Putting the ping into online shopping: Dynamic Heuristics for Vehicle Routing and Slot Offering

Randall, Matt and Letchford, Adam and Kheiri, Ahmed (2024) Putting the ping into online shopping: Dynamic Heuristics for Vehicle Routing and Slot Offering. PhD thesis, Lancaster University.

Text (2024randallphd)
2024randallphd.pdf - Published Version
Restricted to Repository staff only until 25 November 2025.
Available under License None.
Download (18MB)

Abstract

This thesis deals with a real-world optimisation problem arising within the context of attended home delivery of groceries. Customers must choose a delivery slot during which they are able to accept their delivery, and retailers must ensure that each customer is served during their chosen slot. The problem is a stochastic dynamic combinatorial optimisation problem, with three main aspects: vehicle routing, slot offering, and depot allocation. After a literature review in Chapter 2, we consider a sequence of problems of in- creasing complexity, gradually including more aspects of the real-life problem. Chapter 3 introduces a simple dynamic vehicle routing problem, in which we have customer requests arriving whilst routes are being planned. Here we explore the use of insertion heuristics, as well as the application of local search. We find that a specific parallel insertion heuristic, based around a certain “sum-of-squares” insertion criterion, yields particularly strong results. Chapter 4 proposes an approach to incorporate the use of past data into the model from the previous chapter. This is done through the use of k-means clustering to identify “centroids” of customer location clusters, which are then used to “seed” vehicle routes. Chapter 5 focuses on a dynamic vehicle routing and slot offering problem, building on the work in Chapter 3. We focus once again on heuristics, making use of paral- lel insertion and local search, and once again see the merits of the “sum-of-squares” approach. Chapter 6 generalises our work in Chapter 5 to a multi-depot setting. Once more, we experiment with various insertion and local search heuristics, but we also consider heuristics for allocating customers to depots. Finally, some concluding remarks are made in Chapter 7.