Analysis of Incident-Induced Capacity Reductions for Improved Delay Estimation

Abstract

This paper investigates the randomness in incident-induced capacity reductions and discusses the further impacts on delay calculation and modeling. For this purpose, incident and traffic count data sets from four important freeways in California, i.e., I-80, I-280, I-580, and, I-880, dating from February 1 to June 20, 2017, are utilized to analyze the incident capacity reductions. Accordingly, the capacity reduction distributions are identified and compared with the findings from the literature. In addition, the impact of variation in capacity reduction on incident delay is derived analytically for a deterministic queuing model. Through further analysis of the data, it is shown that capacity reduction values vary with respect to traffic conditions (e.g., volume). Accordingly, it is discussed that capacity reduction tables that do not incorporate traffic flow conditions and variance may provide incorrect estimations of delay. Considering the widely used capacity reduction tables for delay calculation, a regression tree approach is utilized to provide similar tables that provide traffic-dependent capacity reduction and variance for practitioner use.