A rectilinear distance location--relocation problem with a probabilistic restriction: mathematical modelling and solution approaches

In this study, we have considered a multi-period centre facility location--relocation problem in the presence of a probabilistic polyhedral barrier uniformly distributed on a horizontal barrier route in rectilinear plane. The objective function of this location--relocation problem is the minimisation of the cost of maximum expected rectilinear barrier distance from demand points to the new facility plus the relocation cost (i.e. a changeover cost at the beginning of each period) in the form of a mixed integer quadratic-constrained mathematical programming. The computational results show that the non-linear solver of commercial software LINGO is only effective in solving small-sized problems. A linear approximation for the system constraints is proposed so that a new mixed integer linear programming model is generated which is solvable via CPLEX optimisation software. Moreover, we proposed a problem decomposition procedure that reduces the multi-period problem into a number of single-period problems with some modifications. To show the efficiency of the model and solution methodologies, a broad range of numerical examples are performed. Results indicate that the developed problem decomposition procedure obtains the near-optimal solution comparatively with the results obtained from the non-linear solver of LINGO, and that the lower bound problem can be useful for large-sized problems in a reasonable time. Moreover, a practical case example to show the model validity in real world is solved and to reality check from practice, results are compared with the problem without barrier.