Failure Detection and Diagnosis in Microsimulation Traffic Models

Abstract

Effective and feasible procedures for validating microscopic, stochastic traffic simulation models are in short supply. Exercising such microsimulators many times may lead to the occurrence of traffic gridlock (or simulation failures) on some or all replications. Whereas lack of failures does not ensure validity of the simulator for predicting performance, the occurrence of failures can provide clues for identifying deficiencies of the simulation model and invite strategies for model improvement. Failure is defined as a severe malfunction in one or more traffic links on the network where vehicles are unable to discharge for an unusually long period. Such malfunctions can be detected with link-based time traces of vehicle trips. Identifying locations where malfunctions arise requires further spatial analyses. A procedure for identifying whether, when, and where failures occur is described. The simulator CORSIM serves as the test-bed simulator for the proposed methodology, but the procedure is applicable to any comparable microscopic model; real-world traffic networks are simulated as case studies. Possible root causes of detected failures are (a) flaws in the simulator behavioral algorithms, (b) improper calibration of inputs, and (c) capacity problems related to the specific network (for example, when heavy traffic demand projections are simulated). Strategies to identify why failures occur are used in the case studies. The results indicate that the proposed failure detection and diagnosis process is an effective (and essential) way to explore the validity of traffic simulators and find improvements where needed.