INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING

Int. J. Numer. Meth. Engng 2005; 63:589–602

Published online 3 March 2005 in Wiley InterScience ([Link]). DOI: 10.1002/nme.1295

Structural shape optimization using equivalent static loads

transformed from dynamic loads

K. J. Park1 , J. N. Lee2 and G. J. Park3,∗, †

1 Departmentof Machine Design and Production Engineering, Hanyang University,
17 Haengdang-dong, Sungdong-gu, Seoul 133-791, South Korea
2 Mobile Display Development Team, Samsung Electronics Co., Ltd, San #24 Nongseo-ri, Giheung-Eup,

Yongin-city, Kyeonggi-do 449-901, South Korea

3 Division of Mechanical Engineering, Hanyang University, 1271 Sa-1 dong, Sangnok-gu, Ansan,

Kyeonggi-do 425-791, South Korea

SUMMARY
In structural optimization, static loads are generally utilized although real external forces are dynamic.
Dynamic loads have been considered only in small-scale problems. Recently, an algorithm for dynamic
response optimization using transformation of dynamic loads into equivalent static loads has been
proposed. The transformation is conducted to match the displacement fields from dynamic and static
analyses. This algorithm can be applied to large-scale problems. However, the application has been
limited to size optimization. The present study applies the algorithm to shape optimization. Because
the number of degrees of freedom of finite element models is usually very large in shape optimization,
it is difficult to conduct dynamic response optimization with conventional methods that directly treat
dynamic response in the time domain. The optimization process is carried out by interfacing an
optimization system and an analysis system for structural dynamics. Various examples are solved to
verify the algorithm. The results are compared to the results from static loads. It is found that the
algorithm using static loads transformed from dynamic loads based on displacement is valid for very
large-scale shape optimization problems. Copyright 䉷 2005 John Wiley & Sons, Ltd.

KEY WORDS: dynamic response optimization; shape optimization; equivalent static loads; dynamic
factor; transient analysis

1. INTRODUCTION

In general structural design, the structure is analysed for prescribed external loads and
designed based on the results of the analysis. Most of the loads applied on structures in

∗ Correspondence to: G. J. Park, Division of Mechanical Engineering, Hanyang University, 1271 Sa-1 dong,
Sangnok-gu, Ansan, Kyeonggi-do 425-791, South Korea.
† E-mail: gjpark@[Link]

Contract/grant sponsor: Center of Innovative Design Optimization Technology

Received 23 March 2004
Revised 2 August 2004
Copyright 䉷 2005 John Wiley & Sons, Ltd. Accepted 17 November 2004
590 K. J. PARK, J. N. LEE AND G. J. PARK

the real world are dynamic loads. It is very important that applied loads are properly evaluated
and reasonably applied in structural optimization when the finite element method (FEM) is
utilized. However, most practical applications assume that the applied loads are static even
though they are dynamic loads. This is due to difficulties on design sensitivity analysis with
dynamic constraints. Design sensitivity analysis of dynamic systems is extremely complicated
because their objective function and constraints are expressed as implicit functions of time and
state variables [1–4]. Therefore, dynamic response optimization to directly handle functions
in the time domain has been confined to small-scale problems with only a few degrees of
freedom [2, 5–8].
Static response optimization is well developed. Many studies have achieved structural
optimization under static loads substituted for dynamic loads. Due to many advantages such
as time, cost, etc., transformation of static loads from dynamic loads is widely utilized in
structural design. Transformation using a dynamic factor is generally used [9]. The dynamic
factor is determined by a design code or the designer’s experiences. This technique, however,
has limitations in that it does not properly evaluate dynamic loads. Recently, an algorithm has
been proposed to handle dynamic loads on structures with many degrees of freedom (DOFs). It
performs dynamic response optimization with equivalent static loads, which generate the same
displacement field as that under dynamic loads. The main idea of this algorithm is transfor-
mation of a dynamic load into a set of equivalent static loads and subsequent static response
optimization with the transformed equivalent static loads as a multiple load case [10–13].
The objective of this study is to conduct shape optimization of a structure subjected to
dynamic loads. Most structural shape optimization problems are large scale problems which have
a large number of DOFs in analysis and many design variables and constraints in optimization.
Conventional methods for dynamic response optimization have difficulties in accommodating
large-scale problems. Therefore, shape optimization with dynamic response is rare.
In this study, shape optimization is conducted by using a dynamic response optimization
algorithm with the proposed equivalent static loads. Some examples using equivalent static loads
show that shape optimization with many DOFs is successfully achieved. Static optimization
using a dynamic factor is also performed. The results from both methods are compared with
each other. The results show that static optimization using a dynamic factor cannot properly
incorporate the dynamic response. The solving process employs commercial systems such as
ABAQUS 6.2 [14] for transient analysis and GENESIS 6.0 [15] for shape optimization. The
interface between transient analysis and optimization is developed to evaluate the equivalent
static loads and update design variables.

2. DYNAMIC RESPONSE OPTIMIZATION USING EQUIVALENT STATIC LOADS

2.1. Equivalent static loads

Dynamic load is the real force where the magnitudes and directions change in a time domain.
Static load is the ideal force where the magnitudes and directions are constant regardless of
time. Static load cannot perfectly represent the vibration of the structure subjected to a dynamic
load. Since static response optimization is easily performed, the dynamic load is transformed
into a set of static loads. The criterion of the transformation is an equivalent static load (ESL).
Equivalent static loads are static loads that generate the same displacement fields as those from

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
 STRUCTURAL SHAPE OPTIMIZATION USING EQUIVALENT STATIC LOADS 591

a dynamic load at any time step [11]. They are defined at all the points in the time domain.
According to general transient analysis and structural optimization associated with FEM, the
structural dynamic behaviour is presented by
M(b)ÿ(t) + K(b)y(t) = r(t) (1)
where M is the mass matrix; K, the stiffness matrix; b, the design variable vector; r, the
dynamic load vector and y, the displacement vector from dynamic loads. Structural damping
is ignored. Considering Equation (1) and the FEM equation for static analysis, the equivalent
static loads are defined as following:
K(b)y(t) = r(t) − M(b)ÿ(t) (2)
or
K(b)y(t) = f eq (3)
Thus,
f eq = r(t) − M(b)ÿ(t) (4)
where f eq is the equivalent static load vector. It includes external forces and inertia forces at
all DOFs and all discrete time steps of transient analysis. Thus, it is physically identical with
the applied dynamic load even though it is a static load.

2.2. Algorithm
A solution of Equation (1) is generally obtained using a numerical method where finite time
parameters are needed. The solution and the ESL is also finite in the discrete time domain.
The solution of Equation (1) can be obtained by transient analysis. The equivalent static loads
at the uth discrete time step are evaluated by the following:
u
f eq = K(b)yu = r(tu ) − M(b)ÿu , u = 1, 2, . . . , q (5)
where q is the number of time grid points in dynamic analysis.
The equivalent static loads at each time step are treated independently and used as multiple
loading conditions for the static optimization process as illustrated in Figure 1. Dynamic
response optimization for a structure under dynamic loads in the discrete time domain can be
formulated as

Find b (6a)

to minimize
(b) (6b)

subject to M(b)ÿu + K(b)yu = r(t), u = 1, . . . , q (6c)

gj u (b, yu , ẏu , ÿu )
0, j = 1, . . . , m, u = 1, . . . , q (6d)

where b is the design variable vector; Equation (6c) is a state equation expressed in terms
of the finite element method; M, K, r and y are mass matrix, stiffness matrix, external load

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
592 K. J. PARK, J. N. LEE AND G. J. PARK

Applied force
Time step: t0 t1 t2 ...... tn . . . . . . Time

Load set: s0 s1 s2 ...... sn ......

Figure 1. Time domain and multiple loading conditions.

Start

Set initial variables

Perform a transient analysis

Update design
Calculate equivalent static loads

Slove static optimization

Satisfy termination
criteria ? NO

YES

End

Figure 2. Optimization process using equivalent static loads.

vector and nodal displacement vector, respectively;
is the objective function; gj u is the j th
constraint at the uth time grid; m is the number of inequality constraints. The equivalent static
load (ESL) is obtained from Equation (5). Dynamic response optimization using equivalent
static load (DROESL) is formulated as follows:

Find b (7a)

to minimize
(b) (7b)

u
subject to K(b)zu = f eq , u = 1, . . . , q (7c)

gj u (b, zu )
0, j = 1, . . . , m, u = 1, . . . , q (7d)

A cycle for dynamic response optimization using equivalent static load is defined by the
process illustrated in Figure 2. Suppose a structure is subjected to the dynamic loads as

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
 STRUCTURAL SHAPE OPTIMIZATION USING EQUIVALENT STATIC LOADS 593

illustrated in Figure 1. The algorithm for the dynamic response optimization with the equivalent
static load is as follows:
Step 1: Set initial design cycle (p = 0) and initial design variables (bp = b0 ).
Step 2: Perform a transient analysis in Equation (1) with bp for the pth design cycle.
Step 3: Calculate equivalent static loads in the time domain as Equation (5).
Step 4: When p = 0, go to Step 5. When p>0, if

q 


p u u 
 f eq − p−1 f eq  < (8)
u=1

terminate the process. Otherwise go to Step 5. p f equ is the equivalent static load

vector at the uth time step in the pth design cycle.

Step 5: Solve the static response optimization problem of Equations (7a)–(7d).
Step 6: Set p = p + 1 and go to Step 2.
At the pth cycle, a static response z is identical with a dynamic response y before the static
optimization process. However, the two responses are different after static optimization. The
difference can be dissolved by repeating the cycle. The process is defined with a sequence of
transient analysis, calculating the equivalent static loads and static optimization. The cycle is
iterated until the convergence criterion is satisfied. It is verified that the final solution obtained
by the proposed algorithm satisfies the Karush–Kuhn–Tucker (KKT) necessary condition (i.e.
the optimality condition) [12, 13].

3. EXAMPLES

Several examples for shape optimization are solved to verify the efficiency and validity
of the proposed method. However, it is almost impossible to perform direct dynamic re-
sponse shape optimization. No reference for shape optimization of dynamic response is found.
Thus, the proposed optimization method is not compared with a conventional optimization
method.
Instead, the optimization results by the equivalent static load method are compared with
the results from static optimization using a dynamic factor. The impact load is defined with a
proper load to easily compare the results between static response optimizations using a dynamic
factor and equivalent static loads.
A load of short duration has little effect on a structure, although it is very large. When
the duration time of a load is longer than 0.5 times of the natural frequency of a system,
the maximum displacement is induced regardless of the duration time of the load [16]. Thus,
transient analysis for this research is conducted with adequate duration time of a load with
respect to the natural frequency of the structure. When a dynamic load is imposed on a spring
without energy loss, the dynamic displacement is twice of the displacement induced from a
static load with the same magnitude as the peak of the dynamic load. In this spring impact
problem, the dynamic factor is obtained as 2 and is the same in many structures. Therefore,
the dynamic factor 2 is used in static optimization of the examples to compare the optimization
results from equivalent static loads.

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
594 K. J. PARK, J. N. LEE AND G. J. PARK

f
22,241N
f
Y
Z X

P1 P2 P3 P4 0.01sec

Figure 3. Loading and boundary condition of an 18-rod truss.

(a) (b)

Figure 4. Shape optimization result of an 18-rod truss: (a) using static loads with a dynamic
factor; and (b) using equivalent static loads.

3.1. Eighteen rod truss

The eighteen-bar truss in Figure 3 is selected. The structure is fixed on the left side and
dynamic loads are imposed on the upper nodal points. The dynamic load is constant for a
certain duration of time as illustrated in Figure 3. The modulus of elasticity is 73.1 GPa and
the density is 2700 kg/m3 . The optimization problem is formulated as follows:
Find design variables
to minimize weight
subject to |max |
13 MPa (9)

FL 2 E
− = −54 GPa
A 2 4
The problem has stress constraints and Euler buckling constraints for all the members of the
structure. To avoid Euler buckling of the elements, the force in each element must be constrained
to be less in magnitude than the Euler buckling load for that element. This constraint can be
expressed as
  2 A2
F  FE = −EI for a circular cross section: I = (10)
L 4
Design variables 1–4 are cross-sectional areas of the top members, bottom members, oblique
members and vertical members, respectively. Other design variables are the locations of x and
y co-ordinates of selected nodal points (P1–P4) in Figure 3. Optimization is performed in two
cases: using static loads with a dynamic factor and using equivalent static loads. As mentioned
earlier, the value of the dynamic factor is 2.
The final results of the two cases are illustrated in Figure 4 for the shape change and shown
in Table I for the change in all design variables. Objective and maximum constraint violation

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
 STRUCTURAL SHAPE OPTIMIZATION USING EQUIVALENT STATIC LOADS 595

Table I. Optimization results for the eighteen-rod truss.

Optimization using Optimization
Initial static loads with a using equivalent
Design variables design dynamic factor static loads

DV1 Cross-sectional areas of the top side

members (mm2 ) 6451.6 5135.3 5338.8
DV2 Cross-sectional areas of the bottom
side members (mm2 ) 13967.7 18221.3 17758.6
DV3 Cross-sectional areas of the oblique
members (mm2 ) 8064.5 5125.6 6208.3
DV4 Cross-sectional areas of the vertical
members (mm2 ) 4561.3 2007.2 3474.4
DV5 The location of x co-ordinate of P4 in
Figure 3 1000 915.2 924.3
DV6 The location of y co-ordinate of P4 in
Figure 3 0.0 187.7 164.1
DV7 The location of x co-ordinate of P3 in
Figure 3 750 648.5 663.2
DV8 The location of y co-ordinate of P3 in
Figure 3 0.0 148.6 137.9
DV9 The location of x co-ordinate of P2 in
Figure 3 500 417.9 419.7
DV10 The location of y co-ordinate of P2 in
Figure 3 0.0 95.4 84.2
DV11 The location of x co-ordinate of P1 in
Figure 3 250 204.4 198.8
DV12 The location of y co-ordinate of P1 in
Figure 3 0.0 20.9 7.7
OBJ Mass of the system (kg) 2916.7 2451.7 2587.8

histories in optimization using ESLs are illustrated in Figure 5. Summary of problem size
and job information is also presented in Table II. Optimum shapes are different in the two
cases and the size design variables are also different. Particularly, the cross-sectional area of
vertical members has a difference of 40%. It is due to the buckling of a vertical member in
transient analysis. In the early time step of transient analysis, the vertical members buckled as
illustrated in Figure 6(a). However, static analysis with a dynamic factor cannot accommodate
this phenomenon as illustrated in Figure 6(b). The buckling of vertical members causes the
violation of the constraints in the optimization process so that the optimization results for the
two cases are different.

3.2. Bracket
A plane bracket with a quadrangular hole in Figure 7 is formulated for a dynamic response
optimization problem. The structure is subjected to the distributed dynamic loads at the top-
side. Right side circular holes are fixed. The modulus of elasticity, the density and Poisson’s
ratio of the structural materials employed are 200 GPa, 7850 kg/m3 and 0.3, respectively. The

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
596 K. J. PARK, J. N. LEE AND G. J. PARK

1.0

0.8

Normalized values
0.6 Objective
Max. constraint violation
0.4

0.2

0.0
0.0 1 2 3 4 5 6 7 8 9 10 11 12 13
No. of iterations

Figure 5. Objective and maximum constraint violation histories of the 18-truss

problem in optimization using ESLs.

Table II. Summary of the 18-truss problem.

Summary of problem size
No. of DOFs of FE model 22
No. of design variables (time independent) 12
No. of constraints (time independent) 36
No. of discrete time points of transient analysis 100
No. of equivalent static load cases 100

Summary of job information

No. of transient analyses 13
No. of design cycles using the proposed method 13
No. of iterations of static optimization in the proposed method 704

(a) (b)

Figure 6. Results of dynamic and static analyses for the initial model of an 18-rod truss:
(a) dynamic analysis; and (b) static analysis.

formulation for dynamic response optimization is represented in Equation (11).

Find design variables

to minimize weight (11)
subject to |max |
300 MPa

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
 STRUCTURAL SHAPE OPTIMIZATION USING EQUIVALENT STATIC LOADS 597

Applied force

fixed

500 N
f

x y

0.01sec
z

fixed

Figure 7. Loading condition for the shape optimization of a bracket.

y x y
x

z z
(a) (b)

Figure 8. Shape optimization for a bracket with holes: (a) using static loads
with a dynamic factor; and (b) using equivalent static loads.

Design variables are the location and the size of the quadrangular hole. The final results
of both methods are illustrated in Figure 8 and Table III. The objective function and the size
of the quadrangular hole may look similar for both cases. However, the vertical location of
the quadrangular hole has an opposite direction. After optimization, both designs are
reanalysed under the original dynamic loads. Both results are presented in Figure 9. The struc-
ture designed by the dynamic factor has stress violation while the one from ESL has active
constraints. Therefore, optimization with dynamic factors can be dangerous. Design history in
optimization using ESLs is illustrated in Figure 10 and the design summary is presented
in Table IV.

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
598 K. J. PARK, J. N. LEE AND G. J. PARK

Table III. Optimization results for the bracket problem.

Optimization using Optimization
static loads with a using equivalent
Design variables Initial design dynamic factor static loads

DV1 Change of the x-dir. location of

the quadrangular hole (mm) 0.0 − 3.0 − 2.1
DV2 Change of the z-dir. location of
the quadrangular hole (mm) 0.0 3.0 − 2.3
DV3 Change of the size of the
quadrangular hole (mm) 0.0 18.4 20.0
OBJ Mass of the system (g) 816.6 559.18 529.48

Max. stress: 329.9 MPa Max. stress: 302.0 MPa

(a) (b)

Figure 9. Comparison of dynamic re-analyses for the optimized bracket: (a) using static loads
with a dynamic factor; and (b) using equivalent static loads.

1.2

1.0 ox ox
Normalized values

0.8

0.6

objective: mass of the system

0.4 DV1: change of the x-dir. location of the hole
o DV2: change of the z-dir. location of the hole
0.2 x DV3: change of the size of the hole
Max. constraint violation

0.0 ox
0 1 2
No. of iterations

Figure 10. Design history of the bracket problem in optimization using ESLs.

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
 STRUCTURAL SHAPE OPTIMIZATION USING EQUIVALENT STATIC LOADS 599

Table IV. Summary of the bracket problem.

Summary of problem size
No. of DOFs of FE model 3348
No. of design variables (time independent) 3
No. of constraints (time independent) 490
No. of discrete time points of transient analysis 100
No. of equivalent static load cases 100

Summary of job information

No. of transient analyses 2
No. of design cycles using the proposed method 2
No. of iterations of static optimization in the proposed method 10

Applied force

fix

11.5 kN
f
(a)

Design variable 1 Design Design Design Design

variable 2 variable 3 variable 4 0.01sec
variable 5
(c)

(b)

Figure 11. Loading condition for the shape optimization of a connecting rod: (a) loading
condition; (b) design variable; and (c) applied load.

3.3. Connecting rod

A connecting rod as illustrated in Figure 11 is optimized. It is popularly used in the development
process of shape optimization methodologies. This example distinctly shows that structural
optimization must be performed under a dynamic load, not a static load. The structure is to
be designed for a fixed left hole and an axial dynamic load is applied at the centre node of
the right semicircle as illustrated in Figure 11(a). The modulus of elasticity, the density and
Poisson’s ratio are 200 GPa, 7260 kg/m3 and 0.25, respectively. The formulation for dynamic
response optimization is represented as follows:

Find design variables

to minimize weight (12)
subject to |max |
250 MPa

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
600 K. J. PARK, J. N. LEE AND G. J. PARK

Figure 12. Dynamic and static analyses for the initial model: (a) static
analysis; and (b) dynamic analysis.

Figure 13. Results of shape optimization of the connecting rod: (a) using static loads with
a dynamic factor; and (b) using equivalent static loads.

Table V. Optimization results for the connecting rod problem.

Optimization using Optimization
static loads with a using equivalent
Design variables Initial design dynamic factor static loads

Design variable 1 0.0 − 0.5 0.0285

Design variable 2 0.0 − 0.6 − 0.5999
Design variable 3 0.0 − 0.6228 − 0.3946
Design variable 4 0.0 − 0.3777 − 0.3663
Design variable 5 0.0 0.6224 − 0.4999
Objective Mass of the system (g) 213.22 168.94 173.18

Design variables are the dimensions of the right and left semicircular shapes and the width
of the rod as illustrated in Figure 11(b).
First, static and transient analyses are performed. Figures 12(a) and (b) are the regions with
maximum stresses. The maximum stress from static analysis occurs at the right end while the
one from transient analysis occurs at the left end.
Figure 13 and Table V present the optimization results from static optimization with the
dynamic factor and optimization with equivalent static loads. They are very different designs.
In order to evaluate the validity of the two cases, transient analysis with their results is
performed. The stress results for the two cases are illustrated in Figure 14. The stress constraint
is considerably violated in Figure 14(a) while it is satisfied in Figure 14(b). This indicates that
the static loads with a dynamic factor cannot properly incorporate the properties of dynamic
effects. However, optimization with equivalent static loads incorporates them well. Design
history in optimization using ESLs is illustrated in Figure 15 and the design summary is
presented in Table VI.

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
 STRUCTURAL SHAPE OPTIMIZATION USING EQUIVALENT STATIC LOADS 601

Max. stress: 303.4 MPa Max. stress: 259.6 MPa

(a) (b)

Figure 14. Comparison of dynamic re-analysis for the optimized connecting rod: (a) using static loads
with a dynamic factor; and (b) using equivalent static loads.

2 2
1.2

1.0 5 5

3
Normalized values

0.8
4 3
4
objective
0.6
1 Design variable 1
2 Design variable 2
0.4 3 Design variable 3
4 Design variable 4
5 Design variable 5
0.2 Max. constraint violation
1

0.0 ox
1
0 1 2
No. of iterations

Figure 15. Design history of the connecting rod problem in optimization using ESLs.

Table VI. Summary of the connecting rod problem.

Summary of problem size
No. of DOFs of FE model 2958
No. of design variables (time independent) 5
No. of constraints (time independent) 432
No. of discrete time points of transient analysis 100
No. of equivalent static load cases 100

Summary of job information

No. of transient analyses 2
No. of design cycles using the proposed method 2
No. of iterations of static optimization in the proposed method 6

4. CONCLUSIONS

A dynamic response optimization method is proposed for the structure subjected to dynamic
loads. The equivalent static loads at each nodal point and each discrete time are obtained from
the displacement and stiffness matrix through transient analysis. Static analysis with a set of

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602
602 K. J. PARK, J. N. LEE AND G. J. PARK

equivalent static loads can include the dynamic effect in a discrete time domain. Structural
shape optimization under dynamic loads is solved with a set of equivalent static loads that are
used as multiple loading conditions. The proposed optimization process is iterative until the
design criterion is satisfied.
Several examples are solved by the proposed method to accomplish structural shape
optimization under a dynamic load. The optimization results are compared with those of static
optimization with a dynamic factor. The shape change of the structure is very different in the
results of both methods. Static optimization with a dynamic factor has obvious limitations to
incorporate a precise dynamic response. When dynamic analysis is performed with the optimum
solution, some constraints are violated. The results from dynamic response optimization using
equivalent static loads, however, satisfy the constraints because a set of equivalent static loads
incorporates the dynamic load effects very well. It is found that the method with equivalent
static loads is valid for shape optimization.

ACKNOWLEDGEMENTS
This research was supported by the Centre of Innovative Design Optimization Technology, which was
funded from the Korea Science and Engineering Foundation. The authors are thankful to Mrs MiSun
Park for her correction of the manuscript.

REFERENCES
1. Haug EJ, Arora JS. Applied Optimal Design. Wiley: New York, USA, 1979.
2. Feng TT, Arora JS, Haug EJ. Optimal structural design under dynamic loads. International Journal for
Numerical Methods in Engineering 1977; 11:39– 62.
3. Haug EJ, Arora JS, Feng TT. Sensitivity analysis and optimization of structures for dynamic response.
Journal of Mechanical Design 1978; 100:311– 318.
4. Hsieh CC, Arora JS. Design sensitivity analysis and optimization of dynamic response. Computer Methods
in Applied Mechanics and Engineering 1984; 43(2):195 – 219.
5. Lim OK, Arora JS. Dynamic response optimization using an active set RQP algorithm. International Journal
for Numerical Methods in Engineering 1987; 24:1827–1840.
6. Paeng JK, Arora JS. Dynamic response optimization of mechanical systems with multiplier methods. ASME
Journal of Mechanism, Transmission and Automation in Design 1989; 37:413 – 430.
7. Wilmert KD, Fox RL. Optimum design of a linear multi-degree of freedom isolation system. ASME Journal
of Engineering for Industry 1972; 94:456 – 471.
8. Heieh CC, Arora JS. A hybrid formulation for treatment of point-wise state constraints in dynamic response
optimization. Computer Methods in Applied Mechanics and Engineering 1985; 48:171–189.
9. Haftka RT, Gurdal Z. Elements of Structural Optimization. Kluwer Academic Publishers: The Netherlands,
1991.
10. Kang BS, Choi WS, Park GJ. Structural optimization under equivalent static loads transformed from dynamic
loads based on displacement. Computers and Structures 2001; 79(2):145 –154.
11. Choi WS, Park GJ. Structural optimization using equivalent static loads at all the time intervals. Computer
Methods in Applied Mechanics and Engineering 2002; 191:2105 – 2122.
12. Kang BS. On optimality of solution by equivalent static load method and its application to flexible multibody
dynamic systems. Ph.D. Thesis, Hanyang University, Korea, 2002.
13. Park GJ, Kang BS. Validation of a structural optimization algorithm transforming dynamic loads into
equivalent static loads. Journal of Optimization Theory and Applications 2003; 118(1):191– 200.
14. ABAQUS/Standard User’s Manual, Version 6.2. HKS Inc.: Pawtucket, RI, USA, 1997.
15. GENESIS User Manual, Version 7.0. VRAND, INC.: CO, USA, 2001.
16. Hibbeler RC. Mechanics of Materials. Prentice-Hall: Englewood Cliffs, NJ, 1979.

Copyright 䉷 2005 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2005; 63:589–602