Open Access

David Fleckenstein

• In a wide variety of business models, logistical service providers must decide on which offers are made to requesting customers and determine a feasible route plan for the fulfillment of orders. Thereby, the objective is often to maximize revenue net of routing cost. Due to structural similarities, the application-specific problems are part of a common family of optimization problems, termed integrated demand management and vehicle routing problems (i-DMVRPs). The cumulative dissertation at hand analyzes i-DMVRPs both from a general, cross-application perspective as well as by focusing on rural shared mobility-on-demand (SMOD) services as one specific business model. The latter is unique due to the high relevance of sustainability objectives. Overall, the dissertation predominantly addresses the demand management subproblem by means of operations research methods. The dissertation consists of six articles. In Article A1, we present the first cross-application survey of the i-DMVRPIn a wide variety of business models, logistical service providers must decide on which offers are made to requesting customers and determine a feasible route plan for the fulfillment of orders. Thereby, the objective is often to maximize revenue net of routing cost. Due to structural similarities, the application-specific problems are part of a common family of optimization problems, termed integrated demand management and vehicle routing problems (i-DMVRPs). The cumulative dissertation at hand analyzes i-DMVRPs both from a general, cross-application perspective as well as by focusing on rural shared mobility-on-demand (SMOD) services as one specific business model. The latter is unique due to the high relevance of sustainability objectives. Overall, the dissertation predominantly addresses the demand management subproblem by means of operations research methods. The dissertation consists of six articles. In Article A1, we present the first cross-application survey of the i-DMVRP literature. It includes a generalized problem definition formalized by a high-level mathematical model, an extensive description and classification of solution concepts and algorithms, and a discussion of future research opportunities based on the existing scientific advances. In Article A2, we provide a formal definition of opportunity cost in the context of i-DMVRPs and use it to prove the general validity of several mathematical properties. These theoretical findings give rise to novel types of approximation approaches that exploit them, which we demonstrate in a numerical study considering many different settings for a stylized problem. Focusing again on opportunity cost approximation, Article A3 presents an explainability technique that allows the characterization of fundamental types of approximation errors by measuring influencing factors and drawing on reward decomposition. For the identified error types, we compile evidence from existing computational studies and discuss algorithmic elements to limit their impact. With Article A4, the focus of the dissertation narrows toward rural SMOD services as a crucial application area for i-DMVRPs. Descriptively analyzing a large real-world data set, we identify demand patterns resulting from an extended booking horizon, which is a typical feature of rural SMOD services. Based on these empirical findings, the article suggests designs for appropriate demand management approaches that exploit opportunities and mitigate negative effects of these patterns. In Article A5, we present a methodology for evaluating different types of demand management that a provider could potentially apply. It is based on a semi-perfect information model, which is a variant of the exact Markov decision process of the provider’s i-DMVRP. The computational results derived from a real-world data set show that there is a considerable potential for performance improvements but also a need to balance the relevant sustainability objectives because there are trade-offs among them. In Article A6, we propose a sustainability-oriented demand management approach that is capable of balancing all sustainability objectives, which we identify using multi-attribute decision analysis. Its basic idea is to optimize prices dynamically such that booked passenger kilometers are maximized. The secondary objectives are also considered in the form of constraints that require the price of any ride to be at least equal to its marginal cost. Comparing our approach to benchmarks representing the state-of-the-art in practice and in the scientific literature, we show its benefits both from the provider’s and the customers’ perspective.…