Applying agent-based modeling and simulation for domino effect assessment in chemical plants

Abstract

Historical data analysis shows that escalation accidents, a so-called domino effect, may have an important role in disastrous accidents in the chemical and process industries. In this study, an agent-based modeling and simulation approach is proposed to study the propagation of domino effects in the chemical and process industries. Different to the analytical or Monte Carlo simulation approaches, which normally study the domino effect phenomenon at probabilistic network levels, the agent-based modeling technique explains the domino effects from a bottom-up perspective. In this approach, the installations involved in a domino effect are modeled as agents while the interactions among the installations (e.g., by means of heat radiation) are modeled via the basic rules of the agents. Application of the developed model to several case studies demonstrates the ability of the model not only in modeling higher-level domino effects and synergistic effects but also in accounting for temporal dependencies. The model can easily be applied to large-scale complicated cases.