This thesis investigates the optimal sizing and siting of energy storage systems (ESS) within a distribution grid, focusing on the provision of ancillary services and the intelligent power control of flexible loads, including electric vehicles (EVs), heat pumps (HPs), and renewable energy sources (RES). The study aims to investigate the use of ESS in combination with EVs to participate in the automatic frequency restoration reserves (aFRR) market. The effect of different seasons and the sensitivity of electricity market prices to the optimal sizing are analysed.

The first level involves day-ahead scheduling of flexible loads (EVs, HPs) and ESS allocation. This stage utilises mixed-integer programming (MIP) to achieve optimal ESS sizing and siting based on day-ahead electricity prices, predicted data on EV arrivals and departures, and building occupancy. Additionally, the aFRR prices are known a day ahead, and the most lucrative time intervals for revenue maximisation are selected. The objective function minimises the total cost, which consists of penalties related to customer satisfaction, grid import/export cost, revenues from aFRR provision, and BESS CAPEX and O\&M cost. Two different MIP formulations were used to solve the optimisation in Gurobi using Python programming language.

The second level involves real-time control, where ESS operation is re-optimized using real-time data. This stage adjusts for forecast errors through rolling horizon optimisation. Within this level, the scheduled aFRR reserves are deployed.

The findings reveal that the optimal integration of ESS, considering additional revenue from aFRR provision, is achieved by using a centralised ESS closer to the substation. In contrast, optimal ESS integration for energy arbitrage alone involves sharing capacity between nodes but requires a lower CAPEX cost of BESS to be economical. The results indicate that optimal BESS allocation can differ based on specific conditions, such as imbalance price distribution, grid limits, peak load and EV flexibility. To maximise revenue and efficiency, the study suggests placing the BESS at nodes with the lowest resistance. Winter and summer results are similar in the optimal placement and sizing of BESS, although the operation of BESS and EVs is different. It was observed that during the winter, V2G utilisation was higher. The main difference for the seasons is in the case of only energy arbitrage. The cost savings from ancillary service provision of either BESS, EVs or both show greater potential for summer. Both of the proposed optimisations can find a solution, although the non-convex MIQCP formulation did not guarantee a global optimum for all cases. However, the iterative method proved to be more robust but had a longer simulation time.