addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwchatcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-checkcircle-with-crosscircle-with-pluscrosseditemptyheartexportfacebookfolderfullheartglobegmailgoogleimageimagesinstagramlinklocation-pinmagnifying-glassmailminusmoremuplabelShape 3 + Rectangle 1outlookpersonplusprice-ribbonImported LayersImported LayersImported Layersshieldstartickettrashtriangle-downtriangle-uptwitteruseryahoo

Using R to forecast power demand of electric customers with distributed energy

Speaker: Jonathon Nelson

As an energy nerd and new R programmer, Jonathan Nelson has been honing his R skills through a project that mines raw net-metered customer billing data to predict demand for backup power from customers with distributed solar energy. Most electric utilities meter their solar customers on a net basis, recording just the difference between consumption and production from distributed generation. This way, customers only pay for their grid usage that’s in excess of their production at the end of the billing period (usually one month). While this has been an effective means to incentivize retail solar customers, it doesn’t allow the utility to see the actual energy demand behind the meter when the solar is producing - they just see the difference. It’s not until a variability event such as shading due to intermittent cloud cover that the utility can see the behind the meter demand as the generation drops off and the customer jumps back onto the grid for backup power. These variability events produce instantaneous spikes in demand for grid power and at a large scale are difficult for a utility to forecast and manage. Jonathan’s model can selectively mine out these variability events to uncover the demand that is likely happening behind the meter at any given moment. Then with this analysis the utility could better forecast the need for backup reserves and more effectively integrate larger penetrations of distributed energy onto the grid.

The model is best characterized as time series data analysis. Packages used so far in the model include: doBy, fields, chron, and lattice. The model deals with manipulation of datetime objects, subsetting, creating dynamic time sequences with settable intervals, creating sampling distributions to compute summary statistics and interval estimates, etc.

As usual, the actual talk will start at 7:30. Come by early to socialize.

Join or login to comment.

  • Galen

    Excellent presentation, and a very interesting topic.

    April 28, 2014

48 went

Our Sponsors

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy