20100820PAC to IIPA 80-82.pdf

~ROCKY MOUNTAIN
POWER
A DISION OF PAFlP
REGE
f'~
~- )
201 South Ma. Suite 2300
Salt Lake City. Uth 84111
~~I. 9: 44in\U ~UG 20 .'.11
August 19,2010
Eric L. Olsen ISB# 4811
RACIN, OLSON, NYE, BUDGE &
BAILEY, CHATERED
P.O. Box 1391; 201 E. Center
Pocatello, Idaho 83204-1391
RE: ID PAC-E-1O-07
LIP A Data Request (80-82)
Please fid enclosed Rocky Mounta Power's responses to IIPA Data Requests 80-82.
If you have any questions, please feel free to call me at [phone].
Sincerely,I~LO~/~
J. Ted Weston
Maager, Regulation
Enclosur:
C.c. Rady Budge/Monsanto
Jea Jewel1lUC
Anthony Yanel
Ben Oto
PAC-E-10-07/Rocky Mountain Power
August 19,2010
LIP A Data Request 80
LIP A Data Request 80
On page 12 (Table Twenty) of the Company's 2009 final report regarding
"Schedule 72 & 72A Idaho Irrigation Load Control Programs", there are figures
regarding the amount (kW) ofloadthat was avoided at various times/dates. Are
these values at sales or generation level? If at sales level, what loss values should
be applied in order to reflect the amount of load that was avoided at the
generation level?
Response to lIP A Data Request 80
On page 12 (Table Twenty) of Schedule 72 & 72A Irrigation Load Control
Programs Final Report 2009 Credit R.dednitiatives, the aggregated values
provided are at sales leveL. To convert the 2009 sales data to generation level data
in the rate case, the Company used the demand loss factor for the secondar
voltage level, as determined by the 2007 MAC loss study. That value is 1.11642.
The associated energy loss factor is 1.10148.
Recordholder:
Sponsor:
Jeff Bumgarer
To Be Determined
:.:-', \.
.~ ,'¡
l¡ .
P AC-E-1 0-07/Rocky Mountain Power
August 19,2010
IIPA Data Request 81
LIP A Data Request 81 t'."
",
, .~,
-:"ü
On page 12 (Table Twenty) of the Company's 2009 final report regarding
"Schedule 72 & 72A Idaho IrrigationLoad Control Programs", there are figures
regarding the amount (kW) ofload that was avoided at various times/dates. The
combination of avoided load for the åispatched and scheduled forward programs
amounted to approximately 240 MW. Please answer the following:
a. The Company's response to lIP A request 24c contaned hourly Idaho retail load
data that was used in the NPC model used in this case. The data did not show any
lowering of the Idaho load that would be reflective of the impact of the Irrigation
Load Control Program. Why not?
b. What would be the impact on the Company's NPCs if the load redllctions in Idaho
as well as in other jurisdictions were lincl:uØed in the loads used to develop theNPC results? .
Response to IIPA Data Request 81
a. The figues in Table Twenty do not represent load that was avoided. Please refer
to page 7 of the same report, which states:
"Based on previous research that showed energy use is 'shifed'
rather than 'avoided', lost revenuis are not included as a cost and, _.l.; .:.
energy savings are not applicable( as ipdicated above. "
b. These types of load control programs shift the timing of load, but do not reduce
energy. The Company has not forecasted the specific time when these events wil
occur and therefore has not modeled this in its hourly data. The hourly load used
in the net power cost study is shaped based on averages of actual historical load.
Recordholder:
Sponsor:
Pete Eelkema
Hui Shu
"
:(, .t- "
ii l
PAC-E-1 0-07/Rocky Mountain Power
August 19,2010
LIP A Data Request 82
lIP A Data Request 82
Please answer the following with respect to the Company's response to LIP A
Request 12:
a. On 7/17/2009 there was listed "Event 1 "regarding the Utah Cool Keeper
Program. The total kW listed was 81,564. Does the 81,564 figure represent the
sumation of the individual customer loads that were interrpted or was it the net
impact on the system over the entire hour? Was the 81,564 kW figue at sales or
generation level?
b. How were the "Res kW" figures developed or each event and each hour within
the event? Please provide a detailed example. Please provide sufficient detail to
explain why/how the "Res kW" figure changes between events and during
different hours durng the same event.
c. What was the total credit paid to thèse custQmers in 2008 and 2009 for their
paricipation in the program?
d. What was the total cost to the Company (absent the credit paid) of these programs
for 2008 and 2009?
e. Why did the value for "Com kW" drop so much in 2009 and why did it not var
as it did in 2008?
Response to lIP A Data Request 82
.'..1':1
(a) The value represents an estimate of the net impact on the system for that event
at that temperatue and was derived from meter data from stratified random
sample of Cool Keeper paricipating sites. The meter data was extrapolated to the
remaining paricipating sites to arve at the 81,564 kilowatt value for the event.
The value is at sales leveL.
(b) Table 1 below details the load reduction results from 2009 curilment events.
The residential results were calculated by averaging of the results two
measurement methods; the "differ~nRing rnethod" and the "duty cycle" method.
The differencing method involves taing thè'difference in average load usage
between the metered results of a controlled sample and non-controlled sample
(grup A and B) of paricipating sites. The sample group sites relied upon for
these measurements are equipped with time based metering and are representative
of the general population of paricipating sites in terms climate zones represented,
size of equipment, housing types, operating conditions, and other relevant factors.
The meter data is collected in 5 minute intervals durng control dispatch events
and the average difference for the honr is;the result from the differencing method.
The duty cycle method involves a simulated reduction of the non-controlled group
l.
PAC-E-1 0-07/Rocky Mountain Power
August 19,2010
LIP A Data Request 82 .,11
based on a nationally recognized mathematical calculation, using sample group
meter data to derive an estimated impact result.
The commercial results for 2009 shown in Table 1 are based only on the duty
cycle method. Due to a smaller number of paricipating commercial sites coupled
with: 1) greater variability of larger compressors; and 2) customer behavior it
wasn't possible to derive a statisticatly relevant baseline in order to apply the
differencing methodology. ' ,. . ,
The varation in load reduction by hour çid temperature is typical of air
conditioning (AlC) equipment which in addition to being temperatue dependent
is subject to constant fluctuations based on individual customer usage patterns and
equipment specifications. Table 2 shows how AC load vares by outside
temperature. Notice also that at a given temperature, AC load can vary widely.
.1'q~le.I=?QQ?Loa4lnipqctRe~uJt~
Co Ke - 2I Di Ev'" .. nd I! ..ip_ _v ~ 1._. .."...........~~......"
1ilOl~~jj4iil~..Ii~Ill~lJii'l.~atd~aiitç ... ............ ....... ....... .................................................. .. .... . ........ ..... . .. .I3loiii3p~III:I1.ii..ni. iisyUl1l97 ..~i.. ai ..idi
-=1 . i~iJ2!~I!l:lrr1! .~i~1l "idTRej 90105:~4t!il~~.~2 i~,n:oo: . 93:~I47l N.tllI!i:OO .!i . ~. ~1l: ti..16:00 ~ '~""'" 47-l!fòl...8I..1i:ÎNT9': 894J'-l .!i.~~3 . 112~/ 'n:ØI9'l,19!fiA lJ; 41l 11.07l&B:C . .... ii''¡;:OIjll~¡Ni~jfiA 8919: 4n' .. 0
'6:lOlll~ .llI~¡fiA.. .......! ........... .. .8!19' ..... ... ...... ....... ..47:.................... . .......0
~.N~~.:~~~:*~=~~=4:~~àël4i.5S.ii~¡~~~~~~;
.~..CI~ti'lI SUda.....il1l.llofai.sw on ll ØI da
Table 2 - AC Load and Temperature 2009
2.25
2.00
:i 1.75
e 1.50
"'
~ 1.25
~ 1.00
~ 0.75
~ 0.50
0.25
0.00
50 60 70 80 90 100 110
Temperature (F)
i,
y = O,0466x - 3.0078
R2=O,8314
!I~ !~; '.:,
¡--~
'.,
PAC-E-1 0-07/Rocky Mountain Power
August 19,2010
LIP A Data Request 82
Differencing Method Calculation
The differencing method involves randomly splitting of the sample meter group
into two groups; sample group Aaiasarple group B. The sample groups are.- . .,
designed to calculate load reduction-with \å 'lhinimum of 90% confdence level and
13% precision for residential sites and a 90% confidence level and 20% precision
for commercial sites. One group is selected to be controlled during a dispatch
event (A or B) while the other is left to ru as it would normally. These two
groups are alternated every other dispatch event to equalize any bias within the
sample groups. 20% of the sample group is replaced each year and the A and B
groups are randomized to fuher reduce any sample site bias and ensure the
sample group remains representative.~Measurements used are from dispatch
events and hours when the ambi~nftemperature is at or above 97 degrees F.
The average load in each 5 minute interval is taken for each group throughoutthe
course of a dispatch event. The differencing result for an event hour is derived
from the difference between the two groups' average load for that hour.
Duty-Cycle Calculation
The duty cycle method relies on metered data and the mathematical formula
(provided below) to simulate the impact of AlC equipment during a dispatch
event (dispatch events when the temperature is 97 degrees F or higher). Meter
data on the usage of the AlC equipment in the sample group not scheduled for
control (A or B) is collected forthtnoUF",irwediately preceding the dispatch
event as well as during the event an4:the S'Ó% duty cycle algorithm is applied tot, ' r., i . . . " ~'
determine the expected result. ..,.
The reduction for the control hour is the average of the curailment estimates for
the two 30-mInute time periods in that controlled hour. It can be calculated as
follows:
Diff= (Diff_0+DifC30)/2 where
Diff 0 is the difference for the 00-29 minutes of the hour
Diff- 30 is the difference for the 3'o-iSl) 111hutes of the hour
if(AvgBefiConnectedLoad ~ 0.'1ftheii"
DifCO = AvgDur_O - Min(AvgDUr-:O, 0¡5*ConnectedLoad)ebe . .
Diff_O = AvgDur _0- Min(AvgDur~O, 0.5* AvgBef)
endif
if(AvgBefiConnectedLoad ~ 0.1) then
Diff_30 = AvgDur_30 - Min(AvgDur_30, O.5*ConnectedLoad)
else
Diff_30 = AvgDur_30- Min(AvgDur_30, 0.5* AvgBef)
endif ,\ :!
The ConnectedLoad is set equa tö'di'99~Øercentile operating load for ths meter
over the entire sumer for those intervals tht are greater than 700 Watts. If the
P AC-E-1 0-07/Rocky Mountain Power
August 19,2010
LIP A Data Request 82
nominal capacity of the air conditioner i~known and the 99th percentile is 1.5
times greater than that value, then the 95th percentile is used.
The AvgBefis the average of the 5-minute interval data for 1 hour prior to the
event.
The AvgDur_O is the average of the 5-minute interval data for minutes 00-29 for
each complete hour durng to the event.
The AvgDur_30 is the average of the 5-minute interval data for minutes 30-59 for
each complete hour during to the event.
The Diff is the calculated expected curilment difference obtained by applying
the 50% duty cycle method for each,$O-miIlute time period.
, 'If,\iExamples:.
1. Curailment event to take place from 2:30pm - 5:00 pm
A vgBef is the average of the interval data for the intervals with sta times of
1:30pm - 2:25 pm (12 intervals)
There would be 2 differences calculated (for the 3pm and 4pm hours)
AvgDur for the 3pm hour is the average of the 2 difference values that were
calculated for the 3:00pm-3:25pm intervals and the 3:30pm-3:55pm intervals.
A vgDur for the 4pm hour is the avera~ú~ of the 2 difference values that were
calculated for the 4:00pm-4:25p:r intervals and the 4:30pm-4:55pm intervals.
2. Curailment event to tae place from 3:00pm - 5:00pm
A vgBef is the average of the intel:al data for the intervals with star times of
2:00pm - 2:55 pm (12 intervals)
There would be 1 difference calculated (for the 4pm hour). There would not be a
difference calculated for the 3pm hour because the first 30 minutes of a control
event are excluded and only estimates for complete hours are calculated.
A vgDur for the 4pm hour is the average of the 2 difference values that were
calculated for the 4:00pm-4:25pm iiittnrals and the 4:30pm-4:55pm intervals.
'1.
Temperature Bin Calculation i.
For each event hour at or above 97 degrees F where a 50% ADI algorith is used,
the average reduction for that hour is taen as the average between the value
obtained from the differencing method and the value obtained from the duty-cycle
method. The results from all hours at a given temperatue level (at or above 97
degrees F) are then averaged into temperatue bins. Table 3, Sample Event
Calculation, provides an example of how: the result of the differencing and duty
cycle methods are used to estimate the average reduction for each hour and
temperatue durng a given dispatchrvept.¡
PAC-E-10-07/Rocky Mountain Power
August 19,2010
LIP A Data Request 82 .
a e -ampie vent a cu atlOn
Event Differencing Method Duty-Cycle Method
Hour Temp Reduction Reduction Average
1 97 0.95 0.95 0.95
2 98 1.10 ' ¡1.15 1.13
3 98 1.05 "c::1.01 1.03
4 103 1.20
"1:1.15 1.18
,
r. bl 3 S 1 E eii .
Table 4, Temperatue Bin Example Results, demonstrates how the event results
would then be averaged by temperatue bin to yield the average event result.
Table 4 r.-em perature in xamTJie
Temp Reduction
Bin (kW)
97 ;1",'0.95
98 1.08
103 1.18
Avera~e 1.07
B' E 1 Results
(c) Cool Keeper paricipation credits paid; 2008 = $1,543,950 and 2009 =
$1,721,805. As noted in Uta's Schedule1l4 customers are paid $20 or $40 per
air conditioner per season depending on the size of their air conditioner.
'iiL.
(d) Cool Keeper expenses less paricipation credits; 2008 = $5,634,898 and 2009
= $8,094,728.
(e) Many factors can impact Cool Keeper results event to event but two of the
most prominent factors are temperatue (preceding and during an event) and event
length. The reason for the dramatic drop in estimated impact from the July 17,
2009 event from those recorded in 2008 was the events short duration and the fact
it occured at 3PM on a day following a cool night. Both conditions led to atypical
impact results from what the systemwould have likely provided in terms of
impact under normal 97 degree F'or.abov'e:'Conditions. Furhermore, the results of
one data point are insuffcient to draw any seasonal value. The varation in 2008
data is what you would expect of eveÎitswithn a given season due to the varng
temperatues (preceding and durng the events) and event durations. The 2008
data collectively was robust enough to derive at a seasonal average for both
residential and commercial paricipating loads.
Recordholder:
Sponsor:
JeffBumgar~r ,:,
"'1 ",To Be Determmed