2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

Genetic Algorithm Optimization of Space Frame

Ghedan Hussein1, Nildem taysi2

1

Department of Civil Engineering, GAZIANTEP University, Turkey

Department of Civil Engineering, GAZIANTEP University, Turkey

ABSTRACT
Structural design of space frames requires appropriate form for a structure so that it can
carry the imposed loads safely and economically. Traditional approaches towards the task of
finding such forms for structures have been by the use of experimental models or by intuition
and experience. The main objective of this paper is to develop and use reliable, creative and
efficient computational tools for the linearly elastic analysis and optimum design of space frame
structures under static loads.
The use of SAP2000 can assist greatly in achieving a safe design. However,
commercially available programs are not designed as optimization tools. In this study for
optimization of multistory structures, home written MATLAB code interface program is
designed to connect SAP2000 which is known as a commercial nonlinear finite element program
and genetic algorithm optimization program.
The design algorithm obtains minimum weight frames by selecting suitable sections from
specified group list, with consideration actual design constraints like, strength, lateral
displacement, inter story drift according to Load and Resistance Factor Design (LRFD). The
improved method is tested on different two dimensional multi story moment resisting frames. It
is concluded that this method can be used as a useful tool in engineering design and
optimization.
Keywords: Optimum design, Genetic algorithm, Steel structure, SAP2000, OAPI
INTRODUCTION
The development and validation of methods for obtaining optimal steel frame designs has
merited significant attention for several decades. The objective of steel frame optimization is the

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

minimization of the cost of frame design, subject typically to strength and serviceability
constraints. The wide-flange shapes provided in the AISC steel construction manual constitute
the variable space in steel optimizations. As steel shapes do not exist on a continuous scale of
cross-sectional area, moment of inertia, or any other section parameter, frame optimization
problems are typically conducted on discrete spaces, rendering deterministic gradient-based
methods impractical. Also, as structural system response is the result of complicated interaction
between various members, steel frame optimizations are also highly nonlinear. Despite these
inherent difficulties, the development of innovative stochastic algorithms and increase in
computing capability has enabled optimal designs for large, discrete structural optimization
problems with various constraints to be obtained within reasonable computational expense.
Genetic Algorithm (GA), interfaced with a SAP2000 commercial package program, is
utilized to produce the optimal solutions. SAP2000 structure analysis program is a well-known
integrated Finite Element (FE) structural analysis tool which already used for modelling and
designing structures according to different design codes. The Open Application Programming
Interface (OAPI) in SAP2000 is a free service in some versions of it, to export and import data
files from and to SAP2000. In SAP2000 after input file being opened, SAP2000 will analyse,
save result and design all members. From one of output files, any required datas, like element
stress and joint displacements can be found to check strength and serviceability constraints [1].

Genetic Algorithm
In this paper Genetic Algorithm (GA) is used so detailed explanation is given for this
method. There are fascinating algorithms. The name came from the way in which they loosely
mimic the process of evolution of organisms, where a problem solution stands in for organisms
genetic string [2]. Features include a survival of the fittest mechanism in which potential
solutions in a population are pitted against each other, as well as recombination of solutions in
mating process and random variations.
The GA is used to solve the following problem.
To minimize
Subjected to

F( s )
g j (s) 0

sil si siu ,

j 1,......., m

i 1, 2, , n

Where, s is the vector of design variables, and F ( s ) is the objective function to be minimized.
sil and siu are the lower and upper bounds on a typical design variable si , g j(s) are the
behavioural constraints.

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

Optimization by Using GA- SAP2000 OAPI

SAP2000 structure analysis program is a well-known integrated (FE) structural analysis
tool which already used for analysing, modelling and design of structures according to different
design codes.
In this paper GA, interfaced with a SAP2000 commercial package program, is utilized to
produce the optimal solutions.SAP2000 could export or import analysis and design data with
extension data base file, Microsoft excel, text file and Microsoft access. The interaction with it
occurs through the input (*$2k) and output files (*out).
The Open Application Programming Interface (OAPI) in SAP2000 is a free service in
some versions of SAP2000, to export and import data files from and to SAP2000,which aims to
offer efficient access to the analysis and design technology of the SAP2000 structural analysis
software.[3]
In SAP2000 after input file being opened, SAP2000 will analyse, save result and design
all members. From one of output files, any required datas, like element stress and joint
displacements can be found to check strength and serviceability constraints [4] Figure 1.
However most of available commercially structural analysis programs are not designed
for optimization [5], but it will be able to achieve this task with preparing open optimization
designed code in any code have ability to interact with it and interfacing them together. The
SAP2000 API offers a broad range of programming languages that it can be used which cover
the vast majority of the modern software development options including Visual Basic.NET,
Visual Basic for Applications (VBA), Visual C#, Visual C++, Visual Fortran and Matlab. [3].

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

Figure 1. Interfacing flowchart between sap2000 and GA

Constraints in Structural Optimization

The introduction of the large-scale digital computers allowed the adaptation of classic
optimization algorithms to realistic engineering problems, as well as the advancement of new
and more powerful techniques to obtain the optimum design of structural systems. Most of them
deal continuous design variables with simple constraints. Only a few of these papers deal with
the discrete design variables and actual design constraints according to different structural design
code [6, 7, and 8], most of them used optimality criteria methods and mathematical programming
techniques with continuous design variables as an optimization tool [9].
The AISC-LRFD specification combines strength, stability and displacement
requirements. Displacement constraints are the allowable interstory drift. These constraints are
implicit constraints because structural responses like stresses, strains, and displacements are
functions of design variables [10].

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

Stress constraints according to LRFD

According to (AISC-LRFD) specification the allowable stress for members subject to bending
and axial force are [11].
For

Pu
0.2
Pn

(1)

M uy
Pu
8 M ux

Pn 9 b M nx b M ny
For

1 .0

Pu
0.2
Pn

(2)

(3)

M ux
M uy
Pu

2Pn b M nx b M ny

1 .0

(4)

In equation 2 and 4 if the axial force is in compression or in tension the terms in above equations
are Pu is the required axial strength (tension or compression); Pn nominal axial strength (tension
or compression), Mux is the required flexural strength about the minor axis, Muy is the required
flexural strength about the minor axis, Mnx is the nominal flexural strength about the major axis,
Mny is nominal flexural strength about the minor axis,
(for 2D structures, Muy is equal to zero); = t resistance factor for tension (equal to 0.90)
=c compression resistance factor and b = flexural resistance reduction factor =0.9.
The nominal compressive strength of a member is computed as

Pn Ag .Fcr

(5)

Fcr 0.658 c 2 .F y

(6)

0.877
.Fy
Fcr
c 2

(7)

Where c1.5

For c>1.5

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

KL F y
r E

(8)

In which Ag is the gross cross-sectional area of a member, K = Effective length factor for braced
and unbraced member [12].
E =modulus of elasticity of a member, r = radius of gyration, L = length of member, and F y =
yield stress of steel.
Serviceability limit states (displacement constraints)
The increasing use and reliance on probability based limit states design methods, such as
the recently adopted AISC- LRFD Specification [11], has concentrated new attention on the
serviceability problems in steel buildings. These methods, along with the development of higherstrength building materials and the use of lighter and less rigid building materials, have led to
more flexible and lightly damped structures than ever before, making serviceability problems
more prevalent.
Lateral frame movement or deflection is usually evaluated for the building as a whole,
where the applicable parameter is total building drift, which is equal to (D/ H) where D is total
top-story drift and H is the total structure height, and for each floor of building which is known
as inter-story drift, and can be defined as the lateral deflection of a floor relative to the one
immediately below it divided by the distance between floors ((dn(dn1)/ h) [13]).
Where dn is the drift of specified floor and h is equal to height of that floor.
According to the ASCE report [14], normally allowable accepted ranges for lateral
displacements are restricted between 1/7501/250 times the building heights with a typical value
of H/400 and the normally accepted limits on the inter-story drift is 1/5001/200 times the story
height with a typical value h/300; where h is the height of an story.
Fabrication constraints
The fabrication constraint is that, structure elements are available in the form of discrete
sections; otherwise the algorithm would not have any practical application.
The standard available steel sections are treated as design variables and the stress and
displacement constraints are taken from the design codes [15]. Traditionally, in design of space
moment frames, frame members (column and beams) are usually selected W-sections, so a file
with different section property is prepared, for beams and columns. For example, consider a
framed structure, where the structure is subjected to design stress, displacement and fabrication
constraints the equation of optimization problem may be expressed as

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

Minimize

W i liAi (i )

(9)

Subjected to

(u ) (l )

(10)

( d u ) d ( d l )

(11)

( Au ) A ( Al )

(12)

i 1

Where , d , A are stress, displacement and cross sectional area and subscripts u and l refer to
prescribed upper and lower boundaries of each constraints.

i
Ai (i )

Index number according to fabrication code

Is the cross sectional area of element i.

NUMERICAL EXAMPLES
5.1 Space Frame Examples
In this part 3D moment frame examples are optimized under static loads. The objective function
is weight minimization under stress and displacement constraints according to AISC-LRFD
specifications, with the 4 load combinations as shown in Table 1.
Table 1 Load combination case
Load case
Combinations
Comb1
1.4Dl
Comb2
1.2Dl+1.6Ll+0.5Rl
Comb3
1.2Dl+0.5Ll+1.6Rl
Comb4
1.2Dl+1.3 Wl+0.5Ll+0.5Rl
5.1.2 Four story space moment frame
This example deals with four-story space moment frame, which is optimized previously by
Tabu-search (TS) [15] and Simulated Annealing (AS) [16], the frame members are divided in to
10 groups. The groups were organized as follows: 1-st group: outer beams of 4-th storey, 2-nd
group: outer beams of 3-rd, 2-nd and 1-st storeys, 3-rd group: inner beams of 4-th storey, 4-th
group: inner beams of 3-rd, 2-nd and 1-st storeys, 5-th group: corner columns of 4-th storey, 6-st
group: corner columns of 3-rd, 2-nd and 1-st storeys, 7-th group: outer columns of 4-th storey, 8th group: outer columns of 3-rd, 2-nd and 1-st storeys, 9-th group: inner columns of 4-th storey,
10-th group: inner columns of 3-rd, 2-nd and 1-st storeys. The height and span lengths of the
structure are as shown in Figure 2 and 3.

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

The structure subjected to live load (Ll= 2.39 kpa) , dead load (Dl=2.78 kpa), roof live load (Rl=
2.39) kpa and wind pressure (p) = cecqqsi. where p is design wind pressure; ce is combined
height, exposure and gust factor coefficient, cq is pressure coefficient is equal to 0.8 and
0.5 for both windward and leeward faces of the structure respectively, qs is wind stagnation
pressure is equal to 0.785 kpa, and the importance factor i=1. wind loads acted on the xdirection for both wind ward and leeward sides. The example is optimized under LRFD
stress constraints and 4.55 cm and 1.52 cm for top and inter-storey drift constraints
respectively. Material properties for the frame are: youngs modulus E 200 10 6 kN/m2,
material density 7850 kg/m3, minimum yield stress fy=248.2 Mpa and modulus of
rigidity g=83 Gpa

Figure 2 Beams and columns of four

story moment frame

Figure 3 Side view, of four story space

frame

The maximum inter-story drift is 0.22 cm which is less than maximum allowable drift =1.52 cm,
and maximum total story drifts =1.4 cm which is less than the maximum allowable top-story
drift = 5.54 cm as recommended by reference [15]and [16].
Table 2 Optimum design variable of four story space frame
Group no.
1
2
3
4
5
6

Degertekin [16]
W 1631
W 1631
W 1840
W 1835
W 835
W 1453

Design variables
Degertekin et al
W
[15]1835
W 1835
W 1835
W 1835
W 831
W 1240

Present Work
W 1226
W 1636
W 1876
W 1835
W 1230
W 1626

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

7
8
9
10
Weight(kg)

W 831
W 835
W 831
W 1468
22405

W 1039
W 1245
W 828
W 1258
23105

W 1253
W 1443
W 620
W 1461
22961.2

10 story space moment frame

This example deals with 10-storey space moment frame with rectangular plane as shown in
Figure 4[9]. The structure is divided in to 9 groups. The groups are organized as follows: 1-st
group: outer beams of top storey, 2-nd group: inner beam of top storey, 3-rd group: outer beams
of storeys from 1 to 9, 4-th group: inner beams of storeys from 1 to 9, 5-th group: outer and
corner columns of 10-th and 9-th storeys, 6-th group: outer and corner columns of 8-th and 7-th
storeys, 7-th group: outer and corner columns of 6-th and 5-th storeys, 8-th group: outer and
corner columns of 4-th and 3-rd storeys, 9-th group: outer and corner columns of 2-nd and 1-st
storeys.. The structure is subjected to the same design loads and load combinations of previous
example. The value of qs is 0.622 kPa with wind load in x-direction, assuming a basic wind
speed of 113 km/h (70mph) and the importance factor is assumed to be one. The maximum
allowable drift is restricted to 18.7 cm. The material is steel with a modulus of elasticity of 200
000MPa and shear modulus of elasticity of 77 000MPa. The yield stress and the unit weight of
material are 344.8MPa and 7850 kg/m3, respectively.

Figure 4 plane of 10 story space moment frame

The results are compared with the other references [9] as shown in table 5.7 which is more close
to them. The maximum displacement is equal to 15.5 cm which is less than the maximum
allowable top story drift = 18.7 cm.
Table 3 Optimum design variable of 10 story space moment frame
Group no.
Design variables

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

1
2
3
4
5
6
7
8
9
Max. Displ. (cm)
Weight(kg)

Hayalioglu [9]
W1426
W1240
W1235
W1235
W1022
W1235
W1468
W1468
W1482
18.1
40976.3

Present Work
W 1430
W 1434
W 1434
W 1438
W 1453
W1448
W 1253
W 1253
W1474
15.5
39970

Conclusion
The use of nonlinear finite element SAP2000 commercial program can assist greatly in achieving
a safe design and is used to check if the applied inner force, member groups and elected sections
are corresponded specified code and constraints or not.
However most of the commercial packages have been developed to be used as verification rather
than the optimization tool, but it is possible to do it by designing an optimization code in an open
file to achieve this task and interfaced with them.
Continued research is allowed for combining with other FEM programs and optimization
methods.

REFERENCES
[55] Kargahi M., Anderson J. C. and Dessouky M. M., (2012). Structural Optimization with
Tabu Search. American Society of Civil Engineers
[56] Practical optimization: a gentle introduction
john w. chinneck, 2006. http://www.sce.carleton.ca/faculty/chinneck/po.html
[57] A. G. Sextos and G. K. Balafas (2011), Using The New Sap2000 Open Application
Programming Interface To Develop An Interactive Front-End For The Modal Pushover
Analysis Of Bridges. 3rd ECCOMAS Thematic Conference on Computational Methods in
Structural Dynamics and Earthquake Engineering, Greece.
[58] Ghozi M., Aji P. and Suprobo P., (2011). Evolutionary parallel SAP2000 for truss
structure optimization. International journal of academic research. 3 (2) 1140-1145
[59] Amar K. (2005),Performance Design Of Reinforced Concrete Slabs Using Commercial
Finite Element Software. Computational Engineering Research Centre Faculty of
Engineering and Surveying The University of Southern Queensland .
[60] Chan, C.M. (1992): An optimality criteria algorithm for tall steel building design using
commercial standard sections. Structural Optimization, 5, 2629

2nd International Balkans Conference on Challenges of Civil Engineering, BCCCE, 23-25 May 2013, Epoka University, Tirana, Albania.

[61] Chan, C.M. and Grierson, D.E., (1993): An efficient resizing technique for the design of
tall steel buildings subject to multiple drift constraints. Journal of. Structural Design Tall
Build. 2, 1732
[62] Soegiarso R.; Adeli, H. (1997): Optimum load and resistance factor design of steel
space-frame structures. Journal of. Structural. Engineering., ASCE 123, 185192.
[63] Hayalioglu, M. S. (2001). Optimum load and resistance factor design of steel space
frames using genetic algorithm.Springer-Verlag journal of Structural and Multidisciplinary
Optimization, 21, 4, 292-299
[64] Pezeshk S., Camp C. V., and chen D. (2000). Design of nonlinear framed structures
using genetic optimization . Journal of structural engineering, 382-388.
[65] American Institute of Steel Construction Ad Hoc Committee (1995). Manual of steel
construction-load and resistance factor design. Chicago.
[66] Papapavlou A., (2008). A genetic algorithm method to generate structurally optimal
Delaunay triangulated space frames for dynamic loads. Degree of Master of Science in
Adaptive Architecture & Computation from University College London.
[67] Wind Drift Design of Steel-Framed Buildings: State-of the- Art Report (1988). Journal of
Structural Engineering, ASCE, 114,.9.
[68] Structural serviceability: a critical appraisal and research needs (1986) Journal of
Structural Engineering ASCE;112, 12, 264664.
[69] Degertekin S.O., Hayalioglu M.S., (2009). Optimum design of steel space frames: tabu
search vs. simulated annealing and genetic algorithms. International Journal of Engineering
and Applied Sciences. 1 (2) 34-45.
[70] Degertekin, S.O., (2007). A comparison of simulated annealing and genetic algorithm
for optimum design of non-linear steel space frames. Structural and Multidisciplinary
Optimization. 34, 347-359.