Traditional loss models tell you what breaks. Agent-based models tell you what happens next — who moves, who waits, and where the plan quietly fails.
Traditional earthquake loss models answer a static question: given this shaking, how many buildings collapse and how many people are injured? That number is essential — and radically incomplete. It says nothing about the hours that follow: which roads are blocked, which hospitals overflow first, where rescue teams queue instead of dig.
Agent-based models start where loss models stop. Each responder, ambulance, hospital and volunteer becomes an agent with its own goals, information and constraints, placed inside a GIS model of the real city. The response emerges from thousands of interactions — exactly as it does on the ground.
The payoff is diagnostic, not decorative. In our district-scale studies, the simulations repeatedly surfaced failure modes that no static analysis could show: a bottleneck intersection that delays every ambulance run, a triage policy that looks fair on paper but starves the busiest hospital, volunteers who converge on visible damage while collapsed schools two blocks away wait.
None of this replaces engineering judgment or field experience. It gives them a rehearsal space. A plan that has failed a hundred times in simulation — and been fixed a hundred times — meets the real event in far better shape than a plan meeting reality for the first time.
Comments
No comments yet — be the first.