Gorge Area Demand Forecast. Prepared for: Green Mountain Power Corporation 163 Acorn Lane Colchester, Vermont Prepared by:

Similar documents
Multivariate Regression Model Results

From Sales to Peak, Getting It Right Long-Term Demand Forecasting

2013 FORECAST ACCURACY BENCHMARKING SURVEY AND ENERGY

Chapter 3. Regression-Based Models for Developing Commercial Demand Characteristics Investigation

2013 WEATHER NORMALIZATION SURVEY. Industry Practices

WEATHER NORMALIZATION METHODS AND ISSUES. Stuart McMenamin Mark Quan David Simons

LOADS, CUSTOMERS AND REVENUE

Defining Normal Weather for Energy and Peak Normalization

COMPARISON OF PEAK FORECASTING METHODS. Stuart McMenamin David Simons

Report on System-Level Estimation of Demand Response Program Impact

2013 Weather Normalization Survey. Itron, Inc El Camino Real San Diego, CA

Into Avista s Electricity Forecasts. Presented by Randy Barcus Avista Chief Economist Itron s Energy Forecaster s Group Meeting

NSP Electric - Minnesota Annual Report Peak Demand and Annual Electric Consumption Forecast

Demand Forecasting Models

Design of a Weather-Normalization Forecasting Model

2018 FORECAST ACCURACY BENCHMARKING SURVEY AND ENERGY TRENDS. Mark Quan

Abram Gross Yafeng Peng Jedidiah Shirey

peak half-hourly New South Wales

Page No. (and line no. if applicable):

Variables For Each Time Horizon

Use of Normals in Load Forecasting at National Grid

peak half-hourly Tasmania

2014 FORECASTING BENCHMARK AND OUTLOOK SURVEY. Mark Quan and Stuart McMenamin September 16, 2014 Forecasting Brown Bag Seminar

SYSTEM BRIEF DAILY SUMMARY

Determine the trend for time series data

Development of Short-term Demand Forecasting Model And its Application in Analysis of Resource Adequacy. For discussion purposes only Draft

UNBILLED ESTIMATION. UNBILLED REVENUE is revenue which had been recognized but which has not been billed to the purchaser.

SYSTEM BRIEF DAILY SUMMARY

BEFORE THE FLORIDA PUBLIC SERVICE COMMISSION DOCKET NO EI

TRANSMISSION BUSINESS LOAD FORECAST AND METHODOLOGY

Estimation of Energy Demand Taking into Account climate change in Southern Québec

Monthly Long Range Weather Commentary Issued: APRIL 18, 2017 Steven A. Root, CCM, Chief Analytics Officer, Sr. VP,

Salem Economic Outlook

Monthly Long Range Weather Commentary Issued: February 15, 2015 Steven A. Root, CCM, President/CEO

Ontario Demand Forecast

peak half-hourly South Australia

Monthly Sales Weather Normalization and Estimating Unbilled Sales. Al Bass Kansas City Power & Light EFG Meeting Las Vegas, NV April 2-3, 2014

Monthly Long Range Weather Commentary Issued: APRIL 25, 2016 Steven A. Root, CCM, Chief Analytics Officer, Sr. VP, sales

PREPARED DIRECT TESTIMONY OF GREGORY TEPLOW SOUTHERN CALIFORNIA GAS COMPANY AND SAN DIEGO GAS & ELECTRIC COMPANY

As included in Load Forecast Review Report (Page 1):

Lecture Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University

Ameren Missouri Peak Load Forecast Energy Forecasting Meeting, Las Vegas. April 17-18, 2013

CWV Review London Weather Station Move

2018 Annual Review of Availability Assessment Hours

Monthly Long Range Weather Commentary Issued: July 18, 2014 Steven A. Root, CCM, President/CEO

EVALUATION OF ALGORITHM PERFORMANCE 2012/13 GAS YEAR SCALING FACTOR AND WEATHER CORRECTION FACTOR

March 5, British Columbia Utilities Commission 6 th Floor, 900 Howe Street Vancouver, BC V6Z 2N3

SEPTEMBER 2013 REVIEW

2006 IRP Technical Workshop Load Forecasting Tuesday, January 24, :00 am 3:30 pm (Pacific) Meeting Summary

2003 Moisture Outlook

BESPOKEWeather Services Monday Afternoon Update: SLIGHTLY BULLISH

BEFORE THE PUBLIC UTILITIES COMMISSION OF THE STATE OF COLORADO * * * * *

Weather Normalization: Model Selection and Validation EFG Workshop, Baltimore Prasenjit Shil

Public Library Use and Economic Hard Times: Analysis of Recent Data

CHAPTER 5 - QUEENSLAND FORECASTS

Forecasting. Copyright 2015 Pearson Education, Inc.

Champaign-Urbana 2001 Annual Weather Summary

= observed volume on day l for bin j = base volume in jth bin, and = residual error, assumed independent with mean zero.

Colorado s 2003 Moisture Outlook

Monthly Long Range Weather Commentary Issued: APRIL 1, 2015 Steven A. Root, CCM, President/CEO

Proposed Changes to the PJM Load Forecast Model

NatGasWeather.com Daily Report

Euro-indicators Working Group

BUSI 460 Suggested Answers to Selected Review and Discussion Questions Lesson 7

Jackson County 2013 Weather Data

Lecture Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University

FEB DASHBOARD FEB JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

Chiang Rai Province CC Threat overview AAS1109 Mekong ARCC

The Colorado Drought of 2002 in Perspective

Drought in Southeast Colorado

YEAR 10 GENERAL MATHEMATICS 2017 STRAND: BIVARIATE DATA PART II CHAPTER 12 RESIDUAL ANALYSIS, LINEARITY AND TIME SERIES

Champaign-Urbana 1999 Annual Weather Summary

STATISTICAL FORECASTING and SEASONALITY (M. E. Ippolito; )

RD1 - Page 469 of 578

Introduction to Forecasting

Interstate Power & Light (IPL) 2013/2014

Short-Term Job Growth Impacts of Hurricane Harvey on the Gulf Coast and Texas

Champaign-Urbana 2000 Annual Weather Summary

Jackson County 2018 Weather Data 67 Years of Weather Data Recorded at the UF/IFAS Marianna North Florida Research and Education Center

particular regional weather extremes

2015 Summer Readiness. Bulk Power Operations

LODGING FORECAST ACCURACY

Climatography of the United States No

SMART GRID FORECASTING

Winter Season Resource Adequacy Analysis Status Report

Time series and Forecasting

Table 01A. End of Period End of Period End of Period Period Average Period Average Period Average

Product and Inventory Management (35E00300) Forecasting Models Trend analysis

Forecasting the Canadian Dollar Exchange Rate Wissam Saleh & Pablo Navarro

GAMINGRE 8/1/ of 7

Normalization of Peak Demand for an Electric Utility using PROC MODEL

BEFORE THE PUBLIC UTILITY COMMISSION OF THE STATE OF OREGON UE 294. Load Forecast PORTLAND GENERAL ELECTRIC COMPANY. Direct Testimony and Exhibits of

A Report on a Statistical Model to Forecast Seasonal Inflows to Cowichan Lake

CAVE CLIMATE COMPARISON ACTIVITY BETWEEN THE SURFACE AND THE CAVERN

The xmacis Userʼs Guide. Keith L. Eggleston Northeast Regional Climate Center Cornell University Ithaca, NY

NatGasWeather.com Daily Report

Corn Basis Information By Tennessee Crop Reporting District

BEFORE THE PUBLIC UTILITIES COMMISSION OF THE STATE OF COLORADO * * * *

Statistics for IT Managers

Climatography of the United States No

Transcription:

Exhibit Petitioners TGC-Supp-2 Gorge Area Demand Forecast Prepared for: Green Mountain Power Corporation 163 Acorn Lane Colchester, Vermont 05446 Prepared by: Itron, Inc. 20 Park Plaza, Suite 910 Boston, Massachusetts 02116 (617) 423-7660 April 25, 2009 "rtf"

Gorge Area Demand Forecast Green Mountain Power (GMP) is evaluating the need to upgrade the transmission system in the Gorge Area (GAR) GMP requested that Itron evaluate historical demand growth trends and develop a peak demand forecast for the transmission planning area. Monthly peak weather normalization and forecast models were developed using historical load data for the GAR planning area covering the period 2003 to 2008. Monthly peaks were derived from historical hourly load data. For the purpose of this analysis Saxon Hill peak demand was subtracted out from the resulting monthly demand series. Figure 1 shows monthly historical GAR peaks. The blue shows monthly GAR peak demand (excluding Saxon Hill) and the red line is a 12-month moving average of the monthly demand. Figure 1: Gorge Area Monthly Peak Demand (MW) 2003-2008 - MPellks.Pellks - MPeaks.MA_Peaks 85.-------------------------------------. eo 75 70 65 55 50 45'-------.-----.---..-------.---.----r-------.-----.----.----.-------.-----l Jan-03 Jul-03 Jan 04 Jul 04 Jan 05 Jul-05 Jan-OO Jul-OO Jan 07 Jul-07 Jan Oe Jul-08

Gorge Area Demand Forecast As depicted in Figure 1, the GAR planning area is summer peaking. The planning area reached a non-weather normalized maximum demand of 83.1 MW in August 2006. The moving average has been increasing over this period indicating positive GAR area demand growth. Figure 2 shows a scatter of monthly GAR peak against average daily peak-day temperature. The summer months are coded red and the winter months are blue. Again, the graph shows the planning area is summer peaking. Figure 2: Monthly Peak Demand vs. Average Peak-Day Temperature MBin.Snmmer MBin.Winter MBin.Sprin!l 100..,---..-----,,.-----.-----.-----.----.---...----..----..----,-----.-----. :...,.. 1It.it4 I.. 75 ' : - ~ -.. + : ~ + -: :... A.i..~A. ~. ~: : : : : : : :.6.;4 It. : I : ": : : A: : "./IIIt.:.,.: : I.. Jt :~... I!.. I.. ' ~ til.:./i :,~: :,.-'" I.. : ~ IlA A tja. ~ 11 t:.,11. : 11 : ~ : 4: : I..I&A- A'" I.. -.4.:1.. :.... :,.. :.. :. : :.. : : I I I I I I I I I 50....T ~ r c t:... r..t r T.. T..r.., I t I I t I I I t I I I I., I I 25 f-.. --_.. -- --:- -------- -~ -- _. -.... ;_..._. -~........ -~--. -...... -~...-... ---- -~-. ----- I --~. ------ I I.. i: - I -: -~--- I I t I' I,. I I. t 'I I t I I I I, t ----. I I I I o L-_~''-_-.i.' i... ~...;...,..;.' i......i......i- I I..;.' I t --'-....J 20 10 o 10 20 30 40 50 60 70 80 90 100 Table 1 shows the annual summer peaks, temperatures occurring on the day and day before the peak, and the maximum average daily temperature for the year. Table 1: GAR Summer Peaks (MW) and Peak-Day Average Temperature Year Month Load (MW) Peak Temp Prior-Day Temp Max Temp 2003 June 79.1 81 76 81 2004 August 73.5 72 78 78 2005 July 81.1 82 83 83 2006 August 83.1 84 83 84 2007 June 81.4 82 81 85 2008 June 79.8 80 80 80 Average 79.7 80 80 82 Itron. Inc. 2

GAR Area Demand Forecast While peak demand is strongly correlated with hot weather, the peak does not necessarily occur on the hottest day. The peak tends to occur after two consecutive days of relatively hot weather. In 2003 and 2007, while the peaks occurred in June, the summer peak demand in July and August are just slightly lower. The exception is in 2008; July and August were extremely cool with the peak-day temperature in July of just 76 degrees and in August 68.5 degrees. Normalized GAR Peak Demand Working with small area load data can be difficult as there tends to be a significant variation in demand due to variation in weather conditions and general noise associated with small area demand data. The first task is to isolate the weather variation in order to evaluate historical demand trends. To normalize historical demands we first constructed a monthly demand weather impact regression model. The model includes peak-day CDD (PKCDD), prior-day CDD (PKCDD _Lag!), peak day HDD (PKHDD), a trend variable to account for non-weather sensitive growth (TrendVar) and a trend variable interactive with CDD to account for changing cooling response over time (CDDPK_Trend); we would expect cooling load response to increase over time as a result of both customer growth and increasing residential air conditioning saturation The model fits the historical data relatively well with an adjusted R2 of 0.88 and an average model error of2.9%. Table 2 shows the estimated model. Actual and predicted results are depicted in Figure 3. Itron. Inc. 3

Gorge Area Demand Forecast Table 2: Estimated Weather Normalization Model Regression Statistics Adjusted Observations 72: Deg. of Freedom for Error R-Squared Adjusted R-Squared Durbin-Watson Statistic F-Statistic Prob (F-Statistic) Model Sum of Squares Sum of Squared Errors Mean Squared Error Std. Error of Regression Mean Abs. Dev. (MAD) Mean Abs. % Err. (MAPE) 63 0.89 0.88 1.98: 64.324 o 3469 425' 6.74, 2.6: 1.89' 2.88% Variable CONST PkCDD PkCDD_Lag1 PkCDD_Trend PkHDD APR MAY APR04 TrendVar Coefficient: 62.835 0.422' 0.412 0.057 0.036' -6.505-6.188; 7.395-0.076 StdErr' 1.21 0.178: 0.138 0.029 0.02. 1.26 1.21 2.874: 0.218 T-Stat 51.93 2.38 2.98' 1.97 1.79' -5.16-5.11 2.57-0.35' P-Value 0.0% 2.1% 0.4% 5.4% 7.8% 0.0% 0.0% 1.3% 72.9% Itron, Inc. 4

GAR Area Demand Forecast Figure 3: Weather Normalization Model Actual and Predicted (MW) 100r--------------------------------. 50------------------------------------------------------------------------- 25 -------------------------------------------------------------------------------------- O'-------..------.------...-----.------.------l Jan-03 Jan-04 Jan-07 Jan-OS Once estimated, the model is used to weather normalize monthly peak demands. Normal peak-day weather conditions are calculated from the extreme temperatures experienced over the last thirty years (1979 to 2008). Normal monthly peak-producing weather is calculated by ranking the maximum temperature from the highest to lowest temperature over the last thirty years within each month. The normal monthly temperature is defined as that with a 50% probability of occurring. Not surprisingly, the warmest weather occurs in July. The annual expected peak producing weather is calculated by ranking the highest temperature in each year regardless of the month. Table 3 shows the result of this ranking. Itron, Inc. 5

Gorge Area Demand Forecast Table 3: Maximum Daily Average Temperature Ranking Year PkDay AvgTemp PkDay CDD65 Probability 2001 89 24 3% 1995 88.5 23.5 7% 1988 86 21 10 ;' 1983 85.5 20.5 13% 1991 85.5 20.5 17% 1987 85 20 20% 1994 85 20 23% 1982 84.5 19.5 27% 2002 84.5 19.5 30% 2007 84.5 19.5 33% 2006 84 19 37% 1981 83.5 18.5 40% 1993 83.5 18.5 43% 1999 83 18 47% 2005 83 18 500;. 1979 82 17 53% 1989 81.5 16.5 57% 1980 81 16 60% 2003 81 16 63% 1984 80.5 15.5 67% 1986 80 15 70% 1996 80 15 73% 2008 80 15 77% 1990 79.5 14.5 80% 1985 79 14 83% 1997 79 14 87% 1998 79 14 90% 1992 78.5 13.5 93% 2004 78 13 97% 2000 77 12 100% Table 3 can be used to calculate expected (50% probability) and extreme (10% probability) peak producing weather. Expected peak-day average temperature over the thirty-year period is 83 degrees or 18 CDD. The extreme peak-day temperature (defined as a 10% probability of occurring) occurs in 1998 with a maximum temperature of 86 degrees or 21 cooling degree-days. In generating normalized historical series we assume that the day prior to the peak is also warm; for the expected case we assume a prior-day CDD of 15 (80 degrees) and in the extreme case a prior-day CDD of20 (85 degrees). normalized July GAR peaks. Table 4 shows the resulting weather Itron, Inc. 6

GAR Area Demand Forecast Table 4: Actual and Weather Normalized July GAR Peaks (MW) Year July Peak WN_July Pk Extreme_JulPk 2003 74.2 79.5 83.1 2004 73.4 79.9 83.7 2005 81.1 80.8 84.7 2006 80.0 82.1 86.2 2007 78.7 83.5 87.8 2008 79.7 85.9 90.2 The peak-day weather over the most recent years has been cooler than normal. Only one year (2006) did the peak-day average temperature exceed normal peak-day temperature. As mentioned before, 2008 was extremely cool. When normalized, the expected 2008 July peak is 85.9 MW. Extreme weather (10% probability weather) results in a demand estimate that is approximately 5.0% higher than in the expected weather case; this translates into a 10% probability demand estimate of90.2 MW in 2008. GAR Peak Demand Forecast Model The next part of this project is to develop a GAR peak demand forecast. The objective is to construct a forecast model that incorporates the impact of projected economic conditions and prices as well as long-term trends in end-use saturation and efficiency. To accomplish this, we constructed a monthly demand model that relates GAR monthly peak demand to end-use energy forecasts for heating, cooling, and other use. An initial peak demand forecast was constructed based on the November 2008 GMP sales forecast; the forecast reflected Ecnonomy.com's November 2008 Vermont economic forecast. The forecast was later updated to reflect Economy.com's March 2009 economic projections. Estimating End-Use Energy Projections Company-level monthly sales forecast models are estimated for each of the primary revenue classes including residential, small commercial, and large commercial revenue classes. The structure of these models allows us to estimate end-use sales for each of these classes. Table 5 shows estimated end-use sales growth for the November 2008 residential and commercial sector (small and large commercial) forecasts. Jtron, Inc. 7

Gorge Area Demand Forecast Table 5: End-Use Sales Growth Projections Residential End-Use Sales Growth Projections Cooling Heating Other Lighting Total 2009 1.0% -0.8% 0.6% -2.6% -0.2% 2010 1.6% 0.0% 1.6% -0.8% 0.8% 2011 2.0% 0.1% 1.9% -0.5% 1.0% 2012 1.6% -0.1% 1.8% -0.5% 0.9% 2013 1.2% -0.8% 2.0% -20.6% -1.9% 2014 1.9% -0.2% 1.8% -5.1% 0.5% 2015 2.4% 0.4% 2.3% -2.3% 1.3% 2016 2.5% 0.4% 2.6% -0.1% 1.7% 2017 2.6% 0.4% 2.1% -0.4% 1.4% 2018 2.7% 0.5% 2.5% 0.5% 1.8% average 1.9% 0.0% 1.9% -3.2% 0.7% Commercial End-Use Sales Estimates Cooling Heating Other Total 2009 0.1% -0.5% -0.2% -0.2% 2010 0.4% 0.0% 1.2% 1.2% 2011 0.7% 0.3% 1.3% 1.2% 2012-0.7% 0.0% 1.1% 1.1% 2013-1.6% -0.5% 0.6% 0.5% 2014-1.3% 0.0% 1.0% 1.0% 2015-0.8% -0.5% 1.2% 1.2% 2016 0.0% -0.5% 1.4% 1.3% 2017 0.3% -0.5% 1.1% 1.0% 2018 0.2% -0.5% 1.2% 1.2% average -0.3% -0.3% 1.0% 1.0% Residential cooling projections are relatively strong as it reflects recent and expected strong growth in room air conditioning saturation. Residential lighting is broken out of the residential other use as there is a significant drop in lighting usage in 2013 as a result of the new Energy Independence and Security Act (EISA) lighting standards; this has a small impact on summer peak demand and a much larger impact on winter peak demand. In the commercial sector, air conditioning sales are flat to declining as improvements in air conditioning efficiency outweigh increases in air conditioning saturation. What little heating there is in the commercial sector is flat to declining. Residential and commercial end-use indices are weighted to reflect the GAR area customer mix. Sales data provided by GMP indicates that approximately 60% ofthe load served is commercial (small and large) and 40% of the load is residential. This information is used to weight commercial and residential enduse energy projections that are incorporated into the peak demand forecast model. Estimated Demand Model The weighted end-use variables capture the stock of cooling, heating, and other equipment in place. The cooling stock is interacted with the peak-day CDD and the heating stock is interacted with the peak-day HDD. The interactive heating and cooling variables and baseltron, Inc. 8

GAR Area Demand Forecast stock estimate are regressed on historical monthly peaks. forecast model. Table 6 shows the estimated peak Itron, Inc. 9

Gorge Area Demand Forecast Table 6: GAR Peak Demand Forecast Model Regression Statistics Adjusted Observations 71 Deg. of Freedom for Error 59 R-Squared 0.912 Adjusted R-Squared 0.896 Durbin-Watson Statistic 1.899 F-Statistic 51.211 Prob (F-Statistic) 0 Model Sum of Squares 3547 Sum of Squared Errors 340 Mean Squared Error 5.77 Std. Error of Regression 2.4 Mean Abs. Dev. (MAD) 1.75 Mean Abs. % Err. (MAPE) 2.63% Variable Coefficient StdErr T-Stat P-Value BaseDmd 1.406 0.021 65.929 0.00% HeatDmd 0.076 0.036 2.14 3.66% CoolDmd 1.039 0.063 16.542 0.00% APR -5.206 1.126-4.624 0.00% MAY -5.618 1.085-5.179 0.00% DEC 2.774 0.92 3.014 0.38% OCT03 6.98 2.238 3.119 0.28% AUG04 6.222 2.37 2.626 1.11 % SEP04 7.525 2.389 3.149 0.26% OCT05 8.21 2.194 3.742 0.04% APR04 11.134 2.369 4.7 0.00% AR(1) 0.494 0.116 4.254 0.01% The forecast model explains historical data well with an adjusted R2 of 0.90 and in sample average error of 2.6%. Figure 4 shows actual and predicted results. Itron. Inc. 10

GAR Area Demand Forecast Figure 4: Actual and Predicted GAR Demand (MW) 100.-----------------------------------, 50 -------------------------------------------------------------------------------------- 25 -------------------------------------------------------------------------------------- o "----.-----r-----...----...-----.------.-----.------l Jan-03 Jan-05 Jan-07 Jan-09 Jan-11 Jan-13 Jan-15 Jan-17 - Actual - Predicted The peak demand forecast is calibrated to starting 2008 weather normalized demand. 7 shows the resulting peak demand forecast for expected and extreme weather conditions based on the November 2008 economic forecast. Table ltron, Inc. 11

Gorge Area Demand Forecast Table 7: GAR Summer Peak Demand Forecast (MW) Year 50% Prob Weather 10% Prob Weather 2003 79.5 83.1 2004 79.9 83.7 2005 80.8 84.7 2006 82.1 86.2 2007 83.5 87.8 2008 85.9 90.2 2009 85.6 89.9 2010 86.8 91.1 2011 88.1 92.5 2012 89.1 93.6 2013 89.8 94.3 2014 90.8 95.4 2015 92.2 96.8 2016 93.7 98.4 2017 95.2 99.9 2018 96.8 101.6 chg 2004 0.5% 0.7% 2005 1.1% 1.2% 2006 1.6% 1.8% 2007 1.7% 1.8% 2008 2.9% 2.7% 2009-0.3% -0.3% 2010 1.3% 1.3% 2011 1.5% 1.5% 2012 1.2% 1.2% 2013 0.8% 0.8% 2014 1.1% 1.1% 2015 1.5% 1.5% 2016 1.7% 1.7% 2017 1.5% 1.5% 2018 1.7% 1.7% 2003-2008 1.6% 1.7% 2008-2018 1.2% 1.2% The historical peak demands (2003 to 2008) represent July weather-normalized demand. From the weather normalization model, we estimate that summer peak demand could swing as high as 5% under more extreme weather conditions. peak The demand forecast reflects GMP's most recent customer class and end-use sales forecast. The sales projections in tum incorporate the most recent economic, price, and end-use saturation and efficiency trends. Summer peak demand is expected to average 1.2% over the next ten years. Demand is expected to actually decline slightly in 2009 largely as a result of Itron, Inc. 12

GAR Area Demand Forecast the recession projected through 2009. Excluding 2009, summer peak demand averages 1.4% growth over the forecast period. Updated Demand Forecast Between November 2008 and April 2009 Vermont along with the country experienced a significant deterioration in economic conditions. The forecast was updated in April to reflect the poorer economic outlook. Economy.com March 2009 economic forecast was executed through the class sales forecast models (residential, general service, and large commercial) and resulting end-use energy forecasts through the peak model. was not re-estimated. The sales and peak model The primary economic drivers include: Number of households Household income Employment Gross State Product These economic variables (in addition to long-term end-use saturation and efficiency trends) drive the GMP sales forecasts which in tum drive the GAR peak demand forecast. We assume that the GAR customer base will respond to market conditions in a manner similar to that of the overall GMP customer base. Economic Projections Not too surprisingly, current economic conditions and near-term projections are significantly worse than that projected last November. Figures 1 to 4 compare forecasted quarterly economic growth (year over year) for the key forecast drivers. The March 2009 forecast is shown in blue and the November 2008 forecast is shown in red. Itron, Inc. 13

Gorge Area Demand Forecast Figure 1: Household Growth Forecast (March 2009 vs. November 2008) 0,0100,---------------------------------------, 0.0015 0.0000 L...-...,----,---..----.-----r--..---...,-----,---,...--...,-----,---..---...,----.---,...---l Q' 03 Q' 04 Q'-05 Q'-06 Q' 07 Q1-OB Q' 09 Q,,O Q,." Q,.,2 Q,.,J Q,.,4 Q' '5 Q'-'6 Q,.,7 Q,.,8 ltron. Inc. 14

GAR Area Demand Forecast Figure 2: Gross Output Growth Forecast (March 2009 vs. Nov 2008) - QEcon.ChlLGRP _Mar09- QEcon.ChlLGRP _NovOS OJJ600,---------------------------------------, 0.0500 0.0400 00300 0.0200 0.0100 OOOOO I-----\-t-H-----l----+----------------------j -0 0100-00200 -0 0300 L---..----.,..---~--~--._--._----r--_,.--,. --. --., - 1 01-05 OHa r,ii-01 01 08 01-09 01-10 OHI QI-I2 01-13 01 14 Q1-15 OHa Itron, Inc. 15

Gorge Area Demand Forecast Figure 3: Household Income Growth Forecast (March 2009 vs. Nov 2008) 0,075,---------------------------------------, 0,050 0,026 0.050.0.075L---~--_._--._-- r_--_._--.--,. --. --.. --_._--_._--_l (11 05 ell 06 0\ 01 0\-08 Q\-09 a 1-10 Q1-I1 0:11 11 0\ 13 QI \& 01-15 0\-\6 ltron, Inc. 16

GAR Area Demand Forecast Figure 4: Employment Growth Forecast (March 2009 vs. Nov 2008) 00500 '----,-----.---..-------,-----.-----.---..-------,:---:----.-----.:---:---.:---:------1 QI OS 0J.D6 QI OI QI OB QI ll9 QI lo Q' I' 01 12 01."3 01 '4 OI-lS QloI6 While household growth was slow to begin with, the March household forecast shows virtually no growth. Household growth recovers by the second quarter of2011 with longerterm household growth slightly higher (0.7%), than that of the November forecast (0.5%). Gross State Product, household income, and employment, decline at a much faster rate in the March forecast. Output growth bottoms out at -2.0% compared with 1.0% growth in the November forecast. Output growth continues to lag the November forecast until the middle of 20 11. There is little change in the long-term output growth rate. Household income and employment growth follow similar patterns. Itron, Inc. 17

Gorge Area Demand Forecast Forecast Impacts Tables 8 and 9 compare the updated summer peak demand forecast (April 2009 economic forecast) with the current summer peak demand forecasts (November 2008 economic forecast). Table 8: Expected Peak Demand Forecast (MW) chg Year Nov 2008 April 2009 Difference 2008 85.9 85.9 2009 85.6 85.3 0.3 2010 86.8 86.3 0.4 2011 88.1 87.6 0.4 2012 89.1 88.8 0.4 2013 89.8 89.4 0.4 2014 90.8 90.4 0.5 2015 92.2 91.7 0.5 2016 93.7 93.3 0.5 2017 95.2 94.7 0.5 2018 96.8 96.3 0.5 2009-0.3% -0.7% 2010 1.3% 1.2% 2011 1.5% 1.5% 2012 1.2% 1.3% 2013 0.8% 0.7% 2014 1.1% 1.1% 2015 1.5% 1.5% 2016 1.7% 1.7% 2017 1.5% 1.5% 2018 1.7% 1.7% 2008-2018 1.2% 1.1% Itron, Inc. 18

GAR Area Demand Forecast Table 9: Design Day Peak Demand Forecast (MW) chg Year November April Difference 2008 90.2 90.2 2009 89.9 89.6 0.4 2010 91.1 90.6 0.5 2011 92.5 92.0 0.5 2012 93.6 93.2 0.4 2013 94.3 93.8 0.5 2014 95.4 94.9 0.5 2015 96.8 96.3 0.5 2016 98.4 97.9 0.5 2017 99.9 99.4 0.5 2018 101.6 101.1 0.5 2009-0.3% -0.7% 2010 1.3% 1.2% 2011 1.5% 1.5% 2012 1.2% 1.3% 2013 0.8% 0.7% 2014 1.1% 1.1% 2015 1.5% 1.5% 2016 1.7% 1.7% 2017 1.5% 1.5% 2018 1.7% 1.7% 2008-2018 1.2% 1.1% With the new economic forecast, demand falls 0.7% in 2009 compared with a 0.3% decline with the November economic projections. Over the longer-term, the new economic forecast reduces demand growth from 1.2% to 1.1%. Design day demand is approximately 0.5 MW lower with the new economic projections. 3066268.1 Itron, Inc. 19