Abstract: Besides minimizing the total cost of the system,the design of distribution system should also meet the time requirements of the demand points.After listing the assumptions for modeling,a time-constrained mixed-integer nonlinear programming(MINLP)model of the single-source and p-median distribution system is built.According to the specific configuration,a heuristic algorithm programmed with Matlab for the model which can find the globally optimal solution is developed.Finally,a comparison among the branch and bound algorithm,rounded relaxation and the heuristic algorithm,on the software platforms of LINGO and Matlab,is presented.The result shows that the heuristic algorithm is better than the others at computing time and seeking for the globally optimal solution.At the end of this paper,a numerical example was given,in which the data were gained from the fieldwork.

Key words: time constraints, distribution system, mixed-integer nonlinear programming, heuristic algorithm, comparison of algorithms