| Much progress has been made in the field of intercity travel demand modelling. However, most of the calibrations of intercity demand models have been in developed countries. When the need for demand modelling arises in developing countries, the general practice is to employ models from developed countries after recalibrating them for the application context. It is shown that this practice if applied to the developing world, results in models having incompatible and inadequate functional forms. A number of significant features salient to developing countries are usually omitted.; Generalised functional forms for total demand modelling and mode choice modelling, applicable to both developed as well as developing countries are suggested herein. For the total demand model, these are based on the inclusion of intrinsic features, variables representing urbanisation, under-developed areas, a mode-abstract functional form for generalised cost and transfers. For the mode choice model, the generalisation includes capacity constraints and comfort measures. The functional forms of the model are consistent with the theory of random utility. These models have been calibrated with data on inter-district travel by public transport in Sri Lanka. Box-Cox transformations have been used in part to derive the best functional forms that can be empirically determined by the data itself. The parameters obtained are satisfactory and support the hypothesis that these are suitable functional forms for intercity travel demand modelling in developing countries.; The models are also examined for temporal and spatial transferability. The transfer of a total demand model has been accomplished after the separation of contextual factors that are transferable from those which are not transferable. Several methods have been examined for the process of updating the model parameters during transfer. A multiple scalar multiplier method gives the best results when only a small sample of data from an application context is available for updating.; A nested hierarchial decision process is considered in the estimation of the mode choice models. These models are calibrated using both aggregate and disaggregate data, providing an opportunity to make some observations on the comparative merit of each approach. |
