Facility LOCATION PROBLEM is a branch of operations research and computational geometry. It covers the best alLOCATION of facilities to minimize transportation costs by considering the factors involved (e. g., avoiding the placement of dangerous materials near the premises and the facilities of competitors). Given the unique customer characteristics and the fi, erce market competition of business-to-consumer e-commerce, the expected value model and chance-constrained model for uncertain facility LOCATION PROBLEMs were constructed. Owing to the intricacies of the competitive market, supply capacity, delivery cost, and customer demand were assumed as uncertain variables. The deterministic equivalent forms of the models were discussed using the inverse uncertainty distribution method. A hybrid algorithm was proposed to solve these models. Some numerical experiments were used to verify the effectiveness of the proposed models and method.

Network LOCATION models comprise one of the main categories of LOCATION models. These models have various applications in regional and urban planning as well as in transportation, distribution, and energy management. In a network LOCATION PROBLEM, nodes represent demand points and candidate LOCATIONs to locate the facilities. If the links network is unchangeably determined, the PROBLEM will be an FLP (Facility LOCATION PROBLEM). However, if links can be added to the network at a reasonable cost, the PROBLEM will then be a combination of facility LOCATION and NDP (Network Design PROBLEM). In previous studies, capacity of facilities was considered to be a constraint while capacity of links was not considered at all. The proposed MIP model considers capacity of facilities and links as decision variables. This approach increases the utilization of facilities and links, and prevents the construction of links and facility LOCATIONs with low utilization. Furthermore, facility LOCATION cost (link construction cost) in the proposed model is supposed to be a function of the associated facility (link) capacity. Computational experiments as well as sensitivity analyses performed indicate the efficiency of the model.