20240814IPC to Staff 35 - Attachment 1 - Load Forecasting Methodology.pdf

Response to Staff Request No.35-Attachment 1
IDAW
&�,.POMR®
An IDACORP Cnmpiny
Load Forecasting Workpaper
Workpaper
Case No. IPC-E-24-07
Page 1 of 7
Load Forecasting

Idaho Power Company Load Forecasting Workpaper
Table of Contents
Tableof Contents.......................................................................................................................... ii
Listof Tables ................................................................................................................................. ii
Listof Figures................................................................................................................................ ii
2024 Test Year Sales Forecast....................................................................................................... 3
DataSources ........................................................................................................................... 3
Weather Normalization........................................................................................................... 3
Energy Efficiency and Electrification .......................................................................................4
ForecastModels......................................................................................................................4
Residential.........................................................................................................................4
Commercial and Industrial ................................................................................................ 5
Irrigation............................................................................................................................ 6
On-Site Generation ........................................................................................................... 6
Forecast Preparation............................................................................................................... 7
List of Tables
Table 1
Residential and Commercial Weighted Normal Weather used in Sales Forecast.......................3
List of Figures
Figure 1
ResidentialData Flow................................................................................................................. 5
Figure 2
Commercial and Industrial Data Flow ........................................................................................6
Case No. IPC-E-24-07
Page 2 of 7
Load Forecasting
Page ii

Idaho Power Load Research and Forecasting Workpaper
2024 Test Year Sales Forecast
Data Source,
The 2024 test year sales forecast uses economic data primarily sourced from Moody's
Analytics and Woods & Poole Economics. The national, state, metropolitan statistical area
("MSA"), and county economic and demographic projections are tailored to Idaho Power's
service area using an in-house historic economic database. Specific demographic projections
are also developed for the service area from national and local census data. Additional data
sources used to substantiate this economic data include, but are not limited to, the Idaho
Department of Labor, Construction Monitor, and Federal Reserve economic databases. All
economic data has been updated to the most recent conditions and assessed in Q3 of 2023.
The Company primarily used National Oceanic and Atmospheric Association ("NOAA") data to
obtain historic weather information within Idaho Power's service area.
Weacher Nurmalizatioi.
In general, Idaho Power assumes normal temperatures and precipitation over a 30-year
meteorological measurement period. Forecasts for the customer categories that are weather
sensitive include the impact from weather conditions to assess the most probable or likely
outcome. The sales and load forecast uses 30-year normal cooling degree days - base 65
("CDD") and heating degree days— base 65 ("HDD"). These are weighted by Idaho Power's
service area customer counts to find the weighted weather normal as defined below.
Table 1
Residential and Commercial Weighted Normal Weather used in Sales Forecast
Residential Commercial
HDD CDD HDD CDD
Jan 1,055 0 1,062 0
Feb 810 0 817 0
Mar 647 0 654 0
Apr 453 2 459 2
May 226 35 231 33
Jun 74 124 77 120
Jul 8 342 9 331
Aug 12 296 14 286
Sep 103 95 109 89
Oct 413 7 421 7
Nov 771 0 777 0
Dec 1,054 0 1,061 0
Case No. IPC-E-24-07
Page 3 of 7
Load Forecasting
Page 3

Idaho Power Load Research and Forecasting Workpaper
Energy Efficiency and Electrification
Assumptions for both energy efficiency and electrification are incorporated into the 2024
sales forecast. Energy efficiency influences on past and future load consist of utility programs,
statutory codes, and manufacturing standards for appliances, equipment, and building
materials that reduce energy consumption. Idaho Power modifies its forecasting models to
capture the impact of economic achievable potential. Guidance on these amounts is
benchmarked to data provided by Applied Energy Group ("AEG") used for validating forecast
model output. The approach on how the energy efficiency data is included differs by customer
segment model. As a note, demand response programs are not incorporated into the sales
forecast.
For transportation electrification, the 2024 test year sales forecast includes an update of the
impact of plug-in electric vehicles ("PEV") on system load to reflect the future impact of this
relatively new and evolving source of energy use. Electric vehicle ("EV") consumer adoption
rates in Idaho Power's service area continue with technological advancements. Improving
attributes of vehicle range, refueling time, and charging availability continue to improve the
competitiveness of these vehicles compared to non-electric models. Idaho Power receives
detailed registration data from the Idaho Transportation Department, and the forecast is
conformed to that information. For building electrification, the load forecast assumes a certain
level of residential heat pumps and up to date usage information on customers that have
moved forward with electrification efforts. The 2024 sales forecast includes the specifications
on home electrification equipment saturations from a 2022 Idaho Power end-use study.
Forecast Models
In developing the sales forecast, unique models are used for different customer groups.
These include residential, commercial and industrial, irrigation, and on-site generation
customers. For each customer group, the source of information used is slightly different
depending on the homogeneity or other characteristics of the customer class.
Residential
The flow of data for the residential customer class is depicted below. A distinguishing
trademark of the residential sales forecast relative to other customer classes is the residential
group is rather homogeneous, as such it is conducive to using a statistically adjusted end-use
framework, in that usage by appliance or energy use type is categorized into heating, cooling,
and non-weather impacts and run through an ordinary least squares regression model to
explain the variation in historical data.
Case No. IPC-E-24-07
Page 4 of 7
Load Forecasting
Page 4

Idaho Power Load Research and Forecasting Workpaper
Figure 1
Residential Data Flow
Residentialesicle�71 ,
Customer Customer
..-
1 Sales Architecture=SAE Framework 1
Training Start=2008
Forecast Dependent Variable=
Monthly Sales
rRe si
•e�nti lal',,
ientia
Customer
ForecastForecastCustomer
Commercial and Industrial
The flow of data for the commercial and industrial customer class is depicted below. This
class is rather heterogeneous, and as such is further segmented into several ordinary least
square models in an effort to isolate economic variables unique to each segment and refine any
correlation between the regressors and the error terms, a primary qualifier for ordinary least
square models. This segmentation enables unique regression models and associated
explanatory variables resulting from the categorization to estimate the relationship between
dependent historical electricity sales and independent variables such as corporate earnings,
economics, price, technological, and demographics. Individual forecasts of customers with
unique energy service agreements that exceed 20 megawatts ("MW") in size are provided to
the Company from the customer.
Case No. IPC-E-24-07
Page 5 of 7
Load Forecasting
Page 5

Idaho Power Load Research and Forecasting Workpaper
Figure 2
Commercial and Industrial Data Flow
Arch tecture=Econometric
Tyarning Stare=early 1990-2000
Dependent Variable=
Annual SalesIPC
Indust
ForecastManu
Model Unique
ow MW_
♦``�
Weather Industrial
Data Models
IPC Aggregate
Economic • ®
Data
Forecastand Comm
Manu
UtilityData A L
Comm
Large
Industrial Model
1
1
1
1
ly
ms Commercial
Ser%nce Sales
Model Forecast
Irrigation
An individual model is used for the irrigation class. Service under this category is applicable
to energy supplied to agricultural-use customers for operating water pumping or water-delivery
systems to irrigate agricultural crops or pasturage. As the irrigation category is highly seasonal
and weather sensitive, a unique model is used to capture the unique characteristics of this
segment relative to the commercial and industrial sectors. In the highly seasonal irrigation
sector, over 97 percent of the annual energy is billed during the six months from May through
October, and nearly half of the annual energy is billed in just two months: July and August.
During the summer, hourly irrigation loads can constitute nearly 900 MW. In a normal July,
irrigation pumping accounts for roughly 25 percent of the energy consumed during the hour of
the annual system peak and nearly 30 percent of the energy consumed during July for general
business sales. The irrigation forecast model is unique from the other classes in that it is a single
ordinary least squares model. That model flows with a more straightforward design as weather
data in the form of growing degree days base 50 ("GDD50"), precipitation data, an economic
variable for farm product producer price index (from Moody's Analytics), and customer growth
are used as estimators for irrigation billed sales.
On-Site Generation
The installation of generating and storage equipment at customer sites will cause the
demand for electricity delivered by Idaho Power to be reshaped throughout the year.
Therefore, the sales forecast reflects the impact of continued growth in on-site generation
Case No. IPC-E-24-07
Page 6 of 7
Load Forecasting
Page 6

Idaho Power Load Research and Forecasting Workpaper
customers. Existing on-site generation customers' billing histories were matched to the billing
histories at that site prior to becoming an on-site generation customer. The resulting average
monthly impact per customer (in kilowatt-hours ("kWh")) was then multiplied by forecasts of
the Schedule 6, 8, and 84 residential, commercial, and irrigation customer counts to estimate
the future energy impact on the sales forecast. The forecast of on-site customers serves as a
function of historical trends and pending regulatory actions. As such, the "delivered-only"
portion of these on-site customers was also analyzed and forecasted for non-legacy customers.
This was done by finding the delivered-only channel of a site's meter for on-site generation
customers and extrapolating that to the forecasted customer growth of on-site generation.
Forecast Preparation
In order to determine forecast customer levels for the 2024 test year, a customer forecast is
prepared based on the primary customer classes described above. For residential, the number
of customers is a function of the number of new service area households as derived from
Moody's Analytics' forecast of housing stock and demographic data for the Company's service
area. From there the number of commercial customers is a function of the number of new
residential customers being added. The industrial and irrigation customer forecasts are based
on a trend of operating area customer counts that are allocated to months and then summed
to monthly class totals.
The final step in determining forecast sales for the 2024 test year is to further categorize
the primary customer classes described above by rate schedule and jurisdiction. The splitting of
the sales and customer count forecasts is done by proportioning the forecast based on
historical actual sales and customer additions.
Case No. IPC-E-24-07
Page 7 of 7
Load Forecasting
Page 7