This research explores the integration of electric vehicles (EVs) into the logistics of bulk-liquid transportation, specifically within Heineken Netherlands’ operations, using a two-echelon network. This study is pivotal as it aligns with global moves towards zero-emission regulations and sustainability in logistics. The primary question it addresses is optimizing a logistic network and truck operations for bulk liquid delivery by transitioning to EVs within a two-echelon framework, ensuring efficiency while adhering to sustainability and regulatory demands.
The investigation employs a sequential exploratory strategy, beginning with qualitative analysis to identify core challenges and opportunities, followed by quantitative methods to refine network design and decision-making. Advanced clustering techniques such as the center of gravity, p-median, and k-means are utilized to determine optimal depot locations, essential for overcoming the operational range and charging limits of EVs. This approach aids in developing a logistics network that is both operationally efficient and environmentally sustainable.
A significant portion of the study focuses on vehicle routing within the two-echelon location-routing model. It considers critical factors like the limited range of EVs, multi-compartment transport requirements for bulk liquids, and specific customer delivery windows. The model integrates these elements to optimize vehicle routes for efficiency and regulatory compliance, illustrating its practical use through the two-echelon multi-compartment electric vehicle routing problem with time windows (2E-MCEVRPTW).
The practical application of this research is demonstrated in a case study of Heineken Netherlands, highlighting the logistical complexities of transitioning to an EV fleet for beer distribution. The study examines operational challenges such as vehicle range and product diversity management, proving the viability and effectiveness of the proposed models.
Results discussion reveals that strategic network design using the center of gravity method significantly enhances kilometer savings and operational efficiencies. However, the benefits diminish with additional hubs, indicating an optimal hub number exists. While transshipment costs pose a significant challenge, outweighing the kilometer savings, potential cost reductions through increased reefer capacity and reduced transshipment times are identified, pointing to possible areas for improvement.
The study concludes that the two-step optimization process, integrating network design and vehicle routing, effectively addresses the research question. It not only shows the potential of EVs in transforming logistics but also underscores the economic and operational challenges of adopting a two-echelon network. The findings lay a groundwork for future innovations in sustainable logistics, though they caution the need for tailored solutions across different operational contexts and suggest further research into computational strategies and customer clustering for enhanced route optimization.

Access to affordable, reliable, and sustainable energy is a key goal of the Sustainable Development Goals (SDG). Despite this, 675 million people globally, especially in rural areas, remain without access to electricity. Sierra Leone exemplifies this issue, with only 4.9% of its rural population having electricity access. The lack of access hinders economic and social development, impacting healthcare, education, and overall quality of life. While grid extension is costly and difficult in remote areas, solar-based micro-grids offer a promising solution for rural electrification, leveraging the country's solar potential.

However, literature reports performance issues with these micro-grids, and while some factors influencing the performance of micro-grids are identified, their impact and mitigation strategies are underexplored in rural contexts. This research aims to identify and classify the factors affecting micro-grid performance and assess their impact on rural developing areas. The study provides insights into mitigation strategies, considering technical, social, economic, and governmental contexts, bridging the gap between qualitative and quantitative research to improve access to electricity.

The research uses a case study approach combined with modelling. Data collection includes site visits, literature reviews, and semi-structured interviews. The modelling uses Python for Power System Analysis (PyPSA) to assess the impact of identified factors on micro-grid performance. The case study focuses on four small communities in Sierra Leone with varying levels of user satisfaction. Findings reveal that micro-grids face economic constraints, technical limitations, and dependency on government support.

The study identifies several key factors affecting micro-grid performance, including high appliance use, low-quality battery design, and a lack of skilled technicians. Battery performance is determined as the most critical factor, directly affecting electricity availability. The research evaluates several mitigation strategies, such as Demand Control (DC), air conditioning, and additional battery capacity. DC is found to be an effective mitigation strategy, especially in evening hours, enhancing electricity access in rural areas.

Demand Control is a promising mitigation strategy and can enhance micro-grid performance, within the rural developing context. Further research is recommended to refine these strategies and explore their broader applicability to other micro-grid systems in rural developing areas.

The Dutch drinking water utilities are legally responsible for supplying drinking water to their assigned customer base. Continuing to supply drinking water for the long term is coming under pressure from challenges on the supply side, related to water quality and quantity on the one hand and developments due to increasing demand for drinking water on the other. Both developments are expected to be negatively influenced by the effects of climate change.
The Dutch drinking water companies face three major challenges regarding strategic investment decisions. First, the current sourcing and production capacity must be expanded to meet future drinking water demand. Second, there is a great demand for End-of-Life replacement of pipes in the drinking water infrastructure. Third, an investment challenge of a lesser financial magnitude but with an expected great impact on business operations is related to gaining operational control over the drinking water distribution network by integrating state-of-the-art sensor technology.

The outcomes of the internal decision-making processes of the drinking water utilities regarding these three strategic challenges will affect the stakeholders of the drinking water utilities. In addition, it offers possibilities for alignment with the goals of the other stakeholders. The main problem that this research seeks to address is a lack of engagement with drinking water utilities' stakeholders in the decision-making processes. A way to engage with stakeholders is by using Participatory Modelling, a technique that is not commonly applied by drinking water utilities.

These possibilities to engage stakeholders in the decision-making process are further backed by the development of new resources that have become available in recent years. These resources are new modelling techniques that have been applied in the field of drinking water research, in recent years. And, a novel perspective on multi-modelling e.g. the Multi-Model Ecology (MME) with Multi-Model Interface (MMI). In the current practice of research for Water Resource Management and other research for drinking water utilities, an MME and MMI (MME+I) have not yet materialised. This study aims to determine if an MME+I can benefit research for drinking water utilities and facilitate Participatory Modelling.

The Participatory Systems Design methodology (PSD methodology) is applied to generate a design for the conceptual model of the MME+I and the logical architecture for the MMI. A Proof of Concept (PoC) use case of model-coupling was applied. Here, an ABM model for Water Demand generates water demand patterns for an EPANET hydraulic model. This is a novel approach in hydraulic modelling for Dutch drinking water utility Oasen, since it introduces agents' behaviour from the ABM model to the modelling of hydraulic networks. It demonstrated that the outcomes of an ABM model affect the performance of the EPANET hydraulic model. In addition, It provided insight into how changes in water demand from scenario studies can affect strategic investment decisions for drinking water utilities.

The swift reduction of human’s carbon footprint is essential to prevent irreversible damage to the climate and to meet climate policy targets. Designing flexible and reliable future energy systems is a big contributor to meeting these goals. While energy system models have improved in the last few decades, they remain vulnerable against parametric and structural uncertainty due to the varying characteristics of parameters and the hardship of modelling all constraints and drivers accurately. This thesis proposes a method that addresses both uncertainty types in energy system modelling by applying SPORES cost optimisation and Monte Carlo scenario modelling simultaneously.

The main case study uses 27 input scenarios with varying outcomes for grid electricity price, solar yield and energy consumption to provide insight in a 100 household neighbourhood energy system with heating, cooling, electricity and hydrogen as energy carriers. With 1377 (near-)optimal solutions, a novel approach in analysis and post processing is used to provide 52 useful configuration options that each have their strengths and weaknesses to different political, economical, social and technical drivers. These configurations are tested for cost, security of supply, CO2 emissions and grid dependency. Those results are visualised through ridge plots and statistical tables to provide a clear overview between each configuration’s trade-offs. An example is included to show how those results can be used for improving energy system design in practice.

This thesis shows that two methods can successfully be combined into one universal one, while providing valuable design insights for energy systems under uncertainty. Furthermore, this method can be applied to a wide variety of energy systems, as long as its possible components, their technical aspects and their allowed interactions are known beforehand. As many future energy system aspects are uncertain, it should be seen as a vital tool to help speed up the decarbonization.

System Dynamics (SD) is an approach to study the nonlinear behaviour of complex systems over time. SD models provide a high­level understanding of the system and aid in designing policies to achieve specific system behaviours. Conventional SD modelling requires an intensive amount of time, human resources and effort. Applying Machine Learning (ML) techniques benefits the modelling process in saving on resources. It also has the potential to provide insights into the system and prevent subjective­ ness of the modeller. This work proposes two methodologies, EvoNN and EvoESN, to learn SD models automatically for the urban system from observations under different levels of prior knowledge. EvoNN solves the automated equation formulation task for a Causal Link Diagram (CLD) and annotates it with Shallow Neural Networks (SNNs) as surrogate equations. The annotated CLD can be further used in simulating the system behaviour. We provide experimental results on a real­world urban system in Am­sterdam as well as the evaluation of the simulation results. The second methodology, EvoESN learns both the structure and the quantitative relations in the model without the prior knowledge about the structure. Trained using observation data, the EvoESN produces satisfactory results on the real­world urban system. We further incorporate the judgement from the domain expert to evaluate the learned model. Applied on a more complex system, EvoESN shows solid reliability and scalability to handle large datasets. Both EvoNN and EvoESN stand as promising supportive tools for SD modellers and remain robust even when lacking system observations.

The increase of the average world temperature, as a result of greenhouse gasses, is one of the greatest challenges the world is currently facing and will be facing in the future. One of the many efforts to reduce the increase of carbon dioxide, is to decarbonize the gas network by replacing natural gas with renewable gasses like green gas. Since the current gas network is not always suited for the injection of big green gas volumes, especially during summer periods, the process of decarbonizing the natural gas network requires adjustments of the current functioning of the gas network. Several technical adjustments for obtaining an increased green gas injection capacity exists, although, up to now, the potential green gas injection capacity obtained per solution was not known. Within this study, a dynamic gas network simulation model was developed wherein different gas network function strategies can be explored to obtain the potential green gas injection capacities. With the developed dynamic gas network simulation model one is capable to implement and simulate different gas distribution network configurations, specify green gas suppliers on the location of choice, simulate different gas demand scenarios and consumer profiles, adjust the city gate station pressure used for simulations of static- and dynamic pressure management, and to model the injection of excess green gas into a storage- and from the storage into the network. Within this study, the gas distribution network of Northeast Friesland, The Netherlands was analyzed on its green gas injection capacity after applying static pressure management, dynamic pressure management, and a pressure management strategy combined with storage. The gas network of Northeast Friesland was explicitly chosen since currently, green gas injection problems are experienced within this network. Following the results obtained from the simulations, a city gate station inlet pressure - demand table was defined. Within this table, the total gas demand measured within the network was plotted against the minimum city gate station inlet pressures, while still in compliance with the lower pressure boundary condition. Using this table, the optimal period to statically decrease- and increase the city gate station inlet pressures from 8.3 to 6.5 bar and from 6.5 bar to 8.3 bar, appeared to be respectively 1 May and 1 October. By changing the pressure to 6.5 bar, a safety margin of 1.5 bar was taken into account. For both static- and dynamic pressure management, green gas injection capacities ranging from 400 to 1600 m³/h, divided over three green gas suppliers, were analysed. The results were depicted against the total injectable hours, providing insight in the maximum green gas injection capacity while remaining eligible for the Stimulation of Sustainable Energy Production (SDE+) subsidy. Since with dynamic pressure management the city gate station can inject at lower inlet pressures, dynamic pressure management results in a greater green gas injection capacity. To point out, static pressure management provide 450 m³/h green gas injection capacity, whereas dynamic pressure management provides 650 m³/h green gas injection capacity without experiencing any injection problems.

The energy system is undergoing a grand transition. In 2019, the Dutch government presented the national climate accord stating that in 2030, 70% of all energy needs to come from a renewable source. In this accord, strong emphasis is put on a regional approach to the energy transition. Across the country, regions are picking up the gauntlet and 141 municipalities in The Netherlands have formulated ambitions to become energy neutral by 2050 or earlier. The road to reach these regional ambitions, however, is not always clear. One of the key issues in energy planning is defining the optimal mix of generation methods to fulfill the electricity demand. Historically, this challenge has been approached only from a least cost perspective. Different stakeholders, however, have a different view on what defines the ‘optimal’ situation and care about more than cost. It is found that minimizing land use and minimizing the visual impact of wind turbines are important objectives to consider when designing an energy system. This research presents a multi-objective optimization that employs a genetic algorithm (NSGA-II) to find the set of pareto-optimal solutions for an optimal generation mix for a regional energy system in The Netherlands minimizing costs, land use and visual impact. Three scenarios are investigated: reducing the total emissions by 70%, 90% and 98%. The results of the optimization are analyzed from a multi-actor perspective to provide insight into the most ideal solutions for different stakeholders. The results show that there are significant trade-offs to be made in designing an energy system: governments, investors and local residents all have a different view about the optimal generation mix. This research presents an average optimal solution: one that may work best for all actors. It shows that by finding a Pareto-optimal set, many optimal solutions can be compared on their desirability, leading to more insight into the functioning of the system and a more feasible design.

In the Netherlands, wind power has the highest potential for future development compared to alternatives such as solar PV and biomass. The intermittent nature of the wind power resource calls for novel approaches to supply and demand management, as well as power quality assurance. Energy storage technologies allow for the separation between power generation and power supply to the grid. The energy and power densities, efficiency and response times of secondary batteries make them highly suitable for utility scale application to wind farms. However, a knowledge gap exists on exact configurations of battery energy storage systems for wind farms within the existing power markets. This study proposes a hybrid power plant approach, combining an onshore wind farm with a battery energy storage system. Literature research identifies balance between power obligation and production as a main design criterium, along with the profitability of a configuration and operational strategy. A technical analysis establishes power fluctuation on multiple time scales as relevant wind power characteristics, and maximum depth of discharge, discharge rates and efficiency as important battery parameters. A detailed analysis of the Dutch power market structure identifies three accessible markets for trade: the Day Ahead Market, the primary reserve (FCR) and the secondary reserve (aFRR). For modelling and optimization purposes, FCR is not considered due to excessive uncertainties regarding bid acceptance and bid activation. The design phase of the study is charged with finding the optimal battery capacity for a given wind farm production, while assuring the obligation to the Day Ahead Market is met. A linear optimization model is developed with the objective to maximize the obtained revenue. It is shown that the optimal battery capacity scales with the power capacity of the wind farm. For a 60 MW wind farm, the optimal battery capacity lies within the range of 60 – 80 MWh. An operational model is designed based on the model of the plant established in the design phase, elaborated with a Model Predictive Control approach. Based on wind power generation forecasts, the algorithm is able to adjust the charge/discharge and imbalance settlement strategy continuously based on expected market price series. However, the low level of accuracy of the applied market price approximations lead to unprofitable results in all simulated cases. Market price approximations with increased accuracy, as well as reduction of battery installation costs may lead to profitable operation of a hybrid power plant.

Even if renewable energy generation is improving and diffusing rapidly, reliable energy access is still a major issue for a consistent part of the global population, with more than 1 billion people still lacking energy access globally. The vast majority of this share of population is living in remote rural areas of developing countries, experiencing major issues in terms of living conditions. While consistent efforts have been done in the past decades to solve the problem, still a lot of work has to be done and novel approaches need to be implemented.
In the past, most of the new energy connections were achieved through national grid extension, which is proving to be a non-adequate short-term solution for a consistent share of the remaining part of the population living in rural areas. This is the reason why decentralised solutions, such as Solar Home Systems and DC micro-grids, are becoming more appealing as alternative ways to improve energy access in developing countries.
In this framework, this Master's thesis will focus on DC solar micro-grids as a solution to the energy access problem. More specifically, the aim will be to develop a methodology to gather, process and analyse data, for planning and evaluation of remote DC micro-grid networks in rural areas of developing countries. One of the main novelty aspects of this proposed methodology is the integrated implementation of Geographic Information Systems and concepts derived from the mathematical field of Graph Theory, together with an electrical analysis.
The methodology is clearly divided into three consecutive steps. The first step focuses on gathering and processing ground-level data using GIS, to compare different micro-grid layouts in term of geometrical length. The second step consists of a graph theory-based dual-objective optimisation algorithm to design meshed micro-grids from a set of starting topologies. The third step implements a DC power flow tool to analyse the operational behaviour of the optimised layouts. The proposed methodology is explained in detail throughout the report, with an example of its application to a sample of villages in different world-wide locations.
The results of this first application of the proposed methodology allow to draw some conclusions on the methodology itself and on the comparison of different micro-grid topologies. First of all, the huge potential of the combination of GIS tools and graph theory applied to micro-grid planning is shown. The results of the layout comparison show how typically implemented micro-grid layouts are generally outperformed by micro-grids designed using novel concepts and this integrated approach. Nonetheless, each specific case studies has peculiar characteristics and conditions that need to be taken carefully into account and can lead to totally different kinds of optimal solutions. It is hence of vital importance to have a methodology which is at the same time well-structured and flexible to adapt to changes and modification of parameters in order to perfectly reflect the specific needs and characteristics of each different rural electrification project.