Tailoring Demand to Match Supply: How Much Flexibility There is in Loads?

Increase in supply side variability due to increases in renewable generation require demand side management strategies to reduce electricity delivery costs. Smart grid technologies provide opportunities for measuring and controlling loads at an unprecedented scale. Yet, understanding their performance requires accurately capturing how loads respond to specific technologies and how consumer behavior affects such response.  Typical demand side management planning and capability studies have been conducted relying on theoretical scenarios of adoption and response. This talk presents a new approach utilizing large scale individual level data made available by AMI deployments (smart meters). We utilize a 500,000+ customer data set and measure the available flexibility as well as potential strategies to elicit this flexibility from consumers.We classify various types of flexibility and demonstrate methods to infer them relying on features from data that reflect both behavior and response to specific technologies.  We demonstrate that flexibility depends both on environmental and geographical considerations as well as appropriately classifying and selecting consumers.The proposed approaches illustrates the value of combining tools from statistical signal processing,  stochastic control, machine learning and economics.

Speaker: Ram Rajagopal, Stanford