residential single family weatherization and hvac measures

24
Residential Single Family Weatherization and HVAC Measures Progress Reports: 1. Estimating Electric and Supplemental Fuels Savings 2. Estimating Value of Emissions Savings Regional Technical Forum June 18, 2013

Upload: bishop

Post on 10-Feb-2016

34 views

Category:

Documents


0 download

DESCRIPTION

Residential Single Family Weatherization and HVAC Measures. Progress Reports: Estimating Electric and Supplemental Fuels Savings Estimating Value of Emissions Savings Regional Technical Forum June 18, 2013. Estimating Electric and Supplemental Fuel Savings. Reminder. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Residential Single Family Weatherization and HVAC Measures

Residential Single Family Weatherization and HVAC Measures

Progress Reports:1. Estimating Electric and Supplemental Fuels Savings

2. Estimating Value of Emissions Savings

Regional Technical ForumJune 18, 2013

Page 2: Residential Single Family Weatherization and HVAC Measures

2

Estimating Electric and Supplemental Fuel Savings

Page 3: Residential Single Family Weatherization and HVAC Measures

3

Reminder• The SEEM Calibration applies

to a specific sub-set of the RBSA homes.– 30% of the 1404 RBSA homes

were used in calibration, the rest weren’t included because:• Incapable of running in SEEM

(foundation type, etc.)• Non-utility fuel use or

equipment (wood, oil, etc.)• Poor billing analysis results

– Note: Gas-heated homes were included in the calibration

975

429

SEEM Calibration

SF RBSA Pie: 1404 Homes

Page 4: Residential Single Family Weatherization and HVAC Measures

4

552

Gas Heated, 249

Electric Heated, 180

In Utility Programs, but not in SEEM

calibration, 423

SEEM Calibration

SEEM is Calibrated … Now What?• The next step is to “bring back into

the analysis” the houses we expect to come in under utility programs.– Utility Program Requirement:

Permanently Installed Electric Heat• No gas, oil, etc. primary heating

systems (FAF or Boiler)• Heat stoves and fireplaces are ok (any

fuel)

• Adjustments Needed– Non-Utility Heating Fuels– Gas Heat Use

• Some houses with “Permanently Installed Electric Heat” use gas (i.e. fireplaces)

– Remaining Calibration FiltersSF RBSA Pie: 1404 Homes

Page 5: Residential Single Family Weatherization and HVAC Measures

5

Overview

• General approach is to estimate adjustment factors that account for electric heating energy differences between the SEEM calibration sample and program population(s).

• A KWh consumption or savings value will ultimately be obtained as:

Intent: this product should “reliably” estimate average electric heating kWh for the target population(s).

Page 6: Residential Single Family Weatherization and HVAC Measures

6

Overview (cont.)• A fundamental question: What is the right level of granularity?

– A single regional true-up factor? – Separate factors for different subpopulations defined by geography,

program screening criteria, or other variables?– What can the data reliably support?

• Some known limitations (for the record):– RBSA data is a snapshot (can’t address changes over time);– RBSA data is observational rather than experimental (lets us

estimate correlation between building characteristics and heating energy—not quite the same as estimating savings caused by program-related measures);

Page 7: Residential Single Family Weatherization and HVAC Measures

7

MethodologyStarting point: Easiest approach would be to calculate a single adjustment factor as a simple ratio,

The problem: This captures the two groups’ differences with respect to all variables that drive heating energy (HDDs, insulation, non-utility heating energy, equipment, partial occupancy, ...).

Want adjustment factor(s) to capture some variables’ effects (e.g., partial occupancy, non-utility heating energy). But other variables (e.g., heating equipment, HDDs, insulation) are specified in SEEM input. Don’t want to capture these variables’ effects (we want to control for these variables).

Page 8: Residential Single Family Weatherization and HVAC Measures

8

Methodology (cont.)• Regression lets us estimate individual variable effects

(when other variables are held constant). – Staff believes current regression model (next slide)…

• Makes physical sense; • Faithfully captures main patterns in the data; and • Is reasonably robust (not overly sensitive to random noise).

– Model development and related technical issues to be provided in a self-contained report.

• Today’s focus: – General framework; – How we use regression results to estimate adjustment factors;– Uncertainty and limitations.

Page 9: Residential Single Family Weatherization and HVAC Measures

9

Regression Summary(Model fit to RBSA sites with permanently installed electric heating system and without non-electric central heating systems and with Electric Heat > 0 kWh/yr)

Variable Definition Coeff. Estimate Std. Error P-value

Natural log of electric heating use in kWh/yr (billing history) n/a n/a n/a n/a

Natural log of UA x HDD65 C1 0.63 0.06 0.00

Indicator variable for “has heat pump” C2 -0.22 0.06 0.00

Indicator variable for occupant-reported non-utility fuel use 0 < kBtu/yr ≤ 40,000 C3 -0.13 0.07 0.06

Indicator variable for occupant-reported non-utility fuel use > 40,000 kBtu/yr C4 -0.60 0.10 0.00

Indicator variable for gas heating fuel use (billing history) 0 < kWh/yr ≤ 5,000 C5 -0.27 0.14 0.06

Indicator variable for gas heating fuel use (billing history) > 5,000 kWh/yr C6 -1.14 0.13 0.00

(Intercept) Intercept C7 -0.35 0.88 0.70

Adjusted R2 = 0.27

ln (Elec tric Heat )     = 𝐶1× ln (UA×HDD )+ 𝐶2×IHeatPump+ 𝐶3×IOtherHeat _ Low  ¿ ¿

Page 10: Residential Single Family Weatherization and HVAC Measures

10

Interpretation• Regression coefficients in logarithmic models:

– Coefficient of describes elasticity means that a 1% increase in is associated with a 0.63% increase in electric heating kWh.

– Each indicator coefficient estimates (roughly) the factor by which electric heating kWh typically differs between houses that have the indicated characteristic and those that do not.

Example: says that (all else being equal) a house that has a heat pump will average about 22% less electric heat kWh than one that does not.

• HDDs, UA, and heat pump presence can be specified in SEEM input. – Want to control for (rather than capture) these characteristics’ effects in

calculating adjustment factors. – and heat pump variables included in the model so that their effects are

not be attributed to other (possibly correlated) variables.

Page 11: Residential Single Family Weatherization and HVAC Measures

11

Interpretation (cont.)The following adjustments will be made to SEEM outputs to determine electric and other fuels consumption and savings• Non-Utility Heating Fuels

– Adjustment based on C3 and C4 and occurrence of non-utility heating fuels within the population we’re interested in

• Gas Heat Use– Adjustment based on C5 and C6 and occurrence of gas heating use within the

population we’re interested in• Remaining Calibration Filters

– A “filtered out of SEEM calibration for other reasons” variable did not show valid results in the regression, meaning there is no adjustment needed (that we can see)

• Electric Heat = 0 kWh/yr– This is a new adjustment, based on the filter applied prior to the regression.– Adjustment based on percentage of population we’re interested in with 0 kWh.

Page 12: Residential Single Family Weatherization and HVAC Measures

12

Example: All Program-Eligible HousesStep 1 – Determine Adjustment Factors:

Step 2 – Use Adjustment Factors to determine Electric Savings, “Wood” Savings, and Gas Savings

Adjustment Category

Regression Variable

% of Homes

Coeff.Coeff. Value

iOtherHeatLOW 29% C3 -0.13 96%iOtherHeatHIGH 12% C4 -0.60 93%

iGasHeatLOW 4% C5 -0.27 99%iGasHeatHIGH 6% C6 -1.14 94%

Electric Heat = 0 n/a 7% n/a n/a 93% 93%77%

Adjustmente(%ofHomes*Coeff.Value)

Non-Utility Heating Fuels

Gas Heat Use

Overall Adjustment

92%

90%

Non-Utility Heating Fuels

Use (kWh)

Gas Heating Use (kWh)

Electric Heat = 0 Adjustment

(kWh)

Baseline 8000 6175 1825 766 558 501

Efficient-Case 6500 5018 1482 622 453 407

Savings 1500 1158 342 144 105 94

Non-Electric Adjustment

(kWh)Case

SEEM Heating Energy Use

(kWh)

Adjusted Electric Heating Energy

Use (kWh)

Page 13: Residential Single Family Weatherization and HVAC Measures

13

Many Different Sub-Populations we could Analyze

Note: Pass billing screen = True if Total Electric Bill kWh/yr > 4.3 * Square Footage + 1000 (this screen can be edited in the workbook)

Non-Utility Heating Fuels

Gas Heat Use

Electric Heat = 0

Overall

All 90% 92% 93% 77% 562Heating Zone 1 91% 92% 95% 80% 464Heating Zone 2 81% 98% 82% 65% 68Heating Zone 3 85% 93% 94% 74% 30eFAF Only 92% 100% 89% 82% 75eZonal Only 89% 98% 96% 83% 267Heat Pump Only 89% 90% 91% 74% 203Don't Pass Screen 84% 79% 56% 37% 44Pass Screen 90% 94% 99% 83% 518

nPopulation

Description (Program-eligible houses only)

Adjustment

Page 14: Residential Single Family Weatherization and HVAC Measures

14

Can we really differentiate effects by heating zone? (Off-grid fuel example)

• These intervals only account for uncertainty in non-utility fuel usage within each group—they do not account for uncertainty related to regression fit.

50%

60%

70%

80%

90%

100%

All HZ1 HZ2 HZ3

Adju

stm

ent f

or N

on-U

tility

Hea

ting

Fuel

s

90% Confidence Interval for the Non-Utility Heating Fuels Adjustment

Page 15: Residential Single Family Weatherization and HVAC Measures

15

“Wood/Other” Heat Screen• Principles

– Has to be “auditable”• Can’t be “How much wood heat do you use”?

– Should use data readily available to the utility• Yes

– Electric consumption– Square footage

• No– Gas usage– UA

• Looked at different screens:– Total Bill normalized by square footage– Electric Heat Usage (i.e. PRISM type analysis) normalized by square footage

• Didn’t find a good screen definition that showed a significant difference between the adjustments for wood– Note the screen on the previous slide is extreme (only 9,600 kWh of total electric use for

a 2,000 ft2 house) and still didn’t show much difference in wood adjustment

Page 16: Residential Single Family Weatherization and HVAC Measures

16

Discussion• The methodology relies on the space heating behavior of the “Program-eligible”

group (green wedge on slide 4) to be similar to the behavior of the “SEEM calibration” group (yellow wedge)

• Are we on the right track?• How much should we try to split things up?

– All Houses– By Climate– By Measure

• By measure efficient and baseline case– By Utility Billing Screen– Combination of the above– Note: The more we split the population, the worse the confidence in the results

• For how long should the results be used?• Should we assemble a subcommittee to go through the details and guide the

final approach?

Page 17: Residential Single Family Weatherization and HVAC Measures

17

Estimating Value of Emissions Savings

Page 18: Residential Single Family Weatherization and HVAC Measures

Wood Heat Emissions Valuation

• Wood fire produces a large amount of pollution.• The most significant health effect is for small

particulates (PM2.5).• Health effects developed over twenty years

focused on lung disease (COPD, Emphysema, Cancer) derived for atmospheric exposure

• This is among the most significant pollutants from wood smoke.

Page 19: Residential Single Family Weatherization and HVAC Measures

Valuation of health effects from PM2.5

• Primary source used:– http://

www.epa.gov/airquality/benmap/models/Source_Apportionment_BPT_TSD_1_31_13.pdf (EPA, 2013)

• The source for woodstove emissions was:– http://www.epa.gov/ttnatw01/burn/woodburn1.pdf (Valenti &

Clayton, 1998)

• Emission valuation taken as the effect of the incremental particulates added to the atmosphere

Page 20: Residential Single Family Weatherization and HVAC Measures

Emission Value, kWh equivalent

Combustion DeviceEmission Rates

Input Heat Stove Output Heat Electric Heat mg/MJ efficiency mg/MJ mg/kWh Fire Place 904 0.2 4520 16272Conventional Wood Stove 786 0.5 1572 5659Certified Wood Stove (catalytic) 425 0.6 708 2550Certified Wood Stove (non-catalytic) 383 0.6 638 2298Pellet (Certified) 110 0.75 147 528

Combustion Device Emission Valuation Low Mean High $/kWh Fire Place 5.73 10.11 14.50Conventional Wood Stove 1.99 3.52 5.04Certified Wood Stove (catalytic) 0.90 1.58 2.27Certified Wood Stove (non-catalytic) 0.81 1.43 2.05Pellet (Certified) 0.19 0.33 0.47

Page 21: Residential Single Family Weatherization and HVAC Measures

Overall Impact on Savings

• Valuation of wood savings about ten time avoided cost of electricity

• Impact on B/C of individual measures dominated by wood heat offsets – Limited to homes with wood heat– Reduces electric savings but increases B/C ratios

• Valuation does not include generation.• Generation reduces emissions offset by 30%

Page 22: Residential Single Family Weatherization and HVAC Measures

National Woodstove Sales Data

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 20120

100,000

200,000

300,000

400,000

500,000

600,000

700,000

800,000

900,000

f(x) = 928649.027265403 exp( − 0.100451881167818 x )R² = 0.883317535159771

Cord Wood Product Sales

Cord Wood Products

Exponential (Cord Wood Products)

Page 23: Residential Single Family Weatherization and HVAC Measures

Woodstove use declining

• Air Quality concerns• Emissions regulation in urban areas• Cost of wood rising relative to alternatives• Sales of wood burning device declining

nationwide– 75% reduction in 15 years

• Improves air quality, increases electric savings

Page 24: Residential Single Family Weatherization and HVAC Measures

24

Subcommittee

• Call for Subcommittee Members for an “Emissions Analysis Subcommittee”– Review the input assumptions to arrive at a

method of monetizing wood/other emission savings