Optimum Design of Structures With Stress and Displacement Constraints Using The Force Method

Available online at www.sciencedirect.
com
International Journal of Mechanical Sciences 45 (2003) 13691389
Optimum design of structures with stress and displacement
constraints using the force method
R. Sedaghati
a,,1
, E. Esmailzadeh
b,2
a
Department of Mechanical and Industrial Engineering, Concordia University, 1455 de Maisonneuve Blvd. West,
Montreal, Quebec, Canada H3G 1M8
b
University of Ontario Institute of Technology, School of Manufacturing Engineering, Oshawa, Ontario,
Canada L1H 7K4
Received 13 August 2002; received in revised form 18 September 2003; accepted 27 September 2003
Abstract
A new structural analysis and optimization algorithm is developed to determine the minimum weight of
structures with the truss and beam-type members under displacement and stress constraints. The algorithm
combines the mathematical programming based on the sequential quadratic programming (SQP) technique
and the nite element technique based on the integrated force method. The equilibrium matrix is generated
automatically through the nite element analysis while the compatibility matrix is obtained directly using
the displacementdeformation relations and the single value decomposition (SVD) technique. By combining
the equilibrium and compatibility matrices with the forcedisplacement relations, the equations of equilibrium
with the element forces as variables are obtained. The proposed method is extremely ecient to analyze and
optimize the truss and beam structures under stress and displacement constraints. The computational eort
required by the force method is found to be signicantly lower than that of the displacement method. The
eect of the geometric nonlinearity in the structural optimization problems under the stress and displacement
constraints were also investigated and it is illustrated that the geometric nonlinearity is not an important issue
in these types of problems and hence, it does not aect the nal optimum solution signicantly. Four examples
illustrate the procedure and allow the results to be compared with those reported in the literatures.
? 2003 Elsevier Ltd. All rights reserved.
Keywords: Optimum structural design; Stress constraints; Finite element force method; Size optimization; Displacement
constraints
Corresponding author. Fax: +1-514-848-8635.

E-mail addresses: sedagha@alcor.concordia.ca (R. Sedaghati), ebrahim.esmailzadeh@uoit.ca (E. Esmailzadeh).
1
Mem. CSME, Mem. ASME.
2
Fellow ASME, Fellow I.Mech.E., Mem. CSME, Mem. SAE.
0020-7403/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmecsci.2003.10.001
1370 R. Sedaghati, E. Esmailzadeh / International Journal of Mechanical Sciences 45 (2003) 13691389
1. Introduction
The concept of equilibrium of forces and compatibility of deformations are fundamental to the
analysis methods in structural mechanics. The underlying principle behind the equilibrium equations
is the force balance. In general, equilibrium equations are not adequate enough in solving the struc-
tural analysis problems and need to be augmented by the compatibility conditions. In other words,
equilibrium equations are indeterminate by nature, and determinacy is hence achieved by inclusion of
the compatibility conditions. Two analytical approaches, namely the displacement and force methods,
have been developed to analyze the determinate or indeterminate structures.
Structural analysis and optimization algorithms developed in recent years have generally been based
on the displacement method [16]. Although, this method is an ecient approach for the stress
displacement type analysis, but it presents some disadvantages in the optimization problems when the
number of stress constraints are larger than the displacement constraints. On the other hand, the force
method has not yet been very popular among researchers for the structural optimization problems.
This is due to the fact that the redundancy analysis required in the force method has not been easily
amenable to the computer automation. However, for the analysis of a not highly redundant structure
in which, the number of redundant elements is lower than the displacement degrees of freedom, or
for a determinate structure the force method is computationally more ecient than the displacement
method.
In the classical form of the force method, it is quite dicult to generate the compatibility con-
ditions. Splitting up the given structure into a determinate based structure and redundant members
would generate the compatibility in the classical force method and the compatibility conditions are
stated by establishing the continuity of deformations between the redundant members and the basis
structure. Navier [7] originally developed this procedure for the analysis of indeterminate trusses.
Prior to the 1960s, the basis structure and redundant members were generated manually. In the
post-1960s, several schemes have been devised to automatically generate the redundant members
and the basis determinate structure [8,9], but with a limited success. In the integrated force method,
developed by Patnaik [10,11], both the equilibrium equations and the compatibility conditions are
considered simultaneously. The generation of the compatibility equations is based on extending the
St. Venants theory of elasticity strain formulation to the discrete structural mechanics by elimination
of the displacement terms in the deformationdisplacement relationship [12].
In the present study, the linear analysis based on the integrated force method has been used to
analyze and optimize the truss and beam structures under both stress and displacement constraints and
the results were compared with those obtained from the geometrically nonlinear analysis. Moreover,
it is intended to investigate the eciency of the force method in the structural optimization of the
truss and beam structures, under displacements and stress constraints, by solving the equilibrium
and compatibility equations simultaneously. A direct method has been developed to generate the
compatibility matrix for indeterminate structures, which, is based on the displacementdeformation
relationship and the singular value decomposition (SVD) technique without the need to select the
consistent redundant members. The equilibrium matrix is also generated automatically through the
nite element analysis.
In most recent works, reported in literatures, the optimization algorithms were mainly based on
the optimality criterion technique because of its computational eciency. For example the optimality
criteria method has been employed to minimize the weight of the truss and beam structures under
R. Sedaghati, E. Esmailzadeh / International Journal of Mechanical Sciences 45 (2003) 13691389 1371
the stress and displacement constraints [1316], stability constraint [17], or frequency constraint [18].
A number of the optimality criterion algorithms are based on satisfying one major critical constraint
in order to avoid the scaling and calculating a large number of Lagrange multipliers. Modern opti-
mality criterion algorithms would involve the case of satisfying the multiple constraints (scaling) and
KarushKuhnTuker (KKT) condition (resizing) alternatively. However when the cross-sectional area
and principal moment of inertia are nonlinearly related (frame structures), the scaling procedures,
normally used in the optimality criterion methods, are approximate in nature and the scaling itself
needs an iteration procedure. Considering these facts the powerful nonlinear mathematical program-
ming method, based on the sequential quadratic programming (SQP) technique, has been utilized as
the optimization algorithm in this study to nd the true optimum solution.
The application and eciency of the proposed methodology is illustrated by minimizing the weight
of the truss and beam structures. It is shown that by using either the force or displacement method, as
an analyzer does not aect the nal optimum solutions of the problems with stress and displacement
constraints. However, the force method is more computationally ecient than the displacement one.
It is also illustrated that in the practical optimization problems, with stress and displacement con-
straints, the nonlinear analysis does not aect the nal optimum solution signicantly. Moreover, it
is found that using the SQP method as the optimizer could lead into a lighter design when compared
with the conventional optimality criterion technique mostly reported in literatures.
In the following sections, a short description of the structural analysis using the force method and
the geometrically nonlinear nite element analysis is presented and the size optimization algorithm
is fully explained. Finally, the application of the algorithm is illustrated by optimizing four dierent
truss and beam-type structures under the stress and displacement constraints.
2. Structural analysis using the force method
A discrete nite element structure can be designated as structure (d, [), where d and [ are the
displacement and force degrees of freedom, respectively. The structure (d, [) has d equilibrium
equations and r = ([ d) compatibility conditions. In static problems the equilibrium equations in
the displacement formulation can be written as
KU = P, (1)
where K is the system stiness matrix of the structure (obtained by assembling the stiness matrices
of the individual elements), U is the nodal displacement vector and P is the external applied load
vector. The compatibility conditions have been satised implicitly during the generation of Eq. (1).
The equivalent form of Eq. (1) in the integrated force formulation can be written as [10,11]:
SF = P
, (2)
where F is the element force vector. The matrix S and vector P
can be obtained through the

combination of the equilibrium matrix as
QF = P (3)
and the compatibility equations can be written as
CA = 0, (4)
where the element deformation vector can be related to the element force vector F in
accordance to
A =GF, (5)
thus
S =
Q

C G
, P
P

0
, (6)
where Q, C and G are the (d [) equilibrium matrix, (r [) compatibility matrix and the
([ [) exibility matrix, respectively. One should note that the matrices Q, C and G are banded
and they have full-row ranks of d, r and [, respectively. The matrices Q and C also depend
on the geometry of the structure and therefore, are independent of the material properties. For a
nite element idealization, the generation of the equilibrium matrix Q and the exibility matrix
G is straightforward and can be obtained automatically. However the automatic generation of the
compatibility matrix C is a laborious task in the standard force method. Moreover, the generation
of C in the integrated force method is based on the elimination of the d displacement degrees of
freedom from the [ elemental deformations. In this study, an ecient method is proposed to derive
the compatibility matrix directly. The method is based on the displacementdeformation relations
and the SVD.
The displacementdeformation relationship for the discrete structures can be obtained by equating
the internal strain energy and the external work as
1
2
P
T
U =
1
2
F
T
A. (7)
By substituting P from Eq. (3) into Eq. (7), we can obtain
1
2
F
T
Q
T
U =
1
2
F
T
A or F
T
(Q
T
U A) = 0. (8)
Since the element force vector F is not a null space, we are able to determine the following
relationship between the member deformation vector and the nodal displacement vector as
A =Q
T
U. (9)
Eq. (9) relates the [ deformations to the d nodal displacement degrees of freedom and hence,
the r = ([ d) compatibility equations can be arrived through the elimination of the d nodal
displacements from the [ deformations. To obtain the compatibility matrix, one may express the
nodal displacements in terms of the member deformations by using Eq. (9) as
U = (QQ
T
)
1
Q A = (Q
T
)
int
A, (10)
where the matrix (Q
T
)
int
denotes the MoorePenrose pseudo-inverse of Q
T
. Considering Eq. (9)
and (10), we may have
[I Q
T
(Q
T
)
int
]A = 0 (11)
or in short AA = 0, (12)
where A = [I Q
T
(Q
T
)
int
] (13)
Eq. (12) is similar to the compatibility equations given by Eq. (4). However, the matrix A is a
([ [) matrix with rank of r. It illustrates that the rows of matrix A are dependent on each other.
In order to extract the (r [) compatibility matrix C form the matrix A, i.e. to reduce the matrix
A to matrix C, the singular value decomposition (SVD) method is used [19]. By applying the SVD
method to A, we obtain
A =R2T
T
, (14)
where R and T are the ([ [) orthogonal matrices and
2 =
0
0 0
([[)
(15)
with = diag{o
1
o
2
o
r
}, and o
1
o
2
o
r
0. It follows that
A =R
C
0
. (16)
Therefore the (r [) compatibility matrix C can be represented by
C = [T
1
T
2
T
i
T
r
]
T
, (17)
where the vector T
i
denotes the ith column of the matrix T.
Although Eq. (9) is quite adequate to determine the element deformations using the nodal dis-
placements, but it is not sucient to obtain the nodal displacements using the element deformations
or forces since the redundant structures are represented by the rectangular equilibrium matrix Q
with no inverse. This implies that the compatibility equations should be merged with the equilibrium
equations. For this reason, using S instead of Q in Eq. (9) and solving for the nodal displacements
U, we obtain
U =JA or U =J G F, (18)
where
J = drows of S
T
. (19)
3. Nonlinear analysis
It is important that the eect of the geometrical nonlinearity on the nal optimum solution must
be fully investigated. In most cases, although the values of strain are quite small and the material
behaves linearly, the response of the structure, as a result of nite rotations and displacements, would
become nonlinear. It is therefore, necessary to express the joint equilibrium in terms of the nal
geometry of the structure. In the case of large displacements, the strain-displacement relationships
contain the nonlinear terms. Consideration of these terms and by using the principle of virtual work,
the system stiness matrix can be evaluated in the form of
K =K
E
+K
G
, (20)
where K represents the system tangent stiness matrix, K
E
is the system linear elastic stiness ma-
trix and K
G
being the system geometric stiness matrix. It is worth noticing that the matrix K
G
is
associated with the changes in the geometry of the structure. These matrices are obtained by the
assemblage of the element linear elastic and the geometric stiness matrices in the global coordi-
nates. Linear and geometry stiness matrices for the truss and beam elements are well documented
in most nite element books [20,21].
To derive the incremental nite element equations, it is assumed that the equilibrium conguration
at a load level is known and the conguration at a slightly higher load level is to be determined.
Using the NewtonRaphson method and updated Lagrangian formulation [21], these equations may
be written as
t+t
K
(k1)
U
(k)
= P
(k1)
=
t+t
:
(k)
P
re[

t+t
P
(k1)
t+t
U
(k)
=
t+t
U
(k1)
+ U
(k)
(21)
where k is the iteration number,
t+t
K is the tangent stiness matrix at time step t +t,
t+t
P
(k1)
is the vector of the nodal resultant member forces at time step t +t, and P
re[
is a given reference
load. Furthermore,
t+t
: is a load factor parameter to denote the external load at time step t + t,
P is the out-of-balance force, U is the vector of increments in the nodal displacements, and
t+t
U is the vector of the nodal displacement at time step t + t. The out-of-balance load vector
P corresponds to a load vector that is not yet balanced by the element forces, and hence an
increment in the nodal displacements is required. This updating of the nodal displacements in the
iteration will be continued until the out-of-balance loads and the incremental displacements are small.
In order to guarantee that both the out-of-balance loads and the incremental displacements are small,
the energy convergence criteria [21] (being a product of the out-of-balance force and the incremental
displacement), has been employed. For the purpose of the present analysis an energy convergence
tolerance c
E
= 10
6
has been selected.
It is noted that any increment in either the load or displacement is conventionally represented as
an evolution in time t. One should note that the problem is pure static and t simply denotes the
incremental steps in the solution.
4. Optimization algorithm
The optimization problem can be dened mathematically as minimizing the structural mass:
Min
M(A) =
n
i=1
j
i
L
i
A
i
(22)
subject to the (n + m) stress and the displacement constraints (behavior constraints)
q
)
(A) = |o
i
} o
i
| 1 60, ) = 1, . . . , n,
q
)
(A) = |U
)
}

U
)
| 1 60, ) = 1, . . . , m (23)
and the n side constraints on the design variables
A
i
A
i
60, i = 1, 2, 3, . . . , n, (24)
where j
i
, A
i
and L
i
are the density, the cross-sectional area and the length of the ith element,
respectively. M is the total mass of the structure and

A
i
is the lower limit on the ith design variable.
Furthermore, o
)
, o
)
are the stresses in the )th and its allowable limit value, respectively, and U
)
,

U
)
are the respective )th constrained displacement and its allowable limit value. It should be pointed
out that the analysis and optimization are two separate modules. The equilibrium and compatibility
equations in the force method, Eq. (2), and the iterative equilibrium equations, Eq. (21) are satised
directly during the nite element analysis (analysis module) and then the results are passed to the
optimization module. Thus it is not necessary to take into account Eq. (2) or Eq. (21) as the equality
constraints inside the optimization algorithm.
In this study, the sequential quadratic programming (SQP) method has been applied to solve the
optimization problem discussed in above. The implementation of the SQP method was performed in
MATLAB [22]. Based on the work done by Powell [23], the method allows one to closely mimic the
Newtons method for the constraint optimization just as it is done for the unconstraint optimization.
SQP is indirectly based on the solution of the KKT conditions.
It must be noted that the stress and displacement gradient functions, in Eq. (23), are not both
smooth and convex functions, thus the local optimum result may be achieved using the gradient-based
algorithms such as the SQP algorithm. In this study, several randomly generated initial points have
been used for the SQP algorithm in order to make sure that the optimal solution is either a global
solution or very close to the global solution.
5. Illustrative examples
Four examples on the analysis of the space truss and frame structures, presented in this section,
illustrate the proposed procedure and allow the results to be compared with those reported in lit-
eratures. It is intended to show that in the structural optimization problems, with the stress and
displacement constraints, the linear analysis procedure (either the force method or the displacement
one) does not aect the nal optimum design. Furthermore, it is to establish the fact that the design
optimization procedure based on the force method is more ecient than the displacement method
and that the geometrical nonlinearity is not an important issue in the truss- and beam-type structures.
5.1. The 25-bar space truss
The 25-bar space truss structure, shown in Fig. 1, has previously been investigated by Saka using
the linear analysis based on the displacement method and the geometrical nonlinear analysis with
the optimality criterion approach [13,14].
The structure has identical symmetries about the XZ and YZ planes, so that the design vari-
able linking is used to impose symmetry on the structure and hence, only eight design vari-
ables are identied here. The material selected is steel with the Youngs modulus of elasticity of
E=207 kN}mm
2
(30, 000, 000 psi) and the mass density of j=7830 kg}m
3
(0.283 lbm}in
3
) also the
structure is subjected to single load case as shown in Table 1. The allowable tensile stress for all the
Fig. 1. The 25-bar space truss structure.
Table 1
Nodal load components (N) for the 25-bar space truss structure
Node Coordinate directions
X Y Z
1 80,000 120,000 30,000
2 60,000 100,000 30,000
3 30,000 0 0
6 30,000 0 0
elements is set as o
at
= 240 MPa (34809 psi) while the allowable compressive stress is determined
in accordance to the AISC codes [24]. It indicates that o
ac
=
2
E}S
2
R
is for the case of S
R
C
c
and
the value of o
ac
=o
at
(1 0.5S
2
R
}C
2
c
) is for S
R
C
c
. The slender ratio of each member is S
R
=L}R
G
,
Table 2
Final design solutions for the cross-sectional areas (mm
2
) for the 25-bar space truss structure
Design variables Member Linear analysis Nonlinear analysis
DM FM
1 1 232.7 232.7 233.0
2 25 1150.6 1150.6 1149.8
3 69 895.1 895.1 894.0
4 10,11 230.4 230.4 235.2
5 12,13 223.3 223.3 227.1
6 1417 1018.4 1018.4 1023.8
7 1821 950.2 950.2 954.4
8 2225 1443.5 1443.5 1440.4
Mass (kg) 649.7 649.7 650.9
No. of iterations 316 424 380
No. of A.C.
a
8 8 8
CPU time (s) 50.64 16.31 209.77
a
Number of active constraints.
where L is the length, R
G
is the radius of gyration for each member and C
c
=
2
2
E}o
at
. Thus, the
value of the allowable compressive stress varies during the optimization process. Stress constraints
have been imposed on all the elements and the displacement constraints of 10 mm have been
imposed on the nodes 1 and 2 in the X- and Y-directions. The minimum cross-sectional area for all
the elements was set at 200 mm
2
. All the members have the pipe-type cross-sections with S
R
=aA
b
,
where A is the area in square centimeter and the constants a and b are selected as 0.4993 and
0.6777, respectively. The number of degrees of freedom for the displacement is m=18, and that of
the force is n = 25. Therefore, the number of the redundancy is found to be r = 7. Without linking
the design variables, the number of the design variables is 25 and, the number of the constraints is
54. On the other hand, by linking the design-variables into eight groups, the number of the design
variables reduces to 8, and the number of the constraints would change to 20.
For the linear analysis, a minimum value of the mass of 649.7 kg (1432.3 lbm) was obtained
using both the displacement method (DM) and the force method (FM). However, when the nonlinear
analysis is carried out the minimum value of the mass was obtained as 650.9 kg (1435 lbm). The
nal results for both the linear and the nonlinear analysis are presented in Table 2 with their
iteration histories illustrated in Fig. 2. The initial cross-sectional area for all the elements is chosen
as 2000 mm
2
(3.1 in
2
). The CPU time required for the FM is signicantly lower than that of the DM
indicating its superior eciency. For the linear and the nonlinear analyses, the compressive stress
constraint in the elements 1, 2, 6, 10, 13, 16, 18 and 24 (one member from each group was selected)
are found to be active. According to Table 2 there is a slight but insignicant dierence between
the optimum results obtained from the linear and the geometrical nonlinear analyses. Nevertheless,
to conrm the feasibility of the linear results, the structure was also analyzed using the nonlinear
analysis based on the optimum cross-sectional areas. It is found that very little stress violations were
occurred in the elements 10, 13, 16 and 18, and since these members are under compression the
optimum results obtained from a linear analysis could result into structural failure.
0 50 100 150 200 250 300 350 400 450
600
700
800
900
1000
1100
1200
1300
1400
Iteration
M
a
s
s

(
K
g
)
Linear analysis (DM)
Linear analysis (FM)
Nonlinear analysis
Fig. 2. Iteration histories for the 25-bar space truss for the linear and nonlinear analysis solutions.
0 200 400 600 800 1000
200
400
600
800
1000
1200
1400
Iteartion
M
a
s
s

(
k
g
)
Initial areas=2000 mm
2
; Feasible guess
Initial areas=500 mm
2
; Infeasible guess
Fig. 3. Iteration histories for dierent initial areas for the 25-bar space truss.
The problem was solved, using dierent initial values of the cross-sectional areas for all the
elements, and results were found to be exactly the same as for the previous case. The iteration
histories for two dierent initial values of the cross-sectional area, using the force method, is shown
in Fig. 3.
A minimum value of the weight of, 921 kg (2030.5 lbm) was obtained by Saka, using the linear
analysis, and of 507 kg (1117.7 lbm) for the nonlinear analysis with the optimality criterion based
only on satisfying the displacement constraints [13,14]. The stress constraints were satised through
the stress-ratio technique in the linear analysis. In the nonlinear analysis, the stress constraints were
satised during the steps of nonlinear analysis, so that when a member force exceeds its allowable
limit, then this limit value is used in the computation of the compensating load. In accordance to
the results found by Saka and the outcome of such redistribution, the truss members do not support
a force, which is more than its critical value.
Flurry and Schmit [5] also solved this problem by using the dual method and the approximation
concept and considering only the linear analysis based on the displacement method. This structure
was also analyzed by the authors using the data provided by them for which, identical results were
obtained.
5.2. The 72-bar space truss
The 72-bar space truss structure, shown in Fig. 4, is a relatively large size problem with the
material chosen as aluminum with the Youngs modulus of elasticity E =69 kN}mm
2
(10
7
psi) and
the mass density j = 2768 kg}m
3
(0.1 lbm}in
3
). The allowable stress for all the members is set as
172.37 MPa (25000 psi) and the stress constraints have been imposed on all the elements. The
displacement constraints of 6.35 mm (0.25 in) are imposed on both the X- and Y-directions for
the nodes 14 and the minimum value of the cross-sectional area for all the elements was set at
64.52 mm
2
(0.1 in
2
).
The structure is subjected to two dierent loading inputs. In the rst case, only the node 1 is
subjected to a pull load of 22.25 kN (5000 lbf ) in the X and Y-directions and a push load of
22.25 kN (5000 lbf ) in the Z-direction. However, for the second case, all the nodes of 14 are
subjected to a push load of 22.25 kN (5000 lbf ) in the Z-direction. The number of displacement
degrees of freedom is m=48, and the number of the force degrees of freedom is n=72, and hence, the
number of the redundancy will be r =24. Without linking together the design variables, the number
of the design variables is 72 and the number of the constraints is found to be 152. However, by
linking the design variables into 16 groups, the number of the design variables becomes 16, and the
associated number of the constraints reduces to 40.
Using the linear analysis, the minimum value of the mass, for both the DM and the FM, was
obtained as 172.20 kg (379.615 lbm). When the same problem was analyzed, using the nonlinear
analysis, the minimum value of the mass was increased slightly to 172.400 kg (380.079 lbm). The
nal results for both the linear and the nonlinear analysis are presented in Table 3 and their itera-
tion histories are illustrated in Fig. 5. The nonlinear analysis steps, required to obtain the response,
was found to be 2 for the rst load case, and 3 steps for the second load case in all the itera-
tions. The required computational time for the FM is signicantly lower than the DM, illustrating
the eciency of the FM over the DM for the analysis of large size problems. A closer examina-
tion of the results reveals that in both the linear and the nonlinear analysis the nodal displacement
constraints at node 1 in the X- and Y-directions for the rst load case, and the stress constraints
in the elements 1 to 4 for the second load case are active. The cross-sectional areas in groups 7,
8, 11, 12, 15 and 16 reached their minima in both analyses. The linear analysis solution matches
exactly with the solution reported by Flurry and Schmit [5], who solved the problem using the dual
Fig. 4. The 72-bar space truss structure.
method and the approximation concepts in a linear analysis, which is based on the displacement
method.
It is noted that there is a very slight dierence, although insignicant, between the optimum
results obtained from the linear and the nonlinear analysis. To conrm the validity of the linear
solution, the structure was analyzed considering the geometrical nonlinearities based on the optimum
Table 3
2
) for the 72-bar space truss structure
Design variable Members Linear analysis Nonlinear analysis
DM FM
1 14 100.97 100.97 101.29
2 512 352.00 352.00 352.39
3 1316 264.77 264.77 265.42
4 17,18 367.55 367.55 368.58
5 1922 337.87 337.87 338.45
6 2330 333.61 333.61 334.06
7 3134 64.52 64.516 64.52
8 35,36 64.52 64.516 64.52
9 3740 818.32 818.32 819.68
10 4148 330.13 330.13 330.32
11 4952 64.52 64.52 64.52
12 53,54 64.52 64.52 64.52
13 5558 1216.90 1216.90 1218.70
14 5966 330.52 330.52 330.71
15 6770 64.52 64.52 64.52
16 71,72 64.52 64.52 64.52
Mass (kg) 172.20 172.20 172.40
No. of A. C. 9 9 9
CPU time (s) 274.23 107.10 1427.6
0 100 200 300 400 500 600
300
400
500
600
700
800
900
Iteration
M
a
s
s

(
x

0
.
4
5

K
g
)
Nonlinear analysis
Fig. 5. Iteration histories for the 72-bar space truss for both the linear and nonlinear solutions.
0 1000 2000 3000 4000 5000
200
300
400
500
600
700
800
900
Iteartion
M
a
s
s

(
x

0
.
4
5

K
g
)
Fig. 6. Iteration histories for the 72-bar space truss with no variables linking.
cross-sectional areas obtained from the linear analysis. Surprisingly enough, both the displacement
constraints at node 1 and the stress constraints in the elements 14 for the second load case were
slightly violated.
Furthermore, the problem was once again solved when the design variables were not linked to-
gether. The results using the FM and the DM were found to be almost identical and the computational
time for the FM was approximately one half of that required by the DM. The optimum value of
the mass was reduced to 288.8 lbm, demonstrating that when the symmetry is not imposed on the
structure a signicantly lower value of the mass can be obtained for the nal design. In this case
the number of active constraints was 47. The displacement constraints at node 1 in both X- and
Y-directions for the rst load case, and the stress constraints in the members 1, 2, 4 and 19 for the
second load case were found to be active. The cross-sectional areas in the elements 5, 8, 9, 1216,
18, 24, 25, 28, 29, 3136, 38, 40, 41, 4854 and 56 were reached to their minima and the iteration
histories for this case is illustrated in Fig. 6.
In order to better understand the eect of the geometrical nonlinearity, the load in all directions
and for both load cases was increased by a multiple of 100 with an increase in the displacement
constraints of 101.6 mm (4 in), simultaneously. The nal results are presented in Table 4 and the
iteration histories are shown in Fig. 7. It is worth noting that the eect of the geometrical nonlinearity
is more pronounced at the nal optimum design and the constraint displacements are quite far from
the limiting value, implying that the design is only controlled by the stress constraints. Moreover,
when using the linear analysis, the stress constraints in members 6, 11, 13, 16, 17, 23, 30, 39, 42,
47, 57, 59 and 66, for the rst load case, and in members 14 and 1922, for the second load
case, are found to be active. However, in the case of nonlinear analysis for the rst load case the
situation is similar to that of the linear analysis, but for the second load case, the stress constraints
in the members 3 and 21 are found to be active.
Table 4
2
) for the 72-bar space truss with increasing load and displacement
constraints
Design variable Members Linear analysis Nonlinear analysis
DM FM
1 14 12,727.00 12,727.00 12,787.00
2 512 5014.60 5014.60 5068.60
3 1316 4412.20 4412.20 4480.50
4 17,18 5664.90 5664.90 5699.70
5 1922 12,911.00 12,911.00 13,168.00
6 2330 4575.30 4575.30 4603.90
7 3134 64.52 64.52 64.52
8 35,36 536.84 536.84 624.51
9 3740 13,865.00 13,865.00 13,998.00
10 4148 4513.50 4513.50 4518.60
11 4952 64.52 64.52 64.52
12 53,54 64.52 64.52 64.52
13 5558 19,153.00 19,153.00 19,322.00
14 5966 4484.20 4484.20 4508.80
15 6770 64.52 64.52 64.52
16 71,72 64.52 64.52 64.52
Mass (kg) 2698.87 2698.87 2723.03
No. of A. C. 15 15 15
CPU time (s) 100.92 46.67 1103.6
0 50 100 150 200 250 300
0
1000
2000
3000
4000
5000
6000
7000
Iteration
M
a
s
s

(
x

0
.
4
5

K
g
)
Nonlinear analysis
Fig. 7. Iteration histories for the 72-bar space truss with increasing load and displacement constraints.
Fig. 8. The 10-member frame structure.
Finally, it may be inferred that for the practical truss design problems, which are optimized for
the size under the stress and displacement constraints, the eect of the geometrical nonlinearity is
not signicant and hence, the linear analysis can provide acceptable solutions. Moreover, for the
test cases studied the nonlinear analysis based on the geometrical nonlinearity does not necessarily
produce a better optimal solution (for lighter structures) when compared to that of the linear analysis.
Nevertheless, an optimized structure based on a linear solution may fail when the values of the stress
at some members would reach beyond the allowable design limit, as can be seen in some of the
test cases presented.
5.3. The 10-member frame (two-story and two-bay)
The 10-member frame structure consists of three stories and two bays, illustrated in Fig. 8, is
made of steel with E = 207 kN}mm
2
(30, 000, 000 psi) and j = 7830 kg}m
3
(0.283 lbm}in
3
). The
stress limit for all the frame members is 165.47 MPa (24000 psi). The horizontal displacements for
all the joints were limited to 0.254 mm (0.01 in), while a minimum cross-sectional area limit of
3225.80 mm
2
(5 in
2
) and a maximum area limit of 64, 516 mm
2
(100 in
2
) were enforced. The initial
cross-sectional area was set as 16, 129 mm
2
(25 in
2
) that being the same for all the elements. The
following empirical relationships were used for the area A, the section modulus S, and the moment
of inertia I [15].
S = 1.6634 A
1.511
I = 4.592 A
2
0 6A615
S = (281.077 A
2
+ 84100)
0.5
290
I = 4.638 A
2
15 A644
S = 13.761 A 103.906
I = 256.229 A 2300
44 A6100,
where A is the area measured in square inches. The above relationship is stated for a steel section
in accordance to the AISC code.
A minimum weight of 3307.23 kg (7291.19 lbm) is obtained with a linear analysis, but with the
use of nonlinear analysis it will increase to 3309.9 kg (7297.08 lbm). The horizontal displacement
constraint at node 4 and the stress constraints on element 6 are identied as active. The horizontal
displacement at node 3 is close to being active and the cross-sectional area of the elements 3, 4 and
10 reached their minimum size. The results were compared with those reported in the literature. As
an example, Khan [15,16] used a displacement based linear analysis with the optimality criterion
technique and has obtained a minimum value of the weight 3391.87 kg (7477.79 lbm) with the
horizontal displacement of the nodes 3 and 4 being active constraints (no active stress constraint).
This problem has also been solved using the CONMIN code and a minimum value of the weight
3969.97 kg (8752.29 lbm) is reported [25].
Table 5
2
) for the 10-member frame structure
Members Linear analysis Nonlinear analysis
DM FM
1 28,387 28,387 28,387
2 23,682 23,682 23,720
3 3226 3226 3226
4 3226 3226 3226
5 46,255 46,255 46,271
6 10,241 10,241 10,238
7 7236 7236 7251
8 16,433 16,433 16,467
9 16,243 16,243 16,266
10 3226 3226 3226
Mass (kg) 3307.23 3307.23 3309.9
No. of A. C. 5 5 5
CPU time (s) 63.52 20.60 144.39
Fig. 9. The 25-member frame structure.
The computational time for the force method was signicantly lower than that required by the
displacement method, again pointing out the eciency of the force method when applying the force
method to the frame-type structures. There is a very small discrepancy between the linear and
nonlinear solutions, however insignicant. Nevertheless, to conrm the validity of the linear analysis
results, the structure was simulated by a nonlinear analysis using the optimum cross-sectional areas
obtained via the linear analysis. It was observed that the stress constraint in element 6 is a little over
active, as shown in Table 5, pointing out that the linear analysis have produced acceptable results.
5.4. The 25-member frame (three-story and three-bay)
The 25-member frame structure, shown in Fig. 9, corresponds to a three stories and three bays
structure. The numerical values of the material properties and the stress limit and the relationship
between the cross sectional area, section modulus and the moment of inertia are all the same as
those mentioned in Section 5.3. The values of the horizontal displacements for the nodes 1, 2,
3, 10, 11 and 12 are limited to 0.127 mm (0.005 in) and a minimum value of the area limit of
3225.80 mm
2
(5 in
2
) and a maximum area limit of 64, 516 mm
2
(100 in
2
) have been dened.
Table 6
Final design results for the cross-sectional areas (mm
2
) for the 25-member frame structure
Members Linear analysis Nonlinear analysis
DM FM
1 10,007 10,007 9483
2 7146 7147 6916
3 4233 4233 3952
4 16,210 16,210 16,730
5 20,247 20,246 20,307
6 3226 3226 3226
7 51,757 51,749 52,280
8 14,878 14,885 14,155
9 31,429 31,429 31,878
10 53,870 53,862 53,237
11 64,516 64,516 64,516
12 20,581 20,581 20,443
13 3226 3226 3226
14 19,581 19,581 19,514
15 3226 3226 3747
16 4826 4832 5067
17 13,153 13,154 12,898
18 3226 3226 3226
19 16,165 16,168 15,072
20 3226 3226 3226
21 18,751 18,754 17,297
22 13,546 13,556 13,765
23 46,754 46,739 47,967
24 3226 3226 3226
25 3226 3226 3226
Mass (kg) 9508.32 9508.32 9487.26
No. of A. C. 18 18 17
CPU time (s) 479.93 299.50 1687.90
The minimum value of the weight obtained using the linear and nonlinear analysis is 9508.32
(20962.26 lbm) and 9487.26 kg (20915.83 lbm), respectively. The horizontal displacement constraints
at the nodes 2 and 10, as well as the stress constraints on the elements 1, 2, 3, 5, 9, 12, 14 and 17,
are active both in the linear and nonlinear analysis. The cross-sectional areas for the frame elements
6, 11, 13, 15, 18, 20, 24 and 25 reached their minimum size. This problem was also solved by
Khan [15] using the displacement based linear analysis and the optimality criterion technique, having
obtained a minimum weight of 10049.77 kg (22155.95 lbm) with just the horizontal displacement at
nodes 2 and 11 being active (no active stress constraint). Once again results obtained from the nite
element force method, performed in this study, indicate a superior advantage over the ones obtained
from the displacement method, as illustrated in Table 6.
6. Conclusions
A structural optimization algorithm, using integrated force formulation technique, has been de-
veloped to minimize the weight of truss- and frame-type structures under the stress and displace-
ment constraints. The required compatibility matrix in the formulation has been derived directly
by utilizing a displacementdeformation relationship and the single value decomposition technique.
Moreover, the sequential quadratic programming method has been adopted to optimize the truss and
frame structures.
The main objective of this study is to investigate the relative performance of the force and
displacement methods in the design and optimization of dierent space structures with the stress
and displacement constraints. It is found that the optimization technique that is based on the force
method is computationally far more ecient than the displacement one. Moreover, it is shown that
the nonlinear analysis does not signicantly aect the optimum solution of these types of problems.
Last but not least, from the results obtained from four dierent examples, it is demonstrated that
the sequential quadratic programming method could result into a lighter optimal design of space
structures when compared to the conventionally used optimality criterion techniques. The proposed
methodology has proved to be extremely ecient in the analysis and optimization of the truss- and
frame-type space structures.
References
[1] Venkayya VB. Structural optimization: a review and some recommendations. International Journal of Numerical
Methods in Engineering 1978;13:20328.
[2] Caneld RA, Grandhi RV, Venkayya VB. Optimum design of structures with multiple constraints. AIAA Journal
1988;26:7885.
[3] Mohr GA. Finite elements for solids, uids, and optimization. Oxford: Oxford University Press; 1992.
[4] Mohr GA. Finite element optimization of structuresI. Computers & Structures 1994;53:121720.
[5] Flurry C, Schmit Jr LA. Dual methods and approximation concepts in structural synthesis. 1980, NASA CR-3226.
[6] Haftka RT, Gurdal Z. Elements of structural optimization, Third ed. Dordrecht: Kluwer Academic Publishers; 1992.
[7] Timoshenko S. History of strength of material. New York: McGraw-Hill; 1953.
[8] Robinson J. Automatic selection of redundancies in the matrix force method: the rank technique. Canadian Aero
Space Journal 1965;11:912.
[9] Kaneko L, Lawo M, Thierauf G. On computational procedures for the force method. International Journal of
Numerical Methods in Engineering 1982;18:146995.
[10] Patnaik SN. The integrated force method versus the standard force method. Computers & Structures 1986;22(2):
15163.
[11] Patnaik SN, Berke L, Gallagher RH. Integrated force method versus displacement method for nite element analysis.
Computers & Structures 1991;38(4):377407.
[12] Patnaik SN, Joseph KT. Generation of the compatibility matrix in the integrated force method. Computer Methods
in Applied Mechanics and Engineering 1986;55:23957.
[13] Saka MP. Optimum design of space trusses with buckling constraints. Proceedings of the Third International
Conference on Space Structures, University of Surrey, Guildford, UK, 1984.
[14] Saka MP, Ulker M. Optimum design of geometrically nonlinear space trusses. Computers & Structures 1992;42:
28999.
[15] Khan MR. Optimality criterion techniques applied to frames having general cross-sectional relationships. AIAA
Journal 1984;22(5):66976.
[16] Khan MR, Willmert KD, Thornton WA. A new optimality criterion method for large scale structures. Proceedings
of AIAA/ASME 19th SDM Conference, Bethesda, USA, 1978, p. 4758.
[17] Sedaghati R, Tabarrok B. Optimum design of truss structures undergoing large deections subject to a system stability
constraint. International Journal of Numerical Methods in Engineering 2000;48(3):42134.
[18] Grandhi RV, Venkayya VB. Structural optimization with frequency constraints. AIAA Journal 1988;26:85866.
[19] Golub GH, Van Loan CF. Matrix computations, Third ed. Baltimore, London: The Johns Hopkins University Press;
1996.
[20] Cook RD, et al. Concepts and applications of nite element analysis. New York: Wiley; 2002.
[21] Bathe KJ. Finite element procedures. Englewood Clis, NJ: Prentice-Hall; 1996.
[22] Coleman T, Branch MA, Grace A. Optimization toolbox for use with Matlab, Users guides, Version 2. Natick, MA:
The MathWorks Inc.; 1999.
[23] Powell MJD. A fast algorithm for nonlinearly constrained optimization calculations. In: Watson GA, editor. Lecture
notes in mathematics and numerical analysis. Berlin: Springer; 1978; p. 630.
[24] AISC, Specication for the design, fabrications and erection of structural steel for buildings, American Institute of
Steel Constructions, Chicago, IL, USA, 1978.
[25] Vanderplaats GN. CONMINa FORTRAN program for constrained function minimization. Users Manual, NASA
TMX-G2, 1973.

Optimum Design of Structures With Stress and Displacement Constraints Using The Force Method

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Optimum Design of Structures With Stress and Displacement Constraints Using The Force Method

Enviado por

Direitos autorais:

Formatos disponíveis

Available online at www.sciencedirect.

Corresponding author. Fax: +1-514-848-8635.

can be obtained through the

Você também pode gostar