The suitability of micro-simulation models to assess the relative safety of road infrastructure for vehicles and pedestrians

Abstract

Paper presented at the 31st Annual Southern African Transport Conference 9-12 July 2012 "Getting Southern Africa to Work", CSIR International Convention Centre, Pretoria, South Africa.

South Africa’s road crash record for fatalities is one of the worst in the world according to various sources and statistics. Of additional concern is that a significant proportion of these fatalities (35% or more) are pedestrians and that they have suffered disproportionately in comparison to vehicle drivers. The results of safety investigations – the causes of crashes – are traditionally aggregated under human, vehicular and environmental factors, but their interaction is seldom investigated as safety programmes, safety initiatives and infrastructure retro-fits continue to be derived mostly from the focus on historic crash statistics and on ‘black-spot’/cluster analysis of incidents. However, issues related to data availability and reliability, methodological challenges posed by the random nature of crashes and the fact that the number of crashes, per location is low, have fostered many complementary approaches to improve road safety assessments. The simulation of traffic conflicts and the use of computer based collision predictors or infrastructure safety indicators are examples of such approaches. With recent increases in computing power and programming skills, a variety of simulation software is now available to the transport profession. To date, the majority of microsimulation models have been used mainly to improve vehicular transport efficiency, but there is a recognition that they can be used to help assess safety risks. Some applications have been developed and used to successfully assess vehicle-vehicle safety. Recent developments in simulation models are now allowing more accurate modelling of pedestrians and, by extension, their interaction with vehicles and the road infrastructure. However, their use in the field of safety analysis is still very limited and depends on their ability to capture complex behavioural relationships that could lead to conflicts and to establish a link between simulated measures and crash risk. Studies of various infrastructure scenarios in Cape Town were undertaken using a suitable simulation model to assess the possibility of its use in the safety field. The results show that the model is able to simulate the difference between different infrastructure measures but that it is unable to accurately capture vehicle-pedestrian interaction for shared surfaces where pedestrians jaywalk.