Sizing of Energy Storage for Spinning Reserve and Efficiency Increase in Isolated Power Systems within a Data Driven Stochastic Unit Commitment Framework

This paper proposes a data-driven stochastic unit commitment (SUC) framework for sizing battery energy storage system (BESS) for spinning reserve and efficiency increase in isolated power system. BESS is used to provide spinning reserve to manage the uncertainties in the power system. Additionally, load-shifting demand response (DR) is incorporated in the data-driven SUC model to reduce the size of the BESS required. Also, battery degradation is modelled in the data-driven SUC formulation to make the results as accurate as possible. Furthermore, to reduce the computational burden a novel scenario reduction method based on the Kantorovich distance metric is proposed for selecting scenarios (from the high number of generated scenarios for load and renewable sources) that describe the problem with limited loss of information. The case study is based on an offshore oil and gas platform in the North Sea which has four identical gas turbine generators and is assumed to integrate offshore wind turbines and an offshore solar farm, thus having three sources of uncertainty (load, wind and solar). Five test cases were analyzed: a baseline case without BESS and flexible load (DR), and other cases including degrading BESS and flexible load, either separately or jointly. First, the selected scenarios were analyzed and seen to retain most of the information in the original historical dataset, which validates the accuracy of the proposed scenario reduction method. The results showed that the methodology proposed in this work can be used to get an accurate optimal BESS capacity and that the inclusion of load-shifting DR can help reduce the optimal BESS capacity. Cases where an optimally sized BESS was considered provided cost reductions in the range 2.54 % to 3.68 % and carbon emissions reductions in the range 3.88 % to 4.5 % compared to the baseline case. Finally, the result also showed that battery degradation modelling is important as the cost reduction increased from 2.54 % to 3.68 % when battery degradation is modelled.