Finite Element Methods as used in

Design, Analysis and Certification

Aspects in the Aircraft Industry Today

J.V.Kamesh, Sc ‘G’
National Civil Aircraft Development Program
National Aerospace Laboratories
Bangalore

1
 Acknowledgements
I would like to sincerely acknowledge a number
of my esteemed colleagues at the
Aeronautical Development Agency for all the
work that I am going to present. I have spent
nearly two decades with all of them.

2
 Outline of the Presentation

 Brief Recapitulation of Finite Element Methods / Principles

 Finite Elements used in Aircraft Structures
 Basis for the development of a Finite Element Model of an Aircraft
 Some Basics of Design Principles
 Analysis and updation of the Model
 Adaptation of the Design to the real structure
 Experiments/Tests to prove the Design – Likely differences and causes
 Comparison of the Test and Analysis results and Updation of the Finite Element
Model
 Analysis as a substitute to Expensive and Time-Consuming Tests for Certification –

 Divergence Analysis
 Dynamic and Flutter Analysis
 Transient Response Analysis – Bird Hit, Shock and Missile Thrust
 Frequency Response Analysis
 Demonstration of Structural Integrity – Effect of the impact of an Arrestor Barrier after a
rejected take off

 Some Modern Areas/Domains of the usage of FE Techniques

3
 ORIGIN OF THE FINITE ELEMENT METHOD

DIFFICULT TO DETERMINE THE ORIGINS

IT HAS EVOLVED OVER THE YEARS TO ITS PRESENT FORM

CONTINUUM

APPROXIMATE EXACT

DISCRETE CONTINUOUS

MATRIX METHODS DIFFERENTIAL EQUATIONS

VARIATIONAL PRINCIPLES, FINITE DIFFERENCES, WEIGHTED
VIRTUAL DISPLACEMENTS, RESIDUALS Etc.
MIN POTENTIAL ENERGY

DISPLACEMENTS
4
 BASICS OF THE FINITE ELEMENT METHOD

IN STRUCTURAL ANALYSIS, WHATEVER BE THE TYPE OF STRUCTURE

TYPE OF LOADING
MATERIAL

ONLY THREE ARGUMENTS NEED TO BE SATISFIED. THEY ARE

EQUILIBRIUM – RELATE STRESS TO APPLIED FORCES
COMPATIBILITY – RELATE STRAIN TO DISPLACEMENTS
STRESS-STRAIN RELATIONSHIP - CONSTITUTIVE LAWS (EXPERIMENT)
GOAL: SOLVE FOR DISPLACEMENTS RESULTING DUE TO APPLIED
EXTERNAL FORCES – DISPLACEMENT METHOD

THE STRUCTURE IS LINEAR. TO OBTAIN A COMPLETE SOLUTION

 DISPLACEMENT COMPATIBILITY &

 EQUILIBRIUM

HAVE TO BE SATISFIED THROUGHOUT THE STRUCTURE.

5
 BASICS OF THE FINITE ELEMENT METHOD

PRINCIPLE OF VIRTUAL DISPLACEMENTS

INTERNAL WORK = EXTERNAL WORK

g# Ng = R t r # , where
g# and r # are virtual displacements

VIRTUAL STRAINS AND DISPLACEMENTS SATISFY

COMPATIBILITY
THIS RESULTS IN AS MANY EQUILIBRIUM EQUATIONS AS
THE NUMBER OF VIRTUAL DISPLACEMENTS
CAN ALSO BE DERIVED BY THE PRINCIPLE OF MINIMUM
POTENTIAL ENERGY, U/ ri = 0

6
 TYPES OF ELEMENTS - USAGE

DEPENDING UPON THE –

 GEOMETRIC DIMENSIONS OF THE STRUCTURE &
 THE TYPE OF LOAD IT C ARRIES
THE FOLLOWING TYPES OF ELEMENTS CAN BE USED
 1 – D ELEMENTS – BARS, RODS, BEAMS
 2 – D ELEMENTS – MEMBRANE, BENDING, SHELL
 3 – D ELEMENTS – SOLID, AXISYMMETRIC
 SPECIAL ELEMENTS SUCH AS SPRING AND DAMPER

RECENT INVENTIONS ARE -

 GAP ELEMENTS
 CONTACT ELEMENTS
 ELEMENTS FOR MODELLING 2-D AND 3-D CRACKS
 ELEMENTS FOR FLUID STRUCTURE INTERACTION
 COUPLED FIELD ELEMENTS – ELECTROMECHANICAL,
PIEZOELECTRIC ETC.

7
ELEMENTS USED IN A FINITE ELEMENT MESH OF AN AIRCRAFT

* BAR
STRINGERS, LONGERONS, FRAME FLANGES & RIBS, WING RIB CAPS, WING SPAR
FLANGE, RIB AND SPAR FLANGES IN FIN AND RUDDER

* BEAM
MAIN AND NOSE UNDER CARRIAGE, RADOME BRACKET, BRAKE PARACHUTE
MOUNTING, PYLONS AND STORES

* QD4
ALL SKINS, SHEAR WALLS, LONGERON WEBS, FRAME WEBS, CANOPY BUBBLE
STUBWING SKIN, DIVERTER SKINS, AIR INATAKE & AIR CHANNEL SKINS, FLOORS,
FUEL TANK WALLS, WING BOX SKIN, NOSE SKIN, ELEV ONBOX SKIN FIN SKIN, NOSE
RIBS, FITTINGS, RUDDER SKIN, NOSE SKIN, RUDDER SPAR WEBS

8
 ELEMENTS USED IN A FINITE ELEMENT MESH OF AN AIRCRAFT

* QD4B
FRAMES, ENGINE MOUNT BRACKETS,
WING SPAR WEBS, WING RIB WEBS, NOSE RIB WEBS
SLAT RIB WEBS, SPAR WEBS FIN RIBS WEBS, FIN SPAR WEBS
* QD4SHEAR
BOTTOM LONGERON WEBS
* QD4PLATE
DOOR SKIN,ENGINE FITTING, RUDDER HORN
* HE8
HONEY COMB FOR ENGINE BAY DOOR, WING - TO SIMULATE INERTIA LOADS
DUE TO FUEL

9
 MESHING
IN ORDER TO DEVELOP A FINITE ELEMENT MESH
THE FOLLOWING NEED TO BE DEFINED –

GEOMETRY
BOUNDARY CONDITIONS
LOADING

AFTER MESHING THE FOLLOWING NEED TO BE

CHECKED –

ASPECT RATIO
ANGULAR DISTORTION – SKEW, TAPER
CURVATURE DISTORTION
MID-SIDE NODE POSITION

10
 SUBSTRUCTURING

A VERY USEFUL TECHNIQUE FOR ANALYSING LARGE

STRUCTURES – ESPECIALLY WHILE PERFORMING
COMPUTATIONALLY INTENSIVE JOBS.
THE STRUCTURE CAN BE DECOMPOSED INTO SEVERAL
SUBSTRUCTURES.
EACH SUBSTRUCTURE IS ANALYSED INDEPENDENTLY AND
IS READY FOR INTEGRATION, IT IS EXPRESSED IN TERMS
OF ITS BOUNDARY DEGREES OF FREEDOM.
COMPONENT MODE SYNTHESIS IS A POPULAR TECHNIQUE.
 Analysis
The importance of Finite Element Methods
need not be over-emphasized.
FINITE ELEMENT MESH OF THE AIRCRAFT

13
 Design Principles - Sizing
 Design requirements
 Structure must remain elastic up to limit loads
 Structure must carry ultimate loads.

 Analytical approach
 Margins of safety (MS) are computed for all members based on maximum
stresses and structural allowables to verify designs.

 MS = F/fmax -1
Allowables (Metals, Composites) Maximum applied stress or strain; developed from finite-element
analysis or traditional procedures

 Validation
 Design is validated by limit loads, ultimate loads, and destruction tests.

14
 SOLUTION OF THE EQUATION OF STATICS

[ K ] x = F - STATIC FORCES

[ K ] IS SYMMETRIC, SPARSE AND BANDED

THE TWO MOST WIDELY USED DIRECT SOLUTION METHODS
ARE -

 GAUSS ELIMINATION

[ K ] = LU

 CHOLESKY DECOMPOSITION (ONLY FOR POSITIVE

DEFINITE MATRICES)

[ K ] = LLT
 SOLUTION OF THE EQUATION OF DYNAMICS

[ M ]x+ +[ C ]x+ [ K ]x = F-INERTIAL+DAMPING+ STATIC

FORCES
[ M ] IS MOST OFTEN DIAGONAL (LUMPED)
IF F=0,FREE VIBRATION,EIGENVALUE PROBLEM
IF F=0,FREQUENCY RESPONSE PROBLEM
1.EIGENVALUE PROBLEM
[ K ]j = j[ M ] j, j = j2
[ K ] IS CORRECTED BY AN ARTIFICE CALLED ‘SHIFTING’ FOR
EXTRACTING ‘FREE-FREE’ MODES
j, ARE ALL ORTHOGONAL, j T[M] i=0, j= i

METHODS USED FOR EIGEN VALUE EXTRACTION (USED IN

MSC/NASTRAN) –
 TRANSFORMATION METHODS-GIVENS, HOUSEHOLDER
 TRACKING METHODS (ITERATIVE) – INVERSE POWER, STURM MODIFIED
 COMBINATION OF BOTH - LANCZOS
SYMMETRIC MODE
ANTISYMMETRIC MODE
FULL AIRCRAFT DYNAMIC TEST
 Causes of Differences
• Modeling of Damping
• Material Properties
• Modeling of Joints
• Finite Element Discretization
• Numerical Errors
• Number of Degrees of Freedom
Reconciliation of the DOF
 Correlation of Modes
• A Comparison of the Mode Shapes has to be done
• The Modal Assurance Criterion (MAC) provides a
measure of consistency (degree of linearity) between
estimates of modal vectors originating from different
sources – Test & Analysis
2
V test . Vana
i j

 
MAC 
 
V test .Vtest Vana .Vana 
i i  j j 

• MAC takes on values of -

Zero - representing no consistent correspondence One
- representing a consistent correspondence
 Correlation of Modes – Case Study
Mode Description % Difference in
Frequency between GVT
and Analysis*
Wing Bending- Symmetric (WSB) -0.5
Fuselage Vertical Bending (FVB) -3.4
Wing Bending - Antisymmetric + Fuselage Roll -8.2
(WAB + FR)
Fuselage Lateral Bending (FLB) -33.5
Fin Bending (FINB) -13.5

Analysis
Frequency
(Hz) FLB WSB WAB + FR FVB FINB

Test WSB 0 99.5 0 2.3 0

FVB 0.5 8.3 14.7 77 0.7

WAB + FR 6.8 0.2 97.9 0.1 13.9

FLB 74.2 0 19 0 94.6

FINB 45.1 0 6.7 0 96.3

MAC Table – Before Model Updating

 Updating The Model
1. Direct Matrix Updating
{XT(test)} [M] {X(test)} = Not Orthogonal

To force Orthogonality
{XT(test)} [M + M] {X(test)}= [ I ]
{XT(test)} [K + K] {X(test)} = [ 2 ]

M and K are the Updates

 …. Updating The Model
2. Physical Property Adjustment

• Certain physical locations on the FE

Model are modified, based on the
MAC.
• This has been done in the present Case
Study
Correlation of Modes – After Updating

Analysis

Frequency WAB +
(Hz) WSB FR FLB FVB FINB

Test WSB 99.5 0 0 6.2 0

FVB 7.9 16.1 1 78.8 0.7

WAB + FR 0.2 97 29.7 0.2 11.8

FLB 0 10.9 96.2 0 91.4

FINB 0 2.6 74.3 0 97.9

A Reasonably Accurate Global Finite Element Model is

Now Available for further Analysis !!!
 Various Forms of Analysis
 STATIC
 VIBRATION
 AEROELASTICITY (STATIC)
 BUCKLING
 OPTIMISATION
 AEROELASTICITY (DYNAMIC-FLUTTER, RESPONSE)
 AERO-SERVO ELASTICITY
 NON-LINEAR (CONTACT, GEOMETRIC/MATERIAL)
 COUPLED BEHAVIOURS
 THERMAL
 DYNAMIC RESPONSE
 DAMAGE/DEFECT ANALYSIS
 FRACTURE STUDIES
 SMART STRUCTURES
 A TYPICAL FLOW CHART FOR STRUCTURAL ANALYSIS AND TESTING

W/T
GROUP AIRCRAFT
AERODYNAMICS
GROUP
STRUCTURAL DYNAMICS &
AEROELASTICITY GROUP
CFD
GROUP AERO DATA

FE MODEL
STRESS
GROUP
ANALYSIS OF THE A/c TESTING OF THE A/c AT THE GROUND TEST CENTRE
FLIGHT
MECHANICS UPDATION OF MODEL

STATIC VIBRATION & FLUTTER GROUND STRUCTURAL IRON BIRD

AEROELASTICITY VIBRATION COUPLING

STRESS CLAW TEAM
MODAL PARAMETERS RESPONSES
GROUP
AERO CLAW TEAM
SERVO NOTCH FILTER
ELASTICITY DESIGN

STRENGTH
TEST CERTIFICATION

FLIGHT
 FREQUENCY RESPONSE PROBLEM

[ M ]x+ [ C ]x+ [ K ]x = F-INERTIAL+DAMPING+ STATIC

FORCES (1)
2. IF F=0,FREQUENCY RESPONSE PROBLEM
DAMPING HAS TO BE SPECIFIED-FROM EXPERIMENT
 DIRECT FREQUENCY RESPONSE
x = [-2[M]+i [C]+[K]]-1 F

 MODAL FREQUENCY RESPONSE

[] T[M] []=[] – MODAL OR GENERALISED MASS
[] T[K] []=[k] – MODAL OR GENERALISED STIFFNESS
ASSUME {x} = []{q}
PREMULTIPLY (1) BY [] T AND SUBSTITIUTE FOR {x}

[] T[M] []q+ [] T[C] []q + [] T[K] []q = [] T F

[]q+ [c]q+ [k]q = [] T{F

SOLVE FOR {q} AND CALCULATE {x}
 RANDOM FREQUENCY RESPONSE PROBLEM

[ M ]x+ [ C ]x+ [ K ]x = F-INERTIAL+DAMPING+ STATIC

FORCES
3. IF F=0 AND RANDOM (SEVERAL SIMULTANEOUS
FREQUENCY COMPONENTS)

 PERFORM THE FREQUENCY RESPONSE OVER THE

DESIRED FREQUENCY RANGE FOR A UNIT INPUT
 THE RESPONSE IS THE TRANSFER FUNCTION OF THE
STRUCTURE - H
 THE INPUT RANDOM SPECTRUM IS SPECIFIED AS A
POWER SPECTRAL DENSITY (e.g. g2/Hz) – pINPUT

pOUTPUT = pINPUT *H2

IN THE ABOVE ANALYSIS THE SIGNAL IS ASSUMED TO BE
STATIONARY & ERGODIC.
 TRANSIENT RESPONSE PROBLEM

THE PROBLEM IS SOLVED IN TIME DOMAIN BECAUSE THE

COMPLETE TIME HISTORY OF THE EXCITATION IS KNOWN.
THE PROBLEM IS SOLVED BY – DIRECT METHOD OR MODAL METHOD
OBJECTIVE: TO ANALYSE THE RESPONE TO THE TRANSIENT
INPUT. THE RESPONSE IS IN THE TIME DOMAIN
THE RESPONSE HAS TO BE DIGITISED (SAMPLED).
NYQUIST CRITERION: fsampling>2fbandwidth
where fs = 1/t, t being the sampling rate

INPUT
SIGNAL

TP

TT
P
Time
P TT Tim
e
Missile Thrust Profile & Aircraft Response
4
x 10
3.5

2.5

THRUST [N]
2

1.5

0.5

0
0 1 2 3 4 5 6
TIME [S]
AEROELASTIC DEFORMATION - 

ANGLE OF ATTACK
AEROELASTIC DEFORMATION - 

ANTISYMMETRIC ELEVON
CAUSING A ROLL
 AEROELASTIC EFFICIENCY

EFFICIENCY  = FLEXIBLE AERO DERIVATIVE

RIGID AERO DERIVATIVE
e.g
CL = RIGID AERO DERIVATIVE FOR LIFT DUE TO ANGLE OF ATTACK
FLEXIBLE AERO DERIVATIVE = f (Flexibility, Dynamic Pressure)
AEROELASTIC EFFICIENCY
A POTATO
FULL AIRCRAFT STATIC TEST
 DIVERGENCE
EXCEEDANCE OF THE AERODYNAMIC MOMENT ABOVE THE
TORSIONAL STIFFNESS OF THE STRUCTURE
REGULATIONS REQUIRE SAFETY MARGIN ON VELOCITY
BEYOND FLIGHT ENVELOPE

TORSIONAL DIVERGENCE OF WING – MOST COMMON

Elastic Restoring Moment

M/ 

Dynamic Pressure qD q
… DIVERGENCE
UNSTEADY AERODYNAMICS
DYNAMIC AEROELASTICITY
(FLUTTER)
FLIGHT FLUTTER TEST
 ASE Tetrahedron
Servodynamics

Collar’s Aeroelastic Triangle

Aerodynamics A

DYNAMIC Elasticity
E
AEROELASTICITY

Inertia

I+E : Structural Vibration

A+I : Flight Mechanics
A+E : Static Aeroelasticity (Divergence, Control Reversal, Load Redistribution, etc.)
A+I+E : Dynamic Aeroelasticity (Flutter, Gust -response, Buffeting, etc.)
S+A+I : Flight Dynamics & Servo Control
S+I+E : Servoelasticity
S+A+E : Static Aeroelasticity influenced by Servo characteristics (Control Effectiveness, etc.)
S+A+E+I: Aeroservoelasticity
 Integrated FCS Schematic

ACTUATORS AIRCRAFT SENSORS

DFCC NF

PILOT
COMMAND
Aircraft in SCT (suspended to simulate free flight BC)
 BIRD-STRIKE ON THE LCA WINDSHIELD
• Bird Strike Requirements of Fighter Aircraft Windshield as per
MIL-W-81752 : should sustain impact of 4 lb bird at 850 Kmph

Stretched Acrylic (SA)

Bird

Polycarbonate (PC)
Polyurethane(PU)

• Bird Impact Tests

ADA/ NAL/ GTRE

• Bird Impact Analysis

ADA - LS-Dyna explicit solver
 ANALYSIS/TEST SEQUENCE FOR LCA WINDSCREEN

Low Inclined Monolithic Sandwich

Simplistic velocity flat Windshield Windshiel Realistic
Impact plates d

6 Configurations

9 Configurations :
5 Monolithic, 4 Sandwich

10 Configurations

16 Configurations :
7 with 1mm PU layers,
9 with > 1mm PU layers
 ANALYSIS/TEST CORRELATION ON FLAT PANELS

Analysis

SA plate, bird v=648 kmph PC plate, bird v=428 kmph

Tests at GTRE