20110718PAC to IIPA 34 -3.pdf

Date: March 24, 2011
To: Jeff Bumgarner
From: Jim Stewart, Hossein Haeri and Brian Hedman
Re: Impacts of Rocky Mountain Power’s Idaho Irrigation Load Control
Program

Rocky Mountain Power retained The Cadmus Group to evaluate the 2009 and 2010 demand
impacts of the company’s irrigation load control program offered to the customers in Idaho. This
document summarizes the results of Cadmus’s study.
Background
In 2009, the Program enrolled 2,032 customers and had approximately 260 MW of participating
load in Schedule 72 (schedule forward) and Schedule 72A (option dispatch). In 2010, the
Program enrolled 1,975 customers and had approximately 283 MW of participating load. In both
years, over 98 percent of the Program load was enrolled through the dispatch option.
During the 2008 Program Season the Company began noticing voltage excursions outside
industry acceptable standards during dispatch events. In 2010 the Company implemented a
process to reduce load and return load to normal operating levels in phases to minimize the
impact on the company’s transmission and distribution system. As a result, the Company was
still unable to take the entire participating load off during the peak time period between 2:00p
and 6:00p. As a consequence, the current level of participation is beyond what RMP can
effectively dispatch. This has reduced the Program’s cost-effectiveness.

Technical Approach
The Cadmus Group estimated the hourly load reductions achieved by the Program in 2009 and
2010. The analysis was conducted using SCADA system data for five sub-stations (Amps, Big
Grassey, Bonneville, Jefferson, and Rigby) that accounted for most (77 percent) of the controlled
irrigation load in Idaho. For each substation and event hour, Cadmus estimated a reference load,
what the load would have been in the absence of the event, and compared it to the observed load
during curtailment events. Results were extrapolated as representative of the remaining circuits
to account for total program loads.
The reference load for an event hour was estimated in two ways: (1) as the unconditional average
load in the same hour of the two weekdays preceding and following the event; and (2) as the
conditional average load estimated using a regression of hourly demand on weather, calendar and
time effects, and indicators for event hours and hours preceding and following the event. The
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 1 of 22
Jeff Bumgarner Page 2 of 22
February 24, 2011
difference between the observed load and the actual yielded the estimate of the load reduction in
the event hour.
For both estimation approaches, the estimated load reduction in each hour was compared to the
expected load reduction (nominal load reduction) adjusted for opt-outs and a load reduction
realization rate was calculated. There are several aspects of this methodology that are worth
noting before considering the results. Nominal load is defined as the sum of customers’ average
billing demands for June, July and August for the two prior years.
• The impact analysis is based on SCADA data at the substation level. Since the majority
of the loads being served by these substations consist of irrigation, the amount of “noise”
in the data resulting from the variability of non-irrigation loads is expected to be minimal.
Moreover, the hourly demand model used to estimate the load impacts largely accounts
for such noise in the substation data.1

• Program management staggers (stair-steps) the dispatching of loads at the beginning and
end of events for grid reliability purposes. The hourly analysis of loads does not account
for the staggering. As a result, the estimated load impacts in the first and last hours are
an estimate of the average load reduction over the hour and may not represent the true
reduction at the beginning (likely to be smaller than estimated) or end of the hour (likely
to be larger).

• The analysis adjusts for, in the calculation of realization rate, the required scheduling of
22 percent of the available participating loads outside of the 2:00p-6:00p time period.
This scheduling restriction was implemented in 2010 to accommodate the Grid control
voltage limitations previously noted. While this did not impact realization rates, it did
impact the decrease in aggregate reduction from 205 MW in 2009 to 156 MW in 2010.
Results Summary and Conclusions

With these limitations in mind, the evaluation team analyzed the substation data for the 2:00p to
6:00p time horizon and reached the following conclusions:
• In 2009, the maximum hourly load reduction on the five substations was 158 MW which
extrapolates to 205 MW for the entire program. This reduction occurred on July 17 and
represented 86 percent of the nominal load (program resources) adjusted for opt-outs in
the hour. The realization rates, which show how much load was shed relative to
expectation, ranged from a low of 17 percent on August 5 to the July 17 high of 86
percent. In 2010, the maximum hourly load reduction at the five substations was 120

1 Of the five substations only the Rigby substation serves other loads, including small businesses, a college, a
hospital and the cities of Rexburg, Rigby, Ririe, Menan, and smaller towns.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 2 of 22
Jeff Bumgarner Page 3 of 22
February 24, 2011
MW which extrapolates to 156 MW for all Idaho irrigation program loads. This occurred
on July 8 and represented 77 percent of the opt-out-adjusted nominal load in the hour.
Program benefits are calculated based on 156 MW of system impact. On July 20, a load
reduction of 120 MW resulted in the maximum realization rate of 82 percent. During
hours when events are traditionally called, realization rates ranged from a low of 29
percent on August 24 to the high of 82 percent on July 20.
• Realization rates were calculated based on expected loads, or in the case of the Rocky
Mountain Program, loads that could safely be dispatched without adversely impacting
line voltages. This is an important distinction worth noting. Had the calculation of
realization rates been based on total participating loads, this would have resulted in lower
realization rates. As program cost-effectiveness is calculated on actual load reductions
relative to a program’s costs (rather than a realization rate), realizations rates should not
be considered the definitive measurement of a program’s effectiveness and value.
• The load reductions and realization rates in any year may not be representative of typical
load impacts the program might achieve because of annual weather-related variations in
irrigation demand.
• Rocky Mountain Power system peak coincides with hours when events are traditionally
called (hours 2:00p to 6:00p). In 2009, all of the top 10 non-event, summer hours
occurred during the traditional event window. Rocky Mountain Power system peak hours
do not coincide with morning and early afternoon / evening hours when loads were
dispatched in 2010 because of transmission and distribution constraints.
• While the Program has been operationally effective, it has not been as cost-effective as it
could be. In 2009 and 2010, the Program enrolled more load on some substations than it
could dispatch during peak hours because of transmission and distribution constraints.
To increase future cost-effectiveness, RMP needs to either upgrade its transmission and
distribution system in Idaho to remove the operating constraints or limit enrollment in the
Program to a level consistent with the system’s ability to dispatch resources during peak
hours.
In addition, since the inception of the program Rocky Mountain Power has been educating
irrigators about efficient irrigation practices and the benefits of irrigating during off-peak
hours. Rocky Mountain Power estimates that because of education irrigators have shifted
between 5 and 7 percent of their loads between 2:00p and 6:00p to off peak. The estimation
of the reference load for this analysis is not taken into consideration in this analysis. If the
benefits from education were taken into consideration the load shifting from education would
have the effect of further improving measured impact or realization rate.

The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 3 of 22
Jeff Bumgarner Page 4 of 22
February 24, 2011
Objectives
The objectives of this evaluation were:
• To estimate the irrigation load reductions from Rocky Mountain Power’s irrigation direct
load control program in 2009 and 2010.
• To estimate ex-post realization rates, the ratio of the ex-post impacts to the nominal program
loads that can be shed.
Program Operations
RMP operates two irrigation load control programs in Idaho. The first is “schedule forward”
(Schedule 72) and involves direct control of irrigation loads on a scheduled basis. Enrollment in
this program has been decreasing annually with the implementation of the dispatch program
option. In July 2009, there were 4.1 MWs of nominal load in this program. The second is the
dispatch option (Schedule 72a). RMP calls “events” with 24 hours advance notice and uses
simplex technology to shed irrigation loads during event hours (a maximum of four hours per
day per customer during weekdays).2 The event hours are typically between 2:00p to 6:00p. In
July of 2009, there were 254 MWs of nominal irrigation load in both programs. In July of 2010,
there were 282 MWs of nominal load.
Event History
In 2009, RMP called six events that each lasted four hours. The events occurred between 2:00p
and 6:00p. Table 1 shows the dates and hours of the events.
Table 1. Event Days and Hours in 2009
Idaho 2009 
  30‐Jun 4 hours 
  17‐Jul 4 hours 
  23‐Jul 4 hours 
  3‐Aug 4 hours 
  5‐Aug 4 hours 
  13‐Aug 4 hours 
Hours for all events occurred 
during hours 2:00p to 6:00p. 

In 2010, RMP called 11 events, excluding three one-hour events in early June and one four-hour
event for irrigators served by the Big Grassey substation and for grid operations purposes.3 In
addition to a larger number of events in 2010, there were also a larger number of hours when

2 Participants may opt out of a maximum of five events per season.
3 The regression models control for the grid operations events, but we do not report the estimated load reductions.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 4 of 22
Jeff Bumgarner Page 5 of 22
February 24, 2011
RMP dispatched program resources. Resources were dispatched during not just 2:00p-6:00p but
also hours before and after this window because of transmission and distribution constraints.
Table 2 shows the dates and number of hours for the 2010 events.

Table 2. Event Days and Hours in 2010
Idaho 2010 
  29‐Jun 8 hours* 
  8‐Jul 8 hours* 
  15‐Jul 8 hours* 
  16‐Jul 8 hours* 
  19‐Jul 12 hours** 
  20‐Jul 12 hours** 
26‐Jul 12 hours** 
  2‐Aug 12 hours** 
  5‐Aug 12 hours** 
  24‐Aug 12 hours** 
  26‐Aug 12 hours** 
*Hours for all substations:
11:00a ‐7:00p. 
** For all substations except 
Big Grassey, event hours 
occurred 11:00a – 7:00p.  
Beginning July 19, RMP also 
dispatched Big Grassey 
customer loads from 7:00a ‐
11:00a. 

Between the first event on June 29, 2010 and the fourth event on July 16, 2010, RMP dispatched
program resources on event days in three blocks over eight hours: 11:00a –3:00p, 2:00p – 6:00p,
and 3:00p–7:00p. Figure 1 illustrates the dispatch of program resources during these time
blocks.
Figure 1. Summer 2010 Irrigation Direct Load Control Dispatch Blocks
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 5 of 22
Jeff Bumgarner Page 6 of 22
February 24, 2011

Beginning with the fifth event on July 19 and ending with the final (11th) event on August 26,
RMP dispatched additional resources between 7 am and 11 am on the Big Grassey substation.4
Resources associated with the other substations continued to be dispatched in three blocks
between 11:00a and 7:00p.
Tables 3 and 4 show loads at the five substations that RMP expected it could shed during each
month of 2009 and 2010 based on the historical demand of enrolled customers. This is known as
the ‘nominal’ load. The estimates of nominal load in Tables 3 and 4 do not take into account
customers that opted out of events.
In 2009, the nominal load varied across months but not hours, as all available program resources
were dispatched during the 2:00p – 6:00p window. Nominal loads were highest during July
when irrigation demand was greatest.
Table 3. Program Nominal Resources (MW) in 2009 for Five Substations

June
(all event 
hours) 
July
(all event 
hours) 
August  
(all event 
hours) 
Program Nominal 
Irrigation Load (MW) 
served by substations 
in estimation sample 

178  

196  

188  
Source: Table 14, Schedule 72 and 72A Idaho Irrigation Load 
Programs 2009 Credit Rider Initiative Final Report and personal 
communications with Bill Marek about percentage of program 
nominal load served by Amps, Big Grassey, Bonneville, Jefferson, 
and Rigby substations.  Loads are not adjusted for opt‐outs.  
Nominal load is the load that RMP expected it could shed based 
on program enrollment and transmission and distribution 
constraints.  

4 In addition, there was an AMD dispatch block on Amps 3 days/week from 6:00p -12:00a. This involved a small
amount of load, approximately 1.75 MW per dispatch or 5.3MW in total. All AMD dispatches from all
substations accounted for ~15 MW of participating load.

The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 6 of 22
Jeff Bumgarner Page 7 of 22
February 24, 2011

In 2010, the nominal loads on the five substations varied between months and event hour, as
program resources were dispatched in several four-hour blocks, as described above. The nominal
loads do not take into account the gradual ramping down and up of loads at the beginning and
end of the period or opt outs.
Table 4. Program Nominal Resources (MW) in 2010 for Five Substations
  7‐10a 11:00a 12:00p 1:00p 2:00p 3:00p 4:00p 5:00p 6:00p 7:00p 8p‐12a
June 0.0 47.0 47.0 49.0 89.3 148.8 148.8 148.8 110.1 5.8 1.6
July 1‐July 
19 0.0 50.7 50.7 53.0 96.5 160.7 160.7 160.7 118.9 6.2 1.8 
July 20‐
July 31 17.1 42.6 42.6 44.9 88.4 151.7 151.7 151.7 109.9 6.2 1.8 
August 16.9 42.0 42.0 44.2 87.1 149.6 149.6 149.6 108.3 6.2 1.7
Source: Schedule 72 and 72A Idaho Irrigation Load Programs 2010 Credit Rider Initiative Final Report and 
personal communications with Bill Marek.  Loads are not adjusted for opt outs.  Nominal load is the load that 
RMP expected it could shed based on program enrollment and transmission and distribution constraints.     

Data
RMP provided Cadmus with 60 second interval data for five substations (Amps, Big Grassey,
Bonneville, Jefferson, Rigby) that served irrigators in its Idaho service territory in 2009 and
2010. The substations accounted for approximately 77 percent of RMP’s irrigation load
subscribed in the program in Idaho in 2010. RMP also provided Cadmus with data about the
days and hours when direct load control resources were dispatched.
Cadmus performed a number of quality checks on and adjustments to the interval data before
analyzing the load impacts. We first put the 60 second interval data on an hourly basis by
calculating average hourly loads for each substation. The hourly load data were then plotted and
examined for irregularities. While the minute interval data did exhibit some random spikes and
drops in load (normal perturbations in electrical Grid operations), these abnormalities were not
evident after the minute interval data were averaged over the hour.
Next, we obtained hourly and daily weather data for Rexburg and Idaho Falls weather stations
from the National Weather Service and merged it with the hourly load data. The weather
variables in the analysis include the daily evapotranspiration rate, temperature (hourly), and
rainfall (hourly).5

5 The evapotranspiration rate was a weighted average of crop-specific ETRs, with weights equal to the share of land
planted in the crops.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 7 of 22
Jeff Bumgarner Page 8 of 22
February 24, 2011
Last, Cadmus mapped information on the occurrence of load control event hours to the data. We
created separate indicator variables for each hour of each event, which were included in the
model.
Impact Estimation Approach
The Cadmus approach to estimating the load reductions in each event was to estimate a reference
load (what demand would have been in each hour of an event if the event had not occurred) in
each hour during an event window. The difference between the actual load and the reference
load in an event hour is the estimate of the program’s impact during that event hour.
Figure 2 illustrates the approach. It shows the hourly loads for the Bonneville substation on July
23, 2009, when RMP called the third event of the summer. The event window was 2:00p to
6:00p. The red (solid) line is the observed load. The blue (dashed) line is the reference load that
was generated with a regression model. The impact of the event in each hour is the difference
between the metered load (red line) and the reference load (blue line). The figure depicts an
estimated average hourly impact of approximately 38 MW.
The reference load can be estimated in several ways. One is a day matching approach. This
involves estimating the (unconditional) average of the loads in the same hour in the two
weekdays immediately preceding and following the event. If irrigation demand conditions,
which are a function of weather, evapotranspiration, crop maturity, and other factors, on the
reference days are similar to those on the event day, the reference load will likely represent well
what demand would have been, and the difference between observed and reference loads will be
an accurate estimate of the true load reduction. However, if any of the demand conditions
change, the load reduction estimates will be biased.
Figure 2. Illustration of Event Impact Estimation Approach

The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 8 of 22
Jeff Bumgarner Page 9 of 22
February 24, 2011
The second approach is multivariate regression in which loads are modeled as a function of
weather, time, and calendar variables. This method accounts for differences in demand
conditions between event and non-event days and will generate a more accurate reference load.
Cadmus determined that because of trends in irrigation demand over the growing season that the
day matching approach would not be appropriate. Reference loads were estimated using an
hourly demand regression models.
Conditional Demand Impact Estimation
Using regression analysis, Cadmus also modeled hourly demand as a function of weather (evapo-
transpiration, temperature, and rainfall), calendar and time effects (week of month, day of the
week, and hour), and load in the same hour in the previous day.6 The models also included
separate indicator variables for each hour of each event and for each of the six hours following
and preceding each event. The coefficients on the event hour variables represent the differences
between the observed loads and the reference loads in the event hours. The Appendix describes
the model specification in greater detail.
Cadmus estimated separate demand models for each of the substations and event months (June,
July, and August). Thus, there were a total of 15 substation models (5 stations x 3 months). We
estimated separate substation month models for two reasons. First, each substation has a
somewhat different load shape over the summer, reflecting differences between stations in
cropping practices and irrigation and non-irrigation demand.7 Second, each substation’s load
shape varies significantly over the summer, reflecting changes in crop maturity, evapo-
transpiration, soil-type temperature, wind, relative humidity, solar radiation, and rainfall over the
growing season.
Model Estimation and Diagnostics
Cadmus estimated the models by Generalized Least Squares (GLS) under the assumption of
auto-correlated errors, that is, load in each hour is assumed to be correlated with the load during
a preceding hour. The error term was modeled as an autoregressive process with lag one.
We performed a number of tests to evaluate the predictive ability of the substation regression
models. These tests included inspection of the signs and statistical significance of the models’
coefficients, estimation of overall explanatory power of each model, represented by R2 statistic,

6 Loads were modeled as a function of the average temperature in the preceding 24 hours, total rainfall in the
preceding 24 hours, and average daily evapo-transpiration over the preceding three days. The week of month
variables capture changes in irrigation demand related to changes in cropping activities. The days of the week
and hour of the day variables capture irrigation demand that varies by day and hour.
7 The Rigby substation is different from the other stations in that it has significant non-irrigation loads.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 9 of 22
Jeff Bumgarner Page 10 of 22
February 24, 2011
and tests of the predictive ability of the models in hours when events could have been called on
non-event days.8 We used the results of the tests in selecting the final model specifications.
The models predict accurately what loads would have been in hours when events were not but
could have been called. Table 6 reports the median absolute percentage error, the median of the
percentage difference between the observed load and the load predicted by the model (|kW-
model predicted kW)|/kW, during non-event hours on July weekdays between 2 and 6 pm.
Table 6. Median Absolute Percentage Error for July 2009
Hour Amps 
Big 
Grassey Bonneville Jefferson Rigby 
2:00 PM 0.74% 1.45% 1.35% 1.54% 0.72% 
3:00 PM 0.86% 1.44% 1.28% 1.48% 0.67% 
4:00 PM 1.62% 1.29% 0.87% 1.28% 0.64% 
5:00 PM 0.95% 1.04% 1.09% 1.85% 0.67% 
Note: Absolute percent error is =|predicted MW –actual MW|/ actual 
MW. 

For example, in 50 percent of the 3 pm non-event hours at the Bonneville station, the regression
model predicts a load that is within 1.28 percent of the actual load. The median absolute
prediction error ranges from less than 0.7 percent to just below two percent. Fifty percent
(N=10) of the substation-hour median percentage errors are less 1.2 percent.
Estimated Load Reductions in 2009
Table 7 reports an estimate of the total load reduction for the Amps, Big Grassey, Bonneville,
Jefferson, and Rigby substations and all Idaho irrigation in each event hour during summer
2009.9 The estimate for Idaho was obtained by dividing the substation estimate by the substation
percentage of the Idaho irrigation load (77 percent). The Table also reports the realization rate
for each event hour (2:00p-6:00p time window), which is the ratio of the estimated total load
reduction in a given hour to the nominal load adjusted for irrigation loads that opted out of the
event.10 The realization rate is a function of the estimated load reduction (the numerator) and
expectations about loads that can be shed (the denominator). It may be less than or equal to 100
percent depending on technical performance of the control equipment (i.e., signals and
transmitted and received and pumps are shut off) and whether irrigation demand during the
season was less than or greater than expected.

8 In general, the coefficients of the models have the expected signs and are statistically significant. Loads were
increasing in the evapo-transpiration rate and temperature and decreasing in rainfall. Loads were generally
highest during the afternoon and early evening hours. Also, based on their R2 statistics, the models explain a
large percentage of the variation in irrigation loads.
9 The Appendix contains estimates of the reduction in load at the substation level in each event hour.
10 Cadmus adjusted the nominal load for an event by subtracting the amount of load that opted out the event.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 10 of 22
Jeff Bumgarner Page 11 of 22
February 24, 2011

Table 7. Estimated Load Reductions and Realization Rates in 2009
Date Event Hour 
Estimated 
Load 
Reduction ‐ 
Five 
substations 
(MW) 
Estimated 
Load 
Reduction 
‐ All 
Idaho 
Irrigation 
(MW) 
Hourly 
Opt‐Out 
Adjusted 
Realization 
Rate 
30‐Jun Event 1 Hour 1 ‐41.8 ‐54.3 24.8% 
    Hour 2 ‐71.8 ‐93.2 42.6% 
    Hour 3 ‐70.7 ‐91.8 42.0% 
    Hour 4 ‐66.4 ‐86.3 39.5% 
17‐Jul Event 2 Hour 1 ‐111.1 ‐144.3 60.8% 
    Hour 2 ‐157.8 ‐204.9 86.3% 
    Hour 3 ‐158.0 ‐205.2 86.4% 
    Hour 4 ‐151.6 ‐196.9 82.9% 
23‐Jul Event 3 Hour 1 ‐102.4 ‐133.0 55.7% 
    Hour 2 ‐137.7 ‐178.9 74.9% 
    Hour 3 ‐138.6 ‐180.0 75.3% 
    Hour 4 ‐136.5 ‐177.2 74.2% 
3‐Aug Event 4 Hour 1 ‐33.6 ‐43.6 18.5% 
    Hour 2 ‐50.0 ‐65.0 27.6% 
    Hour 3 ‐48.1 ‐62.5 26.5% 
    Hour 4 ‐48.0 ‐62.4 26.5% 
5‐Aug Event 5 Hour 1 ‐30.8 ‐40.0 17.0% 
    Hour 2 ‐50.0 ‐65.0 27.6% 
    Hour 3 ‐49.0 ‐63.7 27.1% 
    Hour 4 ‐47.4 ‐61.6 26.2% 
13‐Aug Event 6 Hour 1 ‐36.6 ‐47.6 19.9% 
    Hour 2 ‐45.9 ‐59.6 24.9% 
    Hour 3 ‐45.4 ‐58.9 24.6% 
    Hour 4 ‐45.6 ‐59.2 24.7% 
Notes: Estimates of load reductions for 5 substations based on regression 
model.  Estimated load reductions for all Idaho Irrigation estimated as 5 
substation load reduction divided by 0.77.  Realization rate is the ratio of 
the estimated load reduction to the opt‐out adjusted nominal load. 

The Program reduced irrigation loads in each event hour. The estimated load reductions ranged
from -158 MW to -31 MW and were different from zero at the 5 percent significance level.11
The estimated reductions in Idaho irrigation loads ranged from 40 MW to 205 MW. The

11 The Appendix contains estimated confidence intervals for the estimated load reductions in all event hours.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 11 of 22
Jeff Bumgarner Page 12 of 22
February 24, 2011
estimates also exhibit the expected patterns. First, during each event, the estimated load
reduction in the first hour was the smallest, consistent with the staggering of the event initiation
for grid reliability. (During hours two, three, and four, there is very little difference in the
estimated load reductions.) Second, the load reductions over the summer reflected the seasonal
pattern of irrigation demand. The load reductions were largest in July, when loads and irrigation
demand were at their peak. The maximum load reductions on the five substations of 158 MW
and in Idaho irrigation loads of 205 MW were achieved on July 17 (event 2) during event hour 3.
The estimated load reductions were significantly smaller in June and August, when irrigation
demand was much lower.
The realization rates, which show how much load was shed relative to expectation in any given
hour, ranged from a low of 17 percent during hour 1 of event 5 to a high of 86 percent during
hour 3 of Event 2. As expected, realization rates were significantly higher in July than in June or
August because of irrigation practices and crop maturity. Nominal loads were not adjusted
downward to reflect the lower irrigation demand in June and August. Hence, the low realization
rates were due not to Program performance but rather to below average irrigation demand and
the fact that nominal rates during June and August are lower. RMP may want to consider
adjusting its estimates of nominal loads to reflect changes in irrigation demand.
Estimated Load Reductions in 2010
During events in 2010, program resources were dispatched in three or four blocks over 8 or 12
hours. Loads were dispatched outside of the 2:00p to 6:00p window because of potential adverse
impacts on the transmission and distribution system. Table 8 reports an estimate of the
maximum hourly load reduction in each block of each event during summer 2010.12 Cadmus
reports the maximum in each block of hours instead of in each hour because of the large number
of event hours. The load reductions cover the Amps, Big Grassey, Bonneville, Jefferson, and
Rigby substations. It should be noted that loads that were shed between 7:00a and 10:00a or
11:00a and 1:00p resumed at the end of the event, leaving less opportunity for load reductions in
subsequent hours (note: loads that were were controlled between 7:00a and 10:00a and 11:00a
and 1:00p resumed at the end of the event, leaving less opportunity for load reductions in
subsequent hours).
The load impacts were greatest during 2:00p – 6:00p, when most Schedule 72a resources were
dispatched (see Table 4). The maximum hourly load reduction occurred on July 8, when
irrigation loads on the five substations were reduced by approximately 120 MW and the Idaho
irrigation load was reduced by 156 MW. Load impacts were smaller in June and August, when
irrigation demand was lower.

12 The Appendix contains estimates of the load reduction in each event hour.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 12 of 22
Jeff Bumgarner Page 13 of 22
February 24, 2011
Table 8. Estimated Load Reductions in 2010
Date 

Event 

Load 
7 AM ‐10 
AM 
11 AM ‐1 
PM 
2 PM ‐ 5 
PM 6 PM 
29‐Jun Event 1 5 Substations N/A ‐34.8 ‐87.0 ‐61.7 
    All ID Irrigation N/A ‐45.2 ‐113.0 ‐80.1 
8‐Jul Event 2 5 Substations N/A ‐49.6 ‐119.8 ‐85.7 
    All ID Irrigation N/A ‐64.4 ‐155.5 ‐111.3
15‐Jul Event 3 5 Substations N/A ‐44.2 ‐107.0 ‐86.6 
    All ID Irrigation N/A ‐57.4 ‐139.0 ‐112.5
16‐Jul Event 4 5 Substations ‐39.9 0.0 ‐100.5 ‐77.7 
    All ID Irrigation ‐51.8 0.0 ‐130.5 ‐101.0
19‐Jul Event 5 5 Substations ‐40.2 ‐17.9 ‐103.1 ‐83.3 
    All ID Irrigation ‐52.2 ‐23.2 ‐133.9 ‐108.2
20‐Jul Event 6 5 Substations ‐48.3 ‐15.1 ‐105.4 ‐82.2 
    All ID Irrigation ‐62.7 ‐19.7 ‐136.9 ‐106.7
26‐Jul Event 7 5 Substations ‐36.1 ‐12.2 ‐89.7 ‐75.8 
    All ID Irrigation ‐46.9 ‐15.9 ‐116.5 ‐98.4 
2‐Aug Event 8 5 Substations ‐2.4 ‐3.1 ‐6.7 1.3 
    All ID Irrigation ‐3.1 ‐4.0 ‐8.6 1.7 
5‐Aug Event 9 5 Substations ‐8.7 ‐10.0 ‐42.2 ‐31.5 
    All ID Irrigation ‐11.3 ‐12.9 ‐54.8 ‐41.0 
24‐Aug Event 10 5 Substations ‐25.5 ‐6.0 ‐41.3 ‐31.8 
    All ID Irrigation ‐33.2 ‐7.8 ‐53.6 ‐41.3 
26‐Aug Event 11 5 Substations ‐20.4 ‐2.6 ‐44.3 ‐30.6 
    All ID Irrigation ‐26.5 ‐3.4 ‐57.5 ‐39.7 
Notes: Estimates of load reductions for 5 substations based on regression model.  Estimated load 
reductions for all Idaho Irrigation estimated as 5 substation load reduction divided by 0.77.  
Realization rate is the ratio of the estimated load reduction to the opt‐out adjusted nominal load. 

The hourly MW impacts were smaller in 2010 than in 2009 because load control resources were
dispatched over a larger number of hours. The dispatching of resources in the morning and early
afternoon and early evening to address transmission and distribution issues meant that there was
less potential to reduce loads during peak hours. To put the 2010 load impacts in perspective,
Table 9 reports realization rates, the ratio of the estimated load impact to the nominal load in the
hour adjusted for opt outs.13 The nominal loads during peak hours were smaller in 2010 than in
2009 because programs resources were dispatched before and after the 2:00p – 6:00p period.
The realization rates account for the smaller amount of load that could have been shed between
2:00p and 6:00p.

13 The load opting out was subtracted from the nominal load for hours 2:00p – 6:00p for each event.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 13 of 22
Jeff Bumgarner Page 14 of 22
February 24, 2011
Table 9. Estimated Realization Rates in 2010 (Based on Nominal Capacity)
Date 

Event 
7 AM ‐10 
AM 
11 AM ‐1 
PM 
2 PM ‐5 
PM 6 PM 
29‐Jun Event 1 N/A 71.0%60.3%56.0% 
8‐Jul Event 2 N/A 93.7%77.4%72.1% 
15‐Jul Event 3 N/A 83.5%76.0%72.9% 
16‐Jul Event 4 N/A 0.0%74.0%70.8% 
19‐Jul Event 5 234.9%39.8%76.7%75.8% 
20‐Jul Event 6 282.0%33.7%82.0%74.8% 
26‐Jul Event 7 211.3%27.2%63.9%69.0% 
2‐Aug Event 8 14.1%6.9%4.5%‐1.2% 
5‐Aug Event 9 51.4%22.5%29.7%29.1% 
24‐Aug Event 10 151.4%13.5%28.6%29.4% 
26‐Aug Event 11 121.1%6.0%30.2%28.2% 
Notes: Realization rate is the ratio of the estimated load reduction to the opt‐
out adjusted nominal load.  Opt out loads obtained from Schedule 72 & 72A 
Idaho Irrigation Load Control Programs: 2009 Credit Rider Initiative Final Report. 

During hours when events are traditionally called, the realization rates ranged between 29
percent on August 24 and 82 percent on July 20.14 (We ignore the August 2 event, as load
reductions were uniformly and abnormally low.15) During peak irrigation demand between the
first and third weeks of July, the realization rate ranged between 77 and 82 percent of nominal
load. These impacts are slightly lower than but still close to those in 2009. The difference in
realization rates may reflect the fact that irrigation demand in 2010 was relatively low because of
cooler weather throughout the summer.
Conclusions
Rocky Mountain Power asked Cadmus to evaluate the demand impacts of its Idaho irrigation
load control program. In 2010, the Program enrolled 1,975 customers and had approximately
283 MW of participating load. However, this participating load was more than RMP could
dispatch during peak hours because of transmission and distribution system constraints. This has
had the effect of reducing the Program’s cost-effectiveness.

14 On some event days, the maximum hourly realization rate between 7:00a and 10:00a exceeded 100 percent. This
indicates that in these hours either the Program achieved significantly greater demand reductions than expected,
or the nominal loads are too low,
15 Irrigation demand is typically very low at the beginning of August when hay is harvested and water to field crops
is turned off to initiate the crop maturation process prior to harvest. Accordingly, potential demand reductions
are very small. However, the nominal load covers all of August and does not reflect haying and crop
maturation. The small, negative demand reduction in the 6:00 p hour is statistically indistinguishable from zero.
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 14 of 22
Jeff Bumgarner Page 15 of 22
February 24, 2011
Cadmus estimated the hourly load reductions from the Program in 2009 and 2010 using
regression analysis of SCADA data from five substations in Idaho. In addition, Cadmus
examined the coincidence of the program impacts with the PacifiCorp system peak demands.
There are several noteworthy aspects of the methodology:
• The impact analysis was based on SCADA data at the substation level. Since the majority
of the loads being served by these substations consist of irrigation, the amount of “noise”
in the data resulting from the variability of non-irrigation loads is expected to be minimal.

• The estimation methodology did not consider Rocky Mountain Power’s education of
irrigators about efficient irrigation practices. If the benefits from education were taken
into consideration the load shifting from education would have the effect of improving
measured impact or realization rate.

• The hourly analysis of loads did not account for staggering in the dispatching of loads at
the beginning and end of events for grid reliability purposes. As a result, the estimated
load impacts in the first and last hours are an estimate of the average load reduction over
the hour and may not represent the true reduction at the beginning (likely to be smaller
than estimated) or end of the hour (likely to be larger).

• In the calculation of realization rates, the analysis adjusts for the required scheduling of
22 percent of the available participating loads outside of the 2:00p-6:00p time period.
This scheduling restriction was implemented in 2010 to accommodate the Grid control
voltage limitations previously noted. While this did not impact hourly realization rates, it
did have a significant effect on the difference between the nominal loads and the
aggregated reductions achieved.

Year Nominal Load Aggregated Reduction
2009 260 MW 205 MW
2010 283 MW 156 MW

The analysis of substation loads showed the following:
• In 2009, the maximum hourly load reduction on the five substations was 158 MW or 205
MW for all Idaho irrigation program loads. This represented 86 percent of the nominal
program resources dispatched in that hour. The realization rates, which show how much
load was shed relative to expectation, ranged from a low of 17 percent on August 5 to the
July 17 high of 86 percent. In 2010, the maximum hourly load reduction was 120 MW
or 156 MW for all Idaho irrigation program loads. This occurred on July 8 and
represented 77 percent of the opt-out-adjusted nominal load dispatched in the hour. On
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 15 of 22
Jeff Bumgarner Page 16 of 22
February 24, 2011
July 20, a load reduction of 120 MW resulted in the maximum realization rate of 82
percent.
• Realization rates were calculated based on expected loads, or in the case of the Rocky
Mountain Program, loads that could safely be dispatched without adversely impacting
line voltages. This is an important distinction worth noting. Had the calculation of
realization rates been based on total participating loads, this would have resulted in lower
realization rates. As program cost-effectiveness is calculated on actual load reductions
relative to a program’s costs (rather than a realization rate), realizations rates should not
be considered the definitive measurement of a program’s effectiveness and value.
• The load reductions and realization rates in any year may not be representative of typical
load impacts the program might achieve because of annual variations in irrigation
demand.
• PacifiCorp system peak coincides with hours when events are traditionally called (hours
2:00p-5:00p).
Recommendations
While the Program has achieved significant load reductions, the cost-effective has been
adversely impacted by the level of participation on a megawatt basis. As noted above, in 2009
and 2010, the Program enrolled more load on some substations than it could dispatch during
peak hours because of transmission and distribution constraints. RMP could reduce enrollments
to a level consistent with the system’s ability to dispatch loads. Or if technically feasible, RMP
could increase the Program’s cost-effectiveness by upgrading the transmission and distribution
system to alleviate constraints on when load can be dispatched.

The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 16 of 22
Jeff Bumgarner Page 17 of 22
February 24, 2011
Appendix
Substation Hourly Load Model
Let j=1,2..., J index the events and h=1,2…, H index hours of each event. Also, let MWit be the
electricity load of substation i at time (hour) t. Then (suppressing the index i) substation i’s MW
n at time t (corresponding to a week of the month, day, and hour) can be written as: dema d
72 24 24
∑∑∑
∑ ∑ ∑ ∑ ∑ ∑

The right hand side variables in the model are defined as follows:
• EvapTR72hourt is the average evapo-transpiration rate over the previous 72 hourst at time t.
• Temp24hourt is the average temperature over the previous 24 hours at time t.
• Rainfall24hourt is the total rainfall over the previous 24 hours.
• Weekofmonthwt equals one if time t is in week w, w=1 to 3, and equals zero, otherwise.
Daydt, d=1 to 6, and hourof daykt, k=1 to 23, are defined similarly.
• Eventhourjht equals one if time t is in hour h, h=1 to H, of event j, j=1 to J, and equals zero,
otherwise. Preeventhourjht and Posteventhourjht are defined similarly.
• εt is the error term of the model representing random influences on the demand of customer i
at time t.
The parameters to be estimated and their interpretations are as follows:
• ρhj is the impact of hour h of event j on demand. It is the difference between the estimate of
what demand would have been if an event had not been called (reference load) and the actual
demand in the hour.
• ωhj is the impact of hour h after event j on demand. The coefficients capture any shifting of
irrigation loads in response to the load control events.
• φhj is the impact of hour h before event j on demand. The coefficients capture any shifting of
irrigation loads because of the load control events.
• α0 is substation load at the omitted hour (Sundays at the 12 am hour in the first month).
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 17 of 22
Jeff Bumgarner Page 18 of 22
February 24, 2011
• α1 is the impact of average evapo-transpiration rate in the previous 72 hours on demand. α2
shows the impact of temperature in the previous 24 hours on demand. α3 measures the
impact of rainfall in the previous 24 hours on demand.
• πw, w=1 to 3, is the impact of week of month w on demand.
• δd, d=1 to 6, is the impact of day of the week d on demand.
• γk, k=1 to 23, is the impact of hour k on demand.

The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 18 of 22
Jeff Bumgarner Page 19 of 22
February 24, 2011
Appendix Table A.1. 2010 Estimated Hourly Load Reductions with 95 Percent Confidence
Intervals
Date Event Hour Estimated 
Load 
Reduction ‐ 
Five 
substations 
(MW) 
Lower 
Bound 
95% 
Confidence 
Interval 
Upper 
Bound 
95% 
Confidence 
Interval 
Estimated 
Load 
Reduction 
‐ All Idaho 
Irrigation 
(MW) 
Opt‐Out 
Adjusted 
Nominal 
Load 
Opt‐Out 
Adjusted 
Realization 
Rate 
30‐Jun Event 1 Hour 1 ‐41.8 ‐55 ‐28 ‐54.3 168.4 24.8% 
    Hour 2 ‐71.8 ‐86 ‐57 ‐93.2 168.4 42.6% 
    Hour 3 ‐70.7 ‐86 ‐56 ‐91.8 168.4 42.0% 
    Hour 4 ‐66.4 ‐82 ‐50 ‐86.3 168.4 39.5% 
17‐Jul Event 2 Hour 1 ‐111.1 ‐125 ‐97 ‐144.3 182.8 60.8% 
    Hour 2 ‐157.8 ‐172 ‐144 ‐204.9 182.8 86.3% 
    Hour 3 ‐158.0 ‐172 ‐144 ‐205.2 182.8 86.4% 
    Hour 4 ‐151.6 ‐166 ‐138 ‐196.9 182.8 82.9% 
23‐Jul Event 3 Hour 1 ‐102.4 ‐116 ‐89 ‐133.0 184.0 55.7% 
    Hour 2 ‐137.7 ‐152 ‐124 ‐178.9 184.0 74.9% 
    Hour 3 ‐138.6 ‐153 ‐124 ‐180.0 184.0 75.3% 
    Hour 4 ‐136.5 ‐150 ‐122 ‐177.2 184.0 74.2% 
3‐Aug Event 4 Hour 1 ‐33.6 ‐42 ‐25 ‐43.6 181.5 18.5% 
    Hour 2 ‐50.0 ‐58 ‐42 ‐65.0 181.5 27.6% 
    Hour 3 ‐48.1 ‐57 ‐40 ‐62.5 181.5 26.5% 
    Hour 4 ‐48.0 ‐56 ‐40 ‐62.4 181.5 26.5% 
5‐Aug Event 5 Hour 1 ‐30.8 ‐39 ‐22 ‐40.0 181.0 17.0% 
    Hour 2 ‐50.0 ‐59 ‐41 ‐65.0 181.0 27.6% 
    Hour 3 ‐49.0 ‐58 ‐40 ‐63.7 181.0 27.1% 
    Hour 4 ‐47.4 ‐56 ‐39 ‐61.6 181.0 26.2% 
13‐Aug Event 6 Hour 1 ‐36.6 ‐45 ‐28 ‐47.6 184.2 19.9% 
    Hour 2 ‐45.9 ‐54 ‐37 ‐59.6 184.2 24.9% 
    Hour 3 ‐45.4 ‐54 ‐37 ‐58.9 184.2 24.6% 
    Hour 4 ‐45.6 ‐54 ‐37 ‐59.2 184.2 24.7% 
Notes: Estimates of load reductions for 5 substations based on regression model.  Estimated load reductions for all 
Idaho Irrigation estimated as 5 substation load reduction divided by 0.77.  Realization rate is the ratio of the 
estimated load reduction to the opt‐out adjusted nominal load. 
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 19 of 22
Jeff Bumgarner Page 20 of 22
February 24, 2011
Appendix Table A.2. 2010 Estimated Hourly Load Reductions with 95 Percent Confidence
Intervals
Date Event Hour Block Estimated 
Load 
Reduction ‐ 
5 
Substations 
(MW) 
Lower 
Bound 95% 
Confidence 
Interval 
Upper 
Bound 95% 
Confidence 
Interval 
Estimated 
Load 
Reduction 
‐ All Idaho 
Irrigation 
(MW) 
Opt‐out 
adjusted 
Nominal 
Load 
(MW) 
Realization 
Rate 
Nominal 
Load 
(MW) 
29‐Jun Event 1 11:00 AM 11 AM ‐ 1 PM ‐32.7 ‐42.2 ‐23.1 ‐42.4 47.0 ‐69.6%47.0
29‐Jun Event 1 12:00 PM 11 AM ‐ 1 PM ‐34.8 ‐44.0 ‐25.6 ‐45.2 47.0 ‐74.1%47.0
29‐Jun Event 1 1:00 PM 11 AM ‐ 1 PM ‐28.3 ‐37.1 ‐19.5 ‐36.8 49.0 ‐57.8%49.0
29‐Jun Event 1 2:00 PM 2 PM ‐ 5 PM  ‐49.2 ‐58.9 ‐39.4 ‐63.8 84.8 ‐58.0%89.3
29‐Jun Event 1 3:00 PM 2 PM ‐ 5 PM  ‐87.0 ‐96.8 ‐77.2 ‐113.0 144.3 ‐60.3%148.8
29‐Jun Event 1 4:00 PM 2 PM ‐ 5 PM  ‐82.7 ‐92.5 ‐73.0 ‐107.5 144.3 ‐57.4%148.8
29‐Jun Event 1 5:00 PM 2 PM ‐ 5 PM  ‐75.8 ‐85.3 ‐66.3 ‐98.5 144.3 ‐52.6%148.8
29‐Jun Event 1 6:00 PM 6 PM ‐61.7 ‐70.9 ‐52.5 ‐80.1 110.1 ‐56.0%110.1
8‐Jul Event 2 11:00 AM 11 AM ‐ 1 PM ‐48.7 ‐67.5 ‐29.9 ‐63.2 50.7 ‐96.0%50.7
8‐Jul Event 2 12:00 PM 11 AM ‐ 1 PM ‐49.6 ‐67.9 ‐31.3 ‐64.4 50.7 ‐97.8%50.7
8‐Jul Event 2 1:00 PM 11 AM ‐ 1 PM ‐39.0 ‐56.6 ‐21.4 ‐50.6 53.0 ‐73.6%53.0
8‐Jul Event 2 2:00 PM 2 PM ‐ 5 PM  ‐71.2 ‐90.3 ‐52.1 ‐92.4 90.5 ‐78.6%96.5
8‐Jul Event 2 3:00 PM 2 PM ‐ 5 PM  ‐119.8 ‐138.9 ‐100.6 ‐155.5 154.8 ‐77.4%160.7
8‐Jul Event 2 4:00 PM 2 PM ‐ 5 PM  ‐114.5 ‐133.5 ‐95.5 ‐148.7 154.8 ‐74.0%160.7
8‐Jul Event 2 5:00 PM 2 PM ‐ 5 PM  ‐104.9 ‐123.5 ‐86.2 ‐136.2 154.8 ‐67.8%160.7
8‐Jul Event 2 6:00 PM 6 PM ‐85.7 ‐103.7 ‐67.6 ‐111.3 118.9 ‐72.1%118.9
15‐Jul Event 3 11:00 AM 11 AM ‐ 1 PM ‐41.3 ‐60.1 ‐22.5 ‐53.6 50.7 ‐81.4%50.7
15‐Jul Event 3 12:00 PM 11 AM ‐ 1 PM ‐44.2 ‐62.6 ‐25.9 ‐57.4 50.7 ‐87.2%50.7
15‐Jul Event 3 1:00 PM 11 AM ‐ 1 PM ‐43.1 ‐61.0 ‐25.2 ‐56.0 53.0 ‐81.3%53.0
15‐Jul Event 3 2:00 PM 2 PM ‐ 5 PM  ‐65.6 ‐84.7 ‐46.5 ‐85.2 76.6 ‐85.6%96.5
15‐Jul Event 3 3:00 PM 2 PM ‐ 5 PM  ‐107.0 ‐126.2 ‐87.8 ‐139.0 140.9 ‐76.0%160.7
15‐Jul Event 3 4:00 PM 2 PM ‐ 5 PM  ‐104.4 ‐123.5 ‐85.4 ‐135.6 140.9 ‐74.1%160.7
15‐Jul Event 3 5:00 PM 2 PM ‐ 5 PM  ‐100.1 ‐118.8 ‐81.4 ‐130.0 140.9 ‐71.0%160.7
15‐Jul Event 3 6:00 PM 6 PM ‐86.6 ‐104.7 ‐68.6 ‐112.5 118.9 ‐72.9%118.9
16‐Jul Event 4 7:00 AM 7 AM‐ 10 AM ‐37.5 ‐56.4 ‐18.6 ‐48.7 17.1 ‐219.1%17.1
16‐Jul Event 4 8:00 AM 7 AM‐ 10 AM ‐39.9 ‐58.3 ‐21.5 ‐51.8 17.1 ‐233.2%17.1
16‐Jul Event 4 9:00 AM 7 AM‐ 10 AM ‐35.2 ‐52.8 ‐17.5 ‐45.7 17.1 ‐205.5%17.1
16‐Jul Event 4 10:00 AM 7 AM‐ 10 AM ‐0.2 ‐8.8 8.3 ‐0.3 17.1 ‐1.4%17.1
16‐Jul Event 4 11:00 AM 11 AM ‐ 1 PM 0.0 ‐8.4 8.4 0.0 42.6 0.0%42.6
16‐Jul Event 4 12:00 PM 11 AM ‐ 1 PM 0.1 ‐8.0 8.2 0.2 42.6 0.3%42.6
16‐Jul Event 4 1:00 PM 11 AM ‐ 1 PM 0.4 ‐7.4 8.2 0.5 44.9 0.9%44.9
16‐Jul Event 4 2:00 PM 2 PM ‐ 5 PM  ‐60.6 ‐79.8 ‐41.5 ‐78.7 72.6 ‐83.5%88.4
16‐Jul Event 4 3:00 PM 2 PM ‐ 5 PM  ‐100.5 ‐119.8 ‐81.2 ‐130.5 135.9 ‐74.0%151.7
16‐Jul Event 4 4:00 PM 2 PM ‐ 5 PM  ‐98.6 ‐117.7 ‐79.4 ‐128.0 135.9 ‐72.5%151.7
16‐Jul Event 4 5:00 PM 2 PM ‐ 5 PM  ‐93.4 ‐112.2 ‐74.6 ‐121.3 135.9 ‐68.7%151.7
16‐Jul Event 4 6:00 PM 6 PM ‐77.7 ‐95.9 ‐59.6 ‐101.0 109.9 ‐70.8%109.9
19‐Jul Event 5 7:00 AM 7 AM‐ 10 AM ‐37.4 ‐56.2 ‐18.5 ‐48.5 17.1 ‐218.5%17.1
19‐Jul Event 5 8:00 AM 7 AM‐ 10 AM ‐40.2 ‐58.5 ‐21.8 ‐52.2 17.1 ‐234.9%17.1
19‐Jul Event 5 9:00 AM 7 AM‐ 10 AM ‐39.8 ‐57.5 ‐22.2 ‐51.7 17.1 ‐232.8%17.1
19‐Jul Event 5 10:00 AM 7 AM‐ 10 AM ‐18.1 ‐26.7 ‐9.5 ‐23.5 17.1 ‐105.8%17.1
19‐Jul Event 5 11:00 AM 11 AM ‐ 1 PM ‐17.9 ‐26.3 ‐9.4 ‐23.2 42.6 ‐41.9%42.6
19‐Jul Event 5 12:00 PM 11 AM ‐ 1 PM ‐16.7 ‐24.9 ‐8.6 ‐21.7 42.6 ‐39.3%42.6
19‐Jul Event 5 1:00 PM 11 AM ‐ 1 PM ‐14.6 ‐22.4 ‐6.7 ‐18.9 44.9 ‐32.4%44.9
19‐Jul Event 5 2:00 PM 2 PM ‐ 5 PM  ‐61.3 ‐80.4 ‐42.2 ‐79.6 71.1 ‐86.2%88.4
19‐Jul Event 5 3:00 PM 2 PM ‐ 5 PM  ‐103.1 ‐122.3 ‐83.9 ‐133.9 134.5 ‐76.7%151.7
19‐Jul Event 5 4:00 PM 2 PM ‐ 5 PM  ‐101.2 ‐120.3 ‐82.2 ‐131.5 134.5 ‐75.3%151.7
19‐Jul Event 5 5:00 PM 2 PM ‐ 5 PM  ‐98.6 ‐117.3 ‐80.0 ‐128.1 134.5 ‐73.4%151.7
19‐Jul Event 5 6:00 PM 6 PM ‐83.3 ‐101.4 ‐65.3 ‐108.2 109.9 ‐75.8%109.9
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 20 of 22
Jeff Bumgarner Page 21 of 22
February 24, 2011
Date Event Hour Block Estimated Lower Upper Estimated 
Load Bound 95% Bound 95% Load Opt‐out 
Reduction ‐ Confidence Confidence Reduction adjusted 
5 Interval Interval ‐ All Idaho Nominal Nominal 
Substations Irrigation Load Realization Load 
(MW) (MW) (MW) Rate (MW) 
20‐Jul Event 6 7:00 AM 7 AM‐ 10 AM ‐46.4 ‐65.3 ‐27.5 ‐60.2 17.1 ‐271.1%17.1
20‐Jul Event 6 8:00 AM 7 AM‐ 10 AM ‐48.3 ‐66.6 ‐29.9 ‐62.7 17.1 ‐282.0%17.1
20‐Jul Event 6 9:00 AM 7 AM‐ 10 AM ‐44.4 ‐62.1 ‐26.8 ‐57.7 17.1 ‐259.7%17.1
20‐Jul Event 6 10:00 AM 7 AM‐ 10 AM ‐12.7 ‐21.4 ‐4.1 ‐16.5 17.1 ‐74.5%17.1
20‐Jul Event 6 11:00 AM 11 AM ‐ 1 PM ‐13.6 ‐22.1 ‐5.2 ‐17.7 42.6 ‐32.0%42.6
20‐Jul Event 6 12:00 PM 11 AM ‐ 1 PM ‐14.8 ‐23.0 ‐6.6 ‐19.2 42.6 ‐34.6%42.6
20‐Jul Event 6 1:00 PM 11 AM ‐ 1 PM ‐15.1 ‐23.0 ‐7.3 ‐19.7 44.9 ‐33.7%44.9
20‐Jul Event 6 2:00 PM 2 PM ‐ 5 PM  ‐71.5 ‐90.7 ‐52.3 ‐92.9 65.2 ‐109.6%88.4
20‐Jul Event 6 3:00 PM 2 PM ‐ 5 PM  ‐105.4 ‐124.8 ‐86.1 ‐136.9 128.6 ‐82.0%151.7
20‐Jul Event 6 4:00 PM 2 PM ‐ 5 PM  ‐102.0 ‐121.2 ‐82.8 ‐132.5 128.6 ‐79.3%151.7
20‐Jul Event 6 5:00 PM 2 PM ‐ 5 PM  ‐98.0 ‐116.9 ‐79.1 ‐127.3 128.6 ‐76.2%151.7
20‐Jul Event 6 6:00 PM 6 PM ‐82.2 ‐100.5 ‐63.9 ‐106.7 109.9 ‐74.8%109.9
26‐Jul Event 7 7:00 AM 7 AM‐ 10 AM ‐32.9 ‐51.7 ‐14.0 ‐42.7 17.1 ‐192.1%17.1
26‐Jul Event 7 8:00 AM 7 AM‐ 10 AM ‐36.1 ‐54.5 ‐17.8 ‐46.9 17.1 ‐211.3%17.1
26‐Jul Event 7 9:00 AM 7 AM‐ 10 AM ‐35.0 ‐52.6 ‐17.3 ‐45.4 17.1 ‐204.3%17.1
26‐Jul Event 7 10:00 AM 7 AM‐ 10 AM ‐10.4 ‐18.9 ‐1.8 ‐13.5 17.1 ‐60.7%17.1
26‐Jul Event 7 11:00 AM 11 AM ‐ 1 PM ‐11.0 ‐19.3 ‐2.6 ‐14.2 42.6 ‐25.7%42.6
26‐Jul Event 7 12:00 PM 11 AM ‐ 1 PM ‐11.1 ‐19.3 ‐3.0 ‐14.4 42.6 ‐26.1%42.6
26‐Jul Event 7 1:00 PM 11 AM ‐ 1 PM ‐12.2 ‐20.0 ‐4.4 ‐15.9 44.9 ‐27.2%44.9
26‐Jul Event 7 2:00 PM 2 PM ‐ 5 PM  ‐54.7 ‐73.8 ‐35.6 ‐71.0 76.9 ‐71.1%88.4
26‐Jul Event 7 3:00 PM 2 PM ‐ 5 PM  ‐89.7 ‐108.9 ‐70.5 ‐116.5 140.3 ‐63.9%151.7
26‐Jul Event 7 4:00 PM 2 PM ‐ 5 PM  ‐88.8 ‐107.9 ‐69.7 ‐115.3 140.3 ‐63.3%151.7
26‐Jul Event 7 5:00 PM 2 PM ‐ 5 PM  ‐85.3 ‐104.1 ‐66.5 ‐110.8 140.3 ‐60.8%151.7
26‐Jul Event 7 6:00 PM 6 PM ‐75.8 ‐94.1 ‐57.5 ‐98.4 109.9 ‐69.0%109.9
26‐Jul Event 8 7:00 AM 7 AM‐ 10 AM 24.1 14.1 34.0 31.3 17.1 140.8%17.1
2‐Aug Event 8 8:00 AM 7 AM‐ 10 AM 25.3 15.6 34.9 32.8 17.1 147.7%17.1
2‐Aug Event 8 9:00 AM 7 AM‐ 10 AM 29.7 20.4 38.9 38.6 17.1 173.5%17.1
2‐Aug Event 8 10:00 AM 7 AM‐ 10 AM ‐2.4 ‐6.6 1.7 ‐3.1 17.1 ‐14.1%17.1
2‐Aug Event 8 11:00 AM 11 AM ‐ 1 PM ‐2.0 ‐6.1 2.0 ‐2.6 42.6 ‐4.8%42.6
2‐Aug Event 8 12:00 PM 11 AM ‐ 1 PM ‐2.1 ‐6.0 1.8 ‐2.7 42.6 ‐4.9%42.6
2‐Aug Event 8 1:00 PM 11 AM ‐ 1 PM ‐3.1 ‐6.8 0.6 ‐4.0 44.9 ‐6.9%44.9
2‐Aug Event 8 2:00 PM 2 PM ‐ 5 PM 11.6 1.5 21.7 15.1 83.6 13.9%88.4
2‐Aug Event 8 3:00 PM 2 PM ‐ 5 PM 3.8 ‐6.3 14.0 5.0 146.9 2.6%151.7
2‐Aug Event 8 4:00 PM 2 PM ‐ 5 PM  ‐3.0 ‐13.1 7.0 ‐3.9 146.9 ‐2.1%151.7
2‐Aug Event 8 5:00 PM 2 PM ‐ 5 PM  ‐6.7 ‐16.5 3.2 ‐8.6 146.9 ‐4.5%151.7
2‐Aug Event 8 6:00 PM 6 PM 1.3 ‐8.2 10.9 1.7 109.9 1.2%109.9
5‐Aug Event 9 7:00 AM 7 AM‐ 10 AM ‐8.0 ‐18.0 2.0 ‐10.4 16.9 ‐47.3%16.9
5‐Aug Event 9 8:00 AM 7 AM‐ 10 AM ‐8.7 ‐18.4 1.0 ‐11.3 16.9 ‐51.4%16.9
5‐Aug Event 9 9:00 AM 7 AM‐ 10 AM ‐6.9 ‐16.2 2.4 ‐9.0 16.9 ‐41.0%16.9
5‐Aug Event 9 10:00 AM 7 AM‐ 10 AM ‐8.2 ‐12.4 ‐3.9 ‐10.6 16.9 ‐48.5%16.9
5‐Aug Event 9 11:00 AM 11 AM ‐ 1 PM ‐8.3 ‐12.5 ‐4.2 ‐10.8 42.0 ‐19.9%42.0
5‐Aug Event 9 12:00 PM 11 AM ‐ 1 PM ‐8.6 ‐12.6 ‐4.7 ‐11.2 42.0 ‐20.5%42.0
5‐Aug Event 9 1:00 PM 11 AM ‐ 1 PM ‐10.0 ‐13.7 ‐6.2 ‐12.9 44.2 ‐22.5%44.2
5‐Aug Event 9 2:00 PM 2 PM ‐ 5 PM  ‐19.3 ‐29.5 ‐9.2 ‐25.1 79.6 ‐24.3%87.1
5‐Aug Event 9 3:00 PM 2 PM ‐ 5 PM  ‐41.7 ‐51.9 ‐31.5 ‐54.2 142.0 ‐29.4%149.6
5‐Aug Event 9 4:00 PM 2 PM ‐ 5 PM  ‐42.2 ‐52.4 ‐32.0 ‐54.8 142.0 ‐29.7%149.6
5‐Aug Event 9 5:00 PM 2 PM ‐ 5 PM  ‐39.1 ‐49.1 ‐29.1 ‐50.8 142.0 ‐27.5%149.6
5‐Aug Event 9 6:00 PM 6 PM ‐31.5 ‐41.2 ‐21.9 ‐41.0 108.3 ‐29.1%108.3
24‐Aug Event 10 7:00 AM 7 AM‐ 10 AM ‐25.5 ‐35.5 ‐15.6 ‐33.2 16.9 ‐151.4%16.9
24‐Aug Event 10 8:00 AM 7 AM‐ 10 AM ‐24.9 ‐34.6 ‐15.2 ‐32.3 16.9 ‐147.6%16.9
24‐Aug Event 10 9:00 AM 7 AM‐ 10 AM ‐22.1 ‐31.4 ‐12.9 ‐28.8 16.9 ‐131.3%16.9
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 21 of 22
Jeff Bumgarner Page 22 of 22
February 24, 2011
The Cadmus Group, Inc. 720 SW Washington Street, Suite 400, Portland, OR 97205 503.228.2992 Fax [phone] An Employee-Owned Company www.cadmusgroup.com
Date Event Hour Block Estimated 
Load 
Reduction ‐ 
5 
Substations 
(MW) 
Lower 
Bound 95% 
Confidence 
Interval 
Upper 
Bound 95% 
Confidence 
Interval 
Estimated 
Load 
Reduction 
‐ All Idaho 
Irrigation 
(MW) 
Opt‐out 
adjusted 
Nominal 
Load 
(MW) 
Realization 
Rate 
Nominal 
Load 
(MW) 
24‐Aug Event 10 10:00 AM 7 AM‐ 10 AM ‐5.0 ‐9.3 ‐0.7 ‐6.5 16.9 ‐29.6%16.9
24‐Aug Event 10 11:00 AM 11 AM ‐ 1 PM ‐5.2 ‐9.3 ‐1.0 ‐6.7 42.0 ‐12.3%42.0
24‐Aug Event 10 12:00 PM 11 AM ‐ 1 PM ‐5.5 ‐9.5 ‐1.5 ‐7.1 42.0 ‐13.1%42.0
24‐Aug Event 10 1:00 PM 11 AM ‐ 1 PM ‐6.0 ‐9.8 ‐2.2 ‐7.8 44.2 ‐13.5%44.2
24‐Aug Event 10 2:00 PM 2 PM ‐ 5 PM  ‐32.0 ‐42.1 ‐21.9 ‐41.6 81.9 ‐39.1%87.1
24‐Aug Event 10 3:00 PM 2 PM ‐ 5 PM  ‐40.8 ‐50.9 ‐30.6 ‐52.9 144.3 ‐28.2%149.6
24‐Aug Event 10 4:00 PM 2 PM ‐ 5 PM  ‐39.0 ‐49.0 ‐28.9 ‐50.6 144.3 ‐27.0%149.6
24‐Aug Event 10 5:00 PM 2 PM ‐ 5 PM  ‐41.3 ‐51.1 ‐31.4 ‐53.6 144.3 ‐28.6%149.6
24‐Aug Event 10 6:00 PM 6 PM ‐31.8 ‐41.3 ‐22.3 ‐41.3 108.3 ‐29.4%108.3
26‐Aug Event 11 7:00 AM 7 AM‐ 10 AM ‐20.4 ‐30.4 ‐10.5 ‐26.5 16.9 ‐121.1%16.9
26‐Aug Event 11 8:00 AM 7 AM‐ 10 AM ‐19.0 ‐28.7 ‐9.3 ‐24.7 16.9 ‐112.7%16.9
26‐Aug Event 11 9:00 AM 7 AM‐ 10 AM ‐18.4 ‐27.6 ‐9.1 ‐23.8 16.9 ‐108.8%16.9
26‐Aug Event 11 10:00 AM 7 AM‐ 10 AM ‐2.5 ‐6.7 1.7 ‐3.2 16.9 ‐14.8%16.9
26‐Aug Event 11 11:00 AM 11 AM ‐ 1 PM ‐2.0 ‐6.0 2.1 ‐2.6 42.0 ‐4.7%42.0
26‐Aug Event 11 12:00 PM 11 AM ‐ 1 PM ‐2.6 ‐6.5 1.2 ‐3.4 42.0 ‐6.3%42.0
26‐Aug Event 11 1:00 PM 11 AM ‐ 1 PM ‐2.5 ‐6.2 1.2 ‐3.3 44.2 ‐5.7%44.2
26‐Aug Event 11 2:00 PM 2 PM ‐ 5 PM  ‐31.9 ‐42.0 ‐21.8 ‐41.5 84.0 ‐38.0%87.1
26‐Aug Event 11 3:00 PM 2 PM ‐ 5 PM  ‐44.3 ‐54.4 ‐34.1 ‐57.5 146.4 ‐30.2%149.6
26‐Aug Event 11 4:00 PM 2 PM ‐ 5 PM  ‐40.5 ‐50.6 ‐30.4 ‐52.6 146.4 ‐27.7%149.6
26‐Aug Event 11 5:00 PM 2 PM ‐ 5 PM  ‐37.1 ‐47.0 ‐27.1 ‐48.1 146.4 ‐25.3%149.6
26‐Aug Event 11 6:00 PM 6 PM ‐30.6 ‐40.2 ‐20.9 ‐39.7 108.3 ‐28.2%108.3

ID PAC-E-11-12
IIPA 34 Attachment IIPA 34 -3
Attach IIPA 34 -3.pdf Page 22 of 22