Since the publication of the SACTRA report there has been considerable interest in ways to change appraisal methods, particularly to model the effects of induced traffic. The paper describes models and techniques that allow the numbers of trip origins, trip destinations and their distribution patterns to change with the networks being tested. These models, which need no additional data to set up or calibrate, provide a solution to many of the issues raised by the SACTRA report. The derivation of trip distribution models, whether by the classic entropy maximising methods, or by other means, usually include the constraint that there is a fixed travel expenditure across the network. In practice, when these types of models are used this constraint is ignored and replaced by constraints on the number of trips starting and terminating in each zone. The author has investigated, and his paper sets out these investigations, the implications of retaining the fixed travel expenditure constraint and relaxing the fixed trip end constraint when making forecasts using different networks. This process automatically generates additional trips, rather than simply having the same number of trips travelling further across an improved network. The paper also describes what happens when assignments using trip matrices obtained from the models are fed into conventional evaluation processes.
Abstract