Architectural and structural design optimization in parametric software typically allows only about a thousand evaluations of an expensive model, even for problems with tens of variables. This is a constrained regime where most metaheuristics were not designed to operate effectively.
The paper proposes hybrid metaheuristics specifically for this small evaluation and wall-clock budget. By combining different search strategies, the approach aims to deliver useful design improvements within the limited number of model calls.
The abstract does not detail the specific hybrid combinations or benchmark results, but the motivation is clear: standard optimization methods often assume larger budgets, and this work addresses that mismatch.