Prioritizing highway safety improvement projects using a stochastic optimization model with robust constraints

Abstract

Road authorities need efficient tools to assign a limited budget to safety improvement projects. The process of prioritizing safety improvement projects needs to predict the benefits and costs of projects. In real-life situations and also because of variances of crash frequency, crash modification factor (CMF), and imposed expenses, the forecast of project costs/benefits would be highly affected by uncertainty. Hence, this work develops a model based on stochastic binary programming with robust constraints to treat the inherent uncertainties to prioritize road safety improvement projects. Robust optimization approach ensures a high probability of feasibility for the solutions obtained under uncertain conditions. Efficiency of the suggested model is assessed using data collected from a real case study. Numerical results reveal that neglecting stochastic analysis would lead to a profit loss up to 15% of the total benefits of safety plan and the importance of considering stochastic nature of the problem increases as the budget of a safety plan decreases. Furthermore, to make a comparative analysis, the developed methodology is compared with some conventional methods, e.g., integer programming model and incremental benefit–cost analysis. Main findings demonstrate some deviations concerning how each approach copes with uncertainty which leads to differences in the list of the selected projects with respect to budget limitation. Finally, the developed methodology is recommended to managers as a flexible and robust tool to assess and select projects through setting the level of robustness against cost data uncertainty, which is done according to the decision-makers’ attitudes.