To deal with the asymmetric routes in unban delivery, a two-index vehicle flow model of CVRP was setup and solutions by two algorithms were conducted on many instances. Sensitivity analysis showed that the difference from asymmetric matrix leads to substantial change of objective value and optimal base, which implied that the asymmetric and symmetric problems are, indeed, two different problems. Using the asymmetric matrix from e-map service, the mileage of routes can be effectively shortened. The asymmetric matrix better meets the current logistics requirements and hence more practical and logical than the symmetric one. It is concluded that the node location, demand and routing should be synthesized for urban delivery to balance logistics cost and service.%在建立路径问题二指数流量模型的基础之上，结合非对称性矩阵的变化进行灵敏度分析，并运用两种算法得到求解结果。多个实例的比较分析表明，非对称性问题与对称性问题是两个不同的问题，系数矩阵的差异最终带来目标函数值和基解的差异。城市物流配送实际案例检验表明，运用电子地图路线所得的非对称性距离矩阵，能有效缩短车辆路径总里程；相对于对称性矩阵车辆路径，非对称性距离矩阵车辆路径更符合当前物流的需求和实际情况，更具合理性。因此，城市物流配送应综合考虑各节点需求量和车辆选型，以平衡物流成本与服务要求。
展开▼
