Could not download file: This paper is available to authorised users only.

SIMULATED ANNEALING WITH PARAMETER TUNING FOR ENERGY OPTIMIZATION OF RETAINING WALLS. A CASE STUDY APPLICATION IN EDUCATION

J.V. Martí, V. Yepes, F. Molina-Moreno

Universitat Politècnica de València (SPAIN)
This paper deals with the university training in engineering of projects in the postgraduate course ‘Predictive and optimisation models for concrete structures’, within the Masters in Concrete Engineering of the Universitat Politècnica de València. The course focus in the automated design of concrete structures and provides a state of the art on the application of several heuristic algorithms to the design optimization of real structures, e.g., earth-retaining walls, road portal and box frames, building frames, road vaults, bridge piers, abutments and decks. This communication presents a case study of an infrastructure where the ongoing research of the teaching body is applied as academic material. The case study is on a type of reinforced concrete retaining wall used in the construction of state roads, whose energy consumption is to be measured and heuristically minimized. To this aim, the embodied energy in equipment and material extraction is selected as objective function based on production materials, transport and placement. First, the students must work with a free choice of heuristic parameters for the optimisation problem. Because of the deterministic nature of the decision making problem, students rapidly figure out how complex the parameter’s tuning becomes. Then the algorithmic neighbour search method is conveyed to the students to solve the parameter tuning problem.

The parameters tuning is performed through the simulated annealing algorithm (SA) and it is applied to the optimisation of a wall of 10.00 meters’ height. The model consists of 32 discrete variables describing the structural geometry as well as characteristics of the materials and passive reinforcement. The aim of this study is to present a methodology to obtain the parameters that define the heuristic through a neighbour local search. For this purpose, several indexes need to be evaluated: the objective function, the processing mean time and the results’ scatter. Finally, the case study disclosures to the studentship that the heuristic optimization is a good tool to design buttressed earth-retaining walls by the reduction of the energy consumption. Furthermore, it is showcased how the tune of the algorithm parameters can help in the extraction solid results. This academic procedure unifies in a nutshell the course curricula of the MSc course with the ongoing research of the teaching and research group.