Download this white paper to learn how to transform DSM and energy efficiency business process management through a new system of record approach.
Planning for Efficiency
Forecasting the geographic distribution of demand reductions. Copyright © 2011 Consolidated Edison Company of New York, Inc.
in 2008 significantly complicated the forecasting process. Suddenly, the geographic distribution of efficiency-driven peak load reductions was tied to the relative penetration of new efficiency programs across a very heterogeneous market. Further, a variety of energy efficiency programs targeting differing market sectors with multiple administrators were authorized under the EEPS proceeding. That is, Con Edison had single and multi-family programs, as well as small and large commercial programs, but so did NYSERDA. Each of these programs employed a different set of efficiency measures, and each of these measures had load curves that coincided differently with the various network peaks in Con Edison’s distribution system. These forecasting challenges were additional to the typical regulatory uncertainties around program approval (when would programs be approved and what alterations would be required?), performance uncertainties (how fast would programs ramp up?), and market uncertainties (customer acceptance and macroeconomics). One final complication was the fact that goals for these new programs were expressed in cumulative energy savings rather than peak demand reductions.
Nevertheless, the new EEPS programs were initially expected to result in over 800 MW of peak demand reduction during the 2010 through 2015 implementation period, or approximately 6 percent of Con Edison’s system peak. The company knew that not accounting for such a significant amount of load reduction in its forecasting process could result in substantial capital being invested in unneeded load relief projects. In fact, the inclusion of this EEPS DSM in the demand forecast ultimately resulted in the deferral or cancellation, at least on paper, of more than $1 billion of load relief work over the 10 year planning horizon.
The approach developed by Con Edison to forecast reliable EEPS DSM achievements in each network and load area involved three major steps:
• First, expected program achievements, expressed in annual energy savings, had to be mapped (i.e., allocated) to the 91 separate networks and load areas comprising the distribution system;
• Second, the annual energy savings in each network had to be converted to an expected peak demand reduction based on the coincidence of various efficiency measures with the local network peaks; and
• Third, some measure of uncertainty had to be assigned to the expected demand reductions to reflect the inherent geographic variability in real outcomes—that is, recognizing that the actual impact in each network will vary somewhat from the expectation.
Because these EEPS programs are new, and not all are run by Con Edison, sufficient market data on geographic penetration rates for each program simply aren’t yet established. Thus, we sought a known proxy function that would distribute in a similar way to the expected adoption of energy efficiency measures. Annual energy consumption by account and service class (SC) was selected as the best available proxy. By combining Con Edison’s billing data with the company’s CUFLINK system—which matches accounts to network locations—it was possible to construct a 91 row by 17 column matrix (M) of 2009 annual energy consumption by network (rows) and service class (columns). This matrix was normalized to describe the proportional share of each megawatt-hour of 2009