Energy markets are at the heart of one of the biggest societal challenges of our time - creating a sustainable, reliable and affordable energy provision. The Center for Energy Markets at Technische Universität München brings together economics, finance and engineering approaches to offer applied research contributions to topical real-world energy sector questions.
One research project at the Center for Energy Markets is to understand the economics and future development of Vehicle-to-Grid (V2G) business models. The V2G idea explores ways of capitalizing on the unused potential of parked EVs by aggregating EV batteries for use by the larger power grid. The V2G model has the potential to address the power-generation issues of intermittent renewable energy sources through demand-side management and the provision of ancillary services. Central to this model is the aggregator, who combines a large number of EVs and is granted access to the energy storage capacity of the batteries of parked EVs.
In our research we simulate if the provision of ancillary services with an EV pool is economically feasible. For this purpose, we develop a bottom-up model that encompasses the entire sequence of decisions that an aggregator faces. These include pool composition, bidding activities on electricity markets, and the real-time high-frequency dispatch of single vehicles for the provision of ancillary services. In the simulation, a combination of very high temporal resolution, large vehicle pools, and dispatch optimization, which represents a non-linear and non-smooth optimization problem, result in high complexity and very time consuming calculations. All simulations are implemented in Matlab.
It is essential for our research project to speed up the simulations by using the parallel computing functionalities provided by Matlab. For this purpose, we combine the cloud services of Microsoft Azure with Matlab by using different instances of Microsoft Azure Virtual Machines. The usage of Microsoft Azure helps us to reduce the simulation time by factor 50 and enables us to include additional scenarios and sensitivity analyses in our research.