Intelligent Agent-Based Simulation of Fire Propagation in Multiple Environments

Abstract

Wildfire propagation in urban and forest environments poses significant challenges, particularly in regions like Cuba, where the interplay of diverse environmental factors amplifies the risks. This study develops an agent-based simulation model integrated with Geographic Information Systems (GIS) using the GAMA platform to analyze fire dynamics under varying conditions. The model evaluates critical environmental variables, including wind speed, fuel types, and humidity, and identifies their impact on fire behavior. A multi-phase experimental approach was employed, incorporating sensitivity analysis, a screening experiment, and a full factorial design. The results highlight that wind speed and the number of fire outbreaks are the most influential factors, with high wind speeds doubling the rate of fire spread. Notably, the strategic use of firebreaks reduced the affected areas by up to 40%. Additionally, interactions between factors, such as wind speed and the number of outbreaks, demonstrated significant effects on burned area and fire intensity. This research advances the state of fire propagation modeling by incorporating real-time meteorological data, adapting the model to the Cuban context, and enabling realistic scenario testing. The insights gained provide actionable recommendations for disaster management, urban planning, and firefighter training, offering a robust tool for mitigating fire-related risks in tropical and subtropical environments.