Skip to content

Models ~ demandResponse

Jump to bottom
David Pinney edited this page Oct 14, 2015 · 14 revisions

Introduction

This model takes in historical demand data (hourly for a year) and calculates what demand changes in residential customers could be expected due to demand response programs. Program types the model can calculate are time of use pricing (TOU), peak time rebates (PTR), direct load control (DLC) or critical peak pricing (CPP). These calculations are done using the Brattle Group's PRISM model.

Walkthrough

  1. Fill in the inputs that are blank. The rest are reasonable defaults. You can change them if you want.
  2. The model needs one year of demand data at the level of the utility (or substation, or feeder). This can be uploaded as a CSV file. The format must be:
timestamp,power
01/01/2013 00:00:00,1981
01/01/2013 01:00:00,1903
... [8760 rows total] ...
12/31/2013 23:00:00,1921

If you would like to just try out the model, an example demand file is available here.

FAQs

  • How are commercial and industrial loads handled? The PRISM model does not support industrial and commercial loads. In general, each C&I load is unique, so modeling them is prone to error unless there is lots of data on their behavior. If you input hourly demand data that includes C&I loads, you can exclude them (approximately) from the analysis by reducing the "Load Managed by the Program" input by the amount that those loads contribute to the data set. The best approach, if detailed metering data is available, is to only input demand data from residential loads.

  • What are good values for the price elasticities? Many studies have been done that estimate elasticities for given circuits/climates/locations. We have collected [some example elasticities here](./images/DR Elasticity Estimates.xlsx).

  • What are the details behind this implementation of PRISM? PNNL contributed an Excel implementation and ported that to Python in support of this effort. The [source spreadsheet](./images/PRISM Model Implementation by PNNL - Basis for prismDR.xls) is available at that link.

Clone this wiki locally