Lieferzeitfenstermanagement für Freihauslieferungen: Integration von Revenue Management und Vehicle Routing für stochastische und dynamische Steuerung

RWTH Publications (RWTH Aachen)

Abstract

The ongoing growth in worldwide e-commerce sales yields an increased relevance of attended home deliveries. Products like groceries or fresh flowers require the customer to be present on delivery. Challenges like great cost pressure and high service quality expectations motivate the need for research in regard to respective e-fulfilment operations. The thesis focusses on operational management of delivery time slots for attended home delivery services. That is, it tackles the stochastic, dynamic control problem of assigning time slots to order requests on order acceptance. To this end, order acceptance anticipates the impact of time slots on delivery planning. The thesis proposes an interdisciplinary approach, integrating revenue management and vehicle routing techniques for effective control. Three main research topics lay at the heart of the thesis. First, the thesis contributes alternative approaches tackling the so-called dynamic slotting problem. Given limited delivery resources, controlling the availability of delivery time slots for arriving order requests serves to maximise overall accepted revenue for a delivery period. To this end, modelling customers' choice behaviour enables effective control while increasing problem complexity. The approaches rely on the anticipation of delivery schedules and future acceptable value when making a decision within milliseconds on request arrival. Thus, anticipated delivery schedules result from solving team orienteering problems in advance of the order horizon. Similarly, value approximations result from simulation-based, approximate dynamic programming. The computational study includes a comprehensive analysis of various problem settings, providing decision support in approach selection for empirical problem scenarios. The second topic considers multi-criteria decision making in dynamic slotting to account for other criteria than short-term profit. It introduces a generic approach that allows for an individual set of criteria. New criteria are integrated via constraints or a weighted sum approach, requiring multiple simulation experiments on alternative settings to inform a decision-maker when selecting the final policy. Ann exemplary simulation study examines the criteria visibility via branded trucks and customers' social influence. The research motivates further work in multi-criteria decision making to increase the practical relevance of research contributions in operational time slot management. Third, the thesis focusses on enhancing research practice in e-fulfilment for attended home delivery services by contributing a simulation-based benchmarking system for the interdisciplinary research field. The system provides a pool of benchmarks, a holistic, application-specific data model, and a modular, easily extendable architecture. The thesis discusses a collaborative research process which results in the further development of the system in a community effort. As a result, the system can enable a trend towards more rigorous and relevant e-fulfilment research. The thesis puts its contributions in perspective within a broader research context. Moreover, it provides a discussion in terms of limitations and future research.

Authors 1

1. Magdalena A. K. Lang corresponding Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen