Rochester Institute of Technology

RIT Scholar Works

Theses

Thesis/Dissertation Collections

1989

Investigation of substructuring principles in the

finite element analysis of an automotive space
structure
John E. Martin

Follow this and additional works at: http://scholarworks.rit.edu/theses

Recommended Citation
Martin, John E., "Investigation of substructuring principles in the finite element analysis of an automotive space structure" (1989).
Thesis. Rochester Institute of Technology. Accessed from

This Thesis is brought to you for free and open access by the Thesis/Dissertation Collections at RIT Scholar Works. It has been accepted for inclusion
in Theses by an authorized administrator of RIT Scholar Works. For more information, please contact ritscholarworks@rit.edu.

Investigation of Substructuring Principles

in the Finite Element Analysis of
an Automotive Space Structure

JOHN E. MARTIN
A Thesis Submitted in Partial Fulfillment
of the
Requirements for the Degree of
MASTER OF SCIENCE
ill

Mechanical Engineering
Rochester Institute of Technology
Rochester, New York

May 1989

Approved by:
Dr. JOiSepb S. Torok (AdvisQr)

Dr. Richard G. Budynas

Dr. Hany Ghoneim

Dr. Wayne W. Walter

Dr. Bhalchandra V. Karlekar

Professor and Department Head

ACKNOWLEDGEMENTS

This

work

siblings

is dedicated to my

Chris, Sandy

diversions

provided

Ruth R. Martin, my father Edward L. Martin

Susan Martin.

and

during

mother

the year that

For the love, support,

I have been researching

To my advisor, Dr. Joseph S. Torok. You

when

it.

needed

Thank

you

for

and

and

three

my

encouragement and constant

writing this thesis, thank

were always able to provide a

kind

your support and encouragement on the

you.

word and smile

thesis and

other

professional matters.

To Dr.

Hany Ghoneim,

for his

assistance when

learning

was

to use

ANSYS.

deeply

appreciate your words of encouragement.

To

Tony

Lam

from Stress Technologies Incorporated. I just

appreciate your technical advice

smoothly

as

it did,

To the
and

My

know how

research would never

have

much

gone as

without your assistance.

To my friends, I-en
companionship

concerning ANSYS.

want you to

and

"Samuel"

light diversions

members of

Lin

you

Shashank Kolhatkar,

through all of this

my thesis defense

Dr. Wayne Walter. Thank

and

committee:

for taking the

crazy

time.

who

provided

study my

with

Thanks for the fun.

Dr. Richard G. Budynas, Dr.

time to

me

Hany

Ghoneim

work and make your comments

and suggestions.

Finally, I
helped

me

would

find the

like

to thank

my fellow

members

strength and spiritual growth

Thanks to you, I hope

to go on and accomplish

that

from
I

the

Rochester Chapter

needed to

something

face the

remarkable with

of

challenges

my life.

NSA. You
in my life.

ABSTRACT

The efficiency
using

degree

of

university
of

using substructuring in the dynamic

version of

freedom

ANSYS to

The

model.

model

perform a series of

is based

automotive space structure and was constructed

the model was

model was

divided into

different

divided into one, two,

three and

Based
potential

eigenvalues and the

for saving CPU time.

tolerance of

that agree with the

CPU seconds)
freedom

of

CPU

savings

placed

Minimizing

four

three to

along

time needed to

in CPU time. It is

the substructure

the number of

time needed to

boundary

find

plate elements.

find

of

degrees

a solution with a

study,

used, the

each case

freedom

an

study, a

configurations.

dynamic substructuring has

a great

in

natural

be found for

demonstrated

boundaries has

In

648

for

each case

configurations

respectively.

degree

In

by

a solution were compared.

shown that

could

also

master

master

In the

which resulted

of a non-substructured

digits

examined

case studies on a small,

substructured solutions

frequencies

significant

using STTF 63

for different

Optimal

is

upon the monocoque center section

four substructures,

experiments, it is

upon numerical

frequencies

CPU

of small models

four

substructure configuration.

series of modal analyses was performed

The first four

analysis

baseline

solution

an appreciable

within a

(17.49-70.16

that the number of master

degrees

of

the strongest effect upon solution efficiency.

of

freedom is

desired level

shown

of accuracy.

to minimize the amount

THESIS OUTLINE

ABSTRACT
TABLE OF CONTENTS
CHAPTER 1

INTRODUCTION TO SUBSTRUCTURING

1)

Basic Concepts

2)

Mathematical

Theory

CHAPTER 2.

BACKGROUND/LITERATURE SEARCH

CHAPTER 3.

EXAMPLE PROBLEMS

1)

ID

2)

2D Truss

Spring

CHAPTER 4.

THE GENERAL METHOD OF SUBSTRUCTURING

CHAPTER 5.

SUBSTRUCTURING IN ANSYS

1)

The Generation Pass

2)

The Use Pass

3)

The Stress Pass

4)

Relating

the

Passes

a) General File

Handling

b) Substructure Connectivity
c) Combination

of

Post

Processing Files

(FILE 12) using /AUX1

d) Combining File Handling Commands into

e) Master Degrees

CHAPTER 6.

of

Freedom

Finding

the

Run Files

Totals

CASE STUDIES

1)

Description

of

Problem

a) Modal Analysis

of the

Monocoque Center Section

Space Structure

b) History
c) Material ASTM 5052 Aluminum

d) Other
ii

of an

Automotive

2)

Research
a) One Substructure

b) Two Substructures
c) Three Substructures

d) Four Substructures
CHAPTER 7.

CONCLUSIONS

REFERENCES

APPENDICES

1)

ANSYS COMMANDS

2)

ANSYS RUN FILES FOR THE FOUR CASE STUDIES

BIBLIOGRAPHY

ui

TABLE OF CONTENTS
PAGE

LIST OF FIGURES
LIST OF TABLES

vii

LIST OF SYMBOLS

viii

CHAPTER 1.

INTRODUCTION TO SUBSTRUCTURING

CHAPTER 2.

BACKGROUND / LITERATURE SEARCH

10

CHAPTER 3.

EXAMPLE PROBLEMS

23

CHAPTER 4.

THE GENERAL METHOD OF SUBSTRUCTURING

34

CHAPTER 5.

SUBSTRUCTURING IN ANSYS

39

CHAPTER 6.

CASE STUDIES

49

CHAPTER 7.

CONCLUSIONS

--

82

REFERENCES--

88

APPENDIX 1. ANSYS COMMANDS

90

APPENDIX 2. ANSYS RUN FILES FOR THE FOUR CASE STUDIES

100

IV

LIST OF FIGURES
FIGURE

PAGE

1 1

Models

The Discretization of the Model into Substructures

Substructures into Elements

.2

.3

Demonstrating the Flexibility of the Finite Element Method

A Graphic Explanation

of

the

Master

1.4

A Discrete Dynamic System Model

2. 1

Schematic of the
Highlighted

Boeing 747

2.2

Schematic

Substructures

the

and of

Slave Degrees

of

Freedom

the

Wing-Body Intersection

in the Analysis
Intersection of the Boeing 747

of

Wing-Body

with

and

used

of the

11

12

2.3

Finite Element Models of the Two Substructures

Modeling
the Monocoque Center Section of the
Boeing 747

13

2.4

Finite Element Models

14

the

Wing

Structure

Two Substructures Modeling

Wheel Well Area of the Boeing 747

of the

and

3.1

The Model for

ID

3.2

The Model for

2D Truss

4. 1

A Finite Element Grid

Degrees

of

Spring Substructuring Example Problem

Substructuring Example Problem

Displaying

Freedom, Boundary

Two Kinds
Internal

the

and

of

Two Different Discretizations

the Case Studies

5.1

Row Chart Displaying the File

Multiple Substructure Analysis

5.2

The Generalized Input Files for

the

Generation

5.3

The Generalized Input Files for

the

Stress Passes

5.4

A Method for Finding the TOTAL Values Shown

with a Three-Substructure Model

6. 1

The Basic Monocoque Center Section

with Suspension Attachment Sections

6.2

Plots

6.3

A Table

6.4

The Substructure Discretization for the Four-Substructure

Case Study

6.5

TOTAL Value Calculations for the Four-Substructure Case

of the

the

Results

35

Plate Model Used in

36

Handling for a Generalized

and the

and

41

Use Passes

of the

44

45
-

Center Section

First Four Mode Shapes

Displaying

29

Master

4.2

of the

24

48

50

51
Baseline Analyses

52

55

Study

56

6.6

Flow Chart Displaying the File

Four-Substructure Case Study

6.7

The Substructure Discretization for

Case Study

6.8

TOTAL Value Calculations for the Three-Substructure Case

6.9

Row Chart Displaying the File

Three-Substructure Case Study

6.10

The Substructure Discretization for

Case

58

Handling for the

the

65

the

69

Two-Substructure

Study

TOTAL Value Calculations for the Two-Substructure Case

6.12

Row Chart Displaying the File

Two-Substructure Case Study

6.13

The Substructure Discretization for

6.14

63

Study

Handling for the

6. 1 1

Case

62

Three-Substructure

Handling
the

for

70

Study

72

the

76

One-Substructure

Study

Flow Chart Displaying

One-Substructure Case

the

File

Handling

for

77

the

Study

7. 1

Bar Graph of the CPU Times Needed to Analyze and Post

Process Models Using the Optimum Master Degree of
Freedom Configuration

7.2

Model Frequencies for the Master Degree of Freedom

Configurations of the Two- Substructure Case Study

A. 1

Row chart Displaying the File Handling for

Sample Input File for /AUX1 module

vi

the

85

86

99

LIST OF TABLES
TABLE

PAGE

6. 1

Data for the Four-Substructure Case

6.2

Data for the Three-Substructure Case

6.3

Data for the Two-Substructure Case

6.4

Data for

the

One-Substructure Case

Study

Study

Study
Study

VII

59

66
73
78

LIST OF SYMBOLS

Cross Sectional Area

Ad(v,p)

Generalized Form

Ae

Matrix

Ai

Kss Matrix for Substructure i

Ao

Matrix Assembled

ALPHA

Coefficient Vector used in Numerical Solution

Mode Shapes Related

BDOFi

Number

BETA

Angle

Bi

Ksm Matrix for Substructure i

Bp

Preconditioning Matrix

Damping Matrix

CDOFi

Number

Ci

Kms Matrix for Substructure i

Co

Cosine

Domain in Dirichlet Problem

Young's Modulus

F(t)

Dynamic Load Vector

f(s)

Spatial Distribution

Fl F2

Forces Applied to Nodes 1 2, 3, 4 in

used to

of

of

of the

find ALPHA in

of

to

the

Finite Element Solution

Kmm Matricies for

out of the

of the

of

Inclination

of

of

Dirichlet Problem

Substructures

Dirichlet Problem

Substructure i

General 2D Truss Element

Constrained Degrees

Angle

all

of the

Kron's Matrix

Boundary Degrees of Freedom in

Inclination

of

Dirichlet Problem

Load in

of

of

Freedom in Substructure i

Truss Element

Dynamic Equation

the

First Example Problem

F3F4

in

Basic Substructure

Fa Fb Fc

Nodal Forces

FDOFi

Number

Fi

Forces

at the

Fm

Vector

Containing Loads

Fml

Load Vector for the Master Degrees

of

Freedom

of

Substructure 1

pm2

Load Vector for the Master Degrees

of

Freedom

of

S ubstructure 2

Fmi

Load Vector for

Master Degrees

of

Freedom

of

S ubstructure i

Fj

of

used

the

Free Degrees
Ends

of the

the

of

Freedom in Substructure i

Static

Spring

Element

Applied to Master Degrees

VIII

of

Freedom

FnlUm)

The Condensed Generalized Governing Matrix Equation in Terms

Degree of Freedom Vector JJrri

Fc.

Sum

Fp

The Force that is Applied to Node 4, F4

Es

Vector

Fsl

Load Vector for

the

Slave Degrees

of

Freedom

of

Substructure 1

Fs2

Load Vector for

the

Slave Degrees

of

Freedom

of

Substructure 2

Fsi

Load Vector for the Slave Degrees

of

Freedom

of

Substructure i

Fsub

Condensed Load Vector for

Fsubl

Condensed Load Vector for Substructure 1

Fsub2

Condensed Load Vector for Substructure 2

Fsubi

Condensed Load Vector for Substructure i

Fxi Fyi

Force in

the

and

Y Directions

of the

Fxj Fyj

Force in

the

and

Y Directions

of

Stiffness Matrix

K1K2

Spring

Ks

Spring Stiffness

Kss Ksm
Kms Kmm

Submatricies

Kssl Ksml
Kmsl Kmml

Kss, Ksm, Kms

and

Kmm for Substructure 1

Kss2 Ksm2 Kss, Ksm, Kms

Kms2 Kmm2

and

Kmm for Substructure 2

Kssi Ksmi
Kmsi Kmmi

Kss, Ksm, Kms

and

Kmm for Substructure i

Ksub

Condensed Stiffness Matrix for

Ksubl

Condensed Stiffness Matrix for Substructure 1

Ksub2

Condensed Stiffness Matrix for Substructure 2

Ksubi

Condensed Stiffness Matrix for Substructure i

Operator in Dirichlet Problem

Lg

Length

Mass Matrix

of

Fm Vectors for

all

of

of a

Master

Substructures

Containing Loads Applied

Stiffnesses that

of the

make

to

Fp

Slave Degrees

of

Freedom

Given Substructure

the

i th End

of

Element

End

of

Element

th

up the Substructures in the First Example Problem

Partitioned Stiffness Matrix

Given Substructure

Truss Element

IX

Mss

Msm

Submatricies

Partitioned Mass Matrix

of the

MmsMmm
P

Modal Coordinates

l(t)

Sinusoidal Time Function

R(w)

Kron's Matrix

SO)

Matrix Function Similar to Kron's Matrix

Si

Sine

TBDOF

Angle

of

Inclination

Total Number

of

Boundary Degrees of Freedom in

TDOFi

Total Number

of

Degrees

TMDOF

Total Number

of

Master Degrees

TNBMDOF

Total Number

of

Nonboundary Degrees

TOTLi

TOTAL Value in Substructure i

Vector

Ui

Xs Vector for Substructure i

Um

Master Degree

Uo

Sum

Uv

Approximation Function

Vi Zi

Orthogonal Vectors

that

Hi Can Be Decomposed into

Vo Zo

Orthogonal Vectors

that

Uo. Can Be Decomposed into

Natural

wl w2

The First Four Natural Frequencies for

of

of

of

Truss Element
the

Model

Freedom in Substructure i
of

Freedom in the Model

of

Freedom in

the

Model

Containing Nodal Displacements of the Finite Element Model

of

of

Freedom Vector

Xm Vectors for Substructure i

used

in

the

Numerical Solution

of

Dirichlet Problem

Frequency
the

Model

w3 w4

for x(t)

Amplitude

x(t)

Dynamic Nodal Displacement Vector

x'(t)

First Derivative

x"(t)

Second Derivative

Xa Xb Xc

Nodal Displacements

Xbi

Basis Functions

Xi

End Displacements

Xj

vector

Xm

Vector

Xml

Master Degree

of the

of the

of

Dynamic Nodal Displacement Vector

used

in the Basic Substructure

Subspace S

of

Containing

Dynamic Nodal Displacement Vector

of

(D)

Spring of Stiffness

Master Degrees

of

Ks

Freedom

Freedom Vector for Substructure 1

Xm2

Master Degree

of

Freedom Vector for Substructure 2

Xmi

Master Degree

of

Freedom Vector for Substructure i

Xl

Special Ritz Vectors Xrl

Xri

Xs.

Vector

Xs2

Slave Degree

of

Freedom Vector for Substructure 2

Xsi

Slave Degree

of

Freedom Vector for Substructure i

Xti Yti

and

Y Displacement

of

th

End

of

Truss Element

Xtj Ytj

and

Y Displacement

of

th

End

of

Truss Element

%i

Fraction

th

1st Special Ritz Vector

Special Ritz Vector

Containing

of the

Slave Degrees

Total Number

of

of

Freedom

Free Degrees

XI

of

Freedom in Substructure i

CHAPTER 1

An Introduction to

The

advent of

high

speed

digital

certain

types

of

problems, but

and size.

geometry
use

long enough

and

ANSYS,

are available

Finite

they

are capable of

use on

many

[1], [2]

finite

or

the

[3]. See Figure 1.1 for

analysis of a

large finite

for solving

more efficient

problems of a specific type of

hand, is extremely flexible. It has been in

packages, like NASTRAN

element software

software packages are

dynamic

extremely

large

micro computers to

versatile and

has

finite

made

and

geometry

be found in

examples of models that

They

supercomputers.

analyses on structures with a wide

element method can

element

several smaller problems that can

decomposing
dividing

large finite

the model

into

individually

analyzed

into

may be

solving

structures with unusual

finite

element method the standard

variety

element analysis the

boundary

conditions.

standard references such

demonstrate

the

applicability

of the

element method.

In the

few

to

of these software packages

for analyzing

General information concerning

as

These

of static and

variety

The availability

method of preference

finite

complete

hardware ranging from

performing

of geometries.

limited

are often

element analysis, on the other

that there are

finite

made the

analytical techniques

that are available commercially.

for

has

computers

Certain

method of structural analysis.

Substructuring

"master"

degrees

Once

freedom
which

is

of

freedom. The

this equation

performed once

initial

the

is known

to produce condensed

be

solved

is solved, its
This

for

analysis

substructures
separated

degrees

of

step,

each substructure

substructure.

master and slave

They

are

will

degrees

be

of

provide

are placed

information

in

is

being

The degrees

freedom. The

usually located along

other

expanded

in

consists of

substructures are

in

terms of a

eventually

degrees

of

by

combined

freedom in the

terms of all of the

performed

describes the behavior

into

carried

the substructure with other substructures

freedom

equations are

degrees

of

a series of calculations

is divided into elements,

shows an example of a model

freedom that

The

of

each substructure.

being discretized into elements.

into

as substructures.

into

strategy

Substructuring

of all of the the master

be

used

matrix equations which are

expansion

solution

One commonly

governing

solution can

matrix equation of the substructure which

developed. Figure 1.2

known

condensed

to sub-divide the problem

as substructuring.

governing

in terms

desirable

often

individually.

several sections which are

of each substructure.

During

solved

element model

one equation which can

model.

be

model, it is

in

the main

substructure

degrees

governing

In

about a given point

in

freedom

they

are the

only

are used to

link

cases, master degrees of

master

the substructure and are often

of

where

certain

nonboundary locations. These internal

the substructure are then

equation associated with a

boundaries

the use pass.

be

substructures and of those

freedom in

master

governing

of that substructure can

divided into
of

so that the

degrees

of

freedom

included to improve

FIGURE 1.1
EXAMPLES OF THREE-DIMENSIONAL
FINITE ELEMENT MODELS

Patran Model

Images 3D Model

of a

of

Parabolic Antenna

Supersap

Model

of a

Mounting Bracket

Models from Mitchel, Falk

Valve

Housing

FIGURE 1.2
SUBSTRUCTURE DISCRETIZATION

SS 1

SS2

SS3

Figure 1.2
Discretization

of the

Model into Substructures

SS2

SSI

Figure 1
Discretization

of the

SS3

.2

Substructures into Elements

accuracy in
main

more

by

often performed

Once

equation.
can

be

large

Finally,

form

governing

large

one

then solved

is expanded,

the solution

for

slave

degrees

once the master

master

in terms

degrees

of

of

freedom

degrees

freedom

freedom. Once

degrees

of the master

of

are not used

of

in

the substructure
process

in the governing

formed for each substructure, they

matrix equations are

The

matrix equation which models the entire structure.

in terms

so that

of all of the master

it is in

the

known.

are

freedom This

a series of procedures which condense each matrix

added together to

is

solved

condensed

the condensed

matrix equation

The

analysis.

information concerning

is created, it is

matrix equation

be

They can

equation.

governing

See Figure 1.3 for

is

like dynamic

certain cases

degrees

of

freedom in

degrees

terms of all of the

of

the model.

freedom

of

each

substructure.

Substructure

analysis

is

Generation

pass and the stress pass.

in terms

each substructure

the condensed

resulting

governing

equation

is

structures.

finite

As

matrix equations

solved

often

in terms

because it is

element problems that are too

large

finite

efficiently
savings

with conventional

in CPU time

use pass solves

Time may

among

also

scheme, the

by

major reason that

difficult to

The

model will

finite

not

must

to

be

element

of

the

capacity

Even

have

to

added

in

solving

analyst to take care of

and software to

when

of the

hardware

One

handle. Since

is

reason that

than would

small enough

solved

between iterations

to

for
to

complex

be

solved

can produce a significant

CPU time

be

user

and software available

the model generation tasks

the model

large

the number of

passes, it enables the

can

for in

certain generation or stress passes.

made

not used

in many

will

be saved, is

that the

a standard analysis.

In

only

design iteration

need

to

have

their

For

substructures

certain analysis

problems

are not considered

in

types,

way that

extra

it is

that

much more

file

in

are

a standard analysis.

will minimize the

extra master

to assure the accuracy of the solution.

many details ranging from

is

When substructuring is performed, there

concerning factors that

be discretized into

order

stress passes expand

increase in

an

techniques, substructuring

freedom

substructuring is

be

freedom. The

once, for the first iteration.

needed to compute the solution.

have

be beyond

perform than a nonsubstructured analysis.

many decisions that

of

large for the existing hardware

substructures that remain unchanged

generation passes performed

The

degrees

advanced, there has been

when used effectively.

saved

use pass,

one of the most efficient ways to model certain

several analysts.

for fewer degrees

be

In the

for

all of the substructures are added together and the

Substructuring also makes it easier to divide

element models

matrix equation

of that substructure.

analysis pass with several smaller

solve models that would otherwise

to the analyst.

freedom

governing

level.

used,

replaces one

from

of

of the master

technology has

aerospace

substructuring

degrees

the generation pass, the use

of passes:

passes create the condensed

of the master

the solution on the substructure

Substructuring is

kinds

organized around three

degrees

Substructuring

of

CPU time

freedom

will

also requires the

handling to verifying that the node

FIGURE 1.3
MASTER DEGREES OF FREEDOM IN A SUBSTRUCTURE

<>

Master Degrees of Freedom

Along the Substructure
Boundaries

Internal Master Degrees of Freedom

configuration on both sides of a substructure

for the

opportunities

The

most

important

be

not

part of

substructuring is

Likewise,
from the

degree

the

have

ability

a major

impact

use pass will

help determine

functions

are carried out

equations.

Since

initial

and expansion

the

formulas

In [2], the derivation

governing

degrees

freedom

of

equation

freedom,

the

upon

of the expansion

expansion

the

of

the condensation of the

formula
the

in

accuracy

terms of the

models, the

by formulas

of

of the condensation

degrees

degree

the final solution.

of

of

The

that use components of the

vary for different

analysis

freedom
in

of that

the use pass.

freedom

solution

condensation and
matrix

governing

types, the

condensation

be different.

formulas

of the static analysis condensation and expansion

for

more

matrix equations

of the solution computed

to expand the master

accuracy

matrix equation can

will

many

smaller

governing

The ability

solution.

to translate the properties of each substructure

substructure will

For

are

worth the extra effort.

and the expansion of the master

formulae

There

are compatible.

analyst to make mistakes when substructuring.

in CPU time may

savings

boundary

each substructure.

into the

which enter

The

matrix equation

use pass are separated

is

are

based

upon

partitioned so that the master

from

the

remaining

slave

degrees

freedom.

of

Kss

Ksm

Xs.

ES

(1.1a)

Fm

(1.1

Kms

Xm

and

Xs.

are the

displacement

displacements respectively,

Xs

results

Kmm

and

Xm

vectors

Fm

and

containing

Fs_ are

the

the master and slave

corresponding load

vectors.

degree

of

Solving

b)

freedom

(1.1 a) for

in
-1

Xs.
which can

freedom

Kss*{Fs.

be inserted into (1.1 b) to

vector

(1.2)

Ksm*Xm}

provide an expression that

is in

terms of the master

degree

Xm
-1

Kmm*Xm

Equation (1.3)

can

be

Kms*Kss*

{Fs_

Ksm*Xm}

Kms*Kss*Ksm)*Xm

which

{Fm

Kms*Kss*Fs_}

(1.4)

rewritten as

Ksub*Xm

in

(1.3)

-1

-1

be

Fm

rearranged to obtain

(Kmm

which can

the values

for Ksub

and

Fsub

1.5)

Fsjik
are given

by
6

the

following formulae:

of

-1

Ksub

Kmm

Kms*Kss*Ksm

(1.6 a)

-1

Fsub
Equations (1.6 a)
degree

of

and

freedom

stress passes

(1.6 b)

Fm

Kms*Kss*Fs

(1.6

into terms

are utilized to condense the matrix equation

The

configuration.

using the formula in

master

degree

(1.2)

equation

to

freedom

solution

is

the slave

degrees

of

of

find

of

b)

its

master

then expanded

freedom

in the

of the given

substructure.

For dynamic analysis, displacement

dynamic finite

x'(t)

and

x"(t) are

in

the components
mass

distribution

the

vectors

loads

C*x'(t)

K*x(t)

time.

of

equation of motion

for

(1.7)

and second

on the

degrees

This

K*x(t)

of

results

derivatives

mass

in

matrix,

damping

freedom. For

with respect to time of

which approximates

F(t) is

matrix.

modal

analysis, the

0_

forcing

damping

and

(1.8)
expressed as

(1.9 a)

X*SIN(wt)

the

the

the equation of motion

is assumed, the displacement is

x(t)

The basic

F(t)

M is the

vector x(t).

omitted.

generally

motion

function

containing the first

M*x"(t)
Since harmonic

substructure, and C is the viscous

the

vector which represents

forcing functions are

displacement

of the

is

is

element model

M*x"(t)
where

so that

x"(t)

The

equation of motion

(1.8) is

(1.9

-w*X*SIN(wt)

b)

thus reduced to a the standard eigenvalue problem

2
(K

Equation
vectors

(1.10)

X. This

explained

in [3].

master and slave

can

be

w*M)*X

solved

equation can

for

be

(1.10)

the natural

condensed

degrees

of

freedom
:

equation

using the

(1.10) is in

the

following reduction

form

equation

(1.10) is

as

partitioned so that the

are separated.

2 Mss

Ksm
-

Kms

w, and their corresponding modal

for substructuring using Irons Guyan reduction,

In applying Irons Guyan reduction,

Kss

Once

frequencies,

Kmm

of

Msm

(1.11),

Mms

Mmra

Xs.
=

0.

the stiffness and mass matrices can

formulae:
7

(1.11)

Xm

be

condensed

FIGURE 1.4
DISCRETE DYNAMIC SYSTEM MODEL

-1

Ksub

Kmm

Kms*Kss*Ksm

(1.12 a)

-1

Msub
It is

Mmm

-1

Mms*Kss*Ksm

notable that the condensed stiffness matrix

in

shown

analysis,

formulation that
in the

exists

stiffness
mass

In

master

formulation is identical

formulation,

freedom

expanded

in the

master

The

stress pass

the use pass solution will

mode shape

Xm. The

include

degrees

of

static

an exact

freedom

that

used to condense the

provides an approximation of the

analyst must

the natural

degree

master

for

(1.6 a), is

or

is

that

(1.12b)

to that used

(1.12 a)

configuration that will produce the proper mass

freedom

of

by

(1.12b)

associated with the original mass matrix.

dynamic analysis,
degree

given

relationship between

matrix, the mass matrix condensation formula

of

Mss*Kss*Ksm)

Unlike the formula (1.12 a),

uncondensed matrix.

degree

This

a).

provides the same stiffness

distribution

master

(1.6

equation

-1

Kms*Kss*(Msm

of

be

careful to

input

distribution.

frequencies,

freedom

w, and the

be

mode shape can

following expression:

using the

-1

Xs.
One factor
model.

that

Since the

is

quite significant

in dynamic substructuring is

stiffness relationships

in

a substructure are

to condense the stiffness matrix of a substructure without

the master

degrees

of

freedom

(1.13)

-Kss*Ksm*Xm

of that substructure.

consists of the stiffness matrix and a

readily

condensed

distribution

within the

usually linear in nature, it is

altering

The

the mass

the stiffness

possible

relationship between

static analysis matrix equation

load

vector.

Because there

(1.1)

are no

condensation problems associated with a static model, the substructured solution of a static
problem can

be

distribution is

not a

as accurate as a nonsubstructured solution of the same

linear relationship

an approximation of the mass

use the mass matrix

solutions when

Accuracy

in

and the mass

solved

large

degrees

in the

enough

model.

Mass

and the condensation of the mass matrix will always result

distribution
governing

of

the

matrix

original mass matrix.

equations,

Dynamic problems,

will therefore result

in less

in

which

accurate

substructuring is implemented.

the number of master

master

their

basic

of

distribution

degrees

freedom,

use pass.

If the

of

the substructured model can

freedom in

the user
analyst

to cancel out the

of

the model.

increases the

is

not

However, by increasing

size of the matrix equation

careful, the use pass

potential savings of

be improved

CPU time

matrix equation

afforded

by

by increasing

the number of

(1.5)

that will

be

(1.5) may become

a substructured model.

CHAPTER 2

LITERATURE SEARCH

Although

more attention

has

analysis

seen more

development.

has been focused

developments

Its implementation

computers made use of the

applications of

then, substructuring

of a

this

concerning

investigation

Although
to the

in

747

body

not

was

that there

in 1968

from

in

small sections

in

mentioned

this chapter

form

of

Many

of the

was

early

being

finite

in ANSYS

and

new

in digital

[6] (Figures 2.1-2.4).

a significant volume of

as

the

Since

elements
well.

information

discovered

during

technical

journals.

there are references that

date back

books,

in

to articles

is recent,

element

totally

when advances

The information that

text

not a

of the more complete

is

finite

a notable example

introduced in 1972,

surprising

most of the material presented

1960s. G. Kron

industry,

airliner

the principles of substructuring.

ranged

in the 1960's,

the aerospace
wide

it

where

Given this background, it is

available

applications, substructuring is

have been included in many

capabilities

the eighties, as

element method much more practical.

Boeing

including NASTRAN,

substructuring in

started to spread

were

substructuring

wing-body intersection

codes

finite

and

on

substructuring in his book Diakoptics. in 1963

[7].

Turner

Milsted

forced response,

different

examined several

Kron substructuring [8].

with
and

and

methods of coordinate reduction

Kron substructuring,

uses generalized

forces

and

be

which can

displacements

produce a global system-wide characteristic equation which can

The

equation

characteristic

related

to modal analysis

for compatibility

for both

used

modal analysis

at substructure

be

solved

boundaries

for the desired

R(w) is Kron's

w, squared.

This

problem

extraction method.
uses the

bisection

Creating
large

and

matrix, in

(2.1)

method

solving

substructures.

substructures are

the

A less

Sturm counting

to solve transcendental

the

One

Kron

broken up into

is

is

method of

expensive solution

(2.1)

is

10

frequency,

standard eigenvalue

Williams [9],

which

effective.

can

be extremely

This may

coordinate reduction.

the modal

and

to use a pyramiding scheme

the extra substructuring

in

using the

Wittrick

functions, is

additional substructures.

but

a polynomial

to analyze

characteristic problem

approach

solve the characteristic problem,

expensive.

which each component

is usually too difficult

However,

output.

is

R(w)*B=0

where

to

in

which

reduce the

involved

makes

Coordinate

demanding

for

the original

time needed to

it computationally

reduction

is

a method

FIGURE 2.1
747
THE ANALYZED SECTIONS OF THE BOEING

CARGO

DOOR CABIN ANALYSIS

Intersection Highlighted
747 Schematic With Wing-Body

(From Hansen

11

et.

al.)

FIGURE 2.2
SUBSTRUCTURES USED IN THE ANALYSIS OF THE BOEING 747

Schematic

of

Substructures

used

in the Analysis

of the

(From Hansen

12

Wing-Body

et.

al.)

Intersection

of the

Boeing 747

FIGURE 2.3
MONOCOQUE CENTER SECTION
SUBSTRUCTURES A & B

Finite Element Models

of

Monocoque Center Section Substructures A & B

(From Hansen

13

et.

al.)

FIGURE 2.4
WING & WHEEL WELL AREA
SUBSTRUCTURES C & D

TORQUE BOX

KEEL BEAM

Finite Element Models

of the

Wing Structure

and

(From Hansen

14

Wheel Well

et.

al.)

Area, Substructures C & D

that

is usually invoked

Two

substructure.

at the substructure

common

forms

level to

of coordinate reduction are mass

degrees

of

freedom in the

condensation,

which reduces

reduce the number of

the size of the mass matrix, and modal truncation, in which the number of modes to be extracted
reduced.

using

Both

methods of coordinate reduction can produce accurate results.

truncation, it is important to

modal

The counting

Method,

is

which

finding

slow at

Method

algorithm
used to

find the

upon

quadratically

roots of a

One alternative,

the roots.

find

and can therefore

applied to the

S(A) is

matrix

substructuring

still

faster.

of

implies

scheme

Lanczos

The Lanczos

Kron

In

in [10], is

is

Bisection

to replace the

converges

[11], Newtonian iteration has been

This

(2.2)

equation

problem

is

Newtonian iteration

that

algorithm avoid

iteration

process can

be

the

does

costly

is reliable,

similar to the

be

can

Kron

in Kron

used

terms,

which

into

a tridiagonal

which an orthogonal matrix

Kron

condensation of the

stopped short of complete

faster

is computationally

Sehmi [12]

by

and

as a

way

of

avoiding

matrix

employed.

matrix to a scalar

tridiagonalization,

expensive.

compatibility

indefinite

is

than the

Method. Unfortunately, it

not transform the substructure

converts them

in

above

linearly convergent Bisection

matrix to scalar

algorithm

Kron matrix, but instead

Lanczos

described

algorithm was examined

the result of a similar transformation

In

explored

and

(2.2)

which

above,

to condense the

use of the

the

the roots

sequence which relies on the

condensation.

into

function, has linear convergence

with a significant time savings.

necessary

The

is

The Bisection

The Newton Raphson iteration

is

which

containing functions

characteristic value problem

Sturm-based

transcendental

dynamic transcendental function in

SCOX

where

the Sturm sequence is reliable, but slow.

when

[8].

than are needed

more modes

Newtonian iteration. Unlike the Bisection Method, Newtonian iteration

with

successfully

based

for

solve

However,

is

matrix

equations

form. This is

Not only

can the

quantity, but the

which saves more

CPU time.

other words

The #

of steps

to solve

for

necessary
desired

the

The

<

(2.3)

order of the system

eigenvalues

Kron substructuring is just one,

of

dynamic

substructuring.

shapes and natural

substructuring

is

although somewhat

Dynamic substructuring is

frequencies, but

reduces

no assurance that an

uncommon, method

the number of

important

the results can

degrees

mode shape

of

not

15

an efficient method of

be difficult to defend.

freedom

has

in the

carried

been

into the

missed.

general

evaluating

of

mode

Since dynamic

eigenvalue

One way

category

analysis, there

assuring that

mode

shapes

have

been

not

distribution.

These

The

Ritz

vectors

use of special

Ritz

special

substructuring by providing
use of special

is to

missed

Ritz

vectors

a closer

is

first

function

the

displacement vector, f(s)

time

of

only.

set of special

improve the efficiency

and

accuracy

spatial
of

load

dynamic
The

upon the equation of motion

C*LT

U_ is

derived from the

that are

approximation of the mode shapes of the structure.

is based

M*U_"

where

vectors

in [13].

explored

vectors

Ritz

use

K*U_

f(s)*r(t)

represents the spatial

Ritz vectors, Xl is

(2.4)
distribution

of

loading

and

r(t) is a

added such that

Xl*Y

2.5)
T

Substituting for U

and

premultiplying

Xr

Xr in

by

T
*M*Xr*Y"

(2.6)

equation

can

be

for

special

Ritz

a static analysis of the

be found

using

be

*f(s)*r(t)

using the direct

(2.6)

eigenvalue

formulation

2.7)

the system.

using the

spatial

load distribution, f(s),

as

input

form

the

initial Ritz

f(s)

(2.8)

The remaining Ritz

vector.

be

vectors can

obtained

by

a recurrence relationship.

Xri+1

It has been
solution than

shown

[14],

using the

substructuring

that using this

M*Xri

form

Ritz

same number of exact mode shapes.

with special

Ritz

In

vectors was applied to problems

time needed to run one problem on

the same case using

an exact system of eigenfunctions.

In [16],

(2.9)

of special

freedom. The CPU

large

Xr

by

K*Xrl

where the output will

by

*Xl *M'*Xr]*P

P, that diagonialize

vector can

yields.

Xr *K*Xr*Y

solved

[Xr *K*Xr

to get the modal coordinates,

(2.4)

Xr *C*Xr*Y

For Rayleigh damping,

The first

equation

vector yields a more accurate

one experiment

containing

NASTRAN

was

methods of coordinate reduction are examined with regards

structures.

One

major

difficulty

with

over

1000 degrees

50% less

to the

[15], dynamic

than

it

substructuring

substructuring is that for large structures, there

16

was

of

for

of

are still

degrees

boundary

many

of

freedom

be

size of the eigenvalue problem to

to

define in

Modal

level,

eigensolution at the system

numbers of

interface degrees

easy to use, but

of the mass

One

can

lead to

recursive

This

substructures.

structure

freedom. Static

erroneous results

The Ritz

due

shape.

of

lack

vectors could

freedom

counter-productive

to the

the

boundary

method requires the user to

The Ritz

condensation of

[16].

degrees

of

for

be difficult

requires an extra

problems with

interface degrees

of a clear method

dynamic

et. al. used a

static

substructure

Theory

ones.

by

for

Prakash

of

large

freedom is

the condensation

holds

several

Indian Remote

Sensing

case studies on the

substructuring

preconditioning

accuracy

that static substructures are

level,

where

it is

easier to

can produce results with the

matrix

in

[17]

boundary value problems

Prabhar [16].

and

is broken down into

that was examined, was the

used on the component

general second order elliptic

on

large

was analyzed

investigated using

One factor

freedom,

formulation is based

substructuring,

substructuring,

[16]. However, if

Bramble

the usual methods used in

of

interface degrees
be

which could

procedure was

substructures versus
accurate

deflected

reduction of

possible approach, recursive

Satellite

the

level. This increases

matrix.

In performing

slave

of

the use pass

general problems.

which approximate

some models.

Most

solved.

for

coordinate reduction are not effective

supply the Ritz vectors,

into

which are carried

of

inherently

judge

desired

static

using

less

master and

accuracy.

to perform a substructured analysis of

defined

on two-dimensional

the second order elliptic operator equations associated

domains. The

with

the

Dirichlet

problem.

Lu

f in

where u

bounded domain D in R,

the

on the

boundary

of

(2. 10)

and

SUM( d/dxi(aij(d/dxj)))
vj

The

generalized

Dirichlet bilinear form

of this

formulation is

Ad(v,p)

(2. 1

SUM(aij*(dv/dxi)*(dp/dxj)*dx)
i.j

l.O

1
which

is defined for

Galerkin finite

all

element

and

p in the Solobev Space H(D).

formulation

and

an

approximate

Equation

(2.11) leads

solution.

To

get an algorithm which

better fits

parallel

processing, the space

17

H(D) is divided into

to

1)

suDspaces

(substructures), S^(D),

Ad(U,P)
In

(2.12), Uv is

j, (D)

in

which results

for

(f,P)

the substructure weak

an approximation of the

function

all

S,t,(D)

Uv is

u.

formulation
(2. 12)

composed of the

basis functions

of

follows.

as

Uv

SUM(ALPHAi*Xbi)

(2.13)

i=l

where

ALPHAi is the

coefficient of the respective

basis function Xbi. Equation (2.13)

can

be

rewritten as the expression

Ae*

ALPHA

which can

be

the matrix

Ae is usually "ill-conditioned",

solved

be

situation can

symmetric

for the

changed

(2.14)
ALPHA. The

coefficient vector

both

-l

[17] have discovered

in

by

sides

equation

ALPHA

inverse

This

definite

of a positive

(2.15) is

Bp*F

(2.15)

Bramble, Pasciak

much easier to solve.

preconditioning matricies, Bp,

several

the

that

-l

Bp*Ae*

stated

formulation is

Bp.

matrix

The formulation

this

which makes the problem cumbersome to solve.

by premultiplying

"preconditioning"

difficulty with

(2.15)

that make equation

Schatz

and

easier to

solve.

In [18],

variable order elements are used to make

easier to analyze.

high quality

Substructuring

meshes and can

however, like

mesh

intricate

large

be implemented

when

which

have

be computationally

model.

one

One

solution

CPU time

critical regions of the

using

model substructures

spent

to this problem

less

than perfect.

CPU

time.

forces

is

by

to

have

fine

exhibiting

fine

detailed

mesh

for

detailed
There

models with

are

problems,

Mesh compatibility is

rest of the

in

geometry

both

to use computationally expensive

higher-order

portions of a

relatively

boundary

sides to

with

However, it

with a

around the stress

domain

the portions of

remaining

each substructure

the substructures on

balance accuracy

mesh

geometry.

the

simpler

Unfortunately,

modeling the

intricate features

models that contain substructures with

stress concentrations and rather

often requires the analyst to

against the extra

large

small computers.

It is necessary

expensive to continue

It is desirable to

This

with

with small

the assessment of

relatively

make this technique

a given number of nodes, which

meshes.

on

is working

coarse mesh, which would conserve

have

fine technique for

parts and stress concentrations.

the substructures
would

compatibility that

particularly troublesome
small

is

large models,

have

must

similar

concentration,

superfine mesh.
elements

in the

model, linear elements at less sensitive spots, and variable-order elements in

18

the

intermediate

higher-order

regions.

This

is implemented

method

elements and then

deleting

by

nodes to produce

generating

all of the elements as

lower-order linear,

quadratic and

variable-order elements.

common theme

substructuring to

found in

model types

for

which

usually limited to solving linear,

substructuring to be

Linear finite
problem

applied to

element and

for the Poisson

and

many nonlinear,

substructuring

in

turn

problems that

non-positive

definite

in

involve

small

strains and

algorithms that will allow

problems.

be formulated

problems can

Substructuring is

use.

to solve the

Dirichlet

equation

where u

is the

definite

not

Nicolaides have discovered

^.(?uj=f

where v

the attempt to expand the use of

substructuring is currently

positive

In [19], Gunzburger

deformations.

literature is

the current

much of

The

gradient operator.

discretized into finite

on the

region

elements.

(2.16)

inD

The

boundary

of

D is divided into

subregions, Di

following governing matrix

can

(i=l,n),

which are

be developed for each

subregion.

Kss

Ksm

Us

Es

Kms

Kmm

Um

Fm

(2.17)

In

equation

(2.17), Us.
"master"

boundary

or

together to

form

contains the

degrees

of

system-wide

internal

freedom.

"slave"

These

equation

matrix

degrees

or

of

freedom

the

of

contains the

be

added

form

:B2

Fl
F2

Am

:Bm

Um

Fm

Cm

:Ao

Uo

Fo

:B1

A2

U_m

substructure matrix equations can

Ul
U2

Al

and

(2.18)

CI C2
Where
Ai is the Kss

matrix associated with substructure

Bi is

the

Ksm

matrix associated with substructure

Q is

the

Kms

matrix associated with substructure

Ao is the
Ui

is the

sum of the

vector

Kmm

matrices

containing the

for

slave

all substructures

degrees

of

19

freedom for

substructure

Usually

the coefficient matrices

definite,

so that one can solve

(Ai, Bi, Ci

for

and

Ao)

degrees

boundary

the

of equation
of

(2.18)

are symmetric and positive

freedom, Uo, by using

the

following

expression:

[Ao

-1

SUM(Ci*Ai*Bi)]*Uo

Fo

i=i

The displacements

(2. 19)

i=i

internal degrees

of the

-1

SUM(Ci*Ai*Fi)

of

freedom, Ui,

are computed

as

-l

Ui

The
in

is

above system

which the

decompose

matrices,

the

degree

Ai*(Fi

Bi*Uo)

acceptable

Ai,

of

if the

problem

are nonsingular.

freedom

vectors as

Ui

=Vi

Uo

Vo

l,...,n

is linear, but there

One

method of

are

dealing

(2.20)

many

with

nonlinear problems

is

this

to

orthogonally

follows:

Zi

Zo

l,...,n

(2.21)

l,...,n

(2.22)

so that

Vi*Zi

=0

Vo*Zo

Thus Vo, Zo, Vi

provide

Uo

and

and

Zi

are solved

with

low strains,

nonlinearities are
a turbine

proposed

solvable

rotating beams

theory

NASA

without

used to make a

blade for example,

stiffer than

inverting Ai.

The Vs

and

Zs

are added together to

Ui, i=l...n.

In [20], substructuring is

behavior

for

predicts

large strains, because

using linear

and

large

centrifugal

small

nonlinear

techniques.

space structures.

force due

[21]. In large

manned space

geometrically

space

to

high

Two

problem,

examples

which

containing

In rotating beams,

geometric

speed rotation makes the structure

structures, such as those that

of structural

nonlinear

helicopter rotor or

will

station, large displacements can occur without

deformations

has

components

will

be

used

far

in the

correspondingly

significantly

add

up

form

of

[22].

There is

a unified approach

for treating

all problems with small strains and some

20

geometric
are

nonlinearity [20]. First,

divided into

have

substructures which

The deformation

applicable.

junction degrees

modes of the substructure are

by

to a solution
upon

freedom to form

of

Three

the solution.

modes, the number

interfaces

in

which

common parameters to

of substructures and

are:

the number

linear elasticity theory is

so that

found using linear finite

The

quantity is

vary

nonlinearity is significant,

element

by compatibility constraints imposed

a whole component.

limiting process

using

which geometric

deformations,

small

analysis and the substructures are connected at the

at

in

the components

results can

be

made to converge

varied so that the results converge

deformation

the number of substructure

of

between

constraints

compatibility

substructures.

Two

large

and the other a

In both cases, the

space structure.

converged as the number of substructures were

In [23], the accuracy

communication

that

are then

for the

technology

usual method of

includes

damping

for

assembled

The inclusion
good,

and ground

actual

damping

of

frequencies

reliable models with

in

damping

conditions, like the different

deleting-off

diagonal

there are

limitations

developed for
been

One

are.

enough

it isn't
case

Hermes

data

frequencies agreeing

with

didn't

use

well

in the

was

found

was generated

of

a case

from

the

study

Hermes

for

a mathematical model

main component.

The

substructures

and the mode shapes

substructures.

test

be

the spacecraft which can

gravitational pulls and

case studies

known how

study involved the

collected on the

It

in

examined

using

data into

analyzed

eigenproblem

It is hard

The differences between

lack

of atmospheric

models.

Finally,

More information

to

orbit

damping

in

modal truncation

about substructure

[25].

communication

analytical solutions.

that

and

few documented

substructure synthesis,

for the

terms can contribute to errors.

be found in [24]

synthesis can

Since

of

displacements

the procedures creates certain problems.

space, adds complications to the incorporation

and

is

involves establishing

a structural model

modal

beam

orbit

damping effects

into

analysis

existing data. The data

in its

involving a rotating

approximations of the

for dynamic

each substructure and also

damping factors,

modal

establish

satellite

synthesizing

mathematically

analysis.

results against

one

increased.

of substructure synthesis

comparing substructuring

The

detailed in [20],

examples of geometric nonlinearities were

which

damping

factors

successful substructure synthesis

comparison of several

technology

Hermes

in

satellite against

spacecraft

that

is

found

and what

by
its

substructured solutions

inflight data [23]. There has

it is easy

that substructure synthesis worked

inflight data. Further, the

are

to compare

inflight data

admirably well,

to

with modal

results were consistant with previous models

damping.

The literature

presented above suggests that much of

21

the current research effort

has been

directed

towards two goals:

to analyze

large

advantage of
standard

during
large,

models and

developing
expanding

techniques that save

the use of

substructuring is its ability

finite

this research

focused

substructured models

on methods of

[8]

[18].

governing

expended to

implement the

composite materials
plastic

equations.

by

These

used to

linearize

be

Much

hardly

industry

of the effort

There does

dynamic substructuring

techniques

be

for

lack

that was comparable to the research presented

That

case

study

compared the

models

substructuring

using

an

exact

time using coordinate

Ritz vectors, to the

technique.

22

has been

benefit from

developing

substructure

cases, substructuring

intermediate

in

use of

enough to

large

by

be

used

sized models.

The

this paper was

time required

can

be

time required to analyze models with over

with special

modal

time required to solve

of case studies that could

small or

study

by

using

element techniques to analyze

certain

case

freedom

solve

[19]. The increased

has been invested into

seem to

large to

major

that significant effort

finite

only

CPU

CPU

nonlinear models

[20]. In

too

One

and new techniques to solve the

surprising

and the use of

be

CPU

methods reduced

applied to nonlinear models

nonlinear models.

to verify standard

also

using substructuring

most of the material uncovered

amount of

discretizations,

while

nonlinear problems.

created a class of nonlinear models that are

principles.

techniques that can

surprising that

substructuring in

the aerospace

deformation has

substructuring

use of

It is

not

reducing the

reduction, superior substructure and element

structure's

substructuring to

to analyze models that would

Thus, it is

element techniques.

CPU time

found in [15].

1,000 degrees

analyzing the

of

same

CHAPTER 3

Example Problems to Illustrate

The

following

in Chapter 1.

The first

one-dimensional
was chosen

illustrate

two example problems

for its

It

elements.

spring

is

example

The

simplicity.

will

the use of the

simple

be

static

used to

analysis of a

is

structure
of

outlined
of

comprised

substructuring

and

a two-dimensional truss problem which

on a nontrivial model.

their solutions can

substructuring technique

demonstrate the basics

second example

demonstrates substructuring techniques

elementary discrete elements,

Substructuring Techniques

Since both

be conveniently

models are comprised of

compared to the exact analytical

solutions.

The first
shown

problem

in Figure 3.1

involves

a.

The

the equilibrium analysis of a one-dimensional

model

Kl

two springs, one of stiffness

1 is fixed,

substructure no.

force F

applied to

Since both
same

for both

and

last node,

for

Fi, Fj Xi

and

Xj

Equations (3.1)

can

be

node

4,

Each

substructure contains

The first node,

K2.

on the second substructure

the

spring

element shown

-Ks

Xil

Ks

Xjl

0,

node

has

on

a positive

first step

substructure matrix equation

elements

Kl

is

the

is deduced

by

of the generation pass

and

The

K2.

standard matrix

in Figure 3.1 b is

(3.1)

Fj
corresponding displacements

and

used to assemble a

degrees

master

the

Fi

forces

are the

b. The

basic unit,

the same

of two springs with stiffnesses

freedom,

node

for the two spring

matrix equations

-Ks

of

the

In both cases, the basic

Ks

consisting

and the other of stiffness

involve

substructures.

constitutive equation

where

composed of two substructures.

structure

it.

substructures

assembling the

is

spring

local

Kl
of

and

matrix equation

K2. The

for

springs are

freedom for this

of end nodes

and

j.

the general substructure

joined

substructure are

at the slave

located

degree

at the end

nodes a and c.

Kl
Kl

Equation

(3.2)

freedom from

can then

the master

K2)

-K2

K2

-K2

be

rewritten

degrees

(Kl

of

K2)

-Kl

-K2

Xa
Xb
Xc

-Kl

(Kl

in

the

Fa
Fb
Fc

following form

(3.2)

which separates

the slave

degrees

freedom.

-Kl

-K2

Xb

Fb

Xa
Xc

Fa
Fc

(3.3)
Kl
0

0
K2

23

of

FIGURE 3.1
ID SPRING PROBLEM

Substructure 1

Substructure 2

^AA^AA^
;
V V v
j V V v
Kl
L-> K2
\

->

XI

A,
V

Kl

IaVaw
a^

->

X2

K2

->

X3

X4

Figure 3. 1 a
The Basic Model

Xi

Xj
Figure 3.1 b

Spring Element

MDOF

Nodes a,

SI. DOF: Node b

Figure 3.1 c
Basic Substructure

24

~~"-e

T7

Equation (3.3) is

thus

in

the standard

Kss

Ksm

Kms

Kmm

form

(3.4)

At this point, the

differing boundary
corresponding

0, fixed

generation passes

As

conditions.

its

Fm

for the two

a result, the

0, 1, 2

global coordinates

so that

Xm

have

following reduced matrix

XI

-K2

K2

-K2

condensed so

that

FI

conditions

it is in terms

X2

Xmi

and

Fsubi

and the equivalent

freedom

stiffness

found from

the

are the

of the master

matrix,

displacement

force vector,

following

be

has its first node,

substructure

to

switched to the
node

leaving

the

Ksubi,

(3.5)

F2

have been applied,

Ksubi*Xmi

where

and c will

due

separated

the central row and column removed,

boundary

have been

equation.

(K1+K2)

Once the

b,

coordinates a,

2, 3, 4. The first

or

matrix equation must

local

substructures

the substructure matrix equation

degrees

of

freedom

as

be

can

follows

Fsubi

(3.6)

vector of

the master

degrees

The corresponding

respectively.

of

freedom

of substructure

equivalent master

and the aforementioned equivalent

expressions

(3.5)

degree

force vector, Fsubi.

of

be

can

[2].
-1

Fsubi

=Fmi

(3.7

Kmsi*Kssi*Ksmi

Kmmi

Ksubi

ai

-1

Applying

the

formulae (3.7)

Ksubi

to the

Kmsi*Kssi*Fsi

first

substructure results

in the

K1*K2
K1+K2

Fsubi

(3.8

13..

following

values

for

Ki'

and

(3.S

Fi

a)

(-K2)*1*F1

F2_

b)
Kl

+K2

The

condensed matrix equation

for

the

first

substructure

K1*K2 *X2

Kl

is

thus.

l*o>

K2

25

The

second substructure

generation pass

is the

has

no

fixed degrees

partitioned

(Kl

form

K2)

of

-Kl

of

freedom. The logical starting

its governing

equation

-K2

X3

F3

0
K2

X2
X4

F2
F4

(3.3)

point

which can

be

for the

second

rewritten as.

(3.10)
Kl
0

-Kl

-K2

Using

the matrix equation condensation

(3.10) results in

the

formulae (3.7)

following condensed

to condense the

Ksub2

stiffness matrix,

and

governing

force

vector

matrix equation

Fsub2

-1

Ksub2

Kmrn2

Kms2*Kss2*Ksm2

1K1 0
0 K2

*[-Kl

-Kl

Kl

-K2

K1*K2 *
Kl +K2

(3.11a)
(3.11

-K2]

b)

+K2

(3.11c)

-1

Fsub2

Frn2

Kms2*Kss2*Fs2

-Kl

Fp

-K2

(3.11

1
K1+K2

(3.11 e)

[[0]

(3.11

d)

f)

Fp
so that the condensed substructure matrix equation

K1*K2
K1+K2

Now

second substructure

into

a global matrix equation and solved

displacements.

In this step,

matrix equations

(3.9)

and

which

(3.12)

is known

are combined

K1*K2
Kl +K2

Equation (3.13)

can

be

solved

Now that the

master

degree

-1

for the

Gaussian Elimination. The resulting

Xm

2
-1

X2
X4

of

Fp*

Fp
have been generated, they

into

degree

of

be

freedom

a simple matrix equation.

X2
X4

degree
Kl

(3.13)

Fp

master
of

degree

freedom

of

freedom vector, Xm. using

solution

is

K2

freedom displacements

a simple matter of

terms of the master

can

(3.14)

K1*K2

perform stress passes on each substructure

X3. This step is

(3.12)

as the use pass, the two condensed substructure

following

master

in

is

X2
X4

that the condensed substructure matrix equations

assembled

and

for the

in

order

using the

are

known,

all that needs to

to solve

for the

slave

expansion

26

formula [2]

degrees

be done is to

of

freedom XI

-1

Xsi
Since

the matrix equation

Kssi(Fsi

(3.5)

of

the

Ksmi*Xmi)

first

substructure

(3.15)
has been reduced, the

submatricies of this

equation are all single terms

-1

Kssl

Kl

Fs.

Xm.

Xs

displacement

Kl
The

X2

(-K2)*Fp*

Fp*(Kl + K2)
K1*K2

submatricies of the matrix equation

of

freedom is

Kl +K2
K1*K2

K2

b)

(3.16 c)

-K2

degrees

of the slave

(3.16

=F1=0

Ksml

so that the

(3.16 a)

+K2

(3.16 d)

computed as

(3.17)
Kl

(3.10) pertaining

to the second substructure are

-1

Kss2

Kl

Fs.
Ksm2
Xm

Inserting
Xs_

the quantities

X3

Kl
=

The spring
the

F3

[-Kl

in (3.18) into

(3.18 a)

+K2

X2
X4

-K2

+K2

Fp*(Kl + K2)*
K1*K2

(3.15)

1
2

(3.18d)

yields

l*Fp(Kl + K2)*
K1*K2

(3.19 a)

Fp*fKl + 2*K2)
K1*K2

problem was also solved

substructuring

b)

(3.18 c)

-K2]

equation

0-[-Kl

(3.18

(3.19 b)

using

method and this simple

a standard

base line

force/stiffness

method.

technique are compared

27

below

The

results

for

Force/Stiffness Method

Substructuring
XI

Ffi
Kl

X2

Kl

Fp*(Kl_+K2)

Fp*(Kl + K2)
K1*K2

K1*K2
X3

Fp*(Kl

+ 2*K2)
K1*K2

X4

+ K2)
K1*K2

solutions are

in

The

elements.

exact

Further,

structure.

as would

elements reflect the

both

since

is

to

have

problems

be

in Figure 3.2. This

model

half truss

Element (3) lies between

1 to

solely to

either one.

element placed

in

nodes

and

connection at the

by

truss

Therefore,

in the

and

is

100N force acting in

is

analyzed to

element

in both

top left corner,

truss elements.

figure 3.2 c,

In

the negative

find the deflections

[1],

the

\E

Lg

Co*Co
Co*Si
Co

Si

and

and

Lg

Co

was

divided into

by

Y direction

Co*Si
Si*Si
-Co*Si

-Si*Si

are the cross sectional

The

and

Fyj

are the

on the

is to

with respect

of

is

perfect

to the

-Co*Si

and

substructure

belong

cannot

is simply

half

mathematically
is

element

by

3m X 4m

(3),

which

supported with a pin

node

1. It is loaded

lower right hand corner,

node

2. The

freedom.

the matrix equation

Xti
Yti

for

Xtj
Ytj

modulus and

28

(3)

pattern

lower left corner,

Co*Si
Si*Si

forces

join

a truss

for

the

indivual

element, shown in

horizontal.

-Si*Si

Co*Co
Co*Si

displayed

elements with each

The basic

presented

-Co*Si

area, Young's

two.

develop

-Co*Co

displacements

that

performed

structure

are the respective sine and cosine of the angle of

Fxi, Fyi, Fxj,

3,

diagonal crossmember,

degrees

the

half

two

The division is

and a roller at the

of all

and

the two substructures, element

matrix equation

BETA

-Co*

-Co*Si

A, E

(3)

the two master nodes,

generation pass

following

oriented at an angle

of the

each of which contains two and a

substructures,

substructures.

4,

node

The first step in performing the

distribution considerations,

no mass

sides as elements and a

used

element meshes with the

basic force/deflection relationship

stiffness matrix of the element

has its four

rectangle which

comprised of two

one of the two substructures.

dividing every value

joins

is

finite

static analysis of the simple truss structure

2. Since it is located between

substructure

expected when one performs analyses with

expected.

involves the

second example problem

elements.

be

substructures reflect the stiffness of their

agreement of the solutions

The

2*Fp*fKl + K2)
K1*K2

agreement,

accuracy that the discrete

same

K1*K2

2*Fp*(Kl

The two
discrete

Fp*(Kl_2fK2)

length

Fxi
Fyi

(3.20)

Fxj
Fyj
of

the element, respectively.

inclination, BETA. Xti, Yti, Xtj, Ytj,

applied to the end nodes

and

j.

FIGURE 3.2
2D TRUSS PROBLEM

^^.

3m

100 N

Figure 3.2 a
The Basic Model

MDOF : 1

SI. DOF:2,4

Figure 3.2 b
Substructure Discretization

Figure 3.2 c
Arbitrary Truss Element

29

Since

the truss structure is rectangular, the elements on the opposite sides of the truss
have the

length

same

inclination BETA. Therefore,

and angle of

same cross sectional area

global matrix equation can

Elements m

be

(4)

and

0.002

and are made of the same material with

from

assembled

4.667 E7

0
0
0
0

0
-1

Elements (2)

and

C5)

4m

BETA

3.500 E7

Element

(3)

0
-1

and slave

(4)
Fx4
Fy4
Fx3
Fy3

(3)

0
-1

0
1

12

-9

-12

16

-12

-16

-9

-12

12

12

-16

9
12

used

steps can

degrees

be

of

saved

by

freedom

4.667 E7 N/m

3.500 E7 N/m

(2)

(5)

Xt2
Yt2
Xt3
Yt3

Xti
Yti
Xt4
Yt4

(2)

(5)

Fx2
Fy2

Fxi
Fyi
Fx4
Fy4

Fx3
Fy3
=

(3.21 b)

2.800 E7 N/m

be

12

will

7E10 psi, the

(3.21 a)

Xti
Yti
Xt3
Yt3

16

Fxi
Fyi
Fx3
Fy3

in both substructures, its

multiplier of the stiffness matrix will

A few

(1)
Fxi
Fyi
Fx2
Fy2

1.120 E6

element

Xt4
Yt4
Xt3
Yt3

5m (AE)/L
(0.002m*m)(7 E10 N/m*m)/3m
BETA
53.13 Si 4/5 Co
3/5

Since

(4)

Xti
Yti
xa
Yt2

(AE)/L
(0.002m*m)(7 E10 N/m*m)/3m
90
Si = 1
Co 0

0
0
0
0

(1)
0
0
0
0

0
1
0

0
0
0
0

-1

three element stiffness matrices.

3m (AE)/L (0.002m*m)(7 E10 N/m*m)/3m

BETA 0
Si
0
Co
1

have the

since all of the truss elements

be divided

assembling

the

by

first

two to

(3.21 c)

stiffness must

be halved.

Thus

the

become 0.056 E7.

substructure matrix equation so that the master

are separated.

(3.22)
(4.667

(0
10

0)

(0

0)
3.5)

(0

-3.5

(0

-4.667

Since

0) (0

-3.5

-4.667

(0+12X0.056)
12X0.056) (3.5 + 16X0.056)

9X0.056)

-9X0.056

-12X0.056

-12X0.056

-16X0.056

the pin constraint acts upon the two slave

Xt4

Yt4

-9X0.056

-12X0.056

Xti

-12X0.056

-16X0.056

Yti

Xt3

Yt3

(4.667

9X0.056) (0 + 12X0.056)
(0+12X0.056)
(0+16X0.056)

degrees

30

0
0

of

freedom, Xt4

and

Yt4,

equation

(3.22)

reduces to a master

degree

freedom

of

matrix equation of

Kmm*Xml

the

form

Fml

(3.23 a)

or

4.396
10

-0.672

-0.896

Assembling
the

following

Yti
Xt3
Yt3

-0.896

5.171
0.672

-0.672

0.672
0.896

Fyi
Fx3
Fy3

(3.23 b)

the second substructure matrix

equation, relating elements

(1), (2)

and

(3),

produces

form:

separated

(3.24)
(4.667

(0

0)

10

0) (0

(0

0)
3.5)

-4.667

0
(4.667

+ 9X0.056)
(0 + 12X0.056)
(0+12X0.056)
(0+16X0.056)

-9X0.056

-12X0.056

-3.5

-12X0.056

-16X0.056

There is only

one

fixed degree

the matrix equation

Xt2

-3.5

Yt2

-4.667

of

freedom, Xti, in

(0
(0

-9X0.056

-12X0.056

Xti

-12X0.056

-16X0.056

Yti

9X0.056) (0 + 12X0.056)
12X0.056) (3.5 + 16X0.056)

the second substructure.

Once

0
3.5

0
0

0
0
0

0
0

0.896

0
0

Xt2
Yt2

-3.5

that

is removed,

(3.25)

-0.672

-0.896

0.504
0.672

4.396

-0.672

that equation

Yt3

-100

10

Note

Xt3

becomes.

4.667
0

-3.5

0
-100

-0.896

(3.25) is in

the standard

Yti
Xt3
Yt3

0.672

0
0
0

form

Kss

Ksm

Xs

Fs

Kms

Kmm

Xm

Fm

(3.26)

The

submatricies

equations

Kss2 is

using the

formulae (3.7),

10

used

in

and

by inverting

(1/4.667)

10

(1/3.5)

Fm

first

the

be inverted

can

the

can

be

assembled

into

substructure

substructure example problem.

diagonal

Since

elements.

-7

-7
=

The inverted

into

matrix

diagonal matrix, it

-1

Kss2

Kss, Ksm, Kms, Kmm, Xm. Fs.

submatrix

(3.27)

0.2143
0

0
0.2857

(3.27)

and the other submatrices of matrix equation

the matrix equation reduction

formulae (3.7 a)

31

and

(3.7

b) in

order to

(3.25)

can

find the

be inserted

substructure

degree

master

freedom

of

force vector, Fsub2. for

Ksub2

10

stiffness matrix,

0.896

-0.672

-0.896

-0.672

0.504

0.672

-0.896

0.672

4.396

10

Thus,

0
0
0

7
10

0
0
0

7
10

they

performed
master

by

degree

0.672
0.896

*10

01

0.2857 1*10

(-0.672
(-0.896

0
0.2857

freedom

(3.28a)

(3.28 c)

-100

for the

0.896
0.672
0.896

second substructure reduces to

-0.672

-0.896

Yti

0.504
0.672

0.672
0.896

Xt3
Yt3

0
0

for

matrix

equation

(-0.672

(3.28 d)

-100

(3.23 b)

condensed substructure matrix equations

0.896)
0.672)
0.896)

0
-3.5

0
0

=:

-100

-3.5

freedom
+

of

(3.28 b)

(3.28 d) have been

and

degree

of

freedom displacements in

degrees

of

freedom,

the master

a simple combination of the stiffness matrices and

of

degree

-3.5

0.2143
0

-7

10

0.2143

-7

substructures share the same master

( 4.396
10

-0.896

are combined and solved

Since both

pass.

0.504
0.672

0
0

10

calculated,

the condensed matrix equation

Now that the two

-0.672

-0.672
-0.896

Fsub2

and the substructure master

the second substructure.

0.896
=

Ksub2,

force

the use

be

the combination can

vectors.

Thus,

the global

becomes
(-0.896
(0.672
( 0.896

-0.672)

(5.171+0.504)
( 0.672 + 0.672)

Yti
Xt3
Yt3

0.896)
0.672)
+ 0.896)
+

(0 + 0)
(0 + 0)
(0 100)

(3.29 a)

That is,
5.292
1.344
1.792

10

Solving equation

(3.29

b)

slave

degree

degrees

of

of

master

-1.792

5.675
1.344

1.344

degree

Yti
Xt3
Yt3

of

connection at node

of the

4,

first

simple matrix operation

in

1.792

results

b)

in

1.607

10

(3.30)

-9.643

are

Xt4

only be

and

for

expansion

32

the

Yt4, have been

performed

the stress pass

known, they

Since

the stress pass.

substructure,

which

(3.29

-100

freedom displacements

a stress pass can

0
0

-2.857
=

freedom displacements in

freedom

Yti
XS
Yt3'

using Gaussian Elimination

Xm

Now that the

-1.344

can

be

used to

displacements

partitioned out

substructure

2. The

formula (3.15) is

of

find the

the slave

due to the

stress pass

solved

pin

is

for XsL The

subscript

"i"

refers to the substructure on which the stress pass

appropriate quantities from

Xs2

Xt2
Yt2

10

(3.25), (3.27)

0.2143
0
0
0.2857

(3.30) for

and

is

being performed. Substituting the

the second substructure

10

0 0 0
0 0-3.5

-100

Xs2

The

in the

in

the

the X

element

direction,

Three

substructuring

the

had been

of node

no

element

(1)

method

is

digit

node

must

The

to

cannot pivot

be

loading therefore,
in

in length

zero

for

the

or

node

it

correspondance

for

pivot, so that node 2

2.

In

carried out to a greater number of significant

33

have been

digits.

body

will move

order to compare with

in

find Yt2

in

the

-1.25

methods suggests that the

dimensional truss

correspondance might

solid

X direction due to displacement

between the two

the two

by

can not change

theorem was applied to the same structure to

as accurate

problem.

consistent with results suggested

cannot change

X displacement

Castigliano's

2 is

internal force due

(1) is horizontal,

significant

dimensional spring

(3.32)

Element (1) has

Y direction. Since

m.

(3.31)

E-5

X displacement

substructure analysis,

E-5

1.607

0.0
-1.25

mechanics theory.

length. Since

-2.857

following displacements:

Xt2
Yt2

zero

yields

-9.643

which results

into (3.15)

even

problem as

better if

for

the one

the calculations

CHAPTER 4

The General Method

The basic substructuring techniques

different

complex models with

static

analysis, substructuring

factors. In

about certain
master

degrees

of

Although both

be implimented for

can

other

dynamic analysis, for example,

freedom

distribution throughout the

basic

in Chapter 3

presented

element types.

Substructuring

of

All forms

substructure.

of

analyses, as

equations,

passes.

which are

in terms

of the master

degrees

be

one must

The

the user

of

freedom.

substructure and

including the

find

The

degrees

slave

of

The first step in solving

fit the geometry

combination of
substructure

The

analysis

involve

developed in

are

divide the

In

the one

master

freedom along the

other analysis

required

in

accurately

degrees

of

substructure

is

for improved

define

the same

the generation

freedom

freedom

into
for

"global"

the master

solution

of that

for

each

substructure,

way that the

substructure

is

defined

develop

a condensed

in

being

it into

boundaries

consists of a

if

easier to generate substructures

of the plates.

to

the model and partition

the model that

governing

terms of the master

the

matrix equation that

degrees

of

is discretized into finite

freedom in

elements.

that

When

substructure, it is important to make certain that the element

the substructure.

model

freedom

are selected.

boundaries

to

be

Once

the substructure

static problem requires

selected as master

most

dynamic cases, in

accuracy.

See Figure 4. 1 for

types, like

of

of

boundaries

of a substructure

of the

to

such a

the generation pass, the substructure

enough to

elements, the

is

in Figure 4.2, it is

coincide with the

performing the discretization

is fine

model

For example, if

purpose of the generation pass

substructure.

of

to

flat plates, like

behavior

degree

degrees

terms of all of the

a substructured problem

of the model.

boundaries

can model the

mesh

in

careful

freedom.

It is usually best

substructures.

is

models the mass

condensed substructure matrix equations

stress passes take the master

a solution

involved

"local"

use pass assembles the

more

careful to choose enough

matrix equation that models the whole structure and solves this matrix equation

degrees

larger,

condensed substructure matrix

freedom,

of

long as

accurately

substructuring

procedure with the three types of solution passes.

applied to

of the example problems

that the condensed mass matrix

so

be

can

degrees

which extra master

more

of

is discretized into

only the degrees

freedom. There

are

freedom

are

degrees

information concerning

of

master

degrees

freedom.

Once the basic

then

substructured model

developed for

displacement vector,
reshuffled so

has been generated,

each substructure.

X^ is in any

order

that the master and slave

The

the

matrix equation

desired

by

degrees

of

the user.

freedom

34

finite

element matrix equation

may be

The

is

assembled so that the

matrix equation can

are separated.

of

then

be

It is usually easier,

FIGURE 4.1

MASTER DEGREES OF FREEDOM IN A SUBSTRUCTURE

it
*
4

i?

4>

Master Degrees of Freedom

Along the Substructure
Boundaries

Internal Master Degrees

of

Freedom

35

FIGURE 4.2
DIFFERENT SUBSTRUCTURE DISCRETIZATIONS FOR A PLATE MODEL

Model Discretized into Three Substructures

Model Discretized into Four Substructures

36

however,

to assemble the matrix equation so that the master

and slave

Once

separated.

the

finite

element matrix equation

complete the generation pass is to

then condense the

resulting

performed

partitioned matricies of the

Different

equation.

different formulae

in terms

matrix equation

Matrix condensation is

in

governing

the master

of

conjunction with

all

that the

degrees

formulae

of

user

to perform the matrix condensation.

The governing

has

to

are

do to

conditions and

which use sections of the

Thus,

equations.

freedom

freedom.

equation to produce the matricies used

different governing

analyses use

developed,

of

force, displacement and other boundary

the

apply

is

degrees

in the

condensed

each analysis will require

equation

for

the static case

is
Kss

Ksm

Xs.

Es
=

Kms
The

Kmm

stiffness matrix and

load

Xm

(4.1)

Fm

vector can

be

condensed

by

using the

following formulae from [2]

-1

Ksub

Fsub

=Fm

Kmm

Kms*Kss*Ksm

(4.2 a)

-1

resulting in the

a single substructure.

Once

Fsub

The

of

freedom displacements

use pass

elements.

is like

use of standard

finite

of the model without

certain quantities of

In the

use

interest,

matrix equation which

model contains standard

"global"

the

equation,

just

as

extra

they

stress,

are applied

Using

without

which all of the master

incorporates

standard elements also

for the

be

The

user

find

a stress pass.

degrees

Once

applied to the

the generation passes.

37

lets the

substructures are assembled

use pass.

its

able to change a small portion

of

The

main

into

freedom in the

their matrix equations will also

conditions can

substructures as

combination with substructures.

performing

is developed in the

in

in

terms of all of the master

finite elements,

boundary

pass, in

flexibility of being

matrix equations

is in

matrix equation that

has been created,

element analysis that

a generation pass.

pass, the condensed

use

for every substructure, the

found.

finite

such as

performed

be input to the

elements gives the user the

running

has been

regular elements are used

"global"

If the

are

a standard

In many problems,

(4.3)

a generation pass

condensed substructure matrix equations can

degree

(4.2 b)

condensed equation

Ksub*Xm

for

Kms*Kss*Fs

large

structure.

be incorporated into

"global"

the

degrees

matrix equation
of

limitation

freedom

of

the

of this method of

boundary

applying

is that they

conditions

nonsubstructure degrees of

freedom.

can

only be

After the

degrees

applied to master

boundary

of

freedom

or

have been applied,

conditions

to

the

"global"

matrix equation can

element

degrees

degrees

of

find

Compared

vector created

no major

any

in the

components

The local

is

in terms

displacements

a post

of

master

have

finds

all of

degree

master
of

of

is found,

is implemented.

formula that

This formula

substructure matrix
analyses.

For

can

be

used

is completed, the

degree

master

or

to

the

should

conditions,

of a solution.

vector

be in

master

slave

the static case, this

matrix condensation

formula is

given

is

for

of

of

this

There

nor

is

are

there

The first step in

the

vector and sort

its

the given substructure.

the same order as

degree

degrees

formulae,

in [2],

standard

stress pass

a given substructure.

it

appears

in

freedom displacement

freedom

of the substructure

in the

will consist of sections of the matrices used

and, like the

any

freedom displacement

freedom displacement

local

find the

of

boundary

freedom displacement

Once

and

passes, the stress passes are simple. The

accuracy

of

freedom

freedom.

of

geometry

freedom displacement vector

the original partitioned matrix equation.

vector

degree

of

are related to the standard element

displacements in

the

involving model
master

degrees

use pass

degrees

a major effect on the

"local"

degree

Once the

processing step that takes the

to take the
a

master

stresses, that

the slave

"global"

into

its

of the generation and use

use pass and

would

of

such as

for.

solved

considerations

factor that

stress pass

be

complexity

like

modeling

other

can also

the

to the

stress passes are

solved

freedom. Quantities,

of

freedom

performed to

be

formula

varies

original

for different

as

-1

Xs
where

Fs, Kss

and

assemble all of

to solve
order of

Ksm

Kss*{Fs

derived from

the master and slave

for variables

increasing

After the final

are

of

interest,

stress pass

equation

degrees

of

(4.1).

The last step in

freedom into

such as stresses.

node numbers

(4.4)

Ksm*Xm)

the stress pass

and use

it

organized

in

displacement vector, 2L

This displacement

vector should

be

is to

(XI, Yl, Zl, ROTX1, ROTY1, ROTZ1, X2, Y2,..., ROTZN).

is performed,

the analysis

processing.

38

is

completed and the model

is ready for

post

CHAPTER 5

Substructuring Using
is

Substructuring
be

that must

a multipass method

The

solved.

each of the substructures

substructure

it in terms

passes take the master

the

degrees

The

solved

so that

in the

nodes and

These

freedom

of

it

the master

freedom

be

can

The

steps are performed to

of the master

use pass

in

solution

is to

Fn{Um} is

The

form

function

generation pass

the

basic

of

define

the substructure

and

locate

is just

form

then solved

all

in

of the master

degree

similar

of

to the

freedom.

equation

is

then

so that the equation could

freedom displacement

input file for

See Appendix 1 for list

large

matrix equation

terms of the master

in

degree

of

be

a combination of substructures and standard

finite

in

analyses with minor changes

design

optimization

occurring between

material properties and

portions of the structure that change can

that don't change are modeled as

be

for

(5.1)

the whole structure.

produced

the

The

in

This is why

scheme, where a

In

in

Since only the

use

substructures

models, the use pass contains

The ability to

analysis.

the

assembled

freedom displacements. In effect, the

elements.

39

boundary

input file.

modeled as elements

substructures.

(Um).

of some of the commands unique to

substructured

each

vector

a standard analysis run and

which the elements are substructures.

In many

useful

of

terms of

(5.1)

are sometimes referred to as superelements.

be

stress

include

{0}

generating nodes, elements,

one

a standard analysis

use pass can

to

degrees

This

use pass combines the condensed substructure matrix equations

is

it

geometry in

the master

substructure matrix equation.

generation pass and a generalized sample generation pass

pass

use pass and expand

freedom displacements, Um,

input file is very

conditions as a standard problem.

equation

The

of that structure.

form

the

a vector

generation passes to

the whole

models

create the condensed substructure matrix equation,

boundary conditions

degree

shares the same methods of

The

equation that

freedom

of

found in the

generation pass must

Fn{Um}

where

degrees

pass assembles the condensed

use

governing

for

matrix equations

combined with the other condensed substructure matrix equations and

elements, implement the

input to the

is divided into. The

governing

equation

of each substructure.

use pass.

in terms

produce condensed

a condensed

purpose of the generation pass

like (5.1),
solved

of

degree

into

of

that reduces the size of the governing matrix

passes

that the model

equations

governing

structure and solves

generation

ANSYS

mix elements

model undergoes

an optimization

into

the

several

scheme, the

the use pass, while the portions

elements

in the

use pass will

be

updated, the generation pass,

As has been

or

passes, need only be run once, saving much CPU time.

and uses the same analysis type and option

pass commands
pass

input file,

finds the

which

substructure.
not

is

also

only for

degree

Since

of

degree

of

have been included in

should

is

pass

input commands,

noted that element and material

displacements

the use pass and

stresses, for each

has been

performed after the main problem

solved and

geometry definition, its input file does

of model creation or

which

be

the special use

a generalized use

and related quantities such as

The

are to

has its

stress pass

associated commands to control the stress pass expansions.

for

of

freedom displacements found in

freedom displacements

resemblance to that of a standard analysis run.

each substructure

Some

as a standard analysis.

It

file,

to a standard analysis

nonsubstructure elements.

the stress pass

involve any form

identical

almost

They

in Appendix 1.

stress pass takes the master

slave

setting

have been listed in Appendix 1.

properties are needed

The

input file is

stated earlier, the use pass

be

One

calculated.

own

module,

stress pass must

A list

have any

/STRESS,

be

and

performed on

of some of the

input file

and a generalized example of a stress pass

not

does

are

basic

stress

included in

Appendix 1.

There is
Each

passes.

so that

they

they

can

position which

they

will

be

be

for

moved

be

written

always writes

ANSYS

pass generates and uses several

display the

Files

with

substructuring

and the post

whole structure

in ANSYS have

is determined

its

output

files to

overwritten

other reasons.

accessed at the

by

by

in

than

files.

require some manipulation

have to be

substructures

use and stress

modified so that

use passes must

be

combined

in

to

its

a mode shape plot.

the number

in

form, FILE#,
that

the output

files

and can

filename. A

the same positions, and

Routine Begin level

files

of these

processing files from the

a name of the

These files

just performing generation,

Many

The individual

accessed properly.

be linked together,

can

order to

more to

are

usually

of

ANSYS

moved

and

the

according

are not moved after each pass,

There

are also

using the copy

has

accessed

generation pass or a stress pass

if these files

of the next pass.

be

files that have

to

command which can

be

only

form

/COPY,NFROM,NTO,NORF,NORT

where

NFROM

NORF

and

rewind

key prevents

and

NORT

starting

point.

such as

file

The

NTO

are the numbers of the

are the no rewind

the system

no rewind

from

key

is

keys for

the

file to be

files

copied

"rewinding"

an optional

the specified

function that

stacking.

40

copied

from

file,

can

from

and to.

so that

be

and

useful

If

to,

set to one, the no

it can be
in

respectively.

accessed at

its

some manipulations

FIGURE 5.1
FILE HANDLING
FOR MULTIPLE SUBSTRUCTURES

(file 13)

(ssi

SSNF8)

rV

STRESS PASS 1

STRESS PASS 4

STRESS PASS 2

AUX

2^
41

Another important
ANSYS is

a program with

in the

substructures

freedom coupling

In degree

limited intelligence.
Two

use pass.

It

must

forms

common

substructure connectivity.

be explicitly told how to

connect the

degree

of substructure connection are

of

and node renumbering.

freedom coupling,

freedom together in the

of

as

of

for substructuring in ANSYS involves

consideration

coupling

equation

This coupling

use pass.

is

boundary degrees

used to connect the

equation

is implemented

by

the

CP

command

follows:

CP, ID#, DOF, NODE1, NODE2

DOF is the type

Degree

of

In

This has

which

each

CP coupling

coupling

the

The

likelihood

form

node number as

of an

of

being the

input

degrees

The

other

freedom,

of

It takes many CP coupling

equations

is

must

be

begin level

shown

performed

Thus, by using
up

different

a node

In

files into files

freedom in

which

the

the use

that since

may translate into

equations to connect the

not

only tedious, but it

with

the

by

two modules.

also

of

boundary

which

node

and

FILE3 in the

stress pass of each

AUX5 modules,

renumbering is

renumbers the substructure

freedom displacement file, FTLE13,

before the

and

The

nodes on the

boundary degrees of

can mesh with the other substructures

from FTLE2

the same

module to renumber the

corresponding

AUX4

have

substructures that

renumbering

consult

in the substructuring flow chart, Figure 5.1,

processing file, FILE 12,

is involved in

of

renumbering, doesn't have these

node

connectivity,

that it

degree

AUX4

distinct

degrees

drawback is

the user can establish a connection without the extra

the

master

that
of

two

error occurring.

same node.

a substructure so

information concerning

substructure.

boundary

degrees

many CP coupling

node numbers match those

renumbering

post

so

substructure

renumbers the master

As is

one pair of

boundary

NODE2.

coupled to

boundary

the nuisance of using so many CP coupling equations.

file, FILE8, for

its

only

NODE1 is

the

sides of the

the number of

nodes on one substructure to match

and

both

recognizes two master nodes on

performed at the routine

so that

freedom coupling,

time needed to solve the use pass.

Using

connecting substructure,

AUX5

of

doubling

equation couples

drawbacks. ANSYS

freedom

degree

which

be the direct method, but it has

seem to

equations per node pair.

other

boundary

CPU

the

ROTZ) for

...,

the substructures on

the effect of

substructures together.

increases

freedom (UX,

order to use

increases

pass,

six

of

be input for

must

coupling.

degree

freedom coupling may

disadvantages.
freedom

of

in the

created

in

stress pass.

library

use pass.

the use pass,

AUX5

node

For

more

substructure.

Appendix 1.

each stress pass produces

contains all of the mode shape

deflections for

its

own

a given

order to postprocess the whole structure, the user must assemble the substructure

that contain information

for

the whole structure

42

using the AUX1

module.

This

subprogram, AUX1,
substructure post

processing files into

by AUX1.

Thus if the

iterations, AUX1
not too

difficult

AUX1

module can

In

have

will

operations

from

read

and avoid

found

CPU times

convenient

reason

The
for

an

input file. As

has

to structure the

way

This

stress pass

file copying

stress pass

file, FILE 13,

pass.

The

All

of the

files 2

node

input files

processing files. The AUX1

module

is

for

FTLE8

input files

to structure the

command

file

used

input

following

about the use of the

statement

is

handling

and node

to run the generation,

around the

individual

a generation pass or a use

this, any information

a result of

input file for

on an

meaningless.

has its

It is

easier to

modify

input file.

own

by providing

input files is

the substructures

The investigator has

some natural points at which to

contains

file

out of the

input before

also used

renumbering

only the

for

commands

generation pass

remaining

and

in Figures 5.2

shown

handling
way

5.3.
for

needed

input files, from

for

The first
the

first

the second to

the previous generation pass.

and perform

the use pass

last

of the

and

AUX4

node

renumbering

the commands needed to perform the next

input file

which moves

generation pass

are similar

commands

as well as move

files 2

re-renumbering

AUX5

of the

input file

substructure out of

place and perform

in format

before any

to the generation and use pass

the next stress pass as opposed to

node

first

More information

files 2

and

commands to

cleaning up

to move the master

and

3 back into

boundary

The

after the previous one.

The

It

degree

place.

of

freedom displacement

There

are also commands to

nodes, if needed for the first substructure

contains commands

the way.

before

input files. The

the stress pass.

renumbering, or re-renumbering, are performed

second stress pass

the

is

input file has copy

AUX5

FILE 12, for

into

to

processing file

each post

issued.

and node

to prepare

the whole

steps or

separate

decided

FINISH

input file

pass

structure

AUX4

start the use pass are

perform an

post

in

set

easier to use, the

made the research easier

the previous generation pass are

intent is

was

/INPUT, 27

commands needed to move

The

combines the

involving either N load

understood.

contain the commands needed to perform

and performs the

AUX1

needed to perform each pass.

generation

generation pass.

first

why it

will not allow the

substructure generation pass.

last,

certain passes when each pass

running

data

is interested in N data sets,

are other advantages to this modular approach.

substructure

the

one

only

have been included into the

that this approach

collect the

able to store

its basic logic is

generation or use pass after the

There

begin level.

files that have postprocessing data for

substructuring in ANSYS

One

passes, is that ANSYS

be

routine

be found in Appendix 1.

use and stress passes.

pass to

user

a series of

to combine and output

to use, once

order to render

renumbering

from the

accessed

The investigator has been

structure.
output

be

must

to move the post

processing file,

also contains commands to move

node re -renumbering of the

43

boundary nodes

for the

files 2

second

and

FIGURE 5.2
INPUT FILES FOR SUBSTRUCTURING FN ANSYS
GENERATION & USE PASSES

1st GENERATION PASS INPUT FILE

/PREP7

! 1st GENERATION PASS

FINISH
/EXE
/INPUT,27

2nd GENERATION PASS INPUT FILE

/COPY,2,32
WAY
/COPY,3,33
/AUX4
FINISH
/PREP7

! MOVING 1ST GENERATION

PASS'

FILES 2 & 3 OUT OF

! NODE RENUMBERING FOR 1st SS

! 2nd GENERATION PASS

FINISH
/EXE

/INPUT,27
LAST GENERATION PASS INPUT FILE

/COPY,2,36
OUT OF
/COPY,3,37
/AUX4

! MOVING 2nd TO LAST GENERATION

PASS'

FILES 2 & 3

!WAY
! NODE RENUMBERING FOR 2nd TO LAST SS

FINISH
/PREP7

! LAST GENERATION PASS

FINISH
/EXE
/INPUT,27

USE PASS INPUT FILE

/COPY,2,22
OF WAY
/COPY,3,23
/AUX4

! MOVING LAST GENERATION

! NODE RENUMBERING FOR LAST SS

FINISH
/PREP7

PASS'

! USE PASS

FINISH
/EXE
/INPUT,27

44

FILES 2 & 3 OUT

FIGURE 5.3
INPUT FILES FOR SUBSTRUCTURING FN ANSYS
STRESS PASSES

1st STRESS PASS INPUT FTT

,F.

/COPY,13,8
/COPY,32,2
/COPY,33,3
/AUX5

! MOVE MDOF DISP FILE 13 TO FELE8

! MOVE FILES 2 & 3 BACK INTO PLACE FOR 1st SS
! NODE RERENUMBERTNG FOR 1st SS

FINISH

/STRES,!

! 1ST STRESS PASS

END
FINISH

2nd STRESS PASS INPUT FTI

/COPY, 12,20
/COPY,34,2
/COPY,35,3
/AUX5

,E

! MOVE 1st SS FILE 12 OUT OF WAY

! MOVE FILES 2 & 3 BACK INTO PLACE FOR 2nd SS

! NODE RENUMBERING FOR 2nd SS

FINISH

/STRES, 2

! 2nd STRESS PASS

END
FINISH
LAST STRESS PASS INPUT FILE

/COPY,36,2

! MOVE 2nd TO LAST SS FILE12 OUT OF WAY

! MOVE FILES 2 & 3 BACK INTO PLACE FOR LAST SS

/COPY,37,3
/AUX5

! NODE RENUMBERING FOR LAST SS

FINISH
/STRESS,LAST #

! LAST STRESS PASS

/COPY, 12,23

END
FINISH

/COPY, 12,24
/AUX1

! MOVE LAST SS FTLE12 OUT OF WAY

! ASSEMBLE SS FILE12s INTO FILE12s FOR A FULL
! STRUCTURE

FINISH

45

substructure. All stress pass

basic format. The last

handling
the

operations

There

of

in

used

used

in

be

master

freedom

TOTAL

and

analyses of

a substructure.

freedom

setting

degrees

the

substructures

in

that are neither

proportion

along

To implement

file

value

in

values

are

in

degrees

of

place those that aren't

useful

included in

substructure

that

this method, the user

The

substructure and the

constrained

by the D

free degrees

first

takes

by taking

boundary degrees
FDOFi

next

of

in

the number of

the total

freedom

TDOFi

step is to find the

structure that exist

This brings up

master

inventory

in

in

Another

of

degree

case of

degrees

of

of

dynamic

freedom among

the

freedom

each substructure.

of each substructure.

degrees

degrees

number of

number of

BDOFi

percentage

of

the master

those

of

of

The degrees

freedom (TDOF) in

which are

freedom

find

along the

(CDOF)

degrees

of

freedom for

of

of

that

boundary

which

are

freedom

and

degrees

the

free degrees

is found for

that substructure and

46

a given

subtracting
of

off the

freedom.

(5.2)

of all

percentage

free degrees

freedom for

CDOFi

fraction

This

the number of

free degrees

and the number of constrained

each substructure.

free degrees

stating

the problem

the substructures.

degrees

freedom

of

substructure, one can

The

each substructure.

be found

the use pass as a means of

listed.

Once this information is known for each

substructure can

which

of the model more

The solution, in the

nor constrained

number of constrained

command are also

freedom (FDOF) in

distribution

boundary degrees of freedom (BDOF)

number of

in dynamic analysis, in

accurately defines

the standard structure.

boundaries

placed

already

the substructure generation passes so that the

use pass

to the number of

selects the master

states the total number of master

optimally divided between

in the

directly

of

of each substructure are counted as the total number of

substructure.

taking

substructure

degrees

analysis, is a numerical method that divides the nonboundary

The

in the

master

are needed to make the mass

freedom

of

configuration that exists

number of

file

FILE 12 files linked in

for selecting

useful

be especially

command can

TOTAL

freedom

of

all of the

command

configuration of the structure.

is setting the TOTAL

of the

the same

that after the pre-pass

contained

The M

ANSYS to

commands are also

problem

freedom

has

in

have

substructure generation pass to select the

It is

a substructure.

a substructure and allows

of

of

in the

used

boundaries. TOTAL

The TOTAL

degrees

degree

nonboundary

be

substructure

These M

the master

dynamic

to

cornmancL

extra master
realistic.

freedom (MDOF) in

freedom

of

freedom to be
the

moved and

file handling, is

about

are two commands that can

freedom along

by

the others

stress passes

list

degrees

degrees

it has its FTLE12 file

Information

module.

last

the second to the

input file differs from

stress pass

and the stress pass,

AUX1

master

input files, between

of

freedom in the

a given substructure

by

dividing it by the sum of the

free degrees

freedom in

of

all substructures.

%i

Finally,
must

to

first decide

the

TOTAL

nonboundary
number of

the values

for the TOTAL

This

for the

values

master

degrees

is the

value

total number of

boundaries

freedom for

of

For every
number of

all substructures and

substructure

nonboundary

technique could

be

nonboundary degrees
nonboundary
nonboundary
of

master
master

on the completed

freedom (TMDOF) to be

be

master

used

of

TMDOF

in

used

for

dividing
=

of

degrees

freedom for the last

TOTLi

the use pass

finds

the

of

statement.

To

the total number of

by

subtracting

degrees

of

adding

by

the

freedom.

by

counting

the number of

them at the

boundary degrees

two.

(5.5)

TOTAL

freedom

last

by

of

would add

is found

the percentage

freedom

but in

by

multiplying the total

for that

substructure and

of that substructure.

(5.6)

order to

products of

up, the

are added.

freedom

value

%i*TNBMDOF

substructure,

the total

(5.4)

be found

can

by

that number

BDOFi

freedom

of

total master

the

TBDOF

freedom

boundary degrees
=

TOTAL

in

used

SUM(BDOF)/2

degrees

freedom

degrees

of

structure, or

but the final one,

that value to the number of

degrees

of

one to

degrees

boundary

TOTLi

This

degrees

freedom (TNBMDOF). This is found

TBDOF

adding

the generation and use passes, the user

boundary degrees of freedom (TBDOF) from the

substructure
of

in

substructure generation passes, one

TNBMDOF

The

statements

on the total number of master

problem.

substructuring

find

find

(5.3)

FDOFi/SUM(FDOF)

The

and this value

sum

is

is

be

certain that the number of

the percentages and the total

subtracted

added

to the

from

the number of

number of

boundary

substructure.

BDOFi

+ (TNBMDOF

47

SUM(%j*TNBMDOF)

(5.7)

FIGURE 5.4
A METHOD FOR FINDING TOTAL VALUES
EXAMPLE PROBLEM

The following example problem as illustrates the method for finding TOTAL values. It is
based upon a simple three substructure problem involving a model that will be used in the case
studies.

THE BASICS
TDOF1
TDOF2
TDOF3

252
336
216

BDOFI
BDOF2
BDOF3

108-

252
336
216

SUM(FDOFj)

84

48
=

132

132/528
228/528

168/528

12
24
0
+

=
=

228

%i

PERCENTAGES

%1
%2
%3

CDOF1
CDOF2
CDOF3

108
84
48

FDOFi=TDOFi

FREE DOF:

FDOF1
FDOF2
FDOF3

TMDOF

12
24

200

CDOFi

BDOFi

132
228
168
+

168

528

FDOFi/SUM(FDOFj)

0.250
0.432
0.318

TOTAL VALUES

TBDOF

0.5*SUM(BDOFj)

TNBMDOF

TMDOF

TOTLI
TOTL2
TOTL3

BDOFI

BDOF2

BDOF3

TOTLu

TMDOF

0.5*(108

TBDOF

200

84

120

48)

120

80

108 + 0.250*80
84 + 0.432*80
%2*TNBMDOF
(TNBMDOFSUM(%j*TNBMDOF)

+ %l*TNBMDOF

200

48

108 + 20 = 128
118
84 + 34
48 +(80 20
=

34)

74

CHAPTER 6
CASE STUDIES

DESCRIPTION OF PROBT

The

series of case studies that

modal analysis of

6. 1. The

structure was

beam

elements couldn't

be

Fiat Xl/9

drive

components.

Therefore,
modeled

justify

for

design

problem that

involves the

an automotive space structure shown

frame for

A few early

the use of

on a simple

in Figure

a mid-engined sports car that was to

studies showed that models composed of

substructuring

be

discrete

as well as models made of continuum

The

plate elements.

structure

has

two tubular

section, for attachment to the

and rear of the center

frame sections,

each

suspension system and

train.

The

center section

is

to

be

manufactured

5052

aluminum

good

workability, very good resistance to

structure

10 to 20
not

based

the structure was changed to one with a monocoque center section which

using

front

connected to the

are

a tubular space

originally

upon

elements.

follow

the monocoque center section

based

could

FM

be

noted

for its frequent

has been discretized into

elements.

has

The

version of

ANSYS that

substructure combinations.

The

as

research

into one, two,

is

has

shown

will

of

be

investigator to

The four

in transportation

corrosion and

applications.

the model

does

used to perform

spend more

It

aluminum.

was chosen

high fatigue strength,

(STTF 63). Each

geometry

ASTM

[26].

see

of the center section

limits inherent

the case studies.

The simplicity

time performing

rear comers of the center section

The

panel contains

the

conform to

for its

may

to the

of

the

research on the variation of

have

all of their

degrees

of

in Figure 6.1.

cumulated

three and

0.2 in. ASTM 5052

coarseness of the mesh and the simple

also allowed the

freedom fixed,

use

panels of

a coarse mesh of plate elements

However, simplicity

realistic.

university
model

is

from

four

in

a series of

substructures.

four
Each

case studies

case

49

in

which

the model was

study involves the

discretized

comparison of several

FIGURE 6.1
THE BASIC STRUCTURE

Monocoque Center Section

Monocoque Center Section

with

Suspension Attachment Sections

50

FIGURE 6.2
MODE SHAPE PLOTS

First Mode

Second Mode

Third Mode

Fourth Mode

51

FIGURE 6.3
BASELINE ANALYSIS

In

order to ensure the

reliability

of

the

baseline solution,

the

baseline

analysis was performed

four

times.

NATURAL FREQUENCIES
The

frequencies
listed below.

following

data tables

as

were the same

for

each of the

four

analyses.

They will

be input into the

wl =7.384

w2

9.734

w3

15.55

w4

18.43

CPU TIMES
The CPU times

required

which was within

the values.

The

baseline analysis varied. This variation,

limits for a university version ofANSYS, was corrected by averaging
values were input into the data tables.

to solve and post process each

the normal

averaged

1st Analysis

2nd Analysis

3rd Analysis

4th Analysis

Tsoln

168.45

158.33

160.61

165.95

Tpost

3.58

3.44

3.22

3.56

Averaging

TsolnAvg

0.25*(168.45

TpostAvg

158.33

0.25*(3.58

52

160.61
+

3.44

165.95)
3.22

163.33

3.56)

3.45

different

master

degree

comparison of master
of

the model.

from the

CPU

solution

In the

dubious.
degree

of

from
times

distorting

discretization

the

for the four

discovered

that the

would

first four

plots of the

discretization. The

the solution efficiency

first four natural frequencies

solution with respect to a

first four

mode shapes of the

a standard analysis to solve a model

An initial

results.

concern was that the

Since CPU

the run-to-run variations could make the results of

only

analyses were

is

upon

baseline

distort the base line

performing

averaged, as is

sensitive to
natural

the

master

frequencies

an analysis

indication

requires one solution.

baseline results, the base line

analysis, modal analysis

of the

as the substructured models.

configurations provides an

analysis

be based

by performing

substructured case studies,

freedom

impact The baseline

frequencies

See Figure 6.2 for

time.

given substructure

configurations will

solution was computed

efficiency

for the

a measure of the agreement of the

in the CPU time

run-to-run variation

master

freedom

substructured solutions with the

with the same element

seem

of

configurations

Solution efficiency is

The base line

factor in

freedom

degree

minimum amount of

model.

of

In

of this

order

degree

freedom

53

one analysis

several

different

variation, and lessens its

four

variation

times and the

CPU

of

dynamic

configuration.

It

was also

As

provided an accurate picture of the

solution with a minimum of values that might confuse the reader.

a major

form

in Figure 6.3.
of

any

is

to prevent the run time

model was analyzed

shown

for

time

accuracy

of the

CASE STUDY #1

For the first

are:

case study, the model

the nosepiece; see

along

between the
command

This

is

model

is discritized into four

of the center section

a combination

placed

FOUR SUBSTRUCTURES

floor

and the

Figure 6.4. This discretization

the substructure
substructures

using

used to place

has been

solved

total number of master

backplate,

the

proportionality

these extra master

degrees

master

master

of

degrees

degree

of

side panels and

degrees

of

substructures

of

freedom

freedom be
divided

are

in Chapter 5. The TOTAL

freedom for

freedom ranging from

of

master

method outlined

degrees

using five different

each of the two

120

requires that

boundaries. The remaining

These

substructures.

optimal mass

freedom

distribution.

configurations with

140 (20 nonboundary)

to

the

220 (100

nonboundary).

The

results of each of these

listed for
inherent
the total

These

each master

degree

five
of

to that configuration; the

CPU time

results can

frequencies
postprocess

freedom

configuration:

CPU time

be

compared against the

by

in Table 6.1. The

the

first four

which

a nonsubstructure analysis and also the

baseline

analysis.

54

in

mode shapes

baseline solution,

following

first four frequencies,

needed to solve each pass

needed to solve and plot the

computed

the

analysis runs are shown

items

wl

are

to w4,

that configuration; and

for that

has

CPU time

the

configuration.

first four

needed

natural

to solve and

FIGURE 6.4
SUBSTRUCTURE DISCRETIZATION FOR

THE

FOUR-SUBSTRUCTURE CASE STUDY

The Monocoque Center Section Discretized into Four Substructures

55

FIGURE 6.5
TOTAL VALUES FOR
THE FOUR-SUBSTRUCTURE CASE STUDY
BASICS:
TDOFl=

6*10*6

360

BDOFI

18*6=

108

CDOF1

4*6

24

4*5*6

120

BDOF2=

7*6=

42

CDOF2

2*6=

12

TDOF3=

4*5*6

120

BDOF3=

7*6=

42

CDOF3

2*6=

12

TDOF4=

6*6*6

216

BDOF4=

8*6=

48

CDOF4

TDOF2

TBDOF

0.5*(BDOF1

BDOF2

BDOF3

FDOFi

FREE DOF:
24

228

FDOF2= 120-

42-12=

66
66

FDOF1

360

108

FDOF3

120-

42-12=

FDOF4

216-

48-

SUM(FDOF)

228

%1=

228/528

0.432

%2=

66/528

0.125

%3=

66/528

0.125

TDOFi

BDOF4)

BDOFi

0=168

66

66

168

%i

PERCENTAGES:

528

FDOFi

/SUM(FDOF)

56

0.5*(108

CDOFi

42

42

48)

120

TOTAL VALUES:

TDOF

TOTi

BDOFi

TOT4

BDOF4

TOTu

TDOF

%i*(TDOF

(TDOF

108
42
TOT3= 42
TOT4
48
TOTu = 140
=

0.432*(140

TOT2=

0.125*(140-120)=

108
42

0.125*(140-120)=

42

(140

108
42
42
48
160

0.432*(160

108
TOT2= 42
TOT3= 42
TOT4 = 48

TOT2=
TOT3=

TOT4

TOTu

120

120)

(8

3))

116

+ 3=

45
45
6 54

+ 3=

48

108 + 17 = 125
42+ 5= 47
0.125*(160- 120)= 42+ 5= 47
(160 120 (17 + 5 + 5))
48 + 13

120)

+ 0.125*(160- 120)=
+
+

61

66

=180

TOTI

TOTu

TDOF

SUM(%i*(TDOF

=160

TOTI

TDOF

TBDOF

140

TOTI

TDOF

TBDOF)

0.432*(180

120)

108

26

+ 0.125*(180- 120)=

42+

8=

+ 0.125*(180- 120)=

42+

8=

(180

0.432*(200

120

(26

8))

48

134

50
50
18

=180

200

TOTI

108
TOT2= 42
TOT3= 42
TOT4
48
TOTu
200
=

142
52
42
120)= 42 + 10= 52
120 (34 + 10 + 10)) = 48

120)

108

+ 0.125*(200-

0.125*(200

0.432*(220

34

+ 10=

+ 0.125*(200- 120)=

26

TDOF

220

TOTI

TOT2=

TOT3=

TOT4
TOTu

108
42
42
48
220

120)

+ 0.125*(220- 120)=
+ 0.125*(220- 120)=
+

(220

120

(43

13

108
42
42
+

43

+ 13=

151
55

+ 13=

13))

57

48

55
+ 31

79

74

TBDOF)))

FIGURE 6.6
FILE HANDLING FOR THE-FOUR SUBSTRUCTURE CASE STUDY

GEN PASS 1

GEN PASS 2

GEN PASS 4

GEN PASS 3

(F8_

F8

SS4)

(FILE13 )
FILE8

(SSI

SS4

F8

STRESS PASS 1

STRESS PASS 4

STRESS PASS 3

STRESS PASS 2

(*)
I

31,32,33,34

58

F8)

TABLE 6.1
DATA FOR THE

# OF MDOF

140

FOUR-

160

SUB STRUCTURE CASE STUDY

180

200

220

GEN

SSI

18.52

19.51

18.84

19.32

19.98

GEN

SS2

5.73

5.68

5.10

5.17

5.55

GEN

SS3

5.83

5.99

5.59

5.27

6.18

GEN

SS4

10.65

10.85

11.47

11.82

12.23

40.73

42.03

41.00

41.58

43.94

31.89

45.60

57.02

71.64

96.69

TOTLGENPS

PASS

USE

BASELINE

vl

7.401

7.386

7.385

7.384

7.384

7.384

v2

9.739

9.737

9.735

9.735

9.735

9.734

v3

15.79

15.60

15.56

15.55

15.55

15.55

v4

18.87

18.74

18.49

18.44

18.44

18.43

STRS PS SSI

11.51

11.18

11.40

10.91

11.13

STRS PS SS2

4.08

3.98

4.08

4.24

3.81

STRS PS SS3

4.37

3.59

3.75

3.90

3.68

STRS PS SS4

7.81

7.24

7.41

7.20

7.06

TOTL STRS PS

27.77

25.99

26.64

26.25

25.68

AUX1 FL CMB

5.56

4.66

4.62

4.74

4.90

163.33

POST PROC

5.86

5.51

5.30

5.08

5.44

3.45

134.58

149.29

176.65

166.78

TOTAL

111.81

123.79

59

The

most

striking

CPU time

when

case studies

aspect of this case

using substructuring

was

configuration could

than the

CPU time

time

by

saved

CPU time.

savings of

needed to perform the extra

file

CPU time

handling and calculation

in

analyzed

degree
baseline

these

freedom

of

solution and

was possible

because the

freedom

was greater

fewer degrees

the use pass with

solving

being

the model

close agreement with the

This

for saving

a realistic potential

substructured model, a master

be found in Table 6. 1 that had

yet saved a significant amount of

amount of

Even though

methods.

for

small

relatively

study is that it demonstrates

of

needed

by

the generation

and stress passes.

As the

number of master
converged to the

solution

substructuring is
of

freedom that

baseline

solution.

solved

of

freedom

in the

use pass.

set of equations.
amount of

CPU

configuration,

freedom,

see

significant
reduction

mass

in

This is exactly

Thus, increasing

degrees

freedom

produced a solution

find

of

master

dealing with

freedom

increased.

were

increased

is

time needed to solve the problem.

to solution efficiency,
model comprised of

200

baseline

in CPU time.

60

required more

degrees

The

is the

Increasing

of

increased

solution and saves

the

as the

CPU time

that would

of

be

than a smaller

freedom increased the

degrees

the

baseline

comparable to

degree

containing 200

master

into

the number of master

optimum master

case

degrees

the mass

carried

of the matrix equation

the order

the number of master

correspondence with the

which

the substructured solution also

Solving this larger matrix equation

Table 6.1. The

it is less ideal in

While

anticipated.

the substructured model resemble that of the

time needed to

a model

be

what should

the substructured

relationship between the

the number of master

distribution

of

increased,

the model was

the same stiffness

distribution in

with respect

digit

solution.

in

used

a nonsubstructured model,

degrees

used

freedom

Increasing

The CPU

total number of master

degrees

in

structure.

the mass

of

producing

would exist

This improved

solution.

baseline

capable of

distribution in the
use pass made

degrees

freedom

of

master

total

degrees

freedom has

17.49 seconds,

of

3 to 4

10.5 %

CASE STUDY #7.

The

next case

THREE SUBSTRUCTURES

study involves

combination of the two side panels and the

study; see Figure 6.7.

freedom (120)
freedom

are

as the

divided among the

5. The TOTAL

command

is

used

used to

degree

of master

degrees

The
master

of

freedom

degrees

of

of

freedom

varies

in the first

case study.

Each

as the model

in

model

the

first

from 140 (20 nonboundary)

to

degree

the

configuration:

needed to solve and plot the

compared against the

substructure

postprocess

first four

baseline

first four

each pass

mode shapes

same

basic

which

natural

for that

61

has

This

of

case

degrees

of

degrees

of

in Chapter

freedom in

using five different

the same total number

number of total master

lists

the

following items

frequencies

configuration.

the

master

nonboundary).

the natural

model and

that solution.

solved

floor,

in the first

master

degrees

in that configuration;

solution which shows

analysis of the

220 (100

the

method outlined

has been

case study.

in Table 6.2,

freedom

The remaining

of these configurations

results of each analysis are shown

of

boundary

place these extra master

distribution. This

configuration; the CPU time needed to solve

non-

and the same nosepiece used

using the proportionality

optimally

configurations.

freedom

three substructures:

uses the same number of

substructures

order to assure a near-optimum mass

master

backpiece,

This discretization

discretization

discretized into

a model

each

associated with that

and the total

These

frequencies

CPU

for

CPU time

results can
obtained

time needed to

find

be

by

and

FIGURE 6.7
SUBSTRUCTURE DISCRETIZATION FOR
THE THREE-SUBSTRUCTURE CASE STUDY

The Monocoque Center Section Discretized Into Three Substructures

62

FIGURE 6.8
TOTAL VALUES FOR
THE THREE-SUBSTRUCTURE CASE STUDY
BASICS:
TDOF1

6*7*6

252

BDOFI

108

CDOF1

2*6

12

TDOF2

14*4*6

336

BDOF2=14*6=

84

CDOF2

4*6

24

TDOF3

6*6*6

216

BDOF3

8*6=

48

CDOF3

TBDOF

0.5*(BDOF1

FDOFi

FREE DOF:
=252-

FDOF2

336-

84-24

FDOF3

216-

48

SUM(FDOF)

PERCENTAGES:

132

%i

TDOFi

BDOFi

132/528

0.250

%2

228/528

0.432

BDOF3)

228

0=

168
+

168

0.5*(108

CDOFi

132

FDOFi

228

%1

18*6

BDOF2

108- 12=

FDOFI

528

/SUM(FDOF)

63

84

48)

120

TOTAL VALUES:

TDOF

TOTi

BDOFI

TOT3

BDOF3

TOTu

TDOF

108
84
48
140

+ 0.250*(140-

108
84
48
160

+ 0.250*060-

(160

108
84
48
180

0.250*(180

108
84
48
200

0.250*(200

0.432*(20O-120)

(200

0.250*(220

TOT2=

108
84

0.432*(220-120)

TOT3
TOTu

48
220

(220

TOT2=
TOT3=

TOTu

TDOF=

TOT2=
TOT3=

TOTu

BDOF)

TBDOF

120)
120)

+ 0.432*(140-

(140

-120 -(5

=108

84

+ 8))=

48

5
8
7

=113
=

92
55

+ 0.432*060-

120

120)
120)

(10

+ 10=118
101
84+17
48+13= 61

=108
=

17))

=180

TOTI

TOT2=
TOT3=
TOTu=

TOT2=

TOT3
TOTu

+ 0.432*080-

+ (180-

120

120)
120)

-(15

+ 26))=

108
84
48

108
84
48

+
+

15 = 123
26=110
19= 67

200

TOTI

TDOF

(TDOF

160

TOTI

TDOF

%I*(TDOF

140

TOTI

TDOF

120

120)
(20

34))

20=128
34=118
26= 74

220

TOTI

120

(25

133
108 + 25
127
84 + 43
48 + 32= 80
=

120)

43))

64

SUM(%i*(TDOF

TBDOF)))

FIGURE 6.9
FILE HANDLING FOR THE

THREE-

SUB STRUCTURE CASE STUDY

GEN PASS 3

GEN PASS 2

GEN PASS 1

(F8 SS2)

(F8_ SSI

F8

SS3)

FILE8

USE

PASS

(FILE13)
FILE8

(SS2 F8)

SSI F8

STRESS PASS

SS3

STRESS PASS 3

STRESS PASS 2

31,32,33,34

65

F8)

TABLE 6.2
DATA FOR THE

* OF MDOF

140

THREE-

160

SUB STRUCTURE CASE STUDY

180

200

220

GEN

SSI

10.95

10.71

10.60

10.60

11.29

GEN

SS2

15.90

15.79

16.35

16.36

18.06

GEN

SS3

12.28

12.50

12.12

12.25

13.45

TOTLGENPS

39.13

39.00

39.07

39.21

42.80

USE

33.21

47 48

55.28

69.19

101.89

PASS

BASELINE

vl

7.402

7.390

7.387

7.384

7.384

7.384

v2

9.742

9.737

9.736

9.735

9.735

9.734

v3

15.92

15.65

15.57

15.56

15.55

15.55

v4

18.67

18.47

18.45

18.44

18.44

18.43

STRS PS SSI

8.68

8.11

7.42

6.95

7.62

STRS PS SS2

11.65

10.98

10.03

9.76

10.80

STRS PS SS3

12.22

12.40

10.54

10.28

11.20

TOTL STRS PS

32.55

31.49

27.99

26.99

29.62

163.33

5.66

5.20

5.07

5.21

5.05

3.45

POST PROC

TOTAL

110.55

123.17

127.41

66

140.60

179.36

166.78

Like the

four-substructure

previous

case

time-saving potential of substructuring.

26.18 CPU
freedom
needed

used

to

baseline

seconds over the

find

has

study, the above

simulation also

In this case, choosing 200

solution.

Increasing

master

degree

However,

it had in the

litde

freedom case, the

this savings

four-substructure

model.

closer to the

study.

is less
The

previous case study.

three-

than

natural

frequencies

The first three

difference between
and the

of

freedom

saves

degrees

It is difficult to decide

of the

natural

natural

5.82 %

frequencies from
baseline

frequencies

converge upon

the four- substructure case

solution than those

of the

from the

four-substructure

fourth

natural

three-

frequency

67

of the

baseline

CPU

220

time.

corresponding

study

tend to

be

substructure case

cases tend to

tends to

as the number of master

die frequencies

time

the exception of the

of the time needed to analyze the

frequencies disappears

frequencies

With

CPU

of

whether or not

substructure model saves a small amount of

correspondence, while the correspondence of the

increases

degrees

the same effect upon the solution correspondence and the amount of

a solution as

of

CPU

the number of master

the three-substructure model produces superior solution efficiency.

master

the

demonstrates

be

degrees

solution.

have better
worse.

of

The

freedom

CASE STUDY Jtt

For the third

6.10. One
Figure 6.7

TWO SUBSTRUCTURES

case

is the

substructure

The

substructures

study, the

model was

same

floor

from the

second case

boundary master degrees of freedom.

the two substructures

in the

used

using the

used to place the extra master

Since

this case

two,

previous

nonboundary).

study

a sixth case

The

other

As in the
of

contains

has been
cases

cases

falls into

The
items for

by

the range of

can

have

master

The total

degree

of

each master

degree

of

118

CPU

freedom

model and the

first four

CPU

baseline

natural

time needed to

first four

the other two

solution used

in Chapter 5

and

freedom
of

than the

freedom (10

of

freedom

of the

later

natural

lists

the

following

frequencies

produced

which

that configuration; and the total

case studies.

These

results

The baseline

a nonsubstructure analysis of the

and post process that solution.

68

108

nonboundary).

first four
in

of

degrees

degrees

in the first two

by

requires

divided among

mode shapes of that configuration.

frequencies deduced

find

are

in

configurations that resemble

Table 6.3,

time needed to solve each pass

compared against the same

solution shows the

the

degrees

master

number of master

configuration:

freedom

shown

distribution.

138 (30 nonboundary) to 218 (1 10

freedom

of

in Figure

studies, the TOTAL command was

master

added with a total of

time needed to solve and plot the

be

boundary

#2. It is

that was shown

optimal mass

results of each of these six analyses are shown on

that configuration; the

CPU

degrees

other two case

12 fewer

shown

a combination of

proportionality

freedom for

Study

is

This discretization

welded together.
master

in Case

is

section that

The remaining

those of the two previous case studies.

five

top

same method of

degrees

five

study

previous two case studies.

two substructures as

module that was used

is

other substructure

divided into

basic

FIGURE 6.10
SUBSTRUCTURE DISCRETIZATION FOR
THE TWO-SUBSTRUCTURE CASE STUDY

Substructures
The Monocoque Center Section Discretized into Two

69

FIGURE6.il
TOTAL VALUES FOR

THE TWO-SUBSTRUCTURE CASE STUDY

BASICS:
TDOF1

6*7*6

252

BDOFI

18*6

108

CD0F1=2*6=12

TDOF2

100*6

600

BDOF2

1 8*6

108

CDOF2

TBDOF

0.5*(BDOF1

FDOFi

FREE DOF:
FDOFI

=252

FDOF2

SUM(FDOF)

108

PERCENTAGES:

%1

132/600

TDOFi

0.5*(108

BDOFi

132

%i

24

468

468
=

600

FDOFi /

108)

CDOFi

12=132

108

600

BDOF2)

SUM(FDOF)

0.22

70

108

4*6

24

TOTAL VALUES:

TDOF

TOTi

BDOFi

TOT2

BDOF2

TOTu

TDOF

%i*(TDOF
(TDOF

BDOF)

TBDOF

=118

TOTI
TOT2
TOTu

TDOF

108
108

+ 0.22*018-

=108 +

108)
3)

108

0.22*038
(138 108

108)
7)

+ (118-

108

111
115

=118

=138

TOTI
TOT2
TOTu

TDOF

108
108

23

108

=108

11
39

108

124

=108

16
54

162

=115
=

131

+
+

0.22*058 108)
(158 108-11)

1 19
147

178
=

0.22*(178
(178 108

108)
16)

108)
20)

108)
108-25)

108
108
178

108
108
198

+
+

0.22*098
(198 108

108

0.22*(218

(218

198

TOTI
TOT2
TOTu

TDOF

108

=108

=158

TOTI
TOT2
TOTu

TDOF

=138

TOTI = 108
TOT2=108
TOTu= 158
TDOF

108

20

=108

70

108

+
+

25
85

=108

128
178

218

TOTI
TOT2

TOTu

108
218

71

133
193

% 1 *(TDOF

TBDOF))

FIGURE 6. 12
FILE HANDLING FOR THE

TWO-

SUB STRUCTURE CASE STUDY

GEN PASS 1

GEN PASS 2

31,32,33,34

72

TABLE 6.3
DATA FOR THE

TWO-

SUB STRUCTURE CASE STUDY

#OF MDOF

118

138

158

178

198

218

GEN

SSI

10.50

10.93

10.80

10.53

11.37

10.81

GEN

SS2

29.18

33.80

36.41

37.95

43.16

42.37

TOTL GEN PS

39.68

44.73

47.21

48.48

54.53

53.18

USE PASS

21.99

34.68

41.94

57.31

79.69

90.28

BASELINE

vl

7.432

7.398

7.389

7.386

7.384

7.384

7.384

v2

9.755

9.740

9.737

9.737

9.736

9.736

9.734

v3

16.48

15.76

15.61

15.56

15.56

15.55

15.55

v4

18.97

18.50

18.45

18.44

18.44

18.44

18.43

STRS PS SSI

7.22

8.30

7.37

7.16

8.14

7.61

STRSPSSS2

20.01

21.52

19.92

19.38

20.27

20.52

TOTL STRS PS

27.23

29.82

27.29

26.54

28.41

28.13

163.33

5.38

5.63

5.25

5.44

5.62

5.42

3.45

POST PROC.

TOTAL

94.28

114.86

121.69

137.77

73

168.25

177.01

166.78

While the

general results were similar to the results of the two previous case studies, there were a

few observations that

were surprising.

conserving CPU time

effective at

for this loss in

solution

degrees

It is

also

pattern with

all

four

of

simulations

from

as the ones

efficiency is that the

CPU times resulting from ANSYS

master

The

having

interesting to note

that results

fit into

cause the correlation of the

find

pertaining

a pattern

baseline

in

study

not seem to

One

for 10

extra

be

as

major reason

experienced an

to the two-substructure model

solutions to

If the

model.

present case

two substructures would

have

the

It

yielded solutions that were

analyses of the

latter

between

case

study

efficiency

originally

increase in

nonboundary

increase
be

or

fit into

time needed to

correlated with the number of substructures

Instead,

the other two case studies

in

also tended to require more

expected

pattern, the cases

would also require

the results of the

baseline

terms of

CPU time

is discretized into does

74

not

expected that the results of

decrease. The CPU

study had fit into the

of the model.

did

the number of substructures would

least correlation, but

the number of substructures which a model

effect upon the solution

was

increasing

to analyze than the simulations of the other case studies.

The

case

optimal positions

which

produce a substructured solution was expected to

involving

in this

the solutions of the other two case studies.

in the

study did

freedom.

case studies would

contained

this case

the previous two case studies.

models

to

in

not

less CPU

latter

case

time

study

correspondence.

to yield solutions.

necessarily have

Thus,
direct

CASE STUDY #4

For the last

ONE SUBSTRUCTURE

case study, the whole center section was modeled as one substructure with no

nonsubstructure elements

input into in

Since

the use pass.

this single substructure

to other substructures, or to nonsubstructure elements,

it is

freedom along

other nonoptimum

the substructure

command was used to place

Since

there are no

convenient

analyzed

boundary

every

arbitrarily

master

using

freedom intervals

total of seven

different
using

using 140

of the seven master

degree

that configuration; the

of

CPU

freedom

degrees

degrees

natural

frequencies

standard

20
of

compared against the

CPU

its

to place master

freedom

degrees

of

freedom

the

of

The

model was

configurations.

freedom. The

by

20

lists

master

The

analyses

degree

the

natural

following

items for

frequencies

of

The

baseline solution,

results

each

produced

that configuration; and the total

that case.

was

was analyzed.

first four
in

study, it

case

configurations.

of

which

degrees

most effective position.

freedom in this

master

connect

The TOTAL

location.

freedom increased

of

for

necessary

freedom

time needed to solve each pass

be

and the

of

in Table 6.4,

configurations:

at

degree

degrees

time needed to solve and plot the mode shapes

substructured models can

of

a total of

master

results of the analysis runs are shown

freedom

master

their total number of master

until a case

of

degree

to expand the range of the master

followed had

The

degree

any

placed master

configuration selection started with one

that

or at

not

doesn't

by

CPU

pertaining to the

which shows

the

first four

time needed to perform a non-substructure analysis of the same

model.

75

FIGURE 6.13
SUBSTRUCTURE DISCRETIZATION FOR
THE ONE-SUBSTRUCTURE CASE STUDY

Substructure
The Monocoque Center Section Discretized into One

76

FIGURE 6. 14
FILE HANDLING FOR THE

ONE-

GEN

SUB STRUCTURE CASE STUDY

PASS

( FILE8 )

USE

PASS

(fileh)

( FILE8 )

STRESS PASS

(FILE12)

77

TABLE 6.4
DATA FOR THE

* OF MDOF

ONE-

SUB STRUCTURE CASE STUDY

20

40

60

80

100

120

140

36.97

43.81

46.86

50.23

54.72

59.85

66.30

3.56

4.54

6.63

10.43

15.02

20.38

30.81

vl

7.395

7.389

7.385

7.384

7.384

7.384

7.384

7.384

v2

9.743

9.739

9.736

9.735

9.735

9.735

9.735

9.734

v3

15.80

15.61

15.56

15.56

15.55

15.55

15.55

v4

19.00

18.54

18.44

18.44

18.44

18.44

18.44

18.43

STRS. PASS

22.84

22.97

22.78

24.24

22.82

23.80

25.90

163.33

POST PROC.

4.43

4.42

4.50

4.24

4.06

4.16

4.34

3.45

67.80

75.74

80.77

89.14

96.62

GEN. PASS

USE

PASS

TOTAL

78

108.19 127.35

BASELINE

15.55

166.78

The

results of the

first three

study

number of master

digit correspondence

significant

freedom. This

master

case

The level

case studies.

surprisingly low

of

latter

degrees

case

of

which

freedom

than

degrees
the

high level

correspondence

freedom, every

small number of master

Even the

solutions.

degrees
mode

having

Since
master

with

a mere

the model

degrees

120

of

140

or

of master

degrees

allowing the
above case

case

of

extra

master

from

freedom,

of

time,

which

for

agreement

degrees

of

use more

master

the

first

to

degrees
fewer

less CPU
for this

apparent reason

degrees

distribution.

optimal mass

CPU

level

This

of

Thus,

baseline

containing 20

master

three modes, with the

fourth

comprised

of a

that can

single

model

also

reduces

requires

is

79

as

is

larger

substructure

model.

there are no

baseline

provide

Since

number

by

checking that

in the

52.7 CPU

correlation.

superior

the substructure

boundary

number

the

in Table 6.4. The

a savings of a minimum of

minimum

from

to the time saved

shown

produce a similar

fewer

The larger models,

what allows the simulations

the amount of

the

time.

when compared

multiple-substructure

This significantly

in CPU

efficiency

in

results

an accurate solution with

to the placement of a much

of solution

freedom

in

time than similar size models

savings

common with other substructures,

single-substructure

savings

is insignificant

previous case studies

to use than a

large

is probably due

by ANSYS,

model

easier

selected or renumbered.

The

digit

uses

also requires

correspondence, associated with the model

produce the superior

Not only does the

have boundaries in

The

master

can produce an excellent correlation with

will also produce a

degrees

freedom

seconds over run times

efficiency, but it is

for

placed

is high for

baseline

the

It

100

boundary

since there are no

a single substructure will result

containing

This

100

involving

case

three case studies.

be

can

solution

converges upon a

study

0.05 % deviation from

other

use pass to solve a smaller model.

involving

case

the other three simulations.

freedom

two significant

freedom, it

study to

baseline

results of the

3.09 % deviation.

master

other case studies.

of

baseline

very different from the

for the

solution

is that,

freedom

of

degrees

worst

freedom, had

of

one case

degree

master

from

from the

but

baseline

baseline

case

time to analyze than all

of

freedom. The

of

a maximum of

any

and

of correspondence with the

with

has

interesting,

are quite

solution

does

not

master nodes to

must

be

of solution

be

performed.

passes

and

reduces the amount of

significantly

time and the ease

choice

for many

The

with which

greatest

be larger
The

mesh.

can

The

be implemented,

one-

for

for

large

one-

interesting

the

of

aspect of the

number of master

In the first

freedom increases

there might

be

a similar

ANSYS

latter

case

the

CPU

degrees

study has the largest

to solve the

freedom.

freedom

of

matrix

condensation.

by

used

degree

it

that

The only

Since

is

a restriction on

the master

degree

in

can

investigation

freedom

be

its

number of master

the

with

the

prevented the

to examine the effect of

needed

increasing

to solve the

generation

the number of master

Results

also

Unfortunately,
in

indicated

the

that

degrees

degrees

of

increase in

generation and stress pass

existence questionable.

of

Since

freedom, it has

the

final

provided

freedom increased, the CPU

to the extra time needed to place the extra master

in the

generation pass

that are effected

separation of master and slave

components seem

freedom

with

model.

utilized

that normal fluctuations

or make

to clash

Interference

used.

this

model.

the number of master

major steps

of

of

enough

needed to solve the generation and stress

freedom in the

of

is probably due

these

in

CPU time

on the

generation pass rose

time

being

was obvious that

increase

As the

configuration are

degree

influenced

variation

this question.

This increase in CPU

of

study is

freedom

was small enough

case

degrees

of

master

be large

time needed to solve the use pass.

times would often cancel out the

needed

case

study, it

CPU

time

160

ANSYS

relationship between the CPU time

CPU time, if it did occur,

insight into

for

degrees

passes and the number of master

some

substructure model

substructure model could

the university version of

the generation pass required

and stress passes.

degrees

be feasible.

than the wavefront requirements of an unsubstructured model with a similar element

wavefront restrictions of

increasing

for saving CPU

the generation pass of a one-substructure model

the wavefront restrictions of the particular version of

Another

the great potential

make the one-substructure model an obvious

usefulness of the

wavefront requirements

generation pass

solution of

Both

handling required.

which multiple substructures would not

limitation to the

substructure size.

can

in

problems

it

file

of

the master

freedom

and the

unlikely to have their CPU times significantly

configuration,

80

degrees

by

it is

likely

that the

increase is significantly

influenced by

the placement of extra master

a similar range of variation

display
master

explain

used

solve

of

why the

in the

freedom

stress

generation

of

number of master

in CPU times for the

a change

degrees

in the

degrees

degree

generation pass.

were not placed

by

the

which

had

freedom configurations, did

not

freedom. The
of

other case

This is because

TOTAL

command.

formula,

studies,

a major

This

fraction

of their

observation would also

(1.13),

similar

to the one

pass, does not experience a noticeable increase in the CPU time

needed to

pass,

which relies on a matrix

it.

81

equation

CHAPTER 7

CONCLUSIONS
One

major conclusion of this research

potential

least

for conserving CPU time. The

natural

results corresponded to approximations

to compute the

baseline

for less CPU time

solutions resulted

in

four

of three to

natural

study) to 70.16

digits,

sec

(for the

The

one

is the

results of each of the

four

is

the solution

model

increases,

increases. This larger

amount of

freedom
solution

What

CPU

freedom

also

of

mass

freedom in the

equation

naturally

to

however,

was the extent of

upon solution efficiency.

number of substructures that

on

the

However,

model was

emerge was evidence that the number of master

a

degrees

of

the

accuracy

condensed mass matrix of the

number of master
equation

to the above
a certain

degree

increase in

degrees

in

factors,

this

of

freedom

increasing

the master

of

the

the use pass also

influence

influence, particularly

the solution

efficiency

expected that the number of substructures that the model was

boundaries has

The

time needed to arrive at a

CPU time to solve, thus

have

dynamic

the

degree

of

the

upon

of a model.

expected,

the greatest effect

to the

Due

freedom).

Increasing the

The improvement in

governing

are

sent to the use pass.

a consequence of the

As the

requires more

CPU

case

they

of a

the number of master

amount of

expected.

the size of the system-wide

boundary master degree of freedom configuration,

originally

be

model.

efficiency
is

of

the

models.

of the solution of that model.

distribution is

configuration of a model was expected

was not

increasing

that

distribution in the

time needed to analyze the model.

efficiency

configuration that

increases the

these trends are to

This improved

degrees

number of master

only 648 degrees

degree

demonstrate

for

(for the four-substructure

the solution

of

optimal

tolerance

within a

considerable, because

contains

upon

These

acceptable

impact

improves the accuracy

of

be

greatest

freedom

Both

be

(it

small

a result of superior mass

substructured model.

in the

of

sec

accurate

provided

for solving very large

case studies

degrees

solution.

corresponding
of

master

a particular model

number of master

from 17.49

less

investigated

solution.

considers to

study

strong

were at

over that

baseline frequencies

investigator

in CPU time

the four case studies

find the baseline

of the

most effective

considered to

factor that has the

substructured model

freedom in

is relatively

each case

Many

solution.

one-substructure case study) are

associated with a model that

Substructuring is typically

of

which range

for

a significant savings

that agree with the

which the

These savings,

application considered.

Three

computed

baseline

the

than was needed to

frequencies

significant

frequencies

possessing

solution.

optimal solutions

dynamic substructuring demonstrates

that

frequencies in

rough approximations of the

needed

is

definite

distinct

of the model.

discretized into

relating

solution

freedom

along the

of

placed

of a given model.

of

It

would

not emerge.

efficiency

82

impact

discretized into did

degrees

effect upon the solution

pattern

the

the

was

have

efficiency
What did

substructure

The first three

case

four substructures,

involving

studies

had

a similar number of master

had

similar solution efficiencies.

was

would

had

substructure

level

an

included in the
analyze

improve the

mass

CPU

In

time.

along the

However,

cases,

of

increasing

distribution.

freedom, it

of

freedom

efficiency

of

While

the

this algorithm

CPU time

in

less

This

accurate.

degrees

of

provide a required

efficiency

level

The

of

of

noticeably
a

were placed

freedom

will not

for

to compensate

one

only

CPU time

freedom

is

degrees

of

freedom. Thus,

are

needed to

sufficiently

increased

the

that are placed

solution

demonstrated
small,

are

in

be

be

degrees

needed

of

fraction

of

of

freedom for nonboundary

degrees

an

from

of

freedom

needed to

improvement in the

boundary

master

degrees

of

solution

degrees

freedom in

of

the

to place the extra nonboundary

boundary master degrees of freedom

efficiency

relevant

model

of

the resulting model.

to dynamic analysis, however

to other analysis types as well.

it

with a single substructure.

placed master

model would also

83

degrees

the

this algorithm

the number of

applicable

This

uses an

added to prevent the solution

is usually

have few, if any, arbitrarily

optimal solution efficiency.

master

Increasing

number of master

intended to be

ANSYS

are

number of

the solution

above could

that

boundary master degrees of freedom

master

result

but it has

studies,

the

CPU time

the

is

fewer

uses

increases

by

number of master

it may be desirable to

one-substructure model would

in

amount of

study

positions that will optimize mass

the process.

must

boundary

the decrease

may actually decrease

is relatively

should result

When

freedom

The

this

placing the nonboundary

CPU time in

decreasing

following recommendations

of the principles

of

for

reason

a given model that are placed

to offset the

degrees

The

freedom in

decrease in the total

enough

amount

of

effective at

However, if
the

two-substructure case

and three-substructure case

case studies.

exchange of

be

structure

four

degrees

not

by a small

four

than the

of solution accuracy.

freedom is relatively small,

master

degrees

freedom

study

placed upon

if it

the amount of

degrees

would

it

master

of

had

and

freedom is

as well as

degrees

which case

of

and

accuracy

interior locations, decreases the

The

this rule.

freedom usually decreases the

of the model.

may

freedom

of

of the model

to

needed to analyze the model.

eliminated, nonboundary master

becoming

master

as opposed

of

boundary

increase

accuracy

also uses a significant amount of

the master degrees of freedom

amount of

distribution

discover

to

study consisting

degree

the number of master

to place nonboundary master

algorithm

model

boundary

case

degrees

a master

mass

a solution with a given

difficult

was

master

equation and

exceptions to

degrees

master

the worst solution

master

When

and solution

find

boundaries

the model.

There are, however,

boundary

these

boundaries,

substructure

efficiency

boundary

improve

governing

distribution

most

It

their substructure

along

needed to

These non-optimally located

use pass

the model.

not

placed

The fourth

efficiency.

of solution efficiency.

interior location.

freedom

case studies.

placed

arbitrarily

boundary, it does

substructure

in

no

of

The CPU time

superior solution

yield

superior

degrees

for these first three

close

very

three substructures and two substructures, respectively

be

degrees

easier

of

many
If the

Since

freedom, it

to solve than one

involving
less file

substructures, because

multiple

handling and examination of details

the one-substructure model

using

limitations

of particular

front restrictions,

in

boundaries

boundary
extra

such a

is

way that the

master

Each
be

one that cannot

nonboundary

or

degrees

distribute

master

model generation

of

boundary

degrees

freedom is fixed for

freedom

to

degrees

of

several

freedom

of

distribution

mass

best

the

freedom in the

different

is

substructure

If the

on the substructure

placed on a substructure

also

increases

number of

discretizations, it

boundary

the

It is

also

desirable

to

discretize

in

the structure

such a

The CPU time

stress passes

half

not

performing

may

approach or exceed

solving
a

way

as to avoid

CPU time

to

the number of

handling

and other

These

produced

performing

model the

in CPU time. When

unchanging

in

the

iterations

first iteration. The

can

be

performed

element stresses or other quantities

substructures, CPU time

be

can

found in

saved

only once, namely,

using the

portions of the model that will

modeled either as substructures or as nonsubstructure elements

by

in

be

generation pass
updated can

be

the use pass. If the mode shape,

the stress pass are not required

the

eliminating

is

one

portions of the model as

substructures will require generation passes performed

subsequent

generation

required to solve the problem.

even one pass can produce a significant savings

for the first design iteration. The

files

master

needed to solve all of the generation and

of the total

design iteration scheme, it is best to

substructures.

output

the

required.

and stress passes whenever possible.

Hence,

file

CPU

be best

would

substructures used renders the problem easier to solve

reducing

the

by decreasing

provided

Minimizing

by

Adding

within the model.

distribution

solution

model.

analysts, makes a

use a model that contains the minimum number of substructures.

bookkeeping

wavefront

be used, it is best to discretize

freedom that is

of

improve

For efficiency,

by

structures where either

several

among

substructures are to

degree

greatest restriction to

other type of commercial software.

number of master

degrees

number of

master

any

placed to optimize the mass

time needed to solve the use pass.

of

ANSYS

The

may be limited

size of the model

impractical. If multiple

are minimized.

a single substructure would require

may be desirable for analyzing large

or the need to

one-substructure model
the model

model

containing

such as node renumbering.

is that the

versions of

A multiply-substructured
wave

a model

stress

passes

for
for

certain

those

substructures.

The

results can

demonstrates

solution

degrees

summarized

the effect of the

optimal master

amount of

be

degree

CPU time

of

by

the two graphs

boundary

freedom

master

configurations

to solve and

post-process

degrees

freedom along the

one-substructure model and

has

substructure

no

boundary

of

freedom

from the first three

the results.

efficiency is due to the fact that the first three

of

in Figure 7.1

84

Figure 7.2. Figure 7.1

on solution efficiency.

The

case studies required a similar

It is believed that this similarity in the

case studies

boundaries.

master

and

degrees

have

a similar number of master

The fourth
of

case

study involves

freedom. It is believed that this

FIGURE 7.1
CPU TIMES FOR THE OPTIMUM MASTER DEGREE OF FREEDOM
CONFIGURATION FOR EACH CASE STUDY

TT7T7TTT7-,
s/ssss/s/,

/////////,
/S//S//SS,
ssss/ss/s.

/////////,

s/s/ss/s /

in

ss/ssss/s,

/////////,
//S////S/,
SSS/S//S
/S//S/S/S,
///S////S,
/SS////SS

CO

D
m

O
Q

s
o

sssssssss.

(uuiu^t

rTTTTTTT-rr

IW-TWWW^

\N\\\\\\\\\S\\\S\SS\\\\N\\\\\\S\\\\\\\\\\\\\\\\

xxwxwxxx

WWW WWNW

SWWSWS'

\\\S\\\\V\\\\\\\\\\\\

\\\N\\\S\S

\XXXXXNNNNNN\NXXXXXXX
WXXXXXXNXXXXXXX*

*wwwwwwwww

.WW

wwwww*

wwwww*
wwwww*

wwwww*
wwwwswww WW \W WS WWW WWNWWWWW \w A\NN
NNNXXXNNXXNNXXX xxxxxxx WWW XXXXNXXXXX XXXWXXX W\ ,XNXXXXNN wwwww *
wwwww*
\\\\\\\\\\\\.\\\ xxxxx
wwwwwwwwwwww w\ ,ww
XNXXXXXXXX*
WWWNWWNWS NXNNXW XNXXXXXXXXXXNWXNXXXXW w\ ,WXX
wwwww*
wwwwwwws xxxxxxx
WWWWW NXNXXXXXNN \\\ AWN
wwwww*
xxxxxxxxxxxxxxx
xxxx
WXXXXXXXX XXX ,N\W
nnnn
XXXXXXXXXXNNNNN
WXXXXXXXXNXXXXXXXX \ w WWWWN wwwww*

r\i
oo
U3

V7TrrTTr77T77T77T77777TTTTTTTTTT7TT777
f/Sf//f/SS///////SS/SS/SS/fS//S////SS/
S/SS////SS/SS/S/SS//SSSSS//S//S//S//S/
//S/////S//SS///////S//S/S////////////
f////////S//////////////f/////////////

in
o

X/l
o

r-

.xxxxn

O
in

CO

WW WW WW

WW

xn

XNN

WW NWS

WW

WW
WW

WW

WW

WW

WW

7777TTTTTTTTTTT-,

s/ / /
/ s / s

////////S/S//SS,

On

SS/SS/S//SSS//S,

///////////////,
/////S/////S///,
/ss/s/sssssss/s,

in

CO

/ / / / s / SS/SS//SS//S//S,

f////SS////////S/////S//S//////f//////

/ / / / / / / / / / ///////S//////S,
/ / / ? ?/ ////SS//S//S///

f//S/SS////SS//S/SSSSSS/S//S//S//S//S/

/ s s / s s / / s / S///S//S/SSS/SS.

////////////S/S////////////S/////S////

/ // /

f/////S////////////////////S/////S//S/

oo

CO

///S//S/SS/SS////SS/S//SS//S/S///S/////S/ / s / / / / / s / s y / sssssssss/ss/ss,
f/////////////////f///////////S////////// /S//S//// / / / ////////S////S/
r//sss/////ss//ss/s/ssssss/s//s//s//s//// / s / / s / / s / / s / s/ss/ss/s/ss/ss,
SSSSSS/////SS/SS/SSSSS/SS/S///SS/SSSS//SS

OO

XXX\

w w\

WV
WW WW nnnv
WW WW WW
WW WW WW

7777
' / / /

'SSS/SS/S//SSSSSS/SSSS/SS/SSS/S/SSS/SS

.N

WWWWW*
.

v\\\\\v\\\\\\\\\\ \\ nx
XXXXXXXXXXXNNNNNNNNNNNNNNNXXXXXXXXN
\\\\\s\\\\\\\\\ xxxxxxx
NXXNNNNNNXXXXXNNNNN \\\ AWN
NXXXXXXXXXXXXN
WW wwwwwwwwwwww \ w

in

XNXXXXXXXX*

///////S///S/SS,

O
a

S
o

CO <M

vD

-J

w
TTTTTrrrXT W
^ *. *, ^ ^ V
S.XXXX XXX WW
www
nxxxx XXX xxxx
xnnn
WW xnxnxxxx XXN XXX XXX
XXX N\N XXX
WW XXX \\\w
WW www
XXX XXX X w
WXXXXXXNXXXXXNXXXX \*\WW
www ,\\\\\\N NW x XX x\x
XXX w\
WW NX WW
WWWWW WWWWWW WW
WW nxxxxnn XXX n\n
v"WTWWW*WWWWWW<,

<;

vXNXXXXXXXXXXXXXNXXXXXXX

,wwwsw\wwwwwwww

PQ

NXXXXXXXXXXXXXXXXXXXXXX

in

x\ \ XNNNNNXXXX
XXX XXXXXXXXXX
X\ X
NXXNXXNNX

NNNN XXXN
WW XXXN
XXX XNNNNNXXXX
XXXX NNNN
XXX XNXXXXXXXX
VXXX NNNN
xxxxxxxxx NNXXXXXXXXV
XXXXXX*.
XXXXXXXNNX
WXXXX*xxxxxx wx
nnnnnnn XXX XXX x\x x\\

.XXXXXXX

TTTTTTTTTT
x\x NNNNNNXXXX

.XXXXXXX

zn

O
Q

.XXXNXXXXXXXXXXXXXXXXXXX

eh
(XI
ON
vr

CQ

.XXXXXXXXXXXNXXXXXXXXXXXXX
XNNNNXXNNNNXXNNNNNNNNNNN'

www

WWW WS
.XXX

X*

XXXX xxxxn

wwwww
WWW XXXXX

XXX

WN w\ x\\ XXXXXXXNXX

NXXXXX*-

XXX XXX XXX x\\ XNNNNNXXSX

\xxxx XXX XXX x\x XXN XXXXNXXNNN
X\N XXXXXXXXNX
WWN XXX \w
WWN XXX
XXX XNNNNNNNNN

vNXXXX*

WW
www

WW

vww

.wxxxxxxxxxxxxxxxxxxxxxxxx
"

www

NNNNNN*

NNNN*-

P
in

S
o
o
CM

,WSWWW\WWNNNNVNWN

NNNNNX

CC
01
^r1

CD

O
O

O
Q
LJ
LJ
Ctl
Li_
to

L_
O

CM
o

CD

LlJ
LjJ
01
CD
LjJ
Q

LJ
CD
O

Ctl
LJ
E-h

Ll_

CO

CU
r\i

Ctl
o

CO

LJ

y
CD
LJ
Ctl'

L_

Q
L_J

'

DI

OrT
cn

[ZH)

X3N3riD3dJ
86

-i

lack

boundary

of

demonstrated
degrees

of

of master

master

degrees

in Figure 7.1.

freedom
degrees

of

freedom is

Figure 7.2 demonstrates the

for

on solution correlation

freedom

of

related to the superior solution efficiency

effect of the total number of master

a given substructure

discretization. As the

number

the substructured solution converges on the baseline

increases,

solution.

The

conclusions and recommendations

performed

using

a simple model

a restricted version of

to problems

software packages.
analysis

that

many

involving

However, it is

solution.

As

advised to

solutions

nonboundary
with

basic

not certain

may

not

degrees

of

any modeling technique,

invest in

are

master

of the

little

research

drawn from

heavily loaded DEC

be

learned

principles

larger models, different

types and which principles

number of

section are

simulations that were

ANSYS. The investigations involved

(only 648 degrees of freedom) on

The investigator believes

applicable

in this

applicable.

known.

87

during

minicomputer.

this research should

principles should

For example, in

freedom have little

and to solve

VAX 8810

analysis types and other

how these

an analyst

the modal analysis of

effect on

planning to

several

basic

use

be

hardware

applied

be

and

for

all

a static analysis, the

the

accuracy

substructuring

problems

for

the

of

would

which

be

the

REFERENCES
[1]

Cook, Robert D.; Concepts

[2]

Stasa, Frank L.;

[3]

Hughes, Thomas J. R.; The Finite Element Method: Prentice Hall; 1987

[4]

Mitchell, Robert K.; "Involvement and Integration

Engineering: V107, N6; Jun 1985; PP. 27-9

[5]

Falk, Howard; Beardsley, Charles W.; Finite Element Analysis Packages for Personal
Computers"; Mechanical Engineering: V107, Nl; Jan 1985; PP. 54-71

[6]

Hansen, S. D.; Anderton, G. L.; Connacher, N. E.; Dougherty, C. S.; "Analysis of the
747 Aircraft Wing-Body Intersection"; Proceedings of the 2nd Conference on Matrix
Methods in Structural Mechanics: Wright Patterson Air Force Base, Dayton, Ohio; Oct

and

Applications

of

Finite Element Analysis: Wiley; 1974

Applied Finite Element Analysis for Engineers: CBS Publications;

A Route

to

CAD/CAM";

Mechanical

"

15-17 1968

[7]

Kron, G.; Diakoptics: Macdonald; 1963

[8]

Turner, G. L.;

Milsted, M. G.; Hanks, P.; Adaption of Kron's Method for

Large Finite Element Models"; Journal of Vibrations. Acoustics. Stress
Reliability in Design: V108; Oct 1986; PP. 405-10
with

[9]

Wittrick, W. H.;
Frequencies
263-84

[10]

of

use

and

Williams, F. W.; "A General Algorithm for Computing Natural

Elastic Structures";
Quart. J. Mech. Appl. Math.: V24; PP.

Sehmi, N. S.; "Newtonian Procedure for the Solution of the Kron Characteristic
Value Problem"; Journal of Sound and Vibration: V100, N3; Jun 8, 1985; PP.
409-21

[11]

"The Kron Methodology and Practical Algorithms for Eigenvalue

Analysis of Large Scale Structural Systems"; Aeronautical
Response
Sensitivity
Journal: 1980; PP. 417-33

Simpson, A.;

and

[12]

Sehmi, N. S.; "Lanczos Algorithm Applied to Kron's Method"; International

Journal for Numerical Methods in Engineering; V23; 1986; PP. 1857-72

[13]

Wilson, Edward L.;

"Use of Special Ritz Vectors in Dynamic

Structural Engineering: VI 12, N8; Aug 1986;

Bayo, Eduardo P.;

Substructure Analysis":

Journal

of

PP. 1944-54

[14]

Yuan, M. W.; Dickins, J. M.; "Dynamic Analysis by Direct

Ritz Vectors"; Earth Engineering Structural Dynamics: V10, N6; Nov
1982; PP 613-23

Wilson, E. L.;
Superposition

of

[15]

"Application Ritz Vectors for Dynamic Analysis

Arnold, R. R.;
Structures"; Computers and Structures: V21.N5; 1985; PP. 901-8

[16]

Prakash, B. G.;

of

Large

"Reduction Techniques in Dynamic

Prabhar, M. S. S.;
Computers and Structures: V22, N4; 1986;
Substructures for Large Problems";

PP. 539-52

88

[17]

Bramble,

[18]

Zimmerle, Daniel; "Generation of Variable Order

Generation"; Computers and Structures: V22, N3; 1986;

[19]

Gunzburger, M. D.;

J. H.; Pasciak, J. E.; Shatz, A. H.;

"Construction of Preconditioners
for Elliptic Problems by Substructuring I"; Mathematics of Computation: V47,
N175; Jul 1986; PP. 103-34

Nicolaides, R. A.;

Solution Techniques for

the

"On

Elements
PP. 291-7

Superelement

Algorithms and
Partial Differential
V2; 1986; PP. 243-56

Substructuring

Numerical Approximation

Equations"; Applied Numerical Mathematics-

by

of

[20]

Wu, Shih-Chin;

[21]

Kane, T. R.; Ryne, R. R.; Banargee, A. K.; "Dynamics of a Beam Attached to a

Moving Base"; AAS/AAIA Astrodvnamics Specialist Conference: Vail, Colorado; Aug

Haug, Edward J.; "A Substructure Method for Dynamics of

Flexible Mechanical Systems with
Geometrically Nonlinear Deformation"; Center
for Computer Aided Design; Research Supported
by USAF Office of Scientific
Research; Grant # AFOSR-86-0032; Dec 1986; PP. 377-416

1985

[22]

Turcic, D. A.; Midah, A.; "Generalized Equations of Motion for the Dynamic Analysis of
Elastic Mechanism Systems"; Journal of Dynamic Systems. Measurement and Control:
V106; 1984; PP. 243-260

[23]

Lips, K. W.;
Substructure

Vigneron, F. R.;
"Damping Synthesis for a Spacecraft Using
Component Data"; Journal of Spacecraft and Rockets: V23, N2;

and

Mar-Apr 1986; PP. 158-64

[24]

Vigneron, F. R.; "Ground Test Derived Flight Values of Damping for a Flexible
Procedings of the Symposium on Dynamics and Control of
Spacecraft";
Non-Rigid Spacecraft: Frascati, Italy; May 1977; ESA-SP-117; PP. 325-33

[25]

Hasselman, T. K.; "Damping Synthesis for Substructure Tests"; AAIA Journal:

VI 3; Oct, 1976; PP 1409-18

[26]

Lynch, Charles T.; CRC

and

Handbook

of

Refractory Materials: CRC Press;

Materials Science: Volume II Metals. Composites

1975

89

APPENDIX 1

SUBSTRUCTURING COMMANDS IN ANSYS

Appendix 1

contains specific

information

in Chapter 5. For

are referenced

input file

that shows

variations of a standard

the pass

in

question.

stress pass command

command

handling
for the

lists

in

The
list

stress pass

contains the

is

files

commands and modules that

input files for

of the generation

in

performed

the

generation

of significant commands and a generalized

commands are used.

basic

The

display

generation and use passes are

the commands that are unique to

a separate module called

STRESS.

commands needed to perform a stress pass.

the three

The AUX4

(FTLE8)

module combines the stress pass output

of

list

analysis, so their command lists

and sample

files

there is a

how the listed

a substructured analysis.

output

substructuring

each of the three types of substructure passes

pass, the use pass and the stress pass

sample

about the

auxiliary

and

and use

AUX5

(FTLE13)

files (FILE 12) for

that represent the whole structure.

90

the

There

are

detailed file

modules that perform

modules perform node

The

renumbering

passes,

respectively.

different

substructures

The AUX1

into

a series

GENERATION PASS
INPUT COMMANDS
ANALYSTS TYPE AND OPTIONS

KAN,7

KAY,2,0

,N

KAY,5,N

KAY,6,0

--

,1

,2

,3

KAY,7,0

,1

--

,2

TELLS ANSYS THAT THIS IS A SUBSTRUCTURE GENERATION PASS

PUT SUBSTRUCTURE DATA IN FTLE8
PUT SUBSTRUCTURE DATA IN FTLEN

PLACE SUBSTRUCTURE DATA IN POSITION N OF SUBSTRUCTURE

LIBRARY FILE (USUALLY FTLE8)
STORE ONLY THE STIFFNESS MATRIX ON THE SUBSTRUCTURE
LIBRARY FILE
STORE THE STIFFNESS AND MASS MATRICIES ON THE
SUBSTRUCTURE LIBRARY FILE
STORE THE STIFFNESS, MASS AND DAMPING MATRICIES ON THE
SUBSTRUCTURE LIBRARY FILE
DON'T PRINT OUT EITHER THE MATRICIES OR THE LOAD VECTORS
PRINT OUT BOTH THE MATRICIES AND THE LOAD VECTORS
PRINT OUT THE LOAD VECTORS ONLY

MASTER DEGREES OF FREEDOM

M

TOTAL

FOR USER SELECTION OF MASTER DEGREES OF FREEDOM. NEEDED

TO SELECT DEGREES OF FREEDOM ON SUBSTRUCTURE BOUNDARIES
PROVIDES FOR AUTOMATIC SELECTION OF MASTER DEGREES OF
FREEDOM BY STATING THE TOTAL NUMBER AVAILABLE. CAN BE
USED TO IMPROVE MASS DISTRIBUTION IN DYNAMIC PROBLEM

BOUNDARY CONDITIONS
D

F
P
T
NT
HFLOW

DISPLACEMENT CONSTRAINTS

POINT FORCE

PRESSURE

AMBIENT TEMPERATURE

NODAL TEMPERATURE

HEAT FLOW

91

SAMPLE GENERATION PASS

/PREP7

ENTER MODEL GENERATION MODULE

KAN,7

SUBSTRUCTURE GENERATION PASS

KAY,2,0
KAY,5,M
KAY,6,2
KAY,7,0

PUT SUBSTRUCTURE LIBRARY ON FTLE8

PUT SS ON Mth POSTION ON FILE8
STORE MASS AND STIFFNESS MATRICIES
DONT PRINT OUT MATRICIES OR LOAD VECTORS

ET,1,JSTIF1

INPUT & IDENTIFY ELEMENT TYPES FN

SUBSTRUCTURE
ET,L,JSTIFL
R,l,...

! INPUT REAL CONSTANT SETS

R,0

EX,l,EFORMATl
! INPUT MATERIAL PROPERTIES
DENS,1 J3ENSITY FOR MAT1
NUXY.l, POISSON R FOR MAT1
EX,PJ FOR MATP
DENS,P,DENSITY FOR MATP
NUXY,0,POISSON R FOR MATP

TYPE, FIRST TYPE

REAL, FIRST REAL
MAT, FIRST MATERIAL

! SPECIFY

GENERATE ELEMENTS USING FIRST EL TYPE, FIRST REAL CONST AND FIRST
MATERIAL, THEN CHANGE TYPE, REAL AND MAT AND GENERATE OTHER
ELEMENTS

D,...

! USE D TO FIX DOF AS REQUIRED

M,...

!
!
!
!

TOTAL,#OF MDOF

SELECT BOUNDARY MASTER DEGREES OF

FREEDOM
ALLOW FOR AUTO SELECTION OF MDOF BY
STATTNG TOTAL # OF MDOF FN SS

F,...
P,...
ACEL,.

! CREATE 1st LOAD CASE USING F, P, ACEL AND

! OTHER COMMANDS

LWRTTE

! WRITE 1st LOAD CASE ONTO INPUT FILE

LWRTTE

! WRITE LAST LOAD CASE ONTO INPUT FTLE

AFWRJTE
FINISH

! WRITE ANSYS ANALYSIS FILE

/EXE
/INPUT,27

'

'

EXIT MODEL GENERATION MODULE

START GENERATION PASS ANALYSIS

92

USE PASS
INPUT COMMANDS

ELEMENT TYPE SELECTION

ET,ID#,JSTIF

DOF,UX,...,ROTZ

SELECT ELEMENT TYPE.

ELEMENT

JSTTF

50 FOR SUBSTRUCTURE

SPECIFY DEGREES OF FREEDOM (UX,...,ROTZ) TO BE USED

IN SUBSTRUCTURE PROBLEM

SUBSTRUCTURE ELEMENTS
E,I,J,K

INPUT SUBSTRUCTURE ELEMENT I

I
INDICATES SUBSTRUCTURE POSN. ON LIB. FTLE
J
INDICATES LIB. FILE #, DEFAULT IS FTLE8
KINDICATES NODE # OFFSET BETWEEN GEN. AND USE
PASSES

BOUNDARY CONDITIONS
D
F
P
T
NT
HFLOW

Any boundary
F, NT,

and

DISPLACEMENT CONSTRAINTS
POINT FORCE
PRESSURE
AMBIENT TEMPERATURE
NODAL TEMPERATURE
HEAT FLOW

condition can

HFLOW

can

be

be

applied to the nonstructure element

applied substructure master

93

degrees

of

degrees of freedom, but only D,

freedom.

SAMPLE USE PASS

/PREP7

! ENTER MODEL GENERATION MODULE

KANANAL TYPE #

! CHOOSE ANALYSIS TYPE

KAY,...

! SET ANAL OPTS FOR ANAL TYPE USED

ET,1,50

! INPUT AND ID SS ELEMENT TYPE

ET,2,JSTIF2

! INPUT AND ID NON SS ELEMENT TYPES

ET,L,JSTIFL
R,l,...

! INPUT REAL CONSTANT SETS FOR NON SS

! ELEMENTS

R,0,'...

EX,1,EF0RMAT1
DENS,1 DENSITY FOR MAT1
NUXY,l,POISSON R FOR MAT1

! INPUT MATERIAL PROPERTIES FOR NON SS

! ELEMENTS

EX,P,E FOR MATP

DENS JOENSJTY FOR MATP
NUXY,0,POISSON R FOR MATP
,P

DOF,ALL

! USE ALL 6 DEGREES OF FREEDOM, FROM UX

! TO ROTZ

TYPE,1

! SUBSTRUCTURE ELEMENTS

E,l,...

! INPUT SS 1 TO N

E,N,'...
TYPE,2
REALJTRST REAL
MATJTRST MATERIAL

! SET EL TYPE, REAL CONST AND MAT FOR NON

! SS ELEMENTS

GENERATE NONSUBSTRUCTURE ELEMENTS

M,...
TOTAL/TOTAL DOF

! DISPLAY TOTAL DOF CONFIGURATION IN USE

!PASS

D,...

! USE D TO FIX DOF AS REQUIRED

LWRTTE

! ADD LOAD CASES

AFWRJTE
FINISH

! WRITE ANSYS FILE

! EXIT MODEL GENERATION MODULE
! START USE PASS ANALYSIS

/EXE
/INPUT,27

94

STRESS PASS
INPUT COMMANDS

/STRES S,SENUM,SEOFS

ENTER THE STRESS PASS SOLUTION ROUTINE

SENUM IS THE SUBSTRUCTURE ELEMENT #
SEOFS IS THE SUBSTRUCTURE NODE OFFSET #
BOTH NUMBERS ARE FOUND IN THE SS ELEMENT
COMMAND, E, IN THE USE PASS

NSTRES,NUM

SELECTS THE NUMBER OF

EXPANSIONS TO BE PERFORMED

ITER,NITTER,NPRINT,NPOST

SELECTS THE DATA SETS TO BE EXPANDED.

THESE DATA SETS ARE IDENTrFIED BY DEFINING
NITTER, THE MAXIMUM ITERATION NUMBER. IF
NITTER
NUM, ON THE NSTRES COMMAND, THEN
ALL ITERATIONS ARE EXPANDED BETWEEN 1 AND
NITTER. IF NITTER IS GREATER THAN NUM, THEN
THE EXPANSIONS ARE PERFORMED AT EQUAL
SPACED INTERVALS IN THE RANGE BETWEEN 1
AND NITTER DEFINED BY THE VALUE OF
NnTER/NUM.

STRESS

PASS

FOR EXAMPLE, IF NITTER = 27 AND NUM = 9,

THEN THE FOLLOWING INTERVALS WOULD BE
EXPANDED: 3, 6, 9, 12, 15, 18, 21, 24, 27.

END

TERMINATES COMMAND
STRESS PASS EXPANSIONS

95

INPUT AND

STARTS

SAMPLE STRESS PASS

! SET ANSYS TO EXECUTE STRESS PASS

/EXEC
/STRES

,M

! ENTER STRESS PASS MODULE TO PERFORM STRESS

! PASS ON Mth SS

ITER,P, 1,1

! SET P AS THE NUMBER OF THE MAXIMUM ITERATION

NSTRES

! PERFORM L TTERATIONS

,L

END

! END INPUT AND START ANALYSIS

FINISH

! LEAVE MODULE

96

NODE RENUMBERING

INPUT COMMANDS
/AUX4

(OR/AUX5)

NNUM,NOD#l

,NOD#2

TAKES THE USER FROM THE ROUTINE BEGIN LEVEL

FNTO THE AUX4 OR AUX5 MODULE
PUTS AN ORDER TO RENUMBER NODE NOD#l TO NOD#2
ON A LIST

RENUM,F#1,F#2

EXECUTES ALL NODE RENUMBER ORDERS INPUT TO

LIST BY THE NNUM COMMAND. F#l IS SS LIBRARY
FILE NUMBER, USUALLY 8. THE RENUMBERED SS
LIBRARY FILE IS OUTPUT TO F#2

COPY,F#2,F#l

AN

AUX4/AUX5

LEVEL

VERSION

OF

COMMAND, IS USEFUL FOR MOVING

THE

/COPY

RENUMBERED

FILE TO ORIGINAL POSmON

FINISH

RETURNS THE USER TO THE ROUTINE BEGIN LEVEL

SAMPLE NODE RENUMBERING PASS

/AUX4

! ENTER MODULE

NNUM, 1,11

! RENUMBER NODES 1, 2, 3, 4 AND 5 TO 11, 22, 33, 44 AND

!55

NNUM,2,22
NNUM,3,33
NNUM,4,44
NNUM,5,55
RENUM,8,39

! RENUMBER THE CONTENTS OF FILE8 ACCORDING TO

! THE NNUM COMMANDS AND PUT THE RENUMBERED
! FILE IN FILE39

COPY,39,8

! MOVE RENUMBERED FILE, BACK TO FILE8 POSITION

FINISH

97

AUX1 FILE COMBINATION

INPUT COMMANDS

/AUX1

TAKES USER FROM THE ROUTINE BEGIN LEVEL INTO

THE AUX1 MODULE

SRANGE,LS1,IT1,LS2,IT2

SELECTS THE RANGE OF DATA SETS FOR WHICH THE

SS FILE 12 FILES CAN BE COMBINED. THE RANGE
RUNS FROM SET,LS1,IT1 TO SET LS2,IT2

SCOMB,LS 1

SELECTS THE TWO DATA SETS TO BE COMBINED IN

THE COMB COMMAND. THE SETS SELECTED WOULD
BE SET,LS1,IT1 FROM THE FIRST FILE AND SET
LS2JT2 FROM THE SECOND FILE

,IT1

,LS2,IT2

COMB,F#l,F#2,F#3

COMBINES THE SELECTED DATA SETS FROM FTLE#1

AND FILE#2 AND WRITES THE COMBINED FILE AS
FTLE#3

FINISH

RETURNS THE USER TO THE ROUTINE BEGIN LEVEL

SAMPLE AUX1 FILE COMBINATION PASS

SRANGE,1,4,1,4

! SELECT ITERATIONS 1 TO 4

SCOMB, 1,1, 1,1

! COMBINE FILES FOR rTERATTON #1

COMB,20,22,25
COMB,23,24,26
COMB,25,26,31

SCOMB, 1,2,1,2

! COMBINE FILES FOR ITERATION #2

COMB,20,22,25
COMB,23,24,26
SCOMB, 1,1, 1,1
COMB,25,26,32

! FILES 25 & 26 ARE MADE BY AUX1 AND ONLY HAVE

! ONE DATA SET

SCOMB, 1,3, 1,3

COMB,20,22,25
COMB,23,24,26
SCOMB, 1,1, 1,1
COMB,25,26,33

! COMBINE FILES FOR TTERATION #3

SCOMB, 1,4, 1,4

COMB, 20,22,25

! COMBINE FILES FOR ITERATION #4

COMB,23,24,26
SCOMB, 1,1, 1,1
COMB,25,26,34
FINISH

! RETURN TO RUN BEGIN LEVEL

98

FIGURE A. 1
FILE HANDLING FOR
AUX 1 EXAMPLE PASS

FILE 20

FILE 22

FILE 24

FILE 23

FILE 26

FILE 25

FILE N

*N

31,

32,33,34

99

APPENDIX 2

ANSYS INPUT FILES FOR THE FOUR CASE STUDIES

Appendix 2
each of

consists of the

four

the

remaining three
for

each case

ANSYS input files

case studies.

count

down

with

The first

three, two

study is the FILE18

from FILE 18.

university

version of

case

during one

study

"FILE POOL

passes.

The

ANSYS

error
runs.

run.

four

which

order

lists

The first file

during

the commands used

files that

for

substructures and the

created earlier and referenced

from performing

models

can

be

by

the

accessed

input

in

all of the passes of the

to prevent the system from

shutting down

the

first
with

message, the passes in the four substructure analyses were divide

The first

second run covered the

three case studies required

In

contains

the number of

prevent the user

interactive

file

input files

Limitations in

FULL"

into two interactive

are

study

dynamic

and one substructure respectively.

command

interactive input. The remaining files

commands

case

needed to substructure

fewer files

run covered

AUX1 file

performing the generation,

use and stress

combination and post processing.

and could thus

100

be

performed with one

The

interactive

other

run.

FILE18 FOR BASELINE ANALYSIS

/PREP 7

KAN, 2
KAY,
KAY,
KAY,
KAY,

1,-1

2, 4
3, 4
7, 4

ET,1,63
R,l,.2
EX,1,10.2E6

!
1

DENS,l,2.513E-4
NUXY,1,.33
K,l
K,2,53

ANALYSIS TYPE
MODAL ANALYSIS
MODAL ANALYSIS OPTIONS
=

ELEMENT TYPE 1
PLATE
ELEMENT PROPERTIES SET 1
MATERIAL PROPERTIES SET 1
-

INPUT STRUCTURE GEOMETRY

K, 3, 53, 42
K,4,,42

K,5,39.15,59.7
K,6,13.85,59.7

K, 7, 53,

,27

K,8,,,27

K, 9, 53, 42, 22
K, 10,
K, 11, 45. 3, 63. 7, 19. 5
K, 12, 7. 7, 63. 7, 19. 5
L, 1,2,5
,42,22

L,2,3,4
L,3,4,5
L,4,l,4
Li

.J

,5

L,6,4,2
L,2,7,3

L,7,8,5
L,8,l,3
L,8,10,4

L, 10, 4,3
L,7,9,4
11,9,3,3

L, 9,10, 5
L, 11, 12, 5
L,9,ll,2
L,ll,5,3

L, 10, 12, 2
L,12,6,3

A, 1,2, 3, 4
A, 4, 3, 5, 6
A, 1,2, 7, 8
A, 1,8, 10, 4
A, 2, 7, 9, 3
A, 6, 5, 11, 12
A, 10, 9, 11, 12
AMESH,ALL
D,1,ALL,,,2

CONSTRAIN NODES

WRITE ANSYS

/EXE

SET EXECUTE MODE

/INPUT, 27

PERFORM

43

46

D, 43, ALL
D, 46, ALL
AFWRIT

FILE

FINISH
BASELINE ANALYSIS

101

FOR CASE STUDY

FILE18

INPUT, CS1GN1

1
!

1ST GENERATION PASS

2ND GENERATION PASS

3RD GENERATION PASS

4TH GENERATION PASS

USE PASS

VIEW NATURAL FREQUENCIES

1ST STRESS

PASS

'.

2ND STRESS

PASS

3RD

PASS

FINISH

INPUT, CS1GN2

INISH

INPUT, CS1GN3

FINISH

INPUT, CS1GN4

FINISH
'

INPUT, CS1US4

FINISH
/POST1

SET, 1,1
SET, 1,2
SET

,1,3

SET, 1,4
FINISH
/

INPUT, CS1ST1

FINISH
'

INPUT, CS1ST2

FINISH

INPUT, CS 1ST?

STRESS

FINISH
i

INPUT, CS1ST4

'EOF
/

INPUT, CS1AX1

4TH STRESS PASS

END 1ST RUN

COMBINE OUTPUT FILES

'

PLOT

1ST MODE SHAPE

PLOT

2ND

MODE SHAPE

PLOT

3RD

MODE

SHAPE

PLOT

4TH

MODE

SHAPE

END

FINISH

/COPY, 31, 12
POST1

SHOW,
SET, 1,1
/VIEW, 1,2,1,
,,1

.5

/ANGLE, 1,-90
PLDISP,2
FINISH

/COPY, 32,12
PQST1
SET, 1,1
/VIEW, 1,2,1,. 5
/

/ANGLE, 1,-90
PLDISP,2
FINISH

/COPY, 33, 12
/P0ST1

SET, 1,1
/VIEW, 1,2,1,
/ANGLE, 1,-90

.5

PLDI5P,2
FINISH
/COPY, 34, 12
'P0ST1

SET, 1,1
/VIEW, 1,2,1,

.5

/ANGLE, 1,-90
LDISP,2
FINISH
'EOF

OF

2ND RUN

102

CS1GN1.DAT
/PREP7

KAN, 7
KAY, 5,1
AY, 6, 2

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN FIRST POSITION IN FILE8
STORE K & M MATRICIES IN FILER
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTIES SET ">
MATERIAL PROPERTIES SET 1
=

FT, 1,63
R,l, 2
EX,I,10.2E6

DEN3,l,2.513E-4

MUXY,1,

.33

K,l
K,2,53

INPUT SUBSTRUCTURE GEOMETRY

K. 3, 53,42
K,4,,42
X,5,39. 15,59. 7
K, 6, 13. 85, 59. 7

K, 7, 53,
K

,27

,27

L,l,2,5
L,2,3,4
[,,3,4,5
L 4 1 4
L, 3,5,2
,

L,5,6,5
L,6,4,2
L

3
,

L,8,l,3

A, 1,2, 3, 4
,4,3,5,6

1,2,7,3
AMESH,ALL
D,1,ALL,,,2
^,

D, 43, ALL
D, 46, ALL
M, 7, ALL, 11
M. 16, ALL, 18
M, 31, ALL
M, 33, ALL, 37
M, 44, ALL, 45
M, 51, ALL, 52
TOTAL, 151
AFWRIT
FINISH
/EXE

/INPUT,

27

CONSTRAIN NODES

SELECT MASTER DEGREES

WRITE ANSYS

FILE

SET EXECUTE

MODE

PERFORM

43

&

46

OF FREEDOM

1ST SUBSTRUCTURE GENERATION

103

PASS

CS1GN2.DAT

2, 29
/COPY, 3, 31
-COPY,

MOVE FILES 2 & 3 FROM 1ST SUBSTRUCTURE

GENERATION PASS OUT OF WAY

/PREP7

AN, 7

c>.AY,5,2

'.

KAY, 6, 2

ET,1,6 3

R,l,.2

FX,1,10.2E6

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN 2ND POSITION ON FILE8
STORE K & M MATRICIES ON FILES
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
=

DENS,l,2.513E-4
MUXY,1,.33

OFF SET NODE NUMBERS FOR

SO THAT THEY START AT 61

M,l,,,50
M, 60, 100

2ND

SUBSTRUCTURE

FILL, 1,60
INPUT SUBSTRUCTURE GEOMETRY

K,l
K.,2,,42

K, 3,, 42, 22
K.,4,,,27

L,l,2,4
^

L
L

.\,

3
4

4
3

1,2, 3, 4

AMESH.ALL

0,61, ALL
D, 69, ALL
M, 62, ALL, 66
M, 73, ALL, 74

61

&

'

CONSTRAIN NODES

SELECT MASTER DEGREES

DELETE OFFSET NODES

WRITE ANSYS FILES

69
OF FREEDOM

"0TAL,55
.jDELET,1,60

AFWRIT
FINISH

/EXE

/INPUT, 27

SET EXECUTE MODE

PERFORM 2ND SUBSTRUCTURE GENERATION

104

PASS

CS1GN3.DAT
AUX4

RENUMBER BOUNDARY NODES FOR THE 2ND SUBSTRUCTURE

NNUM, 62, 11
NNUM, 6 3, 13

NUM, 64, 17
NNUM, 6 5, 16
MNUM,7

3,51

NNUM, 7 4, 5 2
RENUM,8,39

COPY, 39,3
FINISH

/COPY, 2, 32
'COPY, 3, 3 3

MOVE FILES 2 & 3 FROM THE 2ND SUBSTRUCTURE

GENERATION PASS OUT OF WAY

-'PREP7

KAN 7
KAY

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN 3RD POSITION ON FILE8
STORE K & M MATRICIES ON FILES
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
=

,5,3

KAY

,6,2

ET,1,63

H,l,.2

EX,1,10.2E6

OENS,l,2.513E
TOXY 1
33
,

M,l,,,50

N, 80, 100
FILL, 1,30

OFFSET NODE NUMBERS FOR 3RD SUBSTRUCTURE

SO THAT THEY START AT 81

K,l,53

INPUT SUBSTRUCTURE GEOMETRY

CONSTRAIN

SELECT

MASTER

DELETE

OFFSET NODES

WRITE ANSYS

SET EXECUTE MODE

PERFORM 3RD SUBSTRUCTURE GENERATION PASS

K, 2, 53, 42
K, 3, 53, 42, 22
K, 4, 53,
,27

,1,2,4

o
_.

"1

J5

L,3,4,4
L,4,l,3

A, 1,2, 3,4
AMESH,ALL
D, 81, ALL
0,8 9, ALL
K, 8 2, ALL, 86

NODES

81

&

DEGREES

89
OF FREEDOM

M, 93, ALL, 94
TOTAL, 55
NDELET,1,80
AFWRIT
FINISH
/EXE
'INPUT, 2 7

105

FILES

CS1GN4.DAT
/

A.UX4
NNUM, 8 2, 7

RENUMBER BOUNDARY NODES FOR

3RD

SUBSTRUCTURE

MNUM,83,3
NUM, 84, 9
.>fNUM,85,lQ

M NUM, 93, 45
NNUM, 9 4, 4 4
RENUM,8, 3 9

COPY, 3 9,8
FINISH

/COPY, 2, 34
'COPY, 3, 35

MOVE FILES
3 FROM 3RD SUBSTRUCTURE
GENERATION PASS OUT OF WAY

'PREP7

KAN

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRIX IN 4TH POSITION ON FILE8
STORE K & M MATRICIES ON FILES
ELEMENT TYPE 1 = PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
=

KAY, 5, 4
KAY

ET,1,63

EX,1,10.2E6
CENS,l,2.513E-4

MUXY

M, 1,130

N,100,,,150
FILL, 1,100
K, 1,53, 42, 22

OFFSET NODE NUMBERS FOR 4TH SUBSTRUCTURE

SO THAT THEY START AT 101

'

INPUT

SELECT

!
!

DELETE

SUBSTRUCTURE GEOMETRY

K, 2,
K, 3, 45. 3, 63. 7, 19. 5
K, 4, 7. 7, 6 3. 7, 19. 5
,42,22

',5,39.15,59.7
<\, 6, 13. 85, 59. 7
L,l,2,5
L,3,4,5
L,5,6,5

_i

L,3,5,3
L,4,6,3

A, 1,3, 4, 2
A, 5, 3, 4, 6
AMESH,ALL

M,
M,
M,
M,
M,

101, ALL
109, ALL
119, ALL

MASTER

DEGREES

OF FREEDOM

122, ALL
125, ALL, 128

TOTAL, 7 9
MDELET, 1,100
AFWP.IT

FINISH
/EXE
' INPUT 2 7
,

i
i

WRITE

OFFSET NODES
FILES

ANSYS

SET EXECUTE MODE

4TH SUBSTRUCTURE GENERATION

PERFORM

106

PASS

FS1US4.DAT
<

AUX4

RENUMBER BOUNDARY NODES FOR 4TH SUBSTRUCTURE

MOVE FILES 2 & 3 FROM 4TH SUBSTRUCTURE

GENERATION PASS OUT OF WAY

NNUM, 10 1,8 6
W NUM, 109, 66
MUM, 119, 31
MNUM,125,34

NNUM, 126, 35
NNUM, 127, 36
NNUM, 123, 3 7
NNUM, 12 2, 3 3

R.ENUM

COPY, 39

1 Q
J

FINISH
'COPY,7

'COPY, 3
/PREP 7
KAN 2

37

KAY, 2, 4

KAY
KAY

ANALYSIS TYPE
MODAL ANALYSIS

MODAL ANALYSIS
OPTIONS

,3,4

,7,4

FT, 1,50
DOF TJX UY US
ELI
E,2
F,3
,

R9TX ROTY
,

!
!

ROT"
,

SUBSTRUCTURE
ELEMENT TYPE 1
DOF FOR EACH NODE
SUBSTRUCTURE ELEMENTS
=

E,4

M, 7, ALL, 11
M, 16, ALL, 18
M, 31, ALL
',33, ALL, 37
A, 51, ALL, 52
M, 44, ALL, 45
M, 66, ALL
M, 86, ALL
TOTAL, 2 20

MASTER

>

AFWRIT
FINISH

WRITE

PERFORM USE

'EXE
/INPUT, 2

DEGREES

OF FREEDOM

SUBSTRUCTURES

ANSYS

FILE

SET EXECUTE MODE

PASS

107

FOR

ALL

CS1ST1

.DAT

MOVE USE PASS OUTPUT FILE (FILE13) TO FILES

MOVE FILES 2 & 3 FOR 1ST SUBSTRUCTURE INTO

PLACE
SET EXECUTE MODE

13,
/COPY, 29, 2
'COPY,

'COPY,
EXEC
/ STRES

ITER

MSTRES

END

-J

A.

1
t

J-

,1

,1

STRESS

PASS

EXPAND

1ST

1ST SUBSTRUCTURE
DATA SETS 'MODES)

FOR
4

,4

PERFORM

1ST

STRESS

PASS

108

CS1ST2.DAT
'COPY, 12
'
AUX5

MOVE OUTPUT FILE FOR

RE-RENUMBER BOUNDARY NODES FOR 2ND SUBSTRUCTURE

32
'COPY, 3 3

'

MOVE FILES

/EXEC

SET EXECUTE MODE

STRESS PASS FOR 2ND SUBSTRUCTURE
EXPAND 1ST 4 DATA SETS (MODES)

NNUM,
NUM,
NNUM,
NNUM,

,2

OF WAY

17, 6 4

NNUM, 52

74
39
3

73

8
3
COPY, 9

RENUM

OUT

13,63
65

PASS

11, 6 2

16
51

NNUM

1ST STRESS

FINISH
/COPY,

'

,4,1

STRES
ITER
NSTRES
END

&

FOR

2ND SUBSTRUCTURE

4
1

PERFORM

2ND

STRESS

PASS

109

INTO

PLACE

\S1ST3.DAT

'COPY, 12, 2
/AUX5

MOVE OUTPUT FILE FOR 2ND STRESS PASS OUT OF WAY

RE-RENUMBER BOUNDARY NODES FOP. 3RD SUBSTRUCTURE

NNUM, 7, 8 2
NUM, 8, 83
cJNUM,9,84

NNUM, 10, 3 5
NNUM, 4 5, 9 3
NNUM, 44 94
RENUM,3

39

COPY, 39,

FINISH

FOR

'COPY, 34,2
'COPY, 3 5, 3

MOVE FILES

/EXEC

SET EXECUTE MODE

'

STRES

ITER, 4,1
M STRES, 4
END

3RD

SUBSTRUCTURE

STRESS PASS FOR 3RD SUBSTRUCTURE

EXPAND 1ST 4 DATA SETS (MODES)
PERFORM

3RD

STRESS

PASS

110

INTO

PLACE

CS1ST4.DAT
/COPY.l
'

MOVE

OUTPUT FILE FOR 3RD STRESS PASS OUT OF WAY

RENUMBER BOUNDARY NODES FOR 4TH SUBSTRUCTURE

AUX5

NNUM, 86
NUM, 66
n NUM,

31

NNUM, 34
35
NNUM, 36
NNUM, 3 7
NNUM, 3 3

NNUM

RENUM

COPY, 3 9

,101

,109

,119

,125

126

,127

,123

,122

,39

,3

FINISH

/COPY, 3 6,2
'COPY, 3 7,3

ITER 4
M STRES
,

MOVE FILES

&

SET EXECUTE MODE

'

STRESS

1,1

EXPAND

4
1

PERFORM

'.

MOVE

/EXEC
/ STRES

END

FROM 4TH SUBSTRUCTURE

INTO

PLACE

PASS FOR 4TH SUBSTRUCTURE

1ST 4 DATA SETS (MODES)
4TH

STRESS

PASS

FINISH
'COPY,!

,24

OUTPUT

FILE FOR THE

111

4TH

STRESS

PASS

OUT

OF THE WAY

7S1AX1.DAT
1

AUX1

3 RANGE

,1,1,1,4

COMB, 1,1, 1,1

ENTER FILE COMBINATION MODULE

RANGE OF DATA SETS = 1ST 4 ITER.

1ST

MODE

SHAPE

2ND

MODE

SHAPE

'.

3RD

MODE

SHAPE

4TH

MODE

SHAPE

COMB, 20, 22, 25

COMB, 23, 24, 26
COMB, 25, 26, 31
SCOMB, 1,2, 1,2
",0MB,20,22,25

COMB, 2 3, 2 4, 26
SCOMB, 1,1, 1,1
COMB, 25, 26, 3 2
SCOMB. 1,3, 1,3
COMB, 20, 22, 25

COMB, 2 3, 24, 26
SCOMB, 1,1, 1,1
COMB, 25, 26, 33
SCOMB, 1,4, 1,4
COMB, 20, 22, 25
COMB, 23, 24, 26
SCOMB,!, 1,1,1
COMB, 2 5, 26, 34

112

FOR

LOAD STEP

FILE18
/ INPUT

CASE STUDY 2
,CS2GN1

1ST GENERATION PASS

,CS2GN2

2ND GENERATION PASS

,CS2GN3

3RD GENERATION PASS

FINISH
/ INPUT
FINISH

/INPUT
FINISH
/ INPUT

USE PASS

,CS2US3

FINISH
/P0ST1

VIEW NATURAL

FREQUENCIES

SET,1, 1
SET,1, 2

SET,1, 3
SET,1, 4
FINISH
/INPUT
FINISH

,C32ST1

1ST STRESS

PASS

/INPUT

,CS2ST2

2ND STRESS

PASS

FINISH
/ INPUT

,CS2ST3

3RD STRESS

PASS

FINISH

/COPY, 31,12

PLOT

1ST MODE SHAPE

PLOT

2ND MODE SHAPE

PLOT

3RD MODE

SHAPE

PLOT

4TH MODE

SHAPE

/POST1

/SHOW,
SET,1, 1
/VIEW, 1,2,1,
,,1

/ANGLE

PLDISP

.5

,1,-90

,2

FINISH

/COPY, 32,12
/POST1

SET, 1,
/VIEW, 1,2,1,
/ANGLE

PLDISP

.5

,1,-90

,2

FINISH

/COPY, 33,12
/POST1

SET,1, 1
/VIEW, 1,2,1,
/ANGLE

PLDISP

.5

,1,-90

,2

FINISH

/COPY, 34,12
/POST1

SET,1,
/VIEW,
/ANGLE
PLDISP
FINISH

1,2,1,-5
,1,-90

,2

/EOF

113

CS2GN1.DAT
/PREP 7

KAN, 7
KAY, 5,1
KAY, 6, 2

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN 1ST POSITION ON FILE18
STORE KSM MATRICIES ON FILE8
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
=

ET,1,63
R,l,.2
EX,1,10.2E6

DENS,l,2.513E-4

NUXY,1,

.33

K,l

INPUT SUBSTRUCTURE GEOMETRY

!
!

CONSTRAIN NODES 1 & 2

SELECT MASTER DEGREES

AFWRIT
FINISH

WRITE ANSYS

/EXE

SET EXECUTE MODE

/INPUT, 27

PERFORM

K,2,53

K, 3, 53, 42
K,4,,42

K,5,39.15,59.7
K,6,13.35,59.7
L,l,2,5
L,2,3,4
L,3,4,5
L,4,l,4
L,5,6,5

L,6,4,2

A, 1,2, 3, 4
A, 4, 3, 5, 6
AMES H, ALL
D,1,ALL, ,
M, 3, ALL, 11
M, 16, ALL, 18
M, 31, ALL
M, 33, ALL, 37
TOTAL, 13 3
,2

OF FREEDOM

FILE

1ST SUBSTRUCTURE GENERATION PASS

114

CS2GN2.DAT

/COPY, 2, 29
/COPY, 3, 31

MOVE FILES

GENERATION PASS OUT OF WAY

/PREP 7
KAN, 7

&

FROM 1ST SUBSTRUCTURE

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN 2ND POSITION ON FILE8
STORE K & M MATRICIES ON FILE8
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
=

KAY, 5, 2
KAY, 6, 2
ET,1,63
R,l,.2

EX,1,10.2E6
DENS,l,2.513E-4
NUXY,1,.33
N,l,,,50
N,50,,,100

!
!

OFF SET NODE NUMBERS FOR

SO THAT THEY START AT 51

INPUT SUBSTRUCTURE GEOMETRY

2ND SUBSTRUCTURE

FILL, 1,50
K,l
K,2,53

K, 3, 53, 42
K,4,,42
K,5,,,27

K,6,53,
K, 7, 53, 42, 22
K, 8,
/VIEW, 1,2,1,
,27

,42,22

.5

L,l,2,5
L,2,3,4
L,3,7,3
L,7,6,4
L,6,5,5
L,5,8,4
L,8,4,3
L,4,l,4
L,l,5,3

L,2,6,3

A, 1,2, 6, 5
A, 1,5, 8, 4
A, 6, 2, 3, 7
/SHOW,,,l
AMESH,ALL
NDELET,1,50

D, 51, ALL,,, 52

DELETE OFFSET NODES

CONSTRAIN NODES 51 52

57

SELECT MASTER DEGREES

OF FREEDOM

WRITE ANSYS FILE

FINISH
/EXE

/INPUT, 27

SET EXECUTE MODE

PERFORM 2ND SUBSTRUCTURE GENERATION PASS

D, 57, ALL
D, 60, ALL
M, 53, ALL, 56
M, 75, ALL
M, 79, ALL
M, 82, ALL, 84
M, 91, ALL, 95

60

TOTAL, 127
AFWRIT

115

US2GN3

/AUX4

RENUMBER BOUNDARY NODES FOR

2ND

SUBSTRUCTURE

NNUM, 53, 3
NNUM, 54, 4
NNUM, 55, 5
NNUM,
NNUM,
NNUM,
NNUM,
NNUM,
NNUM,
NNUM,
NNUM

56, 6
79, 11
84, 18
83, 17
82, 16
91, 7
92, 8

,93,9

NNUM, 94, 10
RENUM,8,39

COPY, 39, 8
FINISH

/COPY, 2, 32
/COPY, 3, 33

MOVE FILES 2 & 3 FROM 2ND SUBSTRUCTURE

GENERATION PASS OUT OF WAY

/PREP 7

KAN, 7
KAY, 5, 3
KAY, 6, 2
ET,1,63
R,l,.2
EX,1,10.2E6
DENS,l,2.513E-4
NUXY,1,.33
N, 1,130

N, 110,150
FILL, 1,110
K, 1,53, 42, 22
K,2,

SUBSTRUCTURE GENERATION PASS

ANALYSIS TYPE
STORE MATRICIES IN 3RD POSITION ON FILE8
STORE K & M MATRICIES ON FILE8
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
-

OFFSET NODE NUMBERS FOR 3RD

SO THAT THEY START AT 111

INPUT SUBSTRUCTURE GEOMETRY

DELETE

SUBSTRUCTURE

,42,22

K,3,45.3,63.7,19.5
K,4,7.7,63.7,19.5
K,5,39.15,59.7

K, 6, 13. 85, 59. 7

L, 1,2,5
L,3,4,5

L,5,6,5
L,l,3,2
L,3,5,3
L,2,4,2

L,4,6,3

A, 1,3, 4, 2
A, 5, 3, 4, 6
AMESH,ALL
NDELET, 1,110

M, 111, ALL
M, 119, ALL
M, 129, ALL

OFFSET NODES
SELECT MASTER DEGREES

WRITE ANSYS FILE

'

SET EXECUTE MODE

PERFORM 3RD SUBSTRUCTURE GENERATION PASS

OF FREEDOM

M, 132, ALL
M, 135, ALL, 138
TOTAL, 80
AFWRIT
FINISH

/EXE

/INPUT, 2 7

116

CS2US3.DAT
/AUX4

3RD SUBSTRUCTURE

RENUMBER

MOVE FILES 2 S 3 FROM 3RD SUBSTRUCTURE

GENERATION PASS OUT OF WAY

BOUNDARY NODES FOR

NNUM,
NNUM,
NNUM,
NNUM,

111, 95
119, 75
129, 31
132, 33
NNUM, 135, 34
NNUM, 136,35
NNUM, 13 7, 36
NNUM, 138, 37
RENUM,8,39
COPY, 39, 8
FINISH
/COPY, 2, 34

/COPY, 3, 35
/PREP 7

KAN, 2
KAY, 2, 4

ANALYSIS TYPE

MODAL ANALYSIS

KAY
KAY

MODAL ANALYSIS

OPTIONS

,3,4

,7,4

ET,1,50

! ELEMENT TYPE 1
SUBSTRUCTURE
I DOF FOR EACH NODE
DOF,UX,UY,UZ,ROTX,ROTY,ROTZ
'
E,l
SUBSTRUCTURE ELEMENTS
E,2
E,3
! MASTER DEGREES OF FREEDOM FOR ALL
M, 3, ALL, 11
=

M, 16, ALL, 18
M, 31, ALL
M, 33, ALL, 37
M, 75, ALL
M, 95, ALL
TOTAL, 220

THREE

AFWRIT

WRITE ANSYS

/EXE

/INPUT, 27

SET EXECUTE MODE

PERFORM USE PASS

SUBSTRUCTURES

FILE

FINISH

117

CS2ST1.DAT

/COPY, 13,8
/COPY, 29, 2
/COPY, 31, 3
/EXEC
/ STRES
ITER
NSTRES

,4,1,1

END

MOVE USE PASS OUTPUT FILE (FILE13) TO FILE8

MOVE FILES 2 & 3 FOR 1ST SUBSTRUCTURE INTO
PLACE
SET EXECUTE MODE
STRESS PASS FOR 1ST SUBSTRUCTURE
EXPAND 1ST 4 DATA SETS (MODES)
PERFORM 1ST STRESS PASS

118

C32ST2.DAT

/COPY, 12, 36

/AUX5

MOVE OUTPUT FILE FOR 2ND STRESS PASS OUT OF WAY

RE-NUMBER BOUNDARY NODES FOR 2ND SUBSTRUCTURE

MOVE FILES

NNUM, 11, 79
NNUM, 16,82
NNUM, 17, 83
NNUM, 18,84
NNUM, 3, 53
NNUM, 4, 54
NNUM, 5, 55
NNUM, 6, 56
NNUM, 8, 92
NNUM, 9, 93
NNUM, 10, 94
NNUM, 7, 91
RENUM,8,39

COPY, 39, 8
FINISH

/COPY, 32, 2
/COPY, 33, 3

ITER

,4,1,1

NSTRES

END

&

FOR

2ND SUBSTRUCTURE

SET EXECUTE MODE

STRESS PASS FOR THE 2ND
EXPAND 1ST 4 DATA SETS

/EXEC
/ STRES

4
i.

PERFORM

2ND

STRESS

119

PASS

SUBSTRUCTURE

INTO

PLACE

CS2ST3.DAT

/COPY, 12, 37

MOVE OUTPUT FILE FOR 2ND STRESS PASS OUT OF WAY

RE-RENUMBER BOUNDARY NODES FOR 3RD SUBSTRUCTURE

/AUX5

NNUM, 95,
NNUM, 75,
NNUM, 31,
NNUM, 33,
NNUM, 34,
NNUM, 35,
NNUM, 36,
NNUM, 37,

111
119
129
132
135
136
137
138

RENUM,8,39

COPY, 39, 8
FINISH

/COPY, 34, 2
/COPY, 35, 3

MOVE FILES

/EXEC

/STRES, 3
ITER, 4, 1,1

SET EXECUTE MODE

STRESS PASS FOR 3RD SUBSTRUCTURE
EXPAND 1ST 4 DATA SETS (MODES)

END
FINISH

PERFORM

/AUX1

SRANGE,1,1,1,4

ENTER DATA FILE COMBINATION MODULE

RANGE OF DATA SETS
1ST 4 ITER. FOR LOAD STEP

NSTRES

&

FOR

3RD

SUBSTRUCTURE

INTO

PLACE

SCOMB, 1,1, 1,1

COMB, 36, 37, 39
COMB, 39, 12, 31

3RD

STRESS

PASS

1ST MODE

SCOMB, 1,2, 1,2

COMB, 36, 37, 39
SCOMB, 1,1, 1,2
COMB, 39, 12, 32

2ND MODE

SCOMB, 1,3, 1,3

COMB, 36, 37, 39
SCOMB, 1,1, 1,3
COMB, 39, 12, 33

3RD MODE

SCOMB, 1,4, 1,4

COMB, 36, 37, 39
SCOMB, 1,1, 1,4
COMB, 39, 12, 34

4TH MODE

120

18 FOR CASE STUDY

FILE

3
1

1ST GENERATION PASS

/ INPUT, C33GN2
"TNISH

2ND GENERATION PASS

INPUT, CS 3 US 2

USE PASS

'

VIEW NATURAL

INPUT, C33GM1

'

FINISH

FINISH
/

POST1

SET,
SET,
SET,
SET,

FREQUENCIES

1,1
1,2
1,3
1,4

FINISH
'INPUT,CS3ST1

1ST

STRESS

PASS

2ND

STRESS

PASS

PLOT

1ST

MODE

SHAPE

PLOT

2ND MODE

SHAPE

PLOT

3RD MODE

SHAPE

PLOT

4TH

FINISH
,'INPUT,CS3ST2

FINISH

/COPY, 3 1,12
,'POSTl

/SHOW,,
SET

,1

,1,1

'VIEW,
'

1,2,1,

ANGLE

.5

,1,-90

PLDISP, 2
FINISH
/COPY, 32, 12
/POST1

SET, 1,1
/VIEW, 1,2,1,
ANGLE, 1,-90

.5

.LDISP,2

FINISH

/COPY, 33, 12
/POST1

SET

,1,1

/VIETL 1,2,1,
'ANGLE, 1,-90
PLDISP, 2

.5

FINISH

/COPY, 34, 12

MODE SHAPE

/POST1

SET, 1,1
'VIEW, 1,2,1,. 5

/ANGLE, 1,-90
PLDISP, 2
FINISH
/EOF

121

CS3GN1.DAT
'PREP 7

KAN,
KAY

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN 1ST POSITION ON FILES
STORE K & M MATRICIES ON FILE3
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1
MATERIAL PROPERTY SET 1
=

,5,1

AY, 6, 2
1 , 63
2
R 1
EX,1,10.2E6
ET

DENS,l,2.513E-4

NUXY,1,

.33

K,l

INPUT SUBSTRUCTURE GEOMETRY

CONSTRAIN NODES 1 & 2

SELECT MASTER DEGREES

K,2,53

K, 3, 53, 42
K.,4,,42
K,5,39. 15,59.7

K, 6, 13. 85, 59.

L,l,2,5
~*

~>

L
L

3
4

4
,

4
1

5
4
n

<-

L,5,6,5

L, 5,4,2
A

A, 4, 3,5,6
\MESH,ALL

D,1,ALL,,,2
M, 3, ALL, 11

'

OF FREEDOM

M, 16, ALL, 18
M, 31, ALL
ALL, 37
.,,33,

TOTAL
AFWRIT
FINISH

,13

3
!

WRITE ANSYS

FILE

/EXE

'.

/INPUT, 27

SET EXECUTE
PERFORM 1ST

MODE
SUBSTRUCTURE GENERATION

122

PASS

CS3GN2.DAT
'

2, 29
'COPY, 3, 31
'COPY,

MOVE FILES 2 & 3 FROM 1ST SUBSTRUCTURE

GENERATION PASS OUT OF WAY

'PREP 7
AN, 7

ANALYSIS TYPE
SUBSTRUCTURE GENERATION PASS
STORE MATRICIES IN 2ND POSITION ON FILE8
STORE K Sl M MATRICIES ON FILES
ELEMENT TYPE 1
PLATE
ELEMENT PROPERTY SET 1

rvAY,5,2

'.

KAY

1
,6,7

ET,1,63

EX,1,10.2E6

'.

MATERIAL PROPERTY SET

M,l,,,50

M, 50,,, 100
FILL, 1,50

OFFSET NODE NUMBERS FOR 2ND SUBSTRUCTURE

SO THAT THEY START AT 51

K,l

DENS,l,2.513E-4

MUXY,1,.33

INPUT

SUBSTRUCTURE GEOMETRY

K.,2,53

K, 3, 53, 42
K,4,,42
K.,5,,,27

K, 6, 53,, 27
K, 7, 53, 42, 22
K, 8,, 42, 22
K, 9, 45. 3,63.7,19.5
K, 10, 7, 7, 63. 7, 19. 5
K, 11, 39. 15, 59. 7
K, 12, 13. 85, 59. 7
L 1, 2 , 5
L,2,3,4
,

'

,3,7,3

7,6,4
L,6,5,5
L,5,8,4
u,

L,8,4,3
L,4,l,4
L,l,5,3
L,2,6,3
1,7,3,5
L,9,10,5
0,11,12,5

L,7,9,2
L,8,10,2
L,9,ll,3

L, 10, 12, 3
A, 1,2, 6, 5
A, 1,5, 8, 4
A, 6, 2, 3, 7
h, 8, 7, 9, 10
A, 12, 11, 9, 10
AMES H, ALL
NDELET,1,50

D, 51, ALL,
D, 57, ALL
D, 60, ALL
M, 5 3, ALL, 56
',79, ALL
rf, 82, ALL, 84
M, 91, ALL, 94
M, 123, ALL, 128
,,52

OFFSET

DELETE

CONSTRAIN

SELECT

NODES

NODES

MASTER

123

51

52

DEGREES

57

& 60

OF FREEDOM

TOTAL, 193
AFWRIT
FINISH

WRITE ANSYS

''EXE

/INPUT, 27

SET EXECUTE MODE

PERFORM 2ND SUBSTRUCTURE GENERATION PASS

124

FILE

CS3US2.DAT
AUX4

RENUMBER BOUNDARY MODES

FOR

2ND

SUBSTRUCTURE

NNUM, 79, 11
NNUM, 82, 16
TNUM,83,17
,t

NUM, 84

NNUM

18

53

.j

NNUM, 54
NNUM, 55
NNUM, 56
NNUM, 92
NNUM, 93
NNUM, 9 4
NNUM, 91
NNUM, 123',31
W NUM, 125 34
NNUM, 126 35
.4

.5

.6

,8

.9

.10

,7

NNUM, 127 36
NNUM, 123 37
33

NNUM, 124
RENUM

8
'TJPY,3 9
,

39
3

FINISH
/

COPY

'COPY, 3

33

&

MOVE FILES

GENERATION PASS

'

ANALYSIS TYPE

FROM 2ND SUBSTRUCTURE

OUT

OF WAY

/PREP7

KAN
KAY

MODAL

,2,4

ANALYSIS

MODAL ANALYSIS

OPTIONS

KAY, 3, 4
"AY, 7,4
ELEMENT TYPE 1
SUBSTRUCTURE
DOF FOR EACH NODE
DOF,UX,UY,US,ROTX,ROTY,ROTZ
E,l
SUBSTRUCTURE ELEMENTS
E,2
! MASTER DEGREES OF FREEDOM FOR BOTH
M, 3, ALL, 11
-

^T,l,50

SUBSTRUCTURES

AFWRIT
FINISH
/EXE

WRITE

/INPUT, 2 7

M, 16, ALL, 18
M, 31, ALL
M, 33, ALL,

37

TOTAL, 218

',

ANSYS

FILE

SET EXECUTE MODE

PASS

PERFORM USE

125

CS3ST1.DAT
'COPY, 13,3
'COPY, 29, 2

MOVE USE PASS OUTPUT FILE (FILE13) TO FILES

MOVE FILES 2 & 3 FOR 1ST SUBSTRUCTURE INTO

/COPY, 31, 3

PLACE

EXEC

STRES
ITER

,4,1,1

NSTRES

END

SET EXECUTE MODE

,

STRESS

PASS

EXPAND

1ST

FOR
4

1ST

SUBSTRUCTURE

DATA SETS

4
PERFORM

1ST

STRESS

PASS

126

(MODES)

CS3ST2.DAT
/COPY, 12,35

MOVE OUTPUT FILE FOR

'AUX5

'

RENUMBER BOUNDARY NODES

1ST

STRESS
FOR

2ND

PASS

OUT OF WAY
SUBSTRUCTURE

NNUM, 11, 79
MUM, 16,82
NNUM

,17,83

NNUM, 18, 8 4
NNUM, 3, 5 3
NNUM, 4, 5 4
NNUM, 5, 55
NNUM, 6, 56
NNUM, 8, 92
NNUM

NNUM

,9,93

,10,94

NNUM, 7, 91
NNUM, 31, 123
NNUM, 34, 125
NNUM, 35, 126
NNUM, 36, 127
NNUM, 37, 128
NNUM, 3 3, 124
RENUM,3,33

COPY, 3 9, 8
FINISH

/COPY, 3 2, 2
'COPY, 3 3, 3

MOVE FILES

/EXEC

SET EXECUTE MODE

/ STRES

ITER, 4, 1,1

STRESS

PASS

&

FOR

2ND

SUBSTRUCTURE

2ND SUBSTRUCTURE
DATA SETS (MODES)

1ST

PERFORM

2ND

STRESS

/COPY, 12,36

MOVE FILE FOR

/AUX1
SRANGE,1,1,1,4

!
'.

ENTER FILE COMBINATION MODULE

RANGE OF DATA SETS = 1ST 4 ITER.

SCOMB, 1,1, 1,1

COMB, 3 5, 36, 31

1ST

SCOMB, 1,2,1,2
COMB, 35, 36, 32

2ND MODE

SCOMB, 1,3,1,3
COMB, 35, 36, 33

3RD MODE

PLACE

FOR

EXPANDS

'STRES

INTO

END

PASS

FINISH

SCOMB

,1,4,1,4

WND

STRESS

MODE

4TH MODE

COMB, 35, 36, 34

127

PASS

OUT

OF THE WAY

FOR

LOAD STEP

FILE

18

FOR

CASE STUDY

INPUT,C34GN1

4
!

GENERATION PASS

USE PASS

VIEW NATURAL

STRESS

PLOT

1ST MODE

SHAPE

PLOT

2ND

MODE

SHAPE

PLOT

3RD

MODE

SHAPE

PLOT

4TH MODE

SHAPE

FINISH
/

INPUT, CS4US1
INISH

'POST1

FREQUENCIES

1 1
SET, 1,2
SET, 1,3
SET, 1,4
SET

FINISH
/

INPUT, CS4ST1

PASS

FINISH
,'POSTl

/SHOW,,,l

'VIEW, 1,2,1,. 5
/ANGLE, 1,-90
SET, 1,1
PLDISP, 2
SET, 1,2
PLDISP, 2
SET, 1,3

PLDISP, 2
SET, 1,4
PLDISP, 2
FINISH
'EOF

128

CS4GN1.DAT
'PREP 7

KAN, 7
KAY, 5,1
AY, 6, 2

SUBSTRUCTURE GENERATION PASS

ANALYSIS TYPE
STORE MATRICIES IN 1ST POSITION ON FILE8
STORE K & M MATRICIES ON FILE8
-

ET,1,63

ELEMENT TYPE

PI

ELEMENT PROPERTY

.2

EX,1,10.2E6

MATERIAL

PLATE
SET 1

PROPERTY

SET

DENS,l,2.513E-4

NUXY,1,

.33

K.,1
K,2,53

INPUT GEOMETRY

K, 3, 53, 42
K,4,,42
K,5,39.15,59.7

K, 6, 13. 85, 59.

K, 7, 53,

,27

K,8,,,27

K, 9, 53, 42, 22
K.,10,

,42,22

K, 11, 45. 3,63.7,19.5

K, 12, 7. 7, 63. 7, 19. 5
r

0,3,4,5
L,4,l,4
0,3,5,2
L,5,6,5
L,6,4,2
,2,7,3

-,7,8,5

L,8,l,3
L 8 10 4
,

L,10,4,3
L,7,9,4
L,9,3,3
L,9,10,5

L, 11, 12, 5
L,9,ll,2
L,ll,5,3

L, 10, 12, 2
L, 12,6,3
h, 1,2, 3, 4
A, 4, 3, 5, 6
A, 1,2, 7, 8
A, 1,8, 10,4
fl,2,7,9,3

A, 6, 5, 11, 12
MO, 9, 11, 12
AMESH,ALL
D,1,ALL,,,2

NODES

CONSTRAIN

WRITE

SET EXECUTE MODE

PERFORM

43

D, 43, ALL
D, 46, ALL

TOTAL, 2 0
AFWRIT
INISH
'EXE
'

INPUT 2 7

ANSYS

FILE

GENERATION

129

PASS

&

46

CS4US1.DAT

2, 3 2

/COPY, 3, 3 3
/ PREP 7
AN, 2

kAY,2,4

'COPY,

KAY
KAY

MOVE FILES 2 & 3 FOR SUBSTRUCTURE

GENERATION PASS OUT OF WAY

ANALYSIS TYPE
MODAL ANALYSIS

MODAL
OPTIONS

.ANALYSIS

,3.4

,7,4

ET,1,50

ELEMENT TYPE

SUBSTRUCTURE

DOF, TJX, UY,UZ,ROTX,ROTY, ROTS

! DOF FOR EACH NODE
i
E,l
SUBSTRUCTURE ELEMENT
20
!
MASTER
DEGREE OF FREEDOM
TOTAL,
i
AFWRIT
WRITE ANSYS FILE
FINISH
/EXE

'

/INPUT, 27

SET EXECUTE MODE

PERFORM USE PASS

130

CS4ST1.DAT
/COPY, 13, 8
/COPY, 32, 2
/COPY, 33,3
EXEC

PASS OUTPUT FILE (FILE13) TO FILES

MOVE FILES 2 & 3 FOR SUBSTRUCTURE INTO

PLACE
SET EXECUTE MODE

/ STRES

ITER
NSTRES

,4,1,1

END

MOVE

STRESS PASS
EXPAND 1ST 4 DATA SETS

PERFORM STRESS

PASS

131

(MODES)