Maximizing Safety Improvement Benefits in Crash Prediction Models with Accident Modification Factors

Abstract

A linear optimization model that maximizes the safety benefits of improve-ments on an existing highway within specific budget constraints is presented. This model works in conjunction with crash prediction models that predict the expected number of crashes for highways by using a base model and productive accident modification factors (AMFs). The base model predicts the expected number of crashes for a base highway. The AMFs modify the prediction on the basis of the actual characteristics of the highway. If there were multiple alternative improvements for several highway features, then there would be many combined alternatives, each with a certain cost and a certain degree of improvement in highway safety. The proposed model uses linear optimization to find the combined alternative that has the best safety improvement within the available budget. The mathematical optimization model is studied in a general form as well as a detailed form in the context of an existing crash prediction model, the crash prediction module of the interactive highway safety design model. The procedure for building the optimization problem is described. A C programming code was developed to build the linear optimization problem. A test case study is defined, and the problem is built and solved by using the CPLEX optimization solver. The variation of the safety measures versus improvement costs is studied, and the results are discussed.