Abstract: To aestablish the initial collection point to facilitate the customers,increase the velocity of prod uctreturns and establish the centralized return center where returned products from retailers or end-customers were collected,sorted,and consolidated into a large shipment destined for manufacturers'or distributors'repair facilities,it is a saving and effective mode of multi-level producet returns.Based on the existing research,this paper considers more variables:initial collection cost,revenue of sell two-hand products selling,revenue of reclaimed products selling,then proposes a three level reverse logistics net work design Optimization model that can connect the customs,centralized return center and manufacturers.Ultimately,sensitivity analysis indicates that genetic algorithm is a valid approach to resolve this kind of problems.

Key words: reverse logistics network design, genetic algorithms, initial collection points, location allocation