JOURNAL OF SOFTWARE ENGINEERING & INTELLIGENT SYSTEMS
ISSN 2518-8739
31
st
December 2017, Volume 2, Issue 3, JSEIS, CAOMEI Copyright © 2016-2017
www.jseis.org
216
Size optimization of steel trusses using a genetic
algorithm in MATLAB
Ersilio Tushaj
Polytechnic University of Tirana, Albania
Email: ersilio.tushaj@gmail.com
ABSTRACT
An important aspect of any engineering design problem is to achieve efficiency and efficacy. This can be in terms
of energy consumption, performance, time, total weight and costs. In many cases, there are multiple solutions to
a problem and you should select the one which satisfies better the criteria. This engineering design process is
known as optimization. Optimization plays an important role in various engineering applications. Engineers are
in continuity, challenged to design structures that use the least amount of resources and satisfy the structural
requirements. The optimal design of structures can be decomposed into three major categories: topology, shape
and size optimization. These methods have evolved with time and they may be divided in two maxi-groups:
deterministic and non-deterministic algorithms. Size optimization of non-deterministic methods with genetic
algorithms (GA) are investigated in this article and applied to some steel trusses in MATLAB soft R2017a. This
is done by building an algorithm consisting in scripts and sub-functions, which are applied to the trusses for
different constraints on stresses, displacements and buckling, depending on the case analyzed. Different values
for the GA parameters are analyzed in such way to achieve the best design. The results are put in comparison with
previous studies.
Keywords: genetic algorithm; steel trusses; structural optimization; engineering; optimization; performance;
1. INTRODUCTION
Reducing costs while meeting performance standards is a common challenge in structural design. Engineers
typically rely on experience and standardized design procedures to make their structures more efficient [1]. A lot
of systematic methods based on mathematical algorithms and grouped under the generic name of Structural
Optimization are available to help designing efficient structures. Optimization is a vast field of mathematics whose
theory is still actively being developed. But when applied to structural engineering, it is essentially regarded as a
helpful to the engineer willing to design more efficient structures.
Optimization of steel trusses has been largely investigated by authors from the beginning of structural
optimization in civil engineering. The first who gave a mathematical formulation of nonlinear optimization of
steel trusses was Schmit in 1960 [2]. Others will follow introducing better performant algorithms which can offer
more reliable solutions at a minor time [3]. Optimization of steel trusses, with the developments of programming
and computers, can be considered as an integration of knowledges in structural matrix analysis, optimization
algorithms, and computer programming. Kirsch [4] in his book Structural optimization: Fundamentals and
applications, reported the necessary step to follow a total layout optimization using matrix analysis of
monodimensional structures and deterministic optimization techniques.
2. PREVIOUS STUDIES AND GENETIC ALGORITHMS IN STRUCTURAL OPTIMIZATION
Different algorithms have been applied successfully in the steel structural design. A survey was prepared by
the same author of this paper in [5]. Previous state of the art and reviews in structural size optimization have been
prepared by different authors [6-9].
Optimization of steel structures have been largely studied in the international literature. Stasa [10] is an
albanian case, from the Polytechnic University of Tirana. In her PhD Dissertation in 1994, she analyzed the
optimal design of steel trusses using two deterministic methods: the Fully Stressed Design (FSD) and the
Sequential Linear Programming with move limits (SLP). For each algorithm applied and constraints imposed,
were given the results of the optimal weight and the final design of the steel elements. The Objective Function
imposed was the minimal weight of the trusses. Stresses, displacement and slenderness criteria were applied to
the problem. Comparisons were reported between the two methods.
Hasancebi, 2009 [11], has studied the performance of some non-deterministic algorithms applied to the
optimum design of steel trusses. Chain, 2015 [12] has done a survey on deterministic approaches applied to steel
structures design. A state of the art in the use of genetic algorithms in structural optimization was prepared by
Pezeshk as a chapter in the Report in Recent Advances in Optimal Structural Design, in 2002 [13].