Computational Approaches for Decision Support in Structural Design and Performance Evaluation

Abstract

Structural design and performance evaluation involves complex decision-making tasks that are both time and cost intensive. Decision Support Systems (DSS) that combine engineer interaction with computer-based tools to provide a joint-cognitive decision-making environment can significantly help engineers. In this study, three different computational approaches that can become part of such a DSS are proposed. Two of these approaches use simulation-based design to assist engineers with the design of (1) bridge trusses and (2) moment-resisting steel frames, and the third is intended for internet-enabled remote control and observation of experimental facilities. A hybrid optimization approach that is composed of two algorithms - an algorithm that generates a good starting solution and a local search, is proposed for truss optimization. The performance of the approach is compared with existing optimization approaches. Then, a GA-based optimization approach is presented for the design of moment-resisting steel frames with rigid connections. The GA uses a unique crossover scheme to perform trade-off study between the number of rigid connections and the total cost. It is observed from the trade-off study that the total cost of the frame is minimum when using only a few rigid connections if they are placed at certain locations in the frame. 'Modeling to Generate Alternatives - MGA' is used to generate good design alternatives that may be more suitable for design than the optimal solution. A generic computational framework for internet-enabled remote control and observation of laboratory experiments is proposed. The framework is illustrated for a shaketable experiment. The framework uses LabVIEW and web technologies to form a secure framework that permits multiple observers to simultaneously view and download measurements from the experiment while a single user controls the shaketable.