Multi-objective battery energy storage optimization for virtual power plant applications

The increasing share of renewable energy sources (RESs) in electricity generation leads to increased uncertainty of generation, frequency and voltage regulation as well as difficulties in energy management. A virtual power plant (VPP), as a combination of dispersed generator units, controllable load and energy storage system (ESS), provides an efficient solution for energy management and scheduling, so as to reduce the cost and network impact caused by the load spikes. This paper proposes a multi-objective optimization (MOO) of battery energy storage system (BESS) for VPP applications. A low-voltage (LV) network in Alice Springs (Northern Territory, Australia) is considered as the test network for this study. The BESS for each customer is used to store and release the energy when required to maintain the voltage regulation performance of the LV network and reduce the cost. To optimize the charge/discharge schedule in each battery, a multi-objective optimization tool (MOOT) is developed, where MOO can directly communicate with DIgSILENT PowerFactory platform to perform multi-time scale simulation. MOO mainly focuses on generating a set of trade-off schedule solutions for each customer over 24 h by considering the customer's cost and network impact. Then according to the stakeholder's prior preference, one of the solutions is selected and verified in DIgSILENT PowerFactory with a realistic LV network. We also design several scenarios with different penetrations of photovoltaics (PVs) and batteries to verify how they influence customer's cost and LV network. With the increasing penetrations of PVs and batteries, the experimental results over the selected solution according to the customer preference show that the customers' cost can be largely reduced by maintaining the network voltage regulation performance.