The explosive growth of online agricultural product purchases in China following the 2020 COVID-19 outbreak highlighted the need for efficient logistics, particularly for perishable goods. Prepositioned warehouses, strategically located closer to consumers than traditional distribution centers, offer a solution. This study focuses on the optimization of a three-level logistics network incorporating prepositioned warehouses under an omni-channel retail model, which integrates both online and offline sales. The omni-channel nature introduces complexities including diverse consumer shopping behaviors, the need to maintain product freshness, and stringent time constraints for delivery. Existing logistics network layouts often fall short in addressing these characteristics, leading to inefficiencies and potential losses for businesses. This research aims to address the optimal location of prepositioned warehouses, inventory management strategies, and efficient delivery routes to minimize both total cost and delivery time. The study utilizes real-world data from Beijing's Haidian District to validate the proposed model and offer actionable managerial insights.

The literature review covers three key areas: omni-channel retailing, prepositioned warehouses, and agricultural logistics networks. Omni-channel retailing research emphasizes the integrated shopping experience and operational benefits across multiple channels. However, studies on the specific challenges of integrating warehouses and distribution within this model, particularly for perishable goods, are limited. Prepositioned warehouse research is relatively nascent, with existing studies focusing on individual aspects like site selection or inventory strategies without fully integrating them within a comprehensive omni-channel framework. Agricultural logistics network research primarily focuses on cost and time minimization, often incorporating freshness penalties for perishable products. This study bridges the gap by integrating these three areas to develop a comprehensive model that considers the unique challenges of omni-channel retailing for perishable agricultural products within a three-level logistics network.

This study develops a multi-objective mixed-integer planning model based on the location-route-inventory problem (LRIP) in city logistics. The model considers a three-level supply chain structure: regional distribution centers, prepositioned warehouses, and consumer groups. The model incorporates the following key features: 1. **Omni-channel Retailing:** The model accounts for both online and offline consumer demand, allowing consumers to choose between in-store purchases and online deliveries depending on the proximity of a prepositioned warehouse. 2. **Perishable Goods:** A freshness penalty cost is incorporated into the model to reflect the deterioration of agricultural products during storage in prepositioned warehouses. 3. **Time Minimization:** Besides cost minimization, the model also aims to minimize total delivery time, reflecting the importance of timely delivery in the omni-channel setting. 4. **Two-Stage Transportation:** The model considers two stages of transportation: (1) regional distribution centers to prepositioned warehouses (using cold chain trucks) and (2) prepositioned warehouses to consumer groups (using electric bicycles). 5. **Inventory Management:** The inventory strategy for prepositioned warehouses follows the classical Economic Order Quantity (EOQ) model. The model's decision variables include the location of prepositioned warehouses, replenishment quantities and cycles, delivery routes, and allocation of consumer demand to warehouses. The model comprises two objective functions: minimizing total cost and minimizing total delivery time. These are combined into a single objective function through normalization and linear weighting. The model is solved using the Gurobi solver, combining its built-in heuristic algorithm for an initial feasible solution and the branch-and-bound algorithm to refine the solution.

The model was applied to a real-world case study of an e-commerce enterprise in Beijing's Haidian District. Key findings include: 1. **Optimal Warehouse Locations:** Four optimal locations for prepositioned warehouses were identified from ten potential sites. The model strategically positioned these warehouses to effectively serve the high-demand areas in the southeast of Haidian District. 2. **Inventory Strategy:** The EOQ model determined optimal replenishment quantities and cycles for each prepositioned warehouse. Replenishment cycles were generally within two days, ensuring product freshness. The high loading rates (over 90% for most replenishment routes) indicate efficient utilization of transportation resources. 3. **Distribution Route Optimization:** Efficient delivery routes were planned for electric bicycles, ensuring that most deliveries were completed within one hour. The routes were optimized to minimize both travel time and cost. 4. **Total Cost and Time:** The total cost of the optimized logistics network was calculated as ¥112,228.14, consisting of site selection cost (¥80,000), inventory cost (¥21,380.97), and transportation cost (¥10,847.17). The total delivery time was 18.8 hours. 5. **Sensitivity Analysis:** Sensitivity analysis revealed that: * Increasing the proportion of consumers choosing offline shopping (β) reduces both total cost and total time. * Increasing the daily freshness penalty cost (Cf) increases total cost but has a negligible impact on total time. This suggests that quality control and minimizing product spoilage are crucial for cost reduction. * Increasing the weight assigned to time minimization (α) in the multi-objective function decreases total time but increases total cost, illustrating the trade-off between cost and time efficiency.

The findings demonstrate the effectiveness of the proposed model in optimizing a three-level logistics network for prepositioned warehouses under an omni-channel retail model for agricultural products. The model successfully integrates the complexities of omni-channel operations, perishable goods, and time sensitivity. The results highlight the importance of considering consumer behavior, product perishability, and the trade-off between cost and time efficiency when designing logistics networks. The case study results are consistent with the observed characteristics of omni-channel prepositioned warehouse operations. The sensitivity analysis provides valuable managerial insights, emphasizing the need for accurate assessment of consumer preferences, effective quality control, and careful consideration of the relative importance of cost versus time efficiency. This model offers a valuable tool for agricultural retail businesses seeking to improve efficiency and profitability.

This study contributes a comprehensive optimization model for the logistics network of prepositioned warehouses under an omni-channel retail model, tailored for perishable agricultural products. The model successfully integrates various factors, including consumer behavior, product perishability, and time constraints. The findings provide practical recommendations for businesses, highlighting the importance of understanding consumer behavior, maintaining product quality, and balancing cost and time efficiency. Future research could expand on this model by incorporating more sophisticated inventory models, considering supplier dynamics, and developing more nuanced freshness penalty functions based on product-specific characteristics. Furthermore, exploring integration with logistics organization and information networks would enrich the model's scope.

The model assumes a static demand pattern and does not consider dynamic fluctuations in consumer demand. The freshness penalty cost is simplified as a constant parameter, ignoring potential variations in spoilage rates across different products or storage conditions. The study focuses on a single case study; applying the model to other regions or contexts would provide further insights into its generalizability. Lastly, the model does not explicitly incorporate aspects of logistics organization and information networks, which could influence operational efficiency.