عنوان مقاله [English]
Customer satisfaction has always been one of the most important issues in public and private service centers. Management of these centers is more sensitive than production centers because the direct presence of customers when receiving services is the main feature of these service centers. In designing service centers, such as medical centers, fire fighting facilities, police stations, and so on, the location of service facilities and the allocation of service calls to servers, affect the congestion of demand. So, location and allocation of these services in an efficient mode, which has the capability of quality and time control, is inevitable. In this paper, we present a Fuzzy Queuing Maximal Covering Location-Allocation Model (FQMCLAM) for congested systems with several types of demand. Previously published papers, including fuzzy queuing maximal covering location-allocation problems, have only one type of demand in their congested systems. We extend a previously proposed model for a state in which there are several types of demand in the congested system. In this paper, a multi-objective integer programming model, in which each objective function represents the amount of covered populations for related demand, is presented. Furthermore, there are some other objective functions in the model that try to concentrate various types of server in a node to decrease the initial set up cost and help customers receive most of their services from a few different service nodes; saving travel time and costs. For example, in emergency service centers, it is efficient to concentrate several service centers in one node to decrease the initial set up cost. It is important to say that these objective functions are the same type of covered population objectives. This issue helps us to sum those objectives that simplify analysis of solution results. In addition, we assume a quality of service constraint in our model by considering service time or queue length in servers. The model is
evaluated using two test problems. These test problems consists of 16 and 25 nodes, correspondingly, which are generated randomly, according to the literature. As mentioned in the literature, QMCLAM is a NP-Hard problem, so, our problem is also NP-Hard, and it is appropriate to develop heuristic or meta heuristic methods to solve the model within a reasonable time. Finally, we investigate genetic and memetic algorithms to solve the proposed model. Both are hybridized with a heuristic algorithm to solve this problem. Solution results show that both genetic and memetic algorithm solutions are appropriate in comparison with solutions achieved by a lingo solver, and also have lower run times. Also, we see that memetic algorithms obtain better solutions than genetic algorithms, even though run time is greater.