In this paper, a closed-loop location-inventory problem for spare parts is presented. The proposed supply chain network includes two echelons, namely (1) distribution centers (DCs) and repairing centers (RCs) and (2) operational bases. Multiple spare parts are distributed among operational bases from distribution centers in the forward supply chain and failed spare parts from operational bases are sent back to the repairing centers to receive multi-modal repair services in the reverse supply chain. The problem is formulated as a mixed-integer nonlinear programming model for the location-allocation problem that tries to choose which DCs and RCs to be opened and to determine the repair service mode of each failed items with objective function minimizing the total cost. Furthermore, the validation of the model is tested by GAMS software for small-sized problem, and particle swarm optimization (PSO) is proposed to solve large-sized problems in a rescannable time. Finally, several sensitivity analyses are presented to evaluate the proposed model.