200403161st Response of ID Irrigation Pumpers to ID Power.pdf

Randall C. Budge, ISB No. 1949
Eric L. Olsen, ISB No. 4811
RACINE, OLSON, NYE, BUDGE &
BAILEY, CHARTERED
O. Box 1391; 201 E. Center
Pocatello, Idaho 83204-1391
Telephone: [phone]
Fax: [phone]
'" ,
..r C-T,:.Lc. 'I,-~
':!L.
IT)
1'-
,~-
')n f\!. ~~
.' ;';'
\ ,i"
;,\
h ,
.. -...-,..'
JTiL1 r;:~~) CDhi'1:SSIOtt
Attorneys for Intervenor
Idaho Irrigation Pumpers Association, Inc.
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
IN THE MATTER OF THE APPLICATION OF
IDAHO POWER COMPANY FOR AUTHORITY)
TO INCREASE ITS INTERIM AND BASE
RATES AND CHARGES FOR ELECTRIC SERVICE.
CASE NO. IPC-O3-
IDAHO IRRIGATION PUMPERS ASSOCIATION, INC.S RESPONSE
TO FIRST INTERROGATORIES AND PRODUCTION REQUEST
OF IDAHO POWER COMPANY
COMES NOW Idaho Irrigation Pumpers Association, Inc., through counsel, and hereby
responds to the First Interrogatories and Production Request ofIdaho Power Company.
REQUEST NO.1: Please provide copies of all data and analyses used to create Figure 6 in
Mr. Yankel's testimony.
RESPONSE 1: Please see the workpapers originally filed by the Irrigators in this case.
Specifically see the Excel file named "Exhibit 26 1998-2003" under the Tab labeled "all less fuel
& PP" in cells C167 through BV168.
REQUEST NO.2: Please provide all data and analyses used to create Exhibit 301.
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 1
RESPONSE 2: Please see the workpapers originally filed by the Irrigators in this case.
Specifically see the Excel file named "Exhibit 16 1998-2003" under the Tab labeled "all less fuel
& PP" in cells BK 167 through BV 168.
REQUEST NO.3: What is the standard deviation of monthly values contained in Figure
RESPONSE 3: Please see the workpapers originally filed by the Irrigators in this case.!
Specifically see the Exfcel file named "Exhibit 26 1998-2003" under the Tab labeled "all less fuel
& PP" in cells F215 through N232 for the statistical results that were provided by Excel with respect
to this analysis.
REQUEST 4: Please provide the r-squared value for the linear fit represented in Figure 6.
RESPONSE 4: Please see the workpapers originally filed by the Irrigators in this case.
Specifically see the Excel file named "Exhibit 26 1998-2003" under the Tab labeled "all less fuel
& PP" in cells F215 through N232 for the statistical results that were provided by Excel with respect
to this analysis.
REQUEST 5: Plase provide copies of all studies or documents that Mr. Yankel is aware of
that demonstrate that a utility s annual expenses move in a linear fashion over time.
RESPONSE 5: Mr. Yankel has not made an analysis to determine what studies or
documents may be available to demonstrate that a utility s annual expenses move in a linear fashion
over time.
REQUEST 6: Please provide a list of all utility regulatory bodies or public utility
commissions that Mr. Yankel is aware of that use trended expense data as the basis for determining
expenses to be included in a utility's revenue requirement.
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 2
RESPONSE 6: Mr. Yankel has not made an analysis to determine what utility regulatory
bodies or public utility commissions use trended expense data as the basis for determining expenses
to be included in a utility's revenue requirement.
REQUEST 7: Please provide any data, study or analysis that supports the contention made
on page 18 of Mr. Yankel' s testimony that irrigation customers have lower distribution and customer
related costs per kWh than residential customers.
RESPONSE 7: Mr. Yankel has performed no specific studies. The statement on page 18
lines 14 and 15 of his direct testimony is based upon the long established principle that similar costs
spread over more units result in a lower per unit cost.
REQUEST 8: Please provide all data, including any analyses performed or used by Mr.
Yankel to support his assertion on pages 18 and 19 of his testimony that irrigation customers
generally have a higher load factor than residential and Schedule 9 - Secondary customers.
RESPONSE 8: Mr. Yankel has not made an analysis to determine the specific load factor
data ofthe irrigation class versus that ofthe residential and Schedule 9 - Second customers. These
comments are just based upon his general knowledge of the types of loads of various customer
groups.
REQUEST 9: Please identify by case number and order number the Idaho Public Utilities
Commission cases in which the 12-CP method has been used to allocate generation and transmission
costs between customer classes as stated on page 27, lines 6 through 8 of Mr. Yankel's direct
testimony.
RESPONSE 9: Mr. Yankel does not have specific Orders and Cases in mind. It is his
recollection that Idaho Power began using a 12-CP method for jurisdictional allocations in the -159
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 3
case and it was approved by the Commission. The case took place about 1979-1980. For class
purposes the Commission also adopted the use of the 12-CP method in the -158 case which occurred
in the same timeframe. It is Mr. Yankels recollection that since the -159 timeframe the Commission
has consistently adopted the use of a 12-CP method for Idaho Power s inter-jurisdictional
allocations. It is also Mr. Yankel's recollection that since the -158 timeframe the Commission has
consistently reviewed the use of a 12-CP method as well as other methods for Idaho Power s class
allocations, but has not come out and specified any favored methodology as it had in the -158 case.
REQUEST 10: Please provide all ofthe data and documentation used to prepare Figure 8
shown on page 35 of Mr. Yankel's direct testimony.
RESPONSE 10: Please see the attached electronic file that was used to develop Figure 8.
REQUEST 11: Please provide a copy ofthe agreement and copies of any other documents
that describe the agreement between the various parties in Utah that reviewed PacifiCorp s cost of
service dta and load research data that is referenced on page 20, lines 18 and 19 of Mr. Yankel' s
testimony. Ifno documents are available, please specify if the agreement is a written agreement or
a verbal agreement and list the parties who have entered into the agreement.
RESPONSE 11: A copy of the Report from Load Research Working Group in Utah is
attached. It is dated July 1 , 2002. PacifiCorp, the Division of Public Utilities (similar to the Idaho
Commission Staff), the Committee of Consumer Services, and various other parties (Mr. Yankel
does not recall names) agreed to the recommendation in the Report.
REQUEST 12: Please provide the rationale for using the allocation factor "gross plt" to
allocate federal and state income taxes to the various customer classes in Mr. Yankel's cost-of-
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 4
service studies included in the files "Yankel COS 1 base case
, "
Yankel COS 1 unweighted
demand"
, "
Yankel COS 3 normalized demand", and "Yankel COS 4 final"
RESPONSE 12: The allocation factor "gross plt" was used under the theory that taxes are
paid on return. Return is based upon plant that can generally be described for allocation purposes
as "Plant in Service" which is similar to "gross plt". There was no attempt at this pont to strongly
advocate the use of any particular allocator - merely an attempt was made to reasonably mimic the
Company s cost-of-service model in a more simplified and user-friendly model. This is the same
allocator that was used in the Company s study for "net provisions for defer inc taxes" and "ITCA"
I continued its uses for taxes because it was not possible to discern what the Company meant by
allocating on the basis of "Direct"
DATED this jlth day of March, 2004.
RACINE, OLSON, NYE, BUDGE & BAILEY
CHARTERED
By Ci v 41~ W~ I'- \'V
For RANDALL C. BUDGE
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 5
CERTIFICATE OF SERVICE
I HEREBY CERTIFY that on the 12th day of March, 2004, a true and complete copy ofthe
foregoing document and attachments were sent electronically to each of the following, and that on
March 15, 2004, a true and complete copy ofthe foregoing document and attachments was sent via
overnight mail to each of the following:
Barton L. Kline
Monica B. Moen
Idaho Power Company
1221 W. Idaho
Boise, Idaho 83707-0070
John R. Gale
VP-Regulatory Affairs
Idaho Power Company
1221 W. Idaho
Boise, Idaho 83707-0070
Lisa Nordstrom
Weldon Stutzman
Deputy Attorney Generals
Idaho Public Utilities Commission
472 W. Washington
Boise, Idaho 83702
Peter J. Richardson
Richardson & O'Leary
99 E. State Street, Suite 200
Eagle, Idaho 83616
Don Reading
Ben Johnson Associates
670 Hill Road
Boise, Idaho 83703
Lawrence A. Gollomp
Assistant General counsel
S. Department of Energy
1000 Independence Ave., SW
Washington, D.c. 20585
Dennis Goins
Potomac Management Group
5801 Westchester Street
Alexandria, VA 22310-1149
Dean J. Miller
McDevitt & Miller, LLP
420 W. Bannock
Boise, Idaho 83701
Jeremiah J. Healy
United Water Idaho, Inc.
O. Box 190420
Boise, ID 83719-0420 (US Mail)
William M. Eddie
Advocates for the West
O. Box 1612
Boise, Idaho 83701 (US Mail)
Nancy Hirsh
Northwest Energy coalition
219 First Ave. South, Suite 100
Seattle, W A 98104
Conley E. Ward
Givens Pursley LLP
601 W. Bannock Street
Boise, Idaho 83701
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 6
Dennis E. Peseau, Ph.
Utility Resources, Inc.
1500 Liberty Street S., Ste 250
Salem, Oregon 97302
Brad M. Purdy
Attorney at Law
2019 N. 17th Street
Boise, Idaho 83702
Michael Karp
147 Appaloosa Lane
Bellingham, Washington 98229
Michael L. Kurtz
Kurt 1. Boehm
37 E. Seventh St., Ste 2110
Cincinnati, OH 45202
Thomas M. Power
Economics Dept Liberal Arts Bldg 407
University of Montana
32 Campus Drive
Missoula, MT 59812
Idaho Irrigation Pumpers Association, Inc.s Responses to First Interrogatories and Production Requests
from Idaho Power Company - 7
Load Research Working Group
Report To The
Utah Public Service Commission
1 July 2002
Table of Contents
Introduction......................................................................................................................
Executive Summary...... ..........
...... ....................... ................................................... .........
Description of Load Research Techniques......................................................................
Quality and Reliability of Load Research Data
As It Relates to Calibration Adjustments............................................................... 9
Use of Load Research Data to Develop Residential Rate Design................................. 13
Irrigation Load Research ...............................................................................................
Appendix..................................................................... ................................. .................. 17
PacifiCorp s Overview Presentation of its Load Research Program.................. A 1
September 2000 Hourly Percentage Calibrations.............................................. A2
September 2000 Hourly MWH Calibrations.......................................................
Utah September 2000 cas Residential Split Summary..................................
Load Research Working Group Report to Utah PSG 1 July 2002
Introduction
In its November 2 2001 Stipulation Order in PacifiCorp general rate case Docket No.
01-035-, the Utah Public Service Commission (Commission) established a Utah Load
Resource Working Group (Working Group). Membership was to be drawn from
PacifiCorp (Company), the Division of Public Utilities (Division), the Committee of
Consumers Services (Committee), and other interested parties. The Committee was
assigned to chair the group. The Commission directed the Working Group to review
PacifiCorp s current and planned load research and studies and submit its report of
findings and recommendations by 1 July 2002.
The Load Research Working Group met five times between January 2002 and June
2002. Representatives of the following parties participated at some or all or those
meetings: Commission Staff, the Committee, Crossroads Urban Center, the Division
Hill Air Force Base, Salt Lake Community Action Program , PacifiCorp, and the Utah
Farm Bureau.
The Load Research Working Group addressed the following issues:
. The techniques used to collect load research data as well as some of the
practical problems associated with gathering reliable data.
. The quality and reliability of the load research data as it relates to
calibration/adjustments that have been performed on this data.
The use of the load research data to help design residential rates for customers
that use more than 1 000 kWh per month during the summer months.
. The need to develop alternatives to the use of historical irrigation load research
data for rate case purposes on a going forward basis.
There was agreement within the Working Group on all issues that were fully reviewed.
For those issues that were not fully reviewed , there was general agreement on progress
achieved and what the data means that has been reviewed thus far. Although some
issues were not fully investigated because of time and other constraints, the Working
Group has been able to better define and clarify these issues so as to make future
discussions regarding these matters more focused with a clearer point from which to
begin.
Load Research Working Group Report to Utah PSG 1 July 2002
Executive Summary
In the Commission s Order of November 2 2001 it adopted a Stipulation that was
presented regarding PacifiCorp Docket No. 01-035-01. As a part of that Order and as a
part of the Stipulation a Load Research Working group was established to review the
current and planned PacifiCorp load research efforts. Although the Stipulation and the
Order did not specify concerns to be addressed , the Working Group focused upon
many of the concerns that were raised during the rate case that are associated with
load research data. The areas reviewed included:
. The quality and reliability of the load research data.
It was decided that with the addition of a large number of sample points for the
Residential Class in early 2000 , that the quality and reliability of the load
research data has been increased. There is still a question about quality and
reliability related to the fact that there is wholesale movement of sample
customers between stratum boundaries on a monthly basis. This is not to say
that there is a problem , merely a concern that needs further investigation.
. The calibration of load research data in order to match assigned Utah
Jurisdictional load with the sum of census data plus load research data.
It was decided that the Company would no longer calibrate load research data in
this manner, thus removing some of the concerns that were raised in the last two
rate cases. This agreement only addresses the impact upon load research data
but does not address the jurisdictional question regarding the appropriateness of
the assignment of jurisdiction demand and energy values.
The appropriateness of developing a good sample for the Irrigation Class.
It was decided that the Company should not establish a new sample for Irrigation
Customers. Instead of using load research data, the Irrigation Class should
simply be given the jurisdictional average change in rates.
. The use of the load research data for rate design purposes as well as cost
allocation purposes.
It was decided that load research data could be used to better define cost of
service boundaries within the Residential Class regarding tailblock pricing during
the summer months. More work should be done in this area before specific
recommendations are made.
Load Research Working Group Report to Utah PSG 1 July 2002
Description of Load Research Techniques
Techniques used by the Company to develop the load research sample design are
appropriate. The load research protocol is designed to produce a sample that is
accurate within +/- 10 percent on 90 percent of the observations. Although the sample
design is aimed at producing an accurate sample, there are practical problems that
require constant review and vigilance in order to insure that the expected level of
accuracy is achieved. The Working Group addressed three areas where possible
improvements could be made in the load research program:
1. The number of residential load research samples being collected could have
been below an acceptable number of samples necessary to produce the desired
precision and accuracy of the sample. This potential problem has been
corrected with the addition of a number of new residential sample sites during
early 2000.
2. Sample customers are grouped according to size (usage) into what are referred
to as Strata. Weightings are then assigned to the various Strata in order to
reflect the relative percentage of the population as a whole that is represented by
customers within that Strata (usage range). A concern has been raised with
respect to sample migration with sample customers moving from one stratum to
another while the data being collected from that customer continues to be
assigned to the original stratum. It is also recognized that the electricity usage of
the population as a whole varies from month to month. Much of this sample
migration could simply mimic the overall change in usage of the general
population. Since the Company is still gathering data on the migration issue , this
Working Group has not had an opportunity to review the data prior to the
deadline for this report. However, the identification of the concern should help to
insure consideration of the issue in future load research efforts.
3. A concern was also raised regarding how the Company checks the accuracy of
its load research data against the total population billed energy usage. At
present, an initial adjustment is made to the load research data if its derived
energy levels differ from the general population by more than +/- 10 percent.
The Company is checking to see if each sample stratum is also found to be
within +/- 10 percent of its respective portion of the general population , however
the Company has stated that the design precision levels are developed for the
whole sample and are not applicable to the individual strata. The identification of
the concern should help to insure that future load research efforts will be more
aware of possible large discrepancies in usage between the various stratum and
the portion of the general population they represent.
Quality of Load Research Data and Calibration Adjustments
In the last two PacifiCorp general rate cases concern has been raised regarding the
impact of calibrations that have been made to the Company s load research data. The
Load Research Working Group Report to Utah PSG 1 July 2002
specific calibrations of concern are the ones that come about when the summation of
the Company s load research data and the directly measured census data do not equal
the Utah Jurisdictional assigned load or "Jurisdictional Border Load" in any given hour.
The Working Group has generally agreed that anyone of three components (load
research data , census data, and/or Utah Border Load data) could have an error that
impacted these calibration factors.
The Working Group agreed on several points with respect to these calibrations that had
historically been applied to the load research data:
. The general conclusion was that there is something occuring within the Utah
Border Load that is more likely the source of the calibration problem than the
load research data or the census data. The Working Group agreed that the
Company should discontinue the practice of calibrating Utah load research data.
Although the above agreement solves the problem of the large and biased
calibration factors that have been placed upon load research data in recent rate
cases, it does not address the source of these high calibration factors or why
measured/sampled retail loads in Utah (plus expected losses) do not equal the
Utah Border Load. It was concluded that the reason for this discrepency was a
result of the losses associated with wholesale and other system transactions that
get assigned to the retail jurisdictions. Since this is an important issue, but
beyond the scope of this Working Group, it is recommended that these losses be
addressed in another forum.
Use of Load Research Data to Develop Residential Rate Desiqn
The review of the Company s load research data resulted in an analysis of the usage
characteristics of the sample customers and how that usage fits the definition of the
various stratum boundaries. It was observed that during the summer months
residential customers using over 1 000 kWh per month were more coincident with the
system peak than were smaller residential customers. This suggested that, during the
summer months, generation and transmission costs were higher for the larger
customers and could support an inverted rate block pricing structure. At the same time
it was noted that the larger customers had higher non-coincident peak load factors than
the smaller residential customers. It was recommended that a cost of service study be
prepared for the summer months that separated the residential class into customers
over 1 000 per month and customers using less than 1 000 kWh per month.
A preliminary study was prepared on an annual basis (not just the 4-summer months)
which segregated the residential class between those customers using more than
000 kWh per year and those using fewer than 12 000 kWh per year. On an annual
basis the study results showed that the generation and transmission component cost of
service was higher per kWh for the larger residential customers. However, this
difference was more than offset by a lower cost of service per kWh for the distribution
and retail components. This suggests that, on an annual basis , an inverted rate was
Load Research Working Group Report to Utah PSG 1 July 2002
not supported. It is important to note that the study did not segregate costs by season.
Without a detailed cost of service analysis it is impossible to determine how significant
the difference in cost of service during the summer months actually is.
While this analysis was discussed , it is agreed by the Working Group that this is a
pricing design issue and not a load research issue.
Irriqation Load Research
From a load research perspective the Irrigation Class has historically been a difficult
group to sample. The Utah Irrigation Class represents only about one percent of the
jurisdiction s total customers and one percent of the jurisdiction s energy requirement.
Compared to its size , however, the Irrigation Class is expensive to sample.
Furthermore, Irrigation Customers can drastically change usage from one year to the
next which could mean that sample meters that are installed one year may not produce
results during the next year, or only during a portion of the irrigation season.
The Company recognizes that its irrigation load research data is stale and some
alternative needs to be found for future rate cases. The Working Group recommends
that the Company not develop a new load research sample for Utah Irrigation
Customers. The Working Group also recommends that whenever rates are changed
for the Utah Jurisdiction as a whole , the Irrigation Customers should get the
jurisdictional average percentage rate change.
Load Research Working Group Report to Utah PSG 1 July 2002
Description of Load Research Techniques
The most significant demand related costs are production and transmission costs
which are related primarily to each rate schedule s demand at the time of the system
peak (coincident peak demand). Unfortunately, most customers do not have demand
meters. Furthermore , most of the customers that do have demand meters have
demand meters that only register the highest 15-minute demand during the month
(which may not be at the time of the system peak). Generally, only customers with
demands equal to or greater than 1 MW have their loads continuously monitored.
Therefore , coincident peak demands for all rate schedules except the very largest must
be estimated through the use of load research techniques. The Company does this by
placing special load research meters on a random sample of customers from each rate
schedule. Data from these samples is used to estimate the coincident peak demands
for each rate schedule from which the 12 CP is calculated.
The Company s Load Research Department made presentations and answered
questions throughout the course of the Working Group meetings. At the 15 January
2002 meeting, the Company presented an overview of its load research efforts in Utah
and other jurisdictions. This presentation is a good point of reference and is included
as Appendix A of this Report.
The load research data is gathered from stratified random samples. Customers are
grouped by annual energy usage and then a specific number of samples are taken from
each of the stratum. The number of samples taken from each stratum is a function of
the size of the stratum and the variation in customer usage within the stratum. For
example , the Residential Class was divided into four strata along the lines of the
following annual usage ranges:
Lower Upper Weighting Sample
Stratum Usa Usa Factor Points
000 43.
001 000 41.
001 000 13.
001 999 999
The energy consumption data is the only known quantity, and thus, it is the basis for
setting the stratum boundaries. Energy consumption is also used as one check on the
validity of the load research results within each rate schedule. The load research
sample results for each rate schedule are scaled up to the overall population of the rate
schedule. The resulting energy values are then compared to the actual energy
Load Research Working Group Report to Utah PSG 1 July 2002
readings taken from the rate schedule population as a whole. If the energy levels from
the load research program result in values in any given month that deviate from the
actual measured energy results for a rate schedule by more than +/- 10 percent, then
an adjustment or calibration is made to bring the load research results for that particular
rate schedule in line with the actual results. This first level calibration should not be
confused with the calibration that is made to all load research data for all customer
classes in order to bring the total measured Utah Load in line with the Utah Border
Load data. Generally, such first level calibrations have not been required for the
residential customers or the small commercial customers taking service under Rate
Schedule 23. Occasionally there has been a need to make such an adjustment to Rate
Schedule 9 , but this adjustment is only made to the sampled group of smaller (under
MW) Schedule 9 customers , which represented only about five percent of the total load
of that schedule.
Although the load research samples are designed to produce results that were accurate
within +/- 10 percent on 90 percent of the observations , there were some practical
problems that may have made this data less accurate than theory would suggest. The
first problem is the number of valid samples actually taken. As pointed out above , the
present design of the residential sample calls for 74 samples to be taken in the first
stratum. Actual sampling data between January 1999 and September 2000 showed
that the number of samples ranged from 50-69 samples. The low number of samples
typical of the sampling in 1999. In early 2000, the residential sample was revised and
new sample customers were added. Since that time , it appears that the number of
Residential Customers being sampled is adequate to meet the study protocol.
A second problem that was addressed is a form of sample migration. Stratum
boundaries have been set up on the basis of annual consumption. For example, the
Residential Stratum 1 Customers are those that have an average monthly usage
between 0 and 500 kWh (annual usage between 0 and 6,000 kWh per year). It was
found that although the appropriate number of customers may have been sampled
within any give stratum certain customers may have used more or less energy than the
range of average usage specified for members of that stratum. In the case of the
Residential Stratum 1 Customers a sample customer may have an average monthly
usage between 0 and 500 kWh , but usage in one or more months may exceed 500
kWh. Thus, during some months this customer would have usage that is similar to that
of a customer in Stratum 2. Likewise, it is possible for a customer from Stratum 2 with
average monthly usage between 501 and 1 000 kWh (annual usage between 6 001
kWh and 12 000 kWh) to occasionally use less than 500 kWh per month--usage similar
to a customer in Stratum 1.
One concern with sample migration is that the data from each sample is weighted by
predetermined values that are based upon annual , not monthly, usage characteristics
of the sample customers as they relate to the usage characteristics of the general
population. If some sample customers are assigned to one stratum , but the usage
characteristics for a given month reflect the usage levels of a different sample stratum
then the weighted sample results may not be reflective of the actual general population
Load Research Working Group Report to Utah PSG 1 July 2002
usage pattern. It was noted that there was generally greater usage by all sample
customers during the summer months, which contributed to this movement of
customers across stratum boundaries. It is possible that this general increase in usage
is reflective of shifts within the general population. The Company maintains that this is
in fact the case. The Working Group has not come to any conclusions regarding the
monthly shifts in usage as they relate to the fixed limits of the stratum boundaries that
are based upon annual consumption.
A related question arose regarding the way the Company checks the accuracy of its
samples each month by comparing the energy usage of the sampled customers
grossed up to the general population , with the actual energy usage of the general
population. As mentioned above, if there is a deviation of greater than +/- 10 percent in
the estimated and sampled energy usage of any given month for a given rate schedule
then an adjustment is made. However, this adjustment is based upon the summation of
all stratum compared to the rate schedule as a whole. Therefore, it is possible that one
or more stratum can be off by more than +/- 10 percent yet the summation of the
stratum are within +/- 10 percent of the general population. As with the sample
migration problem mentioned above, the Working Group has not come to any
conclusions regarding the impact of the accuracy of the sample collection on a stratum-
by-stratum basis.
The Company maintains that seasonal usage fluctuations are present throughout the
entire residential population and that resulting strata migration is accounted for by using
annual energy as the sampling variable. No adjustments should therefore be made to
the stratum level load research estimates.
Load Research Working Group Report to Utah PSG 1 July 2002
Quality and Reliability of Load Research Data
As It Relates to Calibration Adjustments
The Company s Load Research Group has, as a matter of routine , calibrated hourly
load research estimates obtained from all non-census rate groups to the hourly Utah
jurisdictional system loads. These calibrations were applied to all load research
estimates in an attempt to true up the load research data to the "actual" system loads
reported for the State of Utah. This methodology was based on the supposition that the
Utah Jurisdictional Loads 1) accurately depict the actual hourly electrical usage for the
State of Utah , and 2) were appropriately comparable to the summed Utah Firm Retail
Class Loads being estimated by the Load Research Group.
Prior to the Pacific Power - Utah Power merger, class load calibration was performed
on all sample load estimates for Utah , Idaho, and Western Wyoming (UP&L) beginning
with the 1983 annual load reports. Post-merger, ongoing load studies in Idaho and
Wyoming were removed and new load studies were conducted on a cycle basis (each
class is monitored for two-years on a five-year rotation). Cycle rotation was
implemented for these states in an effort to reduce costs. Residential and small
commercial and industrial usage patterns in these jurisdictions did not change
materially from one year to the next. Without benefit of ongoing load research sample
data to compare to the measured jurisdictional loads , calibration was discontinued for
Idaho and Wyoming in 1998.
Load research studies for Oregon , Washington , California and Eastern Wyoming are
conducted on a cycle basis and for this reason calibration has not been performed in
these states.
In the most recent PacifiCorp rate cases , concern was raised about the quality and
reliability of the load research data used in the class cost of service studies. These
concerns centered round the calibrations that were applied to the load research data in
order to make the summation of the load research data and the Census data equal to
the measured Utah Border Loads. There were a number of concerns raised with
respect to these calibrations including:
1) I n Docket 99-035-10 the calibrations that were used for 10 out of the 12
coincident peaks in the test year were positive, thus adding to the peak load
attributed to these customers;
2) The magnitude of these coincident peak calibrations was large, with five out
of the 12 months having calibrations in excess of a positive nine percent;
3) In Docket 01-035-01 it was demonstrated that the calibrations averaged a
positive seven percent over a two-year period; and
Load Research Working Group Report to Utah PSG 1 July 2002
4) In Docket 01-035-01 it was demonstrated that there was a definite seasonal
and time-of-day pattern associated with the magnitude of the calibrations.
The cause of the magnitude of the calibrations as well as the seasonality and time-of-
day nature of these calibrations was the focus of one portion of the Working Group
activities. It was recognized that there were at least four factors that contributed to the
magnitude of the calibrations factors on an hourly basis and ultimately the peak
demand usage of each rate schedule. Those four factors are as follows:
Load research uses sampling to define load requirements for the smaller
customers-reflecting roughly 75 percent of jurisdictional load.
Census data from individual large customers that are individually measured
as opposed to sampled-reflecting the remaining 25 percent.
Utah Jurisdictional Border Load-used to define how much energy is
consumed (including any losses that take place) in the State of Utah.
Average losses (developed in 1991) are applied to the load research data.
These losses likely change over time and temperature.
If the summation of the load research data and the census data do not equal the Utah
Jurisdictional Border Load in any given hour, then a calibration has historically been
made to the load research data only, such that the summation of the load research data
and the census data do equal the Border Load. Anyone of these four components
could have an error that resulted in an impact upon these calibration factors.
The review of the calibration data in Docket 01-035-01 was confined to the average
percentage calibration of the load research data for each hour of each month. Under
the Working Group, a closer look was taken at the level of calibrations that took place
during each hour for the months of January and September 2000. It was discovered
that not only was there a time-of-day pattern to the level of the calibrations, but that this
time-of-day pattern was dependent upon the day of the week, with weekends being
distinctively different than weekdays. Appendix B contains a listing of the hourly
percentage calibrations that were encountered for each hour in September 2000 with
Sundays and Holidays" highlighted. The "Average" hourly calibration for the month of
September 2000 is the same as that used in Exhibit CCS- 7.10 page 1 of 3 in Docket
01-035-01 and serves as the basis for the analysis here. The basic findings were:
Under the first eight hours of the day the "Average" calibration of the same
hour in the month , the "Average" calibration is generally over 10 percent.
During the middle of the day the "Average" calibration dips to five percent or
less, while it generally climbs to six percent or greater during the latter part of
the day.
Load Research Working Group Report to Utah PSG1 July 2002
It was found that the "Sunday/Holiday" calibrations are somewhat higher than
Average" during the first eight hours of the day, significantly higher than the
calibrations during the middle of the day, and moderately higher than the
calibrations during the last eight hours.
. Saturdays calibrations are actually a little less than "Average" during the first
eight hours, more than "Average" but less than Sunday calibrations during the
middle of the day, and unlike either set of values during the last eight hours.
There is a well defined "6 x 16" pattern associated with the calibration data
where the highest calibrations occur during the hours of approximately
midnight to 8:00 a.m. and all day on Sundays and Holidays.
Another observation is that (except for Sundays and Holidays) these
calibrations abruptly go from very high during the first eight hours of the day
to very low calibrations during the next eight hours with virtually no transition.
There appears to be a similar, although less dramatic, jump in the opposite
direction between the hours of 16:00 and 17:00 (4:00 p.m. and 5:00 p.
Up until this point, the calibration problem had only been measured on the basis of the
percentage calibration that was being applied to the load research data. It was
suggested that because the loads vary by hour, it would be of value to also review the
calibration problem on the basis of the actual magnitude of the calibration. Appendix C
lists the actual magnitude of the calibration for each hour of September 2000. Although
this provided a different perspective on the problem, it did not drastically alter the
results. Most noticeably, the major drop in the level of calibrations still occurs at the
8:00 a.m. hour. Sundays still have a high calibration all day long, and calibrations start
a somewhat abrupt increase between the 16:00 and 17:00 hours.
However, the exercise of using actual data as opposed to percentage data brought out
another aspect of the calibration problem. The data used by the Companys Load
Research Department to equate the census and load research data with the Utah
Border Load data includes the use of the peak loss data that has historically been
developed for Utah. The shortcoming with this approach is that peak loss factors are
only appropriate for use during a few hours during the year. Throughout the course of a
day or a month there will be many more hours when the loss factors are closer to the
average loss factors and even less than the average losses. Given the magnitude and
nature of the loss factors, this means that the percentages and the absolute magnitude
of the loss factors listed in Appendix Band C are generally significantly understated.
The Working Group sought to identify the source of the high calibrations that were
being added to the load research data used in the Companys rate cases. Because of
the dramatic (and well-defined) changes in calibrations, it was presumed that the
source of the calibration problem should be detectable by a review of the hourly
Load Research Working Group Report to Utah PSG 1 July 2002
differences in data around the 8:00 a.m. hour for the load research data , the census
data , and the Utah Border Load data. The hourly load research data and census data
for the month of September 2000 around the hour of 8:00 a.m. was reviewed.
Generally speaking, there were no hourly changes for any rate schedule or census
group data of the magnitude found in the calibration data over this timeframe.
The general conclusion was that there is something occuring within the Utah Border
Load that is more likely the source of the calibration problem than the load research
data or the census data. The Working Group agreed that the Company should
discontinue the practice of calibrating Utah load research data such that the sum of the
load research data and census data equate to the Utah Border Load data for the
following reasons:
1. The calibration process requires uniform derivation and application of the
jurisdictional loads, class load studies , and demand loss factors. Deficiencies
any of these contributing functions could contribute to inappropriate calibration
results. The current methodology assumes that all calibration differences are
attributable to load study sample error. We no longer believe this assumption to
be reasonable.
2. Compilation of the Utah Class loads is delayed by months while the jurisdictional
loads are prepared. Jurisdictional loads are input into the calibration process.
3. It is questionable that load research sample estimates are improved in the
calibration process, and in fact from a statistical sense, they may be adversely
impacted.
4. Significant resources are required to perform hourly calibration with no apparent
benefit.
Although the above agreement solves the problem of the large calibration factors that
have been placed upon load research data in recent rate cases , it does not address the
source of these high calibration factors or why measured retail loads in Utah plus
expected losses do not equate to the Utah Border Load. It was concluded that the
reason for this discrepancy was the treatment of the losses associated with wholesale
and other system transactions in the development of the Border Loads for all states.
Given the agreement to remove the calibrations from the load research data , it was
concluded that investigation of the impact of this discrepancy between measured Utah
Retail Load and Utah Border Load is outside the scope of this forum. The Working
Group recommends that the Commission establish or identify another forum and direct
the parties to investigate this discrepancy in that forum.
Load Research Working Group Report to Utah PSG 1 July 2002
Use of Load Research Data to Develop Residential Rate Design
When reviewing the residential load research data , an effort was made to determine if a
relationship exists between monthly usage and monthly coincident peak demand. For
purposes of this analysis, all monthly load research data was combined from all stratas
and then ranked by the amount of energy usage. At the upper usage levels (above
000 kWh per month), there was a great deal of overlap during the four summer
months between customers of various stratum with all four stratum having some
customers with usage greater than 1 000 kWh per month.
It was found that there was a decline in the monthly coincident load factor during the
four summer months (June through September) which means that the larger
usagecustomers had usage more coincident with the monthly system coincident peak
than lower usage customers. This indicates that, during the summer, generation and
transmission costs are higher for the larger residential customers than smaller
residential customers. Although an exact line of demarcation was not established, the
decline in monthly coincident load factor was generally believed to occur near 1 000
kWh of usage. A similar relationship between usage level and monthly coincident peak
was not found for the other eight months of the year. The above data suggests that an
inverted block rate starting at approximately 1 ,000 kWh may be appropriate for the
Residential Class during the summer months. (It is at 400 kWh currently.
A similar review was made of the monthly non-coincident peak data for the Residential
Class. For purposes of this analysis all monthly load research data was combined for
all stratum and then ranked by the amount of energy usage. Unlike the monthly
coincident peak data , it was found that there was an increase in the monthly non-
coincident load factor during the four summer months (June through September) which
means that the larger usage customers used their electricity on a more uniform basis
when compared to the lower usage residential customers.
The conclusion of the Working Group is that this data was significant enough to warrant
further analysis through a cost of service study. It is not only important to identify what
type of customer behavior causes increases in costs, but the actual amount of increase
in costs needs to be quantified as well. Although the load research data serves as the
foundation for much of the data that goes into a cost of service study, it is not a simple
matter to incorporate the above findings into such a study. One way to incorporate the
above data into a cost of service study is to divide the residential customers into two
groups , one that uses over 1 000 kWh during any of the summer months and one that
does not. Under this approach there is a need to review the Company s bill frequency
data and determine how many customers fit this description and how to get the load
research data compiled to properly reflect this bifurcation of the Residential Class.
Further work will need to be done in order to break the cost of service study out into one
more customer category such that the resulting data simply divides the Residential
Class into two groups, but has no impact upon the other rate schedules.
Load Research Working Group Report to Utah PSG 1 July 2002
A preliminary study was prepared which segregated the residential class between those
customers using more than 12 000 kWh per year and those using fewer than 12 000
kWh per year. As suggested in the review of the load research data, the study showed
that the generation and transmission components cost of service were higher, on a
cents per kWh basis, for the larger residential customers. However, this difference was
more than offset by a lower cost of service per kWh for the distribution and retail
components. This suggests that on an annual basis, an inverted rate was not
supported from a cost of service standpoint.
However, the study did not segregate costs by season. Additional analysis that isolates
the effect of the cost differences to the summer period is needed to determine if an
inverted rate for larger residential customers can be supported during that period.
The above work was completed just prior to the submission of this report to the
Commission and has not been thoroughly reviewed by the Working Group. Although
additional analysis is needed , the Commission should be aware that there is data
available upon which it can assess the price impact of summer usage patterns in the
Utah Jurisdiction. Although this information should not be the sole basis for any
residential rate design decisions, the information should serve the Commission well in
understanding the relationship between certain usage patterns and costs.
The Working Group believes that the residential load research data collected since
early 2000 is of sufficient quality to conduct the above described inverted block pricing
study.
Load Research Working Group Report to Utah PSG 1 July 2002
Irrigation Load Research
From a load research perspective the Irrigation Class has historically been a difficult
group to sample. The Irrigation Class has been a small class within the Utah
Jurisdiction with only about one percent of the jurisdiction s total customers and one
percent of the jurisdiction s energy requirement. Compared to its size , the Irrigation
Class is expensive to sample. The Irrigation Class is highly diversified , and thus,
requires more samples per total number of customers in order to get a sample of
desired accuracy and precision. In order to adequately sample the Irrigation Class
100-120 load research meters are required. By comparison , approximately 170 load
research meters are required for the Residential Class that is 30-40 times larger than
the Irrigation Class. Furthermore, Irrigation Customers can drastically change usage
from one year to the next which could mean that sample meters that are installed one
year may not produce results during the next year, or only during a portion of the
irrigation season.
The cost of gathering good load research data for Irrigation Customers has led to the
use of stale load research data in recent rate cases. The Company s cost of service
studies have included values for irrigation loads that are taken from data gathered in
the 1991-1993 timeframe and then massaged in an attempt to simulate test year
results. The resulting cost assignments to the Irrigation Class have been the source of
a great deal of contention during these cases. The Company recognizes that its
historical irrigation load research data is stale and some alternative needs to be found
for future rate cases.
Two alternatives were reviewed. The first alternative was to initiate a new load
research sample for irrigators in Utah. The cost of this alternative was explored and it
was decided that it was not worth the price-especially considering the fact that
Irrigation Customers would continue to change watering requirements and thus usage
from year to year.
The second alternative addressed by the Working Group was proposing a stipulation
that would allow the Company to avoid designing and gathering a new load research
sample for the Irrigation Customers. As a part of this stipulation , it was also agreed that
whenever rates are changed for the Utah Jurisdiction as a whole , that the Irrigation
Customers get the jurisdictional average rate change.
The specific working of the stipulation is as follows:
For at least the last decade , the Irrigation Class comprises less than one
percent of the total kWh sales in Utah. Due to the relatively small size of
the Irrigation Class in Utah, and the potentially high cost of implementing
new load research samples to improve load research studies, the parties
agree that new load research samples are not necessary for the
Load Research Working Group Report to Utah PSG 1 July 2002
foreseeable future. As a result, the parties also agree that in subsequent
cases where rates are changed and where new load research studies have
not been employed for the Irrigation Class, that the proposed rate spread
for Irrigation Customers will be equal to the overall average jurisdictional
percentage change.
It is the intent of all parties that have participated in this Working Group that at the next
appropriate opportunity, the above language will be presented to the Commission for its
consideration with respect to how the revenue requirement for the Irrigation Class
should be calculated in the future without new load research data.
Load Research Working Group Report to Utah PSG 1 July 2002
Appendix
PacifiCorp s Overview Presentation of its Load Research Program
September 2000 Hourly Percentage Calibrations
September 2000 Hourly MWH Calibrations
Cost of Service by Rate Schedule
Actual Actual Actual Actual Actual Actual
demand demand total energy energy total
customers in season out of season in season out of season kwh
2002 24-161 800 3,494 942 319 760 814 702 395 399 710 289 172 641 684 572 351
2001 24-147 862 872 065 440 111 2,431 954 054 384 877 291 137 612 345 522,489
2000 24-152 895 740 379 107 036 847,415 526 573,482 403 202 227 929 775,709
1999 24-147,404 583 866 683 101 266 967 331 084 984 318 614 230 649 699 214
1998 24-143 668 3,411 179 757,402 168 581 152 335 068 271 332 057 1,423 667 125
1997 24-141 528 3,472 159 829 921 302 080 217,419 009 307 791 324 525 210 333
199624-103 764 512 219 733 832 246 051 329 398 112 298 742,446 628 140 558
1995 24-917 360 036 228 392 588,428 089 920 557 210 825 349 300 745 906
1994 24-586 105 838 793 468,466 451 326 179 350 028 639 801 770 419
1993 24-241,493 243 974 525 012 115 852 037 221 784 732 337 980 614
395 810 205
2002 24- T 535 340 875 275,440 200 278 640
2001 24-255 878 133 1,430 319 418,400 848 719
2000 24-4,442 584 026 559,745 837 917 397 662
1999 24- T 051 258 309 746 167 560,429 306 596
1998 24- T 984 608 592 276 249 578 469 854,718
1997 24- T 641 752 393 190 752 849 703 040,455
1996 24- T 903 550 453 2,436 641 562,467 999 108
1995 24- T 796 164 960 163 102 524 919 688 021
1994 24- T
2002
2001
042 182 127 341 259,468 619 387 165 848 785 235
935 352 070 106,422 322,487 308 626 631 113
Schedule 9 secondary service
2002 203 680 306 977
2001 193 137 8 186 838
2000 184,453 7 997 919
643 624 957
691 104,432
573 140 205
0.4359
0.4503
0.4407
Normalized Normalized Normalized Normalized Normalized Normalized
demand demand total energy energy total total in season
in season out of season in season out of season kwh energy demand
504 337 323 570 827 907 281 657 090 278 015 297 559 672 387 108,008%99.732%
872 065 326 392 198,457 282 585 016 355 749 591 638 334 607 82,127%100.000%
740 379 107 036 847,415 1,439 762 818 392 990 272 832 753 090 105.294%100.000%
583 866 683 101 266 967 353 164,427 322 076 209 675 240 636 98.475%100.000%
3,411 179 757,402 168 581 300 689 156 291 876 609 592 565 765 89.395%100,000%
3,472 159 829 921 302 080 270 910 591 272 166 142 543 076 733 98,842%100.000%
511 600 734,451 246 051 325 870 209 303 847 818 629 718 027 99.903%100.018%
358 610 573 035 931 645 254,494 086 308 120 765 562 614 851 83.242%100.042%
586 105 838 793 5,468,466 345,496,406 295,440 044 641 283,404 109.778%100.000%
241 393 243 974 525 012 326 354 801 294 160,408 621 210 221 82.530%100.003%
255 878 133 270,401 370 172 640 573
442 584 026 2,404 502 783 862 188 364
051 258 309 817 669 488 766 306,435
984 608 592 984 211 612 134 596 345
641 752 393 663 705 512 086 175 791
903 550 5,453 2,473 365 555,414 028 779
796 164 960 651 581 827,494 3,479 075
182 127 341 259,468 65,472 866 853 861 326 727
352 070 106,422 867 575 664 181 531 756
Actual Actual
out season total in season out season demand energy
demand demand energy energy in season in season
99,912%99.831 %108,875% 104.013%2002 24-2002 3,494 942 395,400
33,181%57.925%82.208%81.838%2001 24-2001 2 872 065 054 385
100.000%100.000%106.030% 102.599%2000 24-2000 3 740 379 526 573
100.000%100,000%98.368%98,925%1999 24-1999 3 583 866 331 085
100.000%100.000%88.594%92.961%1998 24-1998 3,411 179 152 335
100.000%100.000%95,791 % 113.089%1997 24-1997 3,472 159 217,419
99.964%100.000%100,266%98.320%1996 24-1996 3 512 219 329 398
78.091%93.041%86,881%68.423%1995 24-1995 3 360 036 089 921
100.000%100.000%107.865% 118.477%1994 1994 3 586 105 1,451 326
100.000%100.000%84.129%75.396%1993 1993 3 241 493 115 852
Actual
demand
in season
1993 3 241,493
1994 3 586 105
1995 3 360 036
1996 3 512 219
1997 3,472 159
1998 3,411 179
1999 3 583 866
2000 3 740 379
2001 2 872 065
2002 3,494 942
Actual
energy
in season
115 852
1,451 326
089 921
329 398
217,419
152 335
331 085
526 573
054 385
395,400
Actual
energy
in season
395 399 710
054 384 877
526 573,482
331 084 984
152 335 068
217,419 009
329 398 112
089 920 557
451 326 179
115 852 037
delta actual/normalized
demand demand
in season
395
619
1 ,426
100
Actual
energy
in season
115 852 037
451 326 179
089 920 557
329 398 112
217,419 009
152 335 068
331 084 984
526 573 482
054 384 877
395 399 710
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
810
766 503
619
344 643
monthly ave
Actual
demand
in season
377 999
491 642
369 214
450 338
412,405
390 357
450 910
517 132
357 176
472 696
Actual
energy
in season
810 373
896 526
840 009
878 055
868 040
852 795
895 967
935 095
718 016
873 736
in season
546935
5029
559087
508781
0.462755
0.480306
518502
0.444353
554395
0.471562 000 000
900 000
800 000
700 000
600 000
500 000
400 000
300 000
200 000
100 000
(j)..-..-..-
cr:
..........
Q')Q')Q')Q')Q')Q')Q')Q')Q')
('/')
I'--ex)Q')