To advance the energy transition and prepare for the accompanying increase in electricity demand, the electrical grid of the Netherlands requires rigorous investments for expansion and maintenance. In order to solve and further prevent grid congestion, a lot of projects have to be executed and this demands collaboration of grid operators and local governments. In addition to the technical problems of the transition, there is also a growing part of the population that is suffering from energy poverty and energy injustice which prevents them from benefiting from new innovations, while wealthier households are financially profiting from these. The negative socioeconomic effects are not accounted for in the investment plans of grid operators, even though not addressing these problems will further increase inequality in the population.
In order to research how investments in the electricity grid are related to the socioeconomic status of neighbourhoods and how this affects equality in the province of South-Holland, this thesis is conducted. The study aims to answer the research question: "To what extent can spatial relations between electricity grid infrastructure and socioeconomic characteristics be identified and quantified in the province of South-Holland and what does this relation imply for inequality?"

To answer this question, first a literature study was conducted in order to identify socioeconomic variables related to energy poverty and energy justice to use in the research. Subsequently, data for these selected variables is collected on a neighbourhood level and the electricity infrastructure in the province of South-Holland is mapped. The research focuses on the low and middle voltage grid infrastructure as these grid levels directly influence households and their impact can be assessed on a neighbourhood level.
The collected data was then used in a correlation analysis to discover a possible relationship between the socioeconomic status of a neighbourhood and the amount of low and middle voltage stations located in it. As the electricity demand of wealthier households is higher than the demand of lower income households, it was expected that there would be a positive correlation between prosperity and electricity infrastructure. The found correlations, however, did not result in any definitive conclusions, except for the relation between the amount of inhabitants and the amount of electricity infrastructure, so another analysis was performed that focused more on the spatial correlations. In this hotspot analysis the variables with a strong relation from the correlation analysis were used in a clustering research. This resulted in indications being found that linked areas of socioeconomic prosperity to more investments in electricity infrastructure, as was assumed. This implicates that there is an unbalanced distribution of costs and benefits on the grid as everyone pays the same for the investments, but wealthier neighbourhoods profit more.

In order to learn how this relation came to be and how it affects equality, experts on the topic were interviewed. In total five interviews were conducted with people from main stakeholders in the system, namely: grid operators, municipalities and researchers on energy justice. The interviewees were asked about their thoughts on the results of the analyses, the developments of the electricity grid, its effect on equality and the role of the province of South-Holland in the system. From these interviews it was concluded that the indicated unbalanced distribution of electricity grid investments among different socioeconomic groups in the population is not necessarily a problem for equality on the short term, although it might provide difficulties in the future. The investments follow demand and thus it is a logical consequence that neighbourhoods with a higher electricity usage require more electricity infrastructure.
The problem in the system is that there are socioeconomic groups who have no access to the energy transition and can therefore not benefit from the developments, while others are enriched by it. This unequal balance of costs and benefits causes inequality to grow in the future, so governments will have to ensure lower income households have access to energy transition innovations in order to secure a balanced and sustainable future.

Poor and coastal regions are increasingly at risk from the effects of climate change. These risks are accompanied by a high level of uncertainty, which is also called deep uncertainty. Current planning approaches lose efficacy under deep uncertainty, necessitating new approaches that function better under these conditions. Decision making under deep uncertainty (DMDU) is the name for the family of approaches that attempt to deal with this level of uncertainty. Bangladesh, the Netherlands, and New Zealand have all already adopted the use of DMDU techniques. Research into local implementation of these techniques is also being done.

Two DMDU techniques often deemed as complementary are robust decision making (RDM) and dynamic adaptive policy pathways (DAPP). RDM can be seen as a computational extension of scenario planning, where proposed plans are tested against every potential combination of uncertainties. DAPP is a flexible policy framework that allows decision-makers to keep long-term plans in mind while making short-term decisions. DAPP especially has seen increased adoption in national delta protection plans such as in the Netherlands, Bangladesh, and New Zealand.

To design DAPP, currently a combination of many-objective robust optimization (MORO) and participatory processes are used. These methods both have their own issues. MORO requires the upfront specification of rules and policies and is computationally expensive, while the participatory approach is qualitative and can be insufficient when dealing with complex systems. RDM is seen as a potential improvement in supporting the DAPP policy structure in two main ways. First, RDM can be used to iteratively develop and/or stress-test potential actions and pathways. Second, the vulnerabilities identified through RDM can be used to lay the base for a monitoring system by identifying promising signposts and signals.

While RDM is seen as a potentially helpful tool to support DAPP, there is a lack of studies that have established a systematic and analytical approach which uses the robust decision making process to support the development and monitoring of DAPP. This research proposes a novel approach based on literature to achieve this. The approach uses the vulnerabilities identified through RDM to iteratively inform and develop more robust actions and to lay the basis for the technical side of a monitoring system. This approach is then illustrated by way of the adaptation case of a wastewater treatment plant in Helensville, New Zealand. This wastewater treatment plant serves a small community and will have to retreat at some point in the future due to increasing risks from compound flooding, which are exacerbated by rising sea levels.

The results of the case illustration show the benefits of using RDM to better understand vulnerabilities in the system in two main ways. First, the vulnerability analysis (which included a sensitivity analysis) helped to identify factors most important to the outcomes to inform potentially effective actions. Second, RDM helped in the development of the monitoring system. Those factors making up the identified vulnerabilities formed the basis of the technical signposts selected. Using the coverage-density tradeoff from the scenario discovery results, promising signals could be selected, although timing was not taken into account. This could potentially partially solve a common problem for monitoring DAPP: the selection of trustworthy signals.

There were three main recommendations. The first is to further work through a case such as this, since due to time constraints only the first iteration of the process was followed in this research. This could help identify more potential benefits or issues. A main issue here is also how to identify when an action is fully developed, as the process could continue indefinitely. Second, it is recommended to do further research into determining adaptation tipping points using other scenario discovery methods, and to use the coverage-density tradeoff from the scenario discovery results to modify adaptation tipping points based on policy regret. Third, is to further the monitoring system by posing open questions to support the deliberation on signpost and signal selection, taking timing into account to identify triggers, and by adding a signpost map next to the signal map to visualize signpost interaction, hierarchy, and quality.

In October 2023 congestion is issued in the HIC Rotterdam. As a result businesses can no longer establish, expand or electrify. A potential solution to free up grid capacity is that of an energy hub. This study evaluates future scenarios for the development of an industrial energy hub by utilising the EMA method. A study is done on the details of the congestion problem in the HIC Rotterdam in order to distinguish possible solutions. By exploring technologies applicable in an energy hub and applying exclusion criteria the potential for energy hub integration in the HIC Rotterdam is assessed. Based on the outcomes of a data inquiry a choice is made on an industrial cluster to model. Factors influencing the cluster’s development in the future are drafted and scenarios are designed that describe the potential evolution of the cluster. The electricity system of the industrial cluster is modelled and its data analysed. The model of the electricity system is translated into a model that generates future states of the industrial cluster. Through EMA these scenarios are evaluated.

The congestion problems faced in the HIC will last until project ’Loadpocket Simonshaven’ is commissioned in 2027-2029. Until then, energy hubs might be valuable in creating grid capacity. Making shared use of an electricity connection can already create some extra capacity, however the implementation of large electrification projects is most influenced by the availability of additional grid capacity. As becomes clear from the simulation results. In addition, a promising integration option is the inclusion of an electricity generation unit in the energy hub of which multiple exist in the HIC Rotterdam.

This research examines the resilience of the water sector after a societal collapse, defined by a significant loss of societal complexity. Through Exploratory Modeling and Analysis, this research evaluates how interconnected infrastructures respond to cascading failures caused by disruptions resulting from a societal collapse. The investigation centers on the interdependencies within the water sector, crucial for its function to abstract, treat, store, and distribute water. The model makes use of disruption curves affecting a network of interdependent components to simulate cascading failures, factoring in variables such as population change and contamination rates.

The research makes use of expert interviews and economic Input-Output tables to devise the dependency matrix, finding that accurate representation of dependencies is critical for meaningful results. The model allows for adaptation strategies, specific to the dependencies of a sector, which demonstrates potential improvements in water sector performance and suggests that adaptive methods may be more effective and less costly than redundancies.

While the model simplifies complex interactions and cannot fully encapsulate the breadth of societal collapse effects, it sheds light on the importance of dynamic dependencies and offers a framework for future studies. These could include integrating more intricate domain-specific models to refine the responses of sectors to disruptions and reduced input. The findings highlight the significance of adaptation as a means to enhance the water sector's resilience.

As the world grapples with the escalating challenges of climate change, Carbon Capture and Storage (CCS) emerges as a crucial solution, especially for decarbonizing dominant industries such as steel, cement, and petrochemicals. The Intergovernmental Panel of Climate Change (IPCC) and the International Energy Agency (IEA) have highlighted the potential of CCS in addressing the pressing need for large-scale decarbonization. However, the widespread adoption of CCS is hindered by various challenges, including financial, organizational, technical, and governance aspects. Within this context, the research delves into the central question: "How can an evaluation framework combining both project- and policy-related factors provide a better understanding of the deployment of full-chain CCS projects, and how is it developed and implemented?"

To address this pivotal question, the study employs a multifaceted methodology. Literature research, semi-structured expert interviews, the Best-Worst Method, and a project maturity table are utilized to derive comprehensive insights. Factors influencing CCS projects are systematically categorized into six groups: political-legal, economic, social, technical, ecological, and organizational. Additionally, the research presents two in-depth case studies, ROAD and Longship, offering a practical lens into the challenges and triumphs of real-world CCS projects. The culmination of the research is a robust evaluation framework that provides a systematic approach to assess various facets of CCS projects. This framework emphasizes critical areas for resource allocation, stakeholder communication, legal clarity, and financial stability. In essence, this research not only addresses its foundational question but also crafts a multifaceted tool that stands to benefit stakeholders, guiding informed decision-making and enhancing the prospects of CCS projects' success and longevity.

Yearly, 20.000 fatalities are recorded in road accidents on European roads alone. This situation is far more severe (1.3 million humans) when the global scope of road accidents is considered. Thus, Policymakers are initiating policy programs to overcome this challenge by enhancing safety and maximizing the associated positive economic impact. The EU committed to minimizing fatalities by 50% by 2030 and to be net zero long-term by 2050.

To achieve this, vehicular communication networks (VCN) can play a crucial role. It is anticipated that the connectivity and communication of vehicles with their environment contribute to these goals by enabling safety functionalities or improving the traffic flow. Hereby, it is crucial to understand that this communication requires an infrastructure that assists the vehicles and ensures data transmission and operation. However, due to the European size and heterogeneity of the member states, realizing a vehicular communication network on a large scale requires increased coordination and collaboration efforts. Currently, no vehicular communication network has been introduced working EU-wide. Hence, this resembles a potential for future road improvement. To realize this, decision-makers must understand the complex properties and system interrelationships of such a large-scale infrastructure project so that a common VCN can be designed to ensure interoperability and robust functionalities across the EU. This work contributes to the aforementioned challenge by addressing the following research question:

What are fundamental socio-technical factors to consider in a future European vehicular communication network design?

A mixed approach that combines Peffer´s design science research framework and system engineering methodology is used to synthesize the contributions. By this, a vehicular communication network mission, stakeholder, and system analysis are presented in this work. Further novel and scientific sound requirements and stakeholder insights are synthesized by systematically reviewing 57 articles and interviewing 15 experts from the institutional, scientific, and industry domains. In this context, a stakeholder classification for vehicular communication networks, a 4-layer stakeholder complexity model, and a system requirement structure from a system perspective are proposed to contribute to the VCN understanding and future design attempts. Further, the reflection of the socio-technical interrelations between the technical and social VCN subsystems are subjects of this work.

The conclusion is that the reflection on socio-technical system properties plays a critical role in vehicular communication network design. Further, a future vehicular communication network consists of a magnitude of stakeholders with high interest and power; thus, designers must understand the characteristic of their co-evolutionary cooperative development. Hence, a multidisciplinary understanding and approach are critical for designers. Furthermore, the geographical segregation of designing and decision-making in a future vehicular communication network is identified, and certain goals/issues should be addressed in the respective layer. Another major conclusion is that the VCN discussion is determined more by social and socio-technical conditions, such as stakeholder cooperation/coordination and interoperability, than by technical feasibility.
These findings result in three main research contributions, which are summarized as follows:

- A 4-layer stakeholder complexity model contributing to the understanding of VCN development is contributed.
- An approach of integrating the socio-technical system perspective on the complex, large-scale infrastructure VCN project is contributed. This focuses on the processes and requirements between social and technical subsystems, addressing the integration of heterogeneous stakeholder interests.
- A comparison is made between scientific focus, stakeholder needs and objectives, and expert insights, highlighting the mismatch and alignment of requirements. This contributes valuable insights for adjustment and further research in the VCN community.

Further methodological contributions can be concluded:
- The design science research approach is aligned with systems engineering iso standards and methodology.
- The design science research framework is used as an approach to address the complexity of VCN systems. This novel perspective helps VCN stakeholders design solutions for their field problems.

The results have implications for designing stakeholders in a VCN. Based on the analysis, it is recommended that policymakers identify and extend common objectives with the industry to establish public-private business cases. Further, vehicle manufacturers should participate and embrace the transition to a vehicular communication network by cooperating strongly with stakeholders. In addition, implementing the relevant technologies to communicate with the heterogeneous infrastructure is suggested to shape future infrastructure connectivity development. Lastly, the lack of socio-technical reflection in scientific literature is identified. Hence, scholars should elaborate on the interactions and interrelationships between the social and technical subsystems in future work.

In light of worsening climate change and an increased interest in adapting infrastructure to cope with its effects, model-based decision support has become an essential tool for policy makers. In conditions of deep uncertainty, models may be used to explore a large space of possible system behaviours and so encourage a wider consideration of the possible futures. Such methods, where the focus is intentionally broad, fall under the remit of exploratory modelling and are a potential antidote to traditional predictive modelling methods where only a marginal treatment of uncertainty is attempted. One serious issue limiting the full exploitation of exploratory modelling is its computational intensity, the many computational experiments requiring large amounts on computing power which makes some analyses too expensive to attempt. In order to fully exploit the promise of exploratory modelling new methods of reducing computational intensity are needed. Polynomial chaos expansions (PCEs) are one class of methods which may fulfill this role. Our results conclusively demonstrate that PCEs are capable of accurately reproducing statistical moments and determining Sobol sensitivity indices significantly faster than direct-sampling methods, often requiring orders of magnitude fewer function evaluations. However, we found that the curse of dimensionality rendered conventional PCEs too costly for use with higher-dimensional models. We found that conventional sparse grids were effective at reducing the computational cost associated with fitting PCEs with high-dimensional models, as long as the model output was sufficiently smooth. For models where sparse grids were able to converge with reasonable accuracy, supplementing the PCCs with a modified Gersnter's dimension-adaptive algorithm further improved convergence times. The anisotropic refinement strategy employed by the algorithm allows for accurate determination of Sobol sensitivity indices with a minimum of computational effort.

Climate change will lead to more extreme climate and weather phenomena, and this includes the increased risk and severity of flooding. Flooding is one of the most destructive and common natural hazards globally, and rising sea levels and extreme precipitation mean that many human settlements will be situated in climate-sensitive areas.

However, the increased threat of floods in coastal cities is not currently sufficiently represented in the housing markets. Contemporary property prices are shaped by the locational advantages of coastal cities, and have a tendency to underestimate or ignore flood risk, even after experiencing a flood themselves. This increases the possibility of structural shifts in housing markets when the flood risk is realised by the public, leading to a depreciation of real-estate property values, and may cause cascading effects in other aspects of the economy, such as the collapse of the regional housing market and reduced economic attractiveness to the region. The Netherlands, for example, is arguably the best-protected delta in the world from flooding. Still, it is vulnerable to the "safe development paradox", where public flood protection measures motivate the continued investment and expansion of flood-risky areas. While Dutch safety standards are very high, they cannot guarantee absolute protection from floods; the country is still at risk from rare-but-severe flooding, or the occasional minor flood, both of which can influence the Dutch housing markets.

The deep uncertainty of future flood risk means that empirical research alone cannot sufficently describe possible responses to housing market shocks from flooding. Therefore, modelling and simulation is useful in testing potential variants of a housing market system, such as testing different dynamics of housing market actors and potential policy levers in various flood scenarios. This is further supported by an increase in publicly-available rich datasets, that allow for spatial and empirical representation of the housing market and climate-induced flood scenarios.

\noindent In this thesis, the city of Rotterdam is used as a case study, to explore the usage of empirical data and to model a housing market shocked by various plausible flood scenarios. This is done in two steps: firstly via a data exploration effort in publically-available datasets for the Netherlands, and then consolidating the data into a stylised agent-based model, simulating the transactions of the housing market while several districts experience flooding shocks from different flood scenarios.

Firstly, I have conducted the data exploration aspect using Dutch open data for flood scenarios, housing, local demographics, and empirically-estimated flood discounts. The data was judged on the suitability of the datasets to be incorporated into an empirical model, and on the presence of data gaps while linking between different data. The exploration highlights the potential of modelling the housing market using empirical demographic data based on income level, but only missing certain data to characterise how homebuyers would acquire mortgage financing. Additionally, on the flood damages side, I show the possibility of characterising flood discounts as a function of flood depth, and the depth-damage relation for the Netherlands.

In the modelling part of the thesis study, I have consolidated the data into an agent-based model with a stylised set of relationships for the housing market. The model was designed based on the empirical stylised trends regarding housing markets dynamics, like declining flood risk discounts in property prices over time. In short, the model simulates the purchase transactions of homebuyers, who are discouraged from flood-affected properties, thus leading to a growing demand and price premium for flood-safe properties. This model was then tested with a simple series of experiments, for single flooding and multiple flooding scenarios, based on empirically-grounded severe flood scenarios on Rotterdam. While the model results are limited in terms of prediction for policy purposes directly, the modelling process as a whole illustrates the value of exploratory modelling to refine the understanding of a system. Via this exploration, it was highlighted that there is a need to further characterise flood discounting behaviour around the action of housing market actors.

The use of Automated Decision Making (ADM) systems in the public sector will become increasingly prevalent in the future, making citizens increasingly likely to be confronted with decisions that have been made fully automatically, without human intervention. Ever more digitization increases the opaqueness of the social benefits system in the Netherlands and leads to citizens facing most of the adverse, sometimes unforeseen effects of automation. Why and how unsafeness and unexpected effects emerge is not well known. In this research, these effects, defined to be ripple effects, are explored using a system safety lens and proposing an agent-based model design to explore why and how the effects emerge.

The leading cause of ripple effects was found to be the lack of a hierarchical safety structure, seen in laws being organized per organization, and the lack of an overarching controller to ensure that overlap does not occur.
This is partly due to a lack of an accurate model representing a Dutch citizen. Several policy recommendations are presented, including a rudimentary set-up on how to apply a safety control structure within the social benefits system.

With the population growing and countries becoming more affluent, diets shift. These global dietary shifts contribute greatly to increasing greenhouse gas emissions because of their higher carbon and water intensities. This development is problematic considering both the growth in our global economy as well as our population. It is therefore crucial to find ways to reduce the impact of food systems. The mesopelagic zone in the ocean has recently been found to host a vaster source of biomass than expected, however, designing harvesting policies poses a deep uncertain issue. The mesopelagic zone promises a new source for the human food supply chain, especially serving as fish meal for aquaculture. Aquaculture is currently the worlds’ fastest growing food sector. However, unlocking the potential of the mesopelagic zone has challenges. The working of the mesopelagic zone and its ecosystem are still vastly unknown, as is the magnitude and variety of impacts harvesting might have on this. Moreover, the zone is located at a location beyond national legislation and still lacks overarching regulation. To acknowledge this research gap, this study focuses on how robust decision making methods can support decision making on harvesting in the mesopelagic system. This is done by analysing the current governance arena, the working of the mespelagic zone, and finally applying these findings to a System Dynamics model and performing Many-Objective Robust Decision Making (MORDM). Ultimately, to understand possible mesopelagic futures and thus gain governance relevant insights on harvesting in the mesopelagic zone. Exploring the model behaviour over a variability of plausible futures demonstrated that several uncertainties have a crucial impact on the mesopelagic system's best and worst case behaviour. It seems that profitability of mesopelagic harvesting relative to regular fishing is the most prominent indicator for collapses of the mesopelagic biomass. Additionally, factors concerning low food availability seem to be strongly linked to biomass collapse and vice versa. Moreover, reproductive parameters (e.g., a high female fraction of the biomass) as well as a high ratio of myctophidea to Maurolicus muerelli positively influenced the magnitude of food provision. Furthermore, other noteworthy findings are that an optimal harvesting quota seemed to act as a balancing tool for profitability. Additionally, exploration of the uncertainty space clarified which uncertainties have the strongest influence when no harvesting limits are in place. The two most influential uncertainties are the profitability of mesopelagic fish versus regular fish and the type of risk reward mechanism. This finding reduces the uncertainty space for decision makers by clarifying which indicators to focus on and which uncertainties might not even be relevant to consider for decision making. Moreover, it can be concluded that by harvesting according to the selected policies, the atmospheric carbon levels in 2100 would rise significantly enough to cause added pressure on the climate. This strengthens the call for considering climate change as well as biomass conservation when thinking about mesopelagic governance In conclusion, this research found several governance relevant insights for harvesting in the mesopelagic zone. The set of policies and the insights they provided serve as a promising starting point for stakeholders to learn about harvesting in the mesopelagic zone. Considering the mesopelagic decision arena, this study shows that current marine governance is unprepared to manage the magnitude and variety of the impacts harvesting can have on the mesopelagic biomass and its carbon sequestration function. Also, this research strengthens the call for including the mesopelagic zone and its role in the climate in future governance. These findings facilitate a discussion between stakeholders, which was previously impossible due to the deep uncertain character of the mesopelagic zone. This study is a first demonstration of the added value of MORDM for decision making on an intricate system like the mesopelagic zone.

Agent-based simulation models are rising in popularity recently due to their ability to model real-world problems in a wide range of domains. Inherent to these types of simulations is the fact that an enormous amount of data can be generated, which needs to be analysed in order to make the simulation useful. At the moment, the available tools to perform this analysis are limited for most platforms in terms of functionality or usability. Model designers are often not experienced programmers, so processing and analysing the data from these experiments in Python or R can be challenging. The goal of this master thesis is to determine the best way to make data mining methods accessible to model designers so they can easily execute and analyse model experiments for NetLogo. In order to achieve this goal, the limitations of the currently available tools will be established by communicating with the target group. When there is a clear overview of what exists and what still needs to be done, we will attempt to develop an accessible tool that addresses the current limitations. The tool that was created consists of a GUI that allows the user to design experiments, run simulations and visualise results. With this GUI, the coding aspect of data analysis is taken completely out of the users hands, so anyone with NetLogo experience should be able to conduct extensive analysis of their model and its results. To determine whether the tool achieved its goal of being easy to use while still providing advanced analysis options, a usability study will be conducted with the target group. In this study, the users will execute a certain number of tasks with the tool and the results are recorded. After the tasks are completed, the user will fill in a questionnaire to get more opinion based data to give a complete overview of the usability. This usability study will allow us to determine whether our tool is indeed an improvement for the field and something that people could see themselves actually using during model development. Every measurement goal that was set before the usability study was achieved, most of them with wide margins. The ASQ questionnaires obtained an average result of 4.69 out of 5, where the target was set at 4. Furthermore, the SUS questionnaire obtained an average result of 85 out of 100, with the target being 75. These results show that developing an accessible data mining tool is possible and thus that the goal of this thesis was achieved. This research can be used as a basis for future development in order to provide the agent-based simulation community with more tools for the analysis of models.

EPA: Multi-Modelling for Complex Challenges With the acknowledgement of international grand challenges, the demand for better and ever more complex models is likely to increase a manifold. Where single models may no longer be able to grasp the complexities of a system, multi-modelling (Nikolic, Warnier, et al., 2019; Vangheluwe, 2000) enables the combination of the strengths of different models whilst enabling the development of multi-perspective insights. In this research, I developed a multi-model to provide policymakers in The Netherlands with insights on how to steer the Urban Energy Transition at both national and municipal level. This is done by combining a System Dynamics (Forrester, 1961) model for evaluating market behaviour at the national scale, and an Agent-Based Model (Epstein & Axtell, 1996) for investigating investment behaviour at the neighbourhood level. The results from experiments with this model show that multi-perspective insights can be very helpful in designing policies and strategies for dealing with Sustainable Urban Development. It is found that challenges concerning the conceptualisation phase of the Model Development Cycle, could turn out to be opportunities for model development. As formalism selection and possible overlapping scales between models requires extra effort and attention, this attention results in model concepts that will be better evaluated, resulting in better concepts to be developed altogether. The main challenge that is currently unresolved, is that of the computational performance of multi-models. Although worsened by the choice for Agent-Based Modelling as one of the formalisms used in the multi-model, the computational requirements for the simulation seriously hamper the opportunities for experimentation and analysis. Although solutions can be found to minimise the computational demand or the effects of it, improvements are required in either computational demand or processing supply, for multi-models to become a suitable method for common modelling and simulation practice. CME: Energy Transition in the Built Environment With the EU goals (European Commission, 2018b) to reduce CO2 emissions to zero tons in 2050, The Netherlands has set up a climate agreement (Klimaatakkoord, 2019) in which municipalities get a leading role in reducing the carbon footprint of the Built Environment. To realise these goals, an energy transition is inevitable and strategies need to be developed for remodelling neighbourhoods into sustainable living environments. Before the end of 2021, municipalities have to come up with a plan in which they decide what neighbourhoods will be transformed and when. A problem in this is that currently, municipalities do not have the resources and tools to do so (RTL, 2019; Schuttenhelm, 2020; Straver, 2019). Currently, existing models, such as the Vesta MAIS model (Planbureau voor de Leefomgeving, 2019c) offered by the national government, are missing two crucial aspects which should help municipalities to set up their strategies: time dynamics and population dynamics. The techno-economic approach does present the ’best’ solution but is missing how to get there. Next to this, uncertainty around national policies makes that it is very difficult to decide on a robust policy. To deal with the aforementioned challenges, I use a multi-modelling approach (Nikolic, Warnier, et al., 2019; Vangheluwe, 2000), using System Dynamics (Forrester, 1961) and Agent-Based modelling (Epstein & Axtell, 1996) in which on the one hand, the time dynamics and population dynamics aspects are incorporated, and on the other hand multi-perspective insights on effects of national policies on national goals as well as insights on the heterogeneous effects at the municipal level are provided. An important finding is the crucial role of a CO2 tax on the realisation of national goals. Without this tax, homeowners and housing corporations will not be motivated sufficiently to significantly invest in insulation measures, making that fewer alternative heating systems will be implemented. Next to this, it is found that major differences can be distinguished between rural and urban areas. Heating grids will be a viable starting point for urban municipalities to start their transition due to the availability of residual heat. For rural municipalities, this will be more difficult due to a dependency on low-temperature systems. There, starting with the neighbourhoods that have the most potential to insulate is advised.

Our contemporary society consists of increasingly complex systems-of-systems in which technical and social subsystems influence each other. Multi-models reflect this complexity and are useful to analyse these complex socio-technical systems. They consist of connected simulation models which may each have their focus on a subsystem or modelling paradigm. However, these models contain different types of uncertainties influencing the fidelity of the results. Several methods are available to identify and analyse these uncertainties. It is yet unknown if and how existing uncertainty analysis methods can be applied to multi-model ecologies. In this thesis, we aim to provide for an answer to the following research question: “To what extent can we apply existing uncertainty analysis methods to multi-models?” To answer this question, it is important to first identify additional uncertainties in multi-model ecologies compared to single models. Next, we identify and apply methods to analyse these additional uncertainties. As proof of principle, a multi-model is used which focusses on the expansion and decarbonisation of the energy grid in the Port of Rotterdam. There are three dimensions of uncertainty in simulation models: location, level, and nature. The different locations include the conceptual model, the computer model, input data, the technical model implementation, and the processed output data. In multi-models, we found an additional location: the interface. This is where the exchange of parameters between the models takes place. Within the multi-model locations, we identified some aspects that increase uncertainty. Epistemic opacity and computational expense are properties of multi-models that limit the analyst's knowledge of the multi-model and the feasibility of extensive uncertainty analyses. The methods we identified for analysing these types of uncertainties are divided into sensitivity analysis and calibration. Sensitivity analysis quantifies the contribution of specific uncertainties to the overall uncertainty in the model outcome. Well-established methods are extra-trees feature scoring and Sobol. Calibration methods are based on the notion of equifinality. Using a specified likelihood function, they determine parameter values that lead to results with a high likelihood. Monte Carlo Markov Chain (MCMC) methods are often used for calibration purposes. The applicability of these methods depends on the multi-model configuration. To describe these configurations, two limiting archetypes were used: directed graphs, and undirected graphs with feedback mechanisms over the model components. We found that uncertainty analysis of direct graphs can be carried out on both the model components and the whole. For undirected graphs, the research showed only added value in performing an uncertainty analysis on the whole multi-model. Otherwise, the changing context and emerging path dependencies cannot be included. The application of extra-trees feature scoring, Sobol, and MCMC on the case study model showed that methods for uncertainty analysis are applicable on multi-models by including uncertainties on the interface. Furthermore, it is possible to reduce computational costs by factor fixing, distinguishing between deep and stochastic uncertainties, and assessing the convergence of sensitivity indices. Epistemic opacity can be dealt with by performing multiple replications and by including uncertainties related to the technical implementation of the multi-model. MCMC methods are suitable for scenario discovery, which provides insight into parameter values that lead to specific model results. For future research on this topic, it is recommended to apply uncertainty analysis on multi-models with different network structures and a higher number of model components. This research offers a single case study multi-model, being an undirected graph with two model components. The role of uncertainties on the interface, epistemic opacity, and computational expense should then be further investigated in these different configurations. In addition, other uncertainty analysis methods in the context of multi-modelling could be investigated. Especially the application of moment-independent sensitivity analyses could be interesting to investigate further, since multimodal outcomes can arise from the interaction of heterogeneous models.

This thesis investigates the effect of technological progress on the Dutch labour market. For this, a replication of the Frey and Osborne (2013) study on Dutch data is conducted. The application of the FO model has shown that approximately 12\% of the Dutch workers are at high risk of being replaced due to technological developments. Moreover, when simulating this risk, this study found that on average, 36\% of the Dutch workers can lose their jobs within 20 years due to technological progress.
However, this analysis did not account for the possibility of retraining of workers and the strategic behaviour of workers and firms stemming from their uncertainty due to technological progress. To take these factors into account, an agent-based model was developed. By experimenting with this model, the average unemployment rate after 21 years dropped to 3\%. However, due to policy and scenario testing, the underlying unemployment rates of this average unemployment rate of 3\% varies greatly.

Policymakers who work in the public health sector may rely on the help of quantitative models to support their choice of control strategy against a particular infectious disease. While policymakers have a large number of decision support models to choose from, hardly any of these tools are used to design an intervention strategy that can work well across multiple diseases. This is in part because of the need for large amounts of precise, detailed data and the presence of many unknown or confounding factors, which complicates attempts to make single-disease models, let alone multi-disease models. Is this highly data-intensive and detailed predictive approach necessary to make a good decision about how limited health resources should be allocated against multiple threats? Could the inclusion of multiple pathogens into a single decision support tool change the recommendation of an “optimal” intervention? In the following thesis, a novel multi-disease model is created. Rather than attempting to make a predictive tool to try and foresee a deeply uncertain future, this multi-disease model uses an exploratory approach to systematically evaluate the impacts of uncertain parameters. Many objective optimization techniques are used to find robust intervention strategies that work well for decision-makers who are interested in increasing the impact of their limited resources.