ESTIMATION OF TRIP MATRICES FROM VOLUME COUNTS. VALIDATION OF A MODEL UNDER CONGESTED CONDITIONS

The author has developed a model for the estimation of trip matrices from traffic counts based on an entropy maximising framework. The original model assumed that for each counted link it was possible to identify the proportion of the trips from each origin-destination (o-d) pair which would use it. Furthermore, it is usually assumed that this route choice proportion can be identified independently from the o-d estimation process. These assumptions are perhaps reasonable wherever congestion does not play a key role in route choice. This o-d estimation model was validated using a comprehensive data set collected by TRRL in the central area of Reading. The results of these tests were reported in a paper during the 8th International Symposium on Transportation and Traffic Theory in Toronto. The original model has now been extended to cope with cases where congestion does play a role in route choice. This extension of the model has been validated again with the same data base. It was found that the explanatory power of the model was improved at a moderate extra cost in computer resources. The improved model is likely to be of great interest in practioners as it makes possible the use of relatively inexpensive traffic counts to update and estimate trip matrices under a wide range of conditions. (Author/TRRL)