastran Basic Dynamic Analysis User’s Guide

D
APPENDIX
Solution Sequences for Dynamic
Analysis

■ Overview
336

D.1 Overview
This appendix lists the solution sequences (SOLs) for the dynamic analysis types described
in this guide. The SOLs are listed for the (preferred) structured solution sequences, rigid
formats, and unstructured superelement solution sequences.

The NX Nastran Quick Reference Guide lists all of the solution sequences for all of the
MSC.Nastran analysis types.

Structured Solution Sequences for Basic Dynamic Analysis

SOL Number SOL Name Description

103 SEMODES Normal Modes
108 SEDFREQ Direct Frequency Response
109 SEDTRAN Direct Transient Response
111 SEMFREQ Modal Frequency Response
112 SEMTRAN Modal Transient Response

Rigid Formats for Basic Dynamic Analysis

SOL Number SOL Name Description

3 MODES Normal Modes
8 DFREQ Direct Frequency Response
9 DTRAN Direct Transient Response
11 MFREQ Modal Frequency Response
12 MTRAN Modal Transient Response
 NX Nastran Basic Dynamic Analysis User’s Guide

E
APPENDIX
Case Control Commands for Dynamic
Analysis

■ Overview
338

E.1 Overview
This appendix lists the Case Control commands that are often used for dynamic analysis.
Commands that apply to statics, such as FORCE and STRESS, are not listed. The dynamic
analysis Case Control commands are listed alphabetically. The description of each
command is similar to that found in the NX Nastran Quick Reference Guide. The descriptions
in this guide have been edited to apply specifically to the dynamic analysis capabilities
described herein.

The NX Nastran Quick Reference Guide describes all of the Case Control commands.
The Case Control commands described in this appendix are summarized as follows:

Input Specification
B2GG Direct Input Damping Matrix Selection
K2GG Direct Input Stiffness Matrix Selection
M2GG Direct Input Mass Matrix Selection

Analysis Specification
BC Boundary Condition Identification
DLOAD Dynamic Load Set Selection
FREQUENCY Frequency Set Selection
IC Transient Initial Condition Set Selection
METHOD Real Eigenvalue Extraction Method Selection
SDAMPING Structural Damping Selection
SUPORT1 Fictitious Support Set Selection
TSTEP Transient Time Step Set Selection for Linear Analysis

Output Specification
ACCELERATION Acceleration Output Request
DISPLACEMENT Displacement Output Request
MODES Subcase Repeater
OFREQUENCY Output Frequency Set
OLOAD Applied Load Output Request
OTIME Output Time Set
SACCELERATION Solution Set Acceleration Output Request
SDISPLACEMENT Solution Set Displacement Output Request
SVECTOR Solution Set Eigenvector Output Request
 CHAPTER E 339
Case Control Commands for Dynamic Analysis

SVELOCITY Solution Set Velocity Output Request

VELOCITY Velocity Output Request
340

<E>l

ACCELERATION Acceleration Output Request

Requests form and type of acceleration vector output.

Format:
 ALL 
SORT1 PRINT, PUNCH REAL or IMAG  
ACCELERATION ( , , ) =  n 
SORT2 PLOT PHASE  
 NONE 

Examples:
ACCELERATION = 5
ACCELERATION(SORT2, PHASE) = ALL
ACCELERATION(SORT1, PRINT, PUNCH, PHASE) = 17

Describers Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point.
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
PLOT Computes, but does not print or punch, acceleration output.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Accelerations at all points will be output.
NONE Accelerations at no points will be output.
n Set identification of a previously appearing SET command. Only
accelerations of points with identification numbers that appear on
this SET command will be output. (Integer > 0)
 CHAPTER <E> 341
Case Control Commands for Dynamic Analysis

Remarks:
1. Both PRINT and PUNCH may be requested.
2. Acceleration output is only available for transient and frequency response
problems.
3. See Remark 2 under DISPLACEMENT for a discussion of SORT1 and SORT2.
4. ACCELERATION = NONE allows overriding an overall output request.
5. The PLOT option is used when curve plots are desired in the magnitude/phase
representation and no printer request is present for magnitude/phase
representation.
6. Acceleration results are output in the global coordinate system (see field CD on
the GRID Bulk Data entry).
342

B2GG Direct Input Damping Matrix Selection

Selects a direct input damping matrix.

Format:
B2GG = name

Example:
B2GG = BDMIG

Describer Meaning
name Name of [ B 2gg ] matrix that is input on the DMIG Bulk Data entry.
See “Matrix Assembly Operations in SubDMAP SEMG” on
page 307 of the NX Nastran Reference Manual. (Character)

Remarks:
1. DMIG matrices will not be used unless selected.
2. Terms are added to the damping matrix before any constraints are applied.
3. The matrix must be symmetric and field 4 on the DMIG,name entry must contain
the integer 6.
4. A scale factor may be applied to this input via the PARAM,CB2 entry. See the
“Parameter Descriptions” in Chapter 5 of the NX Nastran Quick Reference Guide.
 CHAPTER <E> 343
Case Control Commands for Dynamic Analysis

BC Boundary Condition Identification

Identifies multiple boundary conditions for normal modes, buckling, and flutter analysis
in SOLs 103, 105, 145, and 200.

Format:
BC = n

Example:
BC = 23

Describer Meaning
n Identification number. (Integer > 0)

Remarks:
1. In SOLs 103, 105, 145, and 200 BC is required in each subcase if multiple boundary
conditions are specified for normal modes, buckling, and flutter analysis.
2. If only one boundary condition is specified, then BC does not have to be specified
and n defaults to zero.
344

DISPLACEMENT Displacement Output Request

Requests the form and type of displacement or pressure vector output. Note that
PRESSURE and VECTOR are equivalent commands.

Format:
 ALL 
SORT1 PRINT, PUNCH REAL or IMAG  
DISPLACEMENT ( , , ) =  n 
SORT2 PLOT PHASE  
 NONE 

Examples:
DISPLACEMENT = 5
DISPLACEMENT(REAL) = ALL
DISPLACEMENT(SORT2, PUNCH, REAL) = ALL

Describer Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point.
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
PLOT Generates, but does not print, displacement data.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Displacements for all points will be output.
NONE Displacement for no points will be output.
n Set identification of a previously appearing SET command. Only
displacements of points with identification numbers that appear
on this SET command will be output. (Integer > 0)

Remarks:
1. Both PRINT and PUNCH may be requested.
2. The defaults for SORT1 and SORT2 depend on the type of analysis:
a. SORT1 is the default in static analysis, requency reponse, steady state heat
transfer analysis, real and complex eigenvalue analysis, flutter analysis, and
buckling analysis.
 CHAPTER <E> 345
Case Control Commands for Dynamic Analysis

b. SORT2 is the default in transient response analysis (structural and heat

transfer). SORT2 is not available for real eigenvalue (including buckling)
complex eigenvalue, or flutter output. If SORT1 is selected for one or more of
the commands ACCE, DISP, ENTH, FORC, HDOT, MPCF, OLOA, SPCF,
STRA, STRE, and VELO, then the remaining commands will also be output in
SORT1 format.
3. VECTOR and PRESSURE are alternate forms and are entirely equivalent to
DISPLACEMENT.
4. DISPLACEMENT = NONE overrides an overall output request.
5. The PLOT option is used when curve plots are desired in the magnitude/phase
representation and no printer output request is present for magnitude/phase
representation.
6. The units or translation are the same as the units of length of the model. Rotations
are in units of radians.
7. Displacement results are output in the global coordinate system (see field CD on
the GRID Bulk Data entry).
346

DLOAD Dynamic Load Set Selection

Selects a dynamic load or an acoustic source to be applied in a transient or frequency

response problem.

Format:
DLOAD = n

Example:
DLOAD = 73

Describer Meaning
n Set identification of a DLOAD, RLOAD1, RLOAD2, TLOAD1,
TLOAD2, or ACSRCE Bulk Data entry. (Integer > 0)

Remarks:
1. RLOAD1 and RLOAD2 may only be selected in a frequency response problem.
2. TLOAD1 and TLOAD2 may only be selected in a transient response problem.
3. Either a RLOADi or TLOADi entry (but not both) must be selected in an
aeroelastic response problem. If RLOADi is selected, a frequency response is
calculated. If TLOADi is selected, then transient response is computed by Fourier
transform. When there are only gust loads (GUST entry), the DLOAD selects a
TLOADi or RLOADi entry with zero load along with field 3 of the GUST
command.
 CHAPTER <E> 347
Case Control Commands for Dynamic Analysis

FREQUENCY Frequency Set Selection

Selects the set of forcing frequencies to be solved in frequency response problems.

Format:
FREQUENCY = n

Example:
FREQUENCY = 17

Describer Meaning
n Set identification number of FREQ, FREQ1, and FREQ2 Bulk Data
entries. (Integer > 0)

Remarks:
1. A frequency set selection is required for a frequency response problem.
2. A frequency set selection is required for transient response by Fourier methods
(SOLution 146).
3. All FREQ, FREQ1 and FREQ2 entries with the same frequency set identification
numbers will be used. Duplicate frequencies will be ignored. f N and f N – 1 are
considered duplicated if

–5
f N – f N – 1 < 10 f MAX – f M IN
348

IC Transient Initial Condition Set Selection

Selects the initial conditions for direct transiet analysis (SOLs 9, 109, 129, and 159)

Format:
IC = n

Example:
IC = 17

Describer Meaning
n Set identification of a TIC Bulk Data entry for structural analysis
or TEMP and TEMPD entries for transient heat transfer analysis
(SOL 159). (Integer > 0)

Remarks:
1. For structural analysis, TIC entries will not be used (therefore, no intial
conditions) unless selected in the Case Control Section.
2. Initial conditions are not allowed in a modal transient analysis (SOLs 12 and 112).
 CHAPTER <E> 349
Case Control Commands for Dynamic Analysis

K2GG Direct Input Stiffness Matrix Selection

Selects a direct input stiffness matrix.

Format:
K2GG = name

Example:
K2GG = KDMIG

Describers Meaning
name Name of a [ K 2gg ] matrix that is input on the DMIG Bulk Data
entry. See the MSC/NASTRAN Reference Manual,
Section 9.4.2. (Character)

Remarks:
1. DMIG matrices will not be used unless selected.
2. Terms are added to the stiffness matrix before any constraints are applied.
3. The matrix must be symmetric and field 4 on DMIG,name entry must contain the
integer 6.
4. A scale factor may be applied to this input via the PARAM,CK2 entry. See the
“Parameter Descriptions” in Chapter 5 of the NX Nastran Quick Reference Guide.
350

M2GG Direct Input Mass Matrix Selection

Selects a direct input mass matrix.

Format:
M2GG = name

Example:
M2GG = MDMIG

Describers Meaning
name Name of a [ M 2gg ] matrix that is input on the DMIG Bulk Data
entry. See the MSC/NASTRAN Reference Manual,
Section 9.4.2. (Character)

Remarks:
1. DMIG matrices will not be used unless selected.
2. Terms are added to the mass matrix before any constraints are applied.
3. The matrix must be symmetric and field 4 on DMIG,name entry must contain the
integer 6.
4. M2GG input is not affected by PARAM,WTMASS. M2GG input must either be
in consistent mass units or PARAM,CM2 may be used.
 CHAPTER <E> 351
Case Control Commands for Dynamic Analysis

METHOD Real Eigenvalue Extraction Method Selection

Selects the real eigenvalue extraction parameters.

Format:
STRUCTURE
METHOD ( ) = n
FLUID

Examples:
METHOD = 33
METHOD(FLUID) = 34

Describer Meaning
STRUCTURE or The referenced EIGR or EIGRL Bulk Data entry is applied to the
FLLUID structural or fluid portion of the model. (Default = STRUCTURE)
n Set identification number of an EIGR or EIGRL Bulk Data entry for
normal modes or modal formulation, or an EIGB or EIGRL entry
for buckling. (Integer > 0)

Remarks:
1. An eigenvalue extraction method must be selected when extracting real
eigenvalues using DMAP modules READ or REIGL.
2. If the set identification number selected is present on both EIGRL and EIGR
and/or EIGB entries, the EIGRL entry will be used. This entry requests the
Lanczos eigensolution method.
3. METHOD(FLUID) permits a different request of EIGR or EIGRL for the fluid
portion of the model from the structural portion of the model in coupled
fluid-structural analysis. See “Additional Topics” on page 439 of the NX Nastran
Reference Manual.
a. If not specified, then the METHOD selection of the structure will be used for
the fluid and modal reduction will not be performed on the fluid portion of
the model in the dynamic solution sequences.
b. The METHOD(FLUID) and METHOD(STRUCTURE) may be specified
simultaneously in the same subcase for the residual structure only. Do not
specify METHOD(FLUID) in a superelement subcase even if the
superelement contains fluid elements.
c. The auto-omit feature (see “Solution Sequences” on page 273 of the NX
Nastran Reference Manual) is not recommended. Therefore, only those
methods of eigenvalue extraction that can process a singular mass matrix
should be used; e.g., EIGRL entry or MGIV and MHOU on the EIGR entry.
352

MODES Subcase Repeater

Repeats a subcase.

Format:
MODES = n

Example:
MODES = 3

Describer Meaning
n Number of times the subcase is to be repeated. (Integer > 0)

Remarks:
1. This command can be illustrated by an example. Suppose stress output is desired
for the first five modes only and displacements for the next two modes and forces
for the remaining modes. The following example would accomplish this:
SUBCASE 1 $ FOR MODES 1 THRU 5
MODES = 5
STRESS = ALL
SUBCASE 6 $ FOR MODES 6 AND 7
DISPLACEMENTS = ALL
MODES = 2
SUBCASE 8 $ FOR MODES 8 AND REMAINING MODES
FORCE = ALL
2. This command causes the results for each mode to be considered as a separate,
successively numbered subcase, beginning with the subcase number containing
the MODES command. In the example above, this means that subcases 1, 2, 3, etc.,
are assigned to modes 1, 2, 3, etc., respectively.
3. If this command is not used, eigenvalue results are considered to be a part of a
single subcase. Therefore, any output requests for the single subcase will apply
for all eigenvalues.
4. All eigenvectors with mode numbers greater than the number of subcases
defined in Case Control are printed with the descriptors of the last subcase. For
example, to suppress all printout for modes beyond the first three, the following
Case Control could be used:
SUBCASE 1
MODES = 3
DISPLACEMENTS = ALL
SUBCASE 4
DISPLACEMENTS = NONE
BEGIN BULK
 CHAPTER <E> 353
Case Control Commands for Dynamic Analysis

OFREQUENCY Output Frequency Set

Selects a set of frequencies for output requests.

Format:

 ALL 
OFREQUENCY =  
 n 

Examples:
OFREQUENCY = ALL
OFREQUENCY = 15

Describers Meaning
ALL Output for all frequencies will be computed.
n Set identification of a previously appearing SET command.
Output for frequencies closest to those given on this SET
command will be output. (Integer > 0)

Remarks:
1. If this command is not specified in the Case Control Section, then output for all
frequencies will be computed.
2. The number of solutions selected will always be equal to the number of quantities
in the selected set. The closest values are used.
3. In flutter analysis (SOL 145), the selected set refers to the imaginary part of the
complex eigenvalues. The physical interpretation of this quantity depends on the
method of flutter analysis as follows:
a. K- or KE-method: Velocity (input units).
b. PK-method: Frequency.
c. In aeroelastic response analysis (SOL 146) with RLOAD selection, the selected
set refers to the frequency (cycles per unit time).
d. In complex eigenvalue analysis (SOLs 7, 10, 107, and 110), the selected set
refers to the imaginary part of the complex eigenvalues.
4. In the modal solution sequences, if this command is specified in more than one
subcase, then it is recommended that the first subcase contain OFREQ = ALL and
subsequent subases contain OFREQ = n. Also, data recovery requests should be
specified only in the subsequent subases. For example:
354

SUBCASE 1
OFREQ = ALL $ 0.0 through 0.5
SUBCASE 2
SET 10 = 0.0 0.1 0.3
OFREQ = 10
DISP = ALL
SUBCASE 3
SET 20 = 0.4 0.5
OFREQ = 20
STRESS = ALL
 CHAPTER <E> 355
Case Control Commands for Dynamic Analysis

OLOAD Applied Load Output Request

Requests the form and type of applied vector output.

Format:
 ALL 
SORT1 PRINT, PUNCH REAL or IMAG  
OLOAD ( , , ) =  n 
SORT2 PLOT PHASE  
 NONE 

Examples:
OLOAD = ALL
OLOAD(SORT, PHASE) = 5

Describers Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point.
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Applied loads for all points will be output. See Remark 2 and
Remark 8.
NONE Applied load for no points will be output.
n Set identification of a previously appearing SET command. Only
loads on points with identification numbers that appear on this
SET command will be output. (Integer > 0)
356

Remarks:
1. Both PRINT and PUNCH may be requested.
2. See Remark 2 under DISPLACEMENT for a discussion of SORT1 and SORT2. In
the SORT1 format, only nonzero values will be output.
3. In a statics problem, a request for SORT2 causes loads at all requested points (zero
and nonzero) to be output.
4. OLOAD = NONE overrides an overall output request.
5. In the statics superelement solution sequences, and in the dynamics SOLs 107
through 112, 118, 145, 146, and 200, OLOADs are available for superelements and
the residual structure only externally applied loads are printed, and not loads
transmitted from upstream superelements. Transmitted loads can be obtained
with GPFORCE requests.
In the nonlinear transient analysis solution sequences SOLS 129 and 159,
OLOADs are available only for residual structure points and include loads
transmitted by upstream superelements.
6. In nonlinear analysis, OLOAD output will not reflect changes due to follower
forces.
7. Loads generated via the SPCD Bulk Data entry do not appear in OLOAD output.
8. In SORT1 format, OLOADs recovered at consecutively numbered scalar points
are printed in groups of six (sextets) per line of output. But if a scalar point is not
consecutively numbered, then it will begin a new sextet on a new line of output.
If a sextet can be formed and it is zero, then the line will not be printed. If a sextet
cannot be formed, then zero values may be output.
9. OLOAD results are output in the global coordinate system (see field CD on the
GRID Bulk Data entry).
10. In inertia relief analysis the OLOAD output is interpreted differently for SOLs 1,
101, and 200:
a. In SOL 1, the output shows only the applied loads.
b. In SOLs 101 and 200, the output includes both the inertia loads and applied
loads.
 CHAPTER <E> 357
Case Control Commands for Dynamic Analysis

OTIME Output Time Set

Selects a set of times for output requests.

Format:

 ALL 
OTIME =  
 n 

Examples:
OTIME = ALL
OTIME = 15

Describers Meaning
ALL Output for all times will be computed.
n Set identification number of a previously appearing SET
command. Output for times closest to those given on this SET
command will be computed. (Integer > 0)

Remarks:
1. If the OTIME command is not supplied in the Case Control Section, then output
for all times will be computed.
2. This command is particularly useful for requesting a subset of the output (e.g.,
stresses at only peak times, etc.).
3. This command can be used in conjunction with the MODACC module to limit the
times for which modal acceleration computations are performed.
4. If this command is specified in more than one subcase in the modal solution
sequences, then it is recommended that the first subcase contain OTIME = ALL
and subsequent subcases contain OTIME = n. Also, data recovery requests
should be specified only in the subsequent subcases. For example:
SUBCASE 1
OTIME = ALL
SUBCASE 2
OTIME = 10
SET 10 = . . .
DISP = ALL
SUBCASE 3
OTIME = 20
SET 20 = . . .
STRESS = ALL
358

SACCELERATION Solution Set Acceleration Output Request

Requests form and type of solution set acceleration output.

Format:
 ALL 
SORT1 REAL or IMAG  
SACCELERATION ( , PRINT, PUNCH, ) =  n 
SORT2 PHASE  
 NONE 

Examples:
SACCELERATION = ALL
SACCELERATION(PUNCH, IMAG) = 142

Describers Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point (or mode number).
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Acceleration for all solution set points (modes) will be output.
NONE Acceleration for no solution set points (modes) will be output.
n Set identification number of a previously appearing SET
command. Only accelerations of points with identification
numbers that appear on this SET command will be output.
(Integer > 0)

Remarks:
1. Acceleration output is only available for transient and frequency response
problems.
2. The defaults for SORT1 and SORT2 depend on the type of analysis and is
discussed in Remark 2 under the DISPLACEMENT command. If SORT1 is
selected for any of the commands SACC, SDIS, and SVEL then the remaining
commands will also be SORT1.
3. SACCELERATION = NONE allows an overall output request to be overridden.
 CHAPTER <E> 359
Case Control Commands for Dynamic Analysis

SDAMPING Structural Damping Selection

Requests modal damping as a function of natural frequency in modal solutions or

viscoelastic materials as a function of frequency in direct frequency response analysis.

Format:
STRUCTURE
SDAMPING ( ) = n
FLUID

Example:
SDAMPING = 77

Describer Meaning
STRUCTURE or Modal damping is requested for the structural or fluid portion of
FLUID the model.
n Set identification number of a TABDMP1 or TABLEDi Bulk Data
entry. (Integer > 0)

Remarks:
1. In the modal solutions (e.g., SOLs 10, 11, 12, 110, 111, 112, 145, 146, and 200),
SDAMPING must reference a TABDMP1 entry.
2. In direct frequency response (e.g., SOL 108), SDAMPING must reference a
TABLEDi entry which defines viscoelastic (frequency-dependent) material
properties. See “Viscoelastic Materials in Frequency Response Analysis” on
page 667 of the NX Nastran Reference Manual.
3. In frequency response analysis, changes in SDAMPING commands between
subcases must be accompanied by a new FREQUENCY command. The new
FREQi entry may be a duplicate of the previous entry except that the SID must be
different.
360

SDISPLACEMENT Solution Set Displacement Output Request

Requests the form and type of solution set displacement output.

Format:
 ALL 
SORT1 REAL or IMAG  
SDISPLACEMENT ( , PRINT, PUNCH, ) =  n 
SORT2 PHASE  
 NONE 

Examples:
SDISPLACEMENT = ALL
SDISPLACEMENT(SORT2, PUNCH, PHASE) = NONE

Describer Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point (or mode number).
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Displacements for all solution set points (modes) will be output.
NONE Displacements for no solution set points (modes) will be output.
n Set identification number of a previously appearing SET
command. Only displacements on points with identification
numbers that appear on this SET command will be output.
(Integer > 0)

Remarks:
1. The defaults for SORT1 and SORT2 depend on the type of analysis and is
discussed in Remark 2 under the DISPLACEMENT command. If SORT1 is
selected for any of the commands SACC, SDIS, and SVEL then the remaining
commands will also be SORT1.
2. SDISPLACEMENT = NONE allows an overall output request to be overridden.
3. The SDISPLACEMENT command is required to output normalized complex
eigenvectors.
 CHAPTER <E> 361
Case Control Commands for Dynamic Analysis

SUPORT1 Fictitious Support Set Selection

Selects the fictitious support set (SUPORT1 entries only) to be applied to the model.

Format:
SUPORT1 = n

Examples:
SUPORT1 = 15
SUPO = 4

Describer Meaning
n Set identification of fictitious support set defined on the SUPORT1
Bulk Data entry. (Integer > 0)

Remarks:
1. SUPORT1 may be used in SOLs 101, 103, and 200 only.
2. SUPORT1 entries will not be used unless selected in Case Control by the
SUPORT1 command.
3. SUPORT entries will be applied in all subcases.
362

SVECTOR Solution Set Eigenvector Output Request

Requests the form and type of solution set eigenvector output.

Format:

 ALL 
 
SVECTOR [(PRINT, PUNCH)] =  n 
 
 NONE 

Examples:
SVECTOR = ALL
SVECTOR(PUNCH) = NONE

Describers Meaning
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
ALL Displacements for all points (modes) will be output.
NONE Displacements for no points (modes) will be output.
n Set identification of a previously appearing SET command. Only
displacements of points with identification numbers that appear
on this SET command will be output. (Integer > 0)

Remarks:
1. Both PRINT and PUNCH may be requested.
2. SVECTOR = NONE overrides an overall output request.
3. Output will be presented as a tabular listing of grid points for each eigenvector.
 CHAPTER <E> 363
Case Control Commands for Dynamic Analysis

SVELOCITY Solution Set Velocity Output Request

Requests the form and type of solution set velocity output.

Format:

 ALL 
SORT1 REAL or IMAG  
SVELOCITY ( , PRINT, PUNCH, ) =  n 
SORT2 PHASE  
 NONE 

Examples:
SVELOCITY = 5
SVELOCITY(SORT2, PUNCH, PRINT, PHASE) = ALL

Describer Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point (or mode number).
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Velocity for all solution points (modes) will be output.
NONE Velocity for no solution points (modes) will be output.
n Set identification of a previously appearing SET command. Only
velocities of points with identification numbers that appear on this
SET command will be output. (Integer > 0)

Remarks:
1. Both PRINT and PUNCH may be requested.
2. Velocity output is only available for transient and frequency response problems.
3. The defaults for SORT1 and SORT2 depend on the type of analysis and is
discussed in Remark 2 under the DISPLACEMENT command. If SORT1 is
selected for any of the commands SACC, SDIS, and SVEL then the remaining
commands will also be SORT1.
4. SVELOCITY = NONE overrides an overall output request.
364

TSTEP Transient Time Step Set Selection

Selects integration and output time steps for linear or nonlinear transient problems.

Format:
TSTEP = n

Example:
TSTEP = 731

Describer Meaning
n Set identification number of a TSTEP or TSTEPNL Bulk Data
entry. (Integer > 0)

Remarks:
1. A TSTEP entry must be selected to execute a transient analysis (SOLs 9, 12, 109,
or 112) and TSTEPNL for a nonlinear transient analysis (SOLs 129 and 159).
2. A TSTEPNL entry must be selected in each subcase to execute a nonlinear
transient problem.
3. For the application of time-dependent loads in modal frequency response
analysis (SOLs 111 and 146), or TSTEP entry must be selected by the TSTEP
command. The time-dependent loads will be recomputed in frequency domain
by a Fourier Transform.
 CHAPTER <E> 365
Case Control Commands for Dynamic Analysis

VELOCITY Velocity Output Request

Requests the form and type of velocity vector output.

Format:
 ALL 
SORT1 PRINT, PUNCH REAL or IMAG  
VELOCITY ( , , ) =  n 
SORT2 PLOT PHASE  
 NONE 

Examples:
VELOCITY = 5
VELOCITY(SORT2, PHASE, PUNCH) = ALL

Describer Meaning
SORT1 Output will be presented as a tabular listing of grid points for each
load, frequency, eigenvalue, or time, depending on the solution
sequence.
SORT2 Output will be presented as a tabular listing of frequency or time
for each grid point.
PRINT The printer will be the output medium.
PUNCH The punch file will be the output medium.
PLOT Generates, but does not print, velocities.
REAL or IMAG Requests rectangular format (real and imaginary) of complex
output. Use of either REAL or IMAG yields the same output.
PHASE Requests polar format (magnitude and phase) of complex output.
Phase output is in degrees.
ALL Velocity for all solution points will be output.
NONE Velocity for no solution points will be output.
n Set identification of a previously appearing SET command. Only
velocities of points with identification numbers that appear on this
SET command will be output. (Integer > 0)
366

Remarks:
1. Both PRINT and PUNCH may be requested.
2. Velocity output is only available for transient and frequency response problems.
3. See Remark 2 under DISPLACEMENT for a discussion of SORT1 and SORT2.
4. VELOCITY = NONE overrides an overall output request.
5. The PLOT option is used when curve plots are desired in the magnitude/phase
representation and no printer request is present for magnitude/phase
representation.
6. Velocity results are output in the global coordinate system (see field CD on the
GRID Bulk Data entry).
 NX Nastran Basic Dynamic Analysis User’s Guide

F
APPENDIX
Bulk Data Entries for Dynamic
Analysis

■ Overview
368

F.1 Overview
This appendix lists the Bulk Data entries that are often used for dynamic analysis. Entries
that apply to generic modeling or statics, such as GRID or FORCE, are not listed. Bulk Data
entries are listed alphabetically. The description of each entry is similar to that found in
the NX Nastran Quick Reference Guide. The descriptions in this guide have been edited to
apply specifically to the dynamic analysis capabilities described herein.

The NX Nastran Quick Reference Guide describes all of the Bulk Data entries.

The Bulk Data entries described in this appendix are summarized as follows:

Mass Properties

CMASS1 Scalar Mass Connection

CMASS2 Scalar Mass Property and Connection
CMASS3 Scalar Mass Connection to Scalar Points Only
CMASS4 Scalar Mass Property and Connection to Scalar Points Only
CONM1 Concentrated Mass Element Connection, General Form
CONM2 Concentrated Mass Element Connection, Rigid Body Form
PMASS Scalar Mass Property

Damping Properties

CDAMP1 Scalar Damper Connection

CDAMP2 Scalar Damper Property and Connection
CDAMP3 Scalar Damper Connection to Scalar Points Only
CDAMP4 Scalar Damper Property and Connection to Scalar Points Only
CVISC Viscous Damper Connection
PDAMP Scalar Damper Property
PVISC Viscous Element Property
TABDMP1 Modal Damping Table

Normal Modes

EIGR Real Eigenvalue Extraction Data

EIGRL Real Eigenvalue Extraction Data, Lanczos Method

Dynamic Loading

DAREA Dynamic Load Scale Factor

DELAY Dynamic Load Time Delay
DLOAD Dynamic Load Combination or Superposition
 CHAPTER F 369
Bulk Data Entries for Dynamic Analysis

DPHASE Dynamic Load Phase Lead

LSEQ Static Load Set Definition
TABLED1 Dynamic Load Tabular Function, Form 1
TABLED2 Dynamic Load Tabular Function, Form 2
TABLED3 Dynamic Load Tabular Function, Form 3
TABLED4 Dynamic Load Tabular Function, Form 4

Frequency Response

FREQ Frequency List

FREQ1 Frequency List, Alternate Form 1
FREQ2 Frequency List, Alternate Form 2
FREQ3 Frequency List, Alternate Form 3
FREQ4 Frequency List, Alternate Form 4
FREQ5 Frequency List, Alternate Form 5
RLOAD1 Frequency Response Dynamic Load, Form 1
RLOAD2 Frequency Response Dynamic Load, Form 2

Transient Response

TIC Transient Initial Condition

TLOAD1 Transient Response Dynamic Load, Form 1
TLOAD2 Transient Response Dynamic Load, Form 2
TSTEP Transient Time Step

Miscellaneous

DMIG Direct Matrix Input at Points

SUPORT Fictitious Support
SUPORT1 Fictitious Support, Alternate Form
370

CDAMP1 Scalar Damper Connection

Defines a scalar damper element.

Format:
1 2 3 4 5 6 7 8 9 10
CDAMP1 EID PID G1 C1 G2 C2

Example:

CDAMP1 19 6 0 23 2

Field Contents
EID Unique element identification number. (Integer > 0)
PID Property identification number of a PDAMP property entry.
(Integer > 0; Default = EID)
G1, G2 Geometric grid point identification number. ( Integer ≥ 0 )
C1, C2 Component number. ( 0 ≤ Integer ≤ 6 ; 0 or up to six unique integers,
1 through 6 may be specified in the field with no embedded blanks.
0 applies to scalar points and 1 through 6 applies to grid points).

Remarks:
1. Scalar points may be used for G1 and/or G2, in which case the corresponding C1
and/or C2 must be zero or blank. Zero or blank may be used to indicate a
grounded terminal G1 or G2 with a corresponding blank or zero C1 or C2. A
grounded terminal is a point with a displacement that is constrained to zero.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. The two connection points (G1, C1) and (G2, C2), must be distinct.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. When CDAMP1 is used in heat transfer analysis, it generates a lumped heat
capacity.
6. A scalar point specified on this entry need not be defined on an SPOINT entry.
7. If Gi refers to a grid point then Ci refers to degrees of freedom(s) in the
displacement coordinate system specified by CD on the GRID entry.
 CHAPTER F 371
Bulk Data Entries for Dynamic Analysis

CDAMP2 Scalar Damper Property and Connection

Defines a scalar damper element without reference to a material or property entry.

Format:
1 2 3 4 5 6 7 8 9 10
CDAMP2 EID B G1 C1 G2 C2

Example:

CDAMP2 16 2.98 32 1

Field Contents
EID Unique element identification number. (Integer > 0)
B Value of the scalar damper. (Real)
G1, G2 Geometric grid point identification number. ( Integer ≥ 0 )
C1, C2 Component number. ( 0 ≤ Integer ≤ 6 ; 0 or up to six unique integers,
1 through 6 may be specified in the field with no embedded blanks.
0 applies to scalar points and 1 through 6 applies to grid points).

Remarks:
1. Scalar points may be used for G1 and/or G2, in which case the corresponding C1
and/or C2 must be zero or blank. Zero or blank may be used to indicate a
grounded terminal G1 or G2 with a corresponding blank or zero C1 or C2. A
grounded terminal is a point with a displacement that is constrained to zero.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. The two connection points (G1, C1) and (G2, C2), must be distinct.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. When CDAMP2 is used in heat transfer analysis, it generates a lumped heat
capacity.
6. A scalar point specified on this entry need not be defined on an SPOINT entry.
7. If Gi refers to a grid point then Ci refers to degrees of freedom(s) in the
displacement coordinate system specified by CD on the GRID entry.
372

CDAMP3 Scalar Damper Connection to Scalar Points Only

Defines a scalar damper element that is connected only to scalar points.

Format:
1 2 3 4 5 6 7 8 9 10
CDAMP3 EID PID S1 S2

Example:

CDAMP3 16 978 24 36

Field Contents
EID Unique element identification number. (Integer > 0)
PID Property identification number of a PDAMP entry. (Integer > 0;
Default = EID)
S1, S2 Scalar point identification numbers. ( Integer ≥ 0; S1 ≠ S2 )

Remarks:
1. S1 or S2 may be blank or zero, indicating a constrained coordinate.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. Only one scalar damper element may be defined on a single entry.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. When CDAMP3 is used in heat transfer analysis, it generates a lumped heat
capacity.
6. A scalar point specified on this entry need not be defined on an SPOINT entry.
 CHAPTER F 373
Bulk Data Entries for Dynamic Analysis

CDAMP4 Scalar Damper Property and Connection to Scalar Points Only

Defines a scalar damper element that is connected only to scalar points and without
reference to a material or property entry.

Format:
1 2 3 4 5 6 7 8 9 10
CDAMP4 EID B S1 S2

Example:

CDAMP4 16 –2.6 4 9

Field Contents
EID Unique element identification number. (Integer > 0)
B Scalar damper value. (Real)
S1, S2 Scalar point identification numbers. ( Integer ≥ 0; S1 ≠ S2 )

Remarks:
1. S1 or S2 may be blank or zero, indicating a constrained coordinate.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. Only one scalar damper element may be defined on a single entry.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. If this entry is used in heat transfer analysis, it generates a lumped heat capacity.
6. A scalar point specified on this entry need not be defined on an SPOINT entry.
374

CMASS1 Scalar Mass Connection

Defines a scalar mass element.

Format:
1 2 3 4 5 6 7 8 9 10
CMASS1 EID PID G1 C1 G2 C2

Example:

CMASS1 32 6 2 1

Field Contents
EID Unique element identification number. (Integer > 0)
PID Property identification number of a PMASS entry. (Integer > 0;
Default = EID)
G1, G2 Geometric grid or scalar point identification number. ( Integer ≥ 0 )
C1, C2 Component number. ( 0 ≤ Integer ≤ 6 ; 0 or up to six unique integers,
1 through 6 may be specified in the field with no embedded blanks.
0 applies to scalar points and 1 through 6 applies to grid points).

Remarks:
1. Scalar points may be used for G1 and/or G2, in which case the corresponding C1
and/or C2 must be zero or blank. Zero or blank may be used to indicate a
grounded terminal G1 or G2 with a corresponding blank or zero C1 or C2. A
grounded terminal is a point with a displacement that is constrained to zero. If
only scalar points and/or ground are involved, it is more efficient to use the
CMASS3 entry.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. The two connection points (G1, C1) and (G2, C2) must be coincident.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. A scalar point specified on this entry need not be defined on an SPOINT entry.
6. If Gi refers to a grid point then Ci refers to degrees of freedom(s) in the
displacement coordinate system specified by CD on the GRID entry.
 CHAPTER F 375
Bulk Data Entries for Dynamic Analysis

CMASS2 Scalar Mass Property and Connection

Defines a scalar mass element without reference to a property entry.

Format:
1 2 3 4 5 6 7 8 9 10
CMASS2 EID M G1 C1 G2 C2

Example:

CMASS2 32 9.25 6 1

Field Contents
EID Unique element identification number. (Integer > 0)
M Value of the scalar mass. (Real)
G1, G2 Geometric grid or scalar point identification number. ( Integer ≥ 0 )
C1, C2 Component number. ( 0 ≤ Integer ≤ 6 ; 0 or up to six unique integers,
1 through 6 may be specified in the field with no embedded blanks.
0 applies to scalar points and 1 through 6 applies to grid points).

Remarks:
1. Scalar points may be used for G1 and/or G2, in which case the corresponding C1
and/or C2 must be zero or blank. Zero or blank may be used to indicate a
grounded terminal G1 or G2 with a corresponding blank or zero C1 or C2. A
grounded terminal is a point with a displacement that is constrained to zero. If
only scalar points and/or ground are involved, it is more efficient to use the
CMASS4 entry.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. The two connection points (G1, C1) and (G2, C2), must be distinct. Except in
unusual circumstances, one of them will be a grounded terminal with blank
entries for Gi and Ci.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. A scalar point specified on this entry need not be defined on an SPOINT entry.
6. If Gi refers to a grid point then Ci refers to degrees of freedom(s) in the
displacement coordinate system specified by CD on the GRID entry.
376

CMASS3 Scalar Mass Connection to Scalar Points Only

Defines a scalar mass element that is connected only to scalar points.

Format:
1 2 3 4 5 6 7 8 9 10
CMASS3 EID PID S1 S2

Example:

CMASS3 13 42 62

Field Contents
EID Unique element identification number. (Integer > 0)
PID Property identification number of a PMASS entry. (Integer > 0;
Default = EID)
S1, S2 Scalar point identification numbers. ( Integer ≥ 0; S1 ≠ S2 )

Remarks:
1. S1 or S2 may be blank or zero, indicating a constrained coordinate.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. Only one scalar mass element may be defined on a single entry.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. A scalar point specified on this entry need not be defined on an SPOINT entry.
 CHAPTER F 377
Bulk Data Entries for Dynamic Analysis

CMASS4 Scalar Mass Property and Connection to Scalar Points Only

Defines a scalar mass element that is connected only to scalar points, without reference to
a property entry.

Format:
1 2 3 4 5 6 7 8 9 10
CMASS4 EID M S1 S2

Example:

CMASS4 23 14.92 23

Field Contents
EID Unique element identification number. (Integer > 0)
M Scalar mass value. (Real)
S1, S2 Scalar point identification numbers. ( Integer ≥ 0; S1 ≠ S2 )

Remarks:
1. S1 or S2 may be blank or zero, indicating a constrainted coordinate. This is the
usual case.
2. Element identification numbers must be unique with respect to all other element
identification numbers.
3. Only one scalar mass element may be defined on a single entry.
4. For a discussion of the scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
5. A scalar point specified on this entry need not be defined on an SPOINT entry.
378

CONM1 Concentrated Mass Element Connection, General Form

Defines a 6 × 6 symmetric mass matrix at a geometric grid point.

Format:
1 2 3 4 5 6 7 8 9 10
CONM1 EID G CID M11 M21 M22 M31 M32

M33 M41 M42 M43 M44 M51 M52 M53

M54 M55 M61 M62 M63 M64 M65 M66

Example:

CONM1 2 22 2 2.9 6.3

4.8 28.6

28.6 28.6

Field Contents
EID Unique element identification number. (Integer > 0)
G Grid point identification number. (Integer > 0)
CID Coordinate system identifcation number for the mass matrix.
( Integer ≥ 0 )
Mij Mass matrix values. (Real)

Remarks:
1. For aless general means of defining concentrated mass at grid points, see the
CONM2 entry description.
2. Element identification numbers must be unique with respect to all other element
identification number.
 CHAPTER F 379
Bulk Data Entries for Dynamic Analysis

CONM2 Concentrated Mass Element Connection, Rigid Body Form

Defines a concentrated mass at a grid point.

Format:
1 2 3 4 5 6 7 8 9 10
CONM2 EID G CID M X1 X2 X3

I11 I21 I22 I31 I32 I33

Example:

CONM2 2 15 6 49.7

16.2 16.2 7.8

Field Contents
EID Element identification number. (Integer > 0)
G Grid point identification number. (Integer > 0)
CID Coordinate system identifcation number for CID of –1; see X1, X2, X3
below. ( Integer ≥ – 1 ; Default = 0)
M Mass value. (Real)
X1, X2, X3 Offset distances from the grid point to the center of gravity of the
mass in the coordinate system defined in field 4, unless CID = –1, in
which case X1, X2, X3 are the coordinates, not offsets, of the center of
gravity of the mass in the basic coordinate system. (Real)
Iij Mass moments of inertia measured at the mass center of gravity in
the coordinate system defined by field 4. If CID = –1, the basic
coordinate system is implied. (For I11, I22, and I33; Real ≥ 0.0 ; for
I21, I31, and I32; Real)

Remarks:
1. Element identification numbers must be unique with respect to all other element
identification numbers.
2. For a more general means of defining concentrated mass at grid points, see the
CONM1 entry description.
3. The continuation is optional.
4. If CID = –1, offests are internally computed as the difference between the grid
point location and X1, X2, X3. The grid point locations may be defined in a
nonbasic coordinate system. In this case, the values of Iij must be in a coordinate
system that parallels the basic coordinate system.
5. The form of the intertia matrix about its center of gravity is taken as:
380

M
M symmetric
M
– I11
– I21 I22
– I31 – I32 I33

where:

M = ∫ ρ dV

I11 = ∫ ρ ( x 22 + x23 ) dV

I22 = ∫ ρ ( x 21 + x23 ) dV

I33 = ∫ ρ ( x 21 + x22 ) dV

I21 = ∫ ρx 1 x2 dV

I31 = ∫ ρx 1 x3 dV

I32 = ∫ ρx 2 x3 dV

and x 1 , x 2 , x 3 are components of distance from the center of gravity in the

coordinate system defined in field 4. The negative signs for the off-diagonal
terms are supplied automatically. A warning message is issued if the inertia
matrix is nonpositive definite, since this may cause fatal errors in dydnamic
analysys modules.
6. If CID ≥ 0 , the X1, X2, and X3 are defined by a local Cartesian system, even if CID
references a spherical or cylindrical coordinate system. This is similar to the
manner in which displacement coordinate systems are defined.
7. See “Grid Point and Coordinate System Definition” on page 35 of the NX
Nastran Reference Manual for a definition of coordinate system terminology.
 CHAPTER F 381
Bulk Data Entries for Dynamic Analysis

CVISC Viscous Damper Connection

Defines a viscous damper element.

Format:
1 2 3 4 5 6 7 8 9 10
CVISC EID PID G1 G2

Example:

CVISC 21 6327 29 31

Field Contents
EID Element identification number. (Integer > 0)
PID Property identification number of a PVISC entry. (Integer > 0;
Default = EID)
G1, G2 Grid point identification numbers of connection points.
(Integer > 0; G1 ≠ G2 )

Remarks:
1. Element identification numbers must be unique with respect to all other element
identification numbers.
2. Only one viscous damper element may be defined on a single entry.
382

DAREA Dynamic Load Scale Factor

Defines scale (area) factors for dynamic loads. DAREA is used in conjunction with
RLOADi and TLOADi entries.

Format:
1 2 3 4 5 6 7 8 9 10
DAREA SID P1 C1 A1 P2 C2 A2

Example:

DAREA 3 6 2 8.2 15 1 10.1

Field Contents
SID Identification number. (Integer > 0)
Pi Grid, extra, or scalar point identification number. (Integer > 0)
Ci Component number. (Integer 1 through 6 for grid point; blank or 0
for extra or scalar point)
Ai Scale (area) factor. (Real)

Remarks:
1. One or two scale factors may be defined on a single entry.
2. Refer to RLOAD1, RLOAD2, TLOAD1 or TLOAD2 entries for the formulas that
define the scale factor Ai.
3. Component numbers refer to the displacement coordinate system.
4. DAREA entries may be used with LSEQ Bulk Data entries. The LSEQ and static
load entries will be used to internally generate DAREA entries.
5. If DAREA is referenced by a GUST entry, Pi must be defined. However, it is only
used if selected through a DLOAD Case Control command. WG from the GUST
entry is used instead of Ai when requested via a GUST entry.
6. In superelement analysis, if Pi is a grid point, then it must be interior to the
residual structure. For points interior to superelements, the LSEQ entry must be
used in conjunction with a static load entry.
 CHAPTER F 383
Bulk Data Entries for Dynamic Analysis

DELAY Dynamic Load Time Delay

Defines the time delay term τ in the equations of the dynamic loading function.

Format:
1 2 3 4 5 6 7 8 9 10
DELAY SID P1 C1 T1 P2 C2 T2

Example:

DELAY 5 21 6 4.25 7 6 8.1

Field Contents
SID Identification number of the DELAY entry. (Integer > 0)
Pi Grid, extra, or scalar point identification number. (Integer > 0)
Ci Component number. (Integer 1 through 6 for grid point, blank or 0
for extra point or scalar point)
Ti Time delay τ for designated point Pi and component Ci. (Real)

Remarks:
1. One or two dynamic load time delays may be defined on a single entry.
2. SID must also be specified on a RLOAD1, RLOAD2, TLOAD1, or TLOAD2 entry.
See those entry descriptions for the formulas that define the manner in which the
time delay τ is used.
3. A DAREA and/or LSEQ entry should be used to define a load at Pi and Ci.
4. In superelement analysis, DELAY entries may only be applied to loads on points
in the residual structure.
384

DLOAD Dynamic Load Combination or Superposition

Defines a dynamic loading condition for frequency response or transient response

problems as a linear combination of load sets defined via RLOAD1 or RLOAD2 entries for
frequency response or TLOAD1 or TLOAD2 entries for transient response.

Format:
1 2 3 4 5 6 7 8 9 10
DLOAD SID S S1 L1 S2 L2 S3 L3

S4 L4 -etc.-

Examples:

DLOAD 17 1.0 2.0 6 –2.0 7 2.0 8

–2.0 9

Field Contents
SID Load set identification number. (Integer > 0)
S Scale factor. (Real)
Si Scale factors. (Real)
Li Load set identification numbers of RLOAD1, RLOAD2, TLOAD1,
and TLOAD2 entries. (Integer > 0)

Remarks:
1. Dynamic load sets must be selected in the Case Control Section with
DLOAD = SID.
2. The load vector being defined by this entry is given by

{ P } = S ∑ Si { P i }
i

3. Each Li must be unique from any other Li on the same entry.

4. SID must be unique from all TLOADi and RLOADi entries.
5. Nonlinear transient load sets (NOLINi entries) may not be specified on DLOAD
entries. NOLINi entries are selected separately in the Case Control Section by the
NONLINEAR command.
6. A DLOAD entry my not reference a set identification number defined by another
DLOAD entry.
7. TLOAD1 and TLOAD2 loads may be combined only through the use of the
DLOAD entry.
8. RLOAD1 and RLOAD2 loads may be combined only through the use of the
DLOAD entry.
 CHAPTER F 385
Bulk Data Entries for Dynamic Analysis

DMIG Direct Matrix Input at Points

Defines direct input matrices related to grid, extra, and/or scalar points. The matrix is
defined by a single header entry and one or more column entries. Only one header entry
is required. A column entry is required for each column with nonzero elements

Header Entry Format:

1 2 3 4 5 6 7 8 9 10
DMIG NAME “0” IFO TIN TOUT POLAR NCOL

Column Entry Format:

DMIG NAME GJ CJ G1 C1 A1 B1

G2 C2 A2 B2 -etc.-

Example:

DMIG STIF 0 1 3 4

DMIG STIF 27 1 2 3 3.+5 3.+3

2 4 2.5+10 0. 50 1.0 0.

Field Contents
NAME Name of the matrix. See Remark 1. (One to eight alphanumeric
characters, the first of which is alphabetic.)
IFO Form of matrix input. IFO = 6 must be specified for matrices selected
by the K2GG, M2GG, and B2GG Case Control commands. (Integer)
1 = Square
9 or 2 = Rectangular
6 = Symmetric
TIN Type of matrix being input. (Integer)
1 = Real, single precision (one field is used per element).
2 = Real, double precision (one field is used per element).
3 = Complex, single precision (two fields are used per
element).
4 = Complex, double precision (two fields are used per
element).
TOUT Type of matrix that will be created. (Integer)
0 = Set by precision system cell. (Default)
1 = Real, single precision
2 = Real, double precision
386

Field Contents
3 = Complex, single precision
4 = Complex double precision
POLAR Input format of Ai, Bi. (Integer = blank or 0 indicates real, imaginary
format; Integer > 0 indicates amplitude, phase format)
NCOL Number of columns in a rectangular matrix. Used only for IFO = 9.
(Integer > 0)
GJ Grid, scalar or extra point identification number for column index.
Integer > 0)
CJ Component number for grid point GJ. ( 0 < Integer ≤ 6 ; blank or
zero if GJ is a scalar or extra point.)
Gi Grid scalar, or extra point identification number for row index.
(Integer > 0)
Ci Component number for Gi for a grid point. ( 0 < CJ ≤ 6 ; blank or
zero if Gi is a scalar extra point.)
Ai, Bi Real and imaginary (or amplitude and phase) parts of a matrix
element. If the matrix is real (TIN = 1 or 2), then Bi must be blank.
(Real)

Remarks:
1. Matrices defined on this entry may be used in dynamics by selection in the Case
Control with K2PP = NAME, B2PP = NAME, M2PP = NAME for [ K pp ] , [ B pp ] , or
[ M pp ] , respectively. Matrices may also be selected for all solution sequences by
K2GG = NAME, B2GG = NAME, and M2GG = NAME. The g-set matrices are
added to the structural matrices before constraints are applied, while p-set
matrices are added in dynamics after constraints are applied. Load matrices may
be selected by P2G = NAME for dynamic and superelement analyses.
2. The header entry containing IFO, TIN and TOUT is required. Each non-null
column is started witha GJ, CJ pair. The entries for each row of that column
follows. Only nonzero terms need be entered. The terms may be input in
arbitrary order. A GJ, CJ pair may be entered more than once, but input of an
element of the matrix more than once will produce a fatal message.
3. Field 3 of the header entry must contain an integer 0.
4. For symmetric matrices (IFO = 6), a given off-diagonal element may be input
either below or above the diagonal. While upper and lower triangle terms may
be mixed, a fatal message will be issued if an element is input both below and
above the diagonal.
5. The recommended format for rectangular matrices requires the use of NCOL and
IFO = 9. The number of columns in the matrix is NCOL. (The number of rows in
all DMIG matrices is always either p-set or g-set size, depending on the context.)
The GJ term is used for the column index. The CJ term is ignored.
 CHAPTER F 387
Bulk Data Entries for Dynamic Analysis

6. If NCOL is not used for rectangular matrices, two different conventions are
available:
a. If IFO = 9, GJ and CJ will determine the sorted sequence, but will otherwise
be ignored; a rectangular matrix will be generated with the columns
submitted being in the 1 to N positions, where N is the number of logical
entries submitted (not counting the header entry).
b. If IFO = 2, the number of columns of the rectangular matrix will be equal to
the index of the highest numbered non-null column (in internal sort). Trailing
null columns of the g- or p-size matrix will be truncated.
7. The matrix names must be unique among all DMIGs.
8. TIN should be set consistent with the number of decimal digits required to read
the input data adequately. For a single-precision specification on a short-word
machine, the input will be truncated after about eight decimal digits, even when
more digits are present in a double-field format. If more digits are needed, a
double precision specification should be used instead. However, note that a
double precision specification requires a “D” type exponent even for terms that
do not need an exponent. For example, unity may be input as 1.0 in single
precision, but the longer form 1.0D0 is required for double precision.
9. On long-word machines, almost all matrix calculations are performed in single
precision and on short-word machines, in double precision. It is recommended
that DMIG matrices also follow these conventions for a balance of efficiency and
reliability. The recommended value for TOUT is 0, which instructs the program
to inspect the system cell that measures the machine precision at run time and sets
the precision of the matrix to the same value. TOUT = 0 allows the same DMIG
input to be used on any machine. If TOUT is contrary to the machine type
specified (for example, a TOUT of 1 on a short-word machine), unreliable results
may occur.
10. If any DMIG entry is changed or added on restart then a complete re-analysis is
performed. Therefore, DMIG entry changes or additions are not recommended
on restart.
388

DPHASE Dynamic Load Phase Lead

Defines the phase lead term θ in the equation of the dynamic loading function.

Format:
1 2 3 4 5 6 7 8 9 10
DPHASE SID P1 C1 TH1 P2 C2 TH2

Example:

DPHASE 4 21 6 2.1 8 6 7.2

Field Contents
SID Identification number of DPHASE entry. (Integer > 0)
Pi Grid, extra, or scalar point identification number. (Integer > 0)
Ci Component number. (Integers 1 through 6 for grid points; zero or
blank for extra or scalar points)
THi Phase lead θ in degrees. (Real)

Remarks:
1. One or two dynamic load phase lead terms may be defined on a single entry.
2. SID must be referenced on a RLOADi entry. Refer to the RLOAD1 or RLOAD2
entry for the formulas that define how the phase lead θ is used.
3. A DAREA and/or LSEQ entry should be used to define a load at Pi and Ci.
4. In superelement analysis, DPHASE entries may only be applied to loads on
points in the residual structure.
 CHAPTER F 389
Bulk Data Entries for Dynamic Analysis

EIGR Real Eigenvalue Extraction Data

Defines data needed to perform real eigenvalue analysis.

Format:
1 2 3 4 5 6 7 8 9 10
EIGR SID METHOD F1 F2 NE ND

NORM G C

Example:

EIGR 13 SINV 1.9 15.6 12

POINT 32 4

Field Contents
SID Set identification number. (Unique Integer > 0)
METHOD Method of eigenvalue extraction. (Character)
INV Inverse Power method.
SINV Inverse Power method with enhancements.
GIV Givens method of tridiagonalization.
MGIV Modified Givens method.
HOU Householder method of tridiagonalization.
MHOU Modified Householder method.
AGIV Automatic selection of METHOD = “GIV” or
“MGIV”. See Remark 15.
AHOU Automatic selection of METHOD = “HOU” or
“MHOU”. See Remark 15.
390

Table F-1 Relationship Between METHOD Field and Other Fields

METHOD Field
Field
INV or SINV GIV, MGIV, HOU or MHOU
F1, F2 Frequency range of interest. F1 Frequency range of interest. If ND is
must be input. If METHOD = not blank, F1 and F2 are ignored. If
“SINV” and ND is blank, then F2 ND is blank, eigenvectors are found
must be input. ( Real ≥ 0.0 ) with natural frequencies that lie in
the range between F1 and F2.
( Real ≥ 0.0 ; F1 < F2)
NE Estimate of number of roots in Not used.
range (Required for METHOD =
“INV”). Not used by “SINV”
method. (Integer > 0)
ND Desired number of roots. If this Desired number of eigenvectors. If
field is blank and METHOD = ND is zero, the number of
“SINV”, then all roots between F1 eigenvectors is determined from F1
and F2 are searched and the limit is and F2. If all three are blank, then
600 roots. (Integer > 0, Default is ND is automatically set to one more
3 < NE for METHOD = “INV” only) than the number of degrees of
freedom listed on SUPORT entries.
( Integer ≥ 0 ; Default = 0)

NORM Method for normalizing eigenvectors. (Character: “MASS”, “MAX”,

or “POINT”; Default = “MASS”)
MASS Normalize to unit value of the generalized mass.
(Default)
MAX Normalize to unit value of the largest component in
the analysis set.
POINT Normalize to a positive or negative unit value of the
component defined in fields 3 and 4 (defaults to
“MAX” if defined component is zero).
G Grid or scalar point identification number. Required only if NORM
= “POINT”. (Integer > 0)
C Component number. Required only if NORM = “POINT” and G is a
geometric grid point. ( 1 ≤ Integer ≤ 6 )

Remarks:
1. The EIGR entry must be selected with the Case Control command
METHOD = SID.
2. See Chapter 3 of the NX Nastran Basic Dynamic Analysis User’s Guide, for a
discussion of method selection.
 CHAPTER F 391
Bulk Data Entries for Dynamic Analysis

3. The units of F1 and F2 are cycles per unit time.

4. The continuation entry is optional. If the continuation entry is not specified, then
mass normalization is performed.
5. If METHOD = “GIVE or “MGIV”, all eigenvalues are found.
6. If METHOD = “GIVE”, the mass matrix for the analysis set (a-set) must be
positive definite. The auto-omit feature removes massless degrees of freedom.
Singularities or near-singularities of the mechanism type in the mass matrix will
produce poor numerical stability for the “GIV” method. The “MGIV” method
should be used for this condition.
7. METHOD = “MGIV” is a modified form of the “GIV” method that allows a non-
positive definite mass matrix for the analysis set (a-set) (i.e., massless degrees of
freedom may exist in the analysis set). The “MGIV” method should give
improved accuracy for the lowest frequency solutions.
8. If NORM = “MAX”, components that are not in the analysis set may have values
larger than unity.
9. If NORM = “POINT”, the selected component should be in the analysis set (a-set).
(The program uses NORM = “MAX” when it is not in the analysis set.) The
displacement value at the selected component will be positive or negative unity.
10. The desired number of roots (ND) includes all roots previously found, such as
rigid body modes determined with the use of SUPORTi entries, or the number of
roots found on the previous run when restarting and appending the eigenvector
file.
11. The “SINV” method is an enhanced version of the “INV” method. It uses Sturm
sequence number techniques to make it more likely that all roots in the range have
been found. It is generally more reliable and more efficient than the “INV”
method.
12. For the “INV” and “SINV” methods, convergence is achieved at 10 –6 .
Convergence is not tested for other methods.
13. For the “SINV” method only, if F2 is blank, the first shift will be made at F1, and
only one eigensolution above F1 will be calculated. If there are no modes below
F1, it is likely that the first mode will be calculated. If there are modes below F1
(including rigid body modes defined by SUPORT entries), a mode higher than the
first mode above F1 may be calculated.
14. The “HOU” and “MHOU” methods are generally faster on machines with vector
processors.
15. If METHOD = “AGIV” or “AHOU”, the program automatically determines the
need for a modified method (“MGIV” or “MHOU”) and makes the proper
selection.
392

EIGRL Real Eigenvalue Extraction Data, Lanczos Method

Defines data needed to perform real eigenvalue (vibration or buckling) analysis with the
Lanczos Method.

Format:
1 2 3 4 5 6 7 8 9 10
EIGRL SID V1 V2 ND MSGLVL MAXSET SHFSCL NORM

option_1 = value_1 option_2 = value_2, etc.

Example:

EIGRL 1 0.1 3.2 10

NORM = MAX NUMS = 2

Field Contents
SID Set Identification number. (Unique integer > 0)
V1, V2 For vibration analysis: frequency range of interest. for buckling
analysis: eigenvalue range of interest. See Remark 4. (Real or blank,
16 16
– 5. × 10 ≤ V1 < V2 ≤ 5. × 10 )
ND Number of roots desired. See Remark 4. (Integer > 0 or blank)
MSGLVL Diagnostic level. ( 0 ≤ Integer ≤ 4 ; Default = 0)
MAXSET Number of vectors in block or set ( 1 ≤ Integer ≤ 15 ; Default = 7)
SHFSCL Estimate of the first flexible mode natural frequency. See Remark 10.
(Real or blank)
NORM Method for normalizing eigenvectors. (Character: “MASS”, “MAX”)
MASS Normalize to unit value of the generalized mass.
Not available for buckling analysis. (Default for
normal modes analysis.)
MAX Normalize to unit value of the largest displacement
in the analysis set. Displacements not in the
analysis set may be larger than unity. (Default for
buckling analysis.)
ALPH Specifies a constant for the calculation of frequencies (Fi) at the
upper boundary segments for the parallel method based on the
following formula. See Remark 14. (Integer > 0.0; Default = 1.0)

i
1. – ALPH
Fi = ( V2 – V1 ) --------------------------------------------
NUMS
1. – ALPH
 CHAPTER F 393
Bulk Data Entries for Dynamic Analysis

Field Contents
NUMS Number of frequency segments for the parallel method.
(Integer > 0; Default = 1)
Fi Frequency at the upper boundary of the i-th segment. See
Remark 14. (Real or blank; V1 < F1 < F2 < ... F15 < V2)
option_i = Assignes a value to the fields above except for SID. ALPH, NUMS,
value_i and Fi must be specified in this format. V1, V2, ND, MSGLVL,
MAXSET, SHFSCL, and NORM may be specified in this format as
long as their corresponding field is blank in the parent entry.

Remarks:
1. Real eigenvalue extraction data sets must be selected with the Case Control
command METHOD = SID.
2. The units of V1 and V2 are cycles per unit time in vibration analysis, and are
eigenvalues in buckling analysis. Each eigenvalue is the factor by which the
prebuckling state of stress is multiplied to produce buckling in the shape defined
by the corresponding eigenvector.
3. NORM = “MASS” is ignored in buckling analysis and NORM = “MAX” will be
applied.
4. The roots are found in order of increasing magnitude; that is, those closest to zero
are found first. The number and type of roots to be found can be determined from
the following table.

Table F-2 Number and Type of Roots Found with EIGRL Entry

V1 V2 ND Number and Type of Roots Found

V1 V2 ND Lowest ND or all in range, whichever is smaller.
V1 V2 blank All in range
V1 blank ND Lowest ND in range [V1, + ∞ ]
V1 blank blank Lowest root in range [V1, + ∞ ]
blank blank ND Lowest ND roots in [ – ∞ , + ∞ ]
blank blank blank Lowest root. See Remark 11.
blank V2 ND Lowest ND roots below V2
blank V2 blank All below V2

5. In vibration analysis, if V1 < 0.0, the negative eigenvalue range will be searched.
(Eigenvalues are proportional to Vi squared; therefore, the negative sign would
be lost.) This is a means for diagnosing improbable models. In buckling analysis,
negative V1 and/or V2 require no special logic.
394

6. Eigenvalues are sorted on order of magnitude for output. An eigenvector is found

for each eigenvalue.
7. MSGLVL controls the amount of diagnostic output during the eigenvalue
extraction process. The default value of zero suppresses all diagnostic output. A
value of one prints eigenvalues accepted at each shift. Higher values result in
increasing levels of diagnostic output.
8. MAXSET is used to limit the maximum block size. It is otherwise set by the region
size or by ND with a maximum size of 15. It may also be reset if there is
insufficient memory available. The default value is recommended.
9. In vibration analysis, if V1 is blank, all roots less than zero are calculated. Small
negative roots are usually computational zeroes which indicate rigid body
modes. Finite negative roots are an indication of modeling problems. If V1 is set
to zero, negative eigenvalues are not calculated.
10. A specification for SHFSCL may improve performance, especially when large
mass techniques are used in enforced motion analysis. Large mass techniques can
cause a large gap between the rigid body and flexible frequencies. If this field is
blank, a value for SHFSCL is estimated automatically.
11. On occasion, it may be necessary to compute more roots than requested to ensure
that all roots in the range have been found. In this case, all roots and eigenvectors
that pass the convergence checks are output and are usually equal to or less than
an integer multiple of MAXSET.
12. NASTRAN MEMSAVE = 1 reduces scratch space usage by 67%. But CPU cost
may increase.
13. NASTRAN SYSTEM(146) provides options for I/O in sparse method only:

Table F-3 SYSTEM(146) Options

SYSTEM(146) Description
2 Increase memory reserved for sparse method by
approximately 100%
3 Increase memory reserved for sparse method by
approximately 400%
4 Increase memory reserved for sparse method by
approximately 500%

14. For the parallel method, the frequency range between V1 and V2 may be
subdivided into segments that can then be analyzed in parallel. NUMS must be
specified greater than 1 to take advantage of the parallel method. NUMS may
also be specified on the NUMSEG keyword of the NASTRAN statement. If both
are specified, then NUMS takes precedence.
 CHAPTER F 395
Bulk Data Entries for Dynamic Analysis

The upper frequencies of each segment may be generated automatically by ALPH

or specified directly in Fi. If both are specified, then Fi takes precedence over
ALPH as long as they are consistent. ALPH if multiplied by 100 may also be
specified on FRQSEQ keyword of the NASTRAN statement.
396

FREQ Frequency List

Defines a set of frequencies to be used in the solution of frequency response problems.

Format:
1 2 3 4 5 6 7 8 9 10
FREQ SID F1 F2 F3 F4 F5 F6 F7

F8 F9 F10 -etc.-

Example:

FREQ 3 2.98 3.05 17.9 21.3 25.6 28.8 31.2

29.2 22.4 19.3

Field Contents
SID Set identification number. (Integer > 0)
Fi Frequency value in units of cycles per unit time. ( Real ≥ 0.0 )

Remarks:
1. Frequency sets must be selected with the Case Control command
FREQUENCY = SID.
2. All FREQi entries with the same frequency set identification numbers will be
used. Duplicate frequencies will be ignored. f N and f N – 1 are considered
duplicated if f N – f N – 1 < DFREQ ⋅ f MAX – f MIN , where DFREQ is a user parameter
with a default of 10 –5 . f MAX and f MIN are the maximum and minimum excitation
frequencies of the combined FREQi entries.
 CHAPTER F 397
Bulk Data Entries for Dynamic Analysis

FREQ1 Frequency List, Alternate Form 1

Defines a set of frequencies to be used in the solution of frequency response problems by

specification of a starting frequency, frequency increment, and the number of increments
desired.

Format:
1 2 3 4 5 6 7 8 9 10
FREQ1 SID F1 DF NDF

Example:

FREQ1 6 2.9 0.5 13

Field Contents
SID Set identification number. (Integer > 0)
F1 First frequency in set. ( Real ≥ 0.0 )
DF Frequency increment. (Real > 0.0)
NDF Number of frequency increments. (Integer > 0; Default = 1)

Remarks:
1. FREQ1 entries must be selected with the Case Control command FREQUENCY =
SID.
2. The units for F1 and DF are cycles per unit time.
3. The frequencies defined by this entry are given by

f i = F1 + DF ⋅ ( i – 1 )

where i = 1 to (NDF + 1).

4. All FREQi entries with the same frequency set identification numbers will be
used. Duplicate frequencies will be ignored. f N and f N – 1 are considered
duplicated if f N – f N – 1 < DFREQ ⋅ f MAX – f MIN , where DFREQ is a user parameter
with a default of 10 –5 . f MAX and f MIN are the maximum and minimum excitation
frequencies of the combined FREQi entries.
398

FREQ2 Frequency List, Aternate Form 2

Defines a set of frequencies to be used in the solution of frequency response problems by

specification of a starting frequency, final frequency, and the number of logarithmic
increments desired.

Format:
1 2 3 4 5 6 7 8 9 10
FREQ2 SID F1 F2 NF

Example:

FREQ2 6 1.0 8.0 6

Field Contents
SID Set identification number. (Integer > 0)
F1 First frequency. (Real > 0.0)
F2 Last frequency. (Real > 0.0, F2 > F1)
NF Number of logarithmic intervals. (Integer > 0; Default = 1)

Remarks:
1. FREQ2 entries must be selected with the Case Control command FREQUENCY =
SID.
2. The units for F1 and F2 are cycles per unit time.
3. The frequencies defined by this entry are given by

( i – 1 )d
f i = F1 ⋅ e
1 F2
where d = -------
- ln ------- , and i = 1, 2, …, ( NF + 1 )
NF F1
In the example above, the list of frequencies will be 1.0, 1.4142, 2.0, 2.8284, 4.0,
5.6569 and 8.0 cycles per unit time.
4. All FREQi entries with the same frequency set identification numbers will be
used. Duplicate frequencies will be ignored. f N and f N – 1 are considered
duplicated if f N – f N – 1 < DFREQ ⋅ f MAX – f MIN , where DFREQ is a user parameter
with a default of 10 –5 . f MAX and f MIN are the maximum and minimum excitation
frequencies of the combined FREQi entries.
 CHAPTER F 399
Bulk Data Entries for Dynamic Analysis

FREQ3 Frequency List, Aternate Form 3

Defines a set of excitation frequencies for modal frequency-response solutions by

specifying number of excitation frequencies between two modal frequencies.

Format:
1 2 3 4 5 6 7 8 9 10
FREQ3 SID F1 F2 TYPE NEF CLUSTER

Example:

FREQ3 6 20.0 200.0 LINEAR 10 2.0

Field Contents
SID Set identification number. (Integer > 0)
F1 Lower bound of modal frequency range in cycles per unit time.
(Real > 0.0)
F2 Upper bound of modal frequency range in cycles per unit time.
(Real > 0.0; F2 ≥ F1 ; Default = F1)
TYPE LINEAR or LOG. Specifies linear or logarithmic interpolation
between frequencies. (Character; Default = “LINEAR”)
NEF Number of excitation frequencies within each subrange including the
end points. The first subrange is between F1 and the first modal
frequency within the bounds. The second subrange is between first
and second modal frequencies between the bounds. The last
subrange is between the last modal frequency within the bounds and
F2. (Integer > 1; Default = 10)
CLUSTER Specifies clustering of the excitation frequency near the end points of
the range. See Remark 6. (Real > 0.0; Default = 1.0)

Remarks:
1. FREQ3 applies only to modal frequency-response solutions (SOLs 11, 111, 146,
and 200) and is ignored in direct frequency response solutions.
2. FREQ3 entries must be selected with the Case Control command FREQUENCY =
SID.
3. In the example above, there will be 10 frequencies in the interval between each set
of modes within the bounds 20 and 2000, plus 10 frequencies between 20 and the
lowest mode in the range, plus 10 frequencies between the highest mode in the
range and 2000.
4. Since the forcing frequencies are near strucutral resonances, it is important that
some amount of damping be specified.
400

5. All FREQi entries with the same set identification numbers will be used.
Duplicate frequencies will be ignored. f N and f N – 1 are considered duplicated if

f N – f N – 1 < DFREQ ⋅ f MAX – f M IN

where DFREQ is a user parameter, with a default of 10 –5 . f MAX and f MIN are the
maximum and minimum excitation frequencies of the combined FREQi entries.
6. CLUSTER is used to obtain better resolution near the modal frequencies where
the response varies the most. CLUSTER > 1.0 provides closer spacing of
excitation frequency towards the ends of the frequency range, while values of less
than 1.0 provide closer spacing towards the center of the frequency range. For
example, if the frequency range is between 10 and 20, NEF = 11, TYPE =
“LINEAR”; then, the excitation frequencies for various values of CLUSTER
would be as shown in Table F-4.

1 1 1 ⁄ CLUSTER
f̂ k = --- ( f̂ 1 + f̂ 2 ) + --- ( f̂ 2 – f̂ 1 ) ξ ⋅ SIGN ( ξ )
2 2
where

ξ = – 1 + 2 ( k – 1 ) ⁄ ( NEF – 1 ) is a parametric coordinate between –1 and 1

k = varies from 1 to NEF (k = 1, 2, ..., NEF)

f̂ 1 = is the lower limit of the frequency subrange

f̂ 2 = is the upper limit of the subrange

f̂ k = is the k-th excitation frequency

f̂ = is the frequency, or the logarithm of the frequency, depending on the

value specified for TYPE

Table F-4 CLUSTER Usage Example

CLUSTER
Excitation
Frequency ξ c = 0.25 c = 0.50 c = 1.0 c = 2.0 c = 4.0
Number
Excitation Frequencies in Hertz
1 –1.0 10.00 10.0 10.0 10.00 10.00
2 –0.8 12.95 11.8 11.0 10.53 10.27
3 –0.6 14.35 13.2 12.0 11.13 10.60
4 –0.4 14.87 14.2 13.0 11.84 11.02
5 –0.2 14.99 14.8 14.0 12.76 11.66
6 0.0 15.00 15.0 15.0 15.00 15.00
7 0.2 15.01 15.2 16.0 17.24 18.34
8 0.4 15.13 15.8 17.0 18.16 18.98
 CHAPTER F 401
Bulk Data Entries for Dynamic Analysis

Table F-4 CLUSTER Usage Example

CLUSTER
Excitation
Frequency ξ c = 0.25 c = 0.50 c = 1.0 c = 2.0 c = 4.0
Number
Excitation Frequencies in Hertz
9 0.6 15.65 16.8 18.0 18.87 19.40
10 0.8 17.05 18.2 19.0 19.47 19.73
11 1.0 20.00 20.0 20.0 20.00 20.00

7. In design optimization (SOL 200), the excitation frequencies generated from this
entry are derived from the natural frequencies computed in the first design cycle
and the excitation frequencies remain fixed through all subsequent design cycles.
In other words, the excitation frequencies will not be readjusted even if the
natural frequencies change.
402

FREQ4 Frequency List, Aternate Form 4

Defines a set of frequencies used in the solution of modal frequency-response problems by

specifying the amount of “spread” around each natural frequency and the number of
equally spaced excitation frequencies within the spread.

Format:
1 2 3 4 5 6 7 8 9 10
FREQ4 SID F1 F2 FSPD NFM

Example:

FREQ4 6 20.0 200.0 0.30 21

Field Contents
SID Set identification number. (Integer > 0)
F1 Lower bound of frequency range in cycles per unit time.
( Real ≥ 0.0 ; Default = 1.0E20)
F2 Upper bound of frequency range in cycles per unit time.
(Real > 0.0; F2 > F1; Default = 0.10)
FSPD Frequency spread, +/– the fractional amount specified for each mode
which occurs in the frequency range F1 to F2. (1.0 > Real > 0.0,
Default = 0.10)
NFM Number of evenly spaced frequencies per “spread” mode.
(Integer > 0; Default = 3. If NFM is even, NFM + 1 will be used.)

Remarks:
1. FREQ4 applies only to modal frequency-response solutions (SOLs 11, 111, 146,
and 200) and is ignored in direct frequency-response solutions.
2. FREQ4 entries must be selected with the Case Control command FREQUENCY =
SID.
3. There will be NFM excitation frequencies between ( 1 – FSPD ) ⋅ f N and ( 1 + FSPD ) ⋅ f N ,
for each natural frequency in the range F1 to F2.
 CHAPTER F 403
Bulk Data Entries for Dynamic Analysis

4. In the example above there will be 21 equally spaced frequencies across a

frequency band of 0.7 ⋅ f N to 1.3 ⋅ f N for each natural frequency that occurs between
20 and 2000. See Figure F-1 for the definition of frequency spread.

( 1 – FSPD ) ⋅ f N fN ( 1 + FSPD ) ⋅ f N

Figure F-1 Frequency Spread Definition

Excitation frequencies may be based on natural frequencies that are not within the
range (F1 and F2) as long as the calculated excitation frequencies are within the
range. Similarly, an excitation frequency calculated based on natural frequencies
within the range (F1 through F2) may be excluded if it falls outside the range.
5. The frequency spread can be used also to define the half-power bandwidth. The
half-power bandwidth is given by 2 ⋅ ξ ⋅ f N , where ξ is the damping ratio.
Therefore, if FSPD is specified equal to the damping ratio for the mode, NFM
specifies the number of excitation frequency within the half-power bandwidth.
See Figure F-2 for the definition of half-power bandwidth.
Peak Response

2 ⋅ ξ ⋅ fN

Half-Power Point (.707 Peak)

Amplitude

fN Frequency
Half-Power
Bandwidth

Figure F-2 Half-Power Bandwidth Definition

6. Since the forcing frequencies are near structural resonances, it is important that
some amount of damping be specified.
404

7. All FREQi entries with the same set identification numbers will be used.
Duplicate frequencies will be ignored. f N and f N – 1 are considered duplicated if

f N – f N – 1 < DFREQ ⋅ f M AX – f M IN

where DFREQ is a user parameter, with a default of 10 –5 . f MAX and f MIN are the
maximum and minimum excitation frequencies of the combined FREQi entries.
8. In design optimization (SOL 200), the excitation frequencies generated from this
entry are derived from the natural frequencies computed in the first design cycle
and the excitation frequencies remain fixed through all subsequent design cycles.
In other words, the excitation frequencies will not be readjusted even if the
natural frequencies change.
 CHAPTER F 405
Bulk Data Entries for Dynamic Analysis

FREQ5 Frequency List, Aternate Form 5

Defines a set of frequencies used in the solution of modal frequency-response problems by

specification of a frequency range and fractions of the natural frequencies within that
range.

Format:
1 2 3 4 5 6 7 8 9 10
FREQ5 SID F1 F2 FR1 FR2 FR3 FR4 FR5

FR6 FR7 -etc.-

Example:

FREQ5 6 20.0 200.0 1.0 0.6 0.8 0.9 0.95

1.05 1.1 1.2

Field Contents
SID Set identification number. (Integer > 0)
F1 Lower bound of frequency range in cycles per unit time.
( Real ≥ 0.0 ; Default = 0.0)
F2 Upper bound of frequency range in cycles per unit time.
(Real > 0.0; F2 > F1; Default = 1.0E20)
FRi Fractions of the natural frequencies in the range F1 to F2. (Real > 0.0)

Remarks:
1. FREQ5 applies only to modal frequency-response solutions (SOLs 11, 111, 146,
and 200) and is ignored in direct frequency response solutions.
2. FREQ5 entries must be selected with the Case Control command FREQUENCY =
SID.
3. The frequencies defined by this entry are given by

f i = FRi ⋅ f N
i

where f N are the natural frequencies in the range F1 through F2.

i
4. In the example above, the list of frequencies will be 0.6, 0.8, 0.9, 0.95, 1.0, 1.05, 1.1,
and 1.2 times each natural frequency between 20 and 2000. If this computation
results in excitation frequencies less than F1 and greater than F2, those computed
excitation frequencies are ignored.
Excitation frequencies may be based on natural frequencies that are not within the
range (F1 and F2) as long as the calculated excitation frequencies are within the
range. Similarly, an excitation frequency calculated based on natural frequencies
within the range (F1 through F2) may be excluded if it falls outside the range.
406

5. Since the forcing frequencies are near structural resonances, it is important that
some amount of damping be specified.
6. All FREQi entries with the same set identification numbers will be used.
Duplicate frequencies will be ignored. f N and f N – 1 are considered duplicated if

f N – f N – 1 < DEFREQ ⋅ f MAX – f MI N

where DFREQ is a user parameter with a default of 10 –5 . The values f MAX and f MIN
are the maximum and minimum excitation frequencies of the combined FREQi
entries.
7. In design optimization (SOL 200), the excitation frequencies generated from this
entry are derived from the natural frequencies computed in the first design cycle
and the excitation frequencies remain fixed through all subsequent design cycles.
In other words, the excitation frequencies will not be readjusted even if the
natural frequencies change.
 CHAPTER F 407
Bulk Data Entries for Dynamic Analysis

LSEQ Static Load Set Definition

Defines a sequence of static load sets.

Format:
1 2 3 4 5 6 7 8 9 10
LSEQ SID DAREA LID TID

Example:

LSEQ 100 200 1000 1001

Field Contents
SID Set identification of the set of LSEQ entries. See Remark 6.
(Integer > 0)
DAREA The DAREA set identification assigned to this static load vector. See
Remark 6. (Integer > 0)
LID Load set identification number of a set of static load entries such as
those referenced by the LOAD Case Control command.
(Integer > 0 or blank)
TID Temperature set identification of a set of thermal load entries such as
those referenced by the TEMP(LOAD) Case Control command.
(Integer > 0 or blank)

Remarks:
1. LSEQ will not be used unless selected in the Case Control Section with the
LOADSET command.
2. The entry is available only in the superelement solution sequences and in SOLs
26, 27, 30, and 31.
3. The number of static load vectors created for each superelement is the number of
unique DAREA IDs on all LSEQ entries in the Bulk Data.
4. DAREA may be referenced by CLOAD, RLOAD1, RLOAD2, TLOAD1, and
TLOAD2 entries in order to apply the static load in nonlinear, static and dynamic
analysis.
5. Element data recovery for thermal loads is not currently implemented in
dynamics.
6. SID and DAREA numbers should be unique with respect to all static and
temperature load set identification numbers.
7. In a nonsuperlement analysis, LID and TID cannot both be blank. In
superelement analysis, they may both be blank as long as static loads are
prescribed in the upstream superelements.
408

PDAMP Scalar Damper Property

Specifies the damping value of a scalar damper element using defined CDAMP1 or
CDAMP3 entries.

Format:
1 2 3 4 5 6 7 8 9 10
PDAMP PID1 B1 PID2 B2 PID3 B3 PID4 B4

Example:

PDAMP 14 2.3 2 6.1

Field Contents
PIDi Property identification number. (Integer > 0)
Bi Force per unit velocity. (Real)

Remarks:
1. Damping values are defined directly on the CDAMP2 and CDAMP4 entries, and
therefore do not require a PDAMP entry.
2. A structural viscous damper, CVISC may also be used for geometric grid points.
3. Up to four damping properties may be defined on a single entry.
4. For a discussion of scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
 CHAPTER F 409
Bulk Data Entries for Dynamic Analysis

PMASS Scalar Mass Property

Specifies the mass value of a scalar mass element (CMASS1 or CMASS3 entries).

Format:
1 2 3 4 5 6 7 8 9 10
PMASS PID1 M1 PID2 M2 PID3 M3 PID4 M4

Example:

PMASS 7 4.29 6 13.2

Field Contents
PIDi Property identification number. (Integer > 0)
Mi Value of scalar mass. (Real)

Remarks:
1. Mass values are defined directly on the CMASS2 and CMASS4 entries, and
therefore do not require a PMASS entry.
2. Up to four mass values may be defined by this entry.
3. For a discussion of scalar elements, see “Scalar Elements (CELASi, CMASSi,
CDAMPi)” on page 135 of the NX Nastran Reference Manual.
410

PVISC Viscous Damping Element Property

Defines properties of a one-dimensional viscous damping element (CVISC entry).

Format:
1 2 3 4 5 6 7 8 9 10
PVISC PID1 CE1 CR1 PID2 CE2 CR2

Example:

PVISC 3 6.2 3.94

Field Contents
PIDi Property identification number. (Integer > 0)
CE1, CE2 Viscous damping values for extension in units of force per unit
velocity. (Real)
CR1, CR2 Viscous damping values for rotation in units of moment per unit
velocity. (Real)

Remarks:
1. Viscous properties are material independent; in particular, they are temperature
independent.
2. One or two viscous element properties may be defined on a single entry.
 CHAPTER F 411
Bulk Data Entries for Dynamic Analysis

RLOAD1 Frequency Response Dynamic Load, Form 1

Defines a frequency-dependent dynamic load of the form

i { θ – 2πfτ }
{ P ( f ) } = { A [ C ( f ) + iD ( f ) ]e }

for use in frequency response problems.

Format:
1 2 3 4 5 6 7 8 9 10
RLOAD1 SID DAREA DELAY DPHASE TC TD

Example:

RLOAD1 5 3 1

Field Contents
SID Set identification number. (Integer > 0)
DAREA Identification number of the DAREA entry set that defines A .
(Integer > 0)
DELAY Identification number of the DELAY entry set that defines τ .
( Integer ≥ 0 )
DPHASE Identification number of the DPHASE entry set that defines θ .
( Integer ≥ 0 )
TC Set identification number of the TABLEDi entry that gives C ( f ) . See
Remark 2. ( Integer ≥ 0 )
TD Set identification number of the TABLEDi entry that gives D ( f ) . See
Remark 2. ( Integer ≥ 0 )

Remarks:
1. Dynamic load sets must be selected with the Case Control command DLOAD =
SID.
2. If any of DELAY, DPHASE, TC, or TD fields are blank or zero, the corresponding
τ , θ , C ( f ) , or D ( f ) will both be zero. Either TC or TD may be blank or zero, but not
both.
3. RLOAD1 loads may be combined with RLOAD2 loads only by specification on a
DLOAD entry. That is, the SID on an RLOAD1 entry must not be the same as that
on an RLOAD2 entry.
4. SID must be unique for all RLOAD1, RLOAD2, TLOAD1, and TLOAD2 entries.
412

RLOAD2 Frequency Response Dynamic Load, Form 2

Defines a frequency-dependent dynamic load of the form

i { φ ( f ) + θ – 2πfτ }
{ P ( f ) } = { A ⋅ B ( f )e }
for use in frequency response problems.

Format:
1 2 3 4 5 6 7 8 9 10
RLOAD2 SID DAREA DELAY DPHASE TB TP

Example:

RLOAD2 5 3 7

Field Contents
SID Set identification number. (Integer > 0)
DAREA Identification number of the DAREA entry set that defines A .
(Integer > 0)
DELAY Identification number of the DELAY entry set that defines τ .
( Integer ≥ 0 )
DPHASE Identification number of the DPHASE entry set that defines θ in
degrees. ( Integer ≥ 0 )
TB Set identification number of the TABLEDi entry that gives B ( f ) .
(Integer > 0)
TP Set identification number of the TABLEDi entry that gives φ ( f ) in
degrees. ( Integer ≥ 0 )

Remarks:
1. Dynamic load sets must be selected with the Case Control command DLOAD =
SID.
2. If any of DELAY, DPHASE, or TP fields are blank or zero, the corresponding τ ,
θ , or φ ( f ) will be zero.

3. RLOAD2 loads may be combined with RLOAD1 loads only by specification on a

DLOAD entry. That is, the SID on an RLOAD2 entry must not be the same as that
on an RLOAD1 entry.
4. SID must be unique for all RLOAD1, RLOAD2, TLOAD1, and TLOAD2 entries.
 CHAPTER F 413
Bulk Data Entries for Dynamic Analysis

SUPORT Fictitious Support

Defines determinate reaction degrees of freedom in a free body.

Format:
1 2 3 4 5 6 7 8 9 10
SUPORT ID1 C1 ID2 C2 ID3 C3 ID4 C4

Example:

SUPORT 16 215

Field Contents
IDi Grid or scalar point identification number. (Integer > 0)
Ci Component numbers. (Integer 0 or blank for scalar points. Any
unique combination of the Integers 1 through 6 for grid points with
no embedded blanks.)

Remarks:
1. The SUPORT entry specifies reference degrees of freedom for rigid body motion.
It is not intended to be used in place of a constraint (i.e., SPCi entry or PS on the
GRID entry).
2. SUPORT and/or SUPORT1 entries are required to perform inertia relief in static
analysis (SOLs 1 and 101).
In SOL 1, the presence of a SUPORT or selection of a SUPORT1 entry
automatically invokes inertia relief analysis.
In SOL 101, PARAM,INREL,–1 must also be specified or the SUPORTi entries will
be treated as constraints.
3. Be careful not to spell SUPORT with two Ps.
4. Degrees of freedom specified on this entry form members of the mutually
exclusive r-set. They may not be specified on other entries that define mutually
exclusive sets. See the “Degree-of-Freedom Sets” on page 701 of the NX Nastran
Quick Reference Guide for a list of these entries.
5. From 1 to 24 support degrees of freedom may be defined on a single entry.
6. See “Rigid Body Supports” on page 263 of the NX Nastran Reference Manual, for
a discussion of supported degrees of freedom (members of the r-set).
7. An alternative to SUPORT is the SUPORT1 entry which is requested by the
SUPORT1 Case Control command.
414

SUPORT1 Fictitious Support, Alternate Form

Defines determinate reaction degrees of freedom (r-set) in a free body analysis. SUPORT1
must be requested by the SUPORT1 Case Control command.

Format:
1 2 3 4 5 6 7 8 9 10
SUPORT1 SID ID1 C1 ID2 C2 ID3 C3

Example:

SUPORT1 5 16 215

Field Contents
SID Identification number of the support set. See Remark 1. (Integer > 0)
IDi Grid or scalar point identification number. (Integer > 0)
Ci Component numbers. (Integer 0 or blank for scalar points. Any
unique combination of the Integers 1 through 6 for grid points with
no embedded blanks.)

Remarks:
1. The SUPORT entry specifies reference degrees of freedom for rigid body motion.
It is not intended to be used in place of a constraint (i.e., SPCi entry or PS on the
GRID entry).
2. SUPORT and/or SUPORT1 entries are required to perform intertia relief in static
analysis (i.e., SOLs 1 and 101).
In SOL 1, the presence of a SUPORT or selection of a SUPORT1 entry
automatically invokes inertia relief analysis.
In SOL 101, PARAM,INREL,-1 must also be specified or the SUPORTi entries will
be treated as constraints.
3. SUPORT1 must be requested by the SUPORT1 Case Control command. The
degrees of freedom specified on SUPORT1 will be combined with those on the
SUPORT entry.
4. Be careful not to spell SUPORT with two P's.
5. Degrees of freedom specified on this entry form members of the mutually
exclusive r-set. They may not be specified on other entries that define mutually
exclusive sets. See the “Degree-of-Freedom Sets” on page 701 of the NX Nastran
Quick Reference Guide for a list of these entries.
6. From 1 to 18 support degrees of freedom may be defined on a single entry.
7. See “Rigid Body Supports” on page 263 of the NX Nastran Reference Manual, for
a discussion of supported degrees of freedom (members of the r-set).
 CHAPTER F 415
Bulk Data Entries for Dynamic Analysis

8. In superelement analysis, SUPORT1 may be specified for points belonging to the

residual structure only.
416

TABDMP1 Modal Damping Table

Defines modal damping as a tabular function of frequency.

Format:
1 2 3 4 5 6 7 8 9 10
TABDMP1 TID TYPE

f1 g1 f2 g2 f3 g3 -etc.-

Example:

TABDMP1 2

2.5 .01057 2.6 .01362 ENDT

Field Contents
TID Table identification number. (Integer > 0)
TYPE Type of damping units. (Character: "G", "CRIT", or "Q"; Default is "G")
fi Natural frequency value in cycles per unit time. ( Real ≥ 0.0 )
gi Damping value. (Real)

Remarks:
1. Modal damping tables must be selected with the Case Control command
SDAMPING = TID.
2. The frequency values fi must be specified in either ascending or descending
order, but not both.
3. Discontinuities may be specified between any two points except the two starting
points or two end points. For example, in Figure F-3 discontinuities are allowed
only between points f2 through f7. Also, if g is evaluated at a discontinuity, then
the average value of g is used. In Figure F-3 the value of g at f = f3 is g = (g3 +
g4)/2.
4. At least one continuation entry must be specified.
5. Any fi or gi entry may be ignored by placing "SKIP" in either of the two fields used
for that entry.
6. The end of the table is indicated by the existence of "ENDT" in either of the two
fields following the last entry. An error is detected if any continuations follow the
entry containing the end-of-table flag "ENDT".
 CHAPTER F 417
Bulk Data Entries for Dynamic Analysis

7. The TABDMP1 uses the algorithm

g = gT ( f )

where f is input to the table and g is returned. The table look-up g T ( f ) is

performed using linear interpolation within the table and linear extrapolation
outside the table using the last two end points. See Figure F-3. No warning
messages are issued if table data is input incorrectly.
y

f value
Discontinuity Range of Table
Allowed Discontinuity
Not Allowed
Linear
Extrapolation
of Segment
f1–f2

f1 f2 f3, f5 f6 f7,
f4 f8

f
Extrapolated

Figure F-3 Example of Table Extrapolation and Discontinuity

8. This form of damping is used only in modal formulations of complex eigenvalue

analysis, frequency response analysis, or transient response analysis. The type of
damping used depends on the solution sequence (structural damping is
displacement-dependent, and viscous damping is velocity-dependent). See
“Formulation of Dynamic Equations in SubDMAP GMA” on page 333 of the
NX Nastran Reference Manual, for the equations used.
9. PARAM,KDAMP may be used in solution sequences that perform modal
frequency and modal complex analysis, to select the type of damping.
KDAMP Result
1 (Default) B Matrix
–1 ( 1 + ig )K
matrix

See “Formulation of Dynamic Equations in SubDMAP GMA” on page 333 of

the NX Nastran Reference Manual, for a full explanation.
10. If TYPE is “G” or blank, the damping values gi, etc., are in units of equivalent
viscous dampers, as follows:
gi
b i = ----- K i
ωi
418

(See “Formulation of Dynamic Equations in SubDMAP GMA” on page 333 of

the NX Nastran Reference Manual) If TYPE is "CRIT", the damping values gi, etc.,
are in the units of fraction of critical damping C ⁄ C 0 . If TYPE is "Q", the damping
values gi are in the units of the amplification or quality factor, Q . These constants
are related by the following equations:
C ⁄ C0 = g ⁄ 2

 1 ⁄ ( 2C ⁄ C 0 )
Q = 
1 ⁄ g
 CHAPTER F 419
Bulk Data Entries for Dynamic Analysis

TABLED1 Dynamic Load Tabular Function, Form 1

Defines a tabular function for use in generating frequency-dependent and time-dependent

dynamic loads.

Format:
1 2 3 4 5 6 7 8 9 10
TABLED1 TID XAXIS YAXIS

x1 y1 x2 y2 x3 y3 -etc.- “ENDT”

Example:

TABLED1 32

–3.0 6.9 2.0 5.6 3.0 5.6 ENDT

Field Contents
TID Table identification number. (Integer > 0)
XAXIS Specifies a linear or logarithmic interpolation for the x-axis. See
Remark 6. (Character: “LINEAR” or “LOG”; Default = “LINEAR”)
YAXIS Same as above, only y-axis.
xi, yi Tabular values. (Real)
“ENDT” Flag indicating the end of the table.

Remarks:
1. xi must be in either ascending or descending order, but not both.
2. Discontinuities may be specified between any two points except the two starting
points or two end points. For example, in Figure F-4 discontinuities are allowed
only between points x2 through x7. Also, if y is evaluated at a discontinuity, then
the average value of y is used. In Figure F-4 the value of y at x = x3 is y = (y3 +
y4)/2. If the y-axis is a LOG axis, then the jump at the discontinuity is evaluated
as y = y3 y 4 .
3. At least one continuation must be specified.
4. Any xi-yi pair may be ignored by placing the character string "SKIP" in either of
the two fields.
5. The end of the table is indicated by the existence of the character string "ENDT"
in either of the two fields following the last entry. An error is detected if any
continuations follow the entry containing the end-of-table flag "ENDT".
6. TABLED1 uses the algorithm
y = yT ( x )
420

where x is input to the table and y is returned. The table look-up is performed
using linear interpolation within the table and extrapolation outside the table
using the two starting or end points. See Figure F-4. The algorithms used for
interpolation or extrapolation are:

XAXIS YAXIS yT ( x )
LINEAR LINEAR xj – x x – xi
----------------- yi + ----------------- yj
xj – xi xj – xi
LOG LINEAR ln ( xj ⁄ x ) ln ( x ⁄ xi )
--------------------------- yi + --------------------------- yj
ln ( xj ⁄ xi ) ln ( xj ⁄ xi )
LINEAR LOG xj – x x – xi
exp ----------------- ln yi + ----------------- ln yj
xj – xi xj – xi

LOG LOG ln ( xj ⁄ x ) ln ( x ⁄ xi )
exp --------------------------- ln yi + --------------------------- ln yj
ln ( xj ⁄ xi ) ln ( xj ⁄ xi )

where xj and yj follow xi and yi.

No warning messages are issued if table data is input incorrectly.
y

x value
Discontinuity Range of Table
Allowed
Discontinuity
Not Allowed
Linear
Extrapolation
of Segment
x1-x2

x
x1 x2 x3 x5 x6 x7
x4 x8
x
Extrapolated

Figure F-4 Example of Table Extrapolation and Discontinuity

7. Linear extrapolation is not used for Fourier transform methods. The function is
zero outside the range of the table.
8. For frequency-dependent loads, xi is measured in cycles per unit time.
9. Tabular values on an axis if XAXIS or YAXIS = LOG must be positive. A fatal
message will be issued if an axis has a tabular value ≤ 0.
 CHAPTER F 421
Bulk Data Entries for Dynamic Analysis

TABLED2 Dynamic Load Tabular Function, Form 2

Defines a tabular function for use in generating frequency-dependent and time-dependent

dynamic loads. Also contains parametric data for use with the table.

Format:
1 2 3 4 5 6 7 8 9 10
TABLED2 TID X1

x1 y1 x2 y2 x3 y3 -etc.-

Example:

TABLED2 15 –10.5

1.0 –4.5 2.0 –4.2 2.0 2.8 7.0 6.5

SKIP SKIP 9.0 6.5 ENDT

Field Contents
TID Table identification number. (Integer > 0)
X1 Table parameter. See Remark 6. (Real)
xi, yi Tabular values. (Real)

Remarks:
1. xi must be in either ascending or descending order, but not both.
2. Discontinuities may be specified between any two points except the two starting
points or two end points. For example, in Figure F-5 discontinuities are allowed
only between points x2 and x7. Also if y is evaluated at a discontinuity, then the
average value of y is used. In Figure F-5 the value of y at x = x3 is y = (y3 + y4)/2.
3. At least one continuation entry must be specified.
4. Any xi-yi pair may be ignored by placing "SKIP" in either of the two fields.
5. The end of the table is indicated by the existence of "ENDT" in either of the two
fields following the last entry. An error is detected if any continuations follow the
entry containing the end-of-table flag "ENDT".
6. TABLED2 uses the algorithm

y = y T ( x – X1 )

where x is input to the table and y is returned. The table look-up is performed
using linear interpolation within the table and linear extrapolation outside the
table using the two starting or end points. See Figure F-5. No warning messages
are issued if table data is input incorrectly.
422

x value
Discontinuity Range of Table
Allowed
Discontinuity
Not Allowed
Linear
Extrapolation
of Segment
x1-x2

x
x1 x2 x3 x5 x6 x7
x4 x8
x
Extrapolated

Figure F-5 Example of Table Extrapolation and Discontinuity.

7. Linear extrapolation is not used for Fourier transform methods. The function is
zero outside the range of the table.
8. For frequency-dependent loads, X1 and xi are measured in cycles per unit time.
 CHAPTER F 423
Bulk Data Entries for Dynamic Analysis

TABLED3 Dynamic Load Tabular Function, Form 3

Defines a tabular function for use in generating frequency-dependent and time-dependent

dynamic loads. Also contains parametric data for use with the table.

Format:
1 2 3 4 5 6 7 8 9 10
TABLED3 TID X1 X2

x1 y1 x2 y2 x3 y3 -etc.-

Example:

TABLED3 62 126.9 30.0

2.9 2.9 3.6 4.7 5.2 5.7 ENDT

Field Contents
TID Table identification number. (Integer > 0)
X1, X2 Table parameters. (Real; X2 ≠ 0.0 )
xi, yi Tabular values. (Real)

Remarks:
1. xi must be in either ascending or descending order, but not both.
2. Discontinuities may be specified between any two points except the two starting
points or two end points. For example, in Figure F-6 discontinuities are allowed
only between points x2 and x7. Also if y is evaluated at a discontinuity, then the
average value of y is used. In Figure F-6 the value of y at x = x3 is y = (y3 + y4)/2.
3. At least one continuation entry must be present.
4. Any xi-yi pair may be ignored by placing "SKIP" in either of the two fields.
5. The end of the table is indicated by the existence of "ENDT" in either of the two
fields following the last entry. An error is detected if any continuations follow the
entry containing the end-of-table flag "ENDT".
6. TABLED3 uses the algorithm

x – X1
y = y T  -----------------
 X2 
where x is input to the table and y is returned. The table look-up is performed
using interpolation within the table and linear extrapolation outside the table
using the two starting or end points. See Figure F-6. No warning messages are
issued if table data is input incorrectly.
424

x value
Discontinuity Range of Table
Allowed
Discontinuity
Not Allowed
Linear
Extrapolation
of Segment
x1-x2

x
x1 x2 x3 x5 x6 x7
x4 x8
x
Extrapolated

Figure F-6 Example of Table Extrapolation and Discontinuity

7. Linear extrapolation is not used for Fourier transform methods. The function is
zero outside the range of the table.
8. For frequency-dependent loads, X1, X2, and xi are measured in cycles per unit
time.
 CHAPTER F 425
Bulk Data Entries for Dynamic Analysis

TABLED4 Dynamic Load Tabular Function, Form 4

Defines coefficients of a power series for use in generating frequency-dependent and time-
dependent dynamic loads. Also contains parametric data for use with the table.

Format:
1 2 3 4 5 6 7 8 9 10
TABLED4 TID X1 X2 X3 X4

A0 A1 A2 A3 A4 A5 -etc.-

Example:

TABLED4 28 0.0 1.0 0.0 100.

2.91 –0.0329 6.51–5 0.0 –3.4–7 ENDT

Field Contents
TID Table identification number. (Integer > 0)
Xi Table parameters. (Real; X2 ≠ 0.0 ; X3 < X4)
Ai Coefficients. (Real)

Remarks:
1. At least one continuation entry must be specified.
2. The end of the table is indicated by the existence of "ENDT" in the field following
the last entry. An error is detected if any continuations follow the entry
containing the end-of-table flag "ENDT".
3. TABLED4 uses the algorithm
N
x – X1 i
y = ∑ Ai  -----------------
X2
i = 0

where x is input to the table, y is returned, and N is the number of pairs.

Whenever x < X3 , use X3 for x ; whenever x > X4 , use X4 for x . There are N + 1
entries in the table. There are no error returns from this table look-up procedure.
4. For frequency-dependent loads, xi is measured in cycles per unit time.
426

TIC Transient Initial Condition

Defines values for the initial conditions of variables used in structural transient analysis.
Both displacement and velocity values may be specified at independent degrees of
freedom. This entry may not be used for heat transfer analysis.

Format:
1 2 3 4 5 6 7 8 9 10
TIC SID G C U0 V0

Example:

TIC 1 3 2 5.0 –6.0

Field Contents
SID Set identification number. (Integer > 0)
G Grid, scalar, or extra point identification number. (Integer > 0)
C Component numbers. (Integer zero or blank for scalar or extra
points, any one of the Integers 1 through 6 for a grid point.)
U0 Initial displacement. (Real)
V0 Initial velocity. (Real)

Remarks:
1. Transient initial conditions sets must be selected with the Case Control command
IC = SID. Note the use of IC in the Case Control command versus TIC on the Bulk
Data entry. For heat transfer, the IC Case Control command selects TEMP or
TEMPD entries for initial conditions and not the TIC entry.
2. If no TIC set is selected in the Case Control Section, all initial conditions are
assumed to be zero.
3. Initial conditions for coordinates not specified on TIC entries will be assumed
zero.
4. Initial conditions may be used only in direct formulation. In a modal formulation
the initial conditions are all zero.
 CHAPTER F 427
Bulk Data Entries for Dynamic Analysis

TLOAD1 Transient Response Dynamic Load, Form 1

Defines a time-dependent dynamic load or enforced motion of the form

{P(t )} = {A ⋅ F(t – τ) }

for use in transient response analysis.

Format:
1 2 3 4 5 6 7 8 9 10
TLOAD1 SID DAREA DELAY TYPE TID

Example:

TLOAD1 5 7 13

Field Contents
SID Set identification number. (Integer > 0)
DAREA Identification number of DAREA entry set or a thermal load set (in
heat transfer analysis) which defines A . (Integer > 0)
DELAY Identification number of DELAY entry set that defines τ .
( Integer ≥ 0 , or blank).
TYPE Defines the nature of the dynamic excitation. See Remark 2.
(Integer 0, 1, 2, 3, Default = 0)
TID Identification number of TABLEDi entry that gives F ( t ) . (Integer > 0)

Remarks:
1. Dynamic load sets must be selected with the Case Control command
DLOAD = SID.
428

2. The nature of the dynamic excitation is defined in field 5 according to the

following table:

Integer Excitation Function

0 Force or Moment (Default)
1 Enforced Displacement
2 Enforced Velocity
3 Enforced Acceleration

See the “Enforced Motion” on page 199 regarding the use of the enforced motion
options. Note that the large mass technique must be used for enforced motion.
For heat transfer and aeroelastic response problems, TYPE (field 5) is ignored and
must be blank.
3. If DELAY is blank or zero, τ will be zero.
4. TLOAD1 loads may be combined with TLOAD2 loads only by specification on a
DLOAD entry. That is, the SID on a TLOAD1 entry may not be the same as that
on a TLOAD2 entry.
5. SID must be unique for all TLOAD1, TLOAD2, RLOAD1, and RLOAD2 entries.
6. DAREA may reference sets containing QHBDY, QBDYi, QVECT, and QVOL
entries when using the heat transfer option.
7. If the heat transfer option is used, the referenced QVECT entry may also contain
references to functions of time, and therefore A may be a function of time.
8. If TLOADi entries are selected in SOL 111 or 146, then a Fourier analysis is used
to tranform the time-dependent loads on the TLOADi entries to the frequency
domain and then combine them with loads from RLOADi entries. Then the
analysis is performed as a frequency response analysis but the solution and the
output are converted to an printed in the time domain.
 CHAPTER F 429
Bulk Data Entries for Dynamic Analysis

TLOAD2 Transient Response Dynamic Load, Form 2

Defines a time-dependent dynamic load or enforced motion of the form

 0 , t < ( T1 + τ ) or t > ( T2 + τ )

{P(t)} = 
 A t̃ B e C t̃ cos ( 2πF t̃ + P ) , ( T1 + τ ) ≤ t ≤ ( T2 + τ )


for use in a transient response problem, where t̃ = t – T1 – τ .

Format:
1 2 3 4 5 6 7 8 9 10
TLOAD2 SID DAREA DELAY TYPE T1 T2 F P

C B

Example:

TLOAD2 4 10 2.1 4.7 12.0

2.0

Field Contents
SID Set identification number. (Integer > 0)
DAREA Identification number of DAREA entry set or a thermal load set (in
heat transfer analysis) that defines A . (Integer > 0)
DELAY Identification number of DELAY entry set that defines τ .
( Integer ≥ 0 , or blank)
TYPE Defines the nature of the dynamic excitation. See Remark 2.
(Integer 0, 1, 2, 3; Default = 0 )
T1 Time constant. ( Real ≥ 0.0 )
T2 Time constant. (Real; T2 > T1)
F Frequency in cycles per unit time. ( Real ≥ 0.0 ; Default = 0.0)
P Phase angle in degrees. (Real; Default = 0.0)
C Exponential coefficient. (Real; Default = 0.0)
B Growth coefficient. (Real; Default = 0.0)

Remarks:
1. Dynamic load sets must be selected with the Case Control command with
DLOAD = SID.
430

2. The nature of the dynamic excitation is defined in accordance with the following
table:

TYPE Excitation Function

0 Force or Moment (Default)
1 Enforced Displacement
2 Enforced Velocity
3 Enforced Acceleration

See the “Enforced Motion” on page 199 regarding the use of the enforced motion
options. Note that "large masses" must be used for enforced motion. For heat
transfer and aeroelastic response problems, TYPE (field 5) is ignored and must be
blank.
3. If DELAY is blank or zero, τ will be zero.
4. TLOAD1 loads may be combined with TLOAD2 loads only by specification on a
DLOAD entry. That is, the SID on a TLOAD1 entry may not be the same as that
on a TLOAD2 entry.
5. SID must be unique for all TLOAD1, TLOAD2, RLOAD1, and RLOAD2 entries.
6. DAREA (field 3) may reference sets containing QHBDY, QBDYi, QVECT, and
QVOL entries when using the heat transfer option.
7. If the heat transfer option is used, the referenced QVECT entry may also contain
references to functions of time, and therefore A may be a function of time.
8. If TLOADi entries are selected in SOL 111 or 146, then a Fourier analysis is used
to tranform the time-dependent loads on the TLOADi entries to the frequency
domain and then combine them with loads from RLOADi entries. Then the
analysis is performed as a frequency response analysis but the solution and the
output are converted to an printed in the time domain. In this case, B will be
rounded to the nearest integer.
9. The continuation entry is optional.
 CHAPTER F 431
Bulk Data Entries for Dynamic Analysis

TSTEP Transient Time Step

Defines time step intervals at which a solution will be generated and output in transient
analysis.

Format:
1 2 3 4 5 6 7 8 9 10
TSTEP SID N1 DT1 NO1

N2 DT2 NO2

-etc.-

Example:

TSTEP 2 10 .001 5

9 0.01 1

Field Contents
SID Set identification number. (Integer > 0)
Ni Number of time steps of value DTi. ( Integer ≥ 1 )
DTi Time increment. (Real > 0.0)
NOi Skip factor for output. Every NOi-th step will be saved for output.
(Integer > 0; Default = 1)

Remarks:
1. TSTEP entries must be selected with the Case Control command TSTEP = SID.
2. Note that the entry permits changes in the size of the time step during the course
of the solution. Thus, in the example shown, there are 10 time steps of value .001
followed by 9 time steps of value .01. Also, the user has requested that output be
recorded for t = 0.0, .005, .01, .02, .03, etc.
3. See the “Guidelines for Effective Dynamic Analysis” on page 269 for a
discussion of considerations leading to the selection of time steps.
4. In modal frequency response analysis (SOLs 111 and 146), this entry is required
only when TLOADi is requested; i.e., when Fourier methods are selected.
5. The maximum and minimum displacement at each time step and the SIL
numbers of these variables can be printed by altering DIAGON(30) before the
transient module TRD1 and by altering DIAGOFF(30) after the module. This is
useful for runs that terminate due to overflow or excessive run times.
6. For heat transfer analysis in SOL 159, use the TSTEPNL entry.
432
 NX Nastran Basic Dynamic Analysis User’s Guide

G
APPENDIX
Parameters for Dynamic Analysis

■ Overview
434

G.1 Overview
This appendix lists some of the parameters that are often used for dynamic analysis. These
parameters are listed alphabetically, beginning below. See the NX Nastran Quick Reference
Guide for a description of all parameters.

ASlNG Default = 0
ASING specifies the action to take when singularities (null rows
and columns) exist in the dynamic matrices (or [ K ll ] in statics). If
ASING = –1, then a User Fatal Message will result.
If ASING = 0 (the default), singularities are removed by
appropriate techniques depending on the type of solution being
performed.
AUTOSPC Default = YES (in SOLutions 21, 38, 39, 61 thru 63, 65 thru 76, 81
thru 91, and 101 through 200 except 106 and 129)
Default = NO (in all other SOLutions)
AUTSPC specifies the action to take when singularities exist in the
stiffness matrix ( [ K gg ] ). AUTOSPC = YES means that singularities
will be constrained automatically. AUTOSPC = NO means that
singularities will not be constrained.
AUTOSPC Singularity ratios smaller than PARAM,EPPRT (default = 1.E-8)
(continued) are listed as potentially singular. If PARAM,AUTOSPC has the
value YES, identified singularities with a ratio smaller than
PARAM,EPZERO (default = 1.E-8) will be automatically
constrained with single-point constraints. If PARAM,EPPRT has
the same value as PARAM,EPZERO (the default case), all
singularities are listed. If PARAM,EPPRT is larger than
PARAM,EPZERO, the printout of singularity ratios equal to
exactly zero is suppressed. If PARAM,PRGPST is set to NO
(default is YES), the printout of singularities is suppressed, except
when singularities are not going to be removed. If
PARAM,SPCGEN is set to 1 (default = 0), the automatically
generated SPCs are placed in SPCi Bulk Data entry format on the
PUNCH file.
AUTOSPC provides the correct action for superelements in all
contexts. It does not provide the correct action for the residual
structure in SOLs 64, 99, or 129. PARAM,AUTOSPCR, not
AUTOSPC, is used for the o-set (omitted set) in the residual
structure in SOLs 66 and 106.
BAILOUT 0 (for superelement sequences)
Default =
-1 (for non-superelement sequences)

See MAXRATIO.
 CHAPTER G 435
Parameters for Dynamic Analysis

CB1, CB2 Default = 1.0

CB1 and CB2 specify factors for the total damping matrix. The
total damping matrix is

x 2
[ B jj ] = CB1 ⋅ [ B jj ] + CB2 ⋅ [ B jj ]

where [ B 2jj ] is selected via the Case Control command B2GG and
[ B jj ] comes from CDAMPi or CVlSC element Bulk Data entries.
x

These parameters are effective only if B2GG is selected in the Case

Control Section.
CK1, CK2 Default = 1.0
CK1 and CK2 specify factors for the total stiffness matrix. The
total stiffness matrix (exclusive of GENEL entries) is

x 2 2
[ K jj ] = CK1 ⋅ [ K j j ] + CK2 ⋅ [ K jj ]

where [ K 2jj ] is selected via the Case Control command K2GG and
[ K jj ] is generated from structural element (e.g., CBAR) entries in
z

the Bulk Data. These are effective only if K2GG is selected in the
Case Control Section. Note that stresses and element forces are
not factored by CK1, and must be adjusted manually.
CM1, CM2 Default = 1.0
CM1 and CM2 specify factors for the total mass matrix. The total
mass matrix is

x 2
[ M jj ] = CM1 ⋅ [ M jj ] + CM2 ⋅ [ M jj ]

where [ M 2jj ] is selected via the Case Control command M2GG and
[ M jj ] is derived from the mass element entries in the Bulk Data
x

Section. These are effective only if M2GG is selected in the Case

Control Section.
COUPMASS Default = -1
COUPMASS > 0 requests the generation of coupled rather than
lumped mass matrices for elements with coupled mass capability.
This option applies to both structural and nonstructural mass for
the following elements: CBAR, CBEAM, CHEXA, CONROD,
CPENTA, CQUAD4, CQUAD8, CROD, CTETRA, CTRIA3,
CTRlA6, CTRIAX6, CTUBE. A negative value (the default) causes
the generation of lumped mass matrices (translational
components only) for all of the above elements.
CP1, CP2 Default = 1.0
436

The load vectors are generated from the equation

 x  2
{ P j } = CP1 ⋅  P j  + CP2 ⋅  P j 
   

where { P 2j } is selected via the Case Control command P2G, and

{ P xj } comes from Bulk Data static load entries.
CURVPLOT Default = -1
PARAM,CURVPLOT,1 requests that X-Y (or curve) plots be made
for displacements, loads, SPC forces, or grid point stresses or
strains. The y values are response values; the x values are related
to grid point locations through the parameter DOPT.
PARAM,CURVPLOT,1 suppresses SORT2-type processing; in
superelement dynamic analysis, SORT1 requests will be honored.
To obtain stress or strain plots, set the CURV parameter to +1.
DOPT controls the x spacing of curves over grid points for the
CURVPLOT module. The default for DOPT is the length between
grid points.
DDRMM Default = 0
By default, the matrix method of data recovery is used in the
modal solutions. DDRMM = –1 will force calculation of complete
g-set solution vectors by the mode displacement method, which is
needed for SORT1 output. SORT1 output is required for deformed
structure plots, grid point force balance output, the mode
acceleration technique, and postprocessing with PARAM,POST,–1
or –2.
DYNSPCF Default = NEW (Structured SOLs only)
PARAM,DYNSPCF,NEW requests that mass and damping
coupled to ground be included in the SPCForce calculations for
the linear dynamic solutions: SOLs 103, 107-112, 115, 118, 145, 146,
and 200. OLD neglects these effects and gives the same SPCForce
results obtained in versions prior to Version 68.
EPZERO Default = 1.E-8
Specifies the minimum value that indicates a singularity. See
AUTOSPC.
G Default = 0.0
 CHAPTER G 437
Parameters for Dynamic Analysis

G specifies the uniform structural damping coefficient in the

formulation of dynamics problems. To obtain the value for the
parameter G, multiply the critical damping ratio, C/Co, by 2.0.
PARAM,G is not recommended for use in hydroelastic or heat
transfer problems. If PARAM,G is used in transient analysis,
PARAM,W3 must be greater than zero or PARAM,G will be
ignored.
HFREQ Default = 1 .+30
The parameters LFREQ and HFREQ specify the frequency range
(LFREQ is the lower limit and HFREQ is the upper limit) of the
modes to be used in the modal formulations. Note that the default
for HFREQ will usually include all vectors computed. A related
parameter is LMODES.
KDAMP Default = 1
If KDAMP is set to –1, viscous modal damping is entered into the
complex stiffness matrix as structural damping.
LFREQ Default = 0.0
See HFREQ.
LMODES Default = 0
LMODES is the number of lowest modes to use in a modal
formulation. If LMODES = 0, the retained modes are determined
by the parameters LFREQ and HFREQ.
MAXRATIO Default = 1.E5
The ratios of terms on the diagonal of the stiffness matrix to the
corresponding terms on the diagonal of the triangular factor are
computed. If, for any row, this ratio is greater than MAXRATIO,
the matrix will be considered to be nearly singular (having
mechanisms). If any diagonal terms of the factor are negative, the
stiffness matrix is considered implausible (non-positive definite).
The ratios greater than MAXRATIO and less than zero and their
associated external grid identities will be printed out. The
program will then take appropriate action as directed by the
parameter BAILOUT.
By default, in the superelement solution sequences the program
will terminate processing for that superelement. A negative value
for BAILOUT directs the program to continue processing the
superelement. Although forcing the program to continue with
near-singularities is a useful modeling checkout technique, it may
lead to solutions of poor quality or fatal messages later in the run.
It is recommended that the default values be used for production
runs. A related parameter is ERROR.
438

In nonsuperelement solution sequences, the default value (–1) of

BAILOUT causes the program to continue processing with near
singularities and a zero value will cause the program to exit if near
singularities are detected.
In SOL 60, and SOLs 101-200 when PARAM,CHECKOUT,YES is
specified, PARAM,MAXRATIO sets the tolerance for detecting
multipoint constraint equations with poor linear independence.
MODACC Default = –1
MODACC = 0 selects the mode acceleration method for data
recovery in dynamic analysis. PARAM,DDRMM,–1 must also be
specified.
NONCUP Default = –1
NONCUP selects either a coupled or non-coupled solution
algorithm in modal transient response analysis. By default, the
noncoupled solution algorithm is selected unless the dynamic
matrices KHH, MHH, or BHH have off-diagonal terms.
NONCUP = 1 requests the coupled algorithm and –2 the
uncoupled algorithm regardless of the existence of off-diagonal
terms in the dynamic matrices. User Information Message 5222
indicates which algorithm is used in the analysis.
NOSORT1 -1 (SOL 30)
Default =
–1 (SOL 31)

NOSORT1 controls the execution of the SDR3 module, which

changes SORT1 modal solutions to SORT2. In SOL 30, where
SORT1 output is the default option, NOSORT1 is set to branch
over this operation. In SOL 31, where SORT2 output is usually
desired, PARAM,NOSORT1,1 may be used when only SORT1
output is desired.
WTMASS Default = 1.0
The terms of the structural mass matrix are multiplied by the
value of this parameter when they are generated. This parameter
is not recommended for use in hydroelastic problems.
W3, W4 Default = 0.0
 CHAPTER G 439
Parameters for Dynamic Analysis

The damping matrix for transient analysis is assembled from the

equation:

1 2 G 1 1 4
[ B dd ] = [ B dd ] + [ B dd ] + --------- [ K dd ] + --------- [ K dd ]
W3 W4
The default values of 0.0 for W3 and W4 cause the [ K 1dd ] and [ K 4dd ]
terms to be ignored in the damping matrix, regardless of the
presence of the PARAM,G or [ K 4dd ] . [ K 1dd ] is the stiffness. [ K 4dd ] is
the structural damping and is created when GE is specified on the
MATi entries.
440
 NX Nastran Basic Dynamic Analysis User’s Guide

H
APPENDIX
File Management Section

■ Overview
■ Definitions

■ MSC.Nastran Database
■ File Management Commands
442

H.1 Overview
The database structure for MSC.Nastran was changed substantially with the
implementation of the Executive System first introduced in Version 66. Furthermore, a
new section called the File Management Section (FMS) was added to the MSC.Nastran
input file. The goal is to make file assignment as uniform as possible across different
computer types.

This appendix is intended to give you an overview of how the database and FMS work so
that you can allocate your computer resources efficiently, especially for large models. For
many problems, due to the default values, the use or knowledge of the FMS is either
transparent or not required by you.
 CHAPTER H 443
File Management Section

H.2 Definitions
Before presenting the details of the database description, it is helpful to define some of the
basic terms that are used throughout this appendix.

DBset Database set. This consists of an MSC.Nastran logical name, such

as MASTER, DBALL, etc., which refers to an entity within the
MSC.Nastran database. It consists of a collection of physical files.
Database Collection of all DBsets assigned to a run.
Data block Matrix or table (e.g., KAA, LAMA) that is stored in the database.
Logical name Local internal name (log-name) used by MSC.Nastran for a file or
DBset.
Word For 32- and 64-bit machines, each word is equivalent to four and
eight bytes, respectively.
Buffsize Length of an MSC.Nastran buffer in terms of words (32- or 64-bit
words). It controls the physical record size for data
storage/transfer that is contained in many MSC.Nastran logical
units. The default and maximum allowable buffsize is machine
dependent. The default value is recommended except for large
problems. It can be modified by using the following NASTRAN
statement:
NASTRAN BUFFSIZE = xxxxx
where xxxxx is the desired buffsize.
Block This is often referred to as a MSC.Nastran GINO block. Each
GINO block contains one MSC.Nastran buffer.
{} A brace indicates that the quantity within this bracket is
mandatory. The underlined item within { } is the default value.
[] A square bracket indicates that the quantity within this bracket is
optional.
444

H.3 MSC.Nastran Database

When submitting a MSC.Nastran job, a series of standard output files is created (e.g., F06
file). Conventions for the filenames are machine dependent. Furthermore, four permanent
(MASTER, DBALL, USRSOU, USROBJ) and one scratch (SCRATCH) DBsets are created
during the run.

MASTER This is the “directory” DBset that contains a list of all the DBsets
used in the job, all the physical file names assigned, and an index
pointing to all the data blocks created and where they are stored.
In addition, it also contains the MSC.Nastran Data Definition
Language (NDDL) used. NDDL is the internal MSC.Nastran
language that describes the database. You do not need to
understand NDDL to use MSC.Nastran effectively. The default
maximum size for MASTER is 5000 blocks.
DBALL This is the DBset where the permanent data blocks are stored by
default. The default maximum size is 25000 blocks.
USRSOU This DBset stores the source file for user-created DMAP. The
default maximum size is 5000 blocks.
USROBJ This DBset stores the object file for user-created DMAP. The
default maximum size is 5000 blocks.
SCRATCH This DBset is used as the temporary workspace for MSC.Nastran.
In general, this DBset is deleted at the end of the run. The default
maximum size is 350100 blocks.

For most solutions, USRSOU and USROBJ are not needed and may be deleted or assigned
as temporary for the duration of the run (see “ASSIGN” on page 446).
For a typical UNIX-based workstation with an MSC.Nastran input file called “dyn1.dat”,
the following sample submittal command can be used:

nastran dyn1 scr=no

where nastran is the name of shell script for executing MSC.Nastran. The following four
physical database files are created as a result of the above command.
dyn1.MASTER
dyn1.DBALL
dyn1.USROBJ
dyn1.USRSOU
Unless otherwise stated, the input filename is assumed to be “dyn1.dat” in this appendix.
 CHAPTER H 445
File Management Section

H.4 File Management Commands

The File Management Section is intended primarily for the attachment and initialization of
MSC.Nastran database sets (DBsets) and FORTRAN files. For many problems, due to the
default values, the use of the FMS is either transparent or not required by you. At first
glance the FMS may be overwhelming due in part to its flexibility and the many options it
has to offer. However, in its most commonly used form, it can be quite simple. Examples
in the most commonly used format are presented throughout this appendix.

There are numerous FMS statements available in MSC.Nastran. This section covers the
commonly used FMS statements in dynamic analysis. They are presented in their most
commonly used format, which in general, is a simplified format.

If an FMS statement is longer than 72 characters, it can be continued to the next line by
using a comma as long as the comma is preceded by one or more blank spaces.

INIT
Purpose
Creates/initializes permanent and/or temporary DBsets. The INIT statement has two
basic formats: one for all the DBsets and one specifically for the SCRATCH DBsets.

Format (Simplified) for All DBsets Except SCRATCH

INIT DBset-name [LOGICAL=(log-name1(max-size1) ,
log-name2(max-size2),...lognamei(max-sizei) ,
...lognamen(max-sizen)]

DBset-name This is the logical name of the DBset being used (e.g., DBALL).
log-namei This is the i-th logical name for the DBset-name referenced in this
INIT statement. You can have up to 10 logical names for each DBset
( 1 ≤ i ≤ 10 ). The i-th physical file is assigned with the “assignment”
statement.
max-sizei This is the maximum allowable number of MSC.Nastran blocks
which may be written to the i-th file.

Example INIT DBALL LOGICAL=(DBALL(50000))

This statement creates the DBALL DBset with a logical name of DBALL and the maximum
size of 50000 MSC.Nastran blocks instead of 25000 blocks, which is the default value for
DBALL. Unless an ASSIGN statement is also used, the physical file is given the name
dyn1.DBALL assuming that your input file is called “dyn1.dat”.
INIT DBALL LOGICAL=(DB1(35000),DBTWO(60000))
This statement creates the DBALL DBset with logical names of DB1 and DBTWO. Two
physical files, dyn1.DB1 and dyn1.DBTWO, are created with a maximum size of 35000 and
60000 MSC.Nastran blocks, respectively.
446

Format (Simplified) for the SCRATCH DBset

INIT SCRATCH [LOGICAL=(log-name1(max-size1) ,
log-name2(max-size2),...log-namei(max-sizei)) ,
SCR300=(log-namei+1(max-sizei+1),...log-namen(max-sizen))]
Log-name1 through log-namei are allocated for regular scratch files as temporary
workspace. This temporary workspace is not released until the end of the job. SCR300 is
a special keyword which indicates that the log-names are members reserved for DMAP
module internal scratch files. The space occupied by these SCR300 files is for the duration
of the execution of the module. This SCR300 space is released at the end of the module
execution. You can have up to a combined total of 10 logical names for the SCRATCH
DBset ( 1 ≤ i ≤ 10 ).

Example INIT SCRATCH LOGICAL=(SCR1(150000),SCR2(100000)) ,

SCR300=(SCRA(250000),SCRB(300000))
This statement creates the SCRATCH DBset with logical names of SCR1, SCR2, SCRA, and
SCRB. Two physical files, dyn1.SCR1 and dyn1.SCR2, are created with a maximum size of
150000 and 100000 blocks, respectively. These two files are regular scratch files. Two
additional physical files, dyn1.SCRA and dyn1.SCRB, are created with a maximum size of
250000 and 300000 blocks, respectively. These two files are SCR300 type files.

ASSIGN
Purpose
Assigns physical filenames to logical filenames or special FORTRAN files that are used by
other FMS statements or DMAP modules.

Format (Simplified) to Assign Logical Files

ASSIGN log-namei=’filenamei’ [TEMP DELETE]
log-namei The i-th logical name for the DBset created by the INIT statement.
TEMP Requests that filenamei be deleted at the end of the job. This is
optional and is often used for USRSOU and USROBJ.
DELETE Requests that filenamei be deleted if it exists before the start of the
run. This is optional; however, if this option is not used and the
FORTRAN file exists prior to the current run, then the job may fail
with the following messages:
 CHAPTER H 447
File Management Section

*** USER FATAL MESSAGE 773 (DBDEF)

THE FOLLOWING PHYSICAL FILE ALREADY EXISTS.
LOGICAL NAME = xxxx
PHYSICAL FILE = yyyy
USER INFORMATION: NO ASSOCIATED DEFAULT FILES OR ASSIGNED DBSETS CAN
EXIST PRIOR TO THE DATA BASE INITIALIZATION RUN.
USER ACTION: DELETE THIS FILE AND RESUBMIT THE JOB.

DELETE is not a suggested option if you are using RESTART since

you can delete your database inadvertently. Manual deletion of
unwanted databases is a safer approach.

Example ASSIGN DB1=’sample.DB1’

INIT DBALL LOGICAL=(DB1(50000))
These statements create a physical file called sample.DB1 for the logical name DB1 in the
current directory. Without the ASSIGN statement, the physical file name created is called
dyn1.DB1, assuming once again that your input file is called dyn1.dat.
ASSIGN DB1=’/mydisk1/se/sample.DB1’
ASSIGN DB2=’/mydisk2/sample.DB2’
INIT DBALL LOGICAL=(DB1(50000),DB2(40000))
Two logical names DB1 and DB2 are created for the DBset DBALL. DB1 points to a
physical file called sample.DB1 that resides in the file system (directory) /mydisk1/se.
DB2 points to a physical file called sample.DB2 that resides in the file system (directory)
/mydisk2.

Format (Simplified) to Assign FORTRAN Files

ASSIGN log-key=’filenamef’ [STATUS={new,old},UNIT=u ,
FORM={FORMATTED,UNFORMATTED},DELETE]

log-key This is the logical keyword for the FORTRAN file being created. The
default values depend on the keyword. Acceptable keywords are
DBC,DBMIG,INPUTT2,INPUTT4,OUTPUT2,OUTPUT4,
DBUNLOAD,DBLOAD, and USERFILE.

You should reference the NX Nastran Quick Reference Guide for detailed descriptions and
the defaults for these keywords.

filenamef This is the physical name of the FORTRAN file.

STATUS Specifies whether the FORTRAN file will be created (STATUS = new)
or is an existing file (STATUS = old).
UNIT Specifies the FORTRAN unit (e.g., UNIT = 12).
FORM Specifies whether the file written is in ASCII (FORM = FORMATTED)
or binary (FORM = UNFORMATTED) format.
DELETE Requests that filenamef be deleted if it exists before the start of the
run.
448

Example ASSIGN OUTPUT2=’sample.out’,STATUS=NEW,UNIT=11,FORM=FORMATTED,

DELETE
This example creates a new FORTRAN file to be used for OUTPUT2 operations. This file
is in ASCII format with a physical filename of sample.out and is assigned to unit 11.

EXPAND
Purpose
Concatenates files into an existing DBset in order to increase the allowable disk space. The
EXPAND statement is normally used in a restart run when you run out of disk space in
your previous run.

Format
EXPAND DBset-name LOGICAL=(log-namei(max-sizei),... )

DBset-name This is the logical name of the DBset to be expanded by the addition
of new members to this existing DBset.
log-namei This is the logical name of the i-th member of the DBset. An ASSIGN
statement should be used to point this logical name to a physical file.
max-sizei This is the maximum allowable number of MSC.Nastran blocks that
may be written to the i-th member.

Example The original run creates a database with a name dyn1.DBALL. This database is filled and
the job fails with the following erorr messages in the F06 file:
*** USER FATAL MESSAGE 1012 (GALLOC)
DBSET DBALL IS FULL AND NEEDS TO BE EXPANDED.
For small to medium size problems, it is best to rerun the job from the beginning with a
larger file allocation. For large problems, if rerunning the job is not practical, then the
database can be expanded with the following statements:

RESTART
ASSIGN MASTER=’dyn1.MASTER’
ASSIGN DBADD=’morespace.DB’
EXPAND DBALL LOGICAL=(DBADD(50000))
These statements create an additional member (with a logical name of DBADD) to the
existing DBset DBALL. This member points to a new physical file called morespace.DB,
which may contain up to a maximum of 50000 MSC.Nastran blocks. In this case, you are
restarting from “dyn1.MASTER”.
 CHAPTER H 449
File Management Section

RESTART

Purpose
Allows you to continue from the end of the previous run to the current run without
resolving the problem from the beginning.

Format
RESTART [PROJECT=’proj-ID’, VERSION={version-ID,LAST},
{KEEP,NOKEEP}]

proj-ID Project identifier used in the original run, which can have up to 40
characters. This is optional and is normally not used. The default
proj-ID is blank.
version-ID The version number you are restarting from. The default is the last
version.
KEEP If this option is used, then the version that you are restarting from is
also saved at the end of the current run.
NOKEEP If this option is used, then the version that you are restarting from is
deleted at the end of the current run. This is the default.

Example RESTART
The current run uses the last version in the database for restart. At the end of the run, this
last version is deleted from the database. This statement is probably the most commonly
used format for RESTART.

RESTART VERSION=5,KEEP
The current run (version 6 or higher) uses version 5 in the database for restart. At the end
of the run, version 5 is also retained in the database. This format is used most often when
you want to ensure that a specific version is saved in the database (i.e., a large normal
modes run).

RESTART PROJ=’xyz’ VERSION=3

The current run uses version 3 with a proj-ID of xyz in the database for restart. At the end
of the run, version 3 with a proj-ID of xyz is deleted from the database.
450

INCLUDE
Purpose
Inserts an external file at the location where this “include” statement is used. This is not a
pure FMS statement because it can be used anywhere in the input file, not just in the FMS
Section. The “include” statement must not be nested; in other words, you cannot attach a
file that contains an “include” statement.

Format
Include ’filename’

filename Physical filename of the external file to be inserted at this location.

Example Sol 101

time 10
cend
include ’sub1.dat’
begin bulk
$
include ’bulk1.dat’
include ’bulk2.dat’
$
$ rest of bulk data file
$
.
.
enddata
This run reads a file called sub1.dat with all the Case Control commands contained in it. It
also reads two additional files (bulk1.dat and bulk2.dat) in the Bulk Data Section. You
may, for example, want to include all your grid Bulk Data entries in the file bulk1.dat and
all your element connectivities in bulk2.dat. As you can see, the “include” statement can
be a handy tool. For parametric studies, you can potentially save a tremendous amount of
disk space by using the “include” statement instead of having multiple files with duplicate
input data.

Summary
Due to the default values, very little knowledge of the MSC.Nastran FMS statements and
database structure is required for small to medium size problems. For large problems,
however, some knowledge of the FMS statements and database structure can help you to
optimize your computer resources.
 NX Nastran Basic Dynamic Analysis User’s Guide

I
APPENDIX
Grid Point Weight Generator

■ Overview
■ Commonly Used Features

■ Example with Direction Dependent Masses

452

I.1 Overview
The grid point weight generator (GPWG) calculates the masses, centers of gravity, and
inertias of the mathematical model of the structure. The data are extracted from the mass
matrix by using a rigid-body transformation calculation. Computing the mass properties
is somewhat complex because a finite element model may have directional mass
properties, that is, the mass may differ in each of the three coordinate directions. From a
mathematical point of view, the MSC.Nastran mass may have tensor properties similar to
the inertia tensor. This complexity is reflected in the GPWG output. All of the
transformations used in calculating the mass properties are shown for the general case.
Since most models have the same mass in each of the three coordinate directions, the
GPWG output provides more information than you generally need.

To avoid unnecessary confusion and at the same time provide the necessary information
for the advanced user, the discussion of the GPWG is separated into two sections. In
Section I.2, a basic discussion is given that should satisfy most users. If you need
additional information, read Section I.3.
In both sections, a simple model consisting of four concentrated masses is used to
demonstrate the GPWG output. In the first section, the mass is the same in each direction.
For the second section, the mass is different in each of the three directions.
 CHAPTER I 453
Grid Point Weight Generator

I.2 Commonly Used Features

To demonstrate the typical output generated by the GPWG, a small model consisting of
four concentrated masses as shown in Figure I-1 is used. This model is typical of most
models because the mass is the same in each coordinate direction. The number of masses
has been kept small so you can better understand the physics.
Concentrated masses are located at four different grid points. The displacement
coordinate system for each of the grid points is the basic coordinate system.

zb
4 (5 Mass Units)
1.0

yb
1.0 3 (3 Mass Units)

2 (3 Mass Units)
1 (2 Mass Units) xb
0.0 0.5 1.0
Figure I-1 Four Concentrated Mass Model

To request the GPWG output, you must add parameter GRDPNT in either the Bulk Data
Section or the Case Control Section as follows:
PARAM,GRDPNT,x
If

x = –1 GPWG is skipped (default).

x=0 The mass properties are computed relative to the origin of the basic
coordinate system.
x>0 The mass properties are computed relative to grid point x. If grid point x
does not exist, the properties are computed relative to the basic coordinate
system.

For the four masses shown in Figure I-1, the resulting GPWG output is given in Figure I-2.
454

O U T P U T F R O M G R I D P O I N T W E I G H T G E N E R A T O R
REFERENCE POINT = 0
M O
* 1.300000E+01 0.000000E+00 0.000000E+00 0.000000E+00 5.000000E+00 -3.000000E+00 *
* 0.000000E+00 1.300000E+01 0.000000E+00 -5.000000E+00 0.000000E+00 7.000000E+00 *
* 0.000000E+00 0.000000E+00 1.300000E+01 3.000000E+00 -7.000000E+00 0.000000E+00 *
* 0.000000E+00 -5.000000E+00 3.000000E+00 8.000000E+00 -1.500000E+00 -2.500000E+00 *
* 5.000000E+00 0.000000E+00 -7.000000E+00 -1.500000E+00 1.000000E+01 0.000000E+00 *
* -3.000000E+00 7.000000E+00 0.000000E+00 -2.500000E+00 0.000000E+00 8.000000E+00 *
S
* 1.000000E+00 0.000000E+00 0.000000E+00 *
* 0.000000E+00 1.000000E+00 0.000000E+00 *
* 0.000000E+00 0.000000E+00 1.000000E+00 *
DIRECTION
MASS AXIS SYSTEM (S) MASS X-C.G. Y-C.G. Z-C.G.
X 1.300000E+01 0.000000E+00 2.307692E-01 3.846154E-01
Y 1.300000E+01 5.384616E-01 0.000000E+00 3.846154E-01
Z 1.300000E+01 5.384616E-01 2.307692E-01 0.000000E+00
I(S)
* 5.384615E+00 -1.153847E-01 -1.923079E-01 *
* -1.153847E-01 4.307692E+00 -1.153846E+00 *
* -1.923079E-01 -1.153846E+00 3.538461E+00 *
I(Q)
* 5.503882E+00 *
* 5.023013E+00 *
* 2.703873E+00 *
Q
* 8.702303E-01 4.915230E-01 3.323378E-02 *
* 3.829170E-01 -7.173043E-01 5.821075E-01 *
* 3.099580E-01 -4.938418E-01 -8.124324E-01 *

Figure I-2 GPWG Output for the Four Concentrated Mass Model

The [ MO ] matrix represents the rigid-body mass properties of the structure and is generally
not needed for model checkout. This matrix represents an intermediate step in computing
the inertia properties of the structure. The [ S ] matrix should always be equal to the identity
matrix when the mass is the same in each coordinate direction, which is the typical case. If
this matrix is not the identity matrix, inspect the model for inconsistent masses.
Following the [ S ] matrix are the mass and center of gravity locations. These are the most
commonly used information of the GPWG output. Because the mass may be different in
the three translational directions, the mass is printed for every coordinate direction. For
the same reason, the center of gravity location is given for each of the three translational
masses. If the mass is the same in all directions, a unique center of gravity exists and is
located at the x-component of the y (or z) mass, the y-component of the x (or z) mass, and
the z-component of the x (or y) mass. If the mass is not the same in all three directions, it
is likely due to the CONM1, CMASSi, or DMIG input. If the reference point is specified
with PARAM,GRDPNT,0 and [ S ] is the identity matrix, then the center of gravity location
is given in the basic coordinate system. For the example, the mass of the structure is 13.0
and the center of gravity location is (0.5384, 0.2307, 0.3846) in the basic coordinate system.
If a grid point ID is used for the reference point, and [ S ] is an identity matrix, then the
center of gravity location is in a coordinate system parallel to the basic coordinate system
with an origin located at the grid point.
 CHAPTER I 455
Grid Point Weight Generator

If the [ S ] matrix is equal to the identity matrix, then the [ I ( S ) ] matrix represents the inertia
matrix of structure for the center of gravity with respect to the basic coordinate system, the
[ I ( Q ) ] matrix is the corresponding principal moments of inertia matrix, and [ Q ] represents
the transformation from the principal directions to the basic coordinate system.
The following additional comments on the GPWG should be noted.
• The scale factor entered with parameter WTMASS is applied to the assembled
element mass before the GPWG. The GPWG module, however, converts mass
back to the original input units that existed prior to the scaling effect of the
parameter WTMASS. (Note that the parameter WTMASS is not applied to M2GG
or M2PP input, but the M2GG mass is assembled into the mass matrix prior to
GPWG. Therefore, for GPWG output only, the M2GG mass is scaled by the same
parameter as the element mass. M2GG input may be scaled independently using
the CM2 parameter.)
• The GPWG is performed on the g-size mass matrix, which is the mass matrix prior
to the processing of the rigid elements, MPCs, and SPCs.
• The mass at scalar points and fluid-related masses are not included in the GPWG
calculation.
• The GPWG for a superelement does not include the mass from upstream
superelements. Therefore, the GPWG for the residual structure includes only the
mass on the residual points. The center of gravity location is also based on the
mass of the current superelement only.
• If a large mass is used for enforced motion, the large mass dominates the
structural mass. For model checkout, it is recommended to remove the large mass
and constrain the driving point. A static analysis is a convenient way to generate
a mass matrix and obtain output from the GPWG.
• The output from the GPWG is for information purposes only and is not used in
the analysis.
456

I.3 Example with Direction Dependent Masses

In the previous section, the mass was the same in each of the three coordinate directions
thereby producing a unique center of gravity location. However, if scalar masses are used,
the total mass may have different properties in each direction, and the center of gravity
may not be a unique location. This effect is shown in the output by providing the direction
and center of gravity for each of the three mass principal components.

When using directional mass, the axes about which the inertia matrix i [ S ] is calculated are
referred to as the principal mass axes. The principal mass axes may not necessarily
intersect. However, these axes provide uncoupled rotation and translation mass
properties. If the structural model is constructed using only real masses, the three principal
mass values printed out are equal, the center of gravity is unique, and the principal mass
axes intersect at the center of gravity.
To demonstrate all of the features of the GPWG module, the four-mass sample problem
discussed in the previous section is modified so that the mass is not equal in each of the
three translational directions (see Figure I-3). Furthermore, different displacement
coordinate systems are used for the grid points. The displacement coordinate system for
grid point 1 is the local rectangular system 1, which is oriented at an angle of 45 degrees
(about the z b axis). The displacement coordinate system for grid point 3 is the local
rectangular system 3, which is oriented at an angle of 60 degrees (about the z b axis). The
grid point locations and masses are summarized in Table I-1.

zb
4
1.0 z3
y3
yb x3

1.0 60°
z1 y1
3
x1
2
1 45° xb
0.0 0.5 1.0

Figure I-3 Four Concentrated Mass Model

Table I-1 Location and Size of Masses

Location Mass
Grid ID Basic System (CP Fields) Global System (CD Fields)
xb yb zb xCD yCD zCD
1 0 0 0 2 3 5
2 1 0 0 2 3 5
3 0.5 1 0 2 3 5
4 0.5 0 1 2 3 5
 CHAPTER I 457
Grid Point Weight Generator

The GPWG output for the four mass model is shown in Figure I-4.

*** USER WARNING MESSAGE 3042 MODULE = GPWG

INCONSISTENT SCALAR MASSES HAVE BEEN USED. EPSILON/DELTA = 3.8054429E-02

O U T P U T F R O M G R I D P O I N T W E I G H T G E N E R A T O R

REFERENCE POINT = 1
M O
* 9.250000E+00 -9.330128E-01 0.000000E+00 0.000000E+00 2.000000E+00 -2.966506E+00 * Rigid-Body Mass
* -9.330128E-01 1.075000E+01 0.000000E+00 -3.000000E+00 0.000000E+00 6.058013E+00 *
Properties Matrix for
* 0.000000E+00 0.000000E+00 2.000000E+01 5.000000E+00 -1.000000E+01 0.000000E+00 *
* 0.000000E+00 -3.000000E+00 5.000000E+00 8.000000E+00 -2.500000E+00 -1.500000E+00 * the Reference Point
* 2.000000E+00 0.000000E+00 -1.000000E+01 -2.500000E+00 9.500000E+00 0.000000E+00 *
* -2.966506E+00 6.058013E+00 0.000000E+00 -1.500000E+00 0.000000E+00 7.495513E+00 *

* 4.321332E-01
S
9.018098E-01 0.000000E+00 *
Transformation from
* -9.018098E-01 4.321332E-01 0.000000E+00 * the Principal Mas to the
* 0.000000E+00 0.000000E+00 1.000000E+00 * Basic Direction
DIRECTION
MASS AXIS SYSTEM (S) MASS X-C.G. Y-C.G. Z-C.G.
Center of Gravity
X 1.119709E+01 3.480388E-02 6.023980E-01 2.512494E-01 Relative to the Reference
Y 8.802916E+00 -6.515555E-03 -4.426965E-02 2.484108E-01 Point in the Principal
Z 2.000000E+01 -9.385824E-03 5.589382E-01 0.000000E+00 Mass Axes System
I(S) Moments of Inertia
* 4.376953E+00 -8.768300E-01 6.624477E-01 * with Respect to
* -8.768300E-01 5.623007E+00 -3.419880E-01 * Principal Mass Axes for
* 6.624477E-01 -3.419880E-01 3.431904E+00 *
the Center of Gravity
I(Q)
* 4.463246E+00 * Principal Moments of
* 6.075616E+00 * Inertia
* 2.893001E+00 *

Q Transformation from the

* 7.201511E-01 4.586845E-01 5.205678E-01 * Principal Direction of the
* -3.717336E-01 8.885992E-01 -2.687111E-01 * Momental Ellipsoid to the
* -5.858298E-01 0.000000E+00 8.104341E-01 * Principal Mass Axes

Figure I-4 Output from the Grid Point Weight Generator

Before showing how each of the matrices are computed, a few items should be noted for
this model:

• User Warning Message 3042 is printed to inform you that inconsistent scalar
masses were used. This message occurs because there are different scalar masses
in the three components. In general, if you are using structural mass and/or
CONM2s, you should not get this message.
• The rigid-body mass matrix [ MO ] is computed with respect to the reference grid
point1 in the basic coordinate system. Grid point 1 is used for this example
because PARAM,GNDPNT,1 is entered in the Bulk Data Section.
• The mass and center of gravity location shown are not in the basic coordinate
system for this example. The mass and center of gravity are computed and
printed in the principal mass coordinate system. This principal mass coordinate
system should not be confused with the principal axes discussed in most text
books. The principal mass axes in MSC.Nastran are the axes that have no
458

coupling terms between the translational masses (diagonal translational mass

matrix). Also, the MSC.Nastran principal mass axes are not the axes of the inertia
ellipsoid. The [ S ] matrix is the transformation from the principal mass direction
to the basic coordinate system.
This additional step may sound confusing, but it is necessary. In real structures,
the mass of structure is generally the same in all directions, so there is no coupling
between the translational mass terms. Since text books are written to solve real
structural problems, there is no need to discuss the principal mass axes, as they
are called in MSC.Nastran. However, with MSC.Nastran you are not restricted to
the same mass in each coordinate direction—a situation that may not be
physically realizable but still is quite useful for certain modeling situations.
Therefore, the additional step of computing principal mass axes is necessary.
• If your model has the same mass in all coordinate directions, then the [ S ] matrix
is the identity matrix indicating that the principal mass axes is the basic
coordinate system. Always check the [ S ] matrix. If it is not the identity matrix,
verify that the mass distribution is correct. Do not use the directional mass and
center of gravity location blindly. Remember, these quantities are in the principal
mass axes.

To fully understand how the GPWG module works, it is useful to trace the steps
MSC.Nastran follows to generate the output shown in Figure I-2. The following shows the
step-by-step procedure for the four mass example.
1. The GPWG module uses the global mass matrix, which is the mass matrix before
any constraints are applied. In this example, there are four grids, each with six
degrees-of-freedom, resulting in a total of 24 degrees-of-freedom in the mass
[ MJJ ] matrix. The matrix is shown in Figure I-5.
 CHAPTER I 459
Grid Point Weight Generator

2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2
0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3
0 0 Grid
0 0 10 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6
0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7
0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8
0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9
0 0 0 0 0 0 0 0 Grid
0 0 20 0 0 0 0 0 0 0 0 0 0 0 0 0 10
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 12
[ MJJ ] =
0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 13
0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 14
0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 15
0 0 0 0 0 0 0 0 0 0 0 0 0 0 Grid
0 0 30 0 0 0 0 0 0 0 16
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 17
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 18
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 19
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 20
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 0 0 0 21
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Grid
0 0 40 0 22
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 23
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Figure I-5 Global Mass Matrix

The [ MJJ ] matrix shows the mass contribution for each of the four grid points.
Note that the coordinate system associated with rows and columns 1 through 6
(grid point 1) is coordinate system 1, the coordinate system associated with rows
and columns 13 through 18 (grid point 3) is coordinate system 3. The remaining
rows and columns are in the basic coordinate system.
2. To generate the 6 × 6 rigid-body mass matrix [ MO ] for the structure, it is necessary
to compute the mass matrix in the basic coordinate system relative to the
reference point. This computation requires the transformation matrix [ D ] that
relates the rigid-body displacements in the global system u g to the six unit
displacements in the basic coordinate system located at the reference grid point
( u o ) as shown in Eq. I-1.

{ u g } = [ D ] { u o } (Reference Point) Eq. I-1

The transformation matrix [ D ] is assembled from the individual transformation

matrices [ d ]i computed for each grid point. Each individual transformation
matrix [ d ]i consists of two transformations: [ Tr ]i , which relates the location of the
grid point to the reference grid point in the basic coordinate system, and [ Ti ]i ,
which relates the global coordinate system at the grid point to the basic
coordinate system.
The [ Tr ]i transformation matrices are first computed by constructing the location
vectors in the basic coordinate system ri for each grid point in the model relative
to the reference point as shown in Eq. I-2.
460

 r1 
 
{ r }i = { R }i – { R }0 =  r2  Eq. I-2
 
 r3 

The location vectors for the example are as follows:

0  1   .5   .5 
       
{ r } 1 =  0  ; { r } 2 =  0  ; { r } 3 =  1 ; { r } 4 =  0  Eq. I-3
       
0  0  0  1

Using the location vectors, the grid point transformation matrix [ Tr ]i is computed
for each grid point by expanding the location vectors to a 3 × 3 matrix as shown
in Eq. I-4.

0 r3 –r2
[ Tr ] = –r3 0 r1 Eq. I-4
i
r2 –r1 0

For the example problem, the grid point transformation matrices are

0 0 0 0 0 0 0 0 –1 0 1 0
[ Tr ] 1 = 0 0 0 ; [ Tr ] 2 = 0 0 1 ; [ Tr ] 3 = 0 0 .5 ; [ Tr ] 1 = – 1 0 .5 Eq. I-5
0 0 0 0 –1 0 1 – .5 0 0 –.5 0

The coordinate system transformation matrices from the global coordinate

system to the basic system coordinates are given by the direction cosine matrices
as follows:

cos ( 45° ) – sin ( 45° ) 0 1 0 0

[ Ti ] 1 = sin ( 45° ) cos ( 45° ) 0 ; [ Ti ] 2 = 0 1 0 ;
0 0 1 0 0 1
Eq. I-6
cos ( 60° ) – sin ( 60° ) 0 1 0 0
[ Ti ] 3 = sin ( 60° ) cos ( 60° ) 0 ; [ Ti ] 4 = 0 1 0
0 0 1 0 0 1

The grid point transformation [ Tr ]i and the coordinate system transformation

[ Ti ] i are combined to form the individual grid point transformation matrix [ d ] i for
each grid point using Eq. I-7.

T T
Ti Ti ⋅ T r
[ d ]i = Eq. I-7
T
0 Ti
 CHAPTER I 461
Grid Point Weight Generator

The rows of each [ d ] i form the columns of the global transformation matrix [ D ]T
as shown in Eq. I-8.

T
[D] = T
d1 d2 … dn
T T Eq. I-8

Using Eq. I-9, the global transformation matrix for the example is

.707 – .707 0 0 0 0 1 0 0 0 0 0 .5 – .866 0 0 0 0 1 0 0 0 0 0

.707 .707 0 0 0 0 0 1 0 0 0 0 .866 .5 0 0 0 0 0 1 0 0 0 0
T
[D] = 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0
0 0 0 .707 –.707 0 0 0 0 1 0 0 0 0 1 .5 – .866 0 0 –1 0 1 0 0
0 0 0 .707 .707 0 0 0 –1 0 1 0 0 0 – .5 .866 .5 0 1 0 – .5 0 1 0
0 0 0 0 0 1 0 1 0 0 0 1 –.067 1.116 0 0 0 1 0 .5 0 0 0 1 Eq. I-9
Grid 1 Grid 2 Grid 3 Grid 4

Using the global transformation matrix [ D ] , the rigid-body mass matrix about the
reference point in the basic coordinate system [ M0 ] is obtained by Eq. I-10.

T
[ M0 ] = [ D ] [ MJJ ] [ D ] Eq. I-10
For the example, [ M0 ] is determined to be

9.25 – .933 0. 0. 2. – 2.967

– .933 10.75 0. – 3. 0. 6.058
[ M0 ] = 0. 0 20. 5. – 10. 0. Eq. I-11
0. –3 5. 8. – 2.5 – 1.5
2. 0 – 10. – 2.5 9.5 0.
– 2.967 6.058 0. – 1.5 0. 7.496

Comparing the results shown in Eq. I-11 to [ M0 ] generated by the GPWG module
(Figure I-2) shows the matrices to be numerically the same.
3. The next step is to inspect the [ M0 ] to determine whether the basic coordinate
system can be used as the principal mass directions. The principal mass axes are
axes that have no coupling between the translational mass components. For real
structures, there is no coupling in the translational mass terms in the inertia
matrix. However, with inconsistent scalar masses (CONM1, CMASSi, or DMIG),
you may define any type of mass matrix you desire.
To determine whether coupling exists between the translational mass terms, [ M0 ]
is partitioned into four 3 × 3 matrices as shown in Eq. I-12.

t tr
[ M0 ] = M M Eq. I-12
rt r
M M

where the superscripts t and r refer to translation and rotation, respectively.

For this example, the translational mass partition is given by
462

t
9.25 – .933 0
[M ] = – .933 10.75 0 Eq. I-13
0 0 20.

A check is made for coupling as follows:

t 2
δ = ∑ ( M ij ) i = j
Eq. I-14
t 2
ε = ∑ ( M ij ) i>j

If ε ⁄ δ is greater than .001, then excessive coupling exists preventing the basic
coordinate system from being used for the principal mass directions and User
.5
Warning Message 3042 is printed. For this problem, δ = ( 9.25 2 + 10.75 2 + 20 2 ) and
2 2 2 .5
ε = ( .933 + 0. + 0. ) . The ratio ε ⁄ δ = .038 agrees with the System Warning
Message 3042 shown in Figure I-2.
If needed, the principal mass directions are computed by performing an
t
eigensolution with the translational mass components. The eigenvectors of [ M ]
are the columns of the transformation matrix [ S ]

[ S ] = [ { e 1 }, { e 2 }, { e 3 } ] Eq. I-15

Using this eigenvector matrix, the partitions of the rigid-body mass matrix with
respect to the principal mass direction are computed by Eq. I-16.

t T t
[M ] = [S] [M ][S]

tr T tr
[M ] = [S ] [M ][ S ] Eq. I-16

r T r
[M ] = [S] [M ][S]

The [ S ] matrix for the example is given by

.432 .902 0.
[S] = – .902 .432 0. Eq. I-17
0. 0. 1.

The [ S ] matrix is printed after the [ M0 ] matrix as shown in Figure I-2. This
represents the transformation relating the basic coordinate system to the principal
mass axes. Again, if there is no coupling between the translational mass
component, which is the case for most problems, the eigensolution is not
required, and the [ S ] matrix is set equal to the identity matrix. This example was
selected to demonstrate all of the features of the GPWG module, but it is not a
typical problem.
4. The next step is to determine the principal masses and the center of gravity
location in the principal mass axes system as shown in Eq. I-18.
 CHAPTER I 463
Grid Point Weight Generator

t
M x = M 11

t
M y = M 22

t
M z = M 33

The “centers of gravity” are

M 11
tr
– M 13
tr
M 12
tr Eq. I-18
X x = ---------- , Y x = ------------- , Z x = ---------- ,
Mx Mx Mx

tr tr tr
M 23 M 22 – M 21
X y = ---------- , Y y = ---------- , Z y = ------------- ,
My My My

tr tr tr
– M 32 M 31 M 33
X z = ------------- , Y z = ---------- , Z z = ----------
Mz Mz Mz

As can be seen, the center of gravity location is not a unique location. The center
of gravity location is computed separately for the for x-, y-, and z-directions
relative to the principal mass axes. Only if the mass is the same in each direction
(which is typical) is there a unique center of gravity location, which is relative to
the reference point in the basic coordinate system.
For this example problem, the center of gravity locations are determined to be

Table I-2 Mass Center of Gravity Locations

Center of Gravity Location in

Mass Principal Mass Axes
Component
X Y L
x - mass .035 –0.602 0.251
y - mass –0.007 –0.044 0.248
z - mass –0.009 –0.559 0

The center of gravity location given in Table I-2 is the same as shown in
Figure I-2.
5. Following the center of gravity calculation is the calculation to determine the
moments of inertia for the center of gravity with respect to the principal mass axes
[ I ( S ) ] as shown in Eq. I-19.
464

(S) r 2 2
I 11 = M 11 – M y Z y – M z Y z

(S) (S) r
I 12 = I 21 = – M 12 – M z X z Y z

(S) (S) r
I 13 = I 31 = – M 13 – M y X y Z y
Eq. I-19
(S) r 2 2
I 22 = M 22 – Mz Xz – Mx Zx

(S) (S) r
I 32 = I 23 = – M 23 – M x Y x Z x

(S) r 2 2
I 33 = M 33 – M x Y x – M y X y

For the example, the inertia matrix [ I ( S ) ] is given by

4.377 – .877 .662

[I(S)] = – .877 5.623 – .342 Eq. I-20
.662 – .342 3.432

6. The final step is to compute the principal moments of inertia and the principal
directions of the momental ellipsoid (commonly referred to as the principal axes
in text books). An intermediate inertia matrix [ Ĩ ] is generated by reversing the
sign on the off diagonal terms of [ I ( S ) ] . For the inertia matrix given in Eq. I-20,
[ Ĩ ] is given by

4.377 .877 – .662

[ Ĩ ] = .877 5.623 .342 Eq. I-21
– .662 .342 3.432

An eigensolution is performed on the [ Ĩ ] matrix to determine the principal

directions. The resulting eigenvalues are the principal moments of inertia, which
are assumed to be the diagonal terms of the principal inertia matrix [ I ( Q ) ] . The
eigenvectors form the columns of the matrix [ Q ] , which is the transformation
relating the intermediate inertia matrix [ Ĩ ] to the principal inertia matrix [ I ( Q ) ] as
shown in Eq. I-22.

T
[ I ( Q ) ] = [ Q ] [ Ĩ ] [ Q ] Eq. I-22
For the example, [ Q ] and [ I ( Q ) ] are given by
 CHAPTER I 465
Grid Point Weight Generator

4.463 0. 0.
[I(Q) ] = 0. 6.076 0.
0. 0. 2.893
Eq. I-23
.720 .459 .521
[Q] = – .372 .889 – .269
– .586 0 .810

The matrices [ S ] and [ Q ] matrices are the coordinate rotation matrices, which
when taken together, relate the principal directions of the momental ellipsoid to
the basic coordinate system. The matrices given in Eq. I-23 are in agreement with
those given in Figure I-2. The example for the four mass model is now complete.
466
 NX Nastran Basic Dynamic Analysis User’s Guide

J
APPENDIX
Numerical Accuracy Considerations

■ Overview
468

J.1 Overview
MSC.Nastran is an advanced finite element analysis program that can solve models in
excess of 1,000,000 degrees-of-freedom. Because speed and accuracy are essential,
MSC.Nastran’s numerical analysis capabilities are continually enhanced to provide the
highest level of each.
This appendix provides a brief overview for detecting and avoiding numerical
ill-conditioning problems, especially as they relate to dynamic analysis. For more
information regarding MSC.Nastran’s numerical analysis algorithms, see the NX Nastran
Numerical Methods User’s Guide.

Linear Equation Solution

The basic statement of the linear equation solution is

[A]{x } = {b} Eq. J-1

where [ A ] is a square matrix of known coefficients (and usually symmetric for structural
models), { b } is a known vector, and { x } is the unknown vector to be determined.
The methods used for solution in MSC.Nastran are based on a decomposition of [ A ] to
triangular matrices and are followed by forward-backward substitution to get { x } . The
equations for this solution technique are

[ A ] = [L ][U ] Eq. J-2

where [ L ] is a lower-triangular matrix and [ U ] an upper-triangular matrix, and

[L]{y} = {b} Eq. J-3

where { y } is an intermediate vector. Eq. J-3 is called the forward pass because the solution
starts with the first row where there is only one unknown due to the triangular form of [ L ] .
The backward pass starts with the last row and provides the solution

[U ]{ x } = { y} Eq. J-4

Eigenvalue Analysis
The general eigensolution equation is

2
[ K + pB + p M ] { u } = 0 Eq. J-5

where p is the complex eigenvalue. This equation can always be transformed to a special
eigenvalue problem for a matrix [ A ]

[A – λ I]{φ} = 0 Eq. J-6

 CHAPTER J 469
Numerical Accuracy Considerations

where [ I ] is the identity matrix. Eq. J-6 is the basis of all the transformation methods of
MSC.Nastran (HOU, GIV, etc.). The iterative methods (INV, SINV) work directly from
Eq. J-5. The Lanczos method uses both. If [ A ] is a symmetric matrix, the eigenvectors are
orthogonal, and they can be normalized such that

T
[φ ][φ] = [I] Eq. J-7

where [ φ ] is a square matrix whose columns contain the eigenvectors { φ i } . With this
normalization convention, then

T
[φ ][A][φ] = [λ] Eq. J-8
and

T
[A] = [φ][λ][φ ] Eq. J-9

where [ λ ] is the eigenvalue diagonal matrix.

Matrix Conditioning
Reordering the previous equations, any matrix [ A ] can be expressed as a sum of its
eigenvalues multiplied by dyadic eigenvector products

T T  T
[ A ] = λ1 { φ1 } { φ1 } + λ 2 { φ 2 } { φ2 } + … + λ n { φn }  φ n  Eq. J-10
 
 
Defining [ Bi ] = { φi }  φ Ti  , which is a full rank 1 matrix, then
 
n
[A] = ∑ λk [ Bk ] Eq. J-11
k = 1

where n is the dimension of [ A ] . On the average, an element of [ Bk ] has the same

magnitude as the corresponding element of [ Bk + 1 ] . Let B max be the magnitude of the
largest coefficient of all [ B k ] matrices. Then

a ij ≤ B max ( λ 1 + λ 2 + … + λ n ) Eq. J-12

This equation shows that the terms of [ A ] are dominated by the largest eigenvalues.
Unfortunately, the smallest eigenvalues are those of greatest interest for structural models.
These small eigenvalues must be calculated by taking the differences of coefficients that are
dominated by the largest eigenvalue. For this reason, the ratio λ n ⁄ λ 1 is called a numerical
conditioning number. If this number is too large, numerical truncation causes a loss of
accuracy when computing the lowest eigenvalues of a system.
470

The assumptions that allow this simple analysis are often pessimistic in practice, that is, the
bounds predicted by the error analysis are conservative. However, the effects it predicts
do occur eventually so that models that produce acceptable results for one mesh size may
produce unacceptable results with a finer mesh size due to the higher eigenvalues included
in the larger-sized matrices occurring from the finer mesh.

Definiteness of Matrices
A matrix whose eigenvalues are all greater than zero is said to be positive definite. If some
eigenvalues are zero but none are less than zero, the matrix is positive semi-definite. A
stiffness matrix assembled from elements is at least positive semi-definite. If all of the
structure’s rigid-body modes are constrained, the stiffness matrix is positive definite.
Another category is the indefinite matrices category. These matrices have zeroes or blocks
of zeroes on the diagonal.

Although definiteness is most concisely defined in terms of eigenvalues, it is not a practical

test for large matrices because of the computational cost of extracting all of the eigenvalues.
However, other operations, particularly linear equation solution and dynamic reduction,
may detect nonpositive definite matrices and provide diagnostics using these terms as
described later in this appendix.

Numerical Accuracy Issues

The numerical operations of MSC.Nastran are executed in a finite 64-bit floating point
arithmetic. Depending on the specific computer’s word structure (number of bits for
mantissa versus exponent), different roundoff errors may occur. The sophistication level
of the actual hardware (or software) arithmetic units also has an influence on the numerical
accuracy. To attain the most numerical accuracy possible, the following strategies are used
in MSC.Nastran.
In the decomposition of positive definite matrices, the Gaussian elimination process
(which is the basis of all decomposition algorithms in MSC.Nastran) does not require
numerical pivots. Since some of the matrices are not positive definite, sparse
decomposition (both symmetric and unsymmetric) employs a numerical pivoting
(row/column interchange) algorithm. These methods consider a pivot term suitable if

THRESH
a kk ≥ 10 ⋅ MAX a ki Eq. J-13
i

which means that a diagonal term is accepted as a pivot if it is greater than the maximum
term in that row multiplied by 10 THRESH (the default for THRESH is –6).

To ensure numerical accuracy in eigenvalue calculations, most MSC.Nastran methods use

a spectral transformation of

1
µ = ---------------- Eq. J-14
λ – λs

where λ s is an eigenvalue shift.

 CHAPTER J 471
Numerical Accuracy Considerations

This transformation is shown in Figure J-1.

µ3

µ2
µ1
λ
λS λ1 λ2 λ3

Figure J-1 Spectral Transformation

The spectral transformation ensures uniform accuracy throughout the frequency by

“shifting” to the area of interest. Another effect of this transformation is the welcomed µ
space separation of closely-spaced eigenvalues. When λ s is close to an eigenvalue λ i , the
decomposition of the “shifted matrix”

T
[ K – λs M ] = [ L ] [ D ] [ L ] Eq. J-15

may produce high MAXRATIO messages. Automatic logic to perturb the λ s value in this
case is implemented in MSC.Nastran.

Sources of Mechanisms
In all of the decomposition methods, a null row or column in [ A ] causes a fatal error
message. All other causes of singularity are not distinguishable from near-singularity
because of the effects of numerical truncation. Only warning messages are usually
provided for these cases.

In standard decomposition

T
[ A ] = [L ][D ][ L ] Eq. J-16

the process starts to compute the first term of [ D ] with the first internal degree of freedom
and then processes each additional degree of freedom and its associated terms. It can be
shown that when processing the k-th row, the k-th row and all rows above it are in effect
free, and all rows below it are constrained to ground. The term of [ D ] at the k-th row is
proportional to the amount of coupling between that degree of freedom and the
degree-of-freedom with a higher value of k. If the terms of the k-th row and above are not
connected to the remaining rows, the k-th term of [ D ] goes to zero. Because of numerical
truncation, the term may be a small positive or negative number instead. If a term of [ D ]
is calculated to be identically zero, it is reset to a small number because of the
472

indeterminacy of its calculation. The existence of such a small term defines a mechanism
in static analysis in MSC.Nastran. A mechanism is a group of degrees-of-freedom that may
move independently from the rest of the structure as a rigid body without causing internal
loads in the structure. A hinged door, for example, is a mechanism with one rigid-body
freedom. If the hinges are disconnected, the door mechanism has six rigid-body freedoms.

Mechanisms are characterized by nondimensional numbers derived by dividing the terms

of [ D ] into the corresponding diagonal term of [ A ] . If these “matrix diagonal to factor
diagonal” ratios are large numbers, various warning and fatal messages are produced,
depending on the context.

Sources of Nonpositive Definite Matrices

A negative semidefinite element stiffness matrix, which is defined as one whose
eigenvalues are all negative or zero, implies that the element has an energy source within
it. MSC.Nastran discourages using such elements by giving fatal messages when negative
element thicknesses or section properties are input. However, there are applications where
such elements are useful, such as when using an element with negative stiffness in parallel
with a passive element to model a damaged or thinned-down element. For this reason,
negative material property coefficients and negative stiffnesses for scalar elements are
allowed. Also, some incorrect modeling techniques, such as the misuse of the
membrane-bending coupling term on PSHELL entries (MID4), can lead to negative
eigenvalues of the element stiffness matrix.

Stiffness matrices with negative eigenvalues cause negative terms in [ D ] . The number of
such terms is automatically output by the standard symmetric decomposition subroutine.
Their existence again causes various warning and fatal messages, depending on the
context. The most common cause of negative terms in [ D ] is true mechanisms, whose
terms are small negative numbers and are actually computational zeroes.

Indefinite matrices occur when using the Lagrange multiplier method for constraint
processing. In this method, the constraints are not eliminated from but are concatenated
to the g-size system of equations. The constraint equations have zeroes on the diagonal,
resulting in indefinite matrices. These systems can be solved using the block pivoting
scheme of the sparse decomposition.

Detection and Avoidance of Numerical Problems

Static Analysis
Models used in static analysis must be constrained to ground in at least a statically
determinate manner even for unloaded directions. For example, a model intended for only
gravity loading must be constrained in horizontal directions as well as vertical directions.
The evidence of unconstrained directions is that the entire model is a mechanism, that is,
the large ratio occurs at the last grid point in the internal sequence.
Another source of high ratios arises from connecting soft elements to stiff elements. Local
stiffness is a function of element thickness (moment per unit rotation) through element
thickness cubed (force per unit deflection) and is inversely proportional to mesh spacing,
 CHAPTER J 473
Numerical Accuracy Considerations

again in linear through cubic ratios. Some relief is possible by sequencing the soft
degrees-of-freedom first in the internal sequence, although this is difficult to control in the
presence of automatic resequencing. More reliable corrections are to replace the very stiff
elements with rigid elements or to place the soft and stiff elements in different
superelements.

A third source of high ratios is the elements omitted through oversight. The corrective
action here is to start with the grid points listed in the diagnostics and track back through
the elements connected to them through the upstream grid points. The missing elements
may be anywhere upstream. PARAM,GPECT,1 output and undeformed structure plots all
provide useful data for detecting missing elements.
At present, there are two major methods of identifying large ratios and
nonpositive-definite matrices. In some solutions, the largest matrix diagonal to factor
5
diagonal ratio greater than 10 (MAXRATIO default) is identified by its internal sequence
number, and the number of negative factor diagonal terms is output. The best method to
identify mechanisms here is to apply checkout loads that cause internal loads in all of the
elements. Then inspect the displacement output for groups of grid points that move
together with implausibly large displacements and common values of grid point rotation.
The only condition that causes a fatal error is a true null column, and NASTRAN
SYSTEM(69)=16 avoids this fatal error by placing a unit spring coefficient on the
degrees-of-freedom with null columns. This option is recommended only for diagnostic
runs because it may mask modeling errors.
In other solution sequences, all ratios greater than 10 5 are printed in a matrix format named
the MECH table. The external sequence number of each large ratio is also printed, which
is the grid point and degree-of-freedom number. If any such ratios exist, the action taken
depends on the value of PARAM,BAILOUT. In the conventional solution sequences, its
default value causes the program to continue after printing the MECH matrix. In the
superelement solution sequences, a different default causes a fatal error exit after printing
the MECH matrix. For both types of solution sequences, the opposite action may be
requested by setting the value of PARAM,BAILOUT explicitly. Also, the criterion used for
identifying large ratios may be changed using PARAM,MAXRATIO.

For static analysis, values between 103 and 106 are almost always acceptable. Values
between 107 and 10 8 are questionable. When investigating structures after finding these
values, some types of structures may be found to be properly modeled. It is still
worthwhile to investigate the structures with questionable values.
The solutions with differential stiffness effects offer another method to obtain nonpositive
definite stiffness matrices. For example, a column undergoing compressive gravity
loading has a potential energy source in the gravity load. A lateral load that is stabilizing
in the absence of gravity (i.e., a decreased load causes a decreased deflection) is
destabilizing when applied in a postbuckled state.
474

Eigensolutions Using the Inverse Iteration and Lanczos Methods

The matrix [ K – λ s M ] is decomposed where λ s is an eigenvalue shift and [ K ] and [ M ] are
the stiffness and mass matrices, respectively. This condition allows the solution of models
with rigid-body modes since the singularities in [ K ] are suppressed if there are
compensating terms in [ M ] . The only conditions that should cause fatal messages due to
singularity are

1. The same null column in both [ K ] and [ M ] . These columns and rows are given
an uncoupled unit stiffness by the auto-omit operation if the default value (0) for
PARAM,ASING is used. If this value is set to –1, a null column in both matrices
is regarded as an undefined degree of freedom and causes a fatal error exit.
2. A massless mechanism. One commonly encountered example is a straight-line,
pinned rod structure made from CBAR elements with no torsional stiffness (J)
defined. The structural mass generated for CBAR elements does not include
inertia for the torsion degrees-of-freedom. The natural frequency for this torsion
mode approaches the limit of zero torsional stiffness divided by zero mass, which
is an undefined quantity. If the elements lie along a global coordinate axis, the
mass term is identically zero, which leads to very large negative or positive
eigenvalues and is usually beyond any reasonable search region. If the elements
are skewed from the global axes, the eigenvalues may be computed at any value
(including negative) because of the indeterminacy caused by numerical
truncation.
The “negative terms on factor diagonal” message generally occurs for every decomposition
performed in the iteration. It can be shown from Sturm sequence theory that the number
of negative terms is exactly equal to the number of eigenvalues below λ s . This condition is
a means of determining if all roots in the range have truly been found.

Eigensolutions Using Transformation Methods

If the GIV or HOU methods (and their variations) are selected, one of the first operations
performed is a Choleski decomposition, which is used to reduce the problem to a special
eigenvalue problem. The mass matrix is given this operation for the straight GIV or HOU
method. Columns that are identically null are eliminated by the auto-omit operation. Poor
conditioning can result from several sources. One example is a point mass input on a
C0NM2 entry that uses offsets in three directions. The grid point to which it is attached has
nonzero mass coefficients for all six degrees-of-freedom. However, only three
independent mass degrees-of-freedom exist, not six. Another example results in a
superelement analysis when most of the elements in a superelement do not have mass and
all interior masses are restricted to only a few structural or mass elements. The boundary
matrix produced for the superelement is generally full, no matter what its rank, i.e.,
regardless of the number of independent mass degrees-of-freedom that it contains.
The presence of a rank-deficient mass matrix when using the GIV or HOU method
produces fatal messages due to the singularity of the mass matrix or produces solutions
with poor numerical stability. Poor stability is most commonly detected when making
small changes to a model and then observing large changes in the solution. Either the
 CHAPTER J 475
Numerical Accuracy Considerations

MGIV and MHOU methods or the AGIV and AHOU methods are the preferred methods
to use when nearly singular mass matrices are expected because these methods decompose
the matrix [ K – λ s M ] instead of the mass matrix. The shift parameter λ s is automatically set
to be large enough to control rigid-body modes. Better modeling practices also reduce the
costs and increase the reliability for the two examples cited above. If an offset point mass
is significant in any mode, it is better to attach it to an extra grid point at its center of gravity
and model the offset with a rigid element. The auto-omit feature then eliminates the
rotational degrees-of-freedom. Similarly, if only a few interior points of a superelement
have mass, it may be more economical to convert them to exterior points, which also
eliminates the singular boundary mass matrix.

It is possible to input negative terms into a mass matrix directly with DMIG terms or scalar
mass elements. This class of problem causes fatal errors due to nonpositive definite mass
matrices for the transformation methods, fatal errors with the Lanczos method, and wrong
answers with the INV method. The complex eigenvalue methods should be used for this
type of problem, which infers that modal or dynamic reduction methods may not be used.

A similar but quite different problem arises because Cholesky decomposition is used on
the generalized mass matrix (named MI in the diagnostics) when orthogonalizing the
eigenvectors with respect to the mass matrix. The existence of negative terms here
indicates that poor eigenvectors were computed. The row number where the negative
term occurs is printed in the diagnostic. This row number does not refer to a physical
degree of freedom but refers instead to an eigenvector number. The usual cause is
computing eigenvectors for computational infinite roots in the modified transformation
methods. This problem can be avoided after the fact by setting ND on the EIGR entry to a
value less than the row number that appears in the diagnostics or before the fact by setting
F2 on the EIGR entry to a realistic value instead.

Frequency Response Analysis

Negative terms on the factor and high factor to matrix diagonal ratios can be expected
when using coupled methods and may often be safely ignored. These messages are merely
an indication that the excitation frequency is above one or some of the natural frequencies
of the system (negative terms), or is near a natural frequency (high ratio).

Transient Response Analysis

A well-conditioned model should have neither negative terms nor high ratios on its factor
terms. The causes of such messages include all of the effects described above. Negative
mass terms can be detected by rapidly diverging oscillations in the transient solution.

The Large Mass Method

One of the methods for enforcing motion is the large mass method. It may be used with
direct methods as well as all of the reduction methods. The basis of the method is to attach
artificial masses to the structure at the degrees-of-freedom where the motion is to be
enforced. These large masses should be orders of magnitude larger than the total mass or
476

the moment of inertia of the structure. A survey of the literature shows recommendations
for mass ratios ranging from 103 to 10 8 with a value of 10 6 as the most common
recommendation.

The mass ratio affects both the accuracy and numerical conditioning, and must be adjusted
in a compromise that meets both criteria. With regard to load accuracy, the error in the
approximation is inversely proportional to the ratio. A ratio of 103 causes an error of ten
percent at resonance for a mode with one-half percent of damping, which represents an
extreme case. Off-resonance excitation or higher damping ratios result in lower errors.
Numerical conditioning problems are much more difficult to predict.

The Lagrange Multiplier Method

Another method to enforce motion is the Lagrange multiplier method. In this method, the
input motion function is described by a constraint equation. This method provides better
accuracy than the large mass method where the numerical error introduced is proportional
to the large mass.
The formulation introduces an indefinite system of linear equations where some numerical
problems may arise from the fact that the system matrix contains terms that have
dimensions of stiffness as well as nondimensional (constraint) terms. The decomposition
of this matrix with the sparse decomposition methods pivoting strategy is stable.
You may control the pivoting strategy with the THRESH DMAP parameter mentioned
earlier. The default value (–6) is adequate for most Lagrange multiplier solutions. In fact,
higher (ranging up to –2) values provide better accuracy, while increasing the number of
pivots may result in performance degradation.
 NX Nastran Basic Dynamic Analysis User’s Guide

K
APPENDIX
Diagnostic Messages for Dynamic
Analysis

■ Overview
478

K.1 Overview
This appendix lists common diagnostic messages for dynamic analysis. The text for each
message is given in uppercase letters and is followed by additional explanatory material,
including suggestions for remedial action. These and other messages are listed and
described in the MSC.Nastran Reference Manual.
The messages in this section have the following format:

*** (SYSTEM/USER) (FATAL/WARNING/INFORMATION) MESSAGE ID, text

where “ID” is a unique message identification number and “text” is the message as
indicated in capital letters for each of the diagnostic messages. Four asterisks (****) in the
message text indicates information that is filled in for a specific use of the message, such as
the number of a grid point or the name of a Bulk Data entry. Some of the messages are
followed by additional explanatory material, including suggestions for remedial action.

Fatal messages cause the termination of the execution following the printing of the
message text. These messages always appear at the end of the MSC.Nastran output.
Warning and information messages appear at various places in the output stream. Such
messages only convey warnings or information to the user. Consequently, the execution
continues in a normal manner following the printing of the message text.
 CHAPTER K 479
Diagnostic Messages for Dynamic Analysis

As an example, consider message number 2025, which appears in the printed output as
follows:
*** USER FATAL MESSAGE 2025, UNDEFINED COORDINATE SYSTEM 102

The three leading asterisks (***) are always present in numbered user and system
diagnostic messages. The word USER indicates that this is a user problem rather than a
system problem. The word FATAL indicates that this is a fatal message rather than a
warning or information message. The number 2025 is the identification number for this
message. The text of the message follows the comma. The number 102 replaces the
asterisks (****) in the general message text and indicates that 102 is the identification
number of the undefined coordinate system. The abbreviation UFM refers to User Fatal
Message, UWM refers to User Warning Message, UIM refers to User Information Message,
and SFM refers to System Fatal Message.
480

UFM 2066 *** USER FATAL MESSAGE 2066, UNDEFINED GRID POINT **** ON DAREA CARD.

A dynamic loading entry references an undefined grid point.

UFM 2069 *** USER FATAL MESSAGE 2069, UNDEFINED GRID POINT **** IN TRANSIENT
INITIAL CONDITION SET ****.
An attempt has been made to specify initial conditions for an undefined grid point. All
degrees-of-freedom with initial conditions must be in the analysis set.

UFM 2071 *** USER FATAL MESSAGE 2071, DYNAMIC LOAD SET **** REFERENCES
UNDEFINED TABLE ****.

A referenced dynamic load table was not present in the Bulk Data.

UFM 2074 *** USER FATAL MESSAGE 2074, UNDEFINED TRANSFER FUNCTION SET ****.

A transfer function set was selected in the Case Control but was not present in the Bulk
Data.

UFM 2079 *** USER FATAL MESSAGE 2079, GRID OR SCALAR POINT **** HAS AN UNDEFINED
COORDINATE REFERENCED ON A DAREA, DELAY, DPHASE CARD.
The “C” or component value for scalar-type points must be zero or one.

UFM 2088 *** USER FATAL MESSAGE 2088, DUPLICATE TABLE ID ****.
All tables must have unique numbers. Check for uniqueness.

UFM 2101A *** USER FATAL MESSAGE 2101A, GRID POINT **** COMPONENT *** ILLEGALLY
DEFINED IN SETS ****.
The above grid point and component are defined in each of the above independent subsets.
A point may belong to a maximum of one independent subset. This error occurs when a
DOF is defined as belonging to two mutually exclusive sets (see NX Nastran Reference
Manual).

A common example of this occurs when a DOF is defined as dependent on an MPC (M-set)
as well as being constrained (s-set) on an SPC entry. The message for this states that the
component is illegally defined in the um (user-defined m-set) and us (user-defined s-set)
sets. These two sets are mutually exclusive because all MPC equations are processed
before the SPCs are applied (the exception is SOL 24) and the m-set DOFs are removed
from the matrix. When the program attempts to apply the SPC, the DOF is no longer
available, and the fatal message is issued. The normal correction for this is to modify the
MPC so that the DOF in question is independent (n-set). Then there is no conflict.

UFM 2107 *** USER FATAL MESSAGE 2107, EIGR-CARD FROM SET **** REFERENCES
DEPENDENT COORDINATE OR GRID POINT ****.

When the point option is used on an EIGR entry, the referenced point and component must
be in the analysis set (a-set) for use in normalization.
 CHAPTER K 481
Diagnostic Messages for Dynamic Analysis

UFM 2109 *** USER FATAL MESSAGE 2109, NO GRID, SCALAR OR EXTRA POINTS DEFINED.

Dynamics problems must have at least one grid, scalar, or extra point.

UFM 2133 *** USER FATAL MESSAGE 2133, INITIAL CONDITION IN SET **** SPECIFIED FOR
POINT NOT IN ANALYSIS SET.
Initial conditions can only be specified for analysis set points. Therefore the
point/component mentioned on TIC entries must belong to the analysis set.

UFM 2135 *** USER FATAL MESSAGE 2135, DLOAD CARD *** HAS A DUPLICATE SET ID FOR
SET ID ***.

The Li Set IDs on a DLOAD entry are not unique. See the DLOAD Bulk Data description.

UFM 2136 *** USER FATAL MESSAGE 2136, DUPLICATE DLOAD, RLOAD, OR TLOAD SET ID
NUMBER = ****** HAS BEEN ENCOUNTERED FOR DLOAD SET = ******.
Dynamic loads may not be defined by giving multiple data entries with the same ID. Use
unique IDs.

UIM 2141 *** USER INFORMATION MESSAGE 2141, TIME ESTIMATE IS **** SECONDS.
PROBLEM SIZE IS ****, SPILL WILL OCCUR FOR THIS CORE AT A PROBLEM SIZE OF
****.
The time estimate includes the time of the tridiagonalization and eigenvalue calculation
when the GIV or HOU methods are used. If ND is given on the EIGR entry, it also includes
the time of the eigenvector generation. If F1 and F2 are used instead, the eigenvector times
are not estimated. This condition can underestimate the time when the range includes
many eigenvectors.

UFM 2200 *** USER FATAL MESSAGE 2200, INCONSISTENT RIGID BODY SYSTEM.
This message occurs if a SUPORT is used and the rigid-body mass matrix is not positive
definite. Possible causes are unconstrained mechanisms or input of negative mass terms.
A diagnostic method is to remove all SUPORT entries and inspect the resulting
eigenvectors for implausible behavior.

UFM 3031 *** USER FATAL MESSAGE 3031, UNABLE TO FIND SELECTED SET (****) IN TABLE
(****) IN SUBROUTINE (****).

A particular set used in the problem was not included in the data. Good examples are
loads, initial conditions, or frequency sets. Include the required data or change the Case
Control commands to select data already in the problem. Set zero (0) has a special
meaning. A set selection was required, but none was made. For example, no METHOD
was selected for an eigenvalue extraction problem, or no FREQ was selected for frequency
response.

This message can also indicate that a LOAD entry references another LOAD entry, which
is not permitted.

This message can also occur if a DLOAD entry references a nonexisting LOAD entry, e.g.,
RLOAD1.
482

UWM 3034 *** USER WARNING MESSAGE 3034, ORTHOGONALITY TEST FAILED. LARGEST
TERM = ****, NUMBER FAILED = ****, PAIR = ****, ****, EPSILON = ****.
This indicates that eigenvector accuracy is in doubt. This message is printed only when the
off-diagonal terms of the modal mass matrix are larger than 1.0E-10. The eigenvectors are
not orthogonal to this extent. This nonorthogonality is especially important if a modal
formulation is used. The pair of eigenvectors listed exhibit the worst behavior. The
number failed is the number of pairs above the criteria. You can improve the numerical
conditioning of the problem by reducing the range of mass ratios, stiffness ratios, and
eigenvalue range.

UIM 3035 *** USER INFORMATION MESSAGE 3035, FOR DATA BLOCKS ****. SUPORT PT. NO.
EPSILON STRAIN ENERGY EPSILONS LARGER THAN .001 ARE FLAGGED WITH
ASTERISKS.

One line of output is printed for each component on a SUPORT entry. Large values of
either EPSILON or STRAIN ENERGY indicate errors in the constraints, MPCs, or offsets.

UWM 3045 *** USER WARNING MESSAGE 3045, INSUFFICIENT TIME TO COMPLETE THE
REMAINING ** SOLUTION(S) IN MODULE ***.

The estimated time for completion of the module is less than the time remaining, as
specified on the Executive Control TIME statement. The module computes one solution
(for example, one excitation frequency in frequency response analysis) and then processes
all output requests. The remaining frequencies can be obtained on restart by adding or
changing a FREQ command.

UFM 3046 *** USER FATAL MESSAGE 3046, YOUR SELECTED LOADING CONDITION, INITIAL
CONDITION, AND NONLINEAR FORCES ARE NULL. A ZERO SOLUTION WILL
RESULT.
Transient solutions must have one of the above nonzero loading condition, initial
condition, or nonlinear forces. Also, make sure that LOADSET is spelled correctly.

UFM 3047 *** USER FATAL MESSAGE 3047, NO MODES WITHIN RANGE AND LMODES=0. A
MODAL FORMULATION CANNOT BE MADE.

The modes used for a modal formulation must be selected by a PARAM entry. Set LFREQ,
HFREQ, or LMODES to request modes.

UFM 3051 *** USER FATAL MESSAGE 3051, INITIAL CONDITION SET **** WAS SELECTED FOR
A MODAL TRANSIENT PROBLEM. INITIAL CONDITIONS ARE NOT ALLOWED IN
SUCH A PROBLEM.

The IC command is not allowed for modal transient response problems.

UWM 3053 *** USER WARNING MESSAGE 3053, THE ACCURACY OF EIGENVALUE **** IS IN
DOUBT. GIV/HOU QR FAILED TO CONVERGE IN **** ITERATIONS.
Each eigenvalue is computed to the precision limits of each machine consistent with the
maximum number of iterations allowed.
 CHAPTER K 483
Diagnostic Messages for Dynamic Analysis

UFM 3057 *** USER FATAL MESSAGE 3057, MATRIX **** IS NOT POSITIVE DEFINITE.

A Cholesky decomposition was attempted on the above matrix, but a diagonal term of the
factor was imaginary or equal to zero such that the decomposition failed. This message is
from the regular (as opposed to sparse) decomposition method.

This message may be produced because of constraint problems. Check the output for
UWM 4698 for large factor diagonal ratios and constrain appropriately.

UWM 4193 *** USER WARNING MESSAGE 4193, A GRID AND COMPONENT SPECIFICATION
ON A (DPHASE/DELAY) SID = **, DOES NOT APPEAR ON A DAREA CARD.

The area specification is set to zero.

SFM 4276 *** SYSTEM FATAL MESSAGE 4276, **** ERROR CODE ****
This message occurs when MSC.Nastran encounters errors that are not otherwise trapped
(including system errors). There are various error codes (EC), each of which has a different
meaning.

In nearly every case, the log file contains further information about the problem, so you
should look there for further clues. In addition, because these are errors that most likely
should be caught in another manner (and with a more explicit error message), you should
look at the recent Error Reports (search for “4276").
In many cases, increasing memory, BUFFSIZE, or disk space resolves the problem,
especially when attempting to run large models on workstations.
This error often indicates machine underflow or overflow, although it is impossible to list
specific reasons for this error code. Observed problems include:
1. Modeling problems. This condition is usually accompanied by an “arithmetic
fault, floating overflow” type message in the log file on some machines. Perform
a static analysis and verify that the value for the maximum factor diagonal ratio
is acceptable.
2. When SFM 4276 is followed by an “access violation”, then it is often due to a lack
of memory or disk space or due to a coding error in MSC.Nastran.

UWM 4312 *** USER WARNING MESSAGE 4312, CONM2 **** HAS NONPOSITIVE-DEFINITE
INERTIA MATRIX.
Most dynamic analysis methods require positive-definite mass matrices. Inserting inertia
data into the wrong fields can result in nonpositive-definite systems, which are not
physically realistic for normal modeling practices. Reduce the size of the off-diagonal
terms to provide a positive-definite determinant.

UFM 4346 *** USER FATAL MESSAGE 4346, FREQUENCY RESPONSE SET, ID = *** IS
UNDEFINED.
Define the set of frequencies to be used for the analysis.
484

UFM 4391 *** USER FATAL MESSAGE 4391, NONUNIQUE DAREA SET ***** HAS BEEN
SPECIFIED FOR LSEQ DEFINED VECTOR *****.
Each LSEQ Bulk Data entry must define a unique DAREA set specification.

UFM 4392 *** USER FATAL MESSAGE 4392, CONTINUATION CARD ERRORS. EXPLANATIONS
FOLLOW LIST OF CARDS IN ERROR.

ERROR
NUMBER .1..2..3...
*** Input echo

EXPLANATION OF ERROR CODES ABOVE FOLLOWS...

1. FIELD 1 IS NOT UNIQUE.
2. MORE THAN ONE CARD HAS FIELD 10 WHICH IS THE SAME AS FIELD 1 OF
THIS CARD.
3. CARD IS AN ORPHAN (I.E., NO PARENT CARD EXISTS).

Continuation mnemonics in field 10 of a parent entry and field 1 of its continuation entry
must be unique. Each continuation entry must have a parent entry. Check all continuation
mnemonics.

UFM 4405 *** USER FATAL MESSAGE 4405, NO EIGENVECTORS COMPUTED FOR
COMPONENT MODE SYNTHESIS OR SYSTEM SOLUTION.
The eigenvectors computed in component mode synthesis (CMS) are used to approximate
the motion of the component. Some vectors must be present to perform this reduction. For
the system solution, an exit is taken if eigenvalues are requested but not eigenvectors.
Reset the “range of frequency” and/or the “number desired” on the EIGR or EIGRL entry.
If the superelement does not have any eigenvalues in the range of interest, remove the CMS
request for that particular superelement.
This message is also issued when the eigenvectors calculated with a transformation
method such as GIV do not pass internal orthogonality checks. This is indicative of a
modeling error.
This message can also be issued if insufficient memory is available for the Lanczos method
with sparse decomposition.
This could occur with UFM 5401 and be related to UWM 5411.

UFM 4407 *** USER FATAL MESSAGE 4407, MR MATRIX HAS NULL DIAGONAL TERM.

The MR matrix contains the rigid-body mass matrix of the structure as measured at the
degrees-of-freedom listed on the SUPORT entry. If any of these degrees-of-freedom have
null mass, they result in invalid eigenvectors. Specify enough masses to define all
rigid-body modes and check attachments between the SUPORT and the rest of the model.
 CHAPTER K 485
Diagnostic Messages for Dynamic Analysis

UIM 4415 *** USER INFORMATION MESSAGE 4415, THE FOLLOWING A-SET
DEGREES-OF-FREEDOM HAVE EITHER NULL MASSES OR NULL MASSES AND
STIFFNESSES.

If the listed degrees-of-freedom have null mass for the GIV, HOU, MHOU, or MGIV
methods, they are automatically omitted. For the INV method, the null
degrees-of-freedom are constrained. For direct frequency or direct transient response, the
null degrees-of-freedom are given a very small mass or stiffness. Inspect the listed
degrees-of-freedom to ensure that masses or stiffnesses are not left out inadvertently.

UFM 4416 *** USER FATAL MESSAGE 4416, NO DYNAMIC LOAD TABLE AVAILABLE.

A frequency response or transient response analysis was requested, but no dynamic load
data is available. Include dynamic load data (DLOAD, RLOADi, TLOADi) in model.

UFM 4417 *** USER FATAL MESSAGE 4417, NO TRANSIENT RESPONSE LIST AVAILABLE.

A transient response dynamic analysis was requested, but no transient response list is
available. Include a TSTEP entry in the Bulk Data Section.

UFM 4418 *** USER FATAL MESSAGE 4418, NO EIGENVALUE EXTRACTION DATA IS
AVAILABLE.

A dynamic analysis was requested, but no eigenvalue extraction data was available.
Include eigenvalue extraction data EIGR or EIGRL in the Bulk Data.
Possible causes are
• The METHOD command in the Case Control Section but no EIGR or EIGRL entry
in the Bulk Data Section.
• The METHOD command in the Case Control Section, EIGRL entry in Bulk Data
Section but no RF3D83 (SOL 3 only).
• No correspondence between Set IDs on the METHOD command and the
EIGR/EIGRL entries.

UWM 4420 *** USER WARNING MESSAGE 4420, THE FOLLOWING DEGREES-OF-FREEDOM ARE
POTENTIALLY SINGULAR.

During decomposition, the degrees-of-freedom listed had pivot ratios greater than
MAXRATIO. Verify that the degrees-of-freedom are not part of a mechanism and that
elements do not have excessive stiffness. In SOLs 61 and higher, this condition causes run
termination. PARAM,BAILOUT may be used to continue the run. See the NX Nastran
Numerical Methods User’s Guide.

UFM 4421 *** USER FATAL MESSAGE 4421, NO FREQUENCY RESPONSE LIST AVAILABLE.
A frequency response dynamic analysis was requested, but no frequency data is available.
Include frequency data (FREQ, FREQ1, FREQ2) in the Bulk Data.
486

UFM 4501 *** USER FATAL MESSAGE 4501, RLOADi CARD SELECTED IN TRANSIENT
ANALYSIS. USE TLOADi.
RLOADi entries are used in frequency response analysis. These entries have no meaning
in transient analysis. Replace RLOADi with TLOADi entries.

UWM 4561 *** USER WARNING MESSAGE 4561, INSUFFICIENT MEMORY FOR MODE
ORTHOGONALITY CHECKS.

The amount of memory needed for eigenvector orthogonalization is 1/2 · [number of

eigenvectors · (number of eigenvectors +1)] + 2 · BUFFSIZE + number of eigenvalues. If
this equation is not met, the modes are orthogonalized, but the checking function is not
performed. However, all outputs from the module are provided. If the check is desired,
you should either increase memory or decrease the number of eigenvectors to satisfy the
above equation.

UFM 4562 *** USER FATAL MESSAGE 4562, TSTEP (TIME STEPS) DATA IS MISSING.
Transient analysis requires the time step data. Add a TSTEP Bulk Data entry, and select it
with a TSTEP Case Control command.

UWM 4582 *** USER WARNING MESSAGE 4582, LSEQ CARD SID = *** REFERS TO A
NONEXISTENT STATIC LOAD MATRIX COLUMN *** (NCOLS = ***).
The most likely cause occurs when changing an LSEQ entry on a restart without
regenerating and assembling the static load matrix.

UFM 4603 *** USER FATAL MESSAGE 4603, THE LSEQ SET ID **** IS NOT UNIQUE WITH
RESPECT TO OTHER STATIC LOAD IDS.
LSEQ set IDs must be unique with respect to all other static load set IDs.

UFM 4645 *** USER FATAL MESSAGE 4645, THE SHIFTED STIFFNESS MATRIX IS NOT POSITIVE
DEFINITE.
The matrix sum K + λ ⋅ M is decomposed by the Cholesky method at the start of the MGIV
method of eigensolution. This decomposition requires that the matrix be positive-definite.
A condition that prevents this is a massless mechanism (for example, a point mass on an
offset with no rotational stiffness).

UFM 4646 *** USER FATAL MESSAGE 4646, THE MASS MATRIX IS NOT POSITIVE DEFINITE,
USING THE GIV/HOU METHOD. USE MGIV/MHOU INSTEAD.

The reduced mass matrix has columns that are not linearly independent. Common causes
are rotation degrees-of-freedom whose only inertia terms result from point masses on
offsets. Use the MGIV or MHOU method instead since it does not require a positive
definite mass matrix.
 CHAPTER K 487
Diagnostic Messages for Dynamic Analysis

UFM 4647 *** USER FATAL MESSAGE 4647, INSUFFICIENT TIME TO COMPLETE *****.

CPU ESTIMATE = **** SEC

CPU REMAINING = **** SEC
I/O ESTIMATE = **** SEC
I/O REMAINING = **** SEC

CPU and I/O limits are supplied on the Executive Control statement TIME (in minutes).
The module where the terminated program is listed. If the time to completion appears
reasonable, you should increase the estimates on the TIME statement and resubmit the run.
For large models, an increase in the system memory request should also be considered.

UWM 4648 *** USER WARNING MESSAGE 4648, THE MODAL MASS MATRIX IS NOT POSITIVE
DEFINITE.
The modal mass matrix cannot be decomposed by the Cholesky algorithm after merging
elastic and free-body modes. (Cholesky decomposition is used to orthogonalize the
eigenvectors with respect to the mass matrix.) The causes include the input of negative
masses and the calculation of eigenvectors for eigenvalues approaching machine infinity.
Inspect the model or ask for fewer eigenvectors using the F2 option. When this condition
occurs, the eigenvectors are not orthogonalized or normalized. The second parameter of
the READ or REIGL module is given a negative sign. This parameter is used in the solution
sequences to branch to an error exit after printing the real eigenvalue table. You may use
a DMAP Alter to print these eigenvectors if the cause of the problem is not apparent in the
eigenvalues. The solution can be forced to completion by changing the sign of this
parameter. You should be aware that a poor-quality solution is provided for this case. This
poor solution may be useful for diagnosing the problem but should not be used for other
purposes.
A possible cause of this error is when large offsets (large relative to the element length) are
used for the BEAM element and coupled mass is selected.

UFM 4671 *** USER FATAL MESSAGE 4671, LOAD COMBINATION REQUESTED BUT LSEQ
CARDS DO NOT EXIST FOR SID = ****.

Check the LSEQ entries.

UFM 4683 *** USER FATAL MESSAGE 4683, MASS/STIFFNESS MATRIX NEEDED FOR
EIGENVALUE ANALYSIS.

The eigensolution module was given a purged (that is, nonexistent) mass or stiffness
matrix. Common causes include the deletion of mass density input on MATi entries, user
restart errors in the superelement solution sequences, or when the auto-selection feature of
generalized dynamic reduction (GDR) determines that there are no natural frequencies
below FMAX. It therefore sets the number of generalized coordinates to zero. This
condition can be detected from UIM 4181. An avoidance is to increase FMAX on the
DYNRED Bulk Data entry if GDR is used. Provide mass matrix generating data by any of
several means including a mass density entry on material entries, concentrated masses,
and g-type DMIG entries.
488

UFM 4683 Possibly no mass matrix is defined. Check for the following:
(cont.)
• RHO entry on MATi
• NSM entry on element properties (i.e., PSHELL, PBAR)
• CONMi or CMASSi
MSC.Nastran needs at least one of the above to compute the mass matrix. Incorrect
cross-sectional properties may also lead to this error but typically show up as another error.

UWM 4698 *** USER WARNING MESSAGE 4698, STATISTICS FOR DECOMPOSITION OF MATRIX
****. THE FOLLOWING DEGREES-OF-FREEDOM HAVE FACTOR DIAGONAL RATIOS
GREATER THAN ****, OR HAVE NEGATIVE TERMS ON THE FACTOR DIAGONAL.
During decomposition, the degrees-of-freedom listed have pivot ratios that are greater
than maxratio or are negative. Verify that the degrees-of-freedom are not part of a
mechanism and that elements do not have excessive stiffness. In SOLs 61 and higher this
condition causes run termination. PARAM,BAILOUT may be used to continue the run to
obtain messages issued by subsequent modules. See the NX Nastran Numerical Methods
User’s Guide.

UIM 5010 *** USER INFORMATION MESSAGE 5010, STURM SEQUENCE DATA FOR
EIGENVALUE EXTRACTION. TRIAL EIGENVALUE = (real), CYCLES = (real),
NUMBER OF EIGENVALUES BELOW THIS VALUE = (integer).
This message is automatic output during eigenvalue extraction using the Lanczos and
SINV methods. This message can be used, along with the list of eigenvalues, to identify the
modes found. See the NX Nastran Numerical Methods User’s Guide.

UFM 5025 *** USER FATAL MESSAGE 5025, LAMA PURGED. DSTA MODULE TERMINATED.
The LAMA data block contains a list of natural frequencies and may be purged because no
eigenvalues were computed or the data block was not properly recovered on restart.

UIM 5218 *** USER INFORMATION MESSAGE 5218, EIGENVALUE APPROACHING INFINITY
AT **** TH MODE. EIGENVECTORS WILL NOT BE COMPUTED BEYOND THIS
POINT.
The MGIV, MHOU, AGIV, and AHOU methods substitute a very large number for
eigenvalues that approach machine infinity. If eigenvectors are computed for these
artificial values, they may be numerical noise, or they may cause overflows. Eigenvector
computation is halted at the first machine infinity instead even if you requested
eigenvectors in this range.
 CHAPTER K 489
Diagnostic Messages for Dynamic Analysis

UIM 5222 *** USER INFORMATION MESSAGE 5222, COUPLED/UNCOUPLED SOLUTION

ALGORITHM USED.
The modal methods use uncoupled solution algorithms, if possible. The uncoupled
algorithms are considerably more economical than the coupled algorithms. Coupled
algorithms are required when any of the following effects are present: transfer functions,
DMIG requests of the p-type, element damping, and PARAM,G.

Consider the use of modal damping (TABDMP1 entry) to reduce the cost of your analysis
in modal solutions.

UFM 5225 *** USER FATAL MESSAGE 5225, ATTEMPT TO OPERATE ON THE SINGULAR
MATRIX **** IN SUBROUTINE DCMP.
This message is preceded by the listing of the grid point ID and degrees-of-freedom for any
null columns.

UIM 5236 *** USER INFORMATION MESSAGE 5236, THE FREQUENCY RANGE HAS BEEN
SPLIT INTO **** SUBREGIONS.
The overall frequency range for eigenanalysis is split into several smaller ranges when
using the SINV option to calculate modes and frequencies.

UFM 5238 *** USER FATAL MESSAGE 5238, THE NUMBER OF ROOTS IN THE DEFINED
FREQUENCY RANGE IS GREATER THAN 600.
More than 600 roots are in the desired frequency range, which is greater than the maximum
allowed using SINV. Decrease the size of the frequency range.

UIM 5239 *** USER INFORMATION MESSAGE 5239, BISECTIONING IN THE ***-*** INTERVAL.
The frequency subregion encompassing eigenvalues xx-yy is cut in half in order to find the
remaining roots.

UIM 5240 *** USER INFORMATION MESSAGE 5240, THE BISECTION VALUE IS: ****

The selected value is midway between the lowest and highest frequencies in the frequency
subregion.

UIM 5241 *** USER INFORMATION MESSAGE 5241, MISSING ROOT(S) IN THE ***-***
INTERVAL.
The Sturm sequence check has indicated that roots are missing in the frequency range, and
they cannot be found by further bisectioning.

If the run terminates with missing roots, decrease the frequency range.

UIM 5242 *** USER INFORMATION MESSAGE 5242, THE ROOT FOUND IS NOT THE LOWEST
ONE ABOVE FMIN.

The Sturm sequence check indicates that at least one unfounded root exists between FMIN
and the lowest frequency root found. Set FMAX close to the lowest frequency found, so
that the lower roots can be found.
490

UIM 5274 *** USER INFORMATION MESSAGE 5274, THE ACTUAL TIME OF (****)
TRIDIAGONALIZATION IS: ****, THE ACTUAL TIME OF EIGENVALUE ITERATION
IS: ****, THE ACTUAL TIME OF EIGENVECTOR GENERATION IS: ****.

The time spent in the major operations of the real eigensolution module are output. Note
that the number of eigenvectors requested has a large effect on solution cost.

UFM 5288 *** USER FATAL MESSAGE 5288, NO ROOT EXISTS ABOVE FMIN.

This message occurs when the eigenproblem of finding any number of roots above FMIN
is expected, but the Sturm number indicates that there is no root above FMIN. Reduce
FMIN in order to attempt to find a root.

SFM 5299 *** SYSTEM FATAL MESSAGE 5299 (This text varies depending on the reason for
termination; see the description given below.)
1. Insufficient storage for the Lanczos method.
2. Factorization error on three consecutive shifts.
3. Stack overflow in the Lanczos method.
4. Unrecoverable termination from the Lanczos method.
5. Insufficient working storage.
6. Finite interval analysis error (see UIM 6361).
This message can also occur for models with two or more widely separated groups of
repeated roots. An avoidance is to search each group separately. This error may also be
caused by a massless mechanism, which can be confirmed by performing a static analysis.

See the NX Nastran Numerical Methods User’s Guide for more information.

UFM 5400 *** USER FATAL MESSAGE 5400, INCORRECT RELATIONSHIP BETWEEN
FREQUENCY LIMITS.
You have incorrectly specified V1 > V2. Check V1, V2 specified on the EIGRL Bulk Data
entry.

SFM 5401 *** SYSTEM FATAL MESSAGE 5401 (REIGL), LANCZOS METHOD IS UNABLE TO
FIND ALL EIGENVALUES IN RANGE. ACCEPTED EIGENVALUES AND
ADDITIONAL ERROR MESSAGES MAY BE LISTED ABOVE. USER ACTION: RERUN
WITH ANOTHER METHOD OR ANOTHER SETTING ON EIGRL ENTRY.

This message can be issued if insufficient memory is available for Lanczos with sparse
decomposition. It can also be issued if UFM 5299 occurs. See the NX Nastran Numerical
Methods User’s Guide.

This condition can be related to the occurrence of UWM 5411.

UWM 5402 *** USER WARNING MESSAGE 5402, ---THE PROBLEM HAS NO STIFFNESS MATRIX.
The problem requires a stiffness matrix. Verify that property entries are specified correctly.
 CHAPTER K 491
Diagnostic Messages for Dynamic Analysis

UIM 5403 *** USER INFORMATION MESSAGE 5403, CPU TIME AT START OF LANCZOS
ITERATION ****.
Since several Lanczos iterations may be executed during one application of the Lanczos
method (each shift is followed by at least one iteration), this information is given to
measure the time required for the individual iterations.

UWM 5404 *** USER WARNING MESSAGE 5404, NEGATIVE MODAL MASS TERM, IS
ENCOUNTERED DURING INVERSE ITERATION. PROCESS ABORTED.
The modal mass matrix should have unit diagonal terms (for mass normalization).
Negative terms may indicate negative eigenvalues. If these negative terms are
computational zeroes (rigid-body modes, for example), then the negative terms are
acceptable. If the negative terms are finite values, there may be a modeling problem.

UWM 5405 *** USER WARNING MESSAGE 5405, ERROR OCCURRED DURING ITERATION.
ERROR NUMBER IS : Y (SEE DESCRIPTION FOR VALUES OF Y AND USER ACTION.)
This message marks the breakdown of the inverse iteration process in the Lanczos method.
See the NX Nastran Numerical Methods User’s Guide for additional values and actions.

Y Value User Action

-11 File open error in interface; see GINO This error should not occur; report error
error message. to MSC.
-12 File open error in postprocessing; see This error should not occur; report error
GINO error message. to MSC.
-13 File read error; see GINO error message. This error should not occur; report error
to MSC.
-21 Insufficient space for blocksize = 1. Increase memory.
-22 Three consecutive factorizations failed Possible ill-conditioning; check model.
at a shift.
-23 Lanczos internal table overflow due to Specify smaller internal; may be
enormous number of shifts. necessary to have several runs.
-31 Internal error in Lanczos (REIGL) This error should not occur; report error
module. to MSC.
-32 No convergence in solving the Possible ill-conditioning; check model.
tridiagonal problem.
-33 Too many eigenvalues were found; Check the orthogonality of the
inconsistency between the roots found eigenvectors; if it is good, then ignore
and Sturm number. this warning.
492

UWM 5406 *** USER WARNING MESSAGE 5406 NO CONVERGENCE IN SOLVING THE
TRIDIAGONAL PROBLEM.
This message signals eigensolution problems in the Lanczos method. There is possible
ill-conditioning; check your model.

UWM 5407 *** USER WARNING MESSAGE 5407, INERTIA (STURM SEQUENCE) COUNT
DISAGREES WITH THE NUMBER OF MODES ACTUALLY COMPUTED IN AN (SUB)
INTERVAL
This message shows a serious problem. Spurious modes were found in the Lanczos
method. Check the multiplicity of the roots given in the interval. See the NX Nastran
Numerical Methods User’s Guide.

UWM 5408 *** USER WARNING MESSAGE 5408, FACTORIZATION FAILED. SHIFT CHANGED
TO ****.

No user action to be taken. This message occurs only for the Lanczos method.

UWM 5411 *** USER WARNING MESSAGE 5411, NEGATIVE TERM ON DIAGONAL OF MASS
MATRIX (VIBRATION) OR STIFFNESS (BUCKLING), ROW ****, VALUE = ****

The message is given from the REIGL module which performs a necessary (but not
sufficient) check on the positive semi-definiteness of the indicated matrix. Look for
evidence of negative mass, such as minus signs on input. Negative terms on the factor of
the indicated matrix must be removed for correct answers.

Something has caused a negative term on the diagonal of the mass or stiffness matrix. Look
for explicitly defined negative mass and/or stiffness terms. Also, check the continuation
entries on the PBEAM entry. An incorrect entry for the SO field may lead to improper mass
definition. For example, if SO is set to NO at a particular X/XB location, the continuation
entry for defining four stress locations on the cross section (C, D, E, F) is not used. If SO is
NO but the C, D, E, and F points are entered in error, negative mass terms can result if
either E1 or E2 entries are entered. The offending DOF can be traced using the USET tables.
The Lanczos method gives wrong answers for indefinite matrices. The existence of
negative diagonal terms indicates a subclass of indefinite matrix.
See the NX Nastran Numerical Methods User’s Guide for more information.

UFM 5429 *** USER FATAL MESSAGE 5429, INSUFFICIENT TIME TO FINISH LANCZOS
ITERATION.
There is not enough time remaining to complete the analysis based on an estimate of how
long the Lanczos iteration will take. Increase the value on the TIME statement in the
Executive Control Section and rerun. In order to ensure that the run can be competed,
overestimate the TIME value using estimates or actual times from a similar model.
 CHAPTER K 493
Diagnostic Messages for Dynamic Analysis

UIM 5458 *** USER INFORMATION MESSAGE 5458, (****) METHOD IS SELECTED. or (****)
METHOD IS (****)
The exact text of this message depends on the METHOD field on the selected EIGR Bulk
Data entry. This message indicates the eigensolution status (all eigenvalues found, not all
found, etc.).

UFM 6133 *** USER FATAL MESSAGE 6133 (DFMSDD), SINGULAR MATRIX IN SPARSE
DECOMPOSITION.
USER ACTION: CHECK MODEL.

This message is often followed by UFM 4645, UFM 4646, or UWM 4648.

UFM 6134 *** USER FATAL MESSAGE 6134 (DFMSDD), MATRIX IS NOT POSITIVE DEFINITE IN
SPARSE DECOMPOSITION.
USER ACTION: CHECK MODEL.
This message is often followed by UFM 4645, UFM 4646, or UWM 4648.

SFM 6135 *** SYSTEM FATAL MESSAGE 6135, ERROR IN READING SYMBOLIC FACTOR IN
SPARSE FBS.

This message may be issued if the FBS module is using a sparse method to solve factors
which are not decomposed by the sparse method. This message can also be caused by a
compatibility or database integrity problem.

UFM 6136 *** USER FATAL MESSAGE 6136 (****), INSUFFICIENT CORE FOR
(SYMBOLIC/NUMERIC) PHASE OF SPARSE DECOMPOSITION.
USER ACTION: INCREASE CORE BY **** WORDS.

USER INFORMATION:
!!! NOW REVERTING BACK TO ACTIVE COLUMN DECOMPOSITION UPON USER
REQUEST !!!
If this message is issued in the symbolic phase, the memory estimate is not necessarily
conservative and even more memory may be required (although this estimate is fairly
accurate for Version 68). Also, the memory increase required is only for the symbolic
phase. It is not unusual for the decomposition phase to require more memory than the
symbolic phase. To increase the chances for a successful run, increase the memory even
more than the amount indicated in this message. After the run is complete, determine the
amount of memory actually used and use this as a guideline for similar runs in the future.
The user information message is written if SYSTEM(166) = 1 (that is, if there is not enough
memory for sparse decomposition, and you should switch to regular decomposition).
494

UWM 6137 *** USER WARNING MESSAGE 6137 (DFMSDD), INPUT MATRIX IS RANK
DEFICIENT, RANK = ****.
USER ACTION: CHECK MODEL.

One of your matrices is singular. See the NX Nastran Numerical Methods User’s Guide for a
discussion of singularity.

UFM 6138 *** USER FATAL MESSAGE 6138 (DFMSB), INSUFFICIENT CORE FOR SPARSE FBS.

USER ACTION: INCREASE CORE BY **** WORDS.

See UFM 6136.

UIM 6214 *** USER INFORMATION MESSAGE 6214, FEWER THAN REQUESTED VECTORS
CALCULATED, DUE TO INSUFFICIENT TIME.
This information message occurs in the READ module when there is insufficient time to
compute eigenvectors. Resubmit the job with an increased time limit (TIME).

UWM 6243 *** USER WARNING MESSAGE 6243 (READ) ---- THE DEGREE OF FREEDOM (D.O.F)
REQUESTED FOR POINT NORMALIZATION HAS NOT BEEN SPECIFIED ON THE
EIGR OR EIGB ENTRY.

USER INFORMATION: THE D.O.F PRECEDING THE REQUESTED D.O.F. IN THE

INTERNAL SEQUENCE LIST WILL BE USED
The point requested was not in the a-set, so another point was chosen.

UIM 6361 *** USER INFORMATION MESSAGE 6361-LANCZOS MODULE DIAGNOSTICS

This message prints various levels of diagnostics for the Lanczos method. The amount of
print depends on the message level set on the EIGRL entry. See the NX Nastran Numerical
Methods User’s Guide for more information.

UIM 6480 *** USER INFORMATION MESSAGE 6480 (REIGLA) ---

EXTERNAL IDENTIFICATION TABLE FOR DECOMPOSITION MESSAGES FOR
MATRIX ****
ROW NUMBER **** = GRID ID **** + COMPONENT ****

This message is output from the REIGL module when using sparse decomposition to
convert the internal (row number oriented) diagnostic messages to external (grid and
component) form.
 NX Nastran Basic Dynamic Analysis User’s Guide

L
APPENDIX
References and Bibliography

■ Overview
■ General References

■ Bibliography
496

L.1 Overview
This appendix includes references of interest in the field of dynamic analysis. Two
categories are included. The first category, General References, lists books that cover the
general range of structural dynamic analysis. The second category, Bibliography, is an
excerpt from the dynamic analysis section of the MSC.Nastran Bibliography, Second Edition.
 CHAPTER L 497
References and Bibliography

L.2 General References

1. Paz, M., Structural Dynamics: Theory and Computation, Van Nostrand Reinhold,
New York, N.Y., 1985.
2. Bathe, K. J. and Wilson, E. L., Numerical Methods in Finite Element Analysis,
Prentice-Hall, Englewood Cliffs, N.J., 1976.
3. Harris, C. M. and Crede, C. E., Shock and Vibration Handbook, McGraw-Hill, New
York, N.Y., 1976.
4. Clough, R. W. and Penzien, J., Dynamics of Structures, McGraw-Hill, New York,
N.Y., 1975.
5. Timoshenko, S., Young, D. H., and Weaver Jr., W., Vibration Problems in
Engineering, John Wiley and Sons, New York, N.Y., 1974.
6. Hurty, W. C. and Rubinstein, M. F., Dynamics of Structures, Prentice-Hall,
Englewood Cliffs, N.J., 1964.
498

L.3 Bibliography
DYNAMICS – GENERAL
Abdallah, Ayman A.; Barnett, Alan R.; Widrick, Timothy W.; Manella, Richard T.; Miller,
Robert P. Stiffness-Generated Rigid-Body Mode Shapes for Lanczos Eigensolution
with Support DOF Via a MSC/NASTRAN DMAP Alter, MSC 1994 World Users’
Conf. Proc., Paper No. 10, June, 1994.
Anderson, William J.; Kim, Ki-Ook; Zhi, Bingchen; Bernitsas, Michael M.; Hoff, Curtis;
Cho, Kyu-Nam. Nonlinear Perturbation Methods in Dynamic Redesign,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 16, March, 1983.

Barber, Pam; Arden, Kevin. Dynamic Design Analysis Method (DDAM) Using
MSC/NASTRAN, MSC 1994 World Users’ Conf. Proc., Paper No. 31, June, 1994.

Bedrossian, Herand; Veikos, Nicholas. Rotor-Disk System Gyroscopic Effect in

MSC/NASTRAN Dynamic Solutions, MSC/NASTRAN Users’ Conf. Proc., Paper
No. 12, March, 1982.

Bernstein, Murray; Mason, Philip W.; Zalesak, Joseph; Gregory, David J.; Levy, Alvin.
NASTRAN Analysis of the 1/8-Scale Space Shuttle Dynamic Model, NASTRAN:
Users’ Exper., pp. 169-242, September, 1973, (NASA TM X-2893).

Berthelon, T.; Capitaine, A. Improvements for Interpretation of Structural Dynamics

Calculation Using Effective Parameters for Substructures, Proc. of the 18th MSC
Eur. Users’ Conf., Paper No. 9, June, 1991.
Birkholz, E. Dynamic Investigation of Automobile Body Parts, Proc. of the 15th
MSC/NASTRAN Eur. Users’ Conf., October, 1988.

Bishop, N. W. M.; Lack, L. W.; Li, T.; Kerr, S. C. Analytical Fatigue Life Assessment of
Vibration Induced Fatigue Damage, MSC 1995 World Users’ Conf. Proc., Paper No.
18, May, 1995.
Blakely, Ken; Howard, G. E.; Walton, W. B.; Johnson, B. A.; Chitty, D. E. Pipe Damping
Studies and Nonlinear Pipe Benchmarks from Snapback Tests at the
Heissdampfreaktor, NUREG/CR-3180, March, 1983.

Blakely, Ken. Dynamic Analysis: Application and Modeling Considerations, J. of

Engineering Computing and Applications, Fall, 1987.

Bramante, A.; Paolozzi, A; Peroni, I. Effective Mass Sensitivity: A DMAP Procedure, MSC
1995 World Users’ Conf. Proc., Paper No. 39, May, 1995.

Brutti, C.; Conte, M.; Linari, M. Reduction of Dynamic Environment to Equivalent Static
Loads by a NASTRAN DMAP Procedure, MSC 1995 European Users’ Conf. Proc.,
Italian Session, September, 1995.

Butler, Thomas G. Dynamic Structural Responses to Rigid Base Acceleration, Proc. of the
Conf. on Finite Element Methods and Technology, Paper No. 8, March, 1981.

Butler, Thomas G. Telescoping Robot Arms, MSC/NASTRAN Users’ Conf. Proc., Paper
No. 10, March, 1984.
 CHAPTER L 499
References and Bibliography

Butler, T. G. Experience with Free Bodies, Thirteenth NASTRAN Users’ Colloq.,

pp. 378-388, May, 1985, (NASA CP-2373).

Butler, Thomas G. Mass Modeling for Bars, Fifteenth NASTRAN Users’ Colloq.,
pp. 136-165, August, 1987, (NASA CP-2481).

Butler, T. G. Coupled Mass for Prismatical Bars, Sixteenth NASTRAN Users’ Colloq.,
pp. 44-63, April, 1988, (NASA CP-2505).
Caldwell, Steve P.; Wang, B. P. An Improved Approximate Method for Computing
Eigenvector Derivatives in MSC/NASTRAN, The MSC 1992 World Users’ Conf.
Proc., Vol. I, Paper No. 22, May, 1992.

Case, William R. Dynamic Substructure Analysis of the International Ultraviolet

Explorer (IUE) Spacecraft, NASTRAN: Users’ Exper., pp. 221-248, September, 1975,
(NASA TM X-3278).

Chang, H. T.; Cao, Tim; Hua, Tuyen. SSF Flexible Multi-Body Control/Structure
Interaction Simulation, The MSC 1993 World Users’ Conf. Proc., Paper No. 15,
May, 1993.
Chang, W. M.; Lai, J. S.; Chyuan, S. W.; Application of the MSC/NASTRAN Design
Optimization Capability to Identify Joint Dynamic Properties of Structure, The
Sixth Annual MSC Taiwan Users’ Conf. Proc., Paper No. 1, November, 1994.
Chargin, M.; Miura, H.; Clifford, Gregory A. Dynamic Response Optimization Using
MSC/NASTRAN, The MSC 1987 World Users Conf. Proc., Vol. I, Paper No. 14,
March, 1987.

Chen, J. T.; Chyuan, S. W.; You, D. W.; Wong, H. T. A New Method for Determining the
Modal Participation Factor in Support Motion Problems Using MSC/NASTRAN,
The Seventh Annual MSC/NASTRAN Users’ Conf. Proc., Taiwan, 1995.

Chen, J. T.; Wong, H. T. Applications of Modal Reaction Method in Support Motion

Problems, Techniques in Civil Engineering, Vol. 4, pp 17 - 30, March, 1996, in
Chinese.

Chen, J. T.; Hong, H. K.; Chyuan, S. W.; Yeh, C. S. A Note on the Application of Large Mass
and Large Stiffness Techniques for Multi-Support Motion, The Fifth Annual MSC
Taiwan Users’ Conf. Proc., November, 1993.
Chen, Yohchia. Improved Free-Field Analysis for Dynamic Medium-Structure Interaction
Problems, The MSC 1992 World Users’ Conf. Proc., Vol. I, Paper No. 13, May, 1992.

Chen, Yohchia. Dynamic Response of Reinforced Concrete Box-Type Structures, The MSC
1992 World Users’ Conf. Proc., Vol. I, Paper No. 24, May, 1992.
Chiu, Chi-Wai. Spacecraft Dynamics During Solar Array Panel Deployment Motion, The
Fifth Annual MSC Taiwan Users’ Conf. Proc., November, 1993.

Chung, Y. T.; Kahre, L. L. A General Procedure for Finite Element Model Check and Model
Identification, MSC 1995 World Users’ Conf. Proc., Paper No. 38, May, 1995.

Cicia, C. Static, Thermal and Dynamic Analysis of the Liquid Argon Cryostat for the
ICARUS Experiment, Proc. of the 15th MSC/NASTRAN Eur. Users’ Conf., October,
1988.
500

Cifuentes, Arturo O. Dynamic Analysis of Railway Bridges Using MSC/NASTRAN, The

MSC 1988 World Users Conf. Proc., Vol. II, Paper No. 44, March, 1988.

Cifuentes, A. O. Dynamic Response of a Beam Excited by a Moving Mass, Finite Elements

in Analysis and Design, Vol. 5, pp. 237-246, 1989.

Citerley, R. L.; Woytowitz, P. J. Ritz Procedure for COSMIC/ NASTRAN, Thirteenth

NASTRAN Users’ Colloq., pp. 225-233, May, 1985, (NASA CP-2373).
Ciuti, Gianluca. Avionic Equipment Dynamic Analysis, MSC 1995 European Users’ Conf.
Proc., Italian Session, September, 1995.

Coates, Dr. Tim; Matthews, Peter. Transient Response in Dynamic and Thermal
Behaviour, The Second Australasian MSC Users Conf. Proc., Paper No. 4,
November, 1988.
Coppolino, Robert N.; Bella, David F. Employment of MSC/STI-VAMP for Dynamic
Response Post-Processing, The MSC 1987 World Users Conf. Proc., Vol. I, Paper
No. 12, March, 1987.
Corder, P. R.; Persh, R. Castigliano and Symbolic Programming in Finite Element
Analysis, Proceedings of the 16th Annual Energy - Sources Technology Conference
and Exhibition, Houston, 1993.
Coyette, J. P.; Wijker, J. J. The Combined Use of MSC/NASTRAN and Sysnoise for
Evaluating the Dynamic Behavior of Solar Array Panels, Proc. of the 20th MSC
European Users’ Conf., Paper No. 16, September, 1993.
Curti, G.; Chiandussi, G.; Scarpa, F. Calculation of Eigenvalue Derivatives of
Acousto-Structural Systems with a Numerical Comparison, MSC 23rd European
Users’ Conf. Proc., Italian Session, September, 1996.
Defosse, H.; Sergent, A. Vibro-Acoustic Modal Response Analysis of Aerospace
Structures, Proc. of the MSC/NASTRAN Eur. Users’ Conf., April, 1985.
Deloo, Ph.; Dunne, L.; Klein, M. Alter DMAPS for the Generation, Assembly and Recovery
of Craig-Bampton Models in Dynamic Analyses, Actes de la 2ème Confèrence
Française Utilisateurs des Logiciels MSC, Toulouse, France, September, 1995.
Denver, Richard E.; Menichello, Joseph M. Alternate Approaches to Vibration and Shock
Analysis Using NASTRAN, Sixth NASTRAN Users’ Colloq., pp. 199-212, October,
1977, (NASA CP-2018).

Detroux, P.; Geraets, L. H. Instability at Restart or Change of Time Step with NASTRAN
in the Presence of Nonlinear Loads, Proc. of the MSC/NASTRAN Eur. Users’ Conf.,
June, 1983.
Deuermeyer, D. W.; Clifford, G. A.; Petesch, D. J. Traditional Finite Element Analysis:
Opportunities for Parallelism?, Computing Systems in Engineering, Vol. 2, No. 2-3,
pp. 157-165, 1991.
Dirschmid, Dr. W.; Nolte, Dr. F.; Dunne, L. W. Application of an FRF-Based Update
Method to the Model Parameter Tuning of an Hydraulic Engine Mounting, Proc. of
the 18th MSC Eur. Users’ Conf., Paper No. 10, June, 1991.
 CHAPTER L 501
References and Bibliography

Drago, Raymond J.; Margasahayam, Ravi N. Resonant Response of Helicopter Gears

Using 3-D Finite Element Analysis, The MSC 1988 World Users Conf. Proc., Vol. I,
Paper No. 20, March, 1988.
Elchuri, V.; Smith, G. C. C.; Gallo, A. Michael. An Alternative Method of Analysis for Base
Accelerated Dynamic Response in NASTRAN, Eleventh NASTRAN Users’ Colloq.,
pp. 89-112, May, 1983.
Everstine, Gordon C.; Schroeder, Erwin A. The Dynamic Analysis of Submerged
Structures, NASTRAN: Users’ Exper., pp. 419-430, September, 1975, (NASA TM
X-3278).

Everstine, Gordon C. Structural Analogies for Scalar Field Problems, Int. J. for Numerical
Methods in Engineering, Vol. 17, No. 3, pp. 471-476, March, 1981.
Everstine, G. C. Dynamic Analysis of Fluid-Filled Piping Systems Using Finite Element
Techniques, J. of Pressure Vessel Technology, Vol. 108, pp. 57-61, February, 1986.

Flanigan, Christopher C. Accurate and Efficient Mode Acceleration Data Recovery for
Superelement Models, The MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 38,
March, 1988.

Flanigan, Christopher C.; Manella, Richard T. Advanced Coupled Loads Analysis Using
MSC/NASTRAN, The MSC 1991 World Users’ Conf. Proc., Vol. I, Paper No. 14,
March, 1991.

Fox, Gary L. Solution of Enforced Boundary Motion in Direct Transient and Harmonic
Problems, Ninth NASTRAN Users’ Colloq., pp. 96-105, October, 1980, (NASA
CP-2151).
Geyer, A.; Schweiger, W. Multiple Support Excitation for NASTRAN Piping Analysis,
Proc. of the MSC/NASTRAN Eur. Users’ Conf., April, 1982.

Geyer, A.; Schweiger, W. Aeroelastic and Stress Analysis of the CHIWEC Chinese Wind
Energy Converter Using MSC/NASTRAN, Proc. of the MSC/NASTRAN Eur. Users’
Conf., April, 1985.

Ghofranian, S.; Dimmagio, O. D. Space Station Dynamic Analysis with Active Control
Systems Using MSC/NASTRAN, The MSC 1988 World Users Conf. Proc., Vol. I,
Paper No. 17, March, 1988.
Gibson, Warren C. Experiences with Optimization Using ASD/NASOPT and
MSC/NASTRAN for Structural Dynamics, The MSC 1987 World Users Conf. Proc.,
Vol. I, Paper No. 13, March, 1987.

Gibson, Warren C.; Austin, Eric. Analysis and Design of Damped Structures Using
MSC/NASTRAN, The MSC 1992 World Users’ Conf. Proc., Vol. I, Paper No. 25,
May, 1992.

Gielen, L.; Brughmans, M.; Petellat, C. A Stepwise Approach for Fatigue Evaluation of
Engine Accessories Prior to Prototyping Using Hybrid Modelling Technology, MSC
1996 World Users’ Conf. Proc., Vol. III, Paper No. 29, June, 1996.

Go, James Chi-Dian. Structural Dynamic and Thermal Stress Analysis of Nuclear Reactor
Vessel Support System, NASTRAN: Users’ Exper., pp. 465-476, September, 1972,
(NASA TM X-2637).
502

Grasso, A.; Tomaselli, L. Whirling Speed Analysis of Multispool Systems, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., May, 1984.

Grimes, Roger G.; Lewis, John G.; Simon, Horst D.; Komzsik, Louis; Scott, David S. Shifted
Block Lanczos Algorithm in MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc.,
Paper No. 12, March, 1985.

Harn, Wen-Ren; Lin, Shyang-Kuang; Chen, Jeng-Tzong. Localization of Dynamic Model

Modification Based on Constrained Minimization Method, The 2nd Annual MSC
Taiwan Users Conf., Paper No. 14, October, 1990.

Herting, David N.; Bella, David F.; Kimbrough, Patty A. Finite Element Simulation of
Coupled Automobile Engine Dynamics, The MSC 1987 World Users Conf. Proc.,
Vol. I, Paper No. 10, March, 1987.
High, Gerald D. An Iterative Method for Eigenvector Derivatives, The MSC 1990 World
Users Conf. Proc., Vol. I, Paper No. 17, March, 1990.

Hill, R. G. Transient Analysis of an IVHM Grapple Impact Test, NASTRAN: Users’

Exper., pp. 161-178, September, 1972, (NASA TM X-2637).
Howells, R. W.; Sciarra, J. J. Finite Element Analysis Using NASTRAN Applied to
Helicopter Transmission Vibration/Noise Reduction, NASTRAN: Users’ Exper.,
pp. 321-340, September, 1975, (NASA TM X-3278).
Howlett, James T. Applications of NASTRAN to Coupled Structural and Hydrodynamic
Responses in Aircraft Hydraulic Systems, NASTRAN: Users’ Exper., pp. 407-420,
September, 1971, (NASA TM X-2378).

Huang, S. L.; Rubin, H. Static and Dynamic Analysis, F-14A Boron Horizontal Stabilizer,
NASTRAN: Users’ Exper., pp. 251-264, September, 1971, (NASA TM X-2378).
Hurwitz, Myles M. New Large Deflection Analysis for NASTRAN, Sixth NASTRAN
Users’ Colloq., pp. 235-256, October, 1977, (NASA CP-2018).
Hussain, M. A.; Pu, S. L.; Lorensen, W. E. Singular Plastic Element: NASTRAN
Implementation and Application, Sixth NASTRAN Users’ Colloq., pp. 257-274,
October, 1977, (NASA CP-2018).
Ishikawa, Masanori; Iwahara, Mitsuo; Nagamatsu, Akio. Dynamic Optimization Applied
to Engine Structure, The MSC 1990 World Users Conf. Proc., Vol. I, Paper No. 31,
March, 1990.

Iwahara, Mitsuo. Dynamic Optimization Using Quasi Least Square Method, The Fifth
MSC/NASTRAN User’s Conf. in Japan, October, 1987, in Japanese.

Jakovich, John; Van Benschoten, John. SDRC SUPERTAB Interactive Graphics as a

Front-End to MSC/NASTRAN Dynamic Analysis, Proc. of the MSC/NASTRAN
Users’ Conf., March, 1979.

Jones, Gary K. The Response of Shells to Distributed Random Loads Using NASTRAN,
NASTRAN: Users’ Exper., pp. 393-406, September, 1971, (NASA TM X-2378).

Kalinowski, Anthony J. Steady Solutions to Dynamically Loaded Periodic Structures,

Eighth NASTRAN Users’ Colloq., pp. 131-164, October, 1979, (NASA CP-2131).
 CHAPTER L 503
References and Bibliography

Kalinowski, A. J. Solution Sensitivity and Accuracy Study of NASTRAN for Large

Dynamic Problems Involving Structural Damping, Ninth NASTRAN Users’
Colloq., pp. 49-62, October, 1980, (NASA CP-2151).
Kasai, Manabu. Real Eigenvalue Analysis by Modal Synthesis Method Taking
Differential Stiffness into Account, The First MSC/NASTRAN User’s Conf. in
Japan, October, 1983, in Japanese.
Kasai, Manabu. Recovery Method for Components by DMAP of Constrained Modal Type,
The Fifth MSC/NASTRAN User’s Conf. in Japan, October, 1987, in Japanese.

Kasai, Manabu. DMAP Program for Modal Mass and Momentum, The Sixth
MSC/NASTRAN User’s Conf. in Japan, October, 1988, in Japanese.

Kienholz, Dave K.; Johnson, Conor D.; Parekh, Jatin C. Design Methods for
Viscoelastically Damped Sandwich Plates, AIAA/ASME/ASCE/AHS 24th
Structures, Structural Dynamics and Materials Conf., Part 2, pp. 334-343, May, 1983.

Lambert, Nancy; Tucchio, Michael. Ring Element Dynamic Stresses, Ninth NASTRAN
Users’ Colloq., pp. 63-78, October, 1980, (NASA CP-2151).
Lee, Jyh-Chiang. Investigation for the Large Stiffness Method, The Fifth Annual MSC
Taiwan Users’ Conf. Proc., November, 1993.
Lee, Ting-Yuan; Lee, Jyh-Chiang. Modal Analysis and Structural Modification for a
Harpoon Launcher, The Fifth Annual MSC Taiwan Users’ Conf. Proc., November,
1993.
LeMaster, R. A.; Runyan, R. B. Dynamic Certification of a Thrust-Measuring System for
Large Solid Rocket Motors, Eleventh NASTRAN Users’ Colloq., pp. 207-225, May,
1983.
Lewis, J.; Komzsik, L. Symmetric Generalized Eigenproblems in Structural Engineering,
SIAM Conf. on Applied Numerical Analysis, 1985.
Lin, Chih-Kai; Harn, Wen-Ren; Lin, Shyang-Kuang. The Dynamic Response of Bridge Due
to Passing of Vehicle, The 2nd Annual MSC Taiwan Users Conf., Paper No. 6,
October, 1990.
Lin, S. L.; Yang, T. W.; Chen, J. T. MSC/NASTRAN Application in Inertia Relief, The 1st
MSC Taiwan Users’ Conf., Paper No. 13, October, 1989, in Chinese.
Lin, Shan. Time-Dependent Restrained Boundary Condition Simulation, The MSC 1988
World Users Conf. Proc., Vol. I, Paper No. 9, March, 1988.

Lu, Ming-Ying; Yang, Joe-Ming. Analysis of Static and Dynamic Responses on Ship
Structures Under Wave Loadings, The Sixth Annual MSC Taiwan Users’ Conf.
Proc., Paper No. 14, November, 1994.

Magari, P. J.; Shultz, L. A.; Murthy, V. R. Dynamics of Helicopter Rotor Blades, Computers
and Structures, Vol. 29, No. 5, pp. 763-776, 1988.
Malcolm, D. J. Dynamic Response of a Darrieus Rotor Wind Turbine Subject to Turbulent
Flow, Engineering Structures, Vol. 10, No. 2, pp. 125-134, April, 1988.

Maritan, M.; Micelli, D. Dynamic Behaviour of a High-Speed Crankshaft, MSC 23rd

European Users’ Conf. Proc., Italian Session, September, 1996.
504

Masters, Steven G. Plant Troubleshooting with MSC/NASTRAN, Proc. of the Conf. on

Finite Element Methods and Technology, Paper No. 12, March, 1981.

Mastrorocco, David T. Predicting Dynamic Environments for Space Structure

Appendages, The MSC 1992 World Users’ Conf. Proc., Vol. II, Paper No. 56, May,
1992.

Mayer, Lee S.; Zeischka, Johann; Scherens, Marc; Maessen, Frank. Analysis of Flexible
Rotating Crankshaft with Flexible Engine Block Using MSC/NASTRAN and DADS,
MSC 1995 World Users’ Conf. Proc., Paper No. 35, May, 1995.

McLaughlin, A. Finite Element Dynamic Analysis of Production Aircraft, 4th Eur.

Rotorcraft and Powered Lift Aircraft Forum, Assoc. Ital di Aeronaut ed Astronaut,
pp. 20.1-20.7, September, 1978.
Melli, R.; Rispoli, F.; Sciubba, E.; Tavani, F. Structural and Thermal Analysis of Avionic
Instruments for an Advanced Concept Helicopter, Proc. of the 15th
MSC/NASTRAN Eur. Users’ Conf., October, 1988.

Mikami, Kouichi. Dynamic Stress Analysis System for Ship’s Hull Structure Under Wave
Loads, The Second MSC/NASTRAN User’s Conf. in Japan, October, 1984, in
Japanese.
Moharir, M. M. NASTRAN Nonlinear Capabilities in Dynamic Solutions,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 9, March, 1985.

Moore, Gregory J.; Nagendra, Gopal K. Dynamic Response Sensitivities in

MSC/NASTRAN, The MSC 1991 World Users’ Conf. Proc., Vol. I, Paper No. 4,
March, 1991.
Mulcahy, T. M.; Turula, P.; Chung, H.; Jendrzejczyk, A. Analytical and Experimental
Study of Two Concentric Cylinders Coupled by a Fluid Gap, NASTRAN: Users’
Exper., pp. 249-258, September, 1975, (NASA TM X-3278).
Murthy, P. L. N.; Chamis, C. C. Dynamic Stress Analysis of Smooth and Notched Fiber
Composite Flexural Specimens, National Aeronautics and Space Administration,
April, 1984, (NASA TM-83694).

Murthy, P. L. N.; Chamis, C. C. Dynamic Stress Analysis of Smooth and Notched Fiber
Composite Flexural Specimens, Composite Materials: Testing and Design (Seventh
Conf.), ASTM, pp. 368-391, 1986, (ASTM STP 893).
Neal, M. Vibration Analysis of a Printed Wiring Board Assembly, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., May, 1984.

Nefske, D. J.; Sung, S. H. Power Flow Finite Element Analysis of Dynamic Systems: Basic
Theory and Application to Beams, American Soc. of Mechanical Engineers, Noise
Control and Acoustics Division, Vol. 3, pp. 47-54, December, 1987.

Nowak, Bill. The Analysis of Structural Dynamic Effects on Image Motion in Laser
Printers Using MSC/NASTRAN, The MSC 1988 World Users Conf. Proc., Vol. I,
Paper No. 10, March, 1988.

Nowak, Bill. Structural Dynamics Analysis of Laser Printers, Sound and Vibration,
Vol. 23, No. 1, pp. 22-26, January, 1989.
 CHAPTER L 505
References and Bibliography

Nowak, William J. Dynamic Analysis of Optical Scan Systems Using MSC/NASTRAN,

Proc. of the Conf. on Finite Element Methods and Technology, Paper No. 10, March,
1981.
Nowak, William; James, Courtney. Dynamic Modeling and Analysis of Spinning Polygon
Assemblies Using MSC/NASTRAN, The MSC 1993 World Users’ Conf. Proc., Paper
No. 66, May, 1993.
Oei, T. H.; Broerse, G. Reduction of Forced Vibration Levels on Ro-Ro Car Ferry-Type
Ships by Means of Minor Changes of the Inner Aft-Body Construction, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., April, 1982.

Ojalvo, I. U. Extensions of MSC/NASTRAN to Solve Flexible Rotor Problems,

MSC/NASTRAN Users’ Conf. Proc., Paper No. 13, March, 1982.
Palmieri, F. Nonlinear Dynamic Analysis of STS Main Engine Heat Exchanger, Proc. of the
15th MSC/NASTRAN Eur. Users’ Conf., October, 1988.

Palmieri, F. W. Analyzing Deployment of Spacecraft Appendages Using MSC/NASTRAN,

The MSC 1990 World Users Conf. Proc., Vol. I, Paper No. 6, March, 1990.
Pamidi, P. R.; Brown, W. K. On Eigenvectors of Multiple Eigenvalues Obtained in
NASTRAN, NASTRAN: Users’ Exper., pp. 285-300, September, 1975, (NASA TM
X-3278).
Paolozzi, A. Structural Dynamics Modification with MSC/NASTRAN, Proc. of the 19th
MSC European Users’ Conf., Paper No. 14, September, 1992.
Parthasarathy, Alwar. Force-Sum Method for Dynamic Stresses in MSC/NASTRAN
Aeroelastic Analysis, The MSC 1991 World Users’ Conf. Proc., Vol. I, Paper No. 8,
March, 1991.
Parthasarathy, Alwar; Elzeki, Mohamed; Abramovici, Vivianne. PSDTOOL-A DMAP
Enhancement to Harmonic/Random Response Analysis in MSC/NASTRAN, The
MSC 1993 World Users’ Conf. Proc., Paper No. 36, May, 1993.
Patel, Jayant S.; Seltzer, S. M. Complex Eigenvalue Solution to a Spinning Skylab Problem,
NASTRAN: Users’ Exper., pp. 439-450, September, 1971, (NASA TM X-2378).
Patel, Jayant S.; Seltzer, S. M. Complex Eigenvalue Analysis of Rotating Structures,
NASTRAN: Users’ Exper., pp. 197-234, September, 1972, (NASA TM X-2637).
Patel, Kirit V. Stress Analysis of Hybrid Pins in a Warped Printed Wiring Board Using
MSC/NASTRAN, MSC 1995 World Users’ Conf. Proc., Paper No. 20, May, 1995.

Paxson, Ernest B., Jr. Simulation of Small Structures- Optics-Controls Systems with
MSC/NASTRAN, The 1989 MSC World Users Conf. Proc., Vol. II, Paper No. 39,
March, 1989.

Pinnament, Murthy. Mode Acceleration Data Recovery in MSC/NASTRAN Dynamic

Analysis with Generalized Dynamic Reduction, MSC/NASTRAN Users’ Conf.
Proc., Paper No. 24, March, 1985.

Raney, John P.; Kaszubowski, M.; Ayers, J. Kirk. Analysis of Space Station Dynamics
Using MSC/NASTRAN, The MSC 1987 World Users Conf. Proc., Vol. I, Paper
No. 11, March, 1987.
506

Reyer, H. A Crash-Down Calculated with NASTRAN, Proc. of the MSC/NASTRAN Eur.

Users’ Conf., April, 1982.

Rose, Ted L. Using Superelements to Identify the Dynamic Properties of a Structure, The
MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 41, March, 1988.

Rose, Ted L. Creation of and Use of ‘Craig-Bampton’ Models Using MSC/NASTRAN, The
MSC 1990 World Users Conf. Proc., Vol. II, Paper No. 51, March, 1990.
Rose, Ted. Using Residual Vectors in MSC/NASTRAN Dynamic Analysis to Improve
Accuracy, The MSC 1991 World Users’ Conf. Proc., Vol. I, Paper No. 12, March,
1991.

Rose, Ted. DMAP Alters to Apply Modal Damping and Obtain Dynamic Loading Output
for Superelements, The MSC 1993 World Users’ Conf. Proc., Paper No. 24, May,
1993.

Rose, Ted.; McNamee, Martin. A DMAP Alter to Allow Amplitude-Dependent Modal

Damping in a Transient Solution, MSC 1996 World Users’ Conf. Proc., Vol. V, Paper
No. 50, June, 1996.
Ross, Robert W. Prediction and Elimination of Resonance in Structural Steel Frames, The
MSC 1988 World Users Conf. Proc., Vol. II, Paper No. 45, March, 1988.
Russo, A.; Mocchetti, R. Dynamic Analysis of Loaded Structures in the Helicopter Field,
Proc. of the MSC/NASTRAN Eur. Users’ Conf., May, 1984.

Salus, W. L.; Jones, R. E.; Ice, M. W. Dynamic Analysis of a Long Span, Cable-Stayed
Freeway Bridge Using NASTRAN, NASTRAN: Users’ Exper., pp. 143-168,
September, 1973, (NASA TM X-2893).

Sauer, G.; Wolf, M. Gyroscopic Effects in the Dynamic Response of Rotating Structures,
Proc. of the MSC/NASTRAN Eur. Users’ Conf., Paper No. 11, May, 1986.
Schips, C. Aero-Engine Turbine Dynamic Analysis, Proc. of the 18th MSC Eur. Users’
Conf., Paper No. 8, June, 1991.

Schmitz, Ronald P. Structural Dynamic Analysis of Electronic Assemblies Using

NASTRAN Restart/Format Change Capability, NASTRAN: Users’ Exper.,
pp. 363-392, September, 1971, (NASA TM X-2378).
Schweiger, W.; de Bruyne, F.; Dirschmid, W. Fluid Structure Interaction of Car Fuel Tanks,
Proc. of the MSC/NASTRAN Eur. Users’ Conf., May, 1984.

Shein, Shya-Ling. Generation of the Space Station Freedom On-Orbit Dynamic Loads
Analysis Model Using MSC/NASTRAN V66A Superelements, The 2nd Annual MSC
Taiwan Users Conf., Paper No. 7, October, 1990.
Shiraki, K.; Hashimoto, H.; Sato, N.; Nasu, S.; Kinno, M. Japanese Experiment Module
(JEM): On-Orbit Structural Dynamic Analysis, 1993 MSC Japan’s 11th User’s Conf.
Proc., Paper No. 10.
Shivaji, M.; Raju, V. S. N. Dynamic Analysis of R. C. C. Chimneys, MSC 1995 World Users’
Conf. Proc., Paper No. 34, May, 1995.
 CHAPTER L 507
References and Bibliography

Singh, Ashok K.; Nichols, Christian W. Derivation of an Equivalent Beam Model From a
Structural Finite Element Model, The MSC 1988 World Users Conf. Proc., Vol. I,
Paper No. 14, March, 1988.
Singh, Sudeep K.; Engelhardt, Charlie. Dynamic Analysis of a Large Space Structure
Using External and Internal Superelements, The MSC 1991 World Users’ Conf.
Proc., Vol. I, Paper No. 27, March, 1991.
Skattum, Knut S. Modeling Techniques of Thin-Walled Beams with Open Cross Sections,
NASTRAN: Users’ Exper., pp. 179-196, September, 1972, (NASA TM X-2637).

Smith, Michael R.; Rangacharyulu, M.; Wang, Bo P.; Chang, Y. K. Application of

Optimization Techniques to Helicopter Structural Dynamics,
AIAA/ASME/ASCE/AHS/ASC 32nd Structures, Structural Dynamics, and
Materials Conf., Part 1, Paper No. 91-0924, pp. 227-237, April, 1991.
Stockwell, Alan E.; Perez, Sharon E.; Pappa, Richard S. Integrated Modeling and Analysis
of a Space-Truss Test Article, The MSC 1990 World Users Conf. Proc., Vol. I, Paper
No. 16, March, 1990.
Struschka, M.; Goldstein, H. Approximation of Frequency Dependant Nonlinearities in
Linear FE-Models, Proc. of the 15th MSC/NASTRAN Eur. Users’ Conf., October,
1988.
Subrahmanyam, K. B.; Kaza, K. R. V.; Brown, G. V.; Lawrence, C. Nonlinear
Bending-Torsional Vibration and Stability of Rotating, Pre-Twisted, Preconed
Blades Including Coriolis Effects, National Aeronautics and Space Administration,
January, 1986, (NASA TM-87207).

Tecco, T. C. Analyzing Frequency Dependent Stiffness and Damping with

MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc., Paper No. 25, March, 1985.
Thornton, E. A. Application of NASTRAN to a Space Shuttle Dynamics Model,
NASTRAN: Users’ Exper., pp. 421-438, September, 1971, (NASA TM X-2378).

Ting, Tienko. Test/Analysis Correlation for Multiple Configurations, The MSC 1993
World Users’ Conf. Proc., Paper No. 74, May, 1993.

Tinti, Francesco Carlo; Scaffidi, Costantino. Structural Dynamics and Acoustic Design of
Engine Component in View of Exterior Noise Reduction Using Numerical
Techniques, MSC 1995 European Users’ Conf. Proc., Italian Session, September,
1995.
Turner, Patrick Ryan. Integrating Finite Element Analysis with Quasi-Static Loadings
from a Large Displacement Dynamic Analysis, The 1989 MSC World Users Conf.
Proc., Vol. II, Paper No. 37, March, 1989.
Tzong, George T. J.; Sikes, Gregory D.; Dodd, Alan J. Large Order Modal Analysis Module
in the Aeroelastic Design Optimization Program (ADOP), The MSC 1991 World
Users’ Conf. Proc., Vol. II, Paper No. 36, March, 1991.
Unger, B.; Eichlseder, Wilfried; Schuch, F. Predicting the Lifetime of Dynamically Stressed
Components, Proc. of the 20th MSC European Users’ Conf., Paper No. 36,
September, 1993.
508

Vance, Judy; Bernard, James E. Approximating Eigenvectors and Eigenvalues Across a

Wide Range of Design, The MSC 1992 World Users’ Conf. Proc., Vol. II, Paper
No. 46, May, 1992.
Visintainer, Randal H.; Aslani, Farhang. Shake Test Simulation Using MSC/NASTRAN,
MSC 1994 World Users’ Conf. Proc., Paper No. 32, June, 1994.

Vitiello, P.; Quaranta, V. SEA Investigation Via a FEM Based Substructuring Technique,
MSC 1995 European Users’ Conf. Proc., Italian Session, September, 1995.

Walton, William B.; Blakely, Ken. Modeling of Nonlinear Elastic Structures Using
MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc., Paper No. 11, March, 1983.

Wamsler, M.; Blanck, N.; Kern, G. On the Enforced Relative Motion Inside a Structure,
Proc. of the 20th MSC European Users’ Conf., September, 1993.
Wang, B. P.; Caldwell, S. P.; Smith, C. M. Improved Eigensolution Reanalysis Procedures
in Structural Dynamics, The MSC 1990 World Users Conf. Proc., Vol. II, Paper
No. 46, March, 1990.
Wang, B. P.; Chang, Y. K.; Lawrence, K. L.; Chen, T. Y. Optimum Design of Structures with
Multiple Configurations with Frequency and Displacement Constraints, 31st
AIAA/ ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials
Conf., Part 1, pp. 378-384, April, 1990.
Wang, B. P.; Caldwell, S. P. Reducing Truncation Error in Structural Dynamic
Modification, The MSC 1991 World Users’ Conf. Proc., Vol. I, Paper No. 11, March,
1991.

Wang, B. P.; Caldwell, S. P. Improved Approximate Method for Computting Eigenvector

Derivatives, Finite Elements in Analysis and Design v 14 n 4 Nov 1993.
Watanabe, Masaaki. Computation of Virtual Mass to Rigid Body Structure by
MSC/NASTRAN, The First MSC/NASTRAN User’s Conf. in Japan, October, 1983,
in Japanese.
Wijker, J. J. Differential Stiffness in Conjunction with Dynamics, NASTRAN User’s Conf.,
June, 1981.
Wijker, J. J. Acoustic Effects on the Dynamic Behaviour of Lightweight Structures, Proc. of
the MSC/NASTRAN Eur. Users’ Conf., Paper No. 3, April, 1985.
Wilhelmy, Dr. Viktor. Dynamic Analysis with Gaps, The 1989 MSC World Users Conf.
Proc., Vol. II, Paper No. 40, March, 1989.

Woytowitz, P. J.; Jiang, K. C.; Bhat, K. P. Dynamic Analysis of Optical Beam Pointing, The
MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 11, March, 1988.
Yang, Jackson C. S.; Frederick, Diana L. Application of NASTRAN in Nonlinear Analysis
of a Cartridge Case Neck Separation Malfunction, NASTRAN: Users’ Exper.,
pp. 389-396, September, 1975, (NASA TM X-3278).
Young, K. J.; Mitchell, L. D. On the Performance of Various Kinds of Rod and Beam Mass
Matrices on a Plane-Frame Structure, Proc. of IMAC-IX, Vol. I, pp. 1057-1065, April,
1991.
 CHAPTER L 509
References and Bibliography

DYNAMICS – ANALYSIS / TEST CORRELATION

Allen, James J.; Martinez, David R. Techniques for Implementing Structural Model
Identification Using Test Data, Sandia National Laboratories, June, 1990,
(SAND90-1185).

Anker, J. C. Checks that Pay, Proc. of the MSC/NASTRAN Eur. Users’ Conf., May, 1984.
Blakely, Ken; Howard, G. E.; Walton, W. B.; Johnson, B. A.; Chitty, D. E. Comparison of a
Nonlinear Dynamic Model of a Piping System to Test Data, 7th Int. Conf. on Struct.
Mech. in Reactor Tech., August, 1983.
Blakely, Ken; Walton, W. B. Selection of Measurement and Parameter Uncertainties for
Finite Element Model Revision, 2nd Int. Modal Analysis Conf., February, 1984.
Blakely, Ken. Updating MSC/NASTRAN Models to Match Test Data, The MSC 1991
World Users’ Conf. Proc., Vol. II, Paper No. 50, March, 1991.
Blakely, Ken. Revising MSC/NASTRAN Models to Match Test Data, Proc. of the 9th Int.
Modal Analysis Conf., April, 1991.
Blakely, Ken. Get the Model Right, then Run the Analysis, Machine Design, October 24,
1991.
Blakely, Ken; Rose, Ted. Cross-Orthogonality Calculations for Pre-Test Planning and
Model Verification, The MSC 1993 World Users’ Conf. Proc., Paper No. 72, May,
1993.

Blakely, Ken; Bush, Richard. Using MSC/NASTRAN to Match Dynamic Test Data, Proc.
of the Int. Conf. on Structural Dynamics Modelling, July, 1993.
Blakely, Ken; Rose, Ted. Cross-Orthogonality Calculations for Pre-Test Planning and
Model Verification, Proc. of the 20th MSC European Users’ Conf., September, 1993.
Blakely, Ken. Matching Frequency Response Test Data with MSC/NASTRAN, MSC 1994
World Users’ Conf. Proc., Paper No. 17, June 1994.
Blakely, Ken. Matching Frequency Response Test Data with MSC/NASTRAN, Proc. of the
21st MSC European User’s Conf., Italian Session, September, 1994.

Brillhart, Ralph; Hunt, David L.; Kammer, Daniel C.; Jensen, Brent M.; Mason, Donald R.
Modal Survey and Test-Analysis Correlation of the Space Shuttle SRM, Proc. of the
6th Int. Modal Analysis Conf., pp. 863-870, February, 1988.
Brughmans, Marc; Leuridan, Jan; Blauwkamp, Kevin. The Application of FEM-EMA
Correlation and Validation Techniques on a Body-in-White, The MSC 1993 World
Users’ Conf. Proc., Paper No. 6, May, 1993.

Brughmans, M.; Lembregts, PhD. F.; Furini, PhD. F.; Storrer, O. Modal Test on the
Pininfarina Concept Car Body “ETHOS 1", Actes de la 2ème Confèrence Française
Utilisateurs des Logiciels MSC, Toulouse, France, September, 1995.

Brughmans, M.; Lembregts, F, Ph.D.; Furini, F., Ph.D. Modal Test on the Pininfarina
Concept Car Body “ETHOS 1", MSC 1995 World Users’ Conf. Proc., Paper No. 5,
May, 1995.
510

Budynas, R.; Kolhatkar, S. Modal Analysis of a Robot Arm Using Finite Element Analysis
and Modal Testing, Proc. of the 8th Int. Modal Analysis Conf., Vol. I, pp. 67-70,
January, 1990.
Budynas, R. G.; Krebs, D. Modal Correlation of Test and Finite Element Results Using
Cross Orthogonality with a Reduced Mass Matrix Obtained by Modal Reduction
and NASTRAN’s Generalized Dynamic Reduction Solution, Proc. of the 9th Int.
Modal Analysis Conf., Vol. I, pp. 549-554, April, 1991.
Butler, Thomas G. Test vs. Analysis: A Discussion of Methods, Fourteenth NASTRAN
Users’ Colloq., pp. 173-186, May, 1986, (NASA CP-2419).

Call, V.; Mason, D. Space Shuttle Redesigned Solid Rocket Booster Structural Dynamic
Predictions and Correlations of Liftoff, AIAA/SAE/ASME/ASEE 26th Joint
Propulsion Conf., Paper No. AIAA 90-2081, July, 1990.
Chung, Y. T. Model Reduction and Model Correlation Using MSC/NASTRAN, MSC 1995
World Users’ Conf. Proc., Paper No. 8, May, 1995.

Coladonato, Robert J. Development of Structural Dynamic Test Evnironments for

Subsystems and Components, Seventh NASTRAN Users’ Colloq., pp. 85-110,
October, 1978, (NASA CP-2062).
Coppolino, Robert N. Integrated Dynamic Test/Analysis Processor Overview,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 5, March, 1986.

Cronkhite, J. D. Development, Documentation and Correlation of a NASTRAN Vibration

Model of the AH-1G Helicopter Airframe, NASTRAN: Users’ Exper., pp. 273-294,
October, 1976, (NASA TM X-3428).
Dascotte, E.; Von Estorff, O.; Wandinger, J. Validation and Updating of MSC/NASTRAN
Finite Element Models Using Experimental Modal Data, Proc. of the 20th MSC
European Users’ Conf., Paper No. 10, September, 1993.
de Boer, A.; Kooi, B. W. A DMAP for Updating Dynamic Mathematical Models with
Respect to Measured Data, Proc. of the MSC/NASTRAN Eur. Users’ Conf., May,
1986.

Deger, Yasar. Modal Analysis of a Concrete Gravity Dam - Linking FE Analysis and Test
Results, Proc. of the 20th MSC European Users’ Conf., September, 1993.
Deutschel, Brian W.; Katnik, Richard B.; Bijlani, Mohan; Cherukuri, Ravi. Improving
Vehicle Response to Engine and Road Excitation Using Interactive Graphics and
Modal Reanalysis Methods, SAE Trans., Paper No. 900817, September, 1991.

Dirschmid, W.; Nolte, F.; Dunne, L. W. Mathematical Model Updating Using

Experimentally Determined Real Eigenvectors, Proc. of the 17th MSC Eur. Users’
Conf., Paper No. 4, September, 1990.

Drago, Raymond J.; Margasahayam, Ravi. Stress Analysis of Planet Gears with Integral
Bearings; 3-D Finite Element Model Development and Test Validation, The MSC
1987 World Users Conf. Proc., Vol. I, Paper No. 4, March, 1987.
 CHAPTER L 511
References and Bibliography

Ferg, D.; Foote, L.; Korkosz, G.; Straub, F.; Toossi, M.; Weisenburger, R. Plan, Execute, and
Discuss Vibration Measurements, and Correlations to Evaluate a NASTRAN Finite
Element Model of the AH-64 Helicopter Airframe, National Aeronautics and Space
Administration, January, 1990, (NASA CR-181973).
Graves, Roger W. Interfacing MSC/NASTRAN with SDRC-IDEAS to Perform Component
Mode Synthesis Combining Test, Analytical, and F. E. Data, The MSC 1988 World
Users Conf. Proc., Vol. II, Paper No. 58, March, 1988.
Hehta, Pravin K. Correlation of a NASTRAN Analysis with Test Measurements for
HEAO-2 Optics, MSC/NASTRAN Users’ Conf. Proc., Paper No. 17, March, 1984.

Herbert, Andrew A.; Currie, A. O.; Wilson, W. Analysis of Automotive Axle Carrier
Assembly and Comparison with Test Data, The MSC 1987 World Users Conf. Proc.,
Vol. I, Paper No. 6, March, 1987.
Herting, D. N. Parameter Estimation Using Frequency Response Tests, MSC 1994 World
Users’ Conf. Proc., Paper No. 18, June, 1994.

Jiang, K. C. Finite Element Model Updates Using Modal Test Data, The 1989 MSC World
Users Conf. Proc., Vol. II, Paper No. 48, March, 1989.

Kabe, Alvar M. Mode Shape Identification and Orthogonalization,

AIAA/ASME/ASCE/AHS/ASC 29th Structures, Structural Dynamics and
Materials Conf., Paper No. 88-2354, 1988.

Kammer, Daniel C.; Jensen, Brent M.; Mason, Donald R. Test-Analysis Correlation of the
Space Shuttle Solid Rocket Motor Center Segment, J. of Spacecraft, Vol. 26, No. 4,
pp. 266-273, March, 1988.
Kelley, William R.; Isley, L. D. Using MSC/NASTRAN for the Correlation of Experimental
Modal Models for Automotive Powertrain Structures, The MSC 1993 World Users’
Conf. Proc., Paper No. 8, May, 1993.
Kelley, William R.; Isley, L. Dean; Foster, Thomas J. Dynamic Correlation Study Transfer
Case Housings, MSC 1996 World Users’ Conf. Proc., Vol. II, Paper No. 15, June, 1996.

Kientzy, Donald; Richardson, Mark; Blakely, Ken. Using Finite Element Data to Set Up
Modal Tests, Sound and Vibration, June, 1989.

Lammens, Stefan; Brughmans, Marc; Leuridan, Jan; Sas, Paul. Application of a FRF Based
Model Updating Technique for the Validation of Finite Element MOdels of
Components of the Automotive Industry, MSC 1995 World Users’ Conf. Proc.,
Paper No. 7, May, 1995.

Lee, John M.; Parker, Grant R. Application of Design Sensitivity Analysis to Improve
Correlations Between Analytical and Test Modes, The 1989 MSC World Users Conf.
Proc., Vol. I, Paper No. 21, March, 1989.

Linari, M.; Mancino, E. Application of the MSC/NASTRAN Program to the Study of a

Simple Reinforced Concrete Structure in Nonlinear Material Field, Proc. of the 15th
MSC/NASTRAN Eur. Users’ Conf., October, 1988.

Lowrey, Richard D. Calculating Final Mesh Size Before Mesh Completion, The MSC 1990
World Users Conf. Proc., Vol. II, Paper No. 44, March, 1990.
512

Marlow, Jill M.; Lindell, Michael C. NASSTAR: An Instructional Link Between

MSC/NASTRAN and STAR, Proceedings of the 11th International Modal Analysis
Conference, Florida, 1993.
Masse, Barnard; Pastorel, Henri. Stress Calculation for the Sandia 34-Meter Wind Turbine
Using the Local Circulation Method and Turbulent Wind, The MSC 1991 World
Users’ Conf. Proc., Vol. II, Paper No. 53, March, 1991.
Mindle, Wayne L.; Torvik, Peter J. A Comparison of NASTRAN (COSMIC) and
Experimental Results for the Vibration of Thick Open Cylindrical Cantilevered
Shells, Fourteenth NASTRAN Users’ Colloq., pp. 187-204, May, 1986, (NASA
CP-2419).

Morton, Mark H. Application of MSC/NASTRAN for Assurance of Flight Safety and

Mission Effectiveness with Regard to Vibration upon Installation of the Stinger
Missile on the AH-64A, The MSC 1991 World Users’ Conf. Proc., Vol. II, Paper
No. 52, March, 1991.

Neads, M. A.; Eustace, K. I. The Solution of Complex Structural Systems by NASTRAN

within the Building Block Approach, NASTRAN User’s Conf., May, 1979.
Nowak, William. Electro-Mechanical Response Simulation of Electrostatic Voltmeters
Using MSC/NASTRAN, The MSC 1993 World Users’ Conf. Proc., Paper No. 65,
May, 1993.
O’Callahan, Dr. John; Avitabile, Peter; Reimer, Robert. An Application of New Techniques
for Integrating Analytical and Experimental Structural Dynamic Models, The 1989
MSC World Users Conf. Proc., Vol. II, Paper No. 47, March, 1989.

Ott, Walter; Kaiser, Hans-Jurgen; Meyer, Jurgen. Finite Element Analysis of the Dynamic
Behaviour of an Engine Block and Comparison with Experimental Modal Test
Results, The MSC 1990 World Users Conf. Proc., Vol. I, Paper No. 14, March, 1990.
Paolozzi, A. Structural Dynamics Modification with MSC/NASTRAN, Proc. of the 19th
MSC European Users’ Conf., Paper No. 14, September, 1992.
Park, H. B.; Suh, J. K.; Cho, H. G.; Jung, G. S. A Study on Idle Vibration Analysis Technique
Using Total Vehicle Model, MSC 1995 World Users’ Conf. Proc., Paper No. 6, May,
1995.

Parker, Grant R.; Rose, Ted L.; Brown, John J. Kinetic Energy Calculation as an Aid to
Instrumentation Location in Modal Testing, The MSC 1990 World Users Conf.
Proc., Vol. II, Paper No. 47, March, 1990.

Preve, A.; Meneguzzo, M.; Merlo, A.; Zimmer, H. Simulation of Vehicles’ Structural
Noise: Numerical/Experimental Correlation in the Acoustic Simulation of the
Internal Noise, Proc. of the 21st MSC European Users’ Conf., Italian Session,
September, 1994.

Rabani, Hadi. Static and Dynamic FEM/Test Correlation of an Automobile Body, The
1989 MSC World Users Conf. Proc., Vol. II, Paper No. 49, March, 1989.

Rainer, I. G. MSC/NASTRAN as a Key Tool to Satisfy Increasing Demand for Numerical

Simulation Techniques, Proc. of the 20th MSC European Users’ Conf., September,
1993.
 CHAPTER L 513
References and Bibliography

Ray, William F. The Use of MSC/NASTRAN and Empirical Data to Verify a Design,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 11, March, 1984.

Scapinello, F.; Colombo, E. An Approach for Detailed Analysis of Complex Structures

Avoiding Complete Models, Proc. of the MSC/NASTRAN Eur. Users’ Conf., May,
1987.

Sok-chu, Park; Ishii, Tetsu; Honda, Shigeki; Nagamatsu, Akio. Vibration Analysis and
Optimum Design of Press Machines, 1994 MSC Japan Users’ Conf. Proc.

Stack, Charles P.; Cunningham, Timothy J. Design and Analysis of Coriolis Mass
Flowmeters Using MSC/NASTRAN, The MSC 1993 World Users’ Conf. Proc., Paper
No. 54, May, 1993.

Su, Hong. Structural Analysis of Ka-BAND Gimbaled Antennas for a Communications

Satellite System, MSC 1996 World Users’ Conf. Proc., Vol. IV, Paper No. 33, June,
1996.

Tawekal, Ricky; Budiyanto, M. Agus. Finite Element Model Correlation for Structures,
The MSC 1993 World Users’ Conf. Proc., Paper No. 73, May, 1993.
Ting, T.; Ojalvo, I. U. Dynamic Structural Correlation via Nonlinear Programming
Techniques, The MSC 1988 World Users Conf. Proc., Vol. II, Paper No. 57, March,
1988.
Ting, Tienko; Chen, Timothy L. C. FE Model Refinement with Actual Forced Responses of
Aerospace Structures, The MSC 1991 World Users’ Conf. Proc., Vol. II, Paper No. 51,
March, 1991.

Ting, Tienko. Test/Analysis Correlation for Multiple Configurations, The MSC 1993
World Users’ Conf. Proc., Paper No. 74, May, 1993.
Ujihara, B. H.; Dosoky, M. M.; Tong, E. T. Improving a NASTRAN Dynamic Model with
Test Data Using Linwood, Tenth NASTRAN Users’ Colloq., pp. 74-86, May, 1982,
(NASA CP-2249).
Zeischka, H. LMS/Link Correlating and Validating F.E.A. for Dynamic Structure
Behaviour with Experimental Modal Analysis, Proc. of the MSC/NASTRAN Eur.
Users’ Conf., May, 1987.

DYNAMICS – COMPONENT MODE SYNTHESIS

Barnett, Alan R.; Ibrahim, Omar M.; Sullivan, Timothy L.; Goodnight, Thomas W.
Transient Analysis Mode Participation for Modal Survey Target Mode Selection
Using MSC/NASTRAN DMAP, MSC 1994 World Users’ Conf. Proc., Paper No. 8,
June, 1994.
Bedrossian, Herand; Rose, Ted. DMAP Alters for Nonlinear Craig-Bampton Component
Modal Synthesis, The MSC 1993 World Users’ Conf. Proc., Paper No. 25, May, 1993.

Brillhart, Ralph; Hunt, David L.; Kammer, Daniel C.; Jensen, Brent M.; Mason, Donald R.
Modal Survey and Test-Analysis Correlation of the Space Shuttle SRM, Proc. of the
6th Int. Modal Analysis Conf., pp. 863-870, February, 1988.
514

Brown, J. J.; Lee, J. M.; Parker, G. R.; Zuhuruddin, K. Procedure for Processing and
Displaying Entire Physical Modes Based on Results Generated Through
Component Mode Synthesis, The MSC 1988 World Users Conf. Proc., Vol. I, Paper
No. 15, March, 1988.
Call, V.; Mason, D. Space Shuttle Redesigned Solid Rocket Booster Structural Dynamic
Predictions and Correlations of Liftoff, AIAA/SAE/ASME/ASEE 26th Joint
Propulsion Conf., Paper No. AIAA 90-2081, July, 1990.
Cifuentes, A. O.; Herting, D. N. Transient Response of a Beam to a Moving Mass Using a
Finite Element Approach, Innovative Numerical Methods in Engineering, Proc. of
the Fourth Int. Symp. on Numerical Methods in Engineering, Springer-Verlag,
pp. 533-539, March, 1986.

Cifuentes, A. O. Non-Linear Dynamic Problems Using a Combined Finite Element-Finite

Difference Technique, Proc. of the 6th Conf. on the Mathematics of Finite Elements
and Application, April/May, 1987.

Carney, Kelly S.; Abdallah, Ayma A.; Hucklebridge, Arthur A. Implementation of the
Block-Krylov Boundary Flexibility Method of Component Synthesis, The MSC
1993 World Users’ Conf. Proc., Paper No. 26, May, 1993.
Del Basso, Steve; Singh, Sudeep; Lindenmoyer, Alan J. Component Mode Synthesis of
Space Station Freedom Using MSC/NASTRAN Superelement Architecture, The
MSC 1990 World Users Conf. Proc., Vol. II, Paper No. 48, March, 1990.
Duncan, Alan E. Application of Modal Modeling and Mount System Optimization to
Light Duty Truck Ride Analysis, 4th Int. Conf. on Veh. Struct. Mech., pp. 113-128,
November, 1981, (SAE #811313).

Flanigan, Christopher C.; Abdallah, Ayman; Manella, Richard. Implementation of the

Benfield-Hruda Modal Synthesis Method in MSC/NASTRAN, The MSC 1992 World
Users’ Conf. Proc., Vol. I, Paper No. 11, May, 1992.
Garnek, Michael. Large Space Structure Analysis Using Substructure Modal Test Data,
AIAA/ASME/ASCE/AHS 25th Structures, Structural Dynamics, and Materials
Conf., AIAA Paper 84-0942-CP, May, 1984.

Ghosh, Tarun. MSC/NASTRAN Based Component Mode Synthesis Analysis Without the
Use of DMAPS, MSC 1996 World Users’ Conf. Proc., Vol. II, Paper No. 18, June,
1996.
Gieseke, R. K. Analysis of Nonlinear Structures via Mode Synthesis, NASTRAN: Users’
Exper., pp. 341-360, September, 1975, (NASA TM X-3278).

Graves, Roger W. Interfacing MSC/NASTRAN with SDRC-IDEAS to Perform Component

Mode Synthesis Combining Test, Analytical, and F. E. Data, The MSC 1988 World
Users Conf. Proc., Vol. II, Paper No. 58, March, 1988.

Halcomb, J. R. Application of Component Modes to the Analysis of a Helicopter, Proc. of

the MSC/NASTRAN Users’ Conf., March, 1979.

Hambric, Stephen A. Power Flow and Mechanical Intensity Calculations in Structural

Finite Element Analysis, ASME J. of Vibration and Acoustics, Vol. 112, pp. 542-549,
October, 1990.
 CHAPTER L 515
References and Bibliography

Herting, David N.; Hoesly, R. L. Development of an Automated Multi-Stage Modal

Synthesis System for NASTRAN, Sixth NASTRAN Users’ Colloq., pp. 435-448,
October, 1977, (NASA CP-2018).
Herting, David N. Accuracy of Results with NASTRAN Modal Synthesis, Seventh
NASTRAN Users’ Colloq., pp. 389-404, October, 1978, (NASA CP-2062).

Herting, D. N. A General Purpose, Multi-Stage, Component Modal Synthesis Method,

Finite Elements in Analysis and Design, Vol. 1, No. 2, 1985.

Hill, R. G.; Merckx, K. R. Seismic Response Evaluation of a Reactor Core Using Modal
Synthesis, Proc. of the 3rd Int. Modal Analysis Conf., Vol. 2, pp. 996-1000, June,
1985.

Hodgetts, P. A.; Maitimo, F. M.; Wijker, J. J. Dynamic Analysis of the Polar Platform Solar
Array Using a Multilevel Component Mode Synthesis Technique, Proc. of the 19th
MSC European Users’ Conf., Paper No. 15, September, 1992.

Ichikawa, Tetsuji; Hagiwara, Ichiro. Frequency Response Analysis of Large-Scale Damped

Structures Using Component Mode Synthesis, Nippon Kikai Gakkai Ronbunshu, C
Hen/Transactions of the Japan Society of Mechanical Engineers, Part C v 60 n 569
Jan 1994.
Jasuja, S. C.; Borowski, V. J.; Anderson, D. H. Finite Element Modeling Techniques for the
Simulation of Automotive Powertrain Dynamics, Proc. of the 5th Int. Modal
Analysis Conf., Vol. II, pp. 1520-1530, April, 1987.
Jay, Andrew; Lewis, Bryan; Stakolich, Ed. Effect of Time Dependent Flight Loads on
Turbofan Engine Performance Deterioration, ASME J. of Engineering for Power,
Vol. 104, No. 3, July, 1982.
Kammer, Daniel C.; Jensen, Brent M.; Mason, Donald R. Test- Analysis Correlation of the
Space Shuttle Solid Rocket Motor Center Segment, J. of Spacecraft, Vol. 26, No. 4,
pp. 266-273, March, 1988.

Kasai, Manabu. Generalized CMS Employing External Superelements, The Fourth

MSC/NASTRAN User’s Conf. in Japan, October, 1986, in Japanese.

Kasai, Manabu. Representation of DMAP by Generalized CMS in System Identification,

The Fourth MSC/NASTRAN User’s Conf. in Japan, October, 1986, in Japanese.
Kasai, Manabu. Approach to CMS Subjected to the Boundary Constrained at Single Point,
The Sixth MSC/NASTRAN User’s Conf. in Japan, October, 1988, in Japanese.

Kim, Hyoung M.; Bartkowicz, Theodoore J.; Van Horn, David A. Data Recovery and
Model Reduction Methods for Large Structures, The MSC 1993 World Users’ Conf.
Proc., Paper No. 23, May, 1993.

Kubota, Minoru. Efficient Use of Component Mode Sysnthesis Using Image Superelements
Applied to Dynamic Analysis of Crankshaft, MSC/NASTRAN Users’ Conf. Proc.,
Paper No. 22, March, 1986.
Lee, W. M. Substructure Mode Synthesis with External Superelement, The 2nd Annual
MSC Taiwan User’s Conf., Paper No. 16, October, 1990, in Chinese.

MacNeal, R. H. A Hybrid Method of Component Mode Synthesis, Computers and

Structures, Vol. 1, No. 4, pp. 581-601, 1971.
516

Martinez, D. R.; Gregory, D. L. A Comparison of Free Component Mode Synthesis

Techniques Using MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc., Paper
No. 18, March, 1983.
Martinez, David R.; Gregory, Danny L. A Comparison of Free Component Mode Synthesis
Techniques Using MSC/NASTRAN, Sandia National Laboratories, June, 1984,
(SAND83-0025).
Murakawa, Osamu. Hull Vibration Analysis by Modal Synthesis Method, The First
MSC/NASTRAN User’s Conf. in Japan, October, 1983, in Japanese.

Murakawa, Osamu; Iwahashi, Yoshio; Sakato, Tsuneo. Ship Vibration Analysis Using
Modal Synthesis Technique, MSC/NASTRAN Users’ Conf. Proc., Paper No. 23,
March, 1984.
Nasu, Syouichi. Modal Synthesis of Experimental Vibration Characteristics Using
MSC/NASTRAN Results as the Reference Model, The Sixth MSC/NASTRAN User’s
Conf. in Japan, October, 1988, in Japanese.

Nefske, D. J.; Sung, S. H.; Duncan, A. E. Applications of Finite Element Methods to Vehicle
Interior Acoustic Design, Proc. of the 1984 Noise and Vibration Conf., Paper
No. 840743, 1984.
Ookuma, Masaaki; Nagamatsu, Akio. Comparison of Component Mode Synthesis Method
with MSC-NASTRAN, Nippon Kikai Gakkai Ronbunshu, C Hen, Vol. 49, No. 446,
pp. 1883-1889, October, 1983, in Japanese.
Ookuma, Masaaki; Nagamatsu, Akio. Comparison of Component Mode Synthesis Method
with MSC-NASTRAN, Bulletin of the JSME, Vol. 27, No. 228, pp. 1294-1298, June,
1984.
Parekh, Jatin C.; Harris, Steve G. The Application of the Ritz Procedure to Damping
Prediction Using a Modal Strain Energy Approach, Damping ’89, Paper No. CCB,
November, 1989.

Philippopoulos, V. G. Dynamic Analysis of an Engine-Transmission

Assembly-Superelement and Component Mode Synthesis, Proc. of the Conf. on
Finite Element Methods and Technology, Paper No. 3, March, 1981.
Reyer, H. Modal Synthesis with External Superelements in MSC/NASTRAN, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., May, 1984.
Sabahi, Dara; Rose, Ted. Special Applications of Global-Local Analysis, The MSC 1990
World Users Conf. Proc., Vol. II, Paper No. 49, March, 1990.

Sabahi, Dara; Rose, Ted. MSC/NASTRAN Superelement Analysis of the NASA/AMES

Pressurized Wind Tunnel, The MSC 1990 World Users Conf. Proc., Vol. II, Paper
No. 50, March, 1990.

Shein, Shya-Ling; Marquette, Brian; Rose, Ted. Superelement Technology Application and
Development in Dynamic Analysis of Large Space Structures, The MSC 1991 World
Users’ Conf. Proc., Vol. I, Paper No. 26, March, 1991.

Suzukiri, Yoshihiro. Component Mode Synthesis Application of MSC/NASTRAN V66, The

2nd Annual MSC Taiwan Users Conf., Paper No. 10, October, 1990.
 CHAPTER L 517
References and Bibliography

Suzukiri, Yoshihiro. Component Mode Synthesis Application of MSC/NASTRAN V66,

Proc. of the First MSC/NASTRAN Users’ Conf. in Korea, Paper No. 17, October,
1990.
Tong, Edward T.; Chang, Craig C. J. An Efficient Procedure for Data Recovery of a
Craig-Bampton Component, MSC 1994 World Users’ Conf. Proc., Paper No. 26, June
1994.
Wamsler, M.; Komzsik, L.; Rose, T. Combination of Quasi-Static and Dynamic System
Mode Shapes, Proc. of the 19th MSC European Users’ Conf., Paper No. 13,
September, 1992.

Wang, Bo Ping. Synthesis of Structures with Multiple Frequency Constraints,

AIAA/ASME/ASCE/AHS 27th Structures, Structural Dynamics and Materials
Conf., Part 1, pp. 394-397, May, 1986.

DYNAMICS – DAMPING
El Maddah, M.; Imbert, J. F. A Comparison of Damping Synthesis Methods for Space
Vehicle Dynamic Analysis, NASTRAN User’s Conf., May, 1979.
Everstine, Gordon C.; Marcus, Melvyn S. Finite Element Prediction of Loss Factors for
Structures with Frequency- Dependent Damping Treatments, Thirteenth
NASTRAN Users’ Colloq., pp. 419-430, May, 1985, (NASA CP-2373).
Gibson, W. C.; Johnson, C. D. Optimization Methods for Design of Viscoelastic Damping
Treatments, ASME Design Engineering Division Publication, Vol. 5, pp. 279-286,
September, 1987.
Gibson, Warren C.; Austin, Eric. Analysis and Design of Damped Structures Using
MSC/NASTRAN, The MSC 1992 World Users’ Conf. Proc., Vol. I, Paper No. 25,
May, 1992.

Johnson, Conor D.; Keinholz, David A. Prediction of Damping in Structures with

Viscoelastic Materials Using MSC/NASTRAN, MSC/NASTRAN Users’ Conf.
Proc., Paper No. 17, March, 1983.
Kalinowski, A. J. Modeling Structural Damping for Solids Having Distinct Shear and
Dilational Loss Factors, Seventh NASTRAN Users’ Colloq., pp. 193-206, October,
1978, (NASA CP-2062).

Kalinowski, A. J. Solution Sensitivity and Accuracy Study of NASTRAN for Large

Dynamic Problems Involving Structural Damping, Ninth NASTRAN Users’
Colloq., pp. 49-62, October, 1980, (NASA CP-2151).
Kienholz, Dave K.; Johnson, Conor D.; Parekh, Jatin C. Design Methods for
Viscoelastically Damped Sandwich Plates, AIAA/ASME/ASCE/AHS 24th
Structures, Structural Dynamics and Materials Conf., Part 2, pp. 334-343, May, 1983.
Li, Tsung-hsiun; Bernard, James. Optimization of Damped Structures in the Frequency
Domain, The MSC 1993 World Users’ Conf. Proc., Paper No. 28, May, 1993.
Lu, Y. P.; Everstine, G. C. More on Finite Element Modeling of Damped Composite
Systems, J. of Sound and Vibration, Vol. 69, No. 2, pp. 199-205, 1980.
518

Mace, M. Damping of Beam Vibrations by Means of a Thin Constrained Viscoelastic

Layer: Evaluation of a New Theory, Journal of Sound and Vibration v 172 n 5 May
19 1994.
Merchant, D. H.; Gates, R. M.; Ice, M. W.; Van Derlinden, J. W. The Effects of Localized
Damping on Structural Response, NASTRAN: Users’ Exper., pp. 301-320,
September, 1975, (NASA TM X-3278).
Parekh, Jatin C.; Harris, Steve G. The Application of the Ritz Procedure to Damping
Prediction Using a Modal Strain Energy Approach, Damping ’89, Paper No. CCB,
November, 1989.

Rose, Ted. DMAP Alters to Apply Modal Damping and Obtain Dynamic Loading Output
for Superelements, The MSC 1993 World Users’ Conf. Proc., Paper No. 24, May,
1993.
Rose, Ted.; McNamee, Martin. A DMAP Alter to Allow Amplitude-Dependent Modal
Damping in a Transient Solution, MSC 1996 World Users’ Conf. Proc., Vol. V, Paper
No. 50, June, 1996.
Shieh, Rong C. A Superefficient, MSC/NASTRAN-Interfaced Computer Code System for
Dynamic Response Analysis of Nonproportionally Damped Elastic Systems, The
MSC 1993 World Users’ Conf. Proc., Paper No. 14, May, 1993.
Tonin, Renzo. Vibration Isolation of Impacts in High-Rise Structures, The Second
Australasian MSC Users Conf., Paper No. 11, November, 1988.

DYNAMICS – FREQUENCY RESPONSE

Balasubramanian, B.; Wamsler, M. Identification of Contributing Modes in
MSC/NASTRAN Modal Frequency Response Analyses, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., May, 1987.
Barnett, Alan R.; Widrick, Timothy W.; Ludwiczak, Damian R. Combining Acceleration
and Displacement Dependent Modal Frequency Responses Using An
MSC/NASTRAN DMAP Alter, MSC 1996 World Users’ Conf. Proc., Vol. II, Paper
No. 17, June, 1996.

Bellinger, Dean. Dynamic Analysis by the Fourier Transform Method with

MSC/NASTRAN, MSC 1995 World Users’ Conf. Proc., Paper No. 10, May, 1995.
Bianchini, Emanuele; Marulo, Francesco; Sorrentino, Assunta. MSC/NASTRAN Solution
of Structural Dynamic Problems Using Anelastic Displacement Fields, Proceedings
of the 36th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and
Materials Conference and AIAA/ASME Adpative Structures Forum, Part 5 (of 5),
New Orleans, 1995.
Blakely, Ken. Matching Frequency Response Test Data with MSC/NASTRAN, MSC 1994
World Users’ Conf. Proc., Paper No. 17, June 1994.
Carlson, David L.; Shipley, S. A.; Yantis, T. F. Procedure for FRF Model Tuning in
MSC/NASTRAN, The MSC 1993 World Users’ Conf. Proc., Paper No. 71, May, 1993.

Charron, Francois; Donato, Louis; Fontaine, Mark. Exact Calculation of Minimum Margin
of Safety for Frequency Response Analysis Stress Results Using Yielding or Failure
Theories, The MSC 1993 World Users’ Conf. Proc., Paper No. 5, May, 1993.
 CHAPTER L 519
References and Bibliography

de la Fuente, E.; San Millán, J. Calculation within MSC/NASTRAN of the Forces

Transmitted by Multipoint Constraints (MPC) and the Forces Generated in Support
Constraints, MSC 1996 World Users’ Conf. Proc., Vol. II, Paper No. 20, June, 1996.
Herbert, S.; Janavicius, P. MSC/NASTRAN Frequency Response Analysis of the Raven
Army Communication Shelter, The Third Australasian MSC Users Conf. Proc.,
Paper No. 14, November, 1989.
Herting, D. N. Parameter Estimation Using Frequency Response Tests, MSC 1994 World
Users’ Conf. Proc., Paper No. 18, June, 1994.

Kajiwara, Itsurou; Nagamatsu, Akio; Seto, Kazuto. New Theory for Elimination of
Resonance Peak and Optimum Design of Optical Servosystem, 1994 MSC Japan
Users’ Conf. Proc., Paper No. 1.
Liew, K. M.; Jiang, L.; Lim, M. K.; Low, S. C. Numerical Evaluation of Frequency Responses
for Delaminated Honeycomb Structures, Computers and Structures v 55 n 2 Apr 17
1995.

Parker, G. R.; Brown, J. J. Evaluating Modal Contributors in a NASTRAN Frequency

Response Analysis, MSC/NASTRAN Users’ Conf. Proc., Paper No. 14, March, 1983.

Rose, Ted. Using Optimization in MSC/NASTRAN to Minimize Response to a Rotating

Imbalance, 1994 MSC Japan Users’ Conf. Proc., Paper No. 28, December 1994.
Shieh, Rong C. A Superefficient, MSC/NASTRAN-Interfaced Computer Code System for
Dynamic Response Analysis of Nonproportionally Damped Elastic Systems, The
MSC 1993 World Users’ Conf. Proc., Paper No. 14, May, 1993.

Soni, Ravi, et al. Development of a Methodology to Predict the Road Noise Performance
Characteristics, The MSC 1993 World Users’ Conf. Proc., Paper No. 9, May, 1993.
Tsutsui, Keicchiro; Nogami, Ray. Development of a Nonlinear Frequency Response
Program for Simulating Vehicle Ride Comfort, MSC 1995 World Users’ Conf. Proc.,
Paper No. 37, May, 1995.
Visintainer, Randal H.; Aslani, Farhang. Shake Test Simulation Using MSC/NASTRAN,
MSC 1994 World Users’ Conf. Proc., Paper No. 32, June, 1994.
Wamsler, Manfred; Krusemann, Rolf. Calculating and Interpreting Contact Forces
Between Brake Disc and Linings in Frequency Response Analysis, Proc. of the 18th
MSC Eur. Users’ Conf., Paper No. 7, June, 1991.

Yen, K. Z. Y.; Hsueh, W. C.; Hsui, T. C. Chatter Suppression of a CNC Lathe in

Inside-Diameter Cutting, The Sixth Annual MSC Taiwan Users’ Conf. Proc., Paper
No. 4, November, 1994.

DYNAMICS – MODES, FREQUENCIES, AND VIBRATIONS

(Korean). Structural Analysis of Solar Array Substate, MSC 1994 Korea Users’ Conf. Proc.,
December, 1994, in Korean.
(Korean). Optimal Design of Chip Mounter Considering Dynamic Characteristics, MSC
1994 Korea Users’ Conf. Proc., December, 1994, in Korean.
520

(Korean). A Study on Vibration Characteristics of a Steering Wheel According to

Geometric Variations, MSC 1994 Korea Users’ Conf. Proc., December, 1994, in
Korean.
(Korean). A Study on Idle Vibration Analysis Technique Using Total Vehicle Model, MSC
1994 Korea Users’ Conf. Proc., December, 1994, in Korean.

(Korean). Vibration Analysis for Outercase in Drum Washer and Floor, MSC 1994 Korea
Users’ Conf. Proc., December, 1994, in Korean.

Ahmad, M. Fouad; Guile, Carl W. Analysis of Coupled Natural Frequencies of

Thin-Walled Beams with Open Cross Sections Using MSC/NASTRAN, The MSC
1990 World Users Conf. Proc., Vol. I, Paper No. 15, March, 1990.

Allen, James J.; Martinez, David R. Techniques for Implementing Structural Model
Identification Using Test Data, Sandia National Laboratories, June, 1990,
(SAND90-1185).

Arakawa, H.; Murakami, T.; Ito, H. Vibration Analysis of the Turbine Generator Stator
Frame, The MSC 1988 World Users Conf. Proc., Vol. II, Paper No. 43, March, 1988.
Armand, Sasan; Lin, Paul. Influence of Mass Moment of Inertia on Normal Modes of
Preloaded Solar Array Mast, The MSC 1992 World Users’ Conf. Proc., Vol. I, Paper
No. 12, May, 1992.
Arora, Tejbir; Birmingham, Lily. Application of MSC/NASTRAN Superelement Dynamic
Reduction Techniques for the Vertical Launching System, The MSC 1988 World
Users Conf. Proc., Vol. I, Paper No. 39, March, 1988.

Barnes, R. A.; Schmid, R.; Adrick, H. C. Rotor Dynamic Analysis with MSC/NASTRAN
via the Important Modes Method, The 1989 MSC World Users Conf. Proc., Vol. I,
Paper 13, March, 1989.

Barnett, Alan R.; Abdallah, Ayma A.; Ibrahim, Omar M.; Manella, Richard T. Solving
Modal Equations of Motion with Initial Conditions Using MSC/NASTRAN DMAP
Part 1: Implementing Exact Mode Superposition, The MSC 1993 World Users’ Conf.
Proc., Paper No. 12, May, 1993.

Barnett, Alan R.; Abdallah, Ayma A.; Ibrahim, Omar M.; Sullivan, Timothy L. Solving
Modal Equations of Motion with Initial Conditions Using MSC/NASTRAN DMAP
Part 2: Coupled vs. Uncoupled Integration, The MSC 1993 World Users’ Conf. Proc.,
Paper No. 13, May, 1993.
Bella, David F.; Steinhard, E. Critical Frequency Determination of a Flexible Rotating
Structure Attached to a Flexible Support, Proc. of the 18th MSC Eur. Users’ Conf.,
Paper No. 28, June, 1991.
Bella, David; Hartmueller, Hans; Muehlenfeld, Karsten; Tokar, Gabriel. Identification of
Critical Speeds of Rotors Attached to Flexible Supports, The MSC 1993 World
Users’ Conf. Proc., Paper No. 34, May, 1993.
Blakely, Ken; Rose, Ted. Cross-Orthogonality Calculations for Pre-Test Planning and
Model Verification, The MSC 1993 World Users’ Conf. Proc., Paper No. 72, May,
1993.
 CHAPTER L 521
References and Bibliography

Blakely, Ken; Rose, Ted. Cross-Orthogonality Calculations for Pre-Test Planning and
Model Verification, Proc. of the 20th MSC European Users’ Conf., September, 1993.

Brughmans, M.; Lembregts, PhD. F.; Furini, PhD. F.; Storrer, O. Modal Test on the
Pininfarina Concept Car Body “ETHOS 1", Actes de la 2ème Confèrence Française
Utilisateurs des Logiciels MSC, Toulouse, France, September, 1995.

Brughmans, M.; Lembregts, F, Ph.D.; Furini, F., Ph.D. Modal Test on the Pininfarina
Concept Car Body “ETHOS 1", MSC 1995 World Users’ Conf. Proc., Paper No. 5,
May, 1995.

Buchanan, Guy. Superelement Data Recovery via the Modal Acceleration Method, The
MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 40, March, 1988.

Budynas, R.; Kolhatkar, S. Modal Analysis of a Robot Arm Using Finite Element Analysis
and Modal Testing, Proc. of the 8th Int. Modal Analysis Conf., Vol. I, pp. 67-70,
January, 1990.

Budynas, R. G.; Krebs, D. Modal Correlation of Test and Finite Element Results Using
Cross Orthogonality with a Reduced Mass Matrix Obtained by Modal Reduction
and NASTRAN’s Generalized Dynamic Reduction Solution, Proc. of the 9th Int.
Modal Analysis Conf., Vol. I, pp. 549-554, April, 1991.
Butler, Thomas G.; Muskivitch, John C. Application of Flanigan’s Mode Acceleration in
MSC/NASTRAN Version 66, The 1989 MSC World Users Conf. Proc., Vol. I, Paper
No. 25, March, 1989.
Caldwell, Steve; Wang, B. P. Application of Approximate Techniques in the Estimation of
Eigenvalue Quality, The MSC 1993 World Users’ Conf. Proc., Paper No. 11, May,
1993.
Campanile, P.; Pisino, E.; Testi, R.; Manzilli, G.; Minen, D. Flexible Structures in Adams
Using Modal Data from NASTRAN, Proc. of the 21st MSC European Users’ Conf.,
Italian Session, September, 1994.

Carlson, Mark. Applications of Finite Element Analysis for an Improved Musical

Instrument Design, MSC 1996 World Users’ Conf. Proc., Vol. I, Paper No. 8, June,
1996.
Carneiro, S. H. S.; Duarte, J. A. A.; Mendonca, C. B. Theoretical and Experimental Modal
Analysis of the VLS (Satellite Launcher Vehicle) Bent, Proc. of the 2nd
MSC/NASTRAN Users’ Conf. in Brazil, Paper No. 3, March, 1993, in Portuguese.
Case, William R. Jr. NASTRAN DMAP Alter for Determining a Local Stiffness
Modification to Obtain a Specified Eigenvalue, NASTRAN: Users’ Exper.,
pp. 269-284, September, 1973, (NASA TM X-2893).
Case, William R. A NASTRAN DMAP Procedure for Calculation of Base Excitation Modal
Participation Factors, Eleventh NASTRAN Users’ Colloq., pp. 113-140, May, 1983.

Cattani, E.; Micelli, D.; Sereni, L.; Cocordano, S. Cylinder Block Eigenfrequencies and
Eigenvectors Prediction with a Linear Brick and Wedge Finite Element Model, Proc.
of the 19th MSC European Users’ Conf., Paper No. 16, September, 1992.
522

Chang, Cuann-yeu; Chang, Yuan-bing. Using MSC/NASTRAN to Obtain Modal

Parameters, The MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 21, March,
1988.
Chargin, M. L.; Dunne, L. W.; Herting, D. N. Nonlinear Dynamics of Brake Squeal, MSC
1996 World Users’ Conf. Proc., Vol. V, Paper No. 47, June, 1996.

Chen, J. T.; Chyuan, S. W.; You, D. W.; Wong, H. T. A New Method for Determining the
Modal Participation Factor in Support Motion Problems Using MSC/NASTRAN,
The Seventh Annual MSC/NASTRAN Users’ Conf. Proc., Taiwan, 1995.

Clary, Robert R. Practical Analysis of Plate Vibrations Using NASTRAN, NASTRAN:

Users’ Exper., pp. 325-342, September, 1971, (NASA TM X-2378).

Cohen, Allan R.; Laurenson, Robert M. Application of a Substructure Technique for

STS/Payload Coupled Modal Analysis, Proc. of the MSC/NASTRAN Users’ Conf.,
March, 1979.

Concilio, A.; Del Gatto, S.; Lecce, L.; Miccoli, G. Simple and Cheap Noise and Vibration
Active Control System Using Collocated Piezoelectric Devices on a Panel,
Proceedings of the 11th International Modal Analysis Conference, Florida, 1993.

Courtney, Roy Leon. NASTRAN Modeling Studies in the Normal-Mode Method and
Normal-Mode Synthesis, NASTRAN: Users’ Exper., pp. 181-200, September, 1971,
(NASA TM X-2378).

Cronkhite, James D.; Smith, Michael R. Experiences in NASTRAN Airframe Vibration

Prediction at Bell Helicopter Textron, American Helicopter Soc. Dynamics
Specialists Mtg., Section 6, Vibrations Session I, Paper No. 1, November, 1989.
Cross, C.; Rao, A. Comparison of Modal Performance of Alternate Compressor Bracket
Design, Seventh Australasian Users Conf. Proc., Sydney, October, 1993.

Deger, Yasar. Modal Analysis of a Concrete Gravity Dam - Linking FE Analysis and Test
Results, Proc. of the 20th MSC European Users’ Conf., September, 1993.
Deutschel, Brian William. A Systematic Approach Using Finite Elements for Improving
Vehicle Ride, CAD/CAM Robotics and Factories of the Future Integration of
Design, Analysis, and Manufacturing (Proc.), Springer-Verlag Berlin, Heidelberg,
Vol. I, pp. 150-154, 1989.
Deutschel, Brian W.; Katnik, Richard B.; Bijlani, Mohan; Cherukuri, Ravi. Improving
Vehicle Response to Engine and Road Excitation Using Interactive Graphics and
Modal Reanalysis Methods, SAE Trans., Paper No. 900817, September, 1991.

Egashira, Yuji. Large Scale Vibration Analysis of Car Body Using Superelement Method,
The Second MSC/NASTRAN User’s Conf. in Japan, October, 1984, in Japanese.

El-Bayoumy, Lotfi. Identification and Correction of Damaging Resonances in Gear

Drives, Gear Technology, Vol. 1, No. 2, pp. 14-19, August/September, 1984.
Ferg, D.; Foote, L.; Korkosz, G.; Straub, F.; Toossi, M.; Weisenburger, R. Plan, Execute, and
Discuss Vibration Measurements, and Correlations to Evaluate a NASTRAN Finite
Element Model of the AH-64 Helicopter Airframe, National Aeronautics and Space
Administration, January, 1990, (NASA CR-181973).
 CHAPTER L 523
References and Bibliography

Flanigan, Chris. Methods for Calculating and Using Modal Initial Conditions in
MSC/NASTRAN, Proc. of the Conf. on Finite Element Methods and Technology,
March, 1980.
Flanigan, C. An Alternate Method for Mode Acceleration Data Recovery in
MSC/NASTRAN, Proc. of the Conf. on Finite Element Methods and Technology,
Paper No. 7, March, 1981.
Flanigan, Christopher C. Efficient and Accurate Procedures for Calculating Data
Recovery Matrices for Superelement Models, The 1989 MSC World Users Conf.
Proc., Vol. II, Paper No. 44, March, 1989.

Flanigan, Christopher C. Implementation of the IRS Dynamic Reduction Method in

MSC/NASTRAN, The MSC 1990 World Users Conf. Proc., Vol. I, Paper No. 13,
March, 1990.
Friberg, Olof; Karlsson, Rune; Akesson, Bengt. Linking of Modal and Finite Elements in
Structural Vibration Analysis, Proc. of the 2nd Int. Modal Analysis Conf. and
Exhibit, Vol. 1, pp. 330-339, February, 1984.
Gallaher, Bruce. Determination of Structural Dynamic Response Sensitivity to Modal
Truncation, MSC/NASTRAN Users’ Conf. Proc., Paper No. 10, March, 1986.
Ghosh, Tarun; Nall, Marsha; Muniz, Ben; Cheng, Joseph. Space Station Solar Array
Pointing System Control/Structure Interaction Study Using CO-ST-IN for Modal
Reduction, The MSC 1993 World Users’ Conf. Proc., Paper No. 68, May, 1993.
Gieseke, R. K. Modal Analysis of the Mated Space Shuttle Configuration, NASTRAN:
Users’ Exper., pp. 221-236, September, 1971, (NASA TM X-2378).
Girard, A.; Boullet, A.; Dardel, R. Dynamic Analysis of a Satellite Using the Normal
Modes of the Appendages, Proc. of the MSC/NASTRAN Eur. Users’ Conf., April,
1985.
Grandle, Robert E.; Rucker, Carl E. Modal Analysis of a Nine-Bay Skin-Stringer Panel,
NASTRAN: Users’ Exper., pp. 343-362, September, 1971, (NASA TM X-2378).

Gupta, Viney K.; Zillmer, Scott D.; Allison, Robert E. Solving Large-Scale Dynamic
Systems Using Band Lanczos Method in Rockwell NASTRAN on Cray X-MP,
Fourteenth NASTRAN Users’ Colloq., pp. 236-246, May, 1986, (NASA CP-2419).
Hardman, E. S. Static and Normal Modes Analysis of an Aircraft Structure Using the
NASTRAN External Superelement Method, Proc. of the MSC/NASTRAN Eur.
Users’ Conf., May, 1986.

Harn, Wen-Ren; Hwang, Chi-Ching. Evaluation of Direct Model Modification Methods

via MSC/NASTRAN DMAP Procedures, The MSC 1990 World Users Conf. Proc.,
Vol. II, Paper No. 43, March, 1990.

Hayashida, Mirihiro. Application of Design Sensitivity Analysis to Reduction of

Vibration of Ship’s Deck Structure, The Sixth MSC/NASTRAN User’s Conf. in
Japan, October, 1988, in Japanese.

Herting, D. N.; Joseph, J. A.; Kuusinen, L. R.; MacNeal, R. H. Acoustic Analysis of Solid
Rocket Motor Cavities by a Finite Element Method, National Aeronautics and
Space Administration, pp. 285-324, September, 1971, (NASA TM X-2378).
524

Herting, David N. Accuracy of Results with NASTRAN Modal Synthesis, Seventh

NASTRAN Users’ Colloq., pp. 389-404, October, 1978, (NASA CP-2062).

Hill, R. G. The Use of MSC/NASTRAN to Determine the Impact Response of a Reactor

Core Due to Seismic Loading, The MSC 1987 World Users Conf. Proc., Vol. I, Paper
No. 26, March, 1987.

Hirano, Tohru. Visual Evaluation Method for the Vibration Analysis Utilizing a Super
Intelligent Color Graphic Display, The First MSC/NASTRAN User’s Conf. in
Japan, October, 1983, in Japanese.

Hsueh, W. c.; Hsui, T. C.; Yen, K. Z. Y. Modal and Frequency Response Analyses of Vertical
Machining Center Structures - VC65, The Fifth Annual MSC Taiwan Users’ Conf.
Proc., November, 1993.
Huang, Jieh-Shan. Detect the Variety of Structural System by the Vibration Test, The 2nd
Annual MSC Taiwan Users Conf., Paper No. 11, October, 1990, in Chinese.

Igarashi, Mitsuo. Eigenvalue Analysis of Shaft Supported by Anti-Isotropic Bearing, The

Fifth MSC/NASTRAN User’s Conf. in Japan, October, 1987, in Japanese.
Ito, Hiroyuki. Application of Modal Analysis Technique for Cars, The Fourth
MSC/NASTRAN User’s Conf. in Japan, October, 1986, in Japanese.
Jabbour, K. N. Normal Mode Analysis of the Radio Astronomy Explorer (RAE) Booms and
Spacecraft, NASTRAN: Users’ Exper., pp. 237-250, September, 1971, (NASA TM
X-2378).
Jasuja, S. C.; Borowski, V. J.; Anderson, D. H. Finite Element Modeling Techniques for the
Simulation of Automotive Powertrain Dynamics, Proc. of the 5th Int. Modal
Analysis Conf., Vol. II, pp. 1520-1530, April, 1987.
Jiang, K. C.; Gahart, R. Analysis and Modal Survey Test of Intelsat VIIA Deployed Solar
Array, MSC 1994 World Users’ Conf. Proc., Paper No. 27, June 1994.
Jiang, L.; Liew, K.M.; Lim, M.K.; Low, S.C. Vibratory Behaviour of Delaminated
Honeycomb Structures: a 3-D Finite Element Modelling, Computers and Structures
v 55 n 5 Jun 3 1995.
Ju, Yeuan Jyh; Ting, Tienko. Modelling and Analysis of an Accelerometer Using
MSC/ARIES and MSC/NASTRAN, MSC 1995 World Users’ Conf. Proc., Paper No.
22, May, 1995.

Kabe, Alvar M. Mode Shape Identification and Orthogonalization,

AIAA/ASME/ASCE/AHS/ASC 29th Structures, Structural Dynamics and
Materials Conf., Paper No. 88-2354, 1988.
Kajiwara, Itsuro; Nagamatsu, Akio. Optimum Design of Structure and Control Systems by
Modal Analysis, 1994 MSC Japan Users’ Conf. Proc.

Kam, T. Y.; Yang, C. M.; Wu, J. H. Determination of Natural Frequencies of Laminated

Composite Space Structures Via The Experiemental and Finite Element Approaches,
The Sixth Annual MSC Taiwan Users’ Conf. Proc., Paper No. 5, November, 1994.
 CHAPTER L 525
References and Bibliography

Kang, J. M.; Kim, J. Y.; Lee, K. J.; Yum, D. J.; Seol, Y. S.; Rashed, S. Kawahara, A. Simulation
of 3-D Sloshing and Structural Response in Ship’s Tanks Taking Account of
Fluid-Structure Interaction, 1994 MSC Japan Users’ Conf. Proc., Paper No. 29,
December 1994.
Kasai, Manabu. Better Accuracy of Response Derived from Modal Analysis, The Second
MSC/NASTRAN User’s Conf. in Japan, October, 1984, in Japanese.
Katnik, Richard B.; Yu, Che-Hsi; Wolf, Walt. Interactive Modal Animation and Structural
Modification, Proc. of the 6th Int. Modal Analysis Conf., Vol. I, pp. 947-952,
February, 1988.

Kelley, William R.; Isley, L. D. Using MSC/NASTRAN for the Correlation of Experimental
Modal Models for Automotive Powertrain Structures, The MSC 1993 World Users’
Conf. Proc., Paper No. 8, May, 1993.
Kientzy, Donald; Richardson, Mark; Blakely, Ken. Using Finite Element Data to Set Up
Modal Tests, Sound and Vibration, June, 1989.

Knott, George; Ishin, Young; Chargin, M. A Modal Analysis of the Violin, The MSC 1988
World Users Conf. Proc., Vol. II, Paper No. 42, March, 1988.

Kodiyalam, Srinivas; Graichen, Catherine M.; Connell, Isobel J.; Finnigan, Peter M. Design
Optimization of Satellite Structures for Frequency, Strength, and Buckling
Requirements, Aerospace Sciences Meeting and Exhibit, AIAA, January, 1993.

Krishnamurthy, Ravi S. Stress and Vibration Analysis of Radial Gas Turbine

Components, Sixteenth NASTRAN Users’ Colloq., pp. 128-137, April, 1988, (NASA
CP-2505).
Lapi, M.; Grangier, H. Modal Effective Parameters: an Application to Shipboard Support
Structures to Reduce Vibrations Transmission, Proc. of the 17th MSC Eur. Users’
Conf., Paper No. 5, September, 1990.
Larkin, Paul A.; Miller, Michael W. MSC/NASTRAN Superelement Analysis,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 10, March, 1982.

Lawrie, Geoff. The Determination of the Normal Modes of a Gliding Vehicle, The Sixth
Australasian MSC Users Conf. Proc., November, 1992.

Lee, Jyh-Chian. Using Residual Vector in MSC/NASTRAN Modal Frequency Response to

Improve Accuracy, The 4th MSC Taiwan Users’ Conf., Paper No. 8, November, 1992,
in Chinese.

Lee, Sang H. Effective Modal Mass for Characterization of Vibration Modes, Proc. of the
Second MSC/NASTRAN Users’ Conf. in Korea, Paper No. 22, October, 1991.
Liepins, Atis A.; Conaway, John H. Application of NASTRAN to Propeller-Induced Ship
Vibration, NASTRAN: Users’ Exper., pp. 361-376, September, 1975, (NASA TM
X-3278).
Lim, Tae W.; Kashangaki, Thomas A. L. Structural Damage Detection of Space Truss
Structures Using Best Achievable Eigenvectors, AIAA Journal v32 n 5 May 1994.

Lipman, Robert R. Computer Animation of Modal and Transient Vibrations, Fifteenth

NASTRAN Users’ Colloq., pp. 111-117, August, 1987, (NASA CP-2481).
526

Liu, Dauh-Churn, Shieh, Niahn-Chung. Vibration Suppression of High Precision Grinding

Machine Using Finite Element Method (MSC/NASTRAN), The Sixth Annual MSC
Taiwan Users’ Conf. Proc., Paper No. 7, November, 1994.
Lu, Y. P.; Killian, J. W.; Everstine, G. C. Vibrations of Three Layered Damped Sandwich
Plate Composites, J. of Sound and Vibration, Vol. 64, No. 1, pp. 63-71, 1979.

Lui, C. Y.; Mason, D. R. Space Shuttle Redesigned Solid Rocket Motor Nozzle Natural
Frequency Variations with Burn Time, AIAA/SAE/ASME 27th Joint Propulsion
Conf., Paper No. AIAA 91-2301, June, 1991.

Lundgren, Gert. MSC’s Solvers Predict the Best Attachment of the Sunbeam Tiger
Fiberglass Front-End, MSC 1996 World Users’ Conf. Proc., Vol. II, Paper No. 16,
June, 1996.
Marcus, Melvyn S. A Finite-Element Method Applied to the Vibration of Submerged
Plates, J. of Ship Research, Vol. 22, No. 2, pp. 94-99, June, 1978.

Marcus, Melvyn S.; Everstine, Gordon C.; Hurwitz, Myles M. Experiences with the
QUAD4 Element for Shell Vibrations, Sixteenth NASTRAN Users’ Colloq.,
pp. 39-43, April, 1988, (NASA CP-2505).

Mase, M.; Saito, H. Application of FEM for Vibrational Analysis of Ground Turbine
Blades, MSC/NASTRAN Users’ Conf., March, 1978.
McMeekin, Michael; Kirchman, Paul. An Advanced Post Processing Methodology for
Viewing MSC/NASTRAN Generated Analyses Results, MSC 1994 World Users’
Conf. Proc., Paper No. 21, June, 1994.

Mei, Chuh; Rogers, James L. Jr. NASTRAN Nonlinear Vibration Analyses of Beam and
Frame Structures, NASTRAN: Users’ Exper., pp. 259-284, September, 1975, (NASA
TM X-3278).

Mei, Chuh; Rogers, James L. Jr. Application of the TRPLT1 Element to Large Amplitude
Free Vibrations of Plates, Sixth NASTRAN Users’ Colloq., pp. 275-298, October,
1977, (NASA CP-2018).

Meyer, Karl A. Normal Mode Analysis of the IUS/TDRS Payload in a Payload

Canister/Transporter Environment, Eighth NASTRAN Users’ Colloq., pp. 113-130,
October, 1979, (NASA CP-2131).
Michiue, Shinsuke. On the Accuracy in Vibration Analysis for Cylindrical
Shell-Comparison Between QUAD4/QUAD8, The First MSC/NASTRAN User’s
Conf. in Japan, October, 1983, in Japanese.

Miller, R. D. Theoretical Analysis of HVAC Duct Hanger Systems, Fifteenth NASTRAN

Users’ Colloq., pp. 222-249, August, 1987, (NASA CP-2481).

Nack, Wayne V.; Joshi, Arun M. Friction Induced Vibration, MSC 1995 World Users’ Conf.
Proc., Paper No. 36, May, 1995.
Nagayasu, Katsuyosi. Method for Prediction of Noise-Oriented Vibration on Pipe Lines
to Refrigerators, The Third MSC/NASTRAN User’s Conf. in Japan, October, 1985,
in Japanese.

Nagendra, Gopal K.; Herting, David N. Design Sensitivity for Modal Analysis, The 1989
MSC World Users Conf. Proc., Vol. I, Paper No. 22, March, 1989.
 CHAPTER L 527
References and Bibliography

Nagy, Lajos I.; Cheng, James; Hu, Yu-Kan. A New Method Development to Predict Brake
Squeal Occurence, MSC 1994 World Users’ Conf. Proc., Paper No. 14, June, 1994.

Neads, M. A.; Eustace, K. I. The Solution of Complex Structural Systems by NASTRAN

within the Building Block Approach, NASTRAN User’s Conf., May, 1979.

Newman, Malcolm; Pipano, Aaron. Fast Modal Extraction in NASTRAN via the FEER
Computer Program, NASTRAN: Users’ Exper., pp. 485-506, September, 1973,
(NASA TM X-2893).

Nishiwaki, Nobukiyo. Coupled Vibration of Rotating Disc and Blades, The Second
MSC/NASTRAN User’s Conf. in Japan, October, 1984, in Japanese.

Overbye, Vern D. MSC/NASTRAN Dynamic Analysis: Modal or Direct?,

MSC/NASTRAN Users’ Conf. Proc., Paper No. 6, March, 1986.
Pamidi, M. R.; Pamidi, P. R. Modal Seismic Analysis of a Nuclear Power Plant Control
Panel and Comparison with SAP IV, NASTRAN: Users’ Exper., pp. 515-530,
October, 1976, (NASA TM X-3428).
Pamidi, P. R. On the Append and Continue Features in NASTRAN, Seventh NASTRAN
Users’ Colloq., pp. 405-418, October, 1978, (NASA CP-2062).
Paolozzi, A. Interfacing MSC/NASTRAN with a Structural Modification Code, Proc. of
the 18th MSC Eur. Users’ Conf., Paper No. 30, June, 1991.
Park, H. B.; Suh, J. K.; Cho, H. G.; Jung, G. S. A Study on Idle Vibration Analysis Technique
Using Total Vehicle Model, MSC 1995 World Users’ Conf. Proc., Paper No. 6, May,
1995.
Parker, G. R.; Brown, J. J. Kinetic Energy DMAP for Mode Identification, MSC/NASTRAN
Users’ Conf. Proc., Paper No. 8, March, 1982.
Parker, G. R.; Brown, J. J. Evaluating Modal Contributors in a NASTRAN Frequency
Response Analysis, MSC/NASTRAN Users’ Conf. Proc., Paper No. 14, March, 1983.

Pulgrano, Louis J.; Masters, Steven G. Self-Excited Oscillation of a 165 Foot Water Tower,
MSC 1995 World Users’ Conf. Proc., Paper No. 32, May, 1995.
Reyer, H. Modal Synthesis with External Superelements in MSC/NASTRAN, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., May, 1984.

Rose, Ted. A Method to Apply Initial Conditions in Modal Transient Solutions, The MSC
1991 World Users’ Conf. Proc., Vol. I, Paper No. 13, March, 1991.
Rose, Ted. Some Suggestions for Evaluating Modal Solutions, The MSC 1992 World Users’
Conf. Proc., Vol. I, Paper No. 10, May, 1992.
Rose, Ted. Using Optimization in MSC/NASTRAN to Minimize Response to a Rotating
Imbalance, MSC 1994 Korea Users’ Conf. Proc., December, 1994.

Rose, Ted. Using Dynamic Optimization to Minimize Driver Response to a Tire

Out-of-Balance, MSC 1994 Korea Users’ Conf. Proc., December, 1994.

Rose, Ted. Using Optimization in MSC/NASTRAN to Minimize Response to a Rotating

Imbalance, 1994 MSC Japan Users’ Conf. Proc., Paper No. 28, December 1994.
528

Rose, Ted L. Using Optimization in MSC/NASTRAN to Minimize Response to a Rotating

Imbalance, The Sixth Annual MSC Taiwan Users’ Conf. Proc., Paper No. D,
November, 1994.
Saito, Hiroshi; Watanabe, Masaaki. Modal Analysis of Coupled Fluid-Structure Response,
MSC/NASTRAN Users’ Conf. Proc. March, 1982.

Salvestro, Livio; Sirocco, Howden; Currie, Andrew. Cyclic Symmetry Analysis of an Air
Blower Fan, Seventh Australasian Users Conf. Proc., Sydney, October, 1993.

Scanlon, Jack; Swan, Jim. A Stand-Alone DMAP Program for Modal Cross-Correlation,
MSC 1995 World Users’ Conf. Proc., Paper No. 40, May, 1995.

Schiavello, D. V.; Sinkiewicz, J. E. DMAP for Determining Modal Participation,

MSC/NASTRAN Users’ Conf. Proc., Paper No. 15, March, 1983.
Schwering, W. Shulze. A DMAP for Identification of Modeshapes, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., June, 1983.
Shalev, D.; Unger, A. Nonlinear Analysis Using a Modal-Based Reduction Technique,
Composite Structures v31 n 4 1995.
Shalev, Doron; Unger, A. Nonlinear Analysis Using a Modal Based Reduction Technique,
The MSC 1993 World Users’ Conf. Proc., Paper No. 51, May, 1993.

Shippen, J. M. Normal Modes Analysis of Spin Stabilised Spacecraft Possessing Cable

Booms, Proc. of the 18th MSC Eur. Users’ Conf., Paper No. 29, June, 1991.

Shirai, Yujiro; Arakawa, Haruhiko; Toda, Nobuo; Taneda, Yuji; Sakura, Kiyoshi. Active
Vibration Control for Aircraft Wing, JSME International Journal, v 36 n 3 Spe 1993.

Shy, Tyson; Hsiu, T. C.; Yen, K. Z. Y. Optimization of Structure Design of a Machining

Center, The Sixth Annual MSC Taiwan Users’ Conf. Proc., Paper No. 6, November,
1994.
Somayajula, Gopichand; Stout, Joseph; Tucker, John. Eigenvalue Reanalysis Using
Subspace Iteration Techniques, The 1989 MSC World Users Conf. Proc., Vol. I,
Paper No. 26, March, 1989.
Stack, Charles P.; Cunningham, Timothy J. Design and Analysis of Coriolis Mass
Flowmeters Using MSC/NASTRAN, The MSC 1993 World Users’ Conf. Proc., Paper
No. 54, May, 1993.

Starnes, James H. Jr. Vibration Studies of a Flat Plate and a Built-Up Wing, NASTRAN:
Users’ Exper., pp. 637-646, September, 1971, (NASA TM X-2378).

Su, Hong. Structural Analysis of Ka-BAND Gimbaled Antennas for a Communications

Satellite System, MSC 1996 World Users’ Conf. Proc., Vol. IV, Paper No. 33, June,
1996.

Subrahmanyam, K. B.; Kaza, K. R. V.; Brown, G. V.; Lawrence, C. Nonlinear Vibration and
Stability of Rotating, Pretwisted, Preconed Blades Including Coriolis Effects, J. of
Aircraft, Vol. 24, No. 5, pp. 342-352, May, 1987.

Sundaram, S. V.; Hohman, Richard L.; Richards, Timothy R. Vibration Modes of a Tire
Using MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc., Paper No. 26, March,
1985.
 CHAPTER L 529
References and Bibliography

Tamba, Richard; Mowbray, Graham; Rao, Ananda. An Effective Method to Increase the
Natural Frequencies of a Transmission Assembly, The Sixth Australasian MSC
Users Conf. Proc., November, 1992.
Tawekal, Ricky; Budiyanto, M. Agus. Finite Element Model Correlation for Structures,
The MSC 1993 World Users’ Conf. Proc., Paper No. 73, May, 1993.

Tawekal, Ricky L.; Miharjana, N. P. Validation of 3650 DWT Semi Containe Ship Finite
Element Model by Full Scale Measurements, MSC 1994 World Users’ Conf. Proc.,
Paper No. 19, June, 1994.

Thornton, Earl A. A NASTRAN Correlation Study for Vibrations of a Cross-Stiffened

Ship’s Deck, NASTRAN: Users’ Exper., pp. 145-160, September, 1972, (NASA TM
X-2637).
Ting, Tienko; Chen, Timothy L. C.; Twomey, William. A Practical Solution to Mode
Crossing Problem in Continuous Iterative Procedure, The 1989 MSC World Users
Conf. Proc., Vol. I, Paper No. 14, March, 1989.

Ting, Tienko; Chen, T.; Twomey, W. Correlating Mode Shapes Based on Modal Assurance
Criterion, The MSC 1992 World Users’ Conf. Proc., Vol. I, Paper No. 21, May, 1992.

Tokuda, Naoaki; Mitikami, Shinsuke; Sakata, Yoshiuki. Accuracy of Vibration Analysis

for Thin Cylindrical Shell by MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc.,
Paper No. 28, March, 1984.

Vaillette, David. Evaluation of the Modal Response of a Pressure Vessel Filled with a
Fluid, The MSC 1991 World Users’ Conf. Proc., Vol. I, Paper No. 24, March, 1991.

Vance, Judy, M.; Bernard, James E. Approximating Eigenvectors and Eigenvalues Across
a Wide Range of Design, Finite Elements in Analysis and Design v 14 n 4 Nov 1993.
Vandepitte, D.; Wijker, J. J.; Appel, S.; Spiele, H. Normal Modes Analysis of Large Models,
and Applications to Ariane 5 Engine Frame, Proc. of the 18th MSC Eur. Users’ Conf.,
Paper No. 6, June, 1991.
Wamsler, M.; Komzsik, L.; Rose, T. Combination of Quasi-Static and Dynamic System
Mode Shapes, Proc. of the 19th MSC European Users’ Conf., Paper No. 13,
September, 1992.

Wang, B. P.; Cheu, T. C.; Chen, T. Y. Optimal Design of Compressor Blades with Multiple
Natural Frequency Constraints, ASME Design Engineering Division Publication,
Vol. 5, pp. 113-117, September, 1987.

Wang, B. P.; Lu, C. M.; Yang, R. J. Topology Optimization Using MSC/NASTRAN, MSC
1994 World Users’ Conf. Proc., Paper No. 12, June, 1994.
Wang, Bo Ping. Minimum Weight Design of Structures with Natural Frequency
Constraints Using MSC/NASTRAN, The MSC 1988 World Users Conf. Proc., Vol. II,
Paper No. 60, March, 1988.
Welte, Y. Vibration Analysis of an 8MW Diesel Engine, Proc. of the MSC/NASTRAN Eur.
Users’ Conf., May, 1986.

West, Timothy S. Approximate Dynamic Model Sensitivity Analysis For Large, Complex
Space Structues, MSC 1996 World Users’ Conf. Proc., Vol. I, Paper No. 6, June, 1996.
530

Wijker, J. J. Substructuring Technique Using the Modal Constraint Method, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., June, 1983.

Wijker, J. J. MSC/NASTRAN Normal Mode Analysis on CRAY Computers, Proc. of the

MSC/NASTRAN Eur. Users’ Conf., June, 1983.

Yang, Howard J. Sorted Output in MSC/NASTRAN, Proc. of the Conf. on Finite Element
Methods and Technology, Paper No. 4, March, 1981.
Yen, K. Z. Y.; Hsueh, W. C.; Hsui, T. C. Chatter Suppression of a CNC Lathe in
Inside-Diameter Cutting, The Sixth Annual MSC Taiwan Users’ Conf. Proc., Paper
No. 4, November, 1994.

Zhu, H.; Knight, D. Finite Element Forced Response Analysis on the Mondeo Front End
Accessory Drive System, Proc. of the 20th MSC European Users’ Conf., September,
1993.

DYNAMICS – RANDOM RESPONSE

Barnett, Alan R.; Widrick, Timothy W.; Ludwiczak, Damian R. Combining Acceleration
and Displacement Dependent Modal Frequency Responses Using An
MSC/NASTRAN DMAP Alter, MSC 1996 World Users’ Conf. Proc., Vol. II, Paper
No. 17, June, 1996.

Chiang, C. K.; Robinson, J. H.; Rizzi, S. A. Equivalent Linearization Solution Sequence for
MSC/NASTRAN, Winter Annual Meeting of the American Society of Mechanical
Engineers, pp. 133-138, November, 1992.

Ciuti, Gianluca. Avionic Equipment Dynamic Analysis, MSC 1995 European Users’ Conf.
Proc., Italian Session, September, 1995.
Coyette, J. P.; Lecomte, C.; von Estorff, O. Evaluation of the Response of a Coupled Elastic
Structure Subjected to Random Mechanical or Acoustical Excitations Using
MSC/NASTRAN and SYSNOISE, MSC European Users’ Conf., Paper No. 21,
September, 1996.

Crispino, Maurizio. A 3-D Model for the Evaluation through Random Analysis of Vertical
Dynamic Overloads in High Speed Railway Lines, MSC 1995 European Users’ Conf.
Proc., Italian Session, September, 1995.

Galletly, Robert; Wagner, R. J.; Wang, G. J.; Zins, John. Random Vibration and Acoustic
Analysis Using ARI RANDOM, a NASTRAN Post Processor, MSC/NASTRAN
Users’ Conf. Proc., Paper No. 26, March, 1984.

Hatheway, A. Random Vibrations in Complex Electronic Structures, MSC/NASTRAN

Users’ Conf. Proc., Paper No. 13, March, 1983.

Hatheway, Alson E. Evaluation of Ceramic Substrates for Packaging of Leadless Chip

Carriers, MSC/NASTRAN Users’ Conf. Proc., Paper No. 16, March, 1982.
Michels, Gregory J. Vibroacoustics Random Response Analysis Methodology, MSC 1995
World Users’ Conf. Proc., Paper No. 9, May, 1995.

Palmieri, F. W. Example Problems Illustrating the Effect of Multiple Cross Correlated

Excitations on the Response of Linear Systems to Gaussian Random Excitations,
The MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 18, March, 1988.
 CHAPTER L 531
References and Bibliography

Palmieri, F. W. A Method for Predicting the Output Cross Power Spectral Density
Between Selected Variables in Response to Arbitrary Random Excitations, The
MSC 1988 World Users Conf. Proc., Vol. I, Paper No. 19, March, 1988.
Parthasarathy, Alwar; Elzeki, Mohamed; Abramovici, Vivianne. PSDTOOL-A DMAP
Enhancement to Harmonic/ Random Response Analysis in MSC/NASTRAN, The
MSC 1993 World Users’ Conf. Proc., Paper No. 36, May, 1993.
Robinson, J. H.; Chiang, C. K.; Rizzi, S. A. Nonlinear Random Response Prediction Using
MSC/NASTRAN, National Aeronautics and Space Administration, Hampton, VA,
Langley Research Center, October, 1993.

Robinson, Jay H.; Chiang, C. K. An Equivalent Linearization Solution Sequence for

MSC/NASTRAN, The MSC 1993 World Users’ Conf. Proc., Paper No. 35, May, 1993.
Schwab, H. L.; Caffrey, J.; Lin, J. Fatigue Analysis Using Random Vibration, MSC 1995
World Users’ Conf. Proc., Paper No. 17, May, 1995.

Shieh, Niahn Chung. Investigation of Swept Sine on Random Load, The 4th MSC Taiwan
Users’ Conf., Paper No. 18, November, 1992, in Chinese.
Zins, J. Random Vibration and Acoustic Analysis Using ARI RANDOM a NASTRAN
Postprocessor, Proc. of the MSC/NASTRAN Eur. Users’ Conf., Paper No. 4, May,
1984.

DYNAMICS – REDUCTION METHODS

Abdallah, Ayman A.; Barnett, Alan R.; Widrick, Timothy W.; Manella, Richard T.; Miller,
Robert P. Stiffness-Generated Rigid-Body Mode Shapes for Lanczos Eigensolution
with Support DOF Via a MSC/NASTRAN DMAP Alter, MSC 1994 World Users’
Conf. Proc., Paper No. 10, June, 1994.

Flanigan, Christopher C. Implementation of the IRS Dynamic Reduction Method in

MSC/NASTRAN, The MSC 1990 World Users Conf. Proc., Vol. I, Paper No. 13,
March, 1990.

Fox, Gary L. Evaluation and Reduction of Errors Induced by the Guyan Transformation,
Tenth NASTRAN Users’ Colloq., pp. 233-248, May, 1982, (NASA CP-2249).

Komzsik, L.; Dilley, G. Practical Experiences with the Lanczos Method, Proc. of the
MSC/NASTRAN Users Conf., Paper No. 13, March, 1985.

Kuang, Jao-Hwa; Lee, Chung-Ying. On a Guyan-Reduction Recycled Eigen Solution

Technique, The 2nd Annual MSC Taiwan Users Conf., Paper No. 13, October, 1990.
Levy, Roy. Guyan Reduction Solutions Recycled for Improved Accuracy, NASTRAN:
Users’ Exper., pp. 201-220, September, 1971, (NASA TM X-2378).
Maekawa, Seiyou. Effect of Guyan Reduction and Generalized Dynamic Reduction, The
Second MSC/NASTRAN User’s Conf. in Japan, October, 1984, in Japanese.

Mera, A. MSC/NASTRAN Normal Mode Analysis with GDR: An Evaluation of

Limitations, MSC/NASTRAN Users’ Conf. Proc., Paper No. 27, March, 1985.

Mera, Andrew. Static Reduction and Symmetry Transformation of Large Finite Element
Models, Proc. of the MSC/NASTRAN Users’ Conf., Paper No. 12, March, 1979.
532

Vandepitte, D.; Wijker, J. J.; Appel, S.; Spiele, H. Normal Modes Analysis of Large Models,
and Applications to Ariane 5 Engine Frame, Proc. of the 18th MSC Eur. Users’ Conf.,
Paper No. 6, June, 1991.
Vollan, Arne; Kaporin, Igor; Babikov, Pavel. Practical Experience with Different Iterative
Solvers for Linear Static and Modal Analysis of Large Finite Element Models, Proc.
of the 21st MSC European Users’ Conf., Italian Session, September, 1994.
Walker, James W. Evaluation of MSC/NASTRAN Generalized Dynamic Reduction and
Response Spectrum Analysis by Comparison with STARDYNE, MSC/NASTRAN
Users’ Conf., March, 1978.

DYNAMICS – RESPONSE SPECTRUM

Barber, Pam; Arden, Kevin. Dynamic Design Analysis Method (DDAM) Using
MSC/NASTRAN, MSC 1994 World Users’ Conf. Proc., Paper No. 31, June, 1994.
Cutting, Fred. Individual Modal Accelerations as the Result of a Shock Response Spectra
Input to a Complex Structure, The 1989 MSC World Users Conf. Proc., Vol. I, Paper
No. 11, March, 1989.

Gassert, W.; Wolf, M. Floor Response Spectra of a Reactor Building Under Seismic
Loading Calculated with a 3-D Building Model, Proc. of the MSC/NASTRAN Eur.
Users’ Conf., June, 1983.

Hirata, M.; Ishikawa, K.; Korosawa, M.; Fukushima, S.; Hoshina, H. Seismic Analysis of
Plutonium Glovebox by MSC/NASTRAN, January, 1993.

Kubota, Minoru. Response Spectrum Analysis of Underground Tank Dome Roof Using
Image Superelement Method, The Fifth MSC/NASTRAN User’s Conf. in Japan,
October, 1987, in Japanese.
Moharir, M. M. Shock Wave Propagation and NASTRAN Linear Algorithms,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 7, March, 1986.
Overbye, Vern D. MSC/NASTRAN Dynamic Analysis: Modal or Direct?,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 6, March, 1986.
Parris, R. A. Aspects of Seismic Analysis Using MSC/NASTRAN, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., Paper No. 7, June, 1983.

Parthasarathy, A. CONSPEC-A DMAP System for Conventional Response-Spectrum

Analysis in MSC/NASTRAN, MSC/NASTRAN Users’ Conf. Proc., Paper No. 8,
March, 1986.
Petteno, L.; Rossetto, P. Tecnomare Experiences in DMAP Processing, Proc. of the
MSC/NASTRAN Eur. Users’ Conf., May, 1984.
Rose, Ted L. Using Superelements for Response Spectrum and Other Handy Alters, The
1989 MSC World Users Conf. Proc., Vol. II, Paper No. 45, March, 1989.

Walker, James W. Evaluation of MSC/NASTRAN Generalized Dynamic Reduction and

Response Spectrum Analysis by Comparison with STARDYNE, MSC/NASTRAN
Users’ Conf., March, 1978.
 CHAPTER L 533
References and Bibliography

DYNAMICS – SEISMIC
Bonaldi, P.; Peano, A.; Ruggeri, G.; Venturuzzo, M. Seismic and Impact Analyses of
Nuclear Island Buildings of Italian Unified Nuclear Design, Proc. of the 15th
MSC/NASTRAN Eur. Users’ Conf., October, 1988.

Burroughs, John W. An Enhancement of NASTRAN for the Seismic Analysis of Structures,

Ninth NASTRAN Users’ Colloq., pp. 79-96, October, 1980, (NASA CP-2151).

Chen, J. T.; Chyuan, S. W.; Yeh, C. S.; Hong, H. K. Comparisons of Analytical Solution and
MSC/NASTRAN Results on Multiple-Support Motion of Long Bridge During
Earthquake, The 4th MSC Taiwan Users’ Conf., Paper No. 20, November, 1992.

Chen, J. T.; Hong, H. K.; Yen, C. S.; Chyuan, S. W. Integral Representations and
Regularizations for a Divergent Series Solution of a Beam Subjected to Support
Motions, Earthquake Engineering and Structural Dynamics, Vol. 25, pp. 909-925,
1996.

Chen, Yohchia. Nonlinear Seismic Analysis of Bridges: Practical Approach and

Comparative Study, The MSC 1993 World Users’ Conf. Proc., Paper No. 57, May,
1993.

Chen, Yohchia. Refined Analysis for Soil-Pipe Systems, MSC 1994 World Users’ Conf.
Proc., Paper No. 38, June, 1994.
Dahlgren, F.; Citrolo, J.; Knutson, D.; Kalish, M. Dynamic Analysis of the BPX Machine
Structure, Proc. of the 14th IEEE/NPSS Symp. on Fusion Engineering, Vol. 1,
pp. 47-49, 1992.

Fallet, P.; Derivery, J. C. MSC/NASTRAN Earthquake Analysis, NASTRAN User’s Conf.,

May, 1979.
Hill, Ronald G. Nonlinear Seismic Analysis of a Reactor Structure with Impact Between
Core Components, NASTRAN: Users’ Exper., pp. 397-418, September, 1975, (NASA
TM X-3278).
Hirata, M.; Ishikawa, K.; Korosawa, M.; Fukushima, S.; Hoshina, H. Seismic Analysis of
Plutonium Glovebox by MSC/NASTRAN, January, 1993.

Liepins, Atis A.; Nazemi, Hamid. Virtual Mass of Fluid in Egg-Shaped Digesters, The MSC
1993 World Users’ Conf. Proc., Paper No. 77, May, 1993.
Nomura, Yoshio. Seismic Response Analysis by MSC/NASTRAN for Coupled Structure -
Ground and Pile, The Fifth MSC/NASTRAN User’s Conf. in Japan, October, 1987,
in Japanese.

Pamidi, M. R.; Pamidi, P. R. Modal Seismic Analysis of a Nuclear Power Plant Control
Panel and Comparison with SAP IV, NASTRAN: Users’ Exper., pp. 515-530,
October, 1976, (NASA TM X-3428).

Tsaur, D. H.; Chyuan, S. W.; Chen, J. T. On the Accuracy of MSC/NASTRAN on Response

of Two-Span Beams to Spatially Varying Seismic Excitation, The 4th MSC Taiwan
Users’ Conf., Paper No. 9, November, 1992.

Yiak, K. C.; Pezeshk, S. Seismic Study of a Cable-Stayed Bridge, Proceedings of the

Structures Congrees ’94, Atlanta, 1994.
534

Zhou, Hongye; Chen, Youping. The Influence of Phase-Difference Effects on Earthquake

Response of Cable-Stayed Bridges, MSC 1994 World Users’ Conf. Proc., Paper No.
37, June, 1994.

DYNAMICS – TRANSIENT ANALYSIS

(Korean). Vibration Analysis for Outercase in Drum Washer and Floor, MSC 1994 Korea
Users’ Conf. Proc., December, 1994, in Korean.

Aslani, Chang; Yatheendar, Manicka; Visintainer, Randal, H.; Rohweder, David S.; Lopez
de Alda, Juan. Simulation of Proving Ground Events for Heavy Truck Cabs Using
Adams, MSC/NASTRAN, and P/FATIGUE, MSC 1994 World Users’ Conf. Proc.,
Paper No. 5, June 1994.
Barber, Pam; Arden, Kevin. Dynamic Design Analysis Method (DDAM) Using
MSC/NASTRAN, MSC 1994 World Users’ Conf. Proc., Paper No. 31, June, 1994.

Barnett, Alan R.; Ibrahim, Omar M.; Sullivan, Timothy L.; Goodnight, Thomas W.
Transient Analysis Mode Participation for Modal Survey Target Mode Selection
Using MSC/NASTRAN DMAP, MSC 1994 World Users’ Conf. Proc., Paper No. 8,
June, 1994.
Bellinger, Dean. Dynamic Analysis by the Fourier Transform Method with
MSC/NASTRAN, MSC 1995 World Users’ Conf. Proc., Paper No. 10, May, 1995.

Chargin, M. L.; Dunne, L. W.; Herting, D. N. Nonlinear Dynamics of Brake Squeal, MSC
1996 World Users’ Conf. Proc., Vol. V, Paper No. 47, June, 1996.

Cifuentes, A. O.; Herting, D. N. Transient Response of a Beam to a Moving Mass Using a

Finite Element Approach, Innovative Numerical Methods in Engineering, Proc. of
the Fourth Int. Symp. on Numerical Methods in Engineering, Springer-Verlag,
pp. 533-539, March, 1986.
Cifuentes, A. O.; Lalapet, S. Transient Response of a Plate to an Orbiting Mass, Proc. of
the Second Panamerican Cong. of Applied Mechanics, January, 1991.
Cifuentes, A. O.; Lalapet, S. A General Method to Determine the Dynamic Response of a
Plate to an Orbiting Mass, Computers and Structures, Vol. 42, No. 1, pp. 31-36,
1992.

Dai, Chung C.; Yang, Jackson C. S. Direct Transient Analysis of a Fuse Assembly by
Axisymmetric Solid Elements, Thirteenth NASTRAN Users’ Colloq., pp. 431-452,
May, 1985, (NASA CP-2373).

Deloo, Ph.; Klein, M. In-Orbit Disturbance Sensitivity Analysis of the Hubble Space
Telescope New Solar Arrays, Proc. of the 19th MSC European Users’ Conf., Paper
No. 11, September, 1992.
Everstine, Gordon C. A NASTRAN Implementation of the Doubly Asymptotic
Approximation for Underwater Shock Response, NASTRAN: Users’ Exper.,
pp. 207-228, October, 1976, (NASA TM X-3428).
Flanigan, Christopher C. Accurate Enforced Motion Analysis Using MSC/NASTRAN
Superelements, MSC 1994 World Users’ Conf. Proc., Paper No. 25, June, 1994.
 CHAPTER L 535
References and Bibliography

Frye, John W. Transient Analysis of Bodies with Moving Boundaries Using NASTRAN,
NASTRAN: Users’ Exper., pp. 377-388, September, 1975, (NASA TM X-3278).

Hirata, M.; Ishikawa, K.; Korosawa, M.; Fukushima, S.; Hoshina, H. Seismic Analysis of
Plutonium Glovebox by MSC/NASTRAN, January, 1993.

Katnik, Richard B.; Deutschel, Brian; Cherukuri, Ravi. Transient Response of a Vehicle
Over Road Bumps Using the Fourier Transform in a Modal Subspace, The MSC 1992
World Users’ Conf. Proc., Vol. I, Paper No. 6, May, 1992.

Kim, Hyoung M.; Bartkowicz, Theodoore J.; Van Horn, David A. Data Recovery and
Model Reduction Methods for Large Structures, The MSC 1993 World Users’ Conf.
Proc., Paper No. 23, May, 1993.

Larkin, Paul A.; Miller, Michael W. STS Coupled Loads Analysis Using MSC/NASTRAN,
MSC/NASTRAN Users’ Conf. Proc., Paper No. 18, March, 1985.

Lee, J. H.; Tang, J. H. K. Dynamic Response of Containments Due to Shock Wave, Proc. of
the Int. Conf. on Containment Design, pp. 25-32, June, 1984.
Lee, Sang H.; Bock, Tim L.; Hsieh, Steve S. Adaptive Time Stepping Algorithm for
Nonlinear Transient Analysis, The MSC 1988 World Users Conf. Proc., Vol. II,
Paper No. 54, March, 1988.
Leifer, Joel; Gross, Michael. Non-Linear Shipboard Shock Analysis of the Tomahawk
Missile Shock Isolation System, 58th Shock and Vibration Symp., Vol. 1, pp. 97-117,
October, 1978, (NASA CP-2488).
Lipman, Robert R. Computer Animation of Modal and Transient Vibrations, Fifteenth
NASTRAN Users’ Colloq., pp. 111-117, August, 1987, (NASA CP-2481).

Mattana, G.; Miranda, D. MSC/NASTRAN Applications in P-180 Analysis, Proc. of the

MSC/NASTRAN First Italian Users’ Conf., October, 1987.
McMeekin, Michael; Kirchman, Paul. An Advanced Post Processing Methodology for
Viewing MSC/NASTRAN Generated Analyses Results, MSC 1994 World Users’
Conf. Proc., Paper No. 21, June, 1994.
McNamee, Martin J.; Zavareh, Parviz. Nonlinear Transient Analysis of a Shock Isolated
Mechanical Fuse, The MSC 1990 World Users Conf. Proc., Vol. I, Paper No. 21,
March, 1990.
Neilson, H. C.; Everstine, G. C.; Wang, Y. F. Transient Response of Submerged
Fluid-Coupled Double-Walled Shell Structure to a Pressure Pulse, J. of the Acoustic
Soc. of America, Vol. 70, No. 6, pp. 1776-1782, December, 1981.

Pamidi, P. R. On the Append and Continue Features in NASTRAN, Seventh NASTRAN

Users’ Colloq., pp. 405-418, October, 1978, (NASA CP-2062).

Rose, Ted. A Method to Apply Initial Conditions in Modal Transient Solutions, The MSC
1991 World Users’ Conf. Proc., Vol. I, Paper No. 13, March, 1991.
Rose, Ted.; McNamee, Martin. A DMAP Alter to Allow Amplitude- Dependent Modal
Damping in a Transient Solution, MSC 1996 World Users’ Conf. Proc., Vol. V, Paper
No. 50, June, 1996.
536

Swan, Jim. A DMAP Alter for Interface Loads Across Superelements in Dynamic Analyses,
The MSC 1992 World Users’ Conf. Proc., Vol. I, Paper No. 23, May, 1992.

Tang, C. C. Space Station Freedom Solar Array Wing: Nonlinear Transient Analysis of
Plume Impingement Load, MSC 1994 World Users’ Conf. Proc., Paper No. 35, June,
1994.

Urban, Michael R.; Dobyns, Alan. MSC/NASTRAN Transient Analysis of Cannon Recoil
Loads on Composite Helicopters, The MSC 1991 World Users’ Conf. Proc., Vol. I,
Paper No. 28, March, 1991.

West, Timothy S. Approximate Dynamic Model Sensitivity Analysis For Large, Complex
Space Structues, MSC 1996 World Users’ Conf. Proc., Vol. I, Paper No. 6, June, 1996.

Wingate, Robert T.; Jones, Thomas C.; Stephens, Maria V. NASTRAN Postprocessor
Program for Transient Response to Input Accelerations, NASTRAN: Users’ Exper.,
pp. 707-734, September, 1973, (NASA TM X-2893).

Yang, Jackson C. S.; Goeller, Jack E.; Messick William T. Transient Analysis Using Conical
Shell Elements, NASTRAN: Users’ Exper., pp. 125-142, September, 1973, (NASA
TM X-2893).
 537

I N D E X
NX Nastran Basic Dynamic Analysis User’s Guide

I N D E X
NX Nastran
Basic Dynamic
Analysis User’s
Guide
A CQUADR, 17
CRAC2D, 17
ACCELERATION, 134, 180 CRAC3D, 17
Acceleration, 3 Critical damping, 6
Acoustic analysis, 301 CROD, 17, 18
Aeroelastic analysis, 312 CSHEAR, 17
Aeroelastic flutter, 312 CTETRA, 18
ASETi, 295 CTRIA3, 18, 72
Automatic Givens method, 46, 48 CTRIA6, 18
Automatic Householder method, 46, 48, 59 CTRIAR, 18
CTRIAX6, 18
CTUBE, 18
B CVISC, 27, 112, 154, 162
B2GG, 31, 154
B2PP, 31, 154
Beating, 184 D
Damped circular natural frequency, 7
Damping, 24, 284
C Damping ratio, 7
DAREA, 118, 123, 124, 135, 136, 146, 148, 150,
CBAR, 17, 83
168, 172, 173, 181, 182, 188, 196, 201, 206,
CBEAM, 17
219
CBEND, 17
DELAY, 117, 118, 123, 135, 168, 172, 181, 182
CDAMPi, 26, 112, 154, 162
Design optimization, 309
CELASi, 72
Design sensitivity, 310
Characteristic frequency, 36
Direct frequency response analysis, 106, 114
Characteristic shape, 36
damping, 106
CHEXA, 17, 77
excitation definition, 115
Circular natural frequency, 5, 39
Direct matrix input, 30
CMASSi, 17, 22, 23, 201, 202
Direct transient response analysis, 153, 164
Complex arithmetic, 106
damping, 154
Complex eigenvalue analysis, 296
excitation definition, 165
Component mode synthesis, 307
initial conditions, 156
Computer resource requirements, 290
DISPLACEMENT, 56, 68, 134, 144, 180
CONEAX, 17
DLOAD, 201, 206
CONMi, 17, 22, 23, 201, 202
DLOAD (Bulk Data), 116, 126, 135, 136, 146,
CONROD, 17 In
166, 175, 181, 182, 188, 219
Consistent mass, 18, 19, 20, 21
DLOAD (Case Control), 116, 117, 123, 124,
Control system, 311
125, 133, 138, 146, 166, 167, 172, 174, 179,
Coupled mass, 17, 18, 19, 20, 21
188
CPENTA, 17, 77
DMAP, 315
CQUAD4, 17, 72
DMIG, 32, 83
CQUAD8, 17
538 INDEX

DPHASE, 117, 119, 135, 136, 146 Frequency response guidelines

Dynamic absorbers, 141 damping, 284
Dynamic aeroelastic response, 313 frequency increment, 284
Dynamic amplification factor, 9, 10 number of retained modes, 284
Dynamic degrees of freedom, 3 Fundamental frequency, 36
Dynamic reduction, 295 Fundamental shape, 36
DYNRED, 295

G
E Generalized dynamic reduction, 295
EIGC, 296 Generalized mass, 41, 60
Eigenvalue, 37, 40 Generalized stiffness, 41, 60
Eigenvalue extraction methods, 45 Geometric nonlinearity, 303
comparison, 48 Givens method, 46
Givens, 45 GPSTRESS, 180
Householder, 45 Gravity, acceleration of, 22
inverse power, 45 GRDSET, 96
Lanczos, 45 Grid point weight generator, 452
modified Givens, 45 Guidelines, 270
modified Householder, 45 Guyan reduction, 295
Sturm modified inverse power, 45
Eigenvalue problem, 37, 39, 468
Eigenvector, 37, 39, 40
MASS normalization, 43
H
Halfpower bandwidth, 132
MAX normalization, 43
Half?power bandwidth, 277
orthogonality, 41
Hertz, 5
POINT normalization, 43
Householder method, 46
scaling, 42
Hydroelastic analysis, 301
strain energy, 44
EIGR, 52, 56, 57, 59, 90, 112, 135, 163, 181
EIGRL, 50, 56, 57, 68, 112, 135, 136, 138, 146,
163, 181, 182, 188 I
ELFORCE, 135, 144, 180 IC, 156, 179, 184
ELSTRESS, 135, 180 INCLUDE, 72
Enforced motion, 200 Include files, 72
EPOINT, 311 Initial conditions, 5
ESE, 68 Inverse power method, 47

F K
File Management Section (FMS), 441 K2GG, 31, 85
FORCE, 125, 135, 175, 180 K2PP, 31
Forced vibration analysis, 8
Free vibration analysis, 4
ex FREQi, 135, 136, 138, 146, 284
FREQUENCY, 133, 138, 146
L
Lanczos, 290
Frequency increment, 284
Frequency response analysis, 104
 INDEX 539

Lanczos method, 45, 48, 50, 68, 72, 77, 91, 474 NOLINi, 305, 311
block size, 51 Nonlinear transient response analysis, 303
diagnostic output, 51 Nonstructural mass, 23, 62
normalizes, 52 Normal frequency, 36
shift, 51, 91 Normal modes analysis, 37, 39
Large mass method, 201, 203, 205, 475 Numerical accuracy, 467
LOADSET, 124, 125, 133, 174, 179
Logarithmic decrement, 277
LSEQ, 124, 125, 135, 136, 148, 150, 173, 175,
181, 182, 196
O
OFREQUENCY, 133
Lumped mass, 17, 19, 20, 21
OLOAD, 134, 144, 180, 280
OMITi, 295
M Orthogonality, 41
OTIME, 179
M2GG, 22, 31, 86 Overdamped system, 6
M2PP, 22, 31
Mass, 17
grid point weight generator, 452
mass density, 22, 62
P
P2G, 31
PARAM,COUPMASS, 17
PARAMs
PARAM,WTMASS, 22
BAILOUT, 473, 485, 488
units, 21
CM1, 22
weight density, 22, 62
CM2, 22
Mass matrix, 39
COUPMASS, 17, 67
Material nonlinearity, 303
DDRMM, 113, 163, 250, 253, 300
MATi, 22, 107, 112, 154, 162
FREQ, 112
GE, 26
G, 26, 107, 154, 155
RHO, 22
GRDPNT, 72, 271, 453
Matrix method, 112, 163
HFREQ, 112, 163
Mechanism, 41, 88, 472
KDAMP, 112
Memory requirements, 290
LFREQ, 163, 203, 220
METHOD, 133, 138, 146, 179, 188
LMODES, 112
Modal frequency response analysis, 108, 114
MAXRATIO, 93, 471
damping, 109
MODACC, 300
excitation definition, 115
MODES, 163
mode truncation, 112
W3, 26, 154, 155, 277
Modal transient response analysis, 158, 164
W4, 26, 154, 155, 277
damping, 159
WTMASS, 22, 62, 202, 219, 275, 455
excitation definition, 165
PBAR, 23
mode truncation, 162
Period of response, 5
Mode acceleration method, 300
Phase lead, 10
Mode displacement method, 112, 163
Phase shift, 104
Mode shape, 36, 39
PLOAD1, 125, 175
MODES, 68
PLOAD4, 136, 148, 150, 182, 196
Modified Givens method, 46
Plotting, 247 In
Modified Householder method, 46
PostScript, 249, 255
Multiple boundary conditions, 79
Principal mass axes, 456

N
Natural frequency, 5, 21, 36
540 INDEX

Q Static degrees of freedom, 3

Stiffness matrix, 39
QSET/QSET1, 295 STRAIN, 135, 180
Quality factor, 25 STRESS, 68, 135, 180
Structural damping, 24
Structure plotting, 249
R Sturm modified inverse power method, 47,
48, 62
Random vibration analysis, 299
Substructuring, 307
Rayleigh's equation, 41
Superelement analysis, 307
RBE2, 203
SUPORT, 90, 91, 92, 93, 94, 96, 203
Repeated roots, 44
SVECTOR, 134, 180
Resonance, 9, 10
SVELOCITY, 134, 180
Resonance frequency, 36
Symmetry, 79
Resonant frequency, 36
Response spectrum analysis, 297
Restarts, 223
Rigidbody drift, 203 T
Rigidbody modes, 41, 88 TABDMP1, 110, 135, 136, 138, 146, 160, 181,
RLOAD1, 116, 123, 125, 136, 148, 150, 196 182, 188, 276
RLOAD2, 117, 136, 146 TABLED1, 120, 123, 125, 136, 146, 150, 169,
172, 175, 182, 196, 206, 219
TABLED2, 121, 170, 206
S TABLED3, 122, 171, 206
TABLED4, 122, 171, 206, 214
SACCELERATION, 134, 180
TABLEDi, 116, 117, 135, 166, 181
SDAMPING, 133, 138, 146, 179, 188, 276
TF, 311
SDISPLACEMENT, 134, 144, 180
TIC, 156, 181, 182, 184
Single degreeoffreedom (SDOF) system, 3
Time step, 154, 159, 177, 286
SOLs
TLOAD1, 166, 172, 175, 182, 202, 208, 219
107, 296
TLOAD2, 167, 182, 188, 202, 208
108, 133
Transient response analysis, 152
109, 179, 305
Transient response guidelines
11, 133
damping, 287
110, 296
duration of computed response, 287
111, 133
integration time step, 286
112, 179, 305
number of retained modes, 286
12, 179
TSTEP, 286
129, 303
TSTEP (Bulk Data), 177, 181, 182, 184, 188
200, 309
TSTEP (Case Control), 177, 179, 184, 188
28, 296
Tuned mass dampers, 141
29, 296
3, 56
67, 296
70, 296 U
8, 133 Underdamped system, 7
ex 9, 179 Units of measure, 28
99, 303 User messages
SORT1, 134, 136 3035, 90, 98, 100
SORT2, 134, 136 784, 232
SPCFORCES, 134, 180
SPOINT, 295, 311
Static condensation, 295
 INDEX 541

V
VECTOR, 56, 134, 180
VELOCITY, 134, 180
Velocity, 3
Virtual fluid mass, 301
Viscous damping, 24

X
XY plotting, 255

In
542 INDEX

ex