Is the value of a multi-period facility location model predictable? An exploratory data analysis
Facility location planning plays a crucial role in the design of a distribution network. Thereby, the aim is to locate facilities and to allocate customers to them in the most favourable way. As it is likely that parameters, such as cost or demand, of a facility location problem change over time, the solution to a single-period facility location problem may not remain optimal over time. In a multi-period facility location model, on the other hand, decisions are made on a time-dependent basis. However, the complexity of the calculation increases significantly. Therefore, the aim of this thesis is to determine characteristics of a facility location problem under which it is worthwhile to deploy a multi-period model. Additionally, it is to be evaluated under what conditions it is advantageous to deploy a static counterpart with time-dependent allocation decisions. For this purpose, a quantitative data analysis is carried out. Within a scenario of increasing customer demand a capacitated phase-in model was applied to assess the relationship between the Value of the Multi-Period Solution (VMPS) and selected parameters of the model. As a major result of the analysis it can be pointed out that there is a significant correlation between the VMPS and parameters like facility’s fixed costs, facility’s capacity and the percentage of demand increase. Furthermore, there is a major impact arising from thelevel of facility’s capacity on the benefit of multi-period allocation decision.