[Truncated] This thesis presents a method of identifying road maintenance and renewal projects and of prioritising and scheduling them in a network during an analysis period, based on the principle of maximising the net benefits from limited expenditure. Previous methods, such as enumeration, linear and goal programming, and dynamic programming, have been restricted to a limited number of projects, and could not deal adequately with the interdependence of projects. A robust genetic algorithm is used to solve the combinatorial optimisation and dynamic expenditure problem, taking full account of interdependence. The entire model includes two important sub-models dealing with travel demand forecasts and prediction of road deterioration. This thesis deals only with road traffic, so that there is no mode split. Because the traffic between origin-destination pairs (OD traffic) is not available, the travel demand model in this study combines the other three stages of the four-stage approach, trip generation, trip distribution and traffic assignment, into a single entity. The combined gravity and logit route choice model, estimated for traffic forecasting, is based directly on the easily obtained zonal information and link flows. The most likely OD matrix is calibrated by the joint model without reference to any prior OD matrix.
|Qualification||Doctor of Philosophy|
|Publication status||Unpublished - 2002|