Heat Pump Research Project Sponsored by the Heat Pump Working Group April 5, 2005

CheckMe! ® Billing Analysis Project Goal: Evaluate impact of refrigerant charge separately from CheckMe! ® tune up.

Sample Distribution and Attrition Raw Data Received Complete Data Analysis Set Participants Non-Participants Participants were dropped if any one of several factors was noted: Customer turnover Otherwise incomplete billing records Participant record included a comment of “withdrawn”.

Analysis Methodology Weather-normalized pre- vs. post- consumption methodology Same difficulties using PRISM ® as in C&RD analysis (wood heat, missing data, a/c). Used a simplified engineering model, EZSim ® Initial plan to use “Test Only” group as control for comparison failed due to similar savings Group of 80 non-participants used as comparison instead

Fit of Model to Bills Example

Analysis Results: Savings Estimates

Analysis Results: NAC Difference (kWh/hr) GroupMeanStd Dev n90% C.I.t-test Refrigerant Adjust3103, Test only5082, Adjust plus repair5532, , Repair only2652, , All Participants4462, Note: Grayed rows indicate statistical significance.

Distribution Profile

Control Group Consumption Note: n= Mean NAC13,08413,62912,913 SD 5,330 5,231 5,415 90% Conf Interval Annual Change Average Change 86

Conclusions EWEB’s CheckMe! ® program provides an effective average savings of 360 annual kWh. These savings are small but statistically significant. There are no additional savings associated with refrigerant adjustment. The savings appear to result from improvements made by the technician during the course of testing and examining the unit.

C&RD/ConAug Billing Analysis Project Goal: Evaluate realization of anticipated savings from C&RD and ConAug heat pump programs in various climates.

Sample Distribution & Attrition GroupRegionExpectedReceivedCompleteAnalyzed TreatedTriCities NW, Kitsap Central OR Coast Portland C&RD1,5211, Non-C&RD Total2,4881,7861,5411,268 UntreatedTriCities NW, Kitsap Central OR Portland Total

Reasons for High Attrition Billing procedural change at utility Data collection change (Energy Trust) Partial vacancy Occupant turnover Incomplete data

Billing Analysis Methodology PRISM ® Regression – Proved Impractical  Wood Heat  Air Conditioning  Missing data Multivariable Regression – Successful  Regression run over range of balance temperatures  Balance temperature optimized for both years together  Individual case review for outliers

Heating Only Temperature Regression Example

Heating / Cooling Temperature Regression Example

Billing Analysis Results Statistically significant (with 90% C.I.) Normalized annual data robust Space heat, cooling and base load disaggregation less robust Little difference between C&RD & non-C&RD groups Tri-Cities represents the only cooling zone

Normalized Annual Consumption (NAC) By RegionBy Equipment Type

Savings Estimates by Region NAC90% C.I.Est. Space Heat Use 90% C.I. TriCities3, , NW, Kitsap5, , Coast3, , Portland4, , C&RD4, , Non-C&RD3, , Total4, ,653226

Savings Estimates by Equipment Type System TypeNAC Savings (kWh/yr) 90% Confidence Interval Heat Pump4, Forced Air Furnace With Central AC 2, Forced Air Furnace Without Central AC 4, Zonal5,3621,023 All Sites4,

Disaggregation

Normalizing Variables CityHDDCDD Richland4, Pt. Angeles5,67128 Hoquium5,16431 Astoria5,11618 Portland4, Not C&RD4, Weighted Average C&RD Group5, Total Study5, Cool 15, Cool 34, C&RD Cool 15, C&RD Cool 35, SystemSF90% CI Heat Pump2, FAF w/CAC2, FAF no CAC2,10674 Zonal2, All Sites2,01245 Vintage BinSvgs (kWh/yr) 90% CI Pre 19884, , Post 19934,1401,080 Average4,263294

Control Group Consumption Note: n = Mean NAC20,16020,46719,932 SD 8,712 8,096 8,036 90% Conf Interval Annual Change Average Change 114

Realization Rates by Program Year Base Prog Year AreanGross NAC kWh/yr Net NAC kWh/yr 90% CIAntici- pated kWh/yr RR 2003Cool 15184,6984, ,28863% Cool 33183,7953, ,81554% All C&RD8364,3544, ,10860%

Realization Rates by Region

Conclusions Study population was smaller than expected due to inability to obtain all the requested data. However, the study group of 1,022 cases is large enough for a relative precise estimate of mean savings. Breakdown of estimates into subsets of the study population suffers from missing information and small sample size for sub-categories. Best estimate of net savings is 4,149 kWh per year and is highly significant. Best estimate for only the C&RD participants is 4,240 kWh per year.

Conclusions (continued) Savings are approximately 60% of the predicted amount. There is ambiguity because specific ex ante savings estimates are not available. There is little difference in overall savings between climate zones. However, Cooling Zone 3 exhibits more cooling consumption and savings, as would be expected. There are significant differences between the type of equipment that was replaced. These differences do not show higher savings for furnace over zonal, as would be expected. The equipment codes also appear to be highly unreliable. Thus, no conclusions are suggested based on equipment type.

Purdue University Lab Tests Project Goal: Establish the charge in heat pump performance resulting from sub-optimum refrigerant charge and air handler flow.

TXV and FEO Comparison Biggest impact in higher temperature bins  47° shows a 10% degradation with FEO & 5% with TXV Much smaller impact in lower temperature bins. At some temperature point, performance is actually degraded by TXV and lower charge variation

TXV vs. FEO System at 47° F Comparison of COPs of TXV system and FEO system at 47 F outdoor temperature, 1300 indoor CFM and different charges

TXV vs. FEO System at 35° F Comparison of COPs of TXV system and FEO system at 35 F outdoor temperature, 1300 indoor CFM and different charges

TXV vs. FEO System at 17° F Comparison of COPs of TXV system and FEO system at 17 F outdoor temperature, 1300 indoor CFM and different charges

Impact of Different Charges and Temperatures on COP COP of heat pump using a TXV at 1300 CFM indoor air flow rate, different charges and outdoor temperatures

Impact of Frost Formation on Heating Capacity Degradation of heating capacity due to frost formation

Impact of Frost Formation on COP Degradation of COP due to frost formation

Lab Tests: Preliminary Results Degradation of COP for TXV and FEO will be less than that seen at 47°. Impact of defrost is potentially significant, although low charge seems to improve performance C D tests more inclusive, but results seem to suggest higher values than those used in tables.