Using Artificial Intelligence to Predict Rail Transit Ventilation Shaft Smoke Re-circulation

Abstract

This study presents an AI-based prediction model for assessing the likelihood of smoke re-circulation between ventilation shafts, using variables such as ventilation shaft size and height, air intake and exhaust directions, separation between air intake and exhaust openings, airflow rates, fire size, wind speed and wind direction. By providing guidance on ventilation shaft configuration during the initial design phase, this model aims to reduce the time and resources required for extensive computational fluid dynamics (CFD) simulations. Following the initial prediction, CFD simulations can then be conducted by designers to confirm that the proposed ventilation shaft layout will not result in smoke re-circulation. While this approach is not a replacement for CFD, its approach offers a more comprehensive design solution and significantly decreases the number of simulations to be performed.