Optimization of distribution routes in an urban context with realistic constraints

The present study was conducted in the context of real data from three important regions. The results obtained from the implementation of algorithms in each region, numerical comparison of performance indicators, stability analysis of algorithms and factors interpretation affecting the differences are discussed. The routing data of this study were collected from three regions of the large city of Tehran and include variables such as distance between points, order timing, and prioritization of goods delivery. One of the most important results is that there is no single metaheuristic algorithm that has optimal performance in all conditions. Rather, depending on the network structure, density, type of constraints and desired goals, different algorithms can provide better performance: Genetic algorithm has stable and accurate performance in high-density environments with complex priorities. ACO has very favorable results in large and diverse urban networks with the ability to simultaneously explore parallel routes. PSO, despite its simplicity of structure, is very suitable for problems with small dimensions and the need for short execution time. This performance variation confirms that in the design of distribution systems, the algorithm selection model should be made as appropriate. Indicators such as "total distance traveled", "number of successful deliveries with high priority", "algorithm execution time", and "performance stability over multiple iterations" have been used. Also, ACO, PSO and GA are all algorithms with a wide range of applications and the results of this study can be generalized to other distribution systems, hospital logistics, electric fleets, and even drones.