Deemed Savings Methodologies

Reid Hart, PE

Associate Director, Technical Research

July 2009 – Regional Technical Forum

2

Deemed Saving Methodologies

• Need for a New Deemed Method• Relevant Baseline Parameters• Approaches• Decision Framework Matrix• Expected Value Deemed Savings• Seeking Approval as Provisional Method

3

Need for a New Deemed Method• Current approaches

– Custom savings require pre-review and approval– Calculators work well with energy auditor– Deemed savings desired for contractor delivery

• Limits on deemed savings– Very few commercial HVAC items– Savings may be excessively conservative– Approval of deemed savings not always timely

• Desired Method– Covers range of conditions– Arrives at appropriate program-wide savings

4

Relevant Baseline Parameters

• Typically: Building Type & Vintage– Vintage not account for retrofit changes– Building types can be generalized

• Meta Parameters (Require Separate Savings)– Major climate zones– Heating fuel type

• Analyzed Baseline Parameter Variation– Internal loads: lighting; density– Envelope: glazing; perimeter/area ratio– Schedule– Measure specific parameters

5

Measure Specific Baseline Parameters

• Premium Ventilation Package as an exmple:– Economizer found changeover– Economizer maximum outside airflow– Minimum outside air setting (Example field data)

Ventilation Air Minimum Setpoint

0

1

2

3

4

5 10 15 20 25 40 55 100 More

Min OA%

Fre

qu

ency

Source: Ecotope EWEB study – 2001

6

Parameters for Premium VentilationParameter Sym Parameter variation

    in baseline BEFORE

    measure is installed

Internal Load L+ 1.8 LPD, 1.5 plug, 100 sf/person

LPD = 1.8 w/sf LPD; eQuest defaults

Density L- 1.0 w/sf LPD; eQuest defaults

Ventilation V+ 37.6 cfm/person

Minimum = 31 cfm/person (typical)

  V- 22.6 cfm/person

Glazing G+ Low-e Argon, double pane

Type = Double pane, solar Bronze

  G- Single Pane

Economizer E++ B, double stage

Changeover E+ C, single stage

  = D or Snap Disk

  E- Failed Economizer

Economizer M+ 80% Max OSA

Max OSA = 65% Max OSA

  M- 50% Max OSA

7

Individual Baseline Parameter Impact on Savings

Savings with Various Baseline ParametersPortland, OR, Individual Changes

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

All = L+ L- E+ E- G+ G- M- M+ V+ V-

KW

h /

sq

uar

e fo

ot

Sav

ing

s

All

Heat

Cool

Fan

8

Parameter Sensitivity

Impact of Baseline on kWh/sf SavingsHeat Pump Heating

0

0.5

1

1.5

2

2.5

LPD/Density EconoChangeover

Glazing VentilationMinimum

Econo Max Combined

Parameter

kWh

/ s

qu

are

foo

t

9

Unit Change Minimizes Some ParameterskWh/square foot to kWh/ton

Impact of Baseline on kWh/ton SavingsHeat Pump Heating

0

200

400

600

800

1000

1200

1400

1600

LPD/Density EconoChangeover

Glazing VentilationMinimum

Econo Max Combined

Parameter

kWh

/ t

on

co

oli

ng

10

Approaches

• Site specific approaches– Custom analysis– Field based monitoring– Field-data driven model– Energy bill adjusted parametric tool– Parametric tool (calculator)– Simplified analysis (e.g. lighting spreadsheet)

• Deemed approaches– Matrix method / decision framework– Deemed savings (vintage, building type, climate)– Unit rebate– Expected value deemed method

11

Decision Framework Matrix• Focus on most sensitive parameters; group results

Relative Impact on Savings of Baseline Parameter ChangesInternal Load (L) and Minimum Ventilation (V)

-80%

-60%

-40%

-20%

0%

20%

40%

60%

80%

L+E+G+V-M-

L+V- V- L+ L-V- L+V+ V+ L- L-V+ L-E-G-V+M+

Ch

an

ge

in T

ota

l Sa

vin

gs

HP kWh Gas Heat kBtu

12

Matrix Results; Provide Savings TableInternal Density

Base Condition L- L= L+

Savings Matrix1.0 w/sf 1.8 w/sf

call center

Ventilation Minimum

V- 15% = - - -

V= 20% + = -

V+ 25% ++ + = Condition  Deemed Savings

From Gas Heat HP Heat

Table Gas Electric Electric

Above therms/ton kWh/ton kWh/ton

- - 36 214 433

- 27 182 557

= 44 363 985

+ 63 406 1336

++ 75 384 1479

13

Expected Value Deemed Savings

• Uses decision analysis methodology

– Typical for supply-side resource risk assessment

– Allows multiple parameters or influences

– Expert judgment can be applied to parameter probability distribution

14

Decision Tree for Expected Value Analysis

• Each influencing variable assigned states, each with– Probability of occurrence– Impact on savings

15

Parameter Savings Impact and ProbabilitiesParameter Sym Parameter variation   Factors for % of neutral savings

    in baseline BEFORE   Gas Heat HP Heat

    measure is installed Probability Gas Electric Electric

Internal Load L+ 1.8 LPD, 1.5 plug, 100 sf/person 20% 0.909 0.732 0.825

LPD = 1.8 w/sf LPD; eQuest defaults 45% 1.000 1.000 1.000

Density L- 1.0 w/sf LPD; eQuest defaults 35% 1.295 1.0003 1.212

Ventilation V+ 37.6 cfm/person 25% 0.659 1.027 1.162

Minimum = 31 cfm/person (typical) 50% 1.000 1.000 1.000

  V- 22.6 cfm/person 25% 1.282 0.979 0.806

Glazing G+ Low-e Argon 10% 0.861 0.929 0.869

Type = Double Bronze 40% 1.000 1.000 1.000

  G- Single Pane 50% 1.079 1.118 1.103

Economizer E++ B, double stage 5%   0.500 0.750

Changeover E+ C, single stage 30%   0.715 0.895

  = D or Snap Disk 45%   1.000 1.000

  E- Failed Economizer 20%   1.109 1.040

Economizer M+ 80% Max OSA 20%   0.897 0.964

Max OSA = 65% Max OSA 70%   1.000 1.000

  M- 50% Max OSA 10%   1.082 1.029

16

Simplified Interaction AdjustmentSimplified correction adjustment factors – limit adjustment

Parameter Combination Impacts and Probabilities

Higher quality as runs limited cf multiple regression

Compare to Neutral LPD/ Density

Econo Changeover Glazing

Ventilation Minimum Econo Max

Factor of All

Combined Run

Combination Adjustment

Plus 1.212 1.040 1.103 1.162 1.029 1.663 1.517 0.912

Minus 0.825 0.895 0.869 0.806 0.964 0.499 0.424 0.850

Simplified approach to adjust for combination impact from multiple parameter changes

    Parameter Combination  Probability Gas Heat HP Heat

    Gas Electric Electric

Interaction Lim+ All parameters increase 10% 0.865 0.924 0.912

factors avg()   25% 0.932 0.962 0.956

from full 1 Single Parameter change 30% 1.000 1.000 1.000

combination avg()   25% 1.215 0.825 0.925

  Lim- All parameters decrease 10% 1.430 0.649 0.850

17

Program-wide Savings DistributionProbability of different savings results based on range of baseline influences and combinations

18

Program-wide Expected Value (EV)

• Range of possible savings shown• Expected value represents program-wide results

19

Compare Matrix & Expected ValueMatrix Framework• Whole building

understanding needed• Multiple saving values• Site specific savings• Gaming inputs possible• Representation difficult

beyond two parameters• Multiple possible savings or

rebates make it difficult for contractor delivery

• 1944 runs required for 5 parameters needing 243 cases (3 states @, 2 climates, 2 heat types)

Expected Value• No need for information

outside discipline• Single saving value• Program-wide savings• Cannot game• Multiple parameters can be

considered• Single savings and rebate

amount makes contractor planning and marketing easy

• 96 runs required for 5 parameters needing 12 cases (3 states @, 2 climates, 2 heat types)

20

Further Research

• Current analysis is an “example” without full development; need – further develop expected values – research into extant building characterization data – “committee of experts” to develop probabilities for parameters

• Evaluate the differences in expected value and range of results for (kWh/unit) vs. (kWh/sf) vs. (kWh/ton) results.

• Use a regression model for high impact parameters to determine acceptability of simplified interactive method

• Find and test other software tools for expected value analysis • Explore past program impacts on long term results as economy of

scale takes hold, explore influence factor for projection• Develop a step-by-step Expected Value Deemed Savings method

for use by others in the region

21

Seeking Approval of Deemed Expected Value as Provisional Savings Method

• Goal: Accelerate new measure adoption with balance between savings accuracy and information needed

• Next step in premium ventilation is to develop pilot program approach with evaluation– For ease of contractor delivery, would prefer single

rebate approach– Seeking approval to use the expected value

method for two-year pilot approach– After pilot and evaluation, can update and

continue or seek different approach• May also consider use of Deemed Expected Value

for other measures that come before RTF

22

Questions?

Contact Information:Reid Hart, PEAssociate Director, Technical Research Portland Energy Conservation, Inc. 503-961-6142rhart@peci.orgwww.peci.org