Flexible, Dynamic, and Collaborative Synchromodal Transport Planning Considering Preferences

Abstract

Freight transport faces a threefold challenge of limited resources, increasing demand for efficient goods movement, and the pressing need to meet ambitious emissions reduction targets in ever shorter timelines. To address these challenges, the industry requires urgent innovation and the adoption of new technologies and logistics systems to change the way goods are transported. The use of intermodal transport has been developed due to the need for efficient, cost-effective, and sustainable freight transport. However, the current state of intermodal transport still faces various barriers to its utilization, such as a lack of flexibility, delays caused by uncertainty, and a lack of cooperation among transport actors. The proposal of synchromodal transport aims to address these barriers. Synchromodal transport represents an advanced form of intermodal transport that dynamically adapts routes and modes while optimizing resource utilization through synchronization and collaboration. Despite the recognition of synchromodal transport as a promising solution, there are still unaddressed gaps in the transport planning field, including the need for flexible, dynamic, preference-based, and collaborative planning. This thesis aims to fill these gaps through the development and evaluation of a series of innovative approaches, which are tested and validated using real-world transport networks. The goal is to advance the field of synchromodal transport planning, enabling the provision of flexible, reliable, and sustainable services that meet the needs of stakeholders.

In order to investigate the potential of flexibility, this thesis presents a mathematical model and a heuristic algorithm (Adaptive Large Neighborhood Search, ALNS) for the simultaneous routing of shipments and vehicles. The proposed approach enables flexible routing and scheduling of vehicles, improving the overall efficiency of the transport system in a static setting as a proof of concept. The results of numerical experiments demonstrate that implementing the proposed approach with flexible services can result in 14% reduction in costs compared to existing methods that do not consider flexibility.

In dynamic planning, this thesis tackles the issue of service time uncertainty in synchromodal transport by using an online Reinforcement Learning (RL) approach, assisted by the ALNS algorithm. The proposed model-assisted RL integrates RL and ALNS to leverage the data-driven strengths of RL and the domain knowledge of ALNS. In this way, the model-assisted RL addresses the "curse of dimensionality" caused by the large state space and complex actions in synchromodal transport. The RL approach dynamically adapts to unexpected events that cause uncertainty by learning from real-time data collected from transport operators, terminal operators, and sensors, without requiring any prior information. The proposed approach was tested in various scenarios that included disturbances, disruptions, and a combination of different types of events, and was found to perform better than traditional waiting and average duration strategies in reducing delay, waiting time, cost, and emissions.

When it comes to preference-based planning, this thesis addresses the challenge of incorporating the heterogeneous and vague preferences of shippers and carriers. To account for carriers' preferences, a multi-objective optimization model that incorporates weight intervals is proposed to handle vague preferences. The model generates a Pareto frontier of solutions that best reflects the carriers' preferences, allowing them to make informed decisions. For shippers' preferences, the thesis employs multiple attribute decision-making and fuzzy set theory to address the heterogeneity and vagueness of preferences, respectively. The results demonstrate that incorporating preferences results in improved satisfaction among shippers by providing solutions with preferred attributes on cost, time, emissions, risk, and delay. By improving shipper satisfaction, carriers can benefit from increased customer loyalty and retention, leading to a competitive advantage in the market. Moreover, by considering various attributes, such as cost, time, emissions, risk, and delay, the model can help carriers make more informed and sustainable decisions, leading to improved environmental performance and compliance with regulations. Overall, incorporating preferences in planning can result in a win-win situation for both shippers and carriers, leading to improved operational performance and a sustainable competitive advantage.

In collaborative planning, this thesis examines the benefits of horizontal collaboration among carriers through the sharing of requests and the consideration of eco-labels. The thesis presents an auction-based mechanism to facilitate collaboration and enable distributed planning. Results indicate that this approach leads to increased request fulfillment, improved sustainability, and reduced costs compared to centralized and non-collaborative planning approaches. On the tested instances, the collaboration between carriers can result in significant increases in the proportion of served requests, with gains of 48% and 11% for synchromodal and unimodal carriers, respectively. Additionally, by taking into account eco-label preferences, the use of the highest or mixed eco-labels can lead to emissions reductions of up to 70% and 15%, respectively, compared to ignoring preferences. Compared to synchromodal carriers, unimodal carriers, especially truck carriers, need to share more requests in collaborative planning to reduce the overall cost. From a policy-making perspective, policymakers can take steps to promote the development of synchromodal transport by implementing incentives for collaborative planning and utilizing eco-labels to achieve sustainable synchromodal transport solutions.

In summary, this thesis provides solutions to address the gaps in synchromodal transport planning by proposing innovative mathematical models and algorithms. These methodologies aim to increase the flexibility, reliability, and sustainability of transport services while also reducing cost, time, emissions, and delay. Additionally, the proposed methodologies consider the preferences of both shippers and carriers, promoting a collaborative and eco-friendly approach to transport planning. The numerical experiments and case studies demonstrate the effectiveness and superiority of the proposed approaches compared to existing methodologies.