Crowd behaviour in Canadian football stadia — Part 2: Modelling

Abstract

In Part 1, a novel data collection exercise of a Canadian Stadium was conducted. Demographic distribution, pedestrian speed, exit and route choice, and areas of congestion were quantified using high resolution cameras. Herein, a numerical model space is built to simulate these observations in conventional software while exemplifying proper techniques. The simulations showed that using real-world behavioural data can significantly improve the accuracy of the model. When using lowest cost inputs rather than behavioural inputs, the maximum percent difference between the model and the observed egress was 15% higher. Parametric simulations showed that individual walking speeds impact overall egress time. This is in addition to crowd density also being a factor that further reduced speed. In simulations with only the fastest and slowest demographics, the maximum percent difference was 9%. Further parametric simulations increased the amount of two directional flow by 10%, which (non-linearly) increased the total egress time.