HomeMy WebLinkAbout20261102Application Exhibit 2.pdf INTERMOUNTAIN GAS COMPANY
CASE No. INT-G-26-06
EXHIBIT No. 2 To APPLICATION
FURNACE BILLING ANALYSIS
•
• f
-
Intermountain Gas Furnace
Billing Analysis Memorandum
July 16 , 2026
Prepared for:
Intermountain Gas
Company
Prepared by:
QUALUS
100 Colonial Center
Parkway, Suite 400
Lake Mary, FL 32746
Intermountain Gas Furnace Billing Analysis Memorandum
Introduction
Qualus LLC (Qualus) conducted a billing analysis to assess the natural gas savings associated
with the Intermountain Gas Company's (IGC) residential furnace energy efficiency program
operating between January 2024 and May 2025. The goal of this analysis was to develop separate
unit energy savings (UES) estimates for IGC's new construction (NC), replace on burnout (ROB),
and early replacement(ER)95%AFUE furnace measures which can be applied to future program
cycles. These estimates were developed in response to a prudence request from the Idaho Public
Utilities Commission (IPUC) in Order No. 36797 to develop separate savings estimates for these
three different installation types based on metered billing data. For NC furnaces, billing data were
used to calculate participant heating load which was in turn used as an input to a standard
engineering algorithm to calculate gas savings. This method aligns with the International
Performance Measurement and Verification Protocol (IPMVP) Option A. Replace on burnout and
ER furnace savings were calculated using a treatment vs. control, Post Program Regression
(PPR) model consistent with IPMVP Option C. This memorandum presents Qualus' savings
analysis methodology, and UES estimates for each furnace installation type. The following
sections detail Qualus' methodology along with key definitions and final furnace savings
estimates.
Intermountain Gas Furnace Billing Analysis Memorandum
1 Savings Methodology
This methodology section is split into two key subsections: Measure Definitions and Technical
Approach. The measure definitions subsection differentiates the New Construction (NC), Replace
on Burnout (ROB), and Early Replacement (ER) furnace installation types while the technical
approach subsection outlines the specific steps Qualus took to calculate furnace gas savings.
Measure Definitions
Intermountain Gas Company's (IGC) 95% AFUE Natural Gas Furnace measure offers
incentives for residential customers that purchase and install qualifying high-efficiency natural
gas furnaces in an existing or new construction home. The measure requires the participant to
be served on the residential rate schedule and to use natural gas exclusively for space heating.
New Construction (NC), replacement on burnout (i.e., replacement at end-of-life), and early
retirement (i.e., retrofit) furnaces are all eligible for rebate provided the installed furnace has an
AFUE of 95% or greater. Per the Regional Technical Forum's (RTF's) Methodology for
Estimating Energy Efficiency Savings in the Northwest', measure installation types can be
differentiated as follows:
• Lost opportunity measures = measures replacing a product on failure or new load on
the system
o The RTF uses a current practice baseline for these measures, which is
represented by either the efficiency required by codes or market-average
efficiency, whichever is more efficient.
• Retrofit measures = discretionary measures that may be purchased at any time
o The RTF uses a pre-conditions baseline to estimate savings for these measures,
for example the existing insulation levels in a home if insulation is being upgraded.
o Early replacement measures = These are a subset of retrofit measures that
represent a voluntary change prior to the unit or system naturally requiring
replacement
• The RTF uses a dual baseline approach. The first baseline is a pre-
conditions baseline which is used for the remaining useful life of the
baseline equipment and the second baseline is a current practice
baseline that is applied to the remaining balance of the installed measure
life.
In the context of the IGC gas furnace measure, Qualus would categorize the three different
furnace installation types as follows:
• New Construction = Lost opportunity measure
lhttps://www.nwcouncil.org/media/filer public/75/22/75227368-dae7-4fd3-9cd2-48ebd103fc08/C2018-
17_Final_MethodologyEESavings_Paper.pdf
Intermountain Gas Furnace Billing Analysis Memorandum
• Replace on Burnout (i.e., "Replacement" in other IGC materials) = Lost opportunity
measure
• Early Replacement (i.e., "Retrofit" in other IGC materials) = Early replacement
subcategory of a retrofit measure
Having defined these three measure installation types in the context of the RTF's standards,
Qualus then evaluated available data based on conversations with IGC to determine the best
available savings methodology. For ROB and ER measures, IGC was able to source at least 12
months of billing data prior to and after measure installation for nearly all customers installing
furnaces. In contrast, given that NC furnaces were installed as or soon after households began
receiving gas bills, IGC could only reliably provide 12 months of billing data after measure
installation. Based on these data limitations, Qualus determined that a pre vs. post regression
analysis methodology could be used for ROB and ER, while a post-only analysis methodology
would be required for NC. More specifically, Qualus decided to employ a pre vs. post, treatment
vs. control PPR analysis methodology for ROB and ER analyses given that it a robust, industry-
standard regression analysis methodology that the IPUC has specifically requested for previous
furnace savings analyses. For the NC analysis, Qualus' decided to employ a heating load
analysis, as it integrates actual IGC customer gas usage data into an RTF- and IPMVP-approved
savings methodology for new construction energy efficiency measures.
Specific details on the models employed are provided in the following subsection, but at a high
level, a pre vs. post regression model estimates the impact of efficient furnace installation by
comparing an individual home's natural gas use in the pre period (i.e., the 12 months prior to
furnace installation) to natural gas use in the post period (i.e., the 12 months after furnace
installation). The treatment vs. control aspect of the regression methodology Qualus employed
adds an additional layer of complexity by incorporating usage matched non-participant customers
into the regression analysis to control for changes in energy usage unrelated to furnace
installation. Lastly, the PPR portion of this analysis (also referred to as a post-only model with
pre-period controls) is a specific type of pre vs. post, treatment vs. control regression. A PPR
analysis includes pretreatment energy consumption as an independent variable in the regression
to account for differences between subjects in their post-treatment consumption. Collectively, this
pre vs. post, treatment vs. control PPR approach is robust and reliable when applied to scenarios
where pre-period billing data is available.
For new construction scenarios, where no pre period exists, Qualus employed a post-only heating
load analysis methodology. In a post-only heating load analysis, a home's first year of billing data
after measure installation is analyzed and the furnace's heating load is isolated. The heating load
is the amount of natural gas energy used to heat that home throughout the year. This estimate,
along with the known AFUE of the installed furnace and a current practice baseline AFUE, is used
in an engineering algorithm to determine the home's energy savings.
1.1 Technical Approach
This subsection details the specific methodologies and equations Qualus employed to calculate
furnace gas savings for NC, ROB, and ER installation scenarios.
Intermountain Gas Furnace Billing Analysis Memorandum
1.1.1 General Billing Analysis Steps
For the purposes of this analysis, a household is considered a treatment household if it has
received a program incentive. To isolate measure impacts, treatment households are eligible to
be included in the billing analysis if they installed only one measure during the 24-month analysis
period (i.e., the pre period and post period combined). Isolation of individual measures is
necessary to provide valid measure-level savings. Households that installed more than one
measure may display interactive energy savings effects across multiple measures that are not
feasibly identifiable. Therefore, instances where households installed isolated measures are used
in the billing analyses. In the following subsections, Qualus provides additional details on the data
preparation steps and regression analyses used in this evaluation.
Data Collection
Prior to running the regression billing analyses, Qualus collected the following raw data:
1. Monthly billing data from January 2023 — March 2026 for program participants who only
installed a 95% AFUE furnace between January 2024 and May 2025
2. Measure and home characteristics data, including pre and post furnace AFUE, furnace
capacity (Btu/h) and household zip code
3. Monthly billing data for a cohort of 20,000 IGC non-participants who had not installed other
energy efficiency measures during the relevant 51-month period
4. National Oceanic and Atmospheric Administration (NOAA)weather data between January
1, 2023, and May 31, 2026)
5. Typical Meteorological Year (TMYx) data
Billing and weather data were obtained for the entire evaluation period and for one year prior to
measure install dates (beginning January 1, 2023). Qualus mapped customer zip codes to the
nearest weather station and extracted temperature data from those weather stations via NOAA's
Global Historical Climatology Network hourly database.' TMY weather stations were assigned to
each zip code in a similar manner, and TMYx weather data was used to extrapolate savings to a
long-run, 30-year average weather. After collecting these key datasets, Qualus conducted the
following data preparation steps.
Data Preparation
In June 2026, IGC shared raw data with Qualus that contained monthly bills from January 2023
through March 2026 for 7,812 customers installing gas furnaces between January 2024 and May
2025. The raw tracking data that Qualus received differentiated between "equipment
replacement" (i.e., either ER or ROB) and "new construction" and also contained a "prior
operational" column that allowed Qualus to distinguish between ER and ROB for equipment
2 Typical Meteorological Year(TMY)is a set of weather data synthesized by NREL representing the long-term climatic conditions for
specific locations.The TMYx dataset was constructed using historical weather data collected from 1976-2005,typically obtained from
weather stations in the vicinity of the location of interest.
3 https://www.ncei.noaa.gov/products/global-historical-climatology-network-hourly
Intermountain Gas Furnace Billing Analysis Memorandum
replacement furnaces. Of note, if IGC's tracking data did not include information on whether a
non-New Construction furnace was previously operational (i.e., the"prior operational"column was
"NA"), Qualus assumed the measure was ER not ROB. In addition, Qualus shared monthly billing
data for a cohort of 20,000 non-participants who, to IGC's knowledge, had not installed energy
efficiency measures during the relevant 51-month period. The following table outlines customer
counts by furnace installation type in the original billing data IGC shared with Qualus.
Table 1-1. Intermountain Gas January 2024— March 2025 Furnace Customer Counts
Furnace Installation Type Analysis Group IGC Customer Count
New Construction Treatment 4,874
Replace on Burnout Treatment 1,719
Early Replacement Treatment 1,219
Non-Participant Control 20,000
As highlighted in Figure 1-1, the distribution of zip codes across Idaho was relatively similar
between different furnace installation type groups. New construction households seemed to be
slightly more concentrated in urban zip codes, while ER and ROB furnace installations were
spread more broadly across IGC's service territory in Southern Idaho.
Figure 1-1. Intermountain Gas Furnace Program Participation by ZIP Code and Install Type
Early Replacement New Construction Replace on Burnout
Furnace Installations
soo
400
300
200
100
SZ) �!
The following steps were taken to prepare the billing data for regression analyses:
1. Exclude any bills missing address information.
2. Clean billing data and tracking data address information by converting addresses to
lowercase and removing whitespace.
3. Subset billing data to only include customers included in the tracking data.
4. Remove any bills missing fuel type/Unit of Measure (UOM).
5. Remove any bills missing usage, billing start date, or billing end date.
Intermountain Gas Furnace Billing Analysis Memorandum
6. Merge baseline and efficient equipment efficiency as well as household zip code from the
tracking data with the billing data by premise ID.
7. Remove any duplicates from tracking and billing data.
8. Calendarize bills; this involves recalculating bill dates, usage, and total billed days such
that bills begin and end at the start and end of each month.
9. Obtain weather data from nearest NOAA weather station using 5-digit zip code per
household.
10. Compute Heating Degree Days (HDD)for a range of setpoints. The Evaluators employed
a setpoint of 650F for HDD calculations.
11. Subset billing data to only include customers with one measure installed during the
analysis years and merge customer measure install dates with billing data (to define pre-
and post-periods).
12. Conduct propensity score matching (PSM) with replacement to match ROB and ER
customers to non-participant customers based on pre-period energy usage. After
matching customers, verify the statistical similarity of the treatment and control groups via
Student's t-tests.
13. Subset customer billing data to only include 12 months of pre-period and 12 months of
post-period billing data for each treatment and control customer.
14. Remove customers with incomplete post-period bills (<10 months total).
15. Remove customers with incomplete pre-period bills (<10 months total); this condition was
ignored for the NC cohort as no pre-period bills were required for the post-only analysis.
1.1.2 Regression Models
After preparing the billing data as detailed in the previous section, Qualus conducted regression
analyses to identify the impact of furnace installation on customer natural gas usage as described
in the following subsections.
Treatment vs. Control PPR Model
The following equation displays the model specification to estimate the average daily savings
associated with furnace installation. This model specification was run once for ROB furnaces and
once for ER furnaces.
Equation 1-1: Post-Program Regression (PPR) Updated Model Specification
ADCit = ao +f31(Treatment)i + R2 (PreUsage)i +f33 (PreUsageSpring)i
+f34(PreUsageSummer)i +f35(PreUsageFa11)i +f36(PreUsageWinter)i
+f37(Month)t +f38(Month x PreUsage)it +fl9(Month x PreUsageSpring)it
+f310(Month x PreUsageSummer)it +fjll(Month x PreUsageFall)it
+f312(Month x PreUsageWinter)it +fl13(HDD)it +F'14(Treatment x HDD)it
+ £it
Intermountain Gas Furnace Billing Analysis Memorandum
Where:
• i= the ith household
• t= the first, second, third, etc. month of the post-treatment period
• ADCit = Average daily usage in therms for reading t for household i during the post-
treatment period
• Treatmenti = A dummy variable indicating treatment status of home i
• Montht = Dummy variable indicating month of month t
• PreUsagei = Average daily usage across household is available pre-treatment billing
reads
• PreUsageSpringi = Average daily usage in the spring months across household is
available pre-treatment billing reads
• PreUsageSummeri = Average daily usage in the summer months across household i s
available pretreatment billing reads
• PreUsageFalli = Average daily usage in the fall months across household is available
pretreatment billing reads
• PreUsageWinteri = Average daily usage in the winter months across household is
available pre-treatment billing reads
• HDDit = Average heating degree days (with 65 'Fahrenheit setpoint) during
period t at home i
• -it = Customer-level random error
• ao= The model intercept for home i
• F'1-14 = Coefficients determined via regression
The coefficient Ql represents the average change in consumption between the pre-period and
post-period for the treatment group and Q14 represents the change in weather-related daily
consumption in the post-period between the groups. Average savings per regression analysis
were estimated by extrapolating the Rl coefficient and the Q14 coefficient with Typical
Meteorological Year (TMYx) HDD data. The equation below displays how savings were
extrapolated for a full year utilizing the coefficients in the first regression model and TMYx data.
Equation 1-2: Regression Savings Extrapolation
Annual Therm Savings = Ql * 365.25 +fllo * TMY HDD
Heating Load Analysis
A heating load estimate provides a way to calculate therms savings for space heating measures
using TRM-based engineering equations, under various scenarios for pre- and post-efficiency
factors and space heating capacities. This approach complies IPMVP Option A, as maintained by
the EVO with sponsorship by the U.S. DOE'. It is often used to calculate deemed savings for new
construction measures.
4 Core Concepts: International Measurement and Verification Protocol. EVO 100000—1:2016,October 2016.
Intermountain Gas Furnace Billing Analysis Memorandum
Heating Load Estimation Methodology
The first step in heating load estimation is calculating TMYx weather normalized average daily
consumption. To do so, customer-specific regressions are run to determine the effect of daily
HDD on average daily consumption. This is a straightforward regression of the following form.
Equation 1-3. Heating Load Regression
ADCi = ao +f31(HDD)i
Where:
i= the ith household
• ADCi = Average daily usage in therms for household i during the post-treatment period
• HDDi = Average heating degree days at home i
• R1 = Coefficient determined via regression
This regression is run separately for each customer to determine F'1, the impact of HDD on
average daily consumption (i.e., the change in therms usage per HDD). From there, F'1 multiplied
by HDD is subtracted from ADC and fl1multiplied by TMY_HDD is added back to ADC to calculate
TMYx weather normalized average daily consumption. The actual HDD attributable therms usage
is subtracted from average daily consumption, and the TMY—HDD attributable therms are added
back in, as outlined in the following equation.
Equation 1-4. Normalized Average Daily Consumption
NADCi = ADCi -F'1 * (HDD)i + N1 * (TMY_HDD)i
Where:
• i= the ith household
NADCi = TMYx normalized average daily usage for household i during the post-period
• F'1 = Customer-specific therms usage per HDD
• ADCi = Average daily usage for household i during the post-treatment period
• HDDi = Average heating degree days at home i
TMY_HDDi = Average TMY heating degree days at home i
Once TMY normalized average daily usage (NADC) is calculated, Qualus performed the following
calculations to determine customer heating load savings:
• Calculated post-period normalized baseload usage (therms) for each participant, where
baseload is equal to average NADC in June, July, and August.
• Calculated TMYx weather normalized average daily heating load (NHL) for each
participant in the post-period by taking the difference between NADC and normalized
baseload usage.
• Set any negative heating loads to zero(assumed to be deviations from average baseload).
Intermountain Gas Furnace Billing Analysis Memorandum
• Merged average NHL back into tracking data.
With weather normalized heating load included in the tracking data for customers installing NC
furnaces, the Evaluators calculated annual therms savings as follows. This equation is outlined in
page 43 of the Arkansas TRM version 9.1.5 but is an industry-standard engineering algorithm for
determining gas furnace savings across the nation.
Equation 1-5. Heating Load Annual Savings
1 1
Savings = NHL * 365.25
BaseEf f InstallEf f
Where:
• Savings = Annual therms savings associated with gas space heating measure
• NHL = Average TMYx normalized daily heating load usage for IGC customers installing a
NC furnace in the post period. Per Qualus' analysis the average normalized daily heating
load was 1.62 therms which equates to an average annual heating load of 593.18 therms.
365.25 = Average number of days per year
• BaseEf f = The baseline efficiency standard for furnaces at the time of measure
installation. Intermountain Gas' tracking data included baseline AFUE (i.e., the efficiency
of the pre-retrofit furnace), which Qualus used to calculate an assumed current practice
baseline efficiency value. Of customers with sufficient data to be included in the billing
analysis, 1,895 had data on baseline AFUE. Qualus used these customers' data to
calculate an average baseline AFUE of 87.6%. While this value is slightly lower than the
88.5% baseline AFUE that the RTF's Furnace UES workbook' assumes, Qualus
determined that it was most appropriate to reference IGC-specific data for this savings
calculation, as opposed to the Pacific Northwest-wide RTF estimates.
• InstallEf f = The installed furnace AFUE per tracking data. New construction customers
had an average efficient furnace AFUE of 96.17%.
The following section details IGC furnace gas savings results by installation type as well as key
conclusions from this analysis.
5 https://apsc.arkansas.gov/wp-content/uploads/AR_TRM_V9.1_Volume_1_2_and_3_on_8-31-22.pdf
s https://rtf.nwcouncil.org/measure/residential-gas-furnaces/
Intermountain Gas Furnace Billing Analysis Memorandum
2 Results and Conclusions
As detailed in the previous section, Qualus employed two different methodologies to calculate the
gas savings associated with IGC's NC, ROB, and ER furnace installations. New construction
savings were calculated using a post-only heating load analysis, while ROB and ER savings were
calculated using pre vs. post, treatment vs. control post program regression (PPR) analyses.
Results from these analyses are presented below.
Table 2-1. Intermountain Gas January 2024— March 2025 Furnace Billing Analysis Results
F Treatment Control Estimated p- Adjusted
Installation Model 7 Customers in Customers Annual Savings
Furnace
Type Analysis in Analysis (therms) value R-squared
New Heating Load 2,209 N/A 59.97 N/A N/A
Construction Analysis
Replace on Post Program 482 467 55.27 <0.001 0.923
Burnout Regression
Early Post Program 735 708 62.96 <0.001 0.932
Replacement Regression
This billing analysis suggests that gas furnaces rebated by Intermountain Gas and installed
between January 2024 and March 2025, were associated with savings between 55.27 and 62.96
therms annually depending on the furnace installation type. Replace on burnout furnaces were
found to be associated with the lowest annual gas savings with an average of 55.27 therms per
household, followed by NC furnaces at 59.31 therms, and ER furnaces at 62.96 therms. The
savings values calculated via PPR analyses were statistically significant at the 0.05 alpha with
ROB and ER analyses yielding a p-value <0.001. Both these treatment vs. control regression
analyses also yielded an adjusted R-squared value above 0.920, suggesting an excellent model
fit. The following table outlines the average annual usage in therms by furnace installation type
during the pre- and post-period.
Table 2-2. Average Annual Household Pre- and Post-Period Usage by Furnace Install Type
Pre-PeriodFurnace Installation Type Average .. Usage(therms)
New Construction' N/A 595.00
Replace on Burnout 841.74 780.65
Early Replacement 890.63 828.32
These findings seem to suggest that participating NC households use substantially less gas than
ROB and ER households, while ROB households use slightly less gas than ER households. This
difference in gas usage could be explained by a variety of factors. The most likely driver is that
NC households are built to stricter energy codes, requiring less energy to be used to keep them
heated during cold season. Furthermore, per Figure 1-1, NC households seem to be in more
urban areas than ROWER households. Households in more urban areas could have slightly
warmer winters due to the urban heat island effect, which could contribute to less gas being used
' New construction pre-period usage is set to"N/A" because almost no pre-period billing data was provided
for NC households.
Intermountain Gas Furnace Billing Analysis Memorandum
on household heating. Similarly, differences in the geography and demographics of ROB and ER
households could contribute to the difference in annual household gas usage highlighted above.
In addition to differences in annual household usage, this analysis also identified slight differences
in estimated furnace gas savings. As highlighted in the previous paragraph, new urban NC
households may have fewer gas appliances than older more rural ER/ROB households, which
could contribute to the higher percentage savings for NC detailed in the following table.
Meanwhile, differences in energy use patterns and/or the geography of ROB and ER households
may have contributed to slight differences in savings between these two equipment replacement
installation types. Furthermore, uncontrolled differences between these two groups could have
contributed to differences in savings. For example, a small proportion of ROB customers could
have installed additional gas appliances along with their gas furnace (unbeknownst to IGC) or
increased their household occupancy, which could have contributed to a reduction in ROB post-
period savings as compared to ER households. Ultimately, the estimated annual savings across
all three furnace installation types are relatively similar, suggesting 95% AFUE gas furnaces in
IGC's service area are an effective gas-saving measure.
In Table 2-1 and following table, Table 2-3, regression-related metrics like p-value and 95%
confidence interval (CI) lower bound are set to "N/A" for the NC analysis due to the fact that
savings were calculated via an engineering equation as opposed to a PPR analysis. The following
table outlines the annual percentage savings and 95% Cl lower and upper bounds of these billing
analyses.
Table 2-3. Idaho Furnace PPR Billing Analysis Percentage Savings and 95% CI Bounds
UpperFurnace Model Estimated Annual Percentage 95%CI Lower 95%Cl
Installation Type Savings(therms) Savings Bound(therms) Bound(therms)
New Heating Load 59.97 10.08% N/A N/A
Construction Analysis
Replace on Post Program 55.27 6.81% 47.61 62.93
Burnout Regression
Early Post Program 62.96 7.44% 56.60 69.33
Replacement Regression
The relatively large sample sizes of the ROB and ER analysis cohorts seem to have contributed
to a relatively narrow 95% confidence interval for both analyses. Across all three furnace
installation types, percentage gas savings were also consistent, ranging from 6.81% to 10.08%.
The following figure outlines the average monthly usage of customers included in billing analyses
by ZIP code and furnace installation type (i.e., ER, NC, and ROB).
Intermountain Gas Furnace Billing Analysis Memorandum
Figure 2-1. Customers in Billing Analyses Average Monthly Usage by ZIP Code and Install Type
Early Replacement New Construction Replace on Burnout
Avg Monthly Usage(therms)
90
70
50
an
All IGC customers included in billing analyses, who installed furnaces between January 2024 and
May 2025, resided in Southern Idaho along relatively similar latitudes. Furthermore, average
monthly customer usage by zip code was relatively consistent, with a mean of 63.55 therms, a
median of 63.13 therms, and a standard deviation of 15.57 therms. This geographic consistency
likely contributed to the relatively consistent gas savings across all three evaluations, with annual
savings only varying by 7.69 therms between ROB and ER.
Ultimately, this analysis suggests that as of 2024 and early 2025, furnace installations by IGC
residential customers yielded gas savings between 55.27 and 62.96 therms annually. These
annual savings estimates are substantially higher than the single-family household deemed
savings estimates between 36.3 and 44.9 therms included in the RTF Residential Gas Furnace
UES workbook v3.1.$ Based on this analysis, in the future it may be most pragmatic for IGC to
calculate furnace gas savings based on a billing analysis as opposed to a deemed savings
method per RTF data. Given that baseline conditions and residential gas usage have the potential
to dramatically change as customers install increasingly efficient measures, Qualus would
recommend IGC reassess furnace savings on a regular basis to ensure the continued accuracy
of these LIES estimates.
8 https://rtf.nwcouncil.org/measure/residential-gas-furnaces/