Optimizing Delhivery's Midmile Logistics Network Using a Hybrid Evolutionary Search Algorithm.

This article focuses on optimizing the midmile logistics network of Delhivery, a leading Indian logistics company, by addressing a complex variant of the vehicle routing problem (VRP) characterized as a multidepot fleet size and mix site-dependent asymmetric distance-constrained VRP with time windows. The authors propose a tailored hybrid evolutionary search algorithm (HESA) integrated with Delhivery's existing systems, which automates route planning and incorporates geographically varying vehicle variable costs, heterogeneous fleets, and multiple operational constraints. Computational experiments on real-world data demonstrate that HESA achieves significant cost savings of approximately 7.3% compared to the previous process, improves vehicle utilization, and reduces execution time by 32% with a novel insertion heuristic variant (r-HESA). The algorithm has been successfully implemented at Delhivery, resulting in streamlined operations, reduced manual intervention, and environmental benefits through decreased travel distances and emissions.