Department of Engineering Sciences, Universidad Andres Bello, Quillota 980, Viña del Mar, Chile.
Waste Manag. 2020 Oct;116:179-189. doi: 10.1016/j.wasman.2020.07.027. Epub 2020 Aug 14.
This study proposes a design of a household waste collection system based on a two-stage procedure. First, the bin location-allocation problem is solved by selecting collection sites from a set of potential sites, and determining the type and number of bins at each selected collection site. Second, bin-to-bin waste collection routes are obtained for a fleet of homogeneous vehicles that are restricted by either work shift duration or vehicle capacity. Mixed integer linear programming (MILP) models are proposed for both stages, considering the particular characteristics of the problem. The models are applied to a real-world instance in the commune of Renca in Santiago, Chile. The results of first stage indicate an important preference for small bins since they have a lower unitary cost. Due to the large size of the real instance, a Large Neighborhood Search (LNS) heuristic is used in the second stage to find good feasible vehicle routing solutions in a reasonable period of time. The results for the routing phase suggest a larger number of routes in the morning work shift since these routes have shorter distances. The LNS heuristic presents a satisfactory behavior when compared to the MILP model with small instances. The proposed bin-to-bin household waste collection vehicle routing presents a more efficient solution than the existing door-to-door waste collection in the commune of Renca with respect to the total daily traveled distance and the average work shift duration. Finally, a sensitivity analysis is presented and discussed for both models.
本研究提出了一种基于两阶段程序的家庭垃圾收集系统设计。首先，通过从一组潜在地点中选择收集地点，并确定每个选定收集地点的垃圾桶类型和数量，解决垃圾桶位置分配问题。其次，为一组同质车辆获得垃圾桶到垃圾桶的废物收集路线，这些车辆受到工作班次持续时间或车辆容量的限制。针对该问题的特定特点，提出了用于两个阶段的混合整数线性规划 (MILP) 模型。这些模型应用于智利圣地亚哥伦卡公社的一个真实实例。第一阶段的结果表明，由于其单位成本较低，小垃圾桶受到了很大的青睐。由于实际实例的规模较大，在第二阶段使用了大邻域搜索 (LNS) 启发式算法，以便在合理的时间内找到良好的可行车辆路径解决方案。路由阶段的结果表明，由于这些路线的距离较短，因此在早上工作班次中有更多的路线。与小实例的 MILP 模型相比，LNS 启发式算法的表现令人满意。与伦卡公社现有的上门收垃圾相比，所提出的垃圾桶到垃圾桶的家庭垃圾收集车辆路径在总每日行驶距离和平均工作班次持续时间方面提供了更有效的解决方案。最后，对两个模型进行了敏感性分析和讨论。
