Potential of Residential Storage Battery Demand Response in Tertiary Balancing Market

Abstract

Non-memberThe recent increase in residential photovoltaic (PV) power generation has become a major factor in the electricity supply instability. As a result, power balancing at the distribution level is becoming increasingly important. In Japan, storage batteries installed in residential homes along with PV systems has reached nearly 3 GW and are expected to contribute to the power system stability. In this paper, we propose a method for a group of residential consumers to participate in the balancing market by managing their total meter value to match the target power. This method consists of the following two stages: bid volume optimization, which optimizes the bid volume based on the past behavior of the consumers, and demand response (DR) dispatch optimization, which quickly determines the dispatch of charging and discharging requests every 5 min when DR is requested. Evaluation simulations based on the actual 1-min power data and consumer response model showed that, by aggregating 1600 consumers, the success criteria of the balancing market can be met. In other words, the total metered power in 1-min and 5-min intervals can be controlled within ± 10% of the bid volume against the target power. From these experiments we have shown that the aggregation of as large as 1600 residential consumers has the potential to participate in the balancing market.

Absorption of PV Power Prediction Errors with Headroom Control by Statistical, Machine Learning and Combined Models

“…[10]. In addition, the previous research on PV systems integrated BESS bidding to the balancing markets can be found, e.g., BESS demand responds in tertiary balancing market based on the past behavior of the consumers, and demand response dispatch optimization [11], and the optimal day‐ahead and intraday market bids via a model predictive control approach considering solar generation prediction [12]. However, only a few studies are related to the absorption of prediction errors by providing headroom in PV without BESS in every time step.…”