An efficient heuristic algorithm for the alternative-fuel station location problem

We have developed an efficient heuristic algorithm for location of alternative-fuel stations. The algorithm is constructed based on solving the sequence of subproblems restricted on a set of promising station candidates, and fixing a number of the best promising station locations. The set of candidates is initially determined by solving a relaxation model, and then modified by exchanging some stations between the promising candidate set and the remaining station set. A number of the best station candidates in the promising candidate set can be fixed to improve computation time. In addition, a parallel computing strategy is integrated into solving simultaneously the set of subproblems to speed up computation time. Experimental results carried out on the benchmark instances show that our algorithm outperforms genetic algorithm and greedy algorithm. As compared with CPLEX solver, our algorithm can obtain all the optimal solutions on the tested instances with less computation time.