Summary

When using traffic simulations in transportation applications, model calibration is of obvious importance, for a poorly calibrated model can produce unreliable results and lead to misinformed decisions with serious consequences. Calibration is a challenging task because a micro simulation model could easily have more than a hundred parameters, each of which impacts simulation results in a manner that is often highly correlated with that of the others. At present calibration of micro traffic simulation models is quite ad hoc, often depends on a user's experience and trial-and-error. The model user is often trapped in a never-ending process of fixing one problem only to find that a new one appears somewhere else. Consequently, calibration is easily the most time-consuming task of a project. This is particularly frustrating to Caltrans district users of micro simulation, who often face tight project deadlines and have limited project resources that should be better spent on other tasks of the project. A streamlined calibration process therefore would greatly benefit these Caltrans users-it would considerably shorten the calibration phase, allowing them to spend more of their productive time solving the transportation problem at hand.

In this project, we plan to utilize our extensive experience with micro-simulation, Paramics in particular, to develop methods, procedures and guidelines that streamline the calibration of micro simulation models. We'll use Paramics to demonstrate our approach since it is at present the predominant micro simulation software used by Caltrans, although the developed method would apply to other micro simulation models with no or minor changes. The main tasks of this project include

1. Survey the current district users to identify key calibration issues
2. Synthesize current practices
3. Break the calibration process into a series of logical, steps, identify key parameters
4. Identify major calibration pitfalls and ways to detect and avoid them
5. Develop advanced procedures to ease, speed up and improve model calibration and postanalysis of results
6. Prepare a user guideline based on the project outcome
7. Conduct workshops to transfer project results to Caltrans local district users and other related parties