Design optimization of large-scale structures

Engineers have to choose among many competitive solutions when designing complex structures in order to find the designs that merit further development and analysis. This task is often in contrast with the constraints on time and computational resources. For this reason, optimisation techniques are very useful if they allow engineers to obtain a large set of designs at low computational cost. Amongst the different optimisation methods, Sequential Linear Programming (SLP) is very popular because of its simplicity and because linear solvers (e.g. Simplex) are easily available. Despite of their theoretical simplicity, “well coded” SLP algorithms may outperform more sophisticated optimisation methods. A practical problem in engineering design is the optimisation of truss structures. Hence, this paper presents the experiences done in design optimisation of large scale truss structures with SLP based algorithms. Sizing and configuration problems of structures under multiple loading conditions with up to 200 design variables and 3500 constraints are considered. Truss structures are widely diffused in engineering and are often chosen as benchmark cases to test the efficiency and reliability of optimisation methods. Therefore, in this work, the relative performance and merits of some SLP based algorithms are compared and the efficiency of an advanced SLP based algorithm called LEAML is tested. The aim of the paper is to show that LEAML is a very powerful design tool for large scale truss structures and even outperforms the Sequential Quadratic Programming method (SQP) which is considered by theoreticians perhaps the best theoretically founded optimisation technique.