The two-echelon distribution networks are very common in last-mile supply chains and urban logistics systems. The problem consists of delivering goods from one depot to a set of satellites and from there to a group of customers. The most common problem is known as the two-echelon vehicle routing problem (2E-VRP), which is known to be computationally difficult to solve. In real-life, uncertainty in travel times increases this complexity to actually define a delivery schedule. To deal with this stochastic behavior, this paper proposes a solution approach based on a combination of simulation and optimization, named simheuristic, to solve the 2E-VRP with stochastic travel times. The model also considers the CO₂ equivalent and fine particles emissions. Experiments are run using real data of a French delivery company for the city of Paris. The performance of two scenarios for freight delivering is evaluated, and the results show that a global optimum gives better results than local optima. Policy makers need to take this into account when defining city policy on freight transport.

两级分销网络在最后一英里供应链和城市物流系统中非常普遍。问题包括将货物从一个仓库运送到一组卫星，然后从那里运送到一组客户。最常见的问题被称为两梯队车辆路径问题 (2E-VRP)，众所周知，该问题在计算上难以解决。在现实生活中，旅行时间的不确定性增加了实际定义交付时间表的复杂性。为了处理这种随机行为，本文提出了一种基于模拟和优化相结合的求解方法，称为 simheuristic，以解决具有随机旅行时间的 2E-VRP。该模型还考虑了 CO ₂等量和细颗粒物排放。实验是使用一家法国快递公司在巴黎市的真实数据进行的。评估了两种货运场景的性能，结果表明全局最优比局部最优给出更好的结果。政策制定者在制定城市货运政策时需要考虑到这一点。