Tackling Variability of Renewable Energy with Stochastic Optimization of Energy System Storage - Solving a Stochastic, Multistage AC Optimal Power Flow problem with the Stochstic Quasi-Gradient Method
MetadataShow full item record
This thesis shows how the Stochastic Quasi-Gradient (SQG) method may be utilized to analyze and optimize the use of energy storage in a power system to facilitate the inclusion of more volatility in the power system and put to use increased variable production by simulating an solving a Stochastic, Multistage Alternating-Current Optimal Power Flow (S-MS-AC-OPF) model. The transformation away from fossil based fuels to more sustainable energy sources present many a challenge. One of these is that the renewable energy to be incorporated into the power system brings with it a high volatility that has not been present in the system before and that the grid and its physical components are not dimensioned to tackle. The technical solution to this is to include temporary energy storage in the energy system. Hence, for the success of the coming transition to a renewable energy system, there is a great need to develop methods that may analysis of how to facilitate the rising variability in the power grid using using energy storage. From a perspective of optimization methods and power system analysis, this type of analysis is not trivial. This is because the analysis of energy storage requires the sought method to be of a dynamic character. It also has to be able to take heed of the uncertainty that will be present, and thus should be a method that deals well with problems of stochastic nature as well. Moreover, for the analysis to be of relevance from a electrical engineering point of view, the problem to be solved has to be able to tackle non-linearity and non-convexity, and hence being able to deal with local minima. This is to fully employ the constraints imposed by security, stability and physical limits of the power system. Therefore, the problem we want to optimize becomes quite complex. Luckily, there exists a method that may deal with all of these issues, and find a solution efficiently, namely the SQG method. By implementing the SQG method with a S-MS-AF-OPF, and testing it with several approaches on several cases, we show how the SQG may serve the desired purpose. Especially the use of a gradient estimate directly from the AC-OPF solution provides a good solution within reasonable time. To conclude, the SQG method coupled with AC-OPF might be an effective tool for analyzing and optimizing energy systems with stochastic and dynamic aspects.