PhD defence Jesus Lago Garcia

28 september 2020 15:00 t/m 17:00 - Locatie: limited audience - Door: DCSC

"Incentivizing renewables and reducing grid imbalances through market interaction: A forecasting and control approach"

As the penetration of renewable energy sources (RESs) increases, so does the dependence of electricity production on weather and, in turn, the uncertainty in electricity generation, the volatility in electricity prices, and the imbalances between production and consumption. In this context, while RES integration does complicate grid balance and increase price volatility, it also opens up opportunities for flexible market agents to reduce grid imbalances. In particular, by using the nature of the interactions between electricity markets and grid balance,  market agents can reduce grid imbalances while increasing their profit. However, despite this obvious win-win situation, traditional research in this field has focused on balancing mechanisms that do not always exploit these relations between  electricity markets and grid balance.

The aim of this thesis is to fill this scientific gap by exploiting the intrinsic relation between electricity markets and grid balancing. Particularly, the goal is to propose new modeling, forecasting, and control algorithms that increase the integration of RES and decrease the grid imbalances by using market interactions. The advantage of the proposed methods is that they allow more energy systems to participate in and contribute to grid balancing. The thesis comprises three parts: a) forecasting algorithms to reduce uncertainty; b) modeling and control of thermal seasonal storage to mitigate imbalances; c) new market mechanisms to ensure a wider participation in grid balancing.

As the uncertainty of RESs hinders their economic profits and makes the grid harder to balance, a first approach to exploit market interactions is to accurately predict future electricity prices and generation of renewables. Particularly, accurate and reliable forecasts lead to better decision making, higher economic profits, and lower uncertainty. This in turn translates into a grid that is easier to balance and larger economic incentives for  integration of RESs. With this in mind, the first part of the thesis advances the field of forecasting methods by contributing to three research areas. First, motivated by the new EU market policies that aim at reaching a single and unified electricity market in Europe, we analyze the effect of market integration in electricity price dynamics and propose new forecasting models that exploit market integration to improve forecasting accuracy. Second, due to the advances of deep learning (DL) methods in several fields, we investigate the application of DL methods for electricity price forecasting and develop new DL forecasting techniques that achieve state-of-the-art results. Third, as forecasting short-term solar irradiance has become key for many applications,  we propose a generalized short-term forecasting model that can  forecast solar irradiance in any location without the need of ground measurements. The new method is paramount as solar generators are geographically dispersed and ground measurements are not always easy to obtain.

Improving the accuracy of forecasting techniques is an indirect approach to reduce the uncertainty in electricity trade and incentivize the integration of renewables. A more direct approach is to use energy storage systems to absorb the grid imbalances.  In this context, while long-term energy storage is arguably one of the most important elements to ensure the success of the energy transition, most of the existing technologies are only economically efficient for short-term and medium-term energy storage. Therefore, in the second part of this thesis, we investigate modeling and control techniques to ensure that seasonal storage systems maximize their profits while operating to reduce grid imbalances. First, as the existing models for thermal seasonal storage systems are too complex and cannot be efficiently integrated in control and optimization problems, we propose a new accurate model that can be integrated in real-time control and optimization applications. Second, we propose control algorithms for seasonal storage systems that, by explicitly exploiting the relation between imbalances and prices, reduce grid imbalances while maximizing profits. These algorithms are novel on their own as the control algorithms that currently exist for market interaction are limited to short-term horizons and are not suited for seasonal storage systems.

A more direct approach to incentivize the integration of renewables and keep the grid balanced is to explicitly modify the structure of electricity markets so that a larger number of energy systems have economic incentives to reduce grid imbalances.  In particular,
as traditional power plants are taken off the grid, it becomes clear that RES systems need to contribute to grid balancing if the grid is to remain stable. However, while some RES systems can potentially contribute to grid balancing, they are not being used for this purpose due to the current rules applied to system balancing. Examples of such systems include solar photovoltaic installations, storage systems such as seasonal storage, or even---in some countries---wind farms. With that motivation, in the third part of the thesis we investigate methods and new market structures that allow these systems to not only participate in balancing the grid, but to also have economic incentives to do so. In detail, we propose a new market framework for providing balancing services by trading with the imbalance settlement mechanism. The new framework incorporates newer systems into the portfolio of balancing providers and gives these systems economic incentives to balance the grid. As an additional advantage, it also incentivizes the use of long-term storage systems, which, as argued before, are key players in the energy transition. 

Promotor: B. De Schutter