(679h) A Data-Driven Optimization Framework for Selection and Operation of Energy Storage Systems

Energy storage systems (ESS) have been widely used to provide benefit to the centralized power system through time shifting to improve the reliability and quality of power supply in the past (IEC, 2011). Along with people’s rising awareness of fossil fuel depletion, climate change and pollution, there is a booming trend towards more research, development and utilization of renewable energy as well as the distributed energy systems based on renewable sources. However, due to their intermittent and fluctuating nature, the availability and amount of electricity generated from these sources are difficult to be matched with the demand of the electricity consumers. Energy storage systems therefore become essential to achieve demand-supply matching, reliability and quality control for the more complicated modern energy systems.

This work will answer two fundamental questions related to ESS: 1) how to choose a suitable energy storage technology for a specific application facing the large variety of technical features that could be considered as well as the diverse and abundant technologies that can be chosen; 2) how to effectively operate energy storage systems to maximize the payoff utility from decentralized demand side management (DSM), meanwhile benefiting the electrical grid.

First, a novel data-driven modeling and decision-making framework for the selection of energy storage systems is proposed to facilitate the decision-making process (Li et al., 2018). The major criteria evaluated include the technical suitability, the economic feasibility, and the environmental impact. A machine learning method, framed as a supervised classification problem, is used to predict the technical suitability of the technology for the application based on the rated power and discharge duration features of the technologies. The up-to-date DOE Global Energy Storage Database is used as the training dataset for the modelling process of this data-driven approach. Furthermore, the economic feasibility of ESS is evaluated using the levelized cost of electricity (LCOE) and cost benefit analysis (CBA), while the environmental impact through life cycle assessment (LCA) method. With the established framework, a multi-objective optimization is employed to maximize technical suitability, maximize profits, and minimize environmental impact by selecting the optimal energy storage technology for a specific application.

Second, a game theoretic approach for DSM model incorporating energy storage components is developed. The proposed model is able to not only reduce the Peak-to-Average ratio to benefit the electrical grid, but also smoothen the dips in load profile caused by supply constraints. In a case study a variety of residential demand types are evaluated within the micro-grid to maximize the payoff of both individual users and the whole system. Moreover, the emerging blockchain technologies are introduced to guarantee the seamless and secure implementation of the proposed scheme, illustrating how a decentralized DSM approach could be implemented in practice.

Through the case study, a variety of energy storage technologies and applications including both traditional energy systems and emerging distributed energy systems are examined using the proposed data-driven decision-making framework. It shows a good prediction and can get easily updated with new input data. With such satisfactory ESS, the demonstrative micro-grid is able to maintain the supply and demand balance, reduce utility bills of consumers and reduce stress on grid in daily operation.

References

IEC (2011) Electrical Energy Storage - White Paper, International Electrotechnical Commission. doi: 10.1002/bse.3280020501.

Li, L. et al. (2018) ‘A Multi-Objective Optimization Approach for Selection of Energy Storage Systems’, Computers & Chemical Engineering. doi: 10.1016/j.compchemeng.2018.04.014.