Comparison of Cost Functions in Modal-Split Models

Using the entropy trip distribution of Wilson as a base, an intercity bimodal-split model was developed based on the generalized cost of travel. Two variants of the model were calibrated and compared against rail and air domestic intercity travel data in the United Kingdom: one using a linear generalized cost function and the other a nonlinear generalized cost function. These were used to predict the modal share for changes in intercity travel times and frequencies that have taken place, and the predicted values compared with actual results. No significant differences were found between the two model variants. It is concluded that for the domestic intercity situation investigated, there is no significant advantage to be gained by departing from the conventional linear generalized cost function.