IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS

Weight Optimization of Square Hollow Steel Trusses Using Genetic

Algorithm
To cite this article: A.N. Ede et al 2018 IOP Conf. Ser.: Mater. Sci. Eng. 413 012044

View the article online for updates and enhancements.

This content was downloaded from IP address 80.248.0.226 on 12/09/2018 at 11:03

ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

Weight Optimization of Square Hollow Steel Trusses Using

Genetic Algorithm
* A.N. Ede1, O.O. Oshokoya1, J.O. Oluwafemi1, S.O. Oyebisi1, O.M. Olofinnade1, A.
I. Akpabot

1
Civil Engineering Department, College of Engineering, Covenant University,
Ota, Ogun State, Nigeria,

Abstract. Conceptual design in structural engineering entails a large amount of trial and errors or
extensive expertise to obtain the most economical and functional design solutions for large
engineering projects. In this paper a modern optimization technique called Genetic algorithm,
adopting its concept from genetic evolution is used to optimize the shape, size and topology of a
plane truss structure with the aim of minimizing the total weight of the truss. A genetic algorithm
developed in MATLAB was implemented in this paper to optimize the weight of plane truss
structures. The objective function of the optimization problem is subjected to constraints such as
stress limits, buckling constraints, tension and compression capacity according to British steel
design code BS 5950. The plane trusses which were subject to point loads were tested in the
genetic algorithm, the resulting optimized truss structures were then subject to real life loading to
determine their feasibility to withstand real life loading. The optimized trusses presented by the
algorithm were modelled in a structural analysis and design software called SAP 2000, where they
were subjected to dead and live loads. After design the weight saving discovered between the
original trusses and the optimized version was between 37 - 47%. The results show that the genetic
algorithm implemented in this study is useful in optimizing the weight of a plane truss structure.
Keywords: Steel, Structural optimization, Genetic Algorithm

1. Introduction

Sustainability of the environment involves the optimum usage of resources to create infrastructures or
projects that meet present day needs and will still be useful to meet future needs [1] An important topic in
the sustainability of civil engineering structures is the topic of structural optimization. Structural
optimization is a concept that is introduced during the conceptual design stage of engineering structures.
During the conceptual design stage, a lot of trial and error or intuition by experts is required to obtain a
good structural solution. However, with structural optimization the designer can define the objectives of
the design and the constraints to help obtain good and optimal structural solutions [2].
Structural optimization can be categorized into shape optimization, topology optimization and size
optimization. Shape Optimization treats the geometry of the truss as the design objective and the design
variable considered is the node coordinates. In topology optimization the connectivity of the members is
the objective while the number of nodes is the variables. For size optimization the members size is the
objective while the design variable is area of available cross-sectional profiles (discrete) or a specified
range of member area (continuous) [3].
Researches in structural optimization can be dated back to the 1900’s when Michell presented theoretical
optimum shapes for statically determinate trusses. Since then, many other researches have continued in
that path, it became more intense with the increase of computational capabilities [4]. This study was
therefore carried out to optimize the weight of plane steel trusses by considering topology, size and shape
using genetic algorithm technique. When the optimization problem is without constraints it is called a non-
constrained optimization but when the optimization process is subjected to one or more constraints, it is
referred to as a constrained optimization [5].
2. Truss Optimization using Genetic algorithm
A truss system consists of straight bars joined together at ends to produce rigid framework which will aid
easy load distribution and transfer to the corresponding support in form of purely axial force [6].
Optimization techniques are categorized into classical and modern methods of optimization. Classical

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

Methods are analytical and make use of differential calculus for locating optimum solutions e.g. Linear
programming, nonlinear programming etc. While modern methods adopt their operation from nature. It is
a probabilistic based approach e.g. genetic algorithm, particle swamp optimization, ant colony optimization
etc. [7].
The genetic algorithm is a method for solving both constrained and unconstrained optimization problems
that is based on natural selection, the process that drives biological evolution. Some of the concepts adopted
form this field are chromosome, genes, reproduction, mutation, crossover, populations, genotype etc. The
technique was first consolidated by John Holland in 1975 [8].
Deb and Gulati [9] used binary encoded genetic algorithms to optimize an 11-member and six node 2D truss
structure that was constrained by stress limits and buckling capacity according to Eurocode 3. The
optimized structure gave an overall weight of 4899.15 kg, a much lighter weight than previous researches.
Osman et al. [10] also developed a genetic algorithm using a different optimization process to find the
optimum weight of plane and space trusses. The proposed genetic algorithm was used to reduce
computation time significantly by reducing the required design variables for the optimization.

2. Development of genetic algorithm in MATLAB

The genetic algorithm and direct search toolbox embedded in the MATLAB software was used as a tool
in developing the genetic algorithm. MATLAB is a powerful language and it has the relevant built-in
functions to run the algorithm [13]. The finite element method was used for the analysis of the truss
structures produced during operation of the algorithm. It is used to determine the fitness values of the
various truss solutions. Finite element method is a robust and effective analysis technique for computer
solutions of various engineering problems. It is used to solve complex engineering problems that analytical
methods cannot accurately solve [11].

For the optimization of the size, topology and shape of the plane truss structures, design constraints were
considered to ensure feasibility of the truss structures. The constraints considered were; material
constraints in which the sections used were limited to the ones available commercially, the truss structure
must have all basic nodes (support nodes and load nodes), the truss structure must be kinematically stable
and the displacement limit of the node is between Lx/250 and Lx/240, as stated in BS 5950. After the
genetic algorithm produces results of an optimized truss structure, the configuration of the optimized truss
and original truss were analyzed in an analysis and design software called SAP 2000. Figure 1 shows a
cross-section of the square hollow steel section used in this paper.

Fig 1: Cross section of a square hollow steel section [12]

3. Results and Discussion

The genetic algorithm developed in MATLAB was tested on two truss problems. The trusses are, a
cantilever truss structure and a pitched truss. Figures 2a and 2b show a representation of the trusses.

2
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

(a) Cantilevered truss with 6 nodes 2-point (b) Pitched truss structure With 6 Nodes
2-point and 2 support nodes
Fig 2: Plane truss test problems

Table 1 Result of optimized cantilever truss in the algorithm

Member Cross- Weight Start End % allowable Stress

sectional coordinates coordinates stress (N/mm2)
profile (x1, y1) (X2, y2)
2 300x300x6.3 528.5232 0, 0 9.144, 0 97.2612 -180.5268
3 150x150x6.3 363.3771 0, 9.144 9.144, 0 90.283 248.2783
4 160x160x5 425.7522 0, 9.144 16.6, 3.1 87.4321 240.4383
5 200x200x5 245.4731 9.144, 0 16.6, 3.1 84.6533 -127.1448
8 120x120x4 61.7711 16.6, 3.1 18.288, 0 99.3574 273.2328
10 160x160x5 220.9704 18.288, 0 9.144, 0 92.5886 -78.2301

Total weight 1849.9258kg

Horizontally 11.267 mm
Displacement Vertically 71.3408 mm

Figures 3,4 and 5 show the original and optimized cantilever truss structure with the details being presented
in tables 1, 2 and 3. The optimized cantilever structure was thereafter modelled in SAP 2000 software and
subjected to dead and live loads. The result of analysis and design showed that all truss members
successfully passed stress checks in accordance to BS 5950 requirements. The total weight saving of the
optimized cantilever truss structure compared to the original truss is 47%. For the cantilever truss the
weight of the original truss when subjected to dead and live loads gave a value of 5970.22 kg and the
optimized truss gave a weight value of 3146.2Kg. From the original truss the number of members were 10
but when the algorithm optimized the truss the number of members reduced to 6 showing that 4 members
in the original truss were non-critical.

3
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

Fig 3: Optimized cantilever truss in the algorithm

Fig 4: SAP 2000 model and section sizes of the original cantilever truss

Fig 5: SAP 2000 model and section sizes of the optimized cantilever truss

Table 2 Sap 2000 design result for the original truss

Member Cross-sectional profile Length (m) Weight (Kg) Stress
(N/mm2)
1 400X400X10 18.288 2231.136 -147.416
2 150X150X5 9.144 206.6544 114.576
3 200X200X8 18.288 872.3376 256.576
4
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

4 200X200X5 9.144 277.976 0.492

5 100X100X4 9.144 108.8136 -122.206
6 300X300X8 12.93157 941.418296 160.037
7 250X250X6.3 12.93157 619.4222 -85.804
8 200X200X6.3 12.93157 491.3997 263.567
9 140X140X5 12.93157 271.5657 244.275

Total weight 5970.723496Kg

Vertical = 53.916mm
Displacement Horizontal = 12.21mm

Table 3 Sap 2000 design result for the optimized cantilever truss
Member Cross-sectional Length (m) Weight (Kg) Stress
profile (N/mm2)
1 350X350X10 9.144 969.264 -122.945
2 250X250X6.3 12.93157 619.44 152.684
4 200X200X8 17.66607 842.6715 125.359
5 200X200X8 8.07477 381.6 -81.612
6 160X160X6.3 3.52978 106.2463 135.073
7 200X200X5 9.144 227.9776 -63.862

Total weight 3147.1994Kg

Vertical = 38.82mm
Displacement Horizontal = 7.02mm

Fig 6: Optimized pitched truss in the algorithm

Table 4 Result of optimized pitched truss in the algorithm

Member Cross weight Start End % of Stress
sectional coordinates coordinates allowable (N/mm2)
profile (x1, y1) (X2, y2) stress
1 120x120x4.9 63.0971 0,0 3,2 78.6274 -
159.8234
3 100x100x5 53.0016 3,2 6,4 85.2432 -
142.3244
4 80x80x3.2 27.5104 3,2 6,0 73.992 -91.6939
5 50x50x4 22.5600 6,4 6,0 98.8142 271.7391

5
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

6 120x120x4 51.9199 6,4 9,2 71.1524 -

145.6985
7 80x80x3.2 27.5104 6,0 9,2 73.992 -91.6939
9 160x160x5 86.8938 9,2 12,0 48.8696 -
116.3081

Total weight 336.6586kg

Horizontally 4.3712 mm, 27.32% of the limit
Displacement Vertically 13.9069 mm, 28.9728% of the limit

Fig 7: SAP 2000 model and section sizes of the original pitched truss

Fig 8: SAP 2000 model and section sizes of the original pitched truss

Figure 6 shows the optimized pitched truss after the optimization process in the algorithm. The original
truss structure had 10 members but after the optimization the truss structure had 5 members. The optimized
pitched truss with the original truss structure were modelled in SAP 2000 as shown in figures 7 and 8. The
reduction in weight of the pitched roof truss is about 37-47 %. As presented in tables 4, 5 and 6, the weight
of the original truss structure after being subject to dead and live load was 1097.9 kg, but after optimization
the truss weight was reduced to 688.08 kg.

Table 5 Sap 2000 design result for the original pitched truss

Member Cross-sectional Length (m) Weight (Kg) Stress

profile (N/mm2)
1 200X200X6.3 7.2111 270.218 -116.383

6
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

2 200X200X6.3 7.2111 274.0218 -116.383

3 200X200X5 12 406.8 48.484.
4 80X80X3.6 3.60555 51.19881 -68.769
5 80X80X3.6 3.60555 51.19881 -68.769
6 80X80X4 4 37.64 150.498
7 40X40X3 2 6.82 -22.084
8 40X40X3 2 6.82 -22.084

Total weight 1097.89742Kg

Vertical – 6.708mm
Displacement Horizontal – 2.16mm

Table 6 Sap 2000 design result for the optimized pitched truss
2
Member Cross sectional profile length Weight (Kg) Stress (N/mm )
1 200X200X6.3 7.2111 274.0218 -120.391
2 200X200X6.3 7.2111 274.0218 -120.391
4 80X80X3.6 3.60555 51.19881 -74.952
5 80X80X3.6 3.60555 51.19881 -74.952
6 80X80X4 4 37.64 153.105

Total Weight 688.08122Kg

Max displacement Vertical – 6.85mm
Horizontal – 2.25mm

4. Conclusion

A genetic algorithm was developed in MATLAB to optimize the weight of plane truss structures and the
optimized trusses were modelled in a structural analysis and design software called SAP 2000 where they
were analyzed and designed under real life loads. The trusses that were optimized showed significant
reduction in the total weight by 30 - 40% showing that the algorithm is a useful tool for engineers in
obtaining structures of optimal weight during design. All structures optimized were within acceptable
stress and displacement limits. For further research works the genetic algorithm used in this paper can be
implemented in the design of other structural elements using other available steel profiles provided in the
design codes. Also, further researches are suggested for the optimization of 3D trusses such as tower cranes
using genetic algorithm technique.

Acknowledgement
The authors wish to show their appreciation to the chancellor and the management of Covenant University
for the Platform created to ensure the successful completion of this research work.

References

[1] Anthony, N. E., Gideon, O. B., Oluwarotimi, M. O., David, O. O., Gideon, A. A., & Ben, U. N.
(2016). mpact of Reliable Built Structures in Driving the Sustainable Development Goals: A look
at Nigerian Building Structures. 3rd International Conference on African Development Issues
(CU-ICADI 2016) (pp. 350-353). Ota: Covenant University Press.
[2] Mark, S., Eric, L., Chung-Soo, D., & David, S. (2009). Optimization Tools for the design
of Structures. SEAOC 2009 Coventio Proceedings, 1-17.

7
ICESW IOP Publishing
IOP Conf. Series: Materials Science and Engineering 413 (2018) 012044 doi:10.1088/1757-899X/413/1/012044
1234567890‘’“”

[3] Kaveh, A., & Kalatjari, V. (2004). Size/geometry optimization of trusses by the force
methodand genetic algorithm. ZAMM – Zeitschrift für Angewandte Mathematik und
Mechanik., 84(5), 347-357.
[4] Rory, C. (2013). Algorithm Selection in Structural Optimization (1st ed.). Massachusetts:
Massachusetts Institute of Technology.
[5] Margaret, E. K. (2012). A Two-Phase Genetic Algorithm for Simultaneous Dimension,
Topology, and Shape Optimization (1st ed.). Ohio: The Ohio State University.
[6] Anthony, N. E., Olatunbosu, A., & Oluwarotimi, M. O. (2015). Modelling, Analysis and Design
of a MultiStorey Storey Helipad-Car Park: a Proposal for Canaan. International Journal of
Innovative Science and Modern Engineering (IJISME), Volume - 3 Issue-4, pp43-47.
[7] Singiresu, S. R. (2009). Engineering Optimization (1st ed.). New Jersey: John Wiley &
Sons,Inc.
[8] Reeves, C. R., & Rowe, J. E. (2002). Genetic Algorithm: Principles and Perspectives- A
Guideto GA Theory. New York.
[9] Deb, K., & Gulati, S. (2000). Design of Truss-Structures for Minimum Weight using
Genetic Algorithms. Finite Elements in Analysis and Design, 447-465.
[10] Osman, S., Atef, E., Tharwat, S., & Osman, H. (2014). Optimization of Plane and Space
TrussesUsing Genetic Algorithms. International Journal of Engineering and Innovative
Technology (IJEIT), 3(7), 66-73.
[11] Austrell, P. E., Dahlblom, O., Lindemann, J., & Olsson, A. (2004). CALFEM- A Finite
Element Toolbox. Lund: Structural Mechanics.
[12] BSI (2000). Structural use of steelwork in building - Part 1: Code of practice for design -
Rolledand welded sections. BSI.
[13] MathWorks. (2017). Genetic Algorithm and Direct Search Toolbox 2.4.2. Retrieved
November 17, 2017, from www.mathworks.com