Novel adaptive cross-entropy method for cost-effective design of steel moment frames

In this paper, we present an adapted Cross-Entropy Method (CEM) for the cost optimization of steel moment frames, aimed at significantly reducing computational effort without compromising solution quality. Traditional CEM, while effective, requires extensive structural analyses to identify elite solutions, leading to high computational costs. Our proposed adaptation introduces a pre-screening phase where particles are initially sorted based on their estimated structural weight, a proxy for cost-effectiveness. Only the most promising particles undergo full structural analysis, and those meeting structural constraints are selected as elite particles. If the number of elite particles falls short of the required elite set in some iterations, the method reverts to the traditional CEM approach for marking elites. This strategy not only reduces the number of structural analyses required but also accelerates the convergence to optimal or near-optimal designs. Comparative studies demonstrate that our adapted method achieves significant computational savings while maintaining, and in some cases improving, the quality of the solutions. The proposed methodology offers a robust and efficient tool for structural engineers and designers seeking to optimize steel moment frames under cost constraints.