Mechanical COADE, Inc.

Engineering
News Volume 18
For the Power, Petrochemical and Related Industries June, 1994

The COADE Mechanical Engineering News Bulletin is has also been run on a machine with DOS 6.2 and Doublespace.
published periodically from the COADE offices in Houston, The software also runs in a DOS box under Windows 3.1.
Texas. The Bulletin is intended to provide information about
software applications and development for
Mechanical Engineers serving the power, petrochemical, Inside This Issue
and related industries. Additionally, the Bulletin will serve
as the official notification vehicle for software errors discov- Network Versions of all Software Available
ered in those Mechanical Engineering programs offered by TANK Program Introduced
COADE. (Please note, this bulletin is published only two to CAESAR II Memory Reduction
three times per year.) CodeCalc Incorporates Rectangular Vessels

Table Of Contents
PC Hardware for the Engineering User (Part 18) ..... 1
Network Versions
Whats New at COADE
TANK - COADE Introduces New Software ......... 3
In order to address user requests for network versions of
Changes for CAESAR II Version 3.21 ................ 4
COADE Software products, COADE has implemented a
Changes for CodeCalc Version 5.3 ...................... 4
new External Software Lock (ESL) from a second supplier.
Seminar Schedules for 1994 .................................. 5
The new ESLs provide transparent access to both a local
Technology You Can Use
installation and a network installation. Additionally, all
Flange Rigidity Calculations Added to
COADE software products will continue to support the
CAESAR II and CodeCalc ...... ...................... 5
current ESL. Through the support for three different ESL
The Influence of Corrosion on Flange Design ...... 6
configurations, COADE products should be able to provide
An Introduction to Rectangular Vessel
the flexibility required to address any DOS/PC installation.
Computations ................................................... 7
Establishing Time History Input - An Example .... 8
According to the vendor, the network protocols supported by
Watching out for Virus Infections ....................... 13
this new ESL are: NETBIOS, IPX, and IPXNOSAP. This
Modifying Printer Configurations ....................... 13
should allow operation under Novell, DLink, Lantastic,
CAESAR II Specifications ................................. 14
Netware, Lan Manager, and 3 COM networks. (Please note,
TANK Specifications .......................................... 15
COADE has implemented a Novell 3.11 network. We have
no experience with any of the other systems listed here.)

All of our software products have been modified to facilitate

PC Hardware & Systems for the
installation and usage on a network. The distribution files
Engineering User (Part 18) have been segregate into three directories; \Program,
\Program\System, and \Program\Examples (where \Program
Occasionally COADE receives inquiries about the is the user- specified installation directory). This organization
compatibility of specific hardware and operating systems of files enables read only access to the program directory
with our products. At the present time, all of our software while still allowing user access to public, system wide files.
products will run on Intel compatible CPUs from the 80386 In addition, the Installation program and the File Manager
up to and including the Pentium. (CodeCalc, PROVESSEL, have both been modified to address discontiguous network
TANK also run on the 8088, and 80286.) All software runs drives.
on DOS versions 3.1 up to and including 6.2. Our software

1
COADE Mechanical Engineering News June, 1994

DOS, and Things That Go "Bump" in the Night directory. To complete the installation, the user must manually
run each (self-extracting) compressed program. Once all of
Some users have experienced a problem installing COADE the files have been extracted, the compressed distribution
software; the program files are not extracted from the files can be deleted, and the configuration program run.
compressed distribution modules. To understand how and Bottom line, if the PC was booted from a floppy disk, the
why this happens one must study the operation of the command installation procedure must be completed "by hand".
interpreter, COMMAND.COM.
In a related matter, users may have noticed that sometimes
When the system is booted, DOS loads COMMAND.COM the DOS command line editing facility (the F1, F2, and F3
in three sections, in three different locations in system keys) does not work, i.e. the system has apparently lost the
memory. The first section, the resident portion of command string. This is actually a function of the last
COMMAND.COM, is loaded as low as possible, just after program run on the system. Unfortunately, DOS stores the
the internal DOS tables and interrupt vectors. The resident command string in the transient portion at the top of memory.
portion of COMMAND.COM maintains several internal When this area is released to the system to run an external
DOS variables and processes interrupt requests. The second program, the buffer containing the command string may be
portion of COMMAND.COM contains the installation and overwritten. If this happens, the F1, F2, and F3 editing keys
setup code. This portion is only used during the boot process. have nothing to work with.
Once the system is up and running, the memory used by this
part of COMMAND.COM is returned to the system. ESL Notes

The third part of COMMAND.COM is the transient portion. The current (green) ESLs have been in use by COADE since
This portion is used to contain additional data tables and to early 1990. Some users who have newer (faster) computers,
process internal commands. When an external command is or computers with substandard parallel ports have
issued (the user requests a program to be run), the memory experienced problems accessing these ESLs. Typically the
used by this transient portion of COMMAND.COM is released error values returned are 5003 and 5001. The Summer 1994
to the system. When the user program terminates, the release of all of our products will contain code to specifically
resident portion of COMMAND.COM must reload the attempt to work around the 5003 error. However, the ESL
transient portion from disk. This is where the potential for vendor has also suggested the following environment settings
a problem lies; the system must be able to find as possible solutions. These environment settings should be
COMMAND.COM on disk. tried individually from the DOS command line. If a 5003 or
5001 problem is resolved, the command should be placed
The system attempts to reload COMMAND.COM from disk permanently in the system AUTOEXEC.BAT file. These
using the COMSPEC environment variable. The COMSPEC environment commands are:
variable defines where on disk COMMAND.COM is located.
If COMSPEC points to a floppy drive, and the diskette has SET SSI_ACT=l,m,n The values of l, m, and n
been changed, the attempt fails. If COMSPEC points to a can vary, but typical values are
different (or altered) version of COMMAND.COM, the 30, 30, and 40. These values are
attempt fails. (For safety, many users copy used instead of the automatic
COMMAND.COM from the root directory to the \DOS timing constants determined by
directory, just so a backup exists. If DOS is upgraded, the the software.
\DOS copy of COMMAND.COM must be upgraded manually
to avoid potential version mismatches, especially if SET SSI_MEM=OLD This setting indicates to the
COMSPEC points to \DOS. During the reload of the transient software that slower ESL
portion, COMMAND.COM performs a checksum to insure memory access should be used.
what gets loaded is the same as what was loaded during the
initial boot process.)
Contact the COADE Sales
During the installation of COADE software, the process of
extracting the program files from the compressed distribution
Department for Network
files is performed by a batch file. Invoking a batch file Upgrade Pricing.
requires reloading COMMAND.COM. Therefore, booting
the system from a floppy drive, or multiple versions of
COMMAND.COM, will cause this batch process to fail.
This leaves the compressed distribution files in the installation

2
COADE Mechanical Engineering News June, 1994

TANK - COADE Introduces a New

API-650/653 Program

In March of 1994, COADE released the first version of

TANK
TANK, a program for the design and analysis of oil storage API 650/653
tanks according to the rules of API-650 (9th edition) and
API-653 (1st edition). One of the major goals of TANK is
Engineering Software
to provide an easy, obvious interface for the user. To this
end, TANK incorporates the best interface features of
CAESAR II and CodeCalc - installation verification
and configuration, context sensitive help for input cells and
menu options, custom units, error checking and diagnostics,
scrolling input/output screens, and of course, Technical
Support.

The sidebar lists the major technical capabilities of TANK

Version 1.00. Anyone interested in additional information
on TANK should contact the COADE Sales staff.
Demonstration disks (the actual program minus the compu-
tation engine) are available on request.
Design & Analysis of
Version 1.10 of TANK is nearing completion and incorporates
the network addressing modifications discussed previously. Oil Storage Tanks Using the
TANK Version 1.1 will be shipped to all users of Version
1.0. In addition, the following enhancements have been Latest API 650 & 653 Codes
added to the software:
Thickness design per variable point, one
Computations according to API-650 Section 3.11 for foot and Appendix A methods
Wind Overturning Stability have been added.
Allowed fluid height
Anchorage design/analysis per API-650 Appendix
E Section E.6 has been added. Seismic requirements of Appendix E

Automatic generation of nozzle material modulus and Internal pressure per Appendix F
expansion coefficient according to API-650 Appendix
Table P-1 added. Wind girder requirements

For API-653 analysis, an additional output table for Nozzle Flexibilities & allowable loads
Retiring Thicknesses and Remaining Corrosion
Allowances has been added. Shell settlement evaluation according to
API 653 Appendix B
An option is available to perform a corroded hydrotest
case for API-653 evaluations. Built-in Appendix P curves

A new loader provides diagnostic capabilities from Built-in materials data base
the Main Menu.
Context sensitive Help & integrated error
A batch stream processor has been added to enable the checking
processing of multiple jobs.
User-defined units
The input and output modules support a pop-up
Introductory price of $895 US
calculator for on-screen computations. The calculator
supports the basic four math operations, trigonometric
Call for more information today!
functions, squares, and square roots.

3
COADE Mechanical Engineering News June, 1994

Changes for CAESAR II Version 3.21 An additional spring hanger manufacturer has been
added Carpenter & Paterson, from the UK.

The enhancements and changes for CAESAR II Version Uniform Building Code (UBC) earthquake spectra
3.21 are also nearing completion. A partial list of these added.
changes is shown below.
Additional changes to the input piping preprocessor
Most of the CAESAR II executable modules have include:
been converted from Microsoft 16 bit FORTRAN to
WATCOM 32 bit FORTRAN. This has reduced the - problem size is now dependent on the amount of
low DOS RAM requirements of the program from free extended memory - the old limit of 400
577k to 475k. elements is now upwards of 8,000 elements.

The software now supports an ESL from a new vendor. - graphics menus automatically turned off for hard
This provides CAESAR II with full networking copies
abilities. The program first checks for a local ESL
(from either vendor), then for a network ESL. - optional node number display for supports,
anchors, hangers, and nozzles in graphics
The SIF scratch-pad from the Miscellaneous processor
(Option C of the Main Menu) has been incorporated - function key map shown on main spreadsheet
into the piping preprocessor. This processor includes
all of the supported piping codes (not just B31.1 and - auxiliary input spreadsheets support on-line help
B31.3 as in Version 3.20) and all of the fittings.
Additionally, any changes made to the scratch-pad CAESAR II Version 3.21 will be shipped to all "current"
data can optionally be transferred directly to the main CAESAR II purchases and leases.
CAESAR II data spreadsheets.

The piping code addenda have been reviewed and any

necessary changes made to the software. The addenda Changes for CodeCalc Version 5.3
include revisions for: ASCE #7, B31.1, B31.8, ASME
NC, and ASME ND.
The CodeCalc program has also been revised. The new
CodeCalc, Version 5.3 includes many enhancements and
The B31.5 piping code has been added.
should ship by June 1994. The major changes to the program
are listed below.
The API-617 and NEMA-SM23 reports have been
overhauled so that the code compliance when using
The A-93 Code Addenda are included.
non-English units systems is consistent.
A new module is available to perform analysis/design
The new Flange Rigidity factor from ASME Section
for rectangular vessels.
VIII has been added (see accompanying article for
details).
The Users Manual has been completely revised.

COADE BBS
The COADE Bulletin Board is a service provided to our users to facilitate the timely transfer of data and
program files. The COADE BBS can be accessed by dailing 713-890-7286 and completing the registration
information. The files available for download are segregated by application. Note, patches for a specific
program version should only be used with that version.

4
COADE Mechanical Engineering News June, 1994

Wind and Earthquake loads are automatically Flange Rigidity Calculations Added to
generated by the Leg&Lug module. CAESAR II and CodeCalc
The new Flange Rigidity factor from Appendix S has
been added (see accompanying article for details). The January 1994 editions of Section VIII, Division 1 and
Bolt circle/weld leg interference checks are performed Division 2 incorporate a new appendix for the determination
in the Flange module. of a flange rigidity factor. This factor is a parameter which
indicates the tendency of the flange to leak.
The software now supports an ESL from a new
vendor. This provides CodeCalc with full As acknowledged by the Code, flanges which have been
networking abilities. The program first checks for a designed based on allowable stress limits alone may not be
local ESL (from either vendor), then for a network sufficiently rigid to control leakage. The flange rigidity
ESL. factor is a parameter which attempts to indicate the potential
for a leakage problem to occur. The flange rigidity factor (J)
A completely new configuration program is available. is computed from one of three equations, depending on the
type of flange, and should be less than 1.0. Should the
An error processor is included to explain fatal computed rigidity factor exceed 1.0, the Code suggests
software conditions. increasing the flange thickness and recalculating all flange
particulars.
The input spreadsheets have been revised to operate
in a scrolling mode, as opposed to the page down/ The various flange rigidity factor equations are listed below.
page up mode. Except for the Kl and KL parameters, all terms are as defined
in ASME Section VIII, Appendix 2.
The input modules support a pop-up calculator for
on-screen computations. The calculator supports the Integral and optional-type flanges:
basic four math operations, trigonometric functions, (Optional flanges may be calculated as either loose or
squares, and square roots. integral.)

CodeCalc Version 5.3 will be shipped to all CodeCalc users

J=
F52.14 M V I
who purchase the next Update/Support/Maintenance Year
which begins June 1, 1994.
G
HLEg h K J
2
0 0 K
0

Loose-type flanges with hubs:

COADE Seminar Schedule
J=
F52.14 M V I
G
HLEg h K J
2
0 0 K
0

L
L

The following seminars have been scheduled for the remainder

of 1994, at the COADE Houston offices.
Loose-type flanges without hubs:
Piping

J=
F 109. 4 M I
September 21-23
September 26-30
Introduction to Pipe Stress
Statics & Dynamics
G
HEt 1naf
3
K K J
K
0

November 7-11 Statics & Dynamics The Summer 1994 release of CAESAR II, and CodeCalc
will include the computation of the flange rigidity factor.
Pressure Vessel

October 17-19 ASME Sect VIII

5
COADE Mechanical Engineering News June, 1994

The Influence of Corrosion on a better picture of what actually happens, two weld neck
Flange Design flanges were analyzed. One with the corrosion question
answered as Yes and one as No. All other input for these jobs
were identical. The input and the results are shown below.
The ASME Code Section VIII Division 1 Appendix 2 One should note that the flange ID and the hub dimensions
addresses the design of flanges for custom applications such are corroded regardless of the answer to this question.
as those types used on pressure vessels and heat exchangers.
Appendix 2 defines several types or styles of flanges. They INPUT VALUES, FLANGE NUMBER 4, Description: NO CORROSION
are, Weld Neck Flanges, Slip on Flanges, Ring Flanges and Description of Flange Geometry (Type) Integral Weld Neck
Reverse Geometry Weld Neck Flanges. Some of these styles Description of Flange Analysis Analysis Only
Design Pressure P 220.00
of flange can be sub classified as being a loose or possibly an psig
Design Temperature 330.00 F
optional type of flange. Our brief study will observe what Corrosion Allowance FCOR .1250 in.
happens when corrosion is considered to be acting on the Use Corrosion Allowance in Thickness Calcs. No

flange face and when it is not acting on the flange face. Flange Inside Diameter B 19.2500 in.
Flange Outside Diameter A 24.3750 in.
Flange Thickness T 1.5625 in.
Thickness of Hub at Small End G0 .3750 in.
Thickness of Hub at Large End G1 .6250 in.
A flange can be thought of as a set of discrete radial beams Length of Hub HL 1.2500 in.
which are being loaded by the effects of pressure and bolting. Flange Material SA105-F
Pressure and bolting produce loads which try to split the Flange Allowable Stress At Temperature SFO 17500.00 psi
Flange Allowable Stress At Ambient SFA 17500.00 psi
flanges apart. Once the loads are known, the distances over
Bolt Material SA193-B7-1
which they act can be determined, moments can be computed Bolt Allowable Stress At Temperature SBO 25000.00 psi
and ultimately stresses can be calculated. In the course of the Bolt Allowable Stress At Ambient SBA 25000.00 psi

stress computation the Code computes various flange stress Diameter of Bolt Circle C 22.7500 in.
Nominal Bolt Diameter DB .7500 in.
factors such as Alpha, Beta, Gamma, Delta and Lambda. All Type of Threads TEMA Thread Series
of these stress factors are a direct function of the flange Number of Bolts 24

thickness except Gamma, which is a function of the flange ID Flange Face Outside Diameter FOD 21.2500 in.
Flange Face Inside Diameter FID 19.5000 in.
and OD. The question now becomes, When we compute the Flange Facing Sketch 1, Code Sketch 1a
stress factors, should the value of the flange thickness involve
Gasket Outside Diameter GOD 21.1250 in.
the corrosion allowance?. The Code does give us some Gasket Inside Diameter GID 20.3750 in.
Gasket Factor, m, M 3.7500
guidance in this matter. In paragraph UG-16 (e), the Code Gasket Design Seating Stress Y 7600.00 psi
clearly states The dimensional symbols used in all design
Column for Gasket Seating 2, Code Column II
formulas throughout this Division represent dimensions in Gasket Thickness .1250 in.
Length of Partition Gasket 20.7500 in.
the corroded condition. Thus it would appear that one
should use (t-ca) in the design formulas for the stress factors
FLANGE ANALYSIS, FLANGE NUMBER 4, Description: NO CORROSION
per the Code. However, in the Forward to the ASME Code, ASME Code, Section VIII, Division 1, 1992, , Appendix 2, A-93
there is some leeway given to designers to account for good Flange Factors for Integral Flange:
engineering judgement. It is the design philosophy of some Factor F per 2-7.2 .811
Factor V per 2-7.3 .211
that the flange stress factors should not be corroded. The Factor f per 2-7.6 1.013
Factors from Figure 2-7.1 K = 1.250
reasoning is that the flange face does not become corroded. T = 1.818 U = 9.703
Some say the fluid contacting the flange face does not have Y = 8.830 Z = 4.556
d = 6.354in ^3 e = .367in^-1
sufficient velocity to corrode the flange face and they believe Stress Factors ALPHA = 1.574
BETA = 1.766 GAMMA = .866
corroding the stress factors leads to a more costly flange. DELTA = .600 LAMBDA = 1.466
Some design engineers say that when a unit is torn down the
Stress Computation Results: OPERATING GASKET SEATING
flange face remains perfectly smooth and uncorroded. There Actual Allowed Actual Allowed
Longitudinal Hub 17739. 26250. 21525. 26250. psi
is certainly a great deal of wisdom in experience. Radial Flange 3165. 17500. 3841. 17500. psi
Tangential Flange 8787. 17500. 10662. 17500. psi
Maximum Average 13263. 17500. 16094. 17500. psi
Unfortunately, the design philosophy on this subject is not Bolting 13490. 25000. 16896. 25000. psi
consistent between fabrication facilities and engineering Estimated M.A.W.P. MIN( OPE., SEATING ) 290.3 psig
design firms. Periodically there is some discussion as to who Estimated Finished Weight of Flange 88.6 lb.
Estimated Unfinished Weight of Forging 139.8 lb.
is right. Fortunately CodeCalc asks a simple question Do APP. S Flange Rigidity Factor for Seating Case .347
APP. S Flange Rigidity Factor for Operating Case .426
you wish to Include Corrosion in the Flange Thickness
Calculations. An appropriate Yes or No answer here causes
CodeCalc to either corrode the flange factors or not. To get

6
COADE Mechanical Engineering News June, 1994

The following results can be produced by answering Yes to An Introduction to Rectangular

the Include Corrosion question. All other input values are Vessel Computations
identical to those above. Note the differences in the flange
stress factors when comparing these results.
Included in the June release of CodeCalc version 5.3 is the
Rectangular Vessel module. This module performs both
RESULTS, FLANGE NUMBER 5, Description: W/CORROSION stress calculations and maximum allowable working pressure
Flange Factors for Integral Flange: calculations per Section VIII, Division I, Appendix 13. At
Factor F per 2-7.2 .811
Factor V per 2-7.3 .211 present, the Rectangular Vessel module encompasses all of
Factor f per 2-7.6 1.013 the analysis types included in Appendix 13 with two
Factors from Figure 2-7.1 K = 1.250
T = 1.818 U = 9.703 exceptions: vessel type 13-2(a)- 6, and vessels subjected to
Y = 8.830 Z = 4.556
d = 6.354 in^3 e = .367 in^-1 external pressure. The vessel type 13-2(a)-6 is a modified
Stress Factors ALPHA = 1.528 version of vessel type 13-2(a)-5 and will be included in
BETA = 1.704 GAMMA = .840
DELTA = .467 LAMBDA = 1.308 future updates of CodeCalc. Likewise, future updates will
Stress Computation Results: OPERATING GASKET SEATING include the external pressure calculations discussed in Section
Actual Allowed Actual Allowed 13-14.
Longitudinal Hub 19884. 26250. 24127. 26250. psi
Radial Flange 4047. 17500. 4910. 17500. psi
Tangential Flange 8984. 17500. 10901. 17500. psi
Maximum Average 14434. 17500. 17514. 17500. psi The stress calculations involve membrane, bending, and
Bolting 13490. 25000. 16896. 25000. psi total stresses. For each of the three types of stress calculations
Estimated M.A.W.P. MIN( OPE., SEATING ) 266.7 psig a maximum allowable working pressure is computed. The
Estimated Finished Weight of Flange 88.6 lb.
Estimated Unfinished Weight of Forging 139.8 lb. only exception is in the analysis of vessel type 13-2(c)(See
APP. S Flange Rigidity Factor for Seating Case .389 Figure 1).
APP. S Flange Rigidity Factor for Operating Case .477

Note the stresses in this abbreviated report. The stresses in

the case of NO CORROSION are nicely within Code
allowables. The W/CORROSION case however has exceeded
the Code allowable in the Maximum Average category.
Thus we can see that this geometry has failed when we
corrode the stress factors. Additionally, all of the flange
stresses were higher in the corroded case, than in the non
corroded case.

The process of flange design is simplified considerably when

using a computerized program such as CodeCalc. This
easily allows the designer to consider both the corroded and Figure 1
uncorroded condition. The magnitude of the difference
between these runs indicates the significance of this decision. For this circular cross section vessel, a single diametral plate
When combined with experience, knowledge of the chemical separates the vessel into two compartments. If the vessel is
process and its effect on the unit, the designer should be able being analyzed with differing pressures in the two
to make the appropriate decision in this matter. compartments, then the M.A.W.P. calculation for the vessel
will not be performed. The reason for the absence of this
calculation can be seen from reviewing the bending stress
calculation for the shell section (Section 13-13,
Equation 8).

L
M O
P
M F I P
c M G 2t1 J
2
3 R2 P
Sb = M
P
3I1 M
M
1G
G J
H2 8 J
K
+ Pa1 P2 f FIt3
P
3 P
P
M 6+ P
M
N HKt1 P
Q

7
COADE Mechanical Engineering News June, 1994

Because this equation considers both the 1st compartment

pressure and the difference in the 1st and 2nd compartment
pressure, there is no single M.A.W.P. that can be computed.
The maximum value of one of the two pressures will be
dependent on the input value of the other pressure.

In addition to vessel type 13-2(a)-6 and vessels with external

pressure mentioned above, vessels 13-2(a)-9 and
13-2(a)-10 are not directly addressed. These two vessels
have stay bars or stay plates separating the vessel into non-
uniform dimensions. In order to analyze these vessels in
CodeCalc, the second method described in Section 13-9(f)
is used. The following discussion addresses each of the
possible cases.
Figure 3
In the event that the vessel has a single stay plate separating
the vessel into two compartments (See Figure 2), the analysis By inspecting the code stress equations for these stayed
is performed per Section 13-9(b). This analysis assumes that vessels, it can be seen that the greater the dimension h is, the
the compartments have equal dimensions, as in Figure 2 (a). greater the code stress is. Because of this direct relationship,
If the vessel has dissimilar compartments, as in Figure 2(b), by using the maximum dimensions throughout, the stress
the method of analysis requires that the vessel be analyzed as analysis will be a conservative analysis. The stresses in the
if the compartments are of the same size, with the dimensions smaller compartments will actually be much lower than the
equal to that of the largest compartment. In the case of Figure calculated values.
2(b), the height dimension h2, would be entered for the long-
side height.

Establishing Time History Input -

An Example

Volume 17 (December 1993) of Mechanical Engineering

News (M.E.N.) described the basics of modal time history
analysis. This capability was first released in CAESAR II
Version 3.2. That first article on time history analysis
approached the subject through a textbook example
dynamic response of a simple cantilever beam to a sudden,
constant load. The intent was to introduce this new capability
and develop user confidence through an intuitive analogy.
Figure 2 Such an article, while properly acquainting the user with the
method, does little in explaining the input requirements.
The second case is that of a vessel with multiple compartments This article will review the time history input with emphasis
(Figure 3). The analysis for this type of vessel is described on the selection of these values.
in Section 13-9(c). The analysis is carried forth as if the
vessel had three similar compartments (Figure 3(a)). In the Introduction
event that the vessel has three, or more, compartments of
dissimilar dimensions, the maximum height dimension is Figure 4 shows a system that has been used in COADEs
entered as the long-side height. In the case of Figure 3(b), the CAESAR II Dynamics Training Seminar for several years.
height h3 would be used for the analysis. This method is SNFAIL examines the dynamic load on a snubber installed
indifferent to the true number of compartments. to reduce the system response to a steam hammer traveling
through the system. The snubber has a rated load capacity
of 2500 pounds and the analysis will predict whether or not
the snubber will fail (ergo the name, SNFAIL). In the past,
this situation was investigated using the force response
spectrum method of analysis. Since this system has a

8
COADE Mechanical Engineering News June, 1994

specific definition of the dynamic load timing, it is a good Load Definition

candidate for time history analysis. Events with a random
duration and content such as earthquakes and relief valve This analysis examines the effect a sudden valve closure at
loads are best suited for the general analysis found in response node 30. Steam is flowing through the line at 45 fps and
spectrum methods. More precise load definition yields suddenly stops. To make matters simple, it is assumes that
better results through time history analysis. Response the valve closes instantaneously. This assumption will allow
spectrum analysis combines the maximum response for each a conservative guess as to the pressure rise in the line. At the
mode of vibration to generate a total response without instant the flow stops at node 30, all the upstream vapor is still
respect to time. Time history follows the modal responses moving at 45 fps towards node 30. Each molecule of vapor
through each time step to produce a system response with will keep on moving towards the valve face and only stop
respect to time. Time history analysis also allows the once it is acted upon by the building mass of stopped vapor.
coordination of separate events in the system. In SNFAIL, The inertia of the moving vapor is converted into the higher
the steam hammer causes a pressure imbalance in each density and pressure of the stopped vapor. So as the vapor
upstream leg one after the other. Force response spectrum is moving down the line it is stopped by a pressure wave
analysis cannot maintain this sequence of events while time traveling upstream. This pressure wave is traveling upstream
history can. The load characteristics of SNFAIL make it an at the speed of sound in the line. The CAESAR II analysis
excellent candidate for time history analysis. will follow this pressure wave up the line.

There are three basic categories of dynamic analysis input for At steady state conditions, the piping system is in equilibrium.
CAESAR II modification to the static analysis model, One component of the load on the system is the pressure
definition of the dynamic load, and specification of the effects on the straight runs of pipe. One was of visualizing
analysis parameters. The model developed for a static this pressure load is to crawl inside the pipe and look
analysis needs only the snubber definition for the dynamic upstream and downstream. The surfaces you can see are the
analysis. This information is entered through Item 2 - surfaces on which the line pressure act. Again, at constant
SNUBBERS of the Dynamic Analysis Input Menu. A one upstream and downstream pressure, the loads are in
million lbf./in. stiffness is entered in the X direction at node equilibrium (Pup) * (Aup) = (Pdn) * (Adn), since Pup = Pdn = P
100. (The input is entered as: 1E6, X, 100.) and Aup = Adn. But as the pressure wave travels up the line into
your straight run of pipe, that run of pipe is no longer in
equilibrium. The downstream elbow (or valve face) is now
loaded with P + DP where DP is the pressure rise due to the
instantaneous valve closure. This imbalance (DP) will
continue until the pressure wave passes the upstream surface
at which time the run is again in balance (at the higher
pressure).

The three pieces to define the load are now established:

1. The load magnitude (from the Piping Handbook):

P A

F = P A = 1433 lbf
P = c V = 14.10 psi
= P / (R T) = 4.051E-4 lbm/in3
P = 500 psi
R = 85 ft-lbf/lbm-R
T = (750+460) R

c = (K R T) = 2074 ft/sec
K = 1.3
V = 45 ft/sec
A = p/4 (OD - 2 wt) = 101.6 in
2

Figure 4 2. The point of application and direction : At the start

9
COADE Mechanical Engineering News June, 1994

of each straight run of pipe upstream of the valve equilibrium when the end of the pressure rise passes the end
with the load pointing downstream. of the leg, 16 ms after this component of the event started.
-X load @ node 30
Figure 5 compares the instantaneous imbalance with the 10
-Y load @ node 100
ms ramp.
+Z load @ node 110
+X load @ node 120

3. Timing : Based on the speed of sound (c) in the medium,

the imbalance starts when the pressure wave enters the
run and ends when the wave leaves the run. The
CAESAR II coordinate list feature makes it easy to
determine the straight run lengths. Time is simply length
divided by the acoustic speed (speed of sound in the
medium, i.e. 2074 ft/sec). The lengths and times are
referenced to node 30 as the zero point.
from 30 to: pipe length(ft): time(ms):
30 0 0
100 137 66
110 191 92
120 221 107
125 235 113

The assumption that the valve will stop the flow

instantaneously is conservative in the calculation of the
pressure rise. (V = 45 fps; steam flowing at 45 fps suddenly
is traveling at 0 fps.) A conservative estimate of the event
timing may not be assured with the same assumption. That
is, the system response to a sudden imbalance may not be as Figure 5
great as to a slower application of the load. To shed more
light on this issue, an adjustment to the timing will be created.
Assume that it takes 10 ms for the valve to stop the flow (45 The CAESAR II Dynamic Analysis Input for this model
fps to 0 fps in 10 ms) and that the pressure imbalance also would take the following form:
takes 10 ms to fully develop (0 lbf to 1433 lbf in 10 ms). For Item 3 - SPECTRUM/TIME HISTORY DEFINITIONS
the first run of pipe (30-100) the event still starts at 0 ms but
does not reach the maximum of 1433 lbf. until 10 ms later. Set #1 (definition) TH1, T, F, LIN, LIN
Likewise, as the wavefront reaches node 100 at 66 ms into the
event, the imbalance starts its drop but this run will not be (data) Time Load
back in balance until 10 ms later at 76 ms. This 10 ms ramp 0 0
is applied to the start of the imbalance and the end of the 10 1
imbalance for each component of the event. 66 1
76 0
What of the last leg, where the wave front enters and leaves
in only 6 ms? Here, the imbalance cannot reach the total of Set #2 (definition) TH2, T, F, LIN, LIN
1433 lbf. as the exiting wavefront will balance out the
still-building wavefront entering the run. So 6 ms into the last (data) Time Load
leg the increasing load entering the leg (at 120) will be 0 0
balanced by the increasing load leaving the leg (at 125). 66 0
Allowing for only 6 ms of the 10 ms rise time means that the 76 1
92 1
maximum imbalance is (0.6 1433) lbf. At 10 ms into the
102 0
last leg, the load entering the leg will be fully developed and
the legs re-balance will begin the flat portion of the
loading will only last 4 ms.. The leg will again reach

10
COADE Mechanical Engineering News June, 1994

Set #3 (definition) TH3, T, F, LIN, LIN Analysis Parameters

(data) Time Load There is only one more Dynamic Analysis Input item required
0 0 to run this time history analysis in CAESAR II B -
92 0 CONTROL PARAMETERS. While some of the settings are
102 1 obvious (e.g. analysis type = Time History), several analysis
107 1 parameters have a range of possible values some having a
117 0 major impact on the accuracy of the analysis. What is the
correct entry for frequency cutoff, time step and load
Set #4 (definition) TH4, T, F, LIN, LIN duration? Should missing mass be used, and if so, how
should it be included? How important is damping? Guidelines
(data) Time Load
for these data are provided here.
0 0
107 0
FREQUENCY CUTOFF The frequency cutoff establishes
113 .6
the number of modes of vibration included in the analysis.
117 .6
With each additional mode included, the accuracy of the
123 0
analysis improves. But there are diminishing returns with the
higher modes the analysis takes much, much longer with
Item 7 - SPECTRUM/TIME HISTORY FORCE SETS
very little change in the results. So how many modes to
include? A conservative approach is to examine the frequency
Load Direction Node Force Set
content of the event and analyze the system at least to the
-1433 X 35 1 point where the frequency response is maximum. The
-1433 Y 100 2 system frequencies not included in the analysis will be
1433 Z 110 3 approximated by the missing mass correction (see M.E.N.
1433 X 120 4 Vol. 16 - Use the Missing Mass Correction Option in
Spectral Analysis). Fortunately the Response Spectrum
Item 9 - SPECTRUM/TIME HISTORY LOAD CASES Generator in CAESAR II (Item 4 in the Dynamic Analysis
Input) can be used to convert the four time histories into their
Time Hist. Factor Dir. Force Set # frequency response curves. The response curve data are
shown in Figure 6. This plot was generated using the
TH1 1 X 1
CAESAR II 2-D Plotting option (Item F on the Main Menu).
TH2 1 Y 2
TH3 1 Z 3
TH4 1 X 4

One note on the magnitudes or factors specified above.

There are three CAESAR II components which combine to
form the total load used in the time history analysis:

a) the force in the time history definition (1 or .6 in

the time history definitions);

b) the force in the force set definition (1433 in the

time history force sets); and

c) the scale factor in the dynamic load case

definition (1 in the time history load cases).

With the total load calculated as ( a b c ), use any

convenient way to enter these data for the three terms. Figure 6

11
COADE Mechanical Engineering News June, 1994

The response curve for the first event (30 - 100) has the mode of vibration or run a modal analysis of the system to
highest response and it peaks at the lowest frequency. This calculate this value. Simply set the analysis type to MODAL
is reasonable as this is the longest duration event and therefore and run the analysis.
can activate lower modes of vibration. The first three
events all tail off on the same line. This, too, is reasonable The first natural frequency of this system occurs at 0.319 Hz.
since they all have the same ramp rate of 10 ms from 0 to with a period of 3.134 seconds (1/0.319). The load duration
maximum load. The ramp rate for the fourth event is shorter should be at least one quarter of this period beyond the end
than the other three (6 ms to maximum load) so its peak of the event. Here the duration of the analysis should be 0.9
occurs at a higher frequency outside the pattern of the other seconds, which corresponds to a 0.123 second event plus a
three. The resonant peak of these four events are below 8 0.784 second quarter period. If the results of the time history
Hz., 22 Hz., 38 Hz., and 55 Hz., respectively. The frequency analysis show that the maximum system response is near the
cutoff for this analysis will be set at 55 Hz. This will ensure end of the reporting period, the analysis should be re-run
that the most significant modes of vibration for all four events using a longer time span.
will be included in the analysis.
# TIME HISTORY OUTPUT CASES - CAESAR II
TIME HISTORY TIME STEP As a rule of thumb for generates a set of output data at equal intervals throughout
accurate analysis, there should be at least ten time steps over the duration of the analysis. The user specifies the number
the period associated with the highest natural frequency used of output sets here. It would be difficult to predict when a
in the analysis. For convenience, the highest natural frequency maximum response occurs during the event. It is best simply
used in the analysis will be the frequency cutoff of 55 Hz. to divide the duration into even numbers and hold the total
(CAESAR II will include the first mode of vibration above number of report times below 20. Such a setting will keep the
the frequency cutoff.) The period of a 55 Hz. vibration is 1/ output manageable. It may also be useful to have a few
55 seconds. The time step should be no more than one-tenth reports while the load is still in progress through the system.
of 1/55 sec. or no more than 1.82 milliseconds. For this Here, where the loading lasts 123 ms and the analysis covers
analysis the time step will be 1.8 ms. 900 ms, 18 reports will be generated. That will produce a
report every 50 ms with two reports while the load is active.
INCLUDE MISSING MASS COMPONENTS The missing
mass contribution to the analysis should be turned on for this DAMPING CAESAR II has a default ratio of critical
analysis. CAESAR II will lump the response of system damping of 0.03 (3% critical damping). Typical values for
modes not included in the analysis as a single static load. The critical damping ratios in piping systems can be anywhere
magnitude of the load will be based on the last system from 0.02 to 0.05. Changing the value to 0.02 may be more
frequency extracted for the analysis. (The default setting conservative but not necessarily more accurate. If changing
uses the load factor for the last frequency extracted. Figure the critical damping ratio from .03 to .02 develops a drastic
6 can be used to see what that load factor may be. The user change in the results, it would be worth the effort to fine tune
may also set the missing mass load factor of 1.0 to approximate the model; not only in terms of the damping ratio but also the
a fully rigid response to the applied load.)1 system and its loading.

LOAD DURATION The load duration requested here is Results

actually the analysis duration. The dynamic loads are set on
and off in the time history definitions. This event is over in A major part of the time history output in CAESAR II is
123 ms. but the system response may peak after the event is similar to the report structure found in the static output
over. The minimum duration of the analysis should be processor (displacements, restraint loads, forces & moments,
greater than one quarter of the period associated with the stresses). In time history output, which has only one load set,
lowest natural frequency for the system plus the event these reports are generated for each report time requested.
duration. At one quarter cycle, the modal response is at its Think of these reports as snapshots in time. But one would
maximum (i.e. for a sine wave the maximum value of 1.0 is be lucky to have a snapshot for the maximum system
reached at 90 degrees through the cycle of 360 degrees). response. Therefore, the time history output also lists an
Using one quarter of the first mode assures that all modes additional report set that contains the system maxima through
achieve their maximum response. Either assume a low first the entire event. Also, the time of the occurrence of the
maximum value is listed with the number. This maximum
value report is a valuable tool for reviewing the analysis.

1
With the frequency cutoff set to 55 Hz., the last mode of vibration
is at 61 Hz. The load factor for the missing mass component will
be about 1.5 for events 1-3 and 1.8 for event 4.

12
COADE Mechanical Engineering News June, 1994

Another informative review technique available is the In the event mismatches are discovered, specialized virus
graphical animation of system displacements. detection software and/or disk maintenance software should
be used. The assistance of an expert in this area may also be
So is the snubber OK? Run the job to find out. required. (Please note, if software patches obtained from
COADE have been installed, these patches will not match the
The data for this analysis is on the COADE Bulletin Board baseline, and consequently generate CRC mismatches.)
in the file SNFAIL.EXE. Download the file and type
SNFAIL at the DOS prompt to unzip the input and output
files (analysis is not required). Running this job after
modifying the time history analysis parameters will develop Modifying Printer Configurations
insight regarding their effect. Try changing values such as
time step, damping, frequency cutoff and missing mass to The COADE software packages CAESAR II and TANK
further investigate this issue. access a file named PRINTER.FMT to acquire printer
configuration strings. It is assumed that this file resides in the
installation directory. (As of the Summer 1994 releases, this
file will reside beneath the installation directory, in a sub-
Watching out for Virus Infections
directory named SYSTEM.)

Everyone knows that computer virus infections should be The configuration strings are used to set the printer to a
avoided at all costs. Most users regularly run virus scanning particular style of printing, such as 132 versus 80 columns,
software or load some type of anti-virus software at system or landscape versus portrait. The configuration data is
startup. However, since new virus programs are continually arranged in sets of two lines for each configuration. The first
being developed, a virus infection is unfortunately still line is a simple text string, used to describe the format set in
possible. the second line. (It is this text string which is displayed by the
configuration program.) For each printer to be addressed,
As an aid to the users of COADE software, we have provided three configuration strings should be defined; normal 80
with each package (CAESAR II, PROVESSEL, CodeCalc, column, 132 column, and the reset string. The PRINTER.FMT
and TANK) a routine to determine if any of our files have file can handle up to twelve configurations, essentially four
been modified. This routine is a small program, CRCCHK, different printers.
which can be run at any time to insure the integrity of
COADE distributed files. The actual formatting string is data acquired from the printer
control code section of the printer manual. The printer
A virus infection usually alters a file in some way - thus manuals define these control codes in terms of numbers,
providing a means of detection if some baseline can be letters, and the [Esc] character. Unfortunately, most editors
established. In implementing CRCCHK, the baseline is a which would be used to modify PRINTER.FMT can not
CRC (cyclic redundancy check) value for each file - essentially handle the [Esc] character! The method employed by
a finger print of the file. COADE establishes the CRC value CAESAR II and TANK is to require the ASCII code for the
for each file and includes this data on the distribution [Esc], which is the number 27.
diskettes, in *.CRC files.
Unfortunately this decision presents some confusion to anyone
The initial installation is verified immediately after the files (or any program) reading the PRINTER.FMT file. How can
are extracted from the compressed distribution format. This you tell when a number is a number or an ASCII code? The
insures the files are intact and have not been damaged by answer to this is that you cant. Therefore, all values in the
either a virus or bad media (disks). Anytime the integrity of PRINTER.FMT file represent ASCII codes except for two
the software is in question, the CRCCHK program should be cases; the character is alphabetic or the number represents
run again to verify the baseline. The CRCCHK program data count. Consider the following set of lines for an HP
must be invoked directly from the installation directory. For Laser Jet as an example.
example, assume CAESAR II has been installed on drive
E in a directory named CAESAR. The proper commands LaserJet_132/6
to verify the baseline are: 27 E 27 & k 50 S 27 & l 54 D 27 & k 48 G
E: [Enter]
CD\CAESAR [Enter]
CRCCHK *.CRC [Enter]

13
COADE Mechanical Engineering News June, 1994

Each item in the configuration string is defined below. Blank A correction for this problem is available from the
lines separate distinct printer commands. COADE BBS as F320D.EXE.

27 the [Esc] code 2) WRC-297 Stress Computations: An error has been

E the capital letter E, the reset command discovered in the computation of the Out-of-Plane
bending component of the longitudinal vessel stresses.
27 the [Esc] code This error exits in Versions 3.15, 3.16, 3.17, 3.18, 3.19,
& the ampersand character and 3.20.
k the small letter k
50 the ASCII code for the character 2 A correction for this problem is available from the
S the capital letter S, sets 132 column printing COADE BBS as F020494.EXE.
27 the [Esc] code
& the ampersand character 3) Piping Error Checker: An error has been discovered in
l the small letter l the setup for the API-650 Appendix P curve interpolation
54 the ASCII code for the character 6 routine. The error caused an invalid pointer to be set
D the capital letter D, sets 6 lines per inch printing when the R/t value was off of the curves.

27 the [Esc] code A correction for this problem is available from the
& the ampersand character COADE BBS as F020494.EXE.
k the small letter k
48 the ASCII code for the character 0 4) Piping Error Checker: An omission has been discovered
G the capital letter G, sets carriage return and line in the determination of the limiting nozzle loads. The
feed sequence properly minimum value of 0.1 for the axis limits of Figures
P-3A and P-3B was not checked.
Modifying this file is relatively simple if the printer codes are
available. It is suggested that the original file distributed by A correction for this problem is available form the
COADE be saved, so that it is possible to recover from a COADE BBS as F320F.EXE.
major editing error.
5) Piping Error Checker: An error was discovered in the
Additional configuration strings defined by CAESAR II conversion of the nozzle delta temperature value for
users are shown in the December 1993 issue of Mechanical API-650 nozzles. The temperature scale offset was
Engineering News. Anyone with comments, suggestions, or used in the conversion. This is not necessary since the
interesting configuration strings is urged to submit them for value is a delta temperature, not an abosolute value.
publication for the benefit of all users. This error only affects the radial nozzle deflection for
jobs using units systems other than English. This error
will be corrected in Version 3.21.
CAESAR II Specifications
Class 2
Listed below are those bugs/errors/omissions in the
CAESAR II program that have been identified since the last 1) Piping Preprocessor: The 3.20 release contained an
newsletter. These items are listed in two classes. error in the text resource file for the individual spring
Class 1 errors are problems or anomalies that might lead to hanger spreadsheets. The error was that the units of
the generation of erroneous results. Class 2 errors are general inches were fixed as part of the text, instead of being
problems that may result in confusion or an abort condition, dependent on the active units file.
but do not cause erroneous results.
This error only affected the display of the spreadsheet.
Class 1 All recipients of Version 3.20 received a letter with the
update which discussed this error and the fix.
1) Dynamic Stress Computations, RCC-M Code: An error
has been discovered in the computation of Equation #10 A correction for this problem is available from the
for dynamic analysis. The Pmax value could possibly be COADE BBS as F320A.EXE
left as zero, causing the longitudinal pressure term to be
zero. This error is in CAESAR II Versions 3.19 and
3.20 only.

14
COADE Mechanical Engineering News June, 1994

2) Structural Input Module: The 3.20 release of the TANK Specifications

structural input module contained compiler debug
information. This increased the memory requirement of
this module to 590k bytes. Listed below are those bugs/errors/omissions in the TANK
program that have been identified since the last newsletter.
A correction for this problem is available from the These items are listed in two classes. Class 1 errors are
COADE BBS as F121193x.EXE, where x is U, problems or anomalies that might lead to the generation of
L, or D, for unlimited users, limited users, and erroneous results. Class 2 errors are general problems that
dealers respectively. may result in confusion or an abort condition, but do not
cause erroneous results.
3) Static & Dynamic Output Modules: The 3.20 output
processors attempt to open the printer with a Class 1
STATUS=OLD parameter, as suggested by Microsoft.
Unfortunately, this causes problems under the Watcom 1) Appendix P Nozzle Limiting Loads: An omission of the
compiler on non-Novell networks. 0.1 limit from Figures P-3A and P-3B allowed negative
numbers to be computed for the limiting nozzle loads.
A correction for this problem is available from the This problem has been corrected and is available on the
COADE BBS as F121193x.EXE, where x is U, COADE BBS as TK100A.EXE.
L, or D, for unlimited users, limited users, and
dealers respectively. 2) Appendix P Units Conversion: An error was discovered
in the conversion of the nozzle delta temperature value.
4) Time History Analysis: Two errors have been corrected The temperature scale offset was used in the conversion.
in the dynamic analysis module related to Time History This is not necessary since the value is a delta temperature,
analysis. The first error is related to a memory not an abosolute value. This error only affects the radial
management problem (only activated if missing mass is nozzle deflection for jobs using units systems other than
turned off) which causes the program to abort. The English. This error will be corrected in Version 1.10.
second error, again related to memory management in
Time History runs, allowed the program to compute Class 2
stresses on rigid elements if the Compute Forces on
Rigids and Expansion Joints flag was turned on in the 1) Wind Girder Computations: For non-specified wind
input. velocities, the velocity factor remains initialized as zero.
This causes the required section modulus of the wind
A correction for this problem is available from the girders to be reported as zero. To correct this situation,
COADE BBS as F320D.EXE. enter the appropriate wind velocity or down-load the
patch TK100A.EXE. (The patch resets a zero velocity
5) Piping Error Checker: An error has been discovered in factor to 1.0.)
the Piping Error Checker for Version 3.20, which causes
a page eject to be issued for each message when printing 2) Output Module: In the report containing the wind data,
errors and warnings. This will be corrected in Version material allowables, thicknesses, and fluid heights, an
3.21. improper conversion factor was applied to the wind
velocity. This resulted in the wrong value of wind
velocity being reported for any unit other than ft/sec.
This problem has been corrected for Version 1.10.

15
COADE Mechanical Engineering News June, 1994

CodeCalc Version 5.3 Addresses Rectangular Vessels

Technical Note
When utilizing the Flange Analysis module of CAESAR II (Option C from the Main Menu), users
should note a change to the 1992 edition of the ASME NC code with regards to the allowable stress
multiplier. Prior to the 1992 edition of NC, paragraph 3658.1(d) allowed a multiplier of 1.5 for the
tangential and radial stresses. The 1992 edition of NC eliminated this paragraph. However, B31.1
Appendix II, Section 4.2.3 still allows the 1.5 multiplier on these two allowables.

For CAESAR II Version 3.21, the spreadsheet will be altered to remove the reference to 3658.1(d).
The user should employ good engineering judgement in deciding whether or not to use a multiplier
greater than 1.0 for these flange allowables. For additional details, see the related article Flange
Allowable Stresses in the October 1991 issue of Mechanical Engineering News.

COADE Engineering Software

12777 Jones Rd. Suite 480, Houston, Texas 77070 Tel: 713-890-4566 Fax: 713-890-3301 BBS: 713-890-7286

16