Comillas Pontifical University. Madrid (Spain)
September 22nd, 2025
Original summary:
Energy system transitions are increasingly shaped by uncertainty about future technological, climate, economic, and social developments. Under such deep uncertainty, traditional energy planning models—typically based on deterministic or probabilistic assumptions—can produce strategies that are fragile or misaligned with real-world complexity. This thesis develops and applies a set of methodological frameworks to improve the robustness, credibility, and decision relevance of long-term energy planning.
A central contribution of the thesis is the development of openMASTER, a modular, open-source, and structurally detailed national energy system optimization model. openMASTER incorporates advanced features such as energy service-based demand, endogenous behavioral dynamics, technology vintages and decommissioning, operational flexibility through technology hibernation and reactivation, and raw material constraints. The model enables transparent, extensible, and reproducible analyses of transition pathways while supporting the integration of advanced decision-support methodologies.
Building on this modeling platform, the thesis introduces three methodological approaches. The first is a hybrid optimization algorithm that combines robust optimization and minimax regret, allowing decision-makers to apply differentiated preferences across uncertainties that affect system feasibility (e.g., demand or resource availability) and those that affect performance (e.g., costs). This framework avoids both excessive conservatism regarding economic uncertainty and protection against vulnerability to infeasibility, resulting in more balanced and credible transition strategies.
The second contribution addresses a common oversimplification in energy models: the assumption of independence among uncertain parameters. A PCA-based method is applied to incorporate empirically-informed correlations between variables such as technology costs and fuel prices into robust planning frameworks. By preserving the internal structure of uncertainties, this approach significantly alters the design of robust strategies and highlights the importance of modeling structurally consistent futures.
The third methodological advance is a scenario-based decision-support framework to design robust policy packages across multiple objectives. It combines exploratory modeling, SHAP-based feature importance analysis, and multi-objective robustness metrics to systematically construct policy portfolios that perform satisfactorily even under adverse conditions. Applied to a national case study, this approach identifies coherent policy combinations—rather than isolated instruments—that support decarbonization, air quality, cost, and energy security under uncertainty.
Together, these contributions form a coherent and original toolbox for robust energy planning, grounded in both conceptual innovation and practical applicability. The thesis demonstrates how combining structural modeling advances with rigorous treatment of uncertainty enhances the robustness, interpretability, and credibility of decision support. The methodologies developed in this thesis offer significant potential for application to pressing real-world challenges, such as the decarbonization of hard-to-abate sectors, the design of robust strategies for sustainable mobility—including the planning of charging infrastructure, modal shifts, and behavioral transitions under uncertainty—, or the strategic management of dependencies on critical materials. By bridging methodological innovation and practical decision needs, this work contributes to the development of more resilient, transparent, and adaptive energy transition strategies capable of withstanding deep and evolving uncertainty.
English summary:
Energy system transitions are increasingly shaped by uncertainty about future technological, climate, economic, and social developments. Under such deep uncertainty, traditional energy planning models—typically based on deterministic or probabilistic assumptions—can produce strategies that are fragile or misaligned with real-world complexity. This thesis develops and applies a set of methodological frameworks to improve the robustness, credibility, and decision relevance of long-term energy planning.
A central contribution of the thesis is the development of openMASTER, a modular, open-source, and structurally detailed national energy system optimization model. openMASTER incorporates advanced features such as energy service-based demand, endogenous behavioral dynamics, technology vintages and decommissioning, operational flexibility through technology hibernation and reactivation, and raw material constraints. The model enables transparent, extensible, and reproducible analyses of transition pathways while supporting the integration of advanced decision-support methodologies.
Building on this modeling platform, the thesis introduces three methodological approaches. The first is a hybrid optimization algorithm that combines robust optimization and minimax regret, allowing decision-makers to apply differentiated preferences across uncertainties that affect system feasibility (e.g., demand or resource availability) and those that affect performance (e.g., costs). This framework avoids both excessive conservatism regarding economic uncertainty and protection against vulnerability to infeasibility, resulting in more balanced and credible transition strategies.
The second contribution addresses a common oversimplification in energy models: the assumption of independence among uncertain parameters. A PCA-based method is applied to incorporate empirically-informed correlations between variables such as technology costs and fuel prices into robust planning frameworks. By preserving the internal structure of uncertainties, this approach significantly alters the design of robust strategies and highlights the importance of modeling structurally consistent futures.
The third methodological advance is a scenario-based decision-support framework to design robust policy packages across multiple objectives. It combines exploratory modeling, SHAP-based feature importance analysis, and multi-objective robustness metrics to systematically construct policy portfolios that perform satisfactorily even under adverse conditions. Applied to a national case study, this approach identifies coherent policy combinations—rather than isolated instruments—that support decarbonization, air quality, cost, and energy security under uncertainty.
Together, these contributions form a coherent and original toolbox for robust energy planning, grounded in both conceptual innovation and practical applicability. The thesis demonstrates how combining structural modeling advances with rigorous treatment of uncertainty enhances the robustness, interpretability, and credibility of decision support. The methodologies developed in this thesis offer significant potential for application to pressing real-world challenges, such as the decarbonization of hard-to-abate sectors, the design of robust strategies for sustainable mobility—including the planning of charging infrastructure, modal shifts, and behavioral transitions under uncertainty—, or the strategic management of dependencies on critical materials. By bridging methodological innovation and practical decision needs, this work contributes to the development of more resilient, transparent, and adaptive energy transition strategies capable of withstanding deep and evolving uncertainty.
Spanish layman's summary:
Esta tesis desarrolla métodos innovadores y un modelo open-source (openMASTER) para planificar transiciones energéticas más robustas. Incorpora incertidumbre profunda, correlaciones realistas y carteras de políticas, ofreciendo herramientas prácticas para una transición resiliente y creíble.
English layman's summary:
This thesis develops innovative methods and an open-source model (openMASTER) to make energy transitions more robust. It integrates deep uncertainty, realistic correlations, and policy packages, providing practical tools for more resilient, transparent, and credible transition planning.
Descriptors: Power technology, Energy, Technological Sciences, Economic Sciences, Sectorial economics
Keywords: Energy planning, energy models, uncertainty, robustness
Citation:
A.F. Rodríguez Matas, "Robust Decision-Making for Long-Term Energy Transitions. Advancing methods to address deep uncertainty in energy system models", PhD. dissertation, Comillas Pontifical University, Madrid, Spain, 2025.
