Energy storage system (ESS) is a key technology to accommodate the uncertainties of renewables. However, ESS at an improper size would result in no-reasonable installation, operation and maintenance costs. With concerns on these costs outweighing ESS operating profit, this paper establishes a stochastic model to size ESS for power grid planning with intermittent wind generation. In the model, the hourly-based marginal distributions with. Energy storage system (ESS) is a key technology to accommodate the uncertainties of renewables. However, ESS at an improper size would result in no-reasonable installation, operation and maintenance costs. With concerns on these costs outweighing ESS operating profit, this paper establishes a stochastic model to size ESS for power grid planning with intermittent wind generation. In the model, the hourly-based marginal distributions with covariance is first derived from historical data of wind generation, and a stochastic cost-benefit analysis model with consideration of the generation fuel cost expectation and ESS amortized daily capital cost is formed. Then a hybrid solution approach combining the Point Estimated method and the parallel Branch and Bound algorithm (PE-BB) is designed to solve the model. Finally, the stochastic model and PE-BB approach are thoroughly tested on the 10-unit and 26-unit systems with uncertain wind generation. Simulation results confirmed the proposed model and PE-BB approach are effective to optimize ESS size for power grid planning with intermittent wind generation. The cost-benefit investigations on four typical ESSs also indicated that the ESS capital cost, charging/discharging efficiency and lifetime are important properties for optimizing ESS size, and it is not always economically justifiable to install ESS in power system.••••A stochastic model is presented to optimize ESS size in power system planning.••The model simultaneously considers expected generation fuel cost and ESS capital cost.••The model is general and flexible to various probabilistic wind generation.••A parallel Branch and Bound algorithm with Point Estimated strategy is proposed.••The. Energy storage system sizingIntermittent wind generationStochastic cost-benefit analysisUnit commitmentEnergy storage system (ESS) is the key technology for reliable and flexible energy integration and has been investigated for various applications in power systems [,, ]. With the premise of instantaneously balancing power generation and consumptions, ESS is often operated to store surplus energy in off-peak hours and release it during energy-deficiency hours such that temporal arbitrages can be obtained via economical scheduling of stored energy.While extensive researches have been conducted on ESS, they could be roughly classified into two categories: 1) the operation analysis of ESS with assumed power and capacity rating, and 2) the determination of ESS optimal sizing.For the operation analysis with pre-set ESS parameters, ESS is usually coordinated with conventional generators and renewables to pursuit the maximum benefits by energy time shifting. For example, this energy shifting concept was adopted in Refs. [4,5] to effectively dispatch energy resources for an integrated thermal-photovoltaic-battery generation system, while a demand response scheme with hybrid electric vehicles was designed and one Nash equilibrium point was obtained to minimize customers' charging cost based on the energy shifting in Ref. In Ref., a security constrained unit commitme. This section proposes the UC-based stochastic cost-benefit model to determine the ESS optimal size for power system planning in presence of correlated uncertain wind generation. Different from the rolling approach with day-by-day deterministic optimizations over multiple years, in this paper the stochastic features of wind generation is first.
