Waste Management Model with Timed Colored Petri Nets

Abstract

Waste management is a key element in the functioning of modern cities.This paper presents a new model of a waste collection system, described as a discrete-event system (DES), implemented using Timed Colored Petri Nets (TCPNs) in combination with an integrated Python server. The model is developed with consideration of container filling dynamics and variable routes, which ensures alignment with real urban conditions.A key element of the developed model is the interface of a vehicle routing problem with capacity constraints, multiple trips, and time windows (MTCVRPTW), which enables vehicles to service containers multiple times during a scheduled period while adhering to volume and time restrictions. The model supports configuration of parameters such as operational delays, container filling and overflow volumes, and vehicle load capacity. The simulation is implemented in CPN IDE using time series as input data, partitioned for efficient processing. Information about container filling levels and road conditions is periodically updated during real-time simulation, enhancing scalability and performance. The model generates event logs—movement, unloading, overflow, and servicing—which are processed by Python scripts to calculate performance metrics.The main performance metrics of the waste collection system were defined, including route distance and time, unloading efficiency, container overflow volume, servicing efficiency, and deviations of planned routes from the schedule.To demonstrate the operation of the model, an experiment was conducted using synthetic data approximating real-world conditions. The locations of 10 containers, unloading points, and depots were determined using the Google My Maps service based on coordinates of real objects in Kyiv. Realistic route distances and travel times were generated using the Google Distance Matrix API. The MTCVRPTW algorithm for two vehicles scheduled two trips per week according to static routes. The simulation of the model generated event logs, which were then used to calculate performance metrics. The analysis of these metrics revealed significant limitations of static route planning and highlighted the need for adaptive strategies that account for the actual state of containers and traffic.The proposed model is a flexible tool for evaluating, analyzing, and improving waste collection strategies in cities.