Waste collection is a difficult problem faced by modern society. Environmental issues and increasing population resulting in more waste production and traffic congestion are examples of negative external effects. This thesis provides two useful models related to the waste collection problem. First, the insertion heuristic is used to solve the Vehicle Routing Problem (VRP). Second, the results from the VRP are used to determine where sensors should be implemented in order to decrease the number of overflow events. The VRP program creates routes for each borough of the city of Rotterdam separately, for every day in one year. Waste accretion is modelled in order to make lists of locations that are eligible to be emptied. This modelling of waste accretion is done using data of inhabitants, waste production and the locations of the waste bins. Using two thresholds, a list with urgent locations and another list with eligible but not urgent locations is constructed. The starting point for a route is the location with currently the highest (estimated) amount of waste. The nearest insertion heuristic is then used to expand the routes until the vehicle’s capacity is reached. Using different numbers of vehicles, different results are obtained. When there are more vehicles available than needed, the program does not use all of them. Overflow occurs less when more vehicles are available. The overflow events are more likely to take place at locations that are either located very far away from their neighbours, have a high standard deviation or a large daily waste accretion. To improve the results, extensions have been added to the VRP including assigning an alternating number of vehicles to a borough and using the forecast of waste accretion to determine which locations should be visited. Locations with a lot of overflow can be equipped with a sensor, meaning that the actual amount of waste in the bin is known and can be used when creating routes. Results show that significant improvements can be observed when sensors are placed in certain locations, especially when combined with the forecast extension to the VRP. Using the results of this thesis in further work, for example with the addition of a cost-benefit analysis, can lead to useful recommendations for municipalities and waste collection companies.

, , , ,
Dollevoet, T.A.B.
hdl.handle.net/2105/45933
Econometrie
Erasmus School of Economics

Rentier, A.M. (2019, February 19). Improving the Waste Collection Problem using Sensorized Containers.. Econometrie. Retrieved from http://hdl.handle.net/2105/45933