Abstract: The joint location-allocation optimization problem for fresh chain enterprises arising in new retail is studted. A nonlinear mixed integer programming model is constructed simultaneously considering location and coverage of multi-format stores, configuration of cold chain facilities, and cold category selection. A hybrid Lagrangian relaxation algorithm is also put forward to solve the model. Results comparisons with CPLEX verify the effectiveness of our algorithm. Based on the actual data of the leading fresh chain enterprise (Guolin) in Chongqing, the optimization results are demonstrated according to the model and algorithm of this paper. Then sensitivity analysis are carried out to explore the impacts of influence factors on urban distribution network and logistics cost, such as demand size, consumer willingness to pick up in a store, online order size and weather changes. The results demonstrate that compared with Guolin's existing distribution network, the optimization logistics cost reduce by 2.52%.Fresh loss cost accounts for more than 70% of logistics cost. And configuration of cold chain facilities only reduces the distribution cost by 0.32%. Demand size has less impact on the urban distribution network and unit logistics cost. Consumers' willingness to pick up in a store, online order size and weather changes have no impact on the distribution network structure and little impact on logistics cost.

Key words: new retail, urban distribution network, last mile delivery, nonlinear mixed integer programming, Lagrangian relaxation algorithm