Multi-objective unequal area pod-structured healthcare facility layout problem with daylight requirements

Outpatient clinic design can be optimized with a proposed mathematical model that formulates an Unequal Area Facility Layout Problem (UAFLP). The proposed non-linear multi-objective model maximizes natural daylight exposure and minimizes total walking distance for patients while considering constraints for modeling daylight, enabling a pod-structure, allowing for rotation of spaces, and ensuring accessibility to rooms within pods. This study identifies efficient algorithms that can tackle several inherent challenges of this NP-hard problem, including computational complexity and approximation of the area constraints for spaces with wide bounds. An adapted Genetic Algorithm for placement (GA-P) proved effective in solving the proposed multi-objective UAFLP model that estimated available daylight by using simulation-optimization. Sensitivity analysis conducted based on the experimental design indicates that the main factors affecting the algorithm's performance are the selection mechanism for best Pareto point, number of spaces requiring lighting, and lighting dataset used. This adapted GA-P achieves better results than a Genetic Algorithm with continuous encoding of space coordinates (GA-C), Particle Swarm Optimization (PSO), and Simulated Annealing for Placement (SA-P) in terms of daylight exposure and travel distance for a multi-objective problem.

Automated summary

A proposed non-linear multi-objective model utilizing Genetic Algorithm is effective in optimizing outpatient clinic design by maximizing natural daylight exposure and minimizing total walking distance for patients. Efficient algorithms are identified to tackle challenges in computational complexity and area constraints approximation for spaces with wide bounds in clinic design. Sensitive analysis reveals that the main factors affecting algorithm performance are the selection mechanism for best Pareto points, number of spaces requiring lighting, and lighting dataset used, with adapted Genetic Algorithm proving superior in achieving better results for a multi-objective problem compared to other optimization algorithms.