Climate change and global warming are resulting in extreme weather events that are occurring with increased frequency and intensity. These events often lead to supply disruptions of repairable assemblies which may be more damaging for regular operations of the main equipment (aircrafts and helicopters). In this paper, we design an inventory system for repairable service parts with a central repair facility (Depot) and several service parts inventories at the bases. When a part fails, it is replaced with part from inventory held at the base. The faulty part is transported back to the depot, repaired, and transported back to any of the bases following the stock allocation policy. However, extreme weather events may result in a scenario where the supply lines between the depot and one or more of the bases are disrupted. In such cases, a stock reallocation policy becomes a prudent approach in which the parts are transported within the bases to avoid a backorder situation. In order to service an aircraft or a helicopter at a base with low-level inventory, parts are optimally diverted from another base with higher stock. In this context, we also have to decide about the optimal time for initiating reallocation. We formulated the stock reallocation model as a Markov Decision Process (MDP) that uses an aggregate queuing model to approximate the relative value function (RVF) of the optimal stock reallocation policy. The results of the numerical experiment show that such a rule performs very close to optimality. We also conduct a sensitivity analysis to evaluate the effects of different problem characteristics on the optimality gap.