Network-based Pedestrian Dynamics Model for Facility Layout Planning in Airport Check-in Systems

Service facility layout planning (SFLP), such as in airports, must account for pedestrian dynamics and behaviour. In practice, SFLP often relies on computationally intensive simulation models, which makes the integration of optimization modules challenging. In this work, we propose a networkbased SFLP framework for service systems like airports. The framework consists of a mesoscopic pedestrian dynamics simulation model that is employable in a simulation-optimization setup. We use a modified genetic algorithm as the optimizer and a large language model (LLM) based agentic system to do post-processing. We propose a set of metrics to evaluate the layout from both passenger and operational perspectives. We experiment with two Indian International airport terminal check-in area layouts to validate the efficacy of the proposed model. Finally, we use social force models to evaluate the optimized layouts. Results show that the proposed model was able to improve the current layouts for both passengers and the operating authorities.