START: The use of sensitivity analysis in traffic model calibration

Project details

Duration

10.2012-09.2016

Sponsor

NA

Staff

Dr. M. Menendez and Q. Ge

Summary

The calibration of traffic simulation models has been a topic of growing attention in the transportation research community. One essential area of model calibration is the Sensitivity Analysis (SA) of the input parameters. Unfortunately, there is no literature suggesting a standard method for the application of SA on traffic models. Users typically choose the important parameters according to former experiences. This often leads to model imprecisions and unrealistic values for the calibrated parameters.

The proper usage of SA can aid in the selection of the correct parameters for calibration and validation. It provides quantitative and qualitative information regarding the effects of the different model parameters on the simulation ouputs. This, in turn, can greatly reduce the efforts required for computation and analysis of the results.

The proposed research is expected to provide both scientific and pragmatic support for the use of SA in the calibration of traffic models. Its objectives include getting a better understanding of the benefits and drawbacks of different SA methods, gaining in-depth knowledge about the behaviour of different microscopic traffic parameters (especially for complex urban networks), and developing useful guidelines to aid model users for selecting and implementing SA methods. Moreover, this research is expected to contribute to the dissemination of best practices on the quantitative calibration and validation of traffic simulation models.