coincident peak load forecasting methodology prepared for june 3, 2010 meeting with division of...
Post on 26-Dec-2015
212 Views
Preview:
TRANSCRIPT
Coincident Peak Load Forecasting Methodology
Prepared for June 3, 2010
Meeting with Division of Public Utilities
© 2
000
PA
CIF
ICO
RP
| P
AG
E 2
Development of Coincident Peak Forecast
Customer Based Forecast
Direct Forecast of Energy by class by
state
Model Driven Forecast
Development of hourly loads and forecast of
system coincident peak
Direct Forecast of Jurisdictional Peaks
© 2
000
PA
CIF
ICO
RP
| P
AG
E 3
Monthly Energy Forecast by Customer Class and by State
Utah Residential Sales Model Results
200
400
600
800
1000
2003 2004 2005 2006 2007 2008 2009
Year
Sal
es (
GW
H)
Actual Pred_Billing Cycle
Utah Residential Number of Customer Model Results
500,000
600,000
700,000
800,000
900,000
2003 2004 2005 2006 2007 2008 2009
Year
Nu
mb
er
of
Cu
sto
mers
Actual Pred
© 2
000
PA
CIF
ICO
RP
| P
AG
E 4
Industrial Sales Forecast Approximately 25% of industrial sales is forecasted
through the model Remainder of industrial customers are separated into
two categories:• Existing customers tracked by the Customer Account
Managers (CAMs)» CAMs provide 3 year forecast for existing customers» Customer is tracked if more than 1 MW
• New large customer or expansions by existing large customers
© 2
000
PA
CIF
ICO
RP
| P
AG
E 5
Jurisdictional Peak Forecast Peak Forecast is developed using an econometric model
Monthly and seasonal peak forecast for each state are developed from historic peak producing weather and several related variables
Utah Peak Model Results
2000
3000
4000
5000
6000
2003 2004 2005 2006 2007 2008 2009
Year
Peak
(MW
)
Actual Pred
© 2
000
PA
CIF
ICO
RP
| P
AG
E 6
Coincident Peak by State The model uses historical state specific hourly load data to
develop an hourly load pattern The hourly load pattern is adjusted for line losses and calibrated
to jurisdictional peaks and energy Hourly loads are aggregated to the total company system level Coincident system peak is identified by month as well as the
contribution of each jurisdiction to those monthly system peaks Model ensures consistency between
• Peak forecast and jurisdictional peak load
• Hourly load and monthly sales with line loss
top related