Decision-making Under Uncertainty for the Operation of Integrated Energy Systems

This thesis deals with the development and application of solution approaches in the form of optimization problems for energy operators and companies that operate under uncertain conditions in an integrated energy system setting. The integration of renewable and partly unpredictable energy sources has increased the need for flexibility in the power systems.

One of the possibilities to provide this flexibility is by integrating different energy systems such as heat and power. The motivation of this thesis is to provide solutions that facilitate the operation of integrated energy systems under uncertain conditions. Nowadays, heat and power systems are coupled by the participation of district heating companies in electricity markets, which is subject to many uncertainties such as volatility in electricity prices and renewable heat and power production.

In this thesis, we propose decision support solutions for district heating producers to optimize their production and create bids for the day-ahead and balancing electricity markets. The proposed methods protect the operator against the different potential realizations of uncertainty. Another solution approach proposed in this thesis is motivated by the replacement of conventional fuel sources with more sustainable fuels such as biomass.

The delivery of this type of sustainable fuels is often settled in long-term contracts before the actual demand is known. In this thesis, we explore new ways of reducing the impact of uncertainty in supply planning and evaluate new contract designs for large combined heat and power producers using decision-making under uncertainty.

Concerning the operation of the power system under uncertain power production, we use the unit commitment problem to explore new ways of handling uncertainty and exploiting the operational modes of large combined heat and power plants to integrate higher shares of wind power production. The proposed methods deal with a large amount of uncertain data that may result in computationally hard optimization problems.

Therefore, we develop new solution approaches that are capable of handling large-scale optimization problems with a significant amount of uncertain data providing suitable solutions for the decision-maker while drastically reducing the solution time of the problem. All the solution approaches presented in this thesis are used for extensive analyses of the realistic systems used as case studies.

These analyses evaluate e.g., how uncertainty affects the obtained solution in terms of operating costs and how the studied systems can react to the uncertainty.