This thesis explores the analysis, periodicity, and scalable modelling of Max-Min-Plus-Scaling systems, a versatile approach to modelling Discrete Event systems. Unlike traditional continuous-time or discrete-time systems that evolve through differential or difference equations, DE systems progress through discrete events. MMPS systems rely only on maximisation, minimisation, addition, and scaling, making them highly suitable for modelling processes with synchronisation and/or competition such as energy delivery, transportation, and manufacturing.
The work is divided into three main segments. First, a new Mixed-Integer Linear Program ming based method is developed for analysing growth rates and fixed points of general implicit MMPS systems. This extends an existing MILP formulation for homogeneous and non-expansive explicit MMPS systems, introducing adaptations for general implicit cases.
A dedicated preprocessing step and search strategy are introduced, resulting in an analysis method that significantly reduces computational requirements. Secondly, the dynamical and stability behaviour of periodic MMPS systems with periods greater than one is examined.
A new canonical form is proposed, enabling the use of existing analysis tools on periodic systems, along with a method for determining the stability of periodic orbits. Thirdly, a modelling framework for transportation systems is introduced, featuring a connectable, node based toolbox and an algorithm that transforms high-level system descriptions into sets of equations.
All developed methods, theories, and tools are demonstrated on a real-world 4-node transportation system. The results confirm the efficiency of the new MILP approach, reveal periodic behaviour and stable periodic orbits, and highlight fixed points, all within the proposed transportation network framework.

Data-driven Inverse Optimization (IO) is a form of Supervised Learning where it is assumed
that the output data is found by means of an optimization problem that depends on the input
data. IO uses this data to approximate the optimization problem as best as possible. In the
case where one wants to emulate an expert operating in a dynamic environment, the dataset
obtained by measuring the expert is often contained in a small, optimal part of the total state
space of the environment. When a model trained on this data finds itself in a different part of
the state space, it can behave erratically.

In this thesis, we will combine Inverse Optimization with the active learning method of Dataset Aggregation (DAgger) to test if this improves model performance in dynamic settings. DAgger is an iterative process where the system is steered by the learner, creating new input data for the expert to find the best actions. This new data is then used to train a new model.

Furthermore, we propose a new algorithm, fast-DAgger, that should converge faster than the
DAgger algorithm, at the possible cost of performance in the final model.

IO models trained with the DAgger and fast-DAgger algorithms are tested and compared to
IO models trained on static datasets. This is done for two case studies: the Dynamic Vehicle
Routing Problem as proposed by the EURO meets Neurips 2022 Vehicle Routing Competition,
and the game of Tetris.

Results show the potential of combining IO with DAgger. However, DAgger is not always
better than training with a static dataset. DAgger can only be helpful when the static training
data is limited to a part of the total state space and when this data does not generalize well to the total state space. The fast-DAgger algorithm did not show a significant speed-up compared to the normal DAgger algorithm in the case studies. However, this is very dependent on the specifics of the model and the hyperparameters of the DAgger algorithm.

Freight transport plays a critical role in supporting global trade, with multimodal transport systems modelling gaining importance due to their potential to optimize efficiency and sustainability. Accurately modeling these multimodal freight chains is essential for infrastructure planning and policy-making. Yet, it remains a persistent challenge due to fragmented datasets, limited granularity, and the absence of observed multimodal chain-level data. Traditional modeling approaches, particularly heuristic-based methods, often struggle to incorporate real-world operational constraints such as port selection logic and cargo handling requirements. Moreover, these models are typically inflexible and computationally intensive.

This research seeks to develop an adaptive multimodal freight chain model that addresses these limitations. Specifically, it introduces a practical path construction framework that integrates port selection based on geographic and functional suitability, aligning cargo handling requirements with port capabilities during the construction of mode chains. This research also tries to address a gap in the literature by applying machine learning to the estimation of multimodal freight flows, a domain traditionally dominated by heuristic and optimization-based methods. To estimate freight demand distribution across the generated chains, this study explores the use of machine learning, particularly the Expectation-Maximization (EM) algorithm, to leverage the abundant but often unstructured transport data available. The EM model enables demand share prediction without relying on labeled training data, reducing the calibration burden and enhancing model responsiveness to observed transport flows.

The proposed modeling framework is applied to a case study based on the NEAC Mode Chain Builder system for inter-country freight movements between the Netherlands and Belgium, two countries with high multimodal connectivity and the largest ports in Europe. The results demonstrate the model’s ability to generate valid mode chain alternatives and to significantly reduce deviations between predicted and observed freight flows, particularly for sea and rail segments. While some deviation increases occur in other segments, these are outweighed by the overall improvement in prediction accuracy. The EM model also shows stable convergence behavior, confirming its potential under data-limited conditions. However, residual deviations suggest that external factors, such as data incompleteness or behavioral uncertainties, still limit full accuracy.

This study highlights the potential of combining graph search algorithms with unsupervised learning to enhance multimodal freight chain modeling, especially under data-constrained conditions. It contributes both a methodological and practical solution for building data-driven multimodal freight transport models that better reflect operational realities and observed empirical data that can be used to improve the freight transport planning and decision-making process.

We present Sadcher, a real-time task assignment framework for heterogeneous multi-robot teams that incorporates dynamic coalition formation and task precedence constraints. Sadcher is trained through Imitation Learning and combines graph attention and transformers to predict assignment rewards between robots and tasks. Based on the predicted rewards, a relaxed bipartite matching step generates high-quality schedules with feasibility guarantees. We explicitly model robot and task positions, task durations, and robots’ remaining processing times, enabling advanced temporal and spatial reasoning and generalization to environments with different spatiotemporal distributions compared to training. Trained on optimally solved small-scale instances, our method can scale to larger task sets and team sizes. Sadcher outperforms other learning-based and heuristic baselines on randomized, unseen prob-
lems for small and medium-sized teams with computation times suitable for real-time operation. Wealso explore sampling-based variants and evaluate scalability across robot and task counts. To address performance limitations identified for large-scale problem instances, we experiment with using Reinforcement Learning (RL) to fine-tune the imitation-learned model. Both discrete and continuous RL formulations are explored, leveraging Proximal Policy Optimization (PPO). This allows for training on larger problem instances, for which optimal solutions for IL are infeasible to obtain. Our RL experiments provide insights into the comparative advantages and trade-offs of IL and RL methods. In addition, we release our dataset of 250,000 optimal schedules to facilitate future research. We include a detailed description of the instance generation and the Mixed-Integer Linear Programming (MILP) formulation used to solve them optimally. Furthermore, this thesis contributes a lightweight
simulation environment with visualization tools for benchmarking different task assignment algorithms.

This research investigates the value of transport mode flexibility in OCCUS supply chains, particularly under uncertain CO2 supply during the early phases of CCUS development. This study aims to develop a strategic decision support model that quantifies the economic and environmental benefits of transport flexibility within the supply chain.
The literature review provides two key insights. First, it identifies the essential steps in the CO2 sup- ply chain: CO2 is captured onboard ships, temporarily stored onboard in solvent, and transported to onshore facilities for regeneration and liquefaction, after which the liquefied CO2 is transported to per- manent underground storage. Second, the review reveals that no comprehensive studies currently model the full OCCUS supply chain while incorporating uncertainty in CO2 supply. Consequently, no established approaches exist to address transport flexibility under such uncertainty within this context. However, real options analysis has been successfully applied in land-based CCUS projects to value in- vestment and operational flexibility under uncertainty. Building on this proven methodology, the present research adopts a real options approach to quantify the value of the option to switch between transport modes.

This research applies the developed real options model to a case study centered on the Port of Rotter- dam. The supply chain model follows the Value Maritime approach: CO2 is captured onboard ships, stored in CO2 -rich solvent and offloaded at the Maasvlakte terminal. Transport from the port to the re- generation and liquefaction facility is done with containerized trucks, with the option to switch to barges. Two barge types are considered: the smaller CEMT-IVa and the larger CEMT-Va or a combination of the two. The LCO2 is subsequently transported by truck to an underground storage site. The model evaluates scenarios under both a fixed average CO2 price (€136/t) and a variable CO2 price increasing over time based on market forecasts.

The results indicate that while the average CO2 price is insufficient to achieve economic viability at any time and outcome, incorporating the option to switch from truck to barge transport adds value if CO2 supply grows. The option to switch to the smaller CEMT-IVa (€630345) barge shows greater economic benefits compared to the larger CEMT-Va (€131555), mainly due to its better alignment with expected supply volumes during the early implementation phase. The combined switching option (€632010) only yields a marginal additional value, as the larger barge is only required at the highest and least probable supply scenario.

Under the variable CO2 price scenario, the truck-only strategy reaches a positive total value by 2031 with a 30% probability. Introducing the option to switch to the smaller CEMT-IVa barge accelerates this to 2029 with a 55% probability, reflecting earlier and more frequent switching. The larger CEMT-Va barge lags behind, with switching and positive value only occurring from 2030 onward and at a lower 11% probability, indicating less frequent and delayed use.

The break-even price for the truck-only transport strategy is €184.21/tCO2 . The inclusion of switch- ing options reduces this threshold across all configurations: the CEMT-IVa barge option achieves a 4.35% reduction to €176.19/tCO2 , while the CEMT-Va barge provides a modest 0.92% reduction to €182.52/tCO2 . The combined strategy yields the largest reduction of 4.36%, lowering the break-even price to €176.17/tCO2 .

The introduction of the shipping container revolutionised global trade by significantly reducing handling costs, improving efficiency and enabling intermodal transportation. This development paved the way for the expansion of international trade and the development of highly interconnected global supply chains. Congestion, sustainability concerns, and vulnerability to disruption have become major obstacles. Recent examples of such obstacles are the COVID-19 pandemic and the blockage of the Suez Canal. Synchromodality has been identified as a potential solution for mitigating some of these concerns.

As a relatively new concept, synchromodality has mainly been studied at a theoretical level, focusing on its definition and potential. As the concept of synchromodality seems to gain attention from a broader public, more recent research has also focused on the more quantitative side. These quantitative studies primarily model the transport planning side of synchromodality. In these studies, some aspects in the supply chain have been overlooked so far, mainly the impact on critical infrastructure such as container terminals.

This research addresses this overlooked area in the existing academic landscape. Aspects such as the loading, unloading and stacking of containers are explicitly modelled in combination with synchromodal transport planning optimisation. This allows for an assessment of how synchromodality influences container terminal operations and how constraints and the dynamics at container terminals influence the transport planning. A multi-agent system approach is used as a framework for this model, as this presents a good option to model the different stakeholders involved in synchromodal transport.

The performance was analysed based on key performance metrics, including container relocation frequency, dwell times, and cost efficiency. The findings indicate that the integration of synchromodal transport planning with container terminal operations yields significant improvements. An iterative feedback loop between the transport planning agent and terminal agents facilitates more effective decision-making, leading to feasible transport planning, smoother operations, and improved resource utilisation.

Scenario analysis yielded further interesting results in terms of how a synchromodal planner would adapt to disruptions. The two most interesting findings are a decrease in the number of transshipments and a modal shift towards faster, more flexible, but also more expensive and more polluting transport modes.

In conclusion, this research demonstrates that the integration of synchromodal transport planning and container terminal operations improves the efficiency and adaptability of synchromodal logistics networks. Through these advances, this research contributes to ongoing efforts in the planning of synchromodal transport and the optimisation of container terminals, offering valuable information for both academia and industry.

Current geopolitical tensions mean that NATO countries must now make greater efforts to deter threats to their territories, having neglected this defence task for years. In case of a major conflict, brigades and their vehicles are ordered to immediately travel to the area of conflict. This vehicle deployment is researched and modelled in a Multi-Integer Linear Programme (MILP), with the objective to minimise the makespan of the deployment, e.g. the arrival time of the last convoy. In this deployment the strategic and operational movements are modelled, meaning the starting point is the Point of Embarkation (POE) and the end point is the Staging Area (SA) or Concentration Area (CA). The model is suited for deployments that use a double modal network, being road and rail transport. Two scenarios are established, distinguishable by their transport networks and vehicle types, and they are subjected to configurations in which parameters are varied. The research is conducted in cooperation with the Royal Netherlands Army (RNLA), and serves as a handle to obtain understanding on deployments when varying certain parameters. It is observed that the usage of the rail mode mainly should be motivated by vehicle suitability, rather than makespan oriented reasoning, as it hardly improves the makespan of the deployment. Furthermore, platooning is a promising technology to implement in the deployment, with the ability to decrease the makespan by at least 11%. The influence of traffic congestion halfway the deployment is highlighted, encompassing an increase in makespan of at least 7-15%. It is found that the convoy amount and the amount of vehicles does not influence the makespan by a lot (pm1.9%). Lastly, the difference between the transport networks of the two scenarios is exposed with the use of a robustness examination, where the makespan of the first scenario improves substantially more (11.7%) than the makespan of the second scenario (0.4%). This research can be expanded in several directions, including the incorporation of additional modalities and a more comprehensive investigation into the robustness of transport networks within these deployments.

This research investigates the optimal design of a charging infrastructure network for heavy-duty battery electric vehicles (HD BEVs) in the Netherlands. It addresses the challenges posed by the limited range of HD BEVs and the need for a robust charging network to support logistics operations. The study identifies key factors influencing fleet owners' charging behaviour and develops a qualitative decision model to estimate the demand distribution between private and public charging. Additionally, an optimization model is proposed to determine the optimal locations for public charging facilities, balancing the objectives of minimising infrastructure costs and maximising the number of trucks that can recharge.

Punctuality in railway transport is a critical concern for passengers, especially during unavoidable disturbances. Therefore, a sufficiently robust timetable is necessary. This thesis proposes a new timetabling method that minimizes passenger delay as the optimization objective, aiming to create a passenger-centric robust timetable. The model is implemented in the Dutch railway network in the Eindhoven-Den Bosch-Tilburg area to verify and validate its correctness and functionality. Experimental results demonstrate that the proposed model significantly reduces passenger delays when dealing with specific disturbances, and the improvement in robustness becomes more pronounced as the severity of the disturbances increases.

The ongoing global push towards energy transformation and decarbonisation underscores the urgent need for reducing greenhouse gas emissions, particularly in the maritime sector, which accounts for a significant portion of global trade and emissions. This study focuses on inland waterway transportation (IWT) as a key area for implementing zero-emission solutions, especially through the use of battery swapping technology. Given the average age and expected lifespan of vessels in the IWT sector, the need for immediate action is critical to make significant strides towards climate neutrality by 2050. This research delves into the complexities of multi-objective optimisation, balancing investment costs with operational efficiencies. It provides an in-depth sensitivity analysis of the impact of battery capacity, battery costs, the availability of docking station spots, and loading state of vessels on the overall system configuration.

The study emphasises the importance of carefully attributing and balancing weights to each objective to optimise network design and total costs. Moreover, the findings support the continued development of higher battery capacities and strategic resource allocation to enhance the efficiency and sustainability of IWT. The significance of this research lies in its potential to inform the deployment of renewable energy-powered, zero-carbon ships, contributing to global efforts to mitigate climate change and promote sustainable development in the maritime sector. Moreover, the insights gained in this thesis inform recommendations, aimed at guiding future research.

Semiconductor product development becomes increasingly challenging due to diminishing product life cycles, miniaturization, introduction of new physical principles, and new manufacturing processes. These problems are compounded in the absence of standardized development processes for the most complex semiconductor products like MEMS technologies, because the manufacturing of these products is often outsourced. Suppliers play a pivoting role in the realization of the product, from product design until process design and ramp-up. The supplier selection problem in this industry denotes the challenges in finding the right supplier while meeting all the technical, process and business requirements. The contribution of this research is in presenting how to develop a generic methodology for data-driven co-development. Thereby, this work presents a novel product development framework that leverages co-development and supply chain integration through data-driven decision-making. Co-development is reached through standardized methods for generating the required engineering output for supplier selection. Supply chain integration is introduced in the early stages of product development. This synthesizes with the outsourced manufacturing processes. Clustering algorithms are used to effectively shortlist suppliers based on their competences, and provide insights into supplier profiles and gaps. The latter is used to draw strategies for developing unattainable technologies. Using the framework, the required engineering output for supplier selection was generated in 77% less time while reducing information asymmetries between actors in the product development process. Furthermore, the framework made it possible to quantify decisions, allowed for supplier profile recognition and gap identification through its hybrid automated approach, and supplier shortlisting in 95% less time. This efficiency does not only showcase the immediate benefits of the proposed methodology, but also lays the foundation for future research towards a fully automated approach in semiconductor product development. The developed framework includes information flows between actors and steps in the product development process, their interfaces and demonstrates its added value and potential for a fully automated yet efficient future approach in semiconductor product development.

The development of the global COVID-19 pandemic from 2020 onward has had significant impact on the world and specifically the maritime industry. Striking examples were COVID-19 outbreaks onboard the Diamond Princess cruise vessel and the U.S.S. Theodore Roosevelt aircraft carrier at the start of the pandemic. Contagious disease management onboard large passenger ships remains a complex issue, amplified by the international character of the industry, confined environment and shared facilities. This report therefore presents an investigation into the effect of ship layout design, operational and behavioral measures on COVID-19 airborne infection risk onboard large passenger vessels. The novelty of this research lies with the integrated infection and crowd behavior model used to calculate agent-specific infection risk, incorporating guest and crew circulation through a passenger ship layout.

Quality control is considered an important process in manufacturing to minimise waste related to the manufacturing process. A way of performing quality control is with help of machine vision. Understanding all decisions that are to be made when designing a machine vision system for quality control is essential, but challenging. This research presents a method that aids decision-makers in the process of designing these machine vision systems for quality control so the efficiency of the decision-making process can be increased. The method results in a knowledge base and a tool that can both be considered as useful to decision-makers. The method was designed by following a Design Science Research approach and the method was applied on a use case in the automotive industry to prove that it addresses its goal.

The Automated Guided Vehicle (AGV) is the most commonly used piece of horizontal movement that uses electric power from an internal battery pack. With much focus on the charging strategy, current charge operation models lack the ability to model large scale terminal operations and to incorporate new types of horizontal movement. The objective of this research is to show the impact of recharge operations on terminal performances. A discrete event simulation model is based on the characteristics of APMT MVII to model the current impact. The results of numerical experiments show charging operations have a limited impact on the productivity of Lift-AGV. An exception is the change in battery type. Using a Li-ion battery lowers the need for recharging and increases total productivity. The findings show the importance of integrating modelling to accurately project terminal operations.

The growing demand for parcel deliveries is causing significant problems for freight logistics, particularly in cities. The increasing number of delivery vehicles leads to traffic congestion, high emissions, and rising costs. To address these problems, new and sustainable delivery methods must be implemented for
the parcel last-mile. However, estimating the impact of a different logistics system is complex, as it depends on consumer adoption of these new delivery methods.
This paper presents a new simulation model that captures and explores the interconnections between multiple delivery methods and consumer preferences for those delivery methods. Consumers’ reaction to the performance and availability of the delivery methods is simulated, next to knowledge progress via
word of mouth and familiarisation. The developed hybrid model uses system dynamics to simulate the evolution of consumer preferences and the last-mile delivery via van, self-collection, crowdshipping and drone delivery at an aggregate level, while an agent-based model provides input on the operational performance of the delivery methods. With this structure, consecutive
interactions can be simulated, and by that, data on consumer preferences and the delivery method operations can be obtained at multiple time points.
Results of a case study on the province of South Holland show that consumers change their preferences due to the introduction of new delivery methods. However, the vehicle kilometres and emissions of van delivery do not reduce at the same rate as the demand. The study also highlights zonal differences in last-mile services and the importance of delivery quality to consumers.

A monopile installation project consists of the whole process of loading monopiles to vessels, transporting monopiles to the wind farm and drilling monopiles in the seabed. Each installation project has different characteristics as the weather at the installation location, the distances to the suppliers and the vessels that are available differ between projects. Different project characteristics require different choices to be made when planning a typical installation project in order to optimally plan or reduce the expected cost of a project. One of these choices is determining which supply chain configuration will be used to get the monopiles from the supplier to the offshore wind farm. The choice of configuration is important as it determines the type of vessels that are required and if intermediate storage ports need to be rented, which requires initial investments.

The purpose of the research study is to assess the performance of different supply chain configurations. The performance of these different configurations is assessed on the key performance indicators project cost, project duration, utilization of the vessels and the waiting on weather duration. In order to assess the effect of using different supply chain configurations on the key performance indicators, 6 different experiments were performed within a developed simulation model.

Planners can use the qualitative insights of this report as a starting point for the choice of the preferred supply chain configuration when planning an offshore wind farm installation project. This will enable planners to make educated choices in the planning phase and give these planners insight into the different factors that are important in the supply chain of offshore wind farm installation projects. From a theoretical point of view, this report contributes to the sparse literature on the supply chain and logistical decisions in the field of offshore wind farm installation. Next to these insights, the weather module used to generate synthetic weather data improves Markov models discussed in literature and provides a solution to the problem of persistence that is present in conventional Markov models.

As connections between customers and suppliers become increasingly complex, the design of transport networks is under growing pressure to achieve efficiency. In response to this challenge, this paper proposes a service network design that takes into account both the forward flows from a production facility to a customer and the returning flows from customers to the production facility. The design allows for products to be shipped directly or to be consolidated and sorted at a depot. To assess the benefits of optimising both the forward and returning flows simultaneously in terms of network costs, the paper formulates an arc-based model and applies it to a real-life case. The results indicate that the network costs can be reduced by optimising simultaneously, as depots are built in the optimal locations and with the right capacity. This contrasts with the sequential method, which does not account for the return flows and can lead to suboptimal depot placement.

The Dutch railway network is a dense network with 7000 kilometers of railway tracks spread over only 42 000 square kilometers [1, 2]. Additionally, the railways are responsible for 1.4 million passengers every working day [1]. These factors make both the infrastructure and timetable of the Dutch railway network extremely intricate. On top of that, the amount of train trips is expected to grow with 40 % by 2040 [3] and with that the complexity of managing the railway network will increase even more. In the case of conflict in the timetable, action must be taken to resolve this conflict in order for the train operations to continue. In current Dutch practice this task resides with the train dispatchers. However it can be difficult for the dispatcher to oversee the situation when multiple trains get involved and taken actions will have consequences on yet another selection of trains. This increasingly complex task has resulted in a grown interest in so-called Traffic Management Systems (TMS), which are an intelligent systems that use a conflict resolution algorithm to find a solution to a conflict. A TMS can be used to support the train dispatcher...

Shipping companies aim to reach the climate goals by following the rules from the International Maritime Oranization (IMO), which includes reducing Scope 3 emissions. Maintenance is a great contributor to the Scope 3 emissions of a shipping company, especially the transportation of spare parts. The current state-of-the-art supply chain optimisation for spare parts is based on cost and risks. This study aims to find the potential influence of including Greenhouse Gas (GHG) emissions into the decision-making process for the logistics of spare parts. The element of the supply chain that makes it possible to plan when a spare part is needed is the maintenance policy. Planning in advance when a part is needed is possible for Preventive Maintenance (PM), which is therefore used in this research. A model can be created by limiting the amount of risk, optimising between the freight cost, the cost of GHG emissions and the cost of capital for alternative delivery locations, using brute force calculations. This model is applied to a case study of a chemical tanker from Stolt Tankers B.V., using available data on its job history and historical location data. From the case study follows that including the GHG emissions in the decision-making process adds an additional saving in emissions and cost with respect to the original situation. It can be concluded that the amount that is saved depends on the choices of the decision-maker. The model presented in this paper is a valuable tool for providing insights into the decision.

Efficient distribution is crucial in many industries today, especially in retail, where customer demand grows rapidly. The rise of online shopping and automation has led to a shift towards using Central Distribution Centers (CDCs). Retailers are part of this trend, which requires them to rethink how they handle their distribution. This study focuses on two main challenges: first, changing the distribution system to adapt to new CDCs, and second, dealing with the computational complexities when optimizing the distribution on a large scale. The aim is to design a large distribution system for the allocation of stores and the routing of vehicles. In this research, we introduce new elements to the mathematical model, which are mostly focused on the routing of vehicles. We also improve the computation by implementing a start solution, adjusting the parameters of the Gurobi optimization tool and including a callback function which bounds the run time. This model will determine the routing decisions for demand from several DCs to the stores and find the optimal vehicle routes. Through a case study at a company in the Netherlands, this paper provides insights into the changing world of distribution systems. It offers practical solutions for businesses trying to make their distribution processes better. The analysis revealed that consolidating demand at CDCs can lead to significant cost improvements while altering transport costs between CDCs and RDCs and adjusting order volumes can influence delivery strategies. Despite the absence of direct relationships between routing decisions, distances, and volumes, a logistic regression model provides guidance.