Sensitivity Evaluation of Traffic Assignment (17-2)

The purpose of this study was to investigate the sensitivity of traffic assignment to input from the preceding modeling phases (i.e., the trip generation and trip distribution phases). The analyses focused not only on sensitivity of assignment results to inaccuracies from the modeling phases, but also the sensitivity of various commonly used measures of assignment accuracy in discerning such inaccuracies. The results indicate that the percent RMS error is the measure most sensitive to trip matrix inaccuracies, while the total vehicle miles of travel (VMT) was the least discriminating. The analyses further demonstrate that, due to the aggregative nature of the assignment procedure, many differences that may be observed at the zonal level and zonal interchange level tend to disappear in the assignment results. Based on the results of these analyses, a "short-cut" (sketch planning) approach is proposed, which would be expected to produce assignment results of sufficient accuracy for preliminary system evaluation and comparison with other alternatives similarly modeled.