An uncertainty-aware optimization approach for flexible loads of smart grid prosumers: A use case on the Cardiff energy grid

This work lies within the context of the smart grid technology and the technological and economic advantages it may provide to users and distributor operators. We focus on the possibility for users to shift their loads through the day and to produce their own electric energy, which allows them saving on their electricity bills and becoming active part in the energy market selling energy and flexibility. In this paper three optimization algorithms are presented and structured in order to minimize costs, discomfort and maximize profits for users for three different grid levels. They use the intrinsic advantages that the energy production/flexibility gives, in a context with pre-established tariffs and leveraging electricity production (e.g. via photovoltaic systems). The interactions between users and providers in a real grid are also shown in a fashion similar to the peak shaving problem. For testing these algorithms a data set obtained from an existing grid in Cardiff, UK and used within the MAS 2TERING research project has been employed. The data set includes data regarding consumption, electricity production and flexible loads of 184 users. Results have been presented, showing that (i) exploiting flexibility of loads and buying/selling energy surpluses may offer users the possibility to have significant cost and discomfort reductions (up to 29% of the total cost), (ii) leveraging flexibility for peak shaving can reduce daily peak loads by over 11%, and (iii) using parallel computation for the algorithms can significantly reduce computational time in real scenarios.