9:15  -  10:15

May 8, 2019

Grand Ballroom B

503: Artificial Intelligence and the Biggest Demand Response Program in Texas

Session Category :  Emerging Technologies 

This session explores using artificial intelligence/machine learning approaches to analyze the amount of demand reduction associated with 4 Coincident Peak events in ERCOT. There are three heuristic methods that have been used for the analysis of the 4CP events:
1. Actual 4CP events are determined by ERCOT after the fact; the first heuristic is a method used to identify ‘Near-CP’ days—days on which there is a significant level of response by customers trying to lower their 4CP charges, but which end up not being ERCOT peak days.
2. Identify which specific customers are responding with significant frequency and magnitude on 4CP and Near-CP days.
3. Identify which of the responders is responding on a specific day and the amount and type of that response.
There are about 13,000 large customers in competitive areas of ERCOT that are subject to 4CP charges, as well as all NOIEs (Non-opt In Entities) The decision as to whether to respond on a specific day is up to the individual customer/NOIE. There has been a high degree of on both the number responding and the volume of the response. Historically, ERCOT has seen up to 2,000 MWs of demand response reduction.
Audience members will learn how artificial intelligence approaches can be applied to assess and quantify customer self-deployed load reductions in response to high dollar incentives.