Novelty search is a state-of-the-art approach focusing on behavioural novelty, rewarding diverging, as opposed to pursuing static objectives. This is relevant for exploratory modelling and analysis, which focuses on exploration of model through open exploration or directed search. Novelty search as an open exploration strategy is being tested against proven methods such as latin hypercube sampling. Using existing evolutionary algorithms and a developed novelty function, the experiments focus on comparisson, impact of the number of functional evaluations and the impact of the goals of the evolutionary algorithm. Finally it can be concluded that novelty search finds novelties in the lake problem, which makes it a relevant search strategy, but not suited for indiviual exploration. That means that it would still be advised to use latin hypercube sampling for earlier exploration.

Humanity faces the unprecedented global challenge of climate change. The sheer complexity and uncertainty of this problem renders mere intuitive reasoning insufficient. To aid global climate negotiations, Integrated Assessment Models (IAMs) are used to analyze the interplay between climate and the economy. More specifically, IAMs account for how greenhouse gas emissions affect climate change, how climate change affects economic production, and how economic production affects GHG emissions. We can use IAMs to project trends in emissions and gross domestic product, assess the costs and benefits of climate policies, and estimate the social carbon cost required to achieve stated emissions reduction targets. Although IAMs are central to informing decision-making to avoid catastrophic consequences, policy recommendations resulting from IAMs commonly prompt a very heterogeneous distribution of risks and benefits across the globe. During the recent 2021 United Nations Climate Change Conference (COP26), it became clear that equity is a central issue in the climate action debate. Emerging economies consider currently suggested abatement policies unjust in light of the historical CO2 generation of high-income countries and the strongly increasing need for energy in low-income countries. The term double inequality has been coined to describe the inverse relationship between the distributions of risks and responsibilities. In fact, the regions that are the least responsible for historical and mostly current CO2 emissions around the world, exhibit the highest degree of vulnerability to climate damages. In order to enable international cooperation and have a shot at meeting the Paris Agreement target, we require policies that promote more equitable mitigation pathways. Equity is therefore an eminently pressing topic, yet most IAM studies largely neglect it due to the implicit use of a utilitarian social welfare function that aggregates risks and benefits over space and time, thus losing sight of distributional consequences. In order to account for distributional justice, we transform the RICE model into a simulation model and embed it in a many-objective simulation-optimization setup such that we can find Pareto-optimal climate mitigation pathways for different problem formulations. Next to using four ethical premises (rooted in utilitarianism, sufficientarianism, egalitarianism, and prioritarianism), we also direct particular attention to the disaggregation of utility and disutility within each of these ethical premises. The reason for this disaggregation is based on the incommensurability of these two. Usually, IAMs maximize aggregate variables such as welfare. If we also consider the minimization of welfare loss, which is based on economic damages as one of the objectives, we can enable a potentially fairer distribution of not only consumption but economic damages. We argue that we can find climate justice behind the veil of aggregation. What we mean by this is that more equitable policy recommendations are obscured and lie hidden behind a bulwark of highly aggregated variables. If we look beyond this obstruction by the means of disaggregation, we are better equipped to find climate justice. In order to get to the bottom of this, we ask the following question: How are Pareto-optimal climate abatement pathways affected by the disaggregation of utility and disutility in alternative ethical problem formulations when using an integrated assessment model under deep uncertainty? To answer this question, we use a framework that is called multi-scenario multi-objective robust decision-making. For each of the eight problem formulations (4 ethical premises x 2 levels of aggregation), we use a multi-objective evolutionary algorithm to find Pareto-optimal policies. We reevaluate their performances under uncertainty by comparing their climate abatement pathways across the problem formulations. On a high-level, we can summarize our key findings as: - dominance of aggregation levels over ethical premises - correlation between low welfare and high welfare loss - general dominance of egalitarian aggregated Pareto-optimal policies - shared misery via egalitarian disaggregated Pareto-optimal policies The effect of disaggregating utility and disutility is stronger than originally expected. Using disaggregated problem formulations yields substantially different pathways even within the same ethical premise. These results are promising as we could transfer these insights to other more complex IAMs such as IMAGE and MESSAGE. Overall, this could be also good news for the equity debate. Using alternative ethical premises and disaggregating incommensurate objectives such as utility and disutility can offer alternative policy recommendations and resulting climate abatement pathways which could in turn enable more equity. What we likely need now, is a stronger dialogue between the modelers and policy analysts on the one side and the stakeholders and decision-makers on the other side. The latter ones should not just blindly trust in the magical outputs of a model but they need to be involved to decide what problem formulations we need to use as there is no correct way to frame a complex real-world problem. As unmitigated climate damages exhibit an independent impact on a region's well-being, we could render IAMs more useful for climate policy if we a) acknowledge that the classical notion of welfare is obsolete, b) use a multi-objective approach, and c) let the decision-makers decide how they want to trade-off the various objectives in post. In this manner, we could use IAMs to advance into the direction of enabling a transition of more climate justice.

Water, an essential resource for diverse purposes like environmental protection, urban water supply, energy generation, and agriculture, faces intensifying demand amid depleting supplies. Multi-Objective Optimization (MOO)-models are vital for addressing complex water system challenges with limited resources. However, varying approaches to distributive justice in these models introduce normative bias, leading to uncertainty in the derived implications from the model. This thesis is the first approach to understanding how the operationalization of distributive justice shapes the implications drawn from the 'optimal' outcomes of MOO-models. This thesis studied: 'How do different operationalization formulations for inequality in existing multi-objective optimization models shift the Pareto front?'. A rival framings approach acknowledges diversity in perspectives, for which it is suitable to contrast the operationalization formulation. The rival framing focused on the inequality metric and the aggregation method over time for this metric, both used for the formulation of inequality in the objective formulation. The case study revolves around the Conowingo Reservoir in the Lower Susquehanna River Basin. Utilizing an Evolutionary Multi-Object Direct Policy Search (EMODPS)-model, the research optimizes water allocation by incorporating an equality objective alongside baseline efficiency goals. Results underscore that the trade-off between equality and efficiency is highly dependent on the chosen operationalization for equality. Moreover, for higher levels of equality, the Pareto front will shift drastically in terms of strength and direction of trade-offs. By unravelling the complexities of justice's integration into MOO-models, advances are made in the comprehension of how distributive justice can inform decision-making. Through the elaboration of justice formulations, a future is reachable where justice is not only considered but reached.

Addressing the global challenge of water scarcity, particularly in the context of Sustainable Development Goal 6, underscores the critical need for effective water resource management. Central to this management is the control of dams in river basins, which poses complex decision-making challenges often labeled as wicked problems. Many-Objective Optimization, a powerful tool in modeling water systems, grapples with conflicting objectives, and the complexity heightens when integrating Distributive Justice principles. Distributive Justice, aiming to equitably distribute resources, is crucial in the context of water distribution for reservoir management. While current modeling approaches predominantly apply Distributive Justice principles A Posteriori, this study focuses on their A Priori incorporation, evaluating their impact on trade-offs and decision-making processes within the framework of Many-Objective Optimization.The investigation delves into Utilitarianism, Egalitarianism, and Prioritarianism, representing varied Distributive Justice perspectives. These frameworks are translated into mathematical models to provide insights into how societal values influence water resource allocation. The research centers on managing contentious shared water resources within the Eastern Nile River Basin. A comparative analysis between A Priori and A Posteriori integration of Distributive justice principles reveals distinct Pareto-optimal trade-offs. The findings underscore significant differences between a priori and a posteriori approaches, with Utilitarian integration enhancing overall system performance, egalitarian integration diversifying solutions, and prioritarian integration yielding sharper trade-offs. This study contributes valuable insights to the ethical considerations in many-objective optimization. It also offers guidance for policymakers and water resource managers by providing clear terminology to discuss and evaluate different approaches that are fitting within the policymaker's scope.

This research has illustrated the abilities of incorporating robust decision making within train maintenance. Both scheduled and unscheduled maintenance processes of the Nederlandse Spoorwegen (NS) have been examined, after which exploratory modeling and analysis enabled developing valuable insights for decision makers. Scenario analysis through simulation allowed for the quantification of the effect of deep uncertainty in relation to train maintenance. From the scenario analysis it is discovered what uncertainties are, according to the simulation model, of significant influence towards maintenance performance. The results presented allow decision makers to support their decisions when evaluating which type of policies to perform within the next decade. While the outcomes are specific for NS, the aim is to show that the approach held in this research is also applicable throughout other fields where decision making information quality should be improved.

The Akosombo dam in Ghana's Lower Volta River Basin provides essential economic benefits through hydropower generation, flood protection, and irrigation opportunities. However, since its construction was completed in 1965, the livelihood of the riverine communities drastically changed. The seasonal peak and low flows in the river, which provided environmental services for many downstream households, are now replaced with a stable river flow throughout the year to favour hydropower production. A study on dam reoperation is not only of interest to look into the possibilities of restoring healthy ecosystems but also to look at how current water demands in the basin can be sustained. Climate change and increased water demands for irrigation are expected to strain the water resources in the Lower Volta. Reoperation is also complicated by the absence of water treaties with neighbouring countries. This can result in lower flows from upstream riparian states due to their increasing water needs. In this study, we aim to understand the vulnerabilities of the Akosombo system while finding robust policies that can cope with challenging climate and demographic conditions. This study highlights the challenges in balancing irrigation, environmental, and hydropower goals while also identifying system vulnerabilities and opportunities for robust operations to meet the Lower Volta River water demands under future challenging conditions. This is done by reevaluating previously found optimal release policies under uncertainty. From this study, it can be concluded that the uncertainties included in the analysis will alter system functioning. Not only will floods occur with the onset of climate change, but energy production and irrigation potential will also become lower if upstream states start to use more water. The most critical abstractions are those of Togo and Côte d'Ivoire. A benefit of this is that these abstractions also provide some protection against floods. It is therefore wise to create cooperation projects with these countries to safeguard the current benefits of the Akosombo and Kpong dam and create new benefits for all parties.

This research explores the effects different cooperation levels between countries have on water resources from a shared river. The water resources that are taken into account are: hydro electricity generated per dam, water that flows to the environment per dam, and irrigation water per station. The Zambezi river basin is selected as study area due to its trans-boundary multi-stakeholder complex character. The cooperation levels that are explored are no cooperation, full cooperation, and section cooperation. No cooperation entails that every hydroelectric dam in the river only tries to maximize its own electricity. With full cooperation the dams releases are adjusted to benefit all the water resources in the whole basin. With section cooperation the dam releases are adjusted to maximize the water resources within the section of the specific dam. A section is a geographical area that is based on country borders. A simulation model is built in order to explore the effects of the different cooperation levels. The release policies that maximize the water resources for full and section cooperation are found with a multi-objective evolutionary algorithm (MOEA). From this research three findings are extracted. First, for the full cooperation level and the section cooperation level policies exist that dominate the no cooperation level policy for every examined water resource. Moreover, for these water resources found policies for either full or section cooperation produce similar results. Therefore, any of the explored cooperation forms could be more beneficial than no cooperation at all. Second, outcomes of this study indicate that owners of water resources located in the upstream of the river have less possible benefits from entering a cooperation agreement in which their dams have to adjust their release policy. These water resources can only improve by a small amount compared to water resources located in the downstream of the river. Third, policies do exist that are able to perform best for certain water resources, but also cause other water resources to decline to lower than their no cooperation value. Therefore, selecting fitting release policies is important since not every found policy leads to a better solution. Last, in this study best performing policies for electricity production and irrigation for both full and section cooperation are compared. It is discovered that adapting the best electricity policy for either one of the cooperation forms would be most beneficial for the involved stakeholders. This is due to the electricity policies having the highest relative resource gains. To conclude, cooperation in a river basin could result in an overall increase in water resources. However, reaching cooperation might prove to be difficult, because of the lower incentives for the stakeholders in the upstream of the river. Furthermore, extra care has to go towards selecting better performing policies, because policies exist that can cause a decline in water resources compared to when there is no cooperation in the river basin. Moreover, stakeholders are expected to select the policy that maximizes the electricity production, since these policies also perform well for other examined water resources.

This thesis evaluates the effects of including automated event data for interstate conflict prediction. Automated event data are web-scraped news stories converted into data and they may allow conflict models to increase their performance. Accurate models can then be used for early-warning purposes. To predict three separate problems, tree ensemble classifiers were used. The three outcomes to be predicted were the occurrence of interstate conflict, its onset, and its escalation. They were predicted globally at the dyad month level, meaning monthly for every country pair, using data from 1995 to 2014. The feature set consisted of eleven structural, slow-changing variables, and 268 event features, which were event counts on a dyad month according to event type. The analysis showed that event data did not increase performance. This held across all three prediction problems. Additionally, it was found that the models for occurrence and escalation and performed well and decently well, respectively, but that the models for conflict onset performed poorly. In conclusion, event data needs further testing in different constellations to be effective in interstate conflict prediction. It seems likely, however, that effective prediction for policy guidance is possible, given the model performance of the occurrence and escalation models.

Global climate change has results in a higher frequency of extreme disaster events and is therefore a serious challenge in disaster impact management. Disaster risk is composed of several components, such as vulnerability, susceptibility, exposure and the probability of occurence and intensity of a hazard. Vulnerability has become a topical issue due to the major role it plays in disaster risk reduction strategies. Hence, this research focuses on the development of a method to understand the dynamics of social vulnerability. The study area comprises Burkina Faso. Therefore the following research question was developed: 'how to calculate a social vulnerability index for Burkina Faso that characterizes the spatial and temporal dynamics of social vulnerability"?" The results showed that despite drawbacks, principal component analysis provides good insight in de internal and externaly dynamics of social vulnerability. However, large differences are found in the ranking of the social vulnerability score of communes when other methods are used. Hence, it is deemed important to develop more research in the semantic meaning of social vulnerability an thus understand better which mathicmatical approach is the most suitable. This research has found the highest social vulnerability in communes prone to conflict which are hosting many IDPs in the North, Centre-Nord, Sahel and East of the country. A statistically significant increase of social vulnerability was found from 2015 - 2017 in Boucle du Mouhoun, the Nord and the Centre-Nord.

Forest BORG, a novel framework using the BORG multi-objective evolutionary algorithm, optimizes policy trees for socio-economic systems. Tested on real-world case studies, Folsom lake and RICE IAM, it addresses multi-objective challenges under deep uncertainty, offering effective and explainable decision support for climate change policies. The 'Forest BORG' framework, based on the BORG multi-objective evolutionary algorithm (MOEA), is introduced as a method for optimizing policy trees in socio-economic systems. This framework accommodates tree-structured decision variables through specialized evolutionary operators, enabling multi-objective and explainable decision support. Tested on the Folsom lake model and the Regional Integrated model of Climate and the Economy (RICE) Integrated Assessment Model (IAM), Forest BORG addresses the challenges of optimizing multi-objective problems under deep uncertainty, primarily driven by climate change. The Folsom lake model and RICE IAM serve as suitable case studies due to their reflection of real-life problems affected by climate change and involving multiple stakeholders. The XLRM framework is employed to operationalize the conceptual models of these case studies, categorizing uncertainties (X), levers (L), relations (R), and metrics (M). Positioned between current MOEA capabilities and the need for explainable policies in real-world multi-objective problems under deep uncertainty, Forest BORG undergoes rigorous testing. Analyses include run-time dynamics, Forest BORG specific evolutionary operator dynamics, controllability, exposed trade-offs, and highlighted policy trees. The framework employs seven mutation operators, crossover with prune and bloat control operations as the recombination operator, and an adaptive tournament size function as the selection operator. The performance and behavior of Forest BORG are evaluated through the five analyses mentioned. Run-time dynamics demonstrate the algorithm's effective convergence based on hypervolume. Search operator dynamics investigate novel mutation operators, showcasing their contribution. Controllability analysis explores the impact of hyperparameter settings on algorithm performance. Exposed trade-offs are visualized through pareto-optimal policy alternatives, emphasizing the selection of compromise policies. The highlighted policy trees provide insight into actionable decisions based on different objectives. Forest BORG proves effective, efficient, and reliable across case studies. Comparative analysis with the original policy optimization tree (POT) algorithm reinforces its effectiveness. Although the expected shift from subtree to point mutations was not observed in all cases, the search operators enhance overall search capability. Hyperparameters are case study-dependent, with modest maximum depths favored to avoid overfitting. Forest BORG successfully produces pareto fronts for each case study, revealing inherent trade-offs. The selected policy trees align with performance metrics, providing valuable insights for decision-makers. However, some results exhibit counter-intuitive behavior, notably in the RICE model trade-offs and small peaks in run-time analysis, attributed to epsilon values. The research suggests further exploration of other IAMs for a more realistic representation of reality. Forest BORG's societal contribution lies in its development and application to real-world case studies, expanding the range of multi-objective decision-making tools for socio-environmental systems. The implications extend to addressing climate change through policy development, emphasizing the potential for broader applications beyond the tested case studies.

New global climate policies must be deemed just in order to be effective. However, global climate policymaking is complex and subject to normative uncertainties, especially in relation to the distribution of resources, risks, and consequences. These diverging views are dependent on what is distributed and are ultimately based on moral rules and principles that prescribe when a distribution is morally just; distributive justice principles. Enhanced understanding of these distributive preferences is necessary to account for them in both policymaking and modelling. A bottom-up evaluation of stakeholder views in negotiations is a time-consuming but useful method to add to this understanding. This research evaluates the potential of using GPT-4o to perform this task, identifying distributive justice principles in High-Level Segment (HLS) speeches from UNFCCC COP. By identifying distributive justice principles—egalitarianism, utilitarianism, prioritarianism, sufficientarianism, and libertarianism—this research examines the moral foundations of climate policy decisions. Manual annotations of 51 HLS speeches created a ground truth dataset, revealing complexities and class imbalances in principles, with prioritarianism being most dominant. GPT-4o’s performance in identifying relevant sentences and distributive justice principles showed promise but struggled with consistency compared to human annotations. Despite limitations, the model demonstrated efficiency, highlighting its potential for pre-processing and classification tasks. The study underscores the importance of a nuanced, bottom-up understanding of distributive justice in climate negotiations, contributing to climate justice and IAM by offering theoretical insights and practical implications.

New global climate policies must be deemed just in order to be effective. However, global climate policymaking is complex and subject to normative uncertainties, especially in relation to the distribution of resources, risks, and consequences. These diverging views are dependent on what is distributed and are ultimately based on moral rules and principles that prescribe when a distribution is morally just; distributive justice principles. Enhanced understanding of these distributive preferences is necessary to account for them in both policymaking and modelling. A bottom-up evaluation of stakeholder views in negotiations is a time-consuming but useful method to add to this understanding. This research evaluates the potential of using GPT-4o to perform this task, identifying distributive justice principles in High-Level Segment (HLS) speeches from UNFCCC COP. By identifying distributive justice principles—egalitarianism, utilitarianism, prioritarianism, sufficientarianism, and libertarianism—this research examines the moral foundations of climate policy decisions. Manual annotations of 51 HLS speeches created a ground truth dataset, revealing complexities and class imbalances in principles, with prioritarianism being most dominant. GPT-4o’s performance in identifying relevant sentences and distributive justice principles showed promise but struggled with consistency compared to human annotations. Despite limitations, the model demonstrated efficiency, highlighting its potential for pre-processing and classification tasks. The study underscores the importance of a nuanced, bottom-up understanding of distributive justice in climate negotiations, contributing to climate justice and IAM by offering theoretical insights and practical implications.

This research identifies a method to evaluate the fairness and stability of existing Pareto optimal policies in the context of transboundary water resource allocations. Using the case study of the Susquehanna river basin, the fairness of efficient policies was evaluated using utilitarianism, egalitarianism and prioritarianism. Stability was evaluated using stability concepts from game theory. The results showed that policies that balance utilitarianism, egalitarianism and prioritarainism can be identified. A status quo bias was seen in non-cooperative games highlighting the cost of getting a large number of actors to agree on a cooperative policy. Fallback bargaining identified an outcome that can potentially be stable in cooperative settings. This outcome also adheres to utilitarian ideas. The method proposed in the study explicitly considers efficiency, fairness and stability as criteria for water allocations and places the research in the context of distributive justice in transboundary water sharing.

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.

Water storage and diversion capacity of reservoirs in the Nile Basin are benefited for various purposes including agricultural irrigation and hydro-energy production. Filling and operation of the recently constructed Grand Ethiopian Renaissance Dam (GERD) have been subject to heated debate for its implications on downstream countries Sudan and Egypt. In this study, we investigated trade-offs between the objectives of the three countries by formulating the problem as many objective optimisation. To this end, we built a simulation model conforming to the evolutionary multi-objective direct policy search (EMODPS) framework. Based on the optimisation results, we discovered two scenarios under which trade-offs exhibit diverging patterns. Our findings suggest that 1) aggressive filling strategies create evident trade-offs between the hydro-energy generation objective of Ethiopia and consumptive uses of Sudan and Egypt, 2) growing demand in the long-term brings the dilemma of maintaining dams operational versus satisfying the demand and 3) increased demand pressure reinforces trade-offs between Egypt’s aggregate deficit minimisation and Ethiopia’s hydropower maximisation objectives. Our results highlight the potential of compromise policies in managing the objectives of all stakeholders without imposing heavy sacrifices. These policies represent an opportunity for cooperative operation of the dams through which multiple challenges facing the basin can be addressed.

Over recent decades there have been advances in the research behind adaptive policy approaches. More recently, emerging qualitative approaches for managing deep uncertainty has drawn the attention of planners in the water industry, particularly in water utilities. Despite interest from both the water industry and the research community to see these novel approaches applied, there have been limited applications and no published guidance to support the operationalisation of these approaches in water utilities. This thesis seeks to bridge this gap by answering the question: “What are the factors influencing the adoption of adaptive policy approaches by water utilities?” To answer this question, a design science approach was used to understand current barriers and enablers to the adoption of such methods in water utilities, and to design a framework to support adoption of adaptive approaches. This work was conducted through a grounded theory analysis of interviews of relevant water utility practitioners in Australia and the Netherlands and members of the decision making under deep uncertainty and adaptive planning research community internationally. The outcome of this study is a maturity framework of barriers and enablers to the uptake of adaptive approaches in water utilities, designed to support utilities and researchers in evaluating their level of adoption and to identify strategies for increasing the implementation of adaptive approaches where appropriate.