This thesis examines how safety and sustainability can be integrated into novel biobased façade panels. In the context of rapid urbanisation, climate change and the construction sector’s significant environmental footprint, biobased materials offer potential but face barriers such as fire safety, durability and long-term performance concerns. The EU’s SSbD framework embeds safety and sustainability from early product development, yet its application to construction, particularly for biobased façades, remains limited.

Using Leafy, a startup producing bamboo-based panels, as a case study, this research operationalises the SSbD framework through a qualitative approach with quantitative elements, drawing on document analysis and stakeholder interviews. Findings indicate that SSbD can be effectively applied to green façade panels through a tiered, multi-criteria assessment that integrates hazard analysis, life cycle thinking and risk mitigation. Responsible design requires meeting minimum safety and sustainability thresholds, guided by precaution and transparency. The study calls for adapting SSbD to address the specific challenges of biobased materials and the evolving needs of the construction sector.

Cities like Rotterdam, facing increasing flood risks, require integrated planning tools that account for the interplay between flooding and urban mobility. Despite advances in hydrodynamic and agent-based transport modelling, practical frameworks for coupling these models remain underdeveloped. This thesis develops an XLRM (eXogenous uncertainties, Levers, Relationships, Metrics) framework to guide the integration of flood simulation models (e.g., 3Di) with transport modelling (e.g., MATSim/SUMO) for robust decision making.
Drawing on literature review and expert interviews, the research identifies key sources of uncertainty and policy levers relevant to Rotterdam’s flood–transport system. The main contribution is the formalisation of the Relationships (R) layer: a set of rule-based protocols translating flood-model outputs into dynamic adjustments in transport models, based on empirically calibrated depth–speed functions and iterative data exchange at operational time steps. Optional feedback loops allow planners to test intervention timing and road management strategies under uncertainty.
Validation with domain specialists confirmed the framework’s technical soundness, logical clarity, and practical transferability. The resulting XLRM register offers planners a blueprint for embedding flood–transport interactions within digital-twin environments, enabling stress testing of adaptation strategies and supporting resilient, evidence-based infrastructure planning under climate uncertainty.

This study investigates whether Large Language Models (LLMs) exhibit victim blaming tendencies when analyzing traffic accident scenarios. Using a systematic factorial design, 144 road safety scenarios were tested across ChatGPT-4o and DeepSeek-V3, varying risk behavior, demographics, injury severity, and driving context. Through mixed-methods analysis of 288 responses, the research reveals that LLMs do not exhibit traditional victim blaming but demonstrate compliance-based attribution. They adapt their analytical approach based on question framing rather than maintaining consistent safety principles. When asked about prevention, models provided comprehensive systems analysis with 89.5% of suggestions targeting systemic factors. However, responsibility attribution followed context-driven patterns: 100% driver blame in private scenarios versus 69.4% company blame in work-related scenarios, regardless of demographics. The convergence of these patterns across different AI architectures suggests fundamental challenges in analytical consistency rather than model-specific biases. These findings have critical implications for AI deployment in safety-critical contexts, revealing that sophisticated responses can mask inappropriate analytical frameworks. The study contributes new theoretical concepts to AI ethics and provides evidence-based guidance for ensuring AI systems support rather than undermine advances in safety science.

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.

Hosting providers are essential for maintaining the security and reliability of digital services, but they continue to face challenges from malicious activities in their network, such as malware and phishing. The European Union’s Digital Services Act (DSA) was introduced to improve accountability and create a safer online environment, but its effectiveness in helping hosting providers mitigate abuse remains unclear. This study investigates whether compliance with the DSA has contributed to reducing malicious activity among Dutch hosting providers and examines the broader relationship between compliance levels and cybersecurity outcomes.

This study evaluates the effectiveness of anti-abuse measures employed by Dutch hosting providers, with a focus on the role of the DSA in helping with compliance and reducing malicious activity. Specifically, it examines whether adherence to the DSA improves the ability of hosting providers to mitigate cyber threats, particularly in reducing the prevalence of malicious IP addresses. Using passive DNS data, the research examines changes in the prevalence of malicious IP addresses before and after the implementation of the DSA. Compliance levels were also analyzed to understand their correlation with malware percentages. The study employed statistical methods, including Interrupted Time Series (ITS) analysis and regression models, to evaluate trends and relationships between compliance and malicious activity.

The findings indicate no statistically significant reduction in malicious IP activity following the implementation of the DSA, suggesting that compliance alone does not automatically translate into improved security outcomes. While the DSA strengthens transparency and procedural accountability, hosting providers continue to face operational challenges in implementing effective anti-abuse measures. Factors such as cybercriminal adaptation, enforcement inconsistencies, and resource constraints likely influence the weak correlation between compliance and actual abuse reduction. These results shows the need for a more holistic approach to cybersecurity regulation, combining technical advancements, industry collaboration, and proactive security enforcement alongside regulatory compliance. Evaluating the effectiveness of frameworks like the DSA is essential to ensuring that they not only establish compliance standards but also provide hosting providers with practical tools to enhance online security and mitigate digital threats effectively.

War is always evolving, becoming more complex over time. With the addition of new technologies and emerging threats the complexity on the battlefield is also growing. Following this trend tools to train new recruits or develop new doctrines also becomes more complex. One of the tools impacted by this is military wargaming. While wargaming is an ancient practice it has developed to follow the growing complexities of war. In the context of military training wargames portray multi-domain battlefield containing artillery, helicopters, tanks, infantry, planes, terrain and much more. To keep the wargames significant and executable the implementation of AI into military wargames was explored. This proved to have an impact on aspects such as data processing, decision making and strategies. Further research shows that while possible and useful the implementation of AI into military wargaming is of an ad hoc nature and lacks any guidance or standardization. That is why this thesis set out to develop a practical and guiding framework for wargame developers for the implementation of AI into military wargaming.

After an exploratory literature review the knowledge gap is identified in the crossing between AI, wargaming and their integration. This lead to the main research question being “Which framework can guide wargame developers in the field of AI and military wargaming through the implementation process ?”. To answer this main research question several sub questions are developed and answered using several research methods.

The first sub question is: “What are the current challenges and constraints in the implementation of AI into military wargaming and how can they be addressed? This question is answered in chapter one with the use of a literature review. It results in more depth into the problem and more insight into the knowledge gap.

The second sub question is: “What are the requirements, that the framework aims to achieve?” This question is answered in chapter two with the use of another literature review and expert interviews. This resulted in three design principles and several requirements. The principles, requirements and their sources can be found in table 4. Furthermore, the result of the expert interviews highlighted the value of a supplementary method, when academic literature is not abundant.

The third sub question is: “How can the design principles and requirements be translated into a framework?” Based on the requirements a preliminary framework is developed. This framework is than tested with a use case to find areas of improvement. The use case was a validation implementation of AI into the military wargame called “Take That Hill”. This resulted in a framework consisting of several phases and activities. The result can be seen in figure 7. Furthermore, the outcome of the use case illustrated the wargame is not properly balanced as it can be won by simply always moving forward. In addition to the direct relation between the requirements and the translation into the framework another finding took place. The use case illustrated the importance of testing the preliminary framework. As the use case uncovered a new dimension in the framework with the core activities, which were not apparent from the theory alone.

The last sub question is: “What is validity of the designed framework?”. This was answered by assessing whether the requirements were met and expert interviews. The requirements assessment can be found in table 6 and the final version of the framework can be found in figure 8. Furthermore, while the validation expert interviews directly improved the framework they also showed the importance of the diversity of experts interviewed. The experts for the requirements focused on time and security, while the experts for the validation focused on the ethical and moral aspects of AI in military wargaming.

In conclusion, this thesis presents a multidisciplinary approach to solving a problem in military training and doctrine development. The development of an practical and guiding framework for AI implementation in military wargaming represents a significant advancement in the field, offering both academic contributions and practical applications. By aligning with the goals of the Master’s program in Complex Systems Engineering and Management, this research exemplifies the integration of complex and complicated domains, ultimately contributing to the enhancement of military preparedness and operational effectiveness.

As this thesis was written as part of a graduation internship at TNO the research recommends the wargame developers of TNO to use this framework as a guide during future implementations of AI into military wargaming and expend on the framework when necessary.  

This thesis investigates the performance of the DeepCASE[13] alert-filtering system when incorporated into a multi-system security detection environment and compares its efficacy to other state-of-the-art alert-filtering tools, such as the Integration Layer in[30]. The main research question addressed is: How does the DeepCASE system perform in a setup with multiple security detection systems, and how does its performance relate to other state-of-the-art alert-filtering systems? This research is of critical importance given the growing complexity of cybersecurity environments, where various detection systems frequently create overwhelming amounts of alerts. Key Findings: • Metrics for Evaluating Alert-Filtering Systems: SQ1 looked into relevant metrics for determining the efficiency and efficacy of alert-filtering systems. It was shown that general performance metrics can be misleading in complex environments. Granular metrics, such as false negative and true positive rates, yield more meaningful evaluations. In this study, the false negative rate was ranked first, followed by workload reduction, as key indicators for evaluating DeepCASE. • DeepCASE Performance with Multiple Detection Systems: SQ2 evaluated DeepCASE’s per- formance in a multi-system setting. DeepCASE achieved promising results, with a 0% false negative rate and an 87% reduction in workload. These findings demonstrate its usefulness in managing alerts across several security detection systems while maintaining accuracy. • Improvements for Multi-System Integration: SQ3 looked at potential improvements to Deep- CASE’s performance in a multi-system environment. We tested with a variety of tactics, including alert aggregation and incorporating security analyst expertise. It was discovered that incorporating security analysts’ expertise via the ”IL Preprocessing approach” considerably increased performance, resulting in a 0% false negative rate and an impressive 96% workload reduction. This emphasizes the significance of expert-driven preparation in improving alert filtering. It should also be noted that out of all of the attempted alert aggregation approach, the 1-Minute window alert aggregation also managed to provide some improvement, achieving 0% false negative rate with a 90% workload reduction. • Comparison with State-of-the-Art Systems: SQ4 compared DeepCASE’s performance to a state- of-the-art alert-filtering system, namely the Integration Layer. DeepCASE performed remarkably well, outperforming the Integration Layer when not all of the latter’s preprocessing approaches were used. However, the paper notes that a properly optimized integration layer may have different outcomes. DeepCASE is highly effective in situations involving multiple security detection systems, significantly reduc- ing workload while retaining zero false negatives. However, expert-driven preprocessing techniques signifi- cantly improve its performance. While it compares favorably to other cutting-edge systems, further research is needed to fully understand the potential of these systems when completely tuned. This thesis advances our understanding of how alert-filtering systems can be evaluated and adjusted in complex, multi-system situations, giving useful insights for improving cybersecurity operations.

This thesis explores the phenomenon of greenwashing, particularly its systemic origins and contributory factors within corporate and regulatory frameworks. Through the analysis of three prominent case studies—Volkswagen, DWS, and Shell—it demonstrates that greenwashing is not merely the result of isolated corporate actions, but a complex, multi-layered symptom of wider sociotechnical influences. The research adopts Rasmussen’s risk management framework alongside Salmon’s systems-thinking approach to identify and map out systemic failures leading to greenwashing. Findings indicate that the most influential factors originate from management policies and regulatory oversight, showcasing the critical role of internal and external governance structures. The study’s conclusions emphasize the need for stricter regulatory audits and the cultivation of transparent corporate cultures. The analysis also provides insights into the ethical implications for sustainability-focused business practices, offering a taxonomy of systemic influences that contribute to greenwashing. This work aims to inform future strategies for regulatory bodies and businesses to mitigate greenwashing and promote genuine environmental responsibility.

This research integrates a socio-technical perspective into the safety risk management of ammonia-powered ships, with a particular focus on the engine room. The objective is to develop strategies that enhance crew safety by examining the complex interactions between humans, organizations, and machines within this high-risk environment.

The international shipping industry, responsible for a significant portion of global greenhouse gas emissions, faces an urgent need to adopt more environmentally friendly technologies. Anhydrous ammonia (NH3) has emerged as a promising alternative fuel due to its carbon-free nature and higher volumetric energy density compared to other fuels. However, ammonia’s toxicity presents substantial challenges, especially in enclosed spaces like the engine room, where crew members are at increased risk of exposure.
Current research predominantly addresses the engineering aspects of ship design to minimize ammonia leakage risk but lacks a focus on safety risk management through a human-centered approach that enhances crew safety. This research addresses this gap by adopting a socio-technical system (STS) approach, emphasizing the intricate dynamics and implications of human interaction with new technology. The study demonstrates that incorporating a socio-technical perspective through the Functional Resonance Analysis Method (FRAM) provides a comprehensive understanding of human-machine interactions, enabling the development of human-oriented modifications.

Employing a multidisciplinary approach, this research combines qualitative and quantitative methods. FRAM is used for qualitative analysis, offering detailed insights into interactions and performance variability within the socio-technical system. This research developed the complex STS relationships through Work-as-Done FRAM, identifying actor roles, time spent, location, and procedures related to engine room activities. Complementing this, Fault Tree Analysis (FTA) and Quantitative Risk Assessment (QRA) provide a deeper understanding of risk factors and their implications.

The findings indicate that adopting a socio-technical approach to safety risk management can significantly enhance the safety of ammonia-powered ships. This research has developed modification frame works that integrate human-machine interaction activities derived from FRAM with necessary system modifications. The study identifies on-board maintenance activities as having the highest probability of crew presence in the engine room (PER). By modifying the engine room layout and work procedures with the on-board maintenance as considerations, the Individual Risk Per Annum (IRPA) can be substantially reduced from the current value of 1.33E − 05 to 2.00E − 06. Normatively, these modifications result in an IRPA level that is lower than that of the currently operating LNG (CH4) powered ship engine room, which has an IRPA value of 3.16E −06. Thus, these modifications meet the risk acceptance criteria as outlined by the IGF Code guidelines.

The research provides several recommendations for stakeholders within the maritime industry. These include modifications to engine room layout and task locations to minimize human exposure to ammonia leaks, advancements in on-board maintenance procedures specifically for ammonia, enhancement of gas sensor response times, and digitization of the engine room logbook. These strategies aim to ensure safe operations and support the adoption of ammonia as a sustainable maritime fuel.

The scale and complexity of projects in the construction sector have grown significantly in recent years. Larger projects typically result in worse performance, including overspending and delays in the calendar.
This research focuses on identifying and analyzing Early Warning Signals (EWS) to mitigate schedule delays during the design phase of infrastructure projects managed by Sweco, a consultancy company in the Netherlands. The primary aim is to explore how effectively EWS can be used to identify potential issues early and prevent project delays.
The research begins by assessing the current knowledge and application of EWS among project managers, controllers and directors at Sweco. Initial findings indicate that while most participants have a basic understanding of the concept of EWS, they do not actively use or recognize it in their daily workflows. Their awareness of these signals primarily stems from gut feelings and extensive experience rather than systematic identification and application.
To identify the most relevant EWS, the study combines insights from existing literature with practical experiences shared by Sweco experts. This process involves creating a comprehensive list of potential EWS, which is then narrowed down and prioritised based on their perceived impact and likelihood. The prioritisation is similar to a risk assessment method, considering both the probability and potential impact of each signal.
The study examines how to measure data related to the signals and proposes measures on how to handle those EWS better. Some of the suggestions include establishing clear working agreements, balancing formal and informal communication, enhancing collaboration through workshops and visual tools, and leveraging advanced technologies like AI and BIM. The study further emphasises the importance of continuous client communication, internal quality checks, and ensuring appropriate expertise for project tasks.
The research provides a comprehensive framework for identifying, measuring, and managing EWS to improve project outcomes and reduce schedule delays in infrastructure projects. The findings emphasise the need for systematic approaches and technological integration to enhance early warning capabilities and project management practices. By implementing the recommended strategies, Sweco can better anticipate potential issues, make informed decisions, and ultimately deliver projects more efficiently.

Keywords: Early Warning Signals (EWS), schedule delay, design phase, infrastructure projects, project management, measurability, prioritisation

This study introduces HydroWizard, an innovative framework addressing critical challenges in water resource modeling through enhanced transparency, efficiency, reproducibility, and extensibility. Integrating a YAML-based Model Specification Language with a sophisticated Execution Engine, HydroWizard enables accessible modeling of complex water systems.

Applied to the Lower Omo-Gibe River Basin in Ethiopia, the study employs Evolutionary Multi-Objective Direct Policy Search to identify 283 Pareto-optimal policies. Findings reveal nuanced trade-offs: irrigation-optimized policies eliminate demand deficits but reduce environmental flows by up to 48%, while environmentally-focused policies show opposite effects. Notably, mean power generation remains relatively consistent across policies, challenging assumptions about water resource allocation trade-offs.

HydroWizard introduces novel visualization techniques, including animated rule curves and system state graphs, enhancing strategy interpretability. Its versatility is demonstrated through application to diverse water systems, including the Zambezi River Basin.

This research marks a significant advancement in water resource modeling, offering an open-source, accessible tool for complex water system analysis. It contributes valuable insights for sustainable water management and sets a new standard for global water resource management studies, emerging as an innovative solution to intensifying water management challenges worldwide.

Construction accidents are known to occur often; in 2022, there were 129 accidents in the Netherlands for every 100,000 workers . Beyond the tendency to place the responsibility on employees, this study investigates the impact of client involvement in enhancing safety performance. The majority of safety research focusses on the building phase, indicating that client influence decreases as projects progress. The study highlights 43 crucial client responsibilities with a focus on pre-construction phases, specifically in the Planning, Inception, and Feasibility; Design, and Procurement and Tendering phases. Moreover, Semi-structured interviews reveal differing practices between infrastructure and oil & gas projects, providing the opportunity to learn from these differences and highlighting areas for improvement, such as establishing stricter safety objectives, early contractor involvement, and enhanced knowledge sharing. The findings suggest that optimizing client roles in pre-construction phases can significantly enhance safety performance in construction projects.

The expansion of offshore wind farms in the North Sea has raised serious concerns about maritime safety, in particular the increased risk of collisions—or 'allisions'—where a ship adrift hits a stationary structure, such as a wind turbine. For Rijkswaterstaat, the executive agency of the Ministry of Infrastructure and Water Management, managing this expansion requires developing a comprehensive understanding of the integral image of allision risks of drifting ships in the Dutch North Sea region. This understanding must be valid, reliable, and insightful, and include factors such as dynamic weather conditions and shipping activity. Unfortunately, current state-of-the-art methods for estimating the probability of allision rely heavily on assumptions, leading to significant variations in risk estimates, as highlighted by Ellis et al. (2008). These methods usually lack a North Sea-wide perspective and instead focus on risks specific to individual turbines.
This study proposes an approach based on trajectory prediction using the OpenDrift model, which is simple and accessible for simulating ship drift trajectories based on wind, wave, and current data. A method is developed to estimate the probability of allision by combining the probability of a ship drifting into a wind farm, analysed through a ship's potential trajectories, with ship density data for specific locations. This method facilitates the calculation of allision risks under the influence of different environmental conditions and can be applied to multiple wind farms, providing a detailed assessment of the origins of potential threats and available response times.
Overall, the assumptions in state-of-the-art methods might not sufficiently capture the risks associated with drifting ships. In addition, the established method broadens the ability to improve monitoring accuracy and optimise the positioning of ERTVs, even for specific environmental conditions. As a next step, this method should be extended to estimate actual risk levels by considering the potential consequences of an allision. With this approach, decision-making regarding allision threats can be contextualised alongside other maritime safety risks, facilitating the development of a robust risk mitigation strategy that enables Rijkswaterstaat to responsibly exploit the abundant potential the North Sea has to offer.

This thesis investigates the integration of Large Language Models (LLMs) within large organizations operating in low resource language (LRL) regions, with a particular focus on Íslandsbanki, one of the prominent banks in Iceland currently grappling with the adoption of LLM technology. The research aims to understand the frameworks governing LLM integration within the bank, explore stakeholder perspectives on the integration process, and identify blind spots and gaps in their understanding of associated risks in order to understand what a comprehensive approach would look like.

The methodology had mixed methods. It consisted of a comprehensive literature review followed by interviews with stakeholders involved in the LLM integration at Íslandsbanki. Subsequently, an Impact-Probability Assessment survey is conducted, where stakeholders are asked to evaluate the identified risks in terms of their potential impact and probability. This approach facilitates the identification and prioritization of risks, thereby enabling future integrators to navigate the integration process more effectively. Furthermore, data analysis techniques were applied to the survey results to provide quantitative insights into the identified gaps and perspectives.

Findings reveal that perspectives on risks associated with LLM integration vary significantly, particularly between individuals with technical backgrounds and decision-makers within the bank. Moreover, the concept of text inflation emerges as a novel risk factor, highlighting the importance of considering diverse perspectives in risk assessment and mitigation strategies. Emphasis is placed on adopting a sociotechnical perspective to comprehensively assess risks at an organizational level.

This research contributes to the understanding of the organizational impact of LLM integration, offers insights into diverse risk perspectives, and underscores the importance of considering sociotechnical factors in the integration process. By shedding light on the challenges and opportunities associated with LLM integration in the financial sector, this study aims to inform future research and practice in this evolving domain.

The abstract outlines a study focusing on improving the design approach for flexible dolphins, vital marine structures used for vessel berthing and mooring. Current design methodologies, particularly those outlined in the CROW C1005 handbook (2018), are questioned due to potential conservatism stemming from insufficiently calibrated partial factors. The study advocates for reliability-based assessments to address these concerns, which consider uncertainties inherent in dolphin design and quantify failure probabilities over their lifespan. The investigation identifies critical failure modes and determines main pile dimensions based on these modes, utilizing the API PY-curves for rapid computation in the design process. Probabilistic assessments, employing Directional Sampling, reveal the significance of berthing load in structural safety and identify soil parameters as dominant variables affecting fixity failure mode. Navigation conditions, ship arrival rates, and variation in ship sizes also influence partial factors, with recommendations provided for adjustments based on different conditions. The study suggests load testing to reduce uncertainties and increase reliability, exemplified by a Bayesian update from Calandkanaal full-scale load tests. Overall, reliability-based assessments yield insights into managing uncertainties in dolphin design, potentially reducing material usage by up to 20% while meeting safety requirements, and offering the potential to decrease failure probabilities tenfold when combined with test loading.

The thesis titled "Towards Smart Maintenance: The Implementation of Predictive Maintenance in the Railway Industry" focuses on integrating predictive maintenance (PdM) technologies into railway operations, emphasizing freight rail. PdM employs data analytics, machine learning, and IoT-enabled sensors to shift from traditional maintenance approaches to a proactive, data-driven methodology. This transformation promises reduced downtime, optimized resource use, and enhanced system reliability.

The research identifies technical, financial, and organizational challenges as key barriers to PdM adoption, including data standardization, algorithm maturity, high initial investment, and resistance to change. Drawing lessons from industries like aviation and infrastructure, the study highlights phased implementation, regulatory frameworks, and stakeholder collaboration as critical enablers.

Employing tools like Interpretive Structural Modeling and Fuzzy MICMAC, the thesis maps interdependencies among barriers and demonstrates PdM's potential for reducing maintenance costs by up to 47.8% over two decades. It concludes with recommendations for systematic adoption, addressing both technical and managerial factors, to ensure safer and more efficient railway maintenance practices.

Phishing attacks are a growing cause of cybersecurity incidents such as data breaches. With these attacks, malicious actors try to gain access to systems by exploiting the vulnerability of employees. Particularly, intruders use different tricks to create convincing phishing emails to force people to behave less securely. Unfortunately, technological applications are insufficient in detecting all possible phishing attempts. As the consequences of a successful breach can be disastrous for a company, especially in the banking sector, pressure is put on employees to recognise and report potential phishing emails. To aid them in this task, phishing training is provided by employers. Employees are taught what phishing emails look like and which actions they should take if an email appears malicious. However, the effectiveness of these training sessions is difficult to evaluate.

To understand the interaction of employees with phishing emails without influencing their behaviour, we study the emails employees report as suspicious to characterise the security culture at a bank. A better understanding of the behaviour provides grounds for recommendations to improve anti-phishing training and create a safer environment. The newfound metrics can provide an alternative to current methods.

This research uses Exploratory Data Analysis (EDA) to evaluate the email reporting behaviour of employees at a bank to answer the Research Question How can email reporting patterns in a large organisation measure the relationship between phishing training, reported emails and employee behaviour? With this case study, we apply EDA to a large dataset containing bank employees' reported emails over 16 months. We analysed the reported emails and related them to the provided phishing training events. Moreover, we did a text analysis of the emails' content using the Term Frequency - Inverse Document Frequency (TF-IDF) method. Additionally, we extract the dominant topics of the emails using topic modelling. Lastly, with the help of interviews, the results are tied to employees' experiences to understand their behaviour.

The major findings of the research are, firstly, the new metrics we identified to measure the security culture of a company. These metrics were found from both the analysis of employee behaviour over time, as well as the analysis of the email content. From the analysis of the reporting behaviour over time, new metrics include the unique reporters in relation to the total reported emails over time. Besides the unique reporters, unique reported emails can uncover the presence of campaigns. For example, a single email can explain the increase from 50 to 350 daily benign reports. The difference between the total reports and unique reports uncovered this. Secondly, topic analysis and content comparison show similarities between benign and malicious reported emails, indicating an increased vigilance of employees on these attributes.

A second finding originates from the analysis of the email components. One of the components was used in all simulation emails, while it was not present in all the benign and malicious reported emails. This shows that the simulation emails can be extended to include different scenarios. Therefore, we recommend the company to extend the phishing simulation emails to contain varied phishing tactics to expose employees to other types of attacks and incorporate all aspects taught during the E-learning.

Lastly, the analysis shows no concrete relation between the number of reported emails and the timing of the simulation wave. Although the reported p-value of benign emails after the simulation is 0.03, this significance can also be explained by external factors.

With these results, we can measure employees' security culture and awareness in real-world circumstances without influencing the employees' behaviour, providing a new approach to investigating phishing behaviour. Adding to the research of Steves et al. (2020),, the click rates can be explained by more than solely the employees' awareness levels, and new explanations come forward to handle phishing threats. Moreover, the absence of a required test environment for the analysis created a solution for existing gaps. For example, as seen in (Hillman et al., 2023).

A limitation of exploratory data analysis is that results are often ambiguous and mainly provide possible directions for future research. Furthermore, external factors influencing the behaviour could provide alternative reasons for the discussed interactions.

To conclude, by using the reported emails to measure security behaviour related to phishing, we found new metrics which do not influence the employees in their daily behaviour while still providing insights to improve the tactics of a company in combating phishing attacks. Reporting behaviour can be used to analyse the current anti-phishing tactics of a company and provide suggestions for improvements.

Future research should explore the differences in applying the method in other companies and across sectors. Overlap and differences can create an understanding of the diversity in security culture and the effect of external factors. Combining these results with a comprehensive understanding of a company's operations can expose directions for improvement in the security approach. Additionally, the effect of the recommendations can be analysed using the metrics we proposed. This can be done with a follow-up analysis of the behaviour to see whether the desired effect can be observed.

The Sahel region has become a focal point of concern due to the alarming escalation of terrorism, culminating in devastating attacks that have claimed numerous lives and disrupted regional stability. Mali, in particular, has experienced a significant surge in terrorist activity. This thesis presents a novel approach of using criticality measures based on network science as pay-offs for a game theoretical model in order to protect connectivity of road networks in conflict areas. Through two comprehensive literature reviews the network scan approach, based on shortest paths, population sizes, travel distances, and travel times and the two-player zero-sum game structure were selected as the basis for the methodology. In order to enhance understanding of dynamics on the road network, semi-structured expert interviews were conducted. The model is applied to the region of Segou in Mali and optimal strategies were compared to empirical terrorist attack data of the ACLED database. It is concluded that road network protection can be strengthened to minimize the impact of terrorist attacks through the allocation of protective resources. Firstly, experts highlight the positive impact of combining military patrols and convoys with explosive ordnance disposal on road safety. Secondly, results revealed the network scan approach's potential as a valuable tool for criticality assessment. Thirdly, contexts lacking empirical traffic data, directing network protection efforts towards roads identified as optimal strategies by the model becomes prudent. Lastly, the comparison of optimal strategies against emperical data provides the means enhance insurgent behaviour's understanding. While the developed model can be considered rudimentary, it is believed to have laid down essential foundational principles. Further research is recommended the extend the model further, including validate the model's effectiveness and determine its applicability to different types of attacks and different insurgency groups. This thesis hopes to have made a contribution to the academic fields encompassing counter-terrorism, network science, and game theory within conflict settings.

As the impact of climate change becomes increasingly urgent, conflicts over transboundary water resources have become more complex, particularly in light of the role of water in shaping socioeconomic and regional power dynamics. The challenge of optimizing objectives and strategic behavior across multiple stakeholders simultaneously makes it difficult to arrive at policies that can enhance long-term transboundary water management in a harmonious manner. In the Eastern Nile Basin (ENB), tensions have risen between upstream Ethiopia and downstream Egypt and Sudan due to the development of the Grand Ethiopian Renaissance Dam (GERD), making successful negotiations for water allocation critical. Ideally, such negotiations should produce agreements with high-stability policies, meaning that no actor has an appealing unilateral gain in utility that would cause them to defect from an agreement.

Several water allocation simulations have been developed in the ENB using both linear and nonlinear programs to aid decision-making, but only two studies have previously examined measures of stability for these optimization results, and neither has been done since the completion of the GERD. Moreover, as simulation complexity has increased, there is a gap in knowledge regarding the measurement of stability using optimization results from closed-loop, multi-objective adaptive simulations.

To address this gap, this research reexamines the stability of policy candidates for water allocations in the ENB using three different solution concepts from cooperative game theory—the Nash-Harsanyi solution, the Shapley value, and the nucleolus. The stability of each policy candidate is assessed using three different stability metrics—the Euclidean distance, the Loehman Power Index, and the propensity to disrupt—to determine their relative stability. The approach yields similar objective trade-offs and utility behaviors for the Nash-Harsanyi and Shapley Values. The most stable policies, when ranked by Euclidean distance, prioritize Ethiopia’s utility, while policies become more unstable with the rapid growth of Egypt’s utility. Furthermore, propensities to disrupt and power indices between Egypt and Ethiopia or Sudan show converging and diverging behaviors, respectively, which explain the negotiation potentials between the players. Our results indicate that Egypt’s willingness to engage in collaboration is directly related to its level of utility; however, at these levels, Ethiopia and Sudan’s benefits from utility are at levels that prompt higher likelihoods of defections from a potential coalition. The results also showed stable policies characterized with high policy efficiencies in instances of basin-wide cooperation, which increases benefits to all nations.

Given the different assumptions and characteristics of each stability concept, the insights on the stability of different policies provide general guidelines for incorporating stability into the optimization formulation itself. The results have larger implications for policy planners in the ENB and the difficulties they may face when finding acceptable solutions in multi-stakeholder decision arenas. Furthermore, the methodological contribution of this study could allow for easy application of this method to other water allocation conflicts to help guide policy planners detect opportunities for utility optimization or risk mitigation in a cooperative setting.