Network design is one of the important and basic issues of supply chain and logistics because the other planning activities and related issues are affected by the network general layout. This importance in service parts logistics systems is much more due to the high price, high costs of equipment downtime and the necessity of representing quick services to customers. Therefore, network designers and decision makers are always looking for a model which consider maximum influential factors in this issue and provide the best decisions by simultaneously considering them. In this paper, we represent a mix in teger mathematical programming model that simultaneously considers facility location, the amount of inventory of each part in each facility, assignment of customers to facilities and traortation mode of each order to its customer and finally take the best decisions. Moreover, there is a service level constraint for each facility that ensures a certain time-based service level. Also when a facility is out of stock of any part, this model allows it to satisfy its assigned customer’s orders by an emergency shipment from the central warehouse. Considering the complexity and largeness of solution for this problem, a genetic algorithm has been used for solving the proposed model. For solving the proposed model by the mentioned method, initial data has been produced randomly with suitable distributions by considering the qualitative data from similar studies and real world. Finally, the proposed model has been solved by GAMS software in small size and by genetic algorithm in small, medium and large size. Also the final results from these two methods have been compared with each other in small problems and sensitivity analysis has been done on time-based service level.