NW Natural High Efficiency Furnace Program Evaluation. Overview of Presentation

Size: px

Start display at page:

Download "NW Natural High Efficiency Furnace Program Evaluation. Overview of Presentation"

Baldwin O’Neal’
5 years ago
Views:

1 NW Natural High Efficiency Furnace Program Evaluation Mark E. Thompson Overview of Presentation Study purpose and objectives Approach sample selection data preparation model specification Results by market segment conversion new construction replacement implications Slide 2 Copyright 22 Page 1

2 Study Objectives Estimate average annual therm savings from NW Natural s High Efficiency Furnace program. By market segment New construction Conversion Replacement (inferred from Conversion results) For program participants April 1999-March 2 Selected to provide most recent participants given billing history requirements. Slide 3 Summary of Findings Realization rates, measured savings as a percent of planned, are near 9% in all segments. Savings are higher than previous evaluation (1997 cohorts), especially in new construction. Replacement segment savings are inferred from conversion segment due to lack of an adequate control group. Annual Therm Savings by Segment Pct of Segment Savings Planned Conversion 85 87% New Construction 12 94% Replacement 12 87% Slide 4 Copyright 22 Page 2

3 Program Participation Program Participation by Year and Segment 3,5 High Eff. Furnace Homes 3, 2,5 2, 1,5 1, YTD July Conversion New Construction Replacement Slide 5 Overview of Methodology Select Sample of Participants Select Control Homes Match Weather Data to Billing Records Aggregate Billing Data for Modeling Model by Segment Slide 6 Copyright 22 Page 3

4 Methodology: Sample Selection Participants April March 2, conversions and new construction segments. Replacement segment (gas-to-gas) not included in study since control group cannot be identified. Furnace Program Database Control group selected for conversion and new construction segments, stratified by size of home to match distribution of home size in participant group. Control homes required to have furnace equipment code. New Sets History Select Sample Sample of Participant and Control Homes n=2,166 Meter records from July 2 - June 21. Reads with Cancel billing code dropped. Square feet data from assessor required to assure similar home sizes between participant and control groups. New const. participants used square feet from program record if assessor data were not available. Monthly Billing Data Assessor Data Conversion New Construction Participant Control Total 1, Slide 7 Number of Participants in Analysis High efficiency furnace installed April 1999 March 2 Site address standardized for match to assessor records by site address Match rate to CIS highest for conversion homes Billing history requirements No more than 2 estimated reads or other irregular events in history 12 months of billing data available (July 2 June 21) Conversion Pct of Total New Construction Pct of Total Program database 1,442 1% 244 1% Address standardized 1,317 91% % Matched to CIS 1,28 89% % Required billing data available % % Slide 8 Copyright 22 Page 4

5 Control Group Selection, Conversion Homes Participant homes larger than system average Stratified sample required for control group System average calculated over same time period as program participants (April 1999 March 2) 2,5 2, 1,5 1, 5 Comparison of Average Square Feet, Conversion Homes 1,649 System Average 1,877 Program Participant Slide 9 Control Group Selection, New Construction Homes Participant homes larger than system average Stratified sample required System average calculated over same time period as program participants (April 1999 March 2) 3, 2,5 2, 1,5 1, 5 Comparison of Average Square Feet, New Construction Homes 2,19 System Average 2,51 Program Participant Slide 1 Copyright 22 Page 5

6 Methodology: Matching Weather to Billing Data Monthly Billing Data Therms Read Date HDD /4/ /6/ /6/ /5/ /4/ /6/98 75 Daily Weather ` File Date Temp HDD65 2/4/ /5/ /6/ /7/ /8/ /5/ Total 581 Slide 11 Methodology: Aggregate Premise Data Premise Billing and Weather File House Read Date Days Therms HDD A 2/1/ B 2/15/ C 2/2/ Total Wt Avg Per Day* *Weighted average calculated by dividing sum of therms or HDD by sum of billing cycle days. Aggregated Billing and Weather File Group Month Therms Per Day HDD Per Day Control Jan Control Feb Control Mar Slide 12 Copyright 22 Page 6

7 Program Impact Model Relate usage to weather and program participation: TPD? a? b( HDD r )? c( HDD r? PD ) Where: TPD = Therms per day a, b, and c = regression model coefficients HDDr = Heating Degree Days per Day at reference temperature r. using 5 standard NOAA values for r (5, 55, 57, 6 and 65) PD = Participation indicator (dummy) variable Slide 13 Graphical Representation of Impact Model Control Group Therms per Day a Slope = b Slope = b + c (c is negative) Program Impact Participant Group Normal Heating Degree Days per Day Slide 14 Copyright 22 Page 7

8 Model Data: Conversion Conversion Modeling 5 Therms per Day Heating Degree Days Per Day Control Participant Slide 15 Model Results: Conversion Constants used in calculations: Normal heating degree days (base 57) 2,514 t-value for confidence interval Average Square Feet 1,877 Base Heat Program Impact Model Coefficients Standard Error t value Annual Therm Usage % Confidence Interval (+/-) Therm Per Square Feet previous furnace evaluation (1997 participants) Slide 16 Copyright 22 Page 8

9 Comparison to Engineering Estimates: Conversion Measured savings 87% of planning estimate Furnace load reduced by 14.7% Planning Measured Percent of Planning Std Eff Furnace load (therms per year) % Gross therm savings per participant 13 Adjustment for market penetration 25% NA (from Furnace Efficiency Survey ) Net Savings % Percent of furnace load 12.2% 14.7% Slide 17 Model Data: New Construction 6 New Construction Modeling Therms per Day Heating Degree Days Control Participant Slide 18 Copyright 22 Page 9

10 Model Results: New Construction Constants used in calculations: Normal heating degree days (base 57) 2,514 t-value for confidence interval Average Square Feet 2,51 Base Heat Program Impact Model Coefficients Standard Error t value Annual Therm Usage % Confidence Interval (+/-) Therm Per Square Feet previous furnace evaluation (1997 participants) Slide 19 Comparison to Engineering Estimates: New Construction Measured savings 94% of planning estimate Furnace load reduced by 2% Planning Measured Percent of Planning Std Eff Furnace load (therms per year) % Gross savings per participant 13 Adjustment for market penetration 2% NA (from Furnace Efficiency Survey ) Net Savings % Percent of furnace load 15.9% 2.5% Slide 2 Copyright 22 Page 1

11 Savings per Participant Savings per Participant, 9% Confidence Interval Therms Conversion New Construction Slide 21 Extending Results to Replacement Segment Using conversion results for replacement segment may understate replacement savings Replacement homes are slightly larger Lower penetration of high efficiency in replacement Average year built is nearly equal (early 195s) 2,5 2, 1,5 1, 5 Average Square Feet Source: Program data 1,991 2,62 4% 3% 2% 1% Penetration of High Efficiency Furnaces Source: 1997 NWN Survey 25% 11% Conversion Replacement % Conversion Replacement Slide 22 Copyright 22 Page 11

12 Calculating Replacement Savings Description Amount Code Formula Conversion savings (therms) 85 A Average square feet, conversion (program data) 1,991 B Conversion savings per square foot C A/B Average square feet, replacement (program data) 2,62 D Replacement savings 88. E C*D Market penetration - conversion (Survey ) 25% F Market penetration - replacement (Survey ) 11% G Difference in market penetration 14% H F-G Replacement savings adjusted for market penetration 12 I E/(1-H) Slide 23 Comparison to Previous Furnace Evaluation Savings are slightly higher in the conversion and replacement segments. New construction savings are significantly higher and more in line with engineering estimates. Average Therms Saved Per Participant Cohort Segment /2 Average Conversion New Construction Replacement Slide 24 Copyright 22 Page 12

13 Supporting Materials Computer files: Aggregated model data [Aggregated model data.xls] SAS printouts, regression model results [SAS Model Results.pdf] Slide 25 Copyright 22 Page 13

Attachment JEM 4 Hearing Exhibit 116 Page 1 of 11] Residential Sales 5,817,938 90,842,431 96,660,368. Normal HDD

Attachment JEM 4 Hearing Exhibit 116 Page 1 of 11] Residential Sales 5,817,938 90,842,431 96,660,368. Normal HDD Residential Sales Page 1 of 11] F=C*(D E)*B H=G F B C D E F G H Month HDD Coefficient Customers HDD Normal HDD Weather Impact Actual Billed WN Billed Jan-16 0.011693 1,256,043 1,183 1,102 1,193,343 18,806,509