Automatic generation of optimization models of systems of flexible energy resources

The energy transition towards a fully decarbonized energy system is one of society’s greatest challenges. An increasing share of variable renewable energy generation requires flexible generation and flexible consumption to counteract the associated volatility. Furthermore, providing energy flexibility holds significant financial saving potential for operators of flexible energy resources. However, the implementation of energy flexibility measures is hindered by limited personnel capacities and a lack of expertise required for developing suitable optimization models. Thus, this thesis presents a methodology for the automatic generation of optimization models for the operational planning of systems of flexible energy resources. To establish a solid methodological foundation for these models, a systematic review of existing modeling approaches for flexible energy resources is conducted. Based on this review, a suitable modeling approach has been elaborated and further refined through an analysis of the required level of detail for optimization models used in the control of flexible energy resources. The results indicate that mixed-integer linear programming optimization models with a low level of detail and temporal resolutions aligned with market intervals are sufficient to represent resource behavior and generate optimized schedules for resource operation. The methodology for automatic model generation builds on a modular generic model structure that represents the flexibility characteristics of resources. Models of individual resources are aggregated to jointly optimize the operation of a system of resources. The generic model is complemented by algorithms for deriving parameter sets from time series data. Additionally, connections between resources are extracted from an information model representing the system structure. Integrating the model structure and parameter derivation into an automated model generation methodology reduces manual effort and streamlines the creation of optimization model instances. Each component of the methodology has been independently evaluated to ensure its accuracy and applicability. Furthermore, the methodology in its entirety was applied in two case studies focusing on a combined heat and power system and a refrigeration system. These case studies demonstrate the applicability of the methodology by showcasing its ability to generate accurate optimization models without manual intervention. The automatically generated optimization models can then be used to plan resource operation within systems. For instance, during periods of high renewable energy generation, resources can be scheduled to maximize energy utilization. By enabling flexible resource operation, this approach contributes to balancing energy supply and demand, thereby supporting a successful energy transition.