This project aims to address the well-known problem of inconsistent travel impedances that exists within FSUTMS' four-step traditional demand model by designing and implementing an automated feedback loop in FSUTMS. The direct method and Method of Successive Averages (MSA) method are used for implementing feedback.
The results are compared against the preassignment scheme as the benchmark. In terms of the effects on the network performance, the feedback process results in higher speeds, shorter travel times, and lower volume-to-capacity ratios. These changes become more pronounced as the congestion level increases, which are consistent with the findings reported in previous studies, except that the effects of feedback in this study are found to be pronounced at earlier stage, i.e., a medium level of congestion. In terms of Root Mean Square Error (RMSE), the model accuracy is improved marginally due to feedback using the friction factor as determined from the pre-assignment scheme. Feedback with on-line trip length calibration and total VHT (Vehicle Hour Traveled) control through proportional adjustment by iteration (PAI) produces a comparable RMSE and other statistics with the pre-assignment scheme. Overall, the results presented in this report do not provide sufficient evidence to support the significant benefit of feedback process to the model accuracy as claimed by previous research. This can be partially attributed to the fact that the proposed feedback model has not been fully calibrated as in the pre-assignment model. Proper calibration of friction factors governing trip length and other important model parameters seems to bear more significance affecting the model accuracy than the feedback process itself.