Operational planning for large State Road Transport Corporation

Abstract

This dissertation addresses to operational planning problem for large public road transport networks typically operated by the state road transport corporations (SRTCs). The problem involves several operational constraints imposed by the vehicle and crew duty conditions. It has several economic and operational objectives like, the fleet-size, the crew-size, the number of vehicle getting daily routine maintenance and so on. Taking fleet-size as the dominating objective, the operational planning problem is viewed in terms of five inter-related sub-problem, namely, the fixed-schedule fleet-size problem, the variable-schedule fleet-size problem, the depot allocation problem, the vehicle scheduling problem and the crew scheduling problem. The fixed-schedule fleet-size problem is concerned with the participating of a given set of trips with fixed departure timings, into the minimum number of sub-sets such that each sub-set of trips forming a chain, is capable of being operated by a single vehicle. The problem considers only the time and the space constraints and ignores the specific operational constraints like the routine maintenance and the crew duty conditions. The problem stand solved in the literature. However, in view of its relevance to the other sub-problems, a detailed exposition of the problem has been given together with some new results on the number of optimal solutions to the problem. The variable-schedule fleet-size problem involves fixing of the departure timings of the trips within the tolerance limits around the most desirable timings such that the fleet-size in minimized. A secondary objective requires the maximum of trips to be fixed at the most desirable timings pre-specified in the provisional timetable. A heuristic approach is developed to perturb the trips within the tolerance limits to reduce the fleet-size keeping number of perturbations to a heuristic minimum. Starting with a provisional allocation of depots, the depot allocation problem seeks to minimize he total fleet-size by reallocating the depots to sue of the trips. The problem also involves a secondary objective namely, the minimization of the number of reallocations with inspect to the provisional allocation. A heuristic approach is developed to solve the problem optimally with respect the fleet-size, keeping the number of reallocations to s heuristic minimum. The vehicle scheduling problem considers a routine maintenance constraint. The constraint requires that every vehicle be provided a routine maintenance at least once every two days at its home depot. A heuristic approach is developed to solve the problem in two stages. In the first stage, a minimum fleet-size problem is solved such that the number 0f maintenance opportunities embedded in the chains are maximized. The problem of maximizing the maintenance opportunities is formulated as an assignment problem and a greedy procedure is developed to solve the problem optimally. The maintenance opportunists are appropriately redistributed among the chains during the second stage through a restructuring process to yield a maintenance feasible solution while heuristically maximizing the number of vehicles getting daily maintainable. The crew scheduling problem considers three specific crew duty conditions, namely, the steering duty limit, the spread over limit and the number of consecutive night-outs at the terminals other than the depot. A heuristic approach is developed to solve the problem. The approach is embedded in the vehicle scheduling algorithm to achieve overlap between the two solutions so that the crew is not required to change the vehicles too frequently. The cannon framework used in the algorithms enables their integration to form an operational planning model for comprehensive solution of the problem. The model is capable of handling a real-life size problem. The lower bounds are developed for the fleet-size, the crew-size and the number of non-maintainable vehicles. The algorithms are evaluated by comparing the actual objective.1unctim value; achieved against the corresponding lover bounds. The model was tested with the real-life and the randomly generated trip data with encouraging results.