Investigating price arbitrage potential based on risk sensitive energy storage policy in turkish electricity

Increasing prices and price deviations in electricity have created economic opportunities in energy storage. The energy storage system can take advantage of electricity prices by buying and storing electricity when the electricity price is low, and by selling it back to the grid when the electricity price is higher. There is not energy storage technology (EDT) in operation for arbitrage in Turkey while academic studies are done in many countries about calculating arbitrage value, but the studies are progressing in this direction. In this study, we investigate price arbitrage potential of 13 different energy storage technologies in Turkish electricity market. The goal is to plan optimal management of storage operations by maximizing the expected cumulative revenue over the time period. An effective energy storage policy must take into account uncertainties in electricity prices as well as uncertainties in demand and supply. Therefore we apply Autoregressive Integrated Moving Average (ARIMA) to forecast electricity prices in Day-Ahead Market and Artificial Neural Networks (YSA) to forecast electricity prices in Balancing Power Market. We then use Quantile Regression to obtain interval estimation and Monte Carlo Simulation to build numerous scenarios for every hour. We propose a two-stage stochastic model to obtain hourly commerce bids that maximize the arbitrage revenue. The number-reduced scenarios using the k-means algorithm and the characteristic data of EDTs in various countries are given as input to this model. Afterwards the risk-neutral market participant has been made more risk-sensitive. This is because the market participant's risk attitude may cause the actual operational cost to differ significantly from the expected cost. This difference indicates that we must consider the risk. For a risksensitive participant, Conditional Value at Risk (CVaR) measurement is integrated into the mathematical model. The results show that there is an arbitrage potential and the level of the risk-aversion significantly affects the expected revenue.