GPT-based Building Performance Simulation

We have developed a GPT agent-based building simulation and decision support system, capable of simulation modeling, result analysis, visualization, and design optimization. Remarkably, we have found that GPT can identify optimal solutions through iterative simulation processes. We started with the implementation of CONTAM, a multi-zone airflow model, and are currently working on its integration with EnergyPlus. Additionally, we plan to expand the system into a Multi-Agent System to offer more sophisticated and diverse functionalities.