Urban transport modelling with flexible travel budgets.

Travel budget models measure travel in terms of distance and, unlike other types of urban transport model where travel is measured in terms of trips, are able to take into account known regularities in the amounts of time and money spent on travel (commonly referred to as travel budgets). the flexible travel budgets model, developed here, differs from zahavi's umot model (unified mechanism of travel) in that the total amount of time and money spent on travel by people living in different categories of income and car ownership is estimated by maximising the utility of travel subject only to upper limits of 24 hours per person per day and the national average expenditure (on all goods and services) per household per day. Tests made with reading survey data showed that the model's (inherently long-term) responses to small changes in the input values were all in the expected direction, in contrast to the umot model where the requirement for households to spend their travel budgets in full produced several elasticity values whose signs were considered counter-intuitive on current evidence and experience. (for example, higher bus fares in the umot model for total travel produced more bus travel and less car travel). average errors in the estimated daily amounts of time and money spent on travel by households in reading were only 6 per cent and 1 per cent, respectively, when compared with amounts calculated from the reading survey data. The model output distances gave errors of less than 7 per cent for all except two of the eight modes available (the error in cycle distance was 17 per cent and external bus distance 26 per cent) though this result served partly as a check on the correct functioning of the model rather than as a wholly independent verifcation of the output distances because the mechanism used to calculate distances in the model was calibrated on the survey data. Further tests are required to see whether this mechanism is transferable from town to town and over time to reduce the amount of recalibration. (Author/publisher)