Automated cold storage warehouses increasingly rely on automation to manage growing throughput demands, labour complexities, and sustainability requirements. However, the design of outbound buffer areas remains underexplored relative to long-term storage and retrieval processes. Fixed buffer lanes, commonly used to stage pallets ahead of truck pickup, create a structural mismatch with the stochastic nature of truck arrivals. When trucks are delayed, buffer lanes fill up, disrupting outbound flow and forcing manual exception handling. Shuttle-based storage and retrieval systems (SBS/RS) are identified in the literature as a promising alternative, given their fast-retrieval, small-footprint, and flexible release characteristics. Their application to outbound dispatch buffering, particularly in a cold-storage context and grounded in real operational data, remains largely unexplored. This research investigates how a shuttle-based dispatch buffer can be designed and controlled to minimise crane congestion and outbound delays in an automated cold-storage warehouse, using NewCold's Piacenza facility as a case study. The research combines a literature review and expert consultation with a system analysis and a discrete-event simulation model, built in Simio, validated against real operational data from the existing Piacenza warehouse. Three scenarios are compared: a validated baseline reflecting the current traditional bufferlane layout, a shuttle-based buffer design where picked pallets return to the highbay before staging, and a variant where picked pallets are routed directly into the shuttle buffer. The results show that a shuttle-based dispatch buffer's core mechanism, decoupling pallet retrieval from truck-arrival timing, substantially reduces the outbound system's dependence on truck-arrival forecasting accuracy. Routing picked pallets directly into the buffer further reduces the total crane workload required per pallet, creating an opportunity to operate with a smaller crane fleet for comparable outbound performance. This benefit does not come without cost: a smaller crane fleet operates closer to its capacity limit, and an under-resourced buffer configuration is shown to destabilise the system rather than absorb demand, confirming that such a design must be deliberately dimensioned rather than assumed to work at any scale. For the wider automated cold-storage warehouse industry, the findings demonstrate that a shuttle-based dispatch buffer's primary value lies in decoupling outbound operations from arrival uncertainty and reallocating pressure away from cranes, rather than in a proportional reduction in total truck dwell time, an important distinction for operators evaluating similar designs on throughput metrics alone. Academically, this research addresses a gap in the shuttle-based storage and retrieval system literature by applying this technology to an outbound dispatch-buffering role, validated against real operational data, and identifies crane capacity as an external coordination dependency not previously characterised in studies treating the shuttle system as self-contained. ... Read more

Despite increasing flood risks, household adaptation often remains below the level required to reduce vulnerability. Understanding how adaptation behavior spreads through communities is therefore an important research challenge. This thesis explores how changing social networks influence the spread of household flood adaptation behavior. In many real-world settings, social connections are not static: people form new relationships, lose contact with others, and interact with varying intensity over time. These processes, referred to as co-evolutionary network dynamics, may affect how households become aware of and decide to adopt flood protection measures. Despite their potential importance, their role in adaptation diffusion remains insufficiently understood.

To address this, this thesis extends the empirically grounded Flood Adaptation via Social Transformation (FAST) agent-based model with co-evolutionary social network dynamics. These include the formation and breaking of social ties as well as changes in the strength of interactions between households. The model is used to systematically explore how different configurations of these dynamics influence both the structure of the social network and the resulting adaptation behavior.... ... Read more

The energy transition depends on storage. Battery Energy Storage Systems (BESS) are being deployed at record pace, and investors compete to acquire the projects behind that growth. Before any acquisition comes screening: an investor receives a short teaser, later a data room full of documents, and must decide within weeks whether a prospect is worth pursuing. This thesis studies that screening step at GIGA Storage B.V. (hereafter: the BESS developer), a Dutch company that develops, acquires, owns, and operates utility-scale BESS, and asks:

How can a Large Language Model-based screening instrument be designed to convert unstructured prospect documentation into a screening record that supports early-stage investment risk assessment for utility-scale BESS acquisition?

The problem underlying this question is first established empirically. Interviews with nine experts and an analysis of the acquisition process reveal five recurring bottlenecks in current practice. Screening relies heavily on information supplied by the selling party (the prospect developer). Several specialists read the same documents in parallel without a shared overview. There is no cross-functional evidence gate that forces a structured go/no-go moment. Assumptions shift between phases and between advisors without being reconciled, and the evaluation is directed mainly at the current state of a project rather than at the state it must reach to become operable and profitable. The literature offers no ready method for these conditions: established techniques need historical data or quantified inputs, and early-stage BESS screening has neither.

Following Design Science Research, the thesis converts this problem into four results that build on one another. A structured literature review, merged with the interview findings, delivers a catalogue of 52 investment risk factors across 12 categories, mapped onto the 4P screening structure used in practice (Plot, Power, Permit, Profit); to the author's knowledge, it is the first factor-level risk catalogue for utility-scale BESS acquisition. Requirements elicitation delivers 41 requirements, of which seven define the design core. A conceptual process design then splits the work: assessing a prospect requires judgement, but most screening effort goes into something more tractable, turning unstructured prospect documents into structured facts. The design decomposes LLM-supported prospect analysis into four blocks that converge on the assessment, and this thesis realises the foundational block, filling the checklist, as a working prototype.

The prototype reads a teaser and a data room and fills a 152-item checklist in three phases. Every answer carries an evidence state (found, probably found, or not found), a source with page number, and a quote that is checked mechanically against the cited document. A human reviewer gates every phase: feedback and approval can be given on the extraction, critical values must be confirmed explicitly, and once a person edits or confirms an item, no later model output silently overwrites it. In the final phase, the tool verifies key claims against publicly available sources such as company registers and grid operators.

The frozen prototype was then evaluated on nine real prospects from the acquisition pipeline. A first quick extraction costs a few cents and completes within a minute; a complete deep run costs a few dollars. The public-source check corrected or extended 62% of the claims it verified, confirming the value of an independent leg in a process that leans on seller-provided information. Three near-identical sibling prospects enabled a consistency probe: 87% of items were classified identically, and the divergence traced mainly to gaps in the item specification rather than to the model. A group session with six practitioners endorsed the design and its scope, and the team asked to start using it. Two record walk-throughs with experts who know one acquired prospect first-hand completed the picture. In the first, the expert judged the record a usable screening basis and identified one substantive extraction error, which was corrected at the review gate. The second, an in-depth due-diligence audit, delivered the evaluation's sharpest finding: the extraction was frequently right in substance but drew on the seller's own overview documents rather than the underlying evidence (61% of quotations, against the 5 to 10% that the auditing expert considers adequate). That shortfall lies in the item specification and in the routing of documents to the model, both properties of the design rather than of the model, and both improvable by the same means this thesis demonstrates.

The result is therefore a tested proof of concept, which provides the foundation on which LLM-supported investment risk assessment can be built. The structuring layer proved technically feasible under the tested configuration, received positive formative feedback from the practitioners who would use it, and is governed by checks and balances that keep the human in control: evidence states that prefer an honest absence, verified sources behind every claim, human decisions that cannot be silently overwritten, spending capped in code, and a model that may not claim checks it did not perform. Two quantities that an adoption decision turns on remain unmeasured, and they define the follow-up study: extraction accuracy and time saved against a known baseline. Beyond the prototype, the thesis contributes the risk-factor catalogue, the requirements specification, a fully documented and reproducible prompt set, and eight design principles for LLM integration in confidential high-stakes document work.

Screening more prospects, more rigorously, with the same expert capacity means that viable storage projects are identified and funded sooner. This thesis contributes one component of that development: a working and formatively evaluated instrument that applies LLMs to the screening step, in support of a more efficient realisation of resilient electricity infrastructure.
... Read more

The Dutch labour market combines the highest flexibility rate of the EU-27 with relatively strong dismissal protection for permanent employees, a segmentation that has historically concentrated the burden of economic shocks on flexible workers. Generative AI differs from earlier waves of automation in that it may substitute for non-routine, cognitive tasks, which raises the question of how quickly Dutch firms will deploy it, which tasks and contract types will be affected, and what this means for wage inequality and unemployment. Existing equilibrium models abstract from firm heterogeneity, bounded rationality and the Dutch institutional setting, while existing agent-based models treat AI adoption as exogenous or as an R&D lottery and do not model labour-market institutions.

This thesis extends the task-based agent-based model of Upreti and Sridhar (2024), implemented in Python (Mesa), with five elements: an explicit investment decision based on net present value, dismissal protection and the chain regulation (ketenregeling), wage formation with collective-agreement-like rigidity, a declining AI price curve, and a Keynesian feedback from unemployment to demand. Eight experiments vary the adoption logic, the institutional configuration, the AI-cost trajectory and the comparative advantage of AI over labour.

Making the adoption decision endogenous tempers automation and inequality considerably relative to a myopic cost-per-unit rule, driven by the expected wage decline through displacement, the demand feedback and a profitability threshold. A decomposition attributes the large majority of this gap to the forward-looking investment logic itself and only a small part to the Dutch employment-protection institutions. The relative price of AI dominates the outcomes more than any institution: both its level and its trajectory select between qualitatively different futures, from near-complete automation under falling costs to a near-standstill under rising costs. Of the institutions examined, only the minimum wage has a clear but double-edged effect, compressing the wage gap for those who keep their jobs while accelerating the displacement of low-skilled workers once it becomes binding. The transition payment and the chain regulation act as a small but persistent brake on the timing rather than on the destination.

Because automation is modelled throughout as all-or-nothing substitution, the model captures the displacing channel of AI but not its augmenting or task-creating effects, so the reported inequality effects are upper bounds. The conclusions are exploratory and hold in the vicinity of a pattern-calibrated baseline; their value lies in showing which mechanisms matter and which uncertainties dominate, above all the price path of AI. ... Read more

This thesis develops a spatial graph-based framework for planning wastewater network expansion in northern Larnaka, Cyprus. The method integrates road-based construction corridors, building-level demand, existing sewer assets, terrain elevation, budget constraints, and pumping feasibility. Five network configurations are generated and compared. Full urban coverage is technically achievable but exceeds the available budget. Budget-constrained alternatives achieve 95.7–97.5% modelled urban coverage at approximately EUR 127–130 million, while differing substantially in municipal distribution. The results highlight the trade-off between aggregate efficiency and balanced service provision, supporting a phased, municipality-aware expansion strategy. ... Read more

A pediatric hospital in the Netherlands is experiencing increasing dissatisfaction among its cardiologists, which is primarily related to the scheduling system. Physician scheduling is a highly complex process due to the involvement of multiple stakeholders with differing needs, where physician satisfaction and organizational efficiency often conflict. In addition, complexity arises from the existence of multiple subspecializations and the wide variety of clinical and non-clinical tasks that must be allocated within limited time resources.

The question addressed in this study is how the current scheduling approach can be improved to better balance stakeholder needs while maintaining acceptable levels of organizational efficiency. To address this, a Soft Systems Methodology approach was first used to gain a comprehensive understanding of the problem situation. Subsequently, a Design Science approach was applied to develop and iteratively refine prototypes, resulting in a proposed final design for an improved scheduling system.

The final design results in a slight decrease in organizational efficiency compared to the current situation. However, this trade-off appears unavoidable due to structural time constraints and the high workload required to complete all assigned tasks within the available time. This reflects a systemic imbalance between workload demand and available physician time.

Future work should focus on further development and implementation of the proposed design in practice. In addition, validation should be conducted with a broader range of stakeholders to better capture the different perspectives on the trade-off between organizational efficiency and physician satisfaction. ... Read more

Electricity grid congestion increasingly limits the ability of large electricity consumers to expand, electrify, or connect new assets. While grid reinforcement remains necessary, it is often too slow to resolve local capacity constraints in the short term. This thesis evaluates the time-duration-based transport right (TDTR) as a complementary instrument for using existing grid capacity more efficiently. TDTR provides firm capacity for part of a customer’s demand and conditional capacity for the remaining share, which may be restricted during predefined congestion-related call-hours.

A customer-level techno-economic optimisation model is developed and applied to three flexibility archetypes: logistics with an electric-vehicle fleet, cold storage, and dairy processing with an industrial heat pump. For each case, the model determines the cost-optimal firm-to-TDTR ratio and evaluates both customer-side savings and grid-side effects. Grid impact is assessed through peak reduction, weighted grid-pressure reduction, and changes in consumption during local call-hours. The cases are tested across four substation call-hour profiles to examine when customer incentives and local grid needs align or diverge.

The results show that TDTR can create a positive business case for all three archetypes, but that the magnitude of savings depends primarily on the share of load that can be shifted in time. Logistics achieves the largest savings, followed by cold storage and dairy processing. However, a positive customer business case does not automatically imply local grid relief. TDTR delivers the strongest grid value when congestion call-hours coincide with hours in which the customer already has an economic incentive to reduce demand. When call-hours coincide with low electricity prices, customer cost minimisation may work against the DSO’s congestion objective.

The main conclusion is that TDTR can align customer value and grid value, but only under specific conditions. For DSOs, this implies that TDTR should be targeted, monitored, and adjusted based on customer flexibility, local call-hour profiles, and realised load reduction rather than contracted TDTR share alone. ... Read more

The Model Context Protocol (MCP) and the Agent-to-Agent (A2A) protocol have quickly become the de facto standards for connecting large-language-model agents to external tools and to one another across organisational boundaries. Their security has begun to receive attention, but their privacy implications, what personal data flows where and under what controls, remain largely unexamined, even as the protocols are adopted in sensitive domains where the people whose data is processed are not parties to the exchange.

This thesis presents a data-subject-centred privacy threat analysis of MCP and A2A using the LINDDUN methodology, applied to a representative financial fraud-detection use case modelled as a multi-agent system. A data-flow model is elicited, threats are systematically derived across the LINDDUN categories, and the findings are validated through interviews with seven domain experts and discussed with a co-author of the LINDDUN framework.

The analysis identifies eight privacy threats across five LINDDUN categories. The most severe arise not from misconfiguration or attack but because the protocols operate as designed: a persistent session identifier enables linkability and cross-store aggregation, unconstrained context-passing exposes identifiable data across organisational boundaries, and the bilateral task record creates a non-repudiable trace. The threats range from protocol-inherent to use-case-inherent, and three form a cross-protocol chain in which exposure compounds as data crosses the boundary between the two protocols; the trust boundary, more than the protocol choice, amplifies severity.

This is the first privacy analysis of MCP and A2A framed from the data subject's perspective. It shows that privacy here is non-compositional, so per-flow assessments miss risks that emerge only when flows combine. Because the decisive design choices are fixed at the protocol layer, the thesis argues that effective privacy governance must engage protocol designers and standards bodies, not deploying organisations alone. ... Read more

Cryptocurrencies play an increasingly prominent role in legal financial activities, while simultaneously creating new opportunities for illicit use. Their speed, ease of crossing borders, and pseudo-anonymous nature strongly contribute to their potential for laundering illegal funds. Cryptocurrency exchanges are considered critical financial services that convert cryptocurrencies to fiat money and vice versa, thereby fulfilling a central role within the crypto ecosystem. As a result, increased attention is directed towards such digital infrastructures that may facilitate and scale illicit financial activity. Gaining insights into the users operating on such platforms, as well as the behaviors they perform, may support law enforcement in understanding how these infrastructures are utilized and which behavioral patterns emerge. Moreover, the data obtained from these vital platforms enhances the potential of data-driven approaches that identify broader behavioral structures and aid differentiating between user groups, which can in turn support the prioritization of investigations.
Understanding user profiles and their behavioral patterns within the internal structures of cryptocurrency exchanges has received limited attention in earlier research, yet understanding (criminal) modus operandi is considered essential for effective intervention. Previous studies mostly rely on identifying suspicious activity through machine learning models, using publicly available blockchain data, often without including the context and perspectives of investigative practice. By examining a unique dataset from the cryptocurrency exchange Cryptex acquired through law enforcement operations in 2024, this study aims to fill this knowledge gap.

The central research question guiding this thesis is: To what extent can latent behavioral profiles be identified within the internal records of a seized cryptocurrency exchange, and what insights do these profiles provide for law enforcement investigations?

By applying a mixed-methods approach, this research conducted a Latent Profile Analysis (LPA) to identify distinct behavioral user profiles based on transactional activity indicators related to deposit and withdrawal operations on a seized cryptocurrency exchange.
First, a conceptual lens was developed based on behavioral indicators related to structuring transactions, informed by policy literature on suspicious transactional activity within cryptocurrency ecosystems. Several methodological selection criteria were established to assess the suitability of the empirical dataset for LPA specifically, after which a study population was defined, consisting only of a subset of users who performed both deposits and withdrawals within a limited timeframe.
Next, the obtained profiles were further enriched by analyzing their differences in variables associated with elevated money laundering risk, including withdrawal address-sharing patterns, illicit exposure of incoming transactions, together with adherence to Know-Your-Customer (KYC) regulations and the use of privacy coins, both of which are associated with increased attribution mitigation.
Finally, an exploratory focus group with law enforcement professionals was conducted to assess the relevance, interpretability, and potential investigative value of the obtained behavioral profiles and the methodological approach applied in this thesis.

In answering the main research question this study demonstrates that distinct latent profiles can be identified within cryptocurrency exchange data, reflecting distinguishable behavioral patterns and broader behavioral strategies within the exchange ecosystem. These profiles primarily differ in their activity intensity, variability in transaction sizes and the extent to which multiple cryptocurrencies were exchanged. High-activity profiles on Cryptex are associated with increased interaction on the blockchain, greater overlap with external wallet addresses, and higher exposure to illicit activity, suggesting more complex or potentially coordinated behavioral patterns. In contrast, lower-activity profiles exhibit limited on-chain interaction and appear more aligned with KYC-procedures, indicating towards potential alternative transaction strategies that are less visible within publicly available blockchain data.
These findings show that user behavior within cryptocurrency exchanges is not homogeneous, but can be meaningfully structured into distinct groups that capture relevant behavioral differences from a money laundering perspective. Rather than directly identifying illicit activity, the findings suggest that behavior related to money laundering risk may emerge through broader behavioral configurations which are built upon combinations of transactional and platform-related indicators. While the profiles do not directly identify illicit activity, according to law enforcement professionals they appear to provide a useful abstraction for reducing complexity in large transactional datasets and can support exploratory analysis, prioritization, and hypothesis generation within law enforcement investigations. However, the method is considered unable to capture important nuance or answer cascading questions which are essential in law enforcement investigations, highlighting the remaining importance of human intelligence.

In doing so, this research provides three key contributions. First, it demonstrates the applicability of LPA as a method to uncover latent behavioral structures in cryptocurrency exchange transactional data, by means of a unique empirical dataset. This is considered to provide a more transparent approach to understanding user behavior and clustering. Second, this study demonstrates that cryptocurrency exchange users exhibit distinct behavioral configurations that become visible when transactional, blockchain-related, and risk-related indicators are analyzed in combination instead of isolation. The findings suggest that potentially elevated money laundering risk is therefore not uniformly distributed. Moreover, such risk does not emerge from a single behavioral dimension alone, but rather through broader combinations of interacting indicators across multiple behavioral dimensions. Therefore, the results suggest that blockchain-based indicators alone may not sufficiently capture the full diversity of transactional strategies, which potentially leaves certain modus operandi underrepresented or overlooked. This highlights the added value of combining internal exchange data with blockchain-related indicators for a more context-sensitive understanding.
From an investigative perspective, the findings demonstrate how behavioral profiling approaches may support exploratory analysis, data structuring, prioritization, and hypothesis generation within intelligence-driven cryptocurrency investigations. At the same time, the study emphasizes that the interpretation of such profiles remains highly dependent on contextual understanding, human expertise and nuance by investigators. ... Read more

Climate change is increasing the frequency and severity of flooding, posing growing challenges for housing markets and regional economies. Riverine regions are particularly exposed, as flood risk interacts with constrained housing supply, residential mobility and spatial inequality. In the Dutch province of Limburg, recent flooding has highlighted the need to understand how households respond to climate related risks when making housing decisions.

Although existing research examines the economic effects of flood risk, less attention has been given to the behavioural mechanisms underlying housing market responses. In particular, limited evidence exists on which reference points households use to evaluate flood losses and how these reference points affect property values, market activity and location choices.

This research applies prospect theory within an agent-based housing market model adapted to the Limburg river area. The model incorporates loss aversion and probability weighting alongside ten formulations based on objective risk perception, single and event-based flood losses, future market value, anticipated financial protection and flood experiences. These formulations are evaluated under medium and high housing demand and annual flood probabilities of 2%, 3% and 5%. Outcomes are assessed through transaction prices, housing utility, transaction frequency and household sorting by income and behavioural characteristics.

The results show that properties in flood-prone areas generally have lower transaction prices, lower utility and fewer transactions than properties outside the flood zone. Reference point selection meaningfully changes the magnitude of these differences, particularly at lower flood probabilities, while differences between formulations narrow at 5% risk. Only anticipated financial protection of 80% reduces flood related market differences below those produced by objective risk perception, while distant flood experience produces outcomes closest to the objective baseline. Higher demand strengthens spatial sorting, with lower income households becoming increasingly concentrated in flood-prone areas. Differences in probability weighting and loss aversion also contribute to sorting, although probability weighting has a stronger effect.

These findings indicate that flood risk policy should account for both behavioural responses and unequal purchasing power. Collective flood defences, clear risk communication and targeted financial protection may reduce both market disruption and unequal exposure. Future research should empirically identify the reference points used in housing decisions and test the model under lower flood probabilities and in different regional and cultural settings.
... Read more

The Dutch industry is responsible for 27% of national greenhouse gas emissions, mainly due to its reliance on natural gas. Green hydrogen is seen as a promising alternative, and the Dutch hydrogen backbone is being developed to connect the five largest industrial clusters. However, many regional industries (Cluster 6) will not be directly connected. These companies are diverse in sector, size and demand, which complicates the design of a cost-efficient supply system. Current research mainly focuses on pipeline networks at a national scale, leaving a knowledge gap on how regional industries can be integrated using multiple transport modes.
This study develops a method to model cost-optimal hydrogen transport networks for Cluster 6 industries using the Optimal Network Layout Tool (ONLT). Industry data was collected and categorised based on demand and distance to existing supply networks. K-means clustering was applied to create industry categories reflecting these characteristics, while DBSCAN clustering was used to group nearby companies and reduce model complexity. The ONLT was adapted to consider three transport modes: pipelines, vessels and trucks.
Results show that while pipelines and waterways are suitable for transporting large quantities of hydrogen, trucks are most cost-efficient for small-scale and flexible transport. The analysis of 291 companies confirms that most regional industries have relatively low hydrogen demand, which strengthens the role of trucks as last-mile solution. The study contributes by filling the gap in literature on regional hydrogen supply, showing how multimodal network design can support the decarbonisation of diverse industrial sectors. ... Read more

The Ostend Declaration (2023) sets out the ambition to transform the North Sea into Europe’s green power plant by significantly increasing offshore wind generation. While essential to meet Europe’s rising energy demand and sustainability targets (North Sea Energy, 2020), this also raises challenges. Transmitting such large volumes of electricity faces technical bottlenecks, including grid congestion and limited storage, and economic barriers such as high cable costs (Beaubouef, 2024). Hydrogen offers a promising pathway. As a flexible, storable energy carrier, it balances the system and reduces reliance on costly grid expansions (Dute et al., 2024; Farahmand et al., 2024). Offshore electrolysis—in turbines, hubs or energy islands—can lower pressure on onshore grids and land use (Ramboll, 2025; Janssen et al., 2025). However, large-scale offshore electrolysis remains underdeveloped: no full-scale projects exist, optimal sites are unclear, and deployment requires cross-border coordination (European Commission, 2025a; Van Wingerden et al., 2023).

Realizing this potential requires collaboration among North Sea countries (Van Wingerden et al., 2023). This involves connecting national offshore systems and integrating electricity and hydrogen into a supranational North Sea energy system (North Sea Energy, 2020; One North Sea, 2021). Energy islands can be strategic assets by linking offshore wind farms to a hydrogen backbone, enabling economies of scale and cross-border flows (Arteaga et al., 2024). Yet the system design faces uncertainties: the hydrogen economy is still developing, national governments mainly plan domestically, and ecological areas, military zones and shipping routes impose spatial constraints, demanding careful multi-use planning (North SEE, n.d.; Staeb, 2025).

Against this background, this thesis develops a conceptual system design for 2050 that minimizes overall costs while accounting for spatial constraints and infrastructure reuse. The societal relevance lies in showing that energy islands can enable large-scale offshore wind and hydrogen production and that a multinational approach is vital for Europe’s climate goals. Academically, it contributes to the still limited literature on multi-energy, multinational offshore systems. The guiding research question is: What is a system design with minimal overall costs for the North Sea, in which energy islands integrate offshore wind farms in an offshore hydrogen network, while accounting for other uses?

Three sub-questions are assessed. The first concerns how many energy islands are needed to balance efficiency and costs. Wind farms are grouped by distance, and cost implications analysed to identify the lowest-cost grouping. The second considers island locations, taking into account spatial constraints. Alternative layouts are compared to identify technically feasible and economically attractive sites. The third examines how these islands can be connected into a cost-efficient hydrogen backbone enabling cross-border flows.

The results show that eight energy islands balance construction and cabling costs, with an estimated €36.4 billion investment. Island locations are shaped by spatial constraints, but accounting for them reduces cabling costs from €14 billion to €0.3 billion by enabling shorter connections. The analysis also shows that the total capacity connected to islands is a key driver of costs, stressing the need for balanced capacity flows.

A hydrogen backbone is then designed. It could integrate 93 GW of hydrogen capacity, requiring 186 GW of electricity, at €8.9 billion. Existing natural gas pipelines can partly be reused, but new pipelines are still required. The combined system—energy islands, offshore wind connections, and a hydrogen backbone—amounts to €31.7 billion. This reflects a coordinated multinational approach; if countries plan separately, costs will be higher and integration weaker.

In sum, integrating offshore wind, hydrogen and energy islands into one North Sea system is technically feasible at substantial but necessary investment costs. These should be seen as strategic opportunities: without them, climate targets may be missed, energy supply less secure, and Europe more dependent on external sources. ... Read more

The global energy transition necessitates the rapid decarbonisation of industrial clusters, with hydrogen emerging as a key solution (kim2023). However, ongoing global crises have led to a shift in policy priorities, potentially delaying decarbonisation efforts and increasing the urgency for strategic and efficient resource planning (Stemerding2025). As climate targets approach, minimising delays becomes essential. This underscores the need for models that can capture the complex behaviour of firms and their mutual influence under uncertainty. Yet, most existing models do not account for how firm level behaviour and interdependencies shape infrastructure adoption and investment decisions in uncertain environments.

This study investigates how different firm characteristics, interdependencies, and scenario conditions influence the development of hydrogen infrastructure over time. The main objective is to understand how early investment decisions affect network formation and spatial outcomes in industrial clusters. To achieve this, a dynamic modelling framework was developed that combines a threshold based adoption model with the Optimal Network Layout Tool (ONLT). This approach incorporates firm level attributes such as hydrogen trade volume, grid connection capacity, plot size, and company type, and uses scenario analyses that vary hydrogen demand, import volumes, and early adopter configurations to simulate firm behaviour.

The results show that network development is highly sensitive to firm interdependencies, adoption behaviour, and external conditions. The timing of adoption depends on each firm's characteristics, with emerging strategic hubs such as Air Liquide, Eneco, and BP accelerating the rollout. In contrast, scenarios with high hydrogen demand might promote more integrated networks, whereas low demand scenarios often lead to fragmentation. Furthermore, the delayed adoption by Air Products, driven by relatively unfavourable characteristics, resulted in inefficient connections that were both long and costly.

The findings inform infrastructure planners and project developers on where to prioritise early incentives. The model provides guidance on investment priorities, supporting a more coordinated and cost effective infrastructure planning process, while also contributing to risk mitigation. By analysing different network layouts, robust segments can be identified that perform consistently across a range of scenario configurations, thereby reducing the risk of stranded assets. This study focuses on the Rotterdam Industrial Cluster as an illustrative case, but the approach could be adapted for application in other clusters beyond Rotterdam. ... Read more

This thesis investigates the comparative performance of three Multi-Objective Evolutionary Algorithms, Epsilon-NSGA-II, Borg, and Generational Borg, within the Evolutionary Multi-Objective Direct Policy Search framework, applied to the optimisation of the JUSTICE Integrated Assessment Model. By combining benchmark problems (DTLZ2 and DTLZ3) with the high-dimensional, multi-modal JUSTICE model, the study evaluates convergence dynamics, computational efficiency, and solution quality across varying levels of parallelisation. Results show that Borg consistently outperforms the other algorithms, particularly under high problem complexity, due to its asynchronous, steady-state architecture and adaptive operator features. Generational Borg performs better than Epsilon-NSGA-II on complex problems, but both suffer from scalability limitations due to synchronous execution. These findings underscore the importance of algorithm selection in climate-economy policy modelling, offering guidance on how MOEA design and computational resources interact to affect optimisation outcomes. ... Read more

With the rise of cloud computing, remote work, and interconnected systems, the traditional model of trusting everything inside the network perimeter has become increasingly risky. In response, Zero Trust Architecture (ZTA) has emerged as a model where no user or device is inherently trusted, regardless of their location. Micro-segmentation, one of ZTA’s key mechanisms, aims to limit lateral movement and enforce fine-grained access control. However, implementing micro-segmentation in environments with legacy systems remains particularly challenging due to outdated technologies, rigid configurations, and limited visibility.

Although the literature offers many advanced segmentation models, these are often designed for cloud-native or modern infrastructures and fail to address the constraints inherent in legacy-heavy environments. This research investigates how organizations with legacy systems can identify and evaluate appropriate micro-segmentation strategies based on their network characteristics and technical constraints. The central research question is:

How can organizations choose and implement micro-segmentation strategies that align with their network architectures and the technical constraints posed by legacy systems?

To address this question, the study began with a structured literature review, which revealed a lack of detailed and context-aware guidance for applying micro-segmentation in constrained environments. To fill this gap, a series of semi-structured interviews were conducted with cybersecurity professionals experienced in industrial and legacy-heavy systems. The interview data were analysed using thematic coding to uncover recurring patterns, practical challenges, and the decision-making logic used in real-world settings.

The findings reveal three principal categories of micro-segmentation strategies: network-based, agent-based, and hybrid. The choice between these depends on technical factors such as system compatibility, network architecture, and the level of administrative control available. In addition to strategy selection, the study identifies a phased implementation approach typically followed by organizations. The study also defines key criteria for evaluating segmentation outcomes, including enforcement of access control, operational continuity, system performance, network visibility, and many more. These insights form the basis of a structured, step-by-step implementation guide and a decision-support framework. Together, they enable organizations to assess their technical constraints and select appropriate strategies. The main contribution of this research is the development of a practice-oriented framework that supports the secure integration of legacy systems into Zero Trust environments. ... Read more

Schiphol Airport aims to achieve fully autonomous airside operations by 2050, driven by the goals of reducing human presence in high-risk environments and enhancing operational efficiency. Within this vision, drones present a promising innovation for airside tasks, offering autonomous inspection capabilities and adaptability to dynamic conditions. National developments, such as the Dutch Ministry of Infrastructure and Water Management's push for drone integration, reinforce this ambition. Schiphol's participation in the Dutch Drone Delta consortium highlights its commitment to exploring drone use cases in operational settings.

To explore drone integration at Schiphol, Runway Condition Reporting (RCR) was selected as a case study due to its repetitive, linear nature and critical safety function. RCR involves collecting runway surface data using a MARWIS sensor system. Currently performed manually by airside vehicles, the study proposes an autonomous drone solution flying between hubs at either end of the Polderbaan runway to perform RCR.

This research addressed autonomous drone-based RCR through three dimensions: reliability, effectiveness, and safety, leading to the central question:

"How can drones be utilized reliably, effectively and safely within Schiphol Airport's operational environment to automate runway condition reporting?"

The research followed a staged case study approach validated through stakeholder input. It began by assessing technical feasibility and system constraints, followed by designing a flight trajectory and operational protocol. A safety risk assessment was then conducted using a HAZOP analysis, and a visual simulation model was developed to communicate system insights.

Reliability was evaluated through performance assessment of commercial drones against Schiphol-specific criteria. A weighted decision-making process, supported by expert interviews, identified the MRTK drone system as suitable. However, wind gusts above 12 m/s—common in afternoons—would limit annual operability to approximately 64%, making wind resistance a key constraint.

Effectiveness was assessed by designing a geofenced, constrained flight path performing RCR in 3 minutes and 41 seconds. The protocol included emergency procedures such as Return-to-Home and Emergency Landing. An agent-based simulation integrated operational data, showing 27 drone inspections per day totaling 1 hour and 45 minutes of flight time. Results suggest potential efficiency improvements, but highlight airspace regulation bottlenecks during simultaneous drone operations.

Safety was assessed via HAZOP with six key stakeholders. The analysis identified 51 potential system deviations, four of which posed high risks: compromised airspace due to wildlife, loss of drone control, emergency protocol failure, and ground collisions. These risks could render runways unusable or halt airside operations. Root causes include wind gusts, system failures, and external interference. Mitigation measures such as stricter geofencing, redundancies, and improved coordination were identified, but require further research.

Despite the risks, this case study provides Schiphol with a reference framework for exploring autonomous drone innovation. It recommends starting with low-risk, non-runway use cases and engaging with regulators and stakeholders to build experience. Scientifically, the study offers a replicable framework combining technical analysis, stakeholder input, and risk assessment for early-stage exploration of autonomous systems.

In conclusion, while drones can perform RCR under certain conditions, safety challenges currently prevent implementation. However, this research provides a structured foundation for advancing autonomous drone integration at Schiphol and similar environments. ... Read more

In an era shaped by systemic risks and long-term challenges such as climate change, technological disruption and geopolitical instability, policymakers must make consequential decisions without knowing what the future holds. This condition, known as deep uncertainty, arises when the relationships between actions and outcomes are contested and traditional prediction methods no longer apply. To address this, analysts have turned to computational tools such as scenario discovery, which explores thousands of simulated futures to identify combinations of factors that lead to policy success or failure. It provides decision-makers with understandable scenarios: data-grounded narratives of the future that highlight critical vulnerabilities and opportunities. These scenarios help enable the design of strategies that are robust and adaptive... ... Read more

Executive Summary

Situation
Dutch first-line healthcare still faces major problems with how its information systems work together. Many of these systems are outdated, don't follow the same standards, and can't easily share data with each other. This makes it difficult for doctors, pharmacists, and nurses to work with complete and up-to-date patient information. It also creates obstacles for third-party developers, like those building telehealth apps or digital tools, who need to connect to these systems to offer new services.

Even though most healthcare providers use digital tools, these tools don’t speak the same language. That means that when one provider enters data, another might not be able to read or use it properly. This lack of data interoperability slows down care, increases paperwork, and limits innovation.

To address these issues, this thesis introduces a modular platform architecture that follows shared rules and standards. The goal is to help different systems work together more smoothly and allow developers to build useful tools without having to create custom connections for each system. The platform uses common APIs and reusable services to exchange data in a secure and structured way, while staying in line with Dutch healthcare policies.

Research Question
This research focuses on the following main question:
How can a platform architecture be designed to enhance data interoperability in first-line healthcare, enabling seamless integration for third-party complementors to support healthcare practitioners?

The question comes from the need to reduce technical complexity and help third-party developers deliver value to healthcare professionals faster and more reliably.

Approach
To answer this question, a systems engineering approach was used. The research started with interviews, where one was completed with a general practitioner and one with an expert from a third-party healthcare IT company. These conversations revealed key issues, such as outdated systems, mismatched data meanings, and a lack of shared standards.

From these insights, a list of requirements was created to guide the platform’s design. Then, several UML diagrams were developed to model how different components of the platform should interact. Based on these models, a working prototype was built using Python (Flask). This included core services like data validation, transformation, and user authentication.

Finally, expert software architects reviewed the prototype. Their feedback confirmed that the design made sense and had potential for real-world use. They also highlighted areas that still need improvement, especially in terms of security and handling data meaningfully and correctly.

Results and Conclusions
The research confirmed that fragmentation, outdated technology, and poor standard usage are the main barriers to data sharing in Dutch first-line healthcare. To address this, a modular platform was developed and tested.

The prototype showed that using shared components for data transformation and validation makes it easier to connect systems. Two services, the DataTransformationModule and the ExternalSystemConnector were seen as essential for creating an interoperable data platform.

The proposed design uses a layered structure and modular services. This makes it scalable, flexible, and easier for external developers to integrate with. While the technical foundation is strong, more work is needed—especially on semantic interoperability (ensuring data has the same meaning everywhere), user access control, and real-life testing in clinics.

Contribution
This thesis adds to both academic knowledge and practical system development.

From a theoretical perspective, it applies platform thinking to healthcare IT, based on Tiwana’s core–periphery model. It also builds on Deshmukh’s layered model of interoperability by turning abstract ideas into working components where each deals with structure, meaning, or security in healthcare data.

On the practical side, it provides a working system design that others can build on. The prototype shows how developers can use reusable components, like data validation or transformation services, to save time and improve data quality. This approach can help reduce technical effort and improve care delivery by making systems more connected.

Recommendations
To make this platform usable in real healthcare environments, several improvements are needed. First, authentication should be upgraded from basic API keys to more secure methods like OAuth2 or federated identity systems. These changes will better protect patient data and support real-world compliance without making the system harder to use.

Second, semantic interoperability needs more attention. Although the platform handles structure and syntax well, it still lacks tools to make sure that medical data means the same thing across systems. Future versions should include support for terminology mapping and validation using Dutch coding systems like SNOMED CT, ICPC, and LOINC.

Third, flexibility should remain a core design principle. Some services, like conflict detection, may only be useful in specific cases and should be added as optional modules rather than built-in features. This keeps the platform adaptable for different types of healthcare organizations.

Next, the platform should be tested in live clinical environments. Real-world deployment would uncover workflow issues, performance bottlenecks, and user experience gaps that are not visible in lab tests. Feedback from actual users, namely: GPs, nurses and pharmacists, is vital to make the system both effective and usable.

Finally, as data standards evolve, the platform must stay up-to-date. Future versions should be able to work with new releases of HL7 FHIR and support multiple profiles. This ensures long-term compatibility and encourages broader adoption.

In summary, this research lays the groundwork for a modular, standards-based platform that makes healthcare data more interoperable. It supports both better patient care and more efficient innovation by reducing the technical barriers that developers and healthcare professionals currently face.
... Read more

This thesis examines how AI-generated content labels are applied on two major short-form video platforms: TikTok and YouTube Shorts. As generative AI becomes more integrated into social media, platforms have introduced disclosure labels to support transparency and help users distinguish between real and synthetic content. However, little is known about how these labels are implemented in practice, whether they align with platform policies, and how they affect user engagement.

Using a dataset of 12,315 videos collected through hybrid scraping and API-based methods, this study compares labeling prevalence, source attribution, policy alignment, and potential effects on user metrics such as likes, comments, and views. The findings reveal notable differences: TikTok relies primarily on creator-applied labels with minimal platform enforcement, while YouTube Shorts applies more labels but does not disclose their source. In both cases, labeling practices are often inconsistent and not fully aligned with platform-specific guidelines or the technical capabilities of C2PA, the standard used for AI content detection.

Although some statistically significant differences in engagement were observed, especially on TikTok, the overall effect of labeling on user behavior was limited. The findings highlight key challenges in current AI labeling practices and point to a growing need for more consistent, transparent, and accountable moderation policies. While regulatory frameworks such as the EU’s AI Act and Digital Services Act aim to improve transparency in digital environments, this study shows that current platform practices often fall short of these goals, especially in terms of visibility, attribution, and enforcement of AI labels. ... Read more

Electric mobility has emerged as a central strategy for decarbonising the transport sector, especially in emerging economies like India. While national-level initiatives such as FAME provide policy momentum, subnational governments play a critical role in implementing Electric Vehicle (EV) transitions. This thesis evaluates the effectiveness of Karnataka’s EV policy mix using the Policy Mix Framework, focusing on four characteristics: consistency, coherence, comprehensiveness, and credibility. Through a qualitative content analysis of 18 key policy documents and 40 state- and municipal-level instruments issued between 2017 and 2025, the study maps interactions and misalignments across manufacturing incentives, infrastructure support, and consumer-facing measures.
Findings reveal moderate consistency, with some reinforcing interactions, such as those between subsidies and concessional tariffs, alongside contradictions in areas like metering standards and tax exemptions. Coherence is hindered by institutional fragmentation, particularly between state and municipal actors. Comprehensiveness is stronger on the supply side (manufacturing and infrastructure) but lacks equitable support for low-income users and rural areas. Due to the document-based methodology, credibility was not assessed.
This study contributes to the literature on sustainability transitions and multi-level governance by applying the Policy Mix Framework to a developing country context. It offers evaluations of a subnational EV ecosystem in India. The findings underscore that effective EV transitions require not only well-designed instruments but also strong coordination and inclusive implementation. Recommendations include creating a dedicated EV task force, expanding digital infrastructure, and introducing inclusive business models. The results provide insights for other Indian states and federal systems aiming to scale sustainable mobility.

... Read more