Avaliação do potencial de modelos de perdas de vida baseados em agentes no aprimoramento de sistemas de alerta e evacuação em inundações naturais e ruptura de barragens

Abstract

The present study investigates the effectiveness of the LifeSim and LSM models in predicting loss of life in flood events, using case studies of the Brumadinho dam failure, Brazil, and floods in the Ahr River valley, Germany. The main objective was to compare and validate the accuracy of these models, analyzing their simulation capacity in disaster scenarios and the influence of warning and evacuation measures on reducing fatalities. In the study undertaken for the rupture of the B-I Dam, an event marked by the absence of effective emergency planning, the modeling revealed a satisfactory representation in the hydrodynamic simulation with HEC-RAS and, in relation to the estimation of loss of life, LifeSim showed excellent adherence to the real event, while LSM displayed overestimated and questionable results. In the case study of the flood that occurred in July 2021, in which extreme rains caused 184 deaths in Germany, 135 in the Ahr River valley, the studies undertaken, as in the previous case, showed that HEC-RAS was useful in the simulation detailed analysis of water behavior, while LifeSim, in addition to presenting results compatible with reality, identified possible factors that contributed to fatalities. It was identified that an early warning could significantly reduce fatalities in both cases, highlighting the importance of effective warnings and population preparation. In general, the case studies demonstrate the importance of accurate models and efficient warning systems for reducing fatalities in disasters, in addition to the need for additional studies to understand the uncertainties of predictive modeling. LifeSim proved to be more effective in representing real events and simulating evacuation scenarios, standing out as a valuable tool for risk management in natural and technological floods.