20141023Yourkowski Direct.pdf

Benjamin otto (lSB No. 8292)
710 N 6th Street
Boise, ID 83701
Ph: [phone] x 12
Fax: [phone]
botto@ idahoconservation.org
Attomey for the ldaho Conservation League
-..- --{,,, i-(. -.. I "." i'-', i;liuv'--
?[iq OCT 23 PH \r 33
. ..,, tr' I itt,.,uir,iii,li tc;l
. i i--l t I -'*
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
IN THE MATTER OF IDAHO I cAsE NO. rpc-E-14-18POWER COMPAIIY',S ) -
APPLICATION TO IMPLEMENT )
soALR TNTEGRATTON RATES AND ) mano coNSERvATroN LEAGUECHARGES. )
DIRECT TESTIMONY
CAMERON YOURKOWSKI
FILED
ocroBER 23,2014
I Q. Please state your name, affiliation, and reason for this testimony.
2 A. My name is Cameron Yourkowski. I have been employed by Renewable Northwest
3 (formerly Renewable Northwest Project) for the past seven years, primarily working on
4 renewable energy transmission and integration issues. The purpose of my testimony in this
5 docket is to review the technical details and merits of Idaho Power Company's ("ldaho Power")
6 Solar Integration Study ("Study'") and the solar integration costs and rates derived therein.
7
8 Q. Please describe your involvement in Idaho Power's Solar Integration Study.
9 A. I participated in the Technical Review Committee ("TRC') involved in the development of
l0 the Study and have reviewed the final Study, the accompanying testimony of Philip DeVol, and
11 data responses submitted in connection with this docket. Initially, fbrmer Renewable Northwest
12 stafTmember, Jimmy Lindsay, participated in the TRC and kept me apprised of TRC
l3 developments and discussions. In April of 2014, following Mr. Lindsay's departure from
14 Renewable Northwest. I began participating directly in the TRC.
15
16 Q. Please summarize your professional opinion of the TRC process.
17 A. As I stated in my June 6.2014 comments to ldaho Power and the TRC. included as Exhibit
l8 201. Renewable Northwest appreciates the opportunity to participate in the TRC and work
l9 constructively and cooperatively with ldaho Power on solar integration issues. The TRC process
20 and the Study itself had both positive and negative aspects. On the positive side. I appreciate
2l that ldaho Power initially followed a set of principles for effectively engaging a technical review
22 committee. The study also did a good job compiling a solar generation data set that is a fair
23 representation of expect actual solar generation in Idaho Power's service territory. However,
IPC-E- l4- l8
YOURKOWSKI, DI
ocToBER 23,20t4 I
I when the Study began analyzing the integration of the solar generation data into the broader
2 system, both the TRC process and ldaho Power's analysis began to break down, and ultimately
3 ended up relying on a flawed methodology.
4
5 Q. Have you reviewed the "Principles for Technical Review (TRC) Involvement" document
6 cited by Idaho Power as a basis for the Study process?
7 A. Yes. I have. This document, authored by the National Renewable Energy Laboratory
8 ("NREL") and the Utility Variable-generation Integration Group ("UVIG"), describes important
9 principles to guide effective participation in a Technical Review Committee. I have included
10 this "Principles" document as Exhibit202 to my testimony.
11
12 Q. Please describe some of the important principles from this document that are relevant
l3 to your critique of the TRC process.
14 A. Most of the important functions and requirements of a working TRC are on the second page
l5 of the Principles document. Four of the principles are particularly relevant here:
16 I ) The TRC will ensure that the findings are based entirely on facts and accurate
17 engineering and science. Project sponsors need to embrace this aim so that the results
l8 and findings are objectively developed and not skewed to support any desired outcome.
19 2) The TRC requires access to all relevant information needed to properly evaluate the
20 work and the results.
21 3) The TRC requires assurance that project sponsors will describe the project as having
22 the benefit of expert review by a TRC only if the TRC has clearly expressed its
23 acceptance of and agreement with the results of the study.
IPC-E,r4-18
YOURKOWSKI, DI
ocToBER 23,2014
| 4) The TRC requires assurance that, in the event agreement is not reached by the TRC
2 and other project participants, any reference to the TRC will be removed from the final
3 report and any associated documents or publicity.
5 Q. In your view, did Idaho Power's TRC process live up to these principles?
6 A. No. While the TRC process adhered to some of the principles, it departed from certain critical
7 principles. The TRC process started out in a very productive and congenial manner and helped
8 Idaho Power do a good job collecting and developing a good solar data set. However, when the
9 Study moved into the analysis phase:
10 l) The TRC was not able to ensure the findings of the Study were based entirely on facts
11 and accurate engineering and science.
12 2) The TRC did not have access to all relevant information needed to properly evaluate
13 the work and the results.
14 3) The TRC has not clearly expressed its acceptance of and agreement with the results of
15 the Study.
16 4) Idaho Power has not accurately portrayed the comments and level of support in the
17 final Study, associated documents and publicity.
l8
19 Q. Please summarize your view of the TRC process as it relates to these principles.
20 A. The TRC could have provided more value if these principles were adhered to. Ultimately, the
2l accuracy and meaningfulness of the Study was compromised because these principles were not
22 adhered to. Adhering to these principles in the future is a necessary element to ensure that future
23 integration studies provide a more accurate estimate of solar integration costs.
IPC-E-r4-18
YOURKOWSKI, DI
ocToBER 23,20t4 3
1
2
3
4
5
6
7
8
9
l0
ll
t2
13
l4
15
16
17
l8
19
20
2t
22
Q. Please summarize what you view as the positive aspects of this Study.
A. In general, the Study did a good job gathering solar data and modeling the output and
performance of different solar buildout scenarios. As I described in my commentsr to the TRC, I
view the positive aspects of this Study to be:
I . Solar Data: Based on my understanding of the process and ef-fort prior to my joining the
TRC, I think that Idaho Power has done a good job collecting a quality set of solar data
for ldaho Power's service territory for use in the Study.
2. Solar Diversity: The Study has appropriately analyzed both concentrated and dispersed
solar buildout scenarios. The "diversity value" associated with dispersed buildouts
significantly intluences the need fbr balancing reserves and, based on the information
available to me at this time, the Study has captured this effect sufficiently.
3. Wavelet Transformation: The Study's use of the Sandia National Lab's "wavelet-based
variability model" to transform the raw single-point irradiance data into plant-level
generation data is appropriate. This data transformation is important because the raw
single-point irradiance data always exhibits more variability than actual plant-level
generation will. The wavelet model is the best approach to handling this data
transformation issue that I am aware of.
4. Hour-Ahead Forecast Methodology: The Study incorporated a lot of positive work on
developing an hour-ahead fbrecast for solar generation. In concept, the use of a
persistence forecast adjusted forthe known curvature ofthe irradiance curve, based on a
"clear sky index" for that day ofthe year. is a sound approach.
' Exhibit 20 l.
IPC-E-14- l8
YOURKOWSKI, DI
ocToBER 23,20t4
I Q. Please describe in more detail where and how the TRC process broke down.
2 A. The TRC process associated with Idaho Power's Solar Integration Study did not provide for a
3 thorough rbview of the Study methodology or its findings. The compressed timeline at the end
4 of the process--{uring the analysis phase-diminished the TRC members' ability to
5 comprehensively review the details of this Study and the merits of its findings. In order for
6 ldaho Power to work constructively and cooperatively with stakeholders, it must be receptive to
7 TRC member input on best practices, and must build time into the study process to adjust the
8 study methodology to account for flaws or shortcomings identified by TRC members.
9 The most important process failure was the rushed nature of the integration study design
10 and the production cost modeling which started and abruptly concluded in the spring of 2014.
I I Beginning in May of 2014,ldaho Power began rushing the process in order to produce results
12 more quickly. In f'act. Idaho Power held only one TRC meeting. on May 16.2014, to review
13 ldaho Power's proposed study design and only one meeting on May 29,2014, to review the
14 results of the single production cost model run and resulting outputs. On June 2. ldaho Power
l5 provided the TRC with a Study draft to review. Following this, on June 5, 2014, the TRC
l6 received an email from Idaho Power, included here as exhibit 203. stating that it intended to
17 finalize the Study regardless of the comments received from the TRC. On June 6th.2014,1
18 submitted my written comments. On June l7th.2014.ldaho Power filed the Study with the
l9 Commission. The rushed nature of this phase of the Study significantly limited the ability of the
20 TRC to provide value during the most important phase of the Study.
21 Another important failure is Idaho Power's disrnissal of the critiques it received from
22 TRC members on the draft Study as mere items for "future study." The problem is that some of
23 these critiques identified fundamental flaws in the methodology that undermine the accuracy of
IPC.E.I4.I8
YOURKOWSKI, DI
ocToBER 23,2014
I
2
3
4
5
6
7
8
9
10
11
t2
t3
t4
15
16
L7
18
t9
20
21
22
23
24
25
26
27
the Study. For example. Idaho Power did not adjust the study methodology to account for the
basic concept that the estimated scheduling error of solar facilities should be netted with the
scheduling error ofother resources and load.
One member of the TRC commented:
In the report, it is important that the text provide an accurate representation of the
elements of the study that were available and reviewed by the TRC and those that were
not. I stated in the last meeting of the TRC that th is is not a study but rather it is a
document presenting the results of a single computer model run of an internal Idaho
Power computer model that was not reviewed by the TRC for accuracy, sensitivity and
variability.
In conclusion. the items that are Iisted as "Phase II" study tasks are the tasks that need to
be done now to consider the work an actual study. Absent that work, the work to date
does not constitute a "study", and therefore should be labeled more accurately as a single
computer model output run.l
Q. Do you agree with this TRC member's comments?
A. Yes.
Q. Do you have a sense of why Idaho Power rushed the critical final phase of the Study?
A. In the above-ref'erenced June 5,2014 email that ldaho Power sent to the TRC members, Idaho
Power stated that "the recent interest sunounding potential PURPA solar development in our
service territory. as well as the IPUC's directives fiom its May 28 Order. requires the urgent
completion of the study." The email provided a link to the Commission's Order in Docket No.
IPC-E- l4-09, in which the Commission "ordered that ldaho Power complete its solar integration
study as soon as possible."
t Exhibit 20+
IPC-E- l4-18
YOURKOWSKI. DI
ocToBER 23,2014
I Q. What is your understanding of the Commission's order as it relates to the timeline for
2 completing the Study?
3 A. I will leave the legal interpretations of the Orderto the lawyers. However, my understanding
4 is that the Commission's directive did not require Idaho Power to sacrifice quality or accuracy in
5 exchange for rapid completion of the Study. In my opinion, it was ldaho Power's own
6 interpretation of the Commission's direction on the timing of the Study that resulted in a
7 breakdown of the TRC process and a flawed Study.
8
9 Q. Setting aside these process issues for now, what is your professional opinion of the
10 technical aspects ofthe Study as it stands today?
11 A. My opinion is that the Study has some positive components and some shortcomings,
12 including one fundamental flaw. I mentioned the positive aspects above, which relate to
l3 developing a data set that provides a fair to characterization of the expected solar generation in
14 ldaho Power's service territory. However, when it came to modeling how this expected solar
15 generation would integrate into Idaho Power's existing system, the Study has several
16 shortcomings, including a fundamental flaw that systematically overstates the integration costs
17 associated with solar generating resources. By including this fundamental f'law, the Study also
18 ignores the potential to reduce the amount of load following reserves; reducing reserve
19 requirements for following load could benefit Idaho Power's retail customers.
20
2l Q. Please summarize what you view as the shortcomings of this Study.
IPC-E-14-I8
YOURKOWSKI, DI
ocToBER 23,20t4 7
I A. The shortcomings relate to how ldaho Power modeled the integration of the expected solar
2 generation into its broader system. As I described in my commentss on the draft Study, the
3 major shortcomings of this Study are:
4 l. No Netting Effect: The Study approach analyzes the incremental reserve requirement
5 and associated production costs for solar assuming that there is no netting effect between
6 the balancing reserves needed for wind, solar, other generation, and load. This
7 assumption is inaccurate as the scheduling errors for wind, solar, other generation, and
8 load offset each other, thereby reducing the total system reserve requirement and
9 reducing the production costs attributable to all three components, including solar. This
10 is an important shortcoming of this Study that has the effect of systematically
l1 overestimating the reserves needed for solar and the associated production costs. I
12 discuss the issue in greater detail below.
13 2. Schedule Lead-Time: Idaho Power uses an extremely conservative assumption that it
14 must use a 45-minute lead-time when calculating solar generation forecasts and
15 submitting solar generation schedules. The amount of lead-time assumed for submitting
16 schedules greatly influences the accuracy of persistence-based forecasts-the less lead-
17 time, the more accurate the persistence-based forecast and the less incremental balancing
18 reserves Idaho Power needs for solar, thereby reducing solar integration costs. NERC
19 and WECC standards allow for schedules to be submitted up until20 minutes before the
20 hour.a A 3O-minute lead-time is reasonable, as it allows for compliance with the 20-
2l minute timeframe for submitting schedules, while also allowing l0 minutes to conduct
rExhibit 20t.
'Available at: htlp://wrvwnerqsS-llc1qq,sl,qg/i,s/lnterchange-[e,ference-Cr(dlqes-V'?-211!?-(]2-12:!irr.rl.pdt'
IPC-E-14-18
YOURKOWSKI, DI
OCTOBER 23.2014 8
I trading activities before schedules are submitted. Other integration studies use a 30-
2 minute lead-time.
3 3. Confidence Interval: The Study's use of a 95% confidence interval is significantly higher
4 than the 90% confidence interval Idaho Power used in its wind integration study. A
5 larger confidence interval greatly increases the amount of balancing reserves required.
6
7 Q. Of the three shortcomings you just stated, which is the most important flaw in the
8 Study?
9 A. The most important flaw is that the Study fails to account for the netting effect between solar
l0 scheduling errors, load scheduling error, wind scheduling error. and the scheduling error of all
I I other generation. As soon as I uncovered this flaw, I brought it to the attention of the TRC and
12 ldaho Power.
13
14 Q. What do you mean, in general, by the term 'onetting effect'o?
l5 A. In maintaining the reliable operation of the electricalgrid, a utility must constantly ensure that
l6 load and generation match. In order to accomplish this, utilities fbrecast their expected load and
17 generation and schedule their generation to meet the expected load. In doing so, the load
l8 forecast (or "schedule") will have error associated with it; the solar. wind. and other generation
l9 forecasts will also all have errors associated with them. Utilities hold operating reserves (or
20 "balancing reserves") in order to manage this unavoidable fbrecast and schedule error and
2l maintain the reliable operations of the system. The term "netting eff'ect" refers to the tact that in
22 order to maintain the reliability of its system, Idaho Power rreed only balance the net variability
23 and scheduling error of the total collective system. not the individual variability and scheduling
IPC-E-14-18
YOURKOWSKI, DI
ocToBER 23, t0l4 9
I error of each separate element (load. wind. solar and other generation). It is standard utility
2 practice to use statistical methods to account for this netting effect while ensuring that a utility
3 still has sufficient balancing reserves on hand, but not more than ratepayers need to pay for.
4 ldaho Power's Study methodology ignores the standard utility practice that is commonly referred
5 to as "netting.'"
6
7 Q. What is the effect of Idaho Power not netting the forecast error of loads and
8 generation?
9 A. The effect of this flaw in the Study is to systematically overestimate the reserve requirement
l0 and integration costs not only for solar resources, but also for other generation and load.
I I Correcting this flaw would lower the costs of solar integration and the cost of following load that
12 are passed on to retail customers.
13
14 Q. Is it common utility practice to account for this netting effect in solar and wind
l5 integration studies such as the one ldaho Power has conducted here?
l6 A. Yes. Absolutely it is. Every other integration study fbr wind and solar resources that I am
17 aware of accounts fbr this netting effect. It is a bedrock principle of integration cost analysis to
l8 account fbr netting. Allof the relevant integration study guidance documents and other utility
l9 integration studies cited by ldaho Power in this docket are fbunded upon and/or utilize the
20 principle of netting. For example, "Evolution of Wind Power lntegration Studies: Past, Present.
2l and Future" describes this principle on page two: NV Energy's "Large-Scale PV Integration
22 Study" describes this principle on page eleven at bullet number two: and. also, Arizona Public
23 Service Company's "Solar Photovoltaic (PV) Integration Cost Study" discusses the same
IPC.E.I4- 18
YOURKOWSKI, DI
ocToBER 23.20t4 l0
I principle in Section 4 on pages two and fbur. I included excerpts from these studies as exhibit
2 205.lndeed, the only utility that I am aware of that does not incorporate netting into its
3 integration studies is Idaho Power.
4
5 Q. Please explain in more detail the concept of "netting scheduling error.'o
6 A. ''Netting" is simply the concept that positive and negative numbers cancel each other out;
7 positive 4 plus negative 3 equals l, not 7. Netting is a mathematical fact that underlies allof the
8 statistical analysis used in solar and wind integration analysis. Electrically speaking, it is a
9 physical reality that utilities comply with their reliability requirements to meet Control
l0 Performance Standards ("CPS") by balancing the netted variability of their collective system.
I I The reason that solar and load scheduling/forecasting eror should be netted is because
12 sometimes the load forecast errs in a positive direction (e.9.+ 20 MW) and at the same time, the
13 solar forecast errs in a negative direction (e.g. - I 5 MW). In this example, the solar and load
14 errors cancel each other out for all but 5 MW, and the system operator only needs to dispatch 5
l5 MW of balancing reserves.' Five megawatts. in this simple example, would be the incremental
l6 balancing reserve need attributable to solar resources. By not accounting for netting, Idaho
' Whil" solar and load fbrecasting errors do not necessarily cancel each other out every second ofthe yeal and can
sometimes even be additive. on average. over the course ofa y'ear. thel do oftset each other in a manner that reduces
the balancing reserve requirement fbr both load and solar. Balancing reserve requirements are designed to measure
and prepare fbl the maximum amount of total system fbrecast error identifled in the data. up to a predetermined
confidence interval (e.g. 90% or 95%). Llsualll,. a solar integration stud1, r.r,ill account fbr the netting efhct by
tiltering a combined data set of solar. load. and other generation schedulin-q error to identitl, the moments *'ith the
greatest total netted slstem error. Once these moments are identified. the study rvill look at the contribution oleach
individual component's (load. solar. etc.) fbrecast etror to the total netted systern error during these defining
moments. Solar's contribution to the netted total s),stem error during these moments is the appropriate measure firr
calculating solar's incremental reserve requirenrent and is the appropriate method tbr calculating integration costs.
IPC-E- r4- 18
YOURKOWSKI, DI
ocToBER 23,20t4
1 Power's Study proposes to charge solar resources for all 20 MW of balancing reserves. In this
2 manner, failing to account for netting can have a significant impact on integration costs.6
3 The netting of diffbrent sources of variability and forecasting eror on the system,
4 including that of loads and generators, has been a cornerstone of efficient utility operations for
5 more than a century. Idaho Power and other utilities already integrate hundreds of thousands of
6 different loads on a daily basis, each with unique properties of variability and uncertainty.
7 Statistical techniques readily allow grid operators to determine the optimal reserve
8 requirement needed to accommodate the combined variability and uncertainty of all loads and
9 generation on the power system; this same principle applies when solar generation is added to
l0 the power system.
t1 The methodology used in ldaho Power's Study is incorrect because it does not factor in
12 this netting effect. The methodology Idaho Power uses is to first separately calculate the amount
l3 of balancing reserves required for load (3% of load) and calculate the production costs associated
14 with this portfolio. Next, Idaho Power separately calculates the amount of balancing reserves
t5 needed for solar (roughly 4.5%o of installed capacity) and then again calculates the production
l6 cost associated with the solar case. Idaho Power compares the difference between these two
17 production cost runs to determine the incremental cost associated with balancing solar
l8 generation. which directly feeds into ldaho Power's proposed solar integration rates.
6 Statisticalll, speaking. only if solar and load variations rvere pert'ectlv negativell'correlated with a correlation
coeftlcientof-l.0rvouldldahoPorverbecorrectthatitisnotnecessary'tonettheirofliettin-e!ariabilit). Well-
established statistical principles dictate that the combined variabilitl ofuncorrelated sources ofvariability is equal to
the square root ofthe sum olthe squarcs ofthe individual sources' variabilitl. As an exanrple. given a t'ictitious
pouer s-v. stem with solar variability of 30 MW per hour. load variability'of 100 MW per hour. and conventional
generator variability,of 40 MW per hour. the method fbr accurately.calculating total power s1'stenr variabilitl is as
fbllorvs:
Sum of squares variabilitl' : ( 30r + 1 69r + 40r) : 900+ I 0.000+ I .600: | 2.500
Total Pouer S),stern Variability : Square Root of (30: + 100: + 40r): lll.80 MW'this method is essential to accuratel) capture the statistical tact that the combined variability ofseveral uncorrelated
tactors is less than the sum of their palts. hence uhy in the example above the cornbinc'd variabilitl,olS0 MW.40
MW. and 100 MW is onlv I I 1.80 MW and not 170 MW.
IPC-E-I4- 18
YOURKOWSKI, DI
ocToBER 23,20t4 t2
1 The problem with this approach is that in reality, utilities do not balance the forecast
2 errors ofload and solar resources separately; utilities balance a netted electrical signal that is the
3 sum of all load and generation forecast error. Reliability standards require Idaho Power to keep
4 its system frequency at 60 Hertz. System frequency is an electrical signal determined by the
5 system's load-resource balance and the netted scheduling errors of all loads and all resources on
6 the system, collectively.T By including a methodological flaw in its Study that is inconsistent
7 with mathematical principles and with how ldaho Power will comply with reliability standards,
8 ldaho Power systematically overstates the balancing reserves needs for all components of the
9 system-including solar, wind, other generators, and load.
10
l1 Q. How should ldaho Power conduct a solar integration study to account for this netting
L2 effect?
13 A. What most other studies do is develop a time-synchronized five-minute dataset of all existing
14 and anticipated sources of variability and scheduling error on their system, including solar, wind,
15 other generation, and load. Idaho Power would then analyze the total system variability and
16 scheduling error collectively as a whole. This single additional step would account for the
17 netting effect. Idaho Power could then conduct its study much the same as it does today: First it
18 would estimate the collective netted reserve requirement of load, wind and other generation-
l9 leaving out solar-and run the production cost model. Then, it would estimate the collective
20 netted reserve requirement of load, wind, other generation and solar, and run the production cost
'Reliability standards (CPS) require utilities to keep their system frequency at 60 herts. Frequency is an electrical
signal that is the collective product ofall loads and resoruces on the system.
http://www.nerc.com/palStand/Reliabilitf/o20Standards/BAL-00 I - l.pdf
http://www.nerc.com/docs/oc/rsiNERCo/o20Balancingo/o20ando/o20Frequencf/o2OControlo/o200405201I l.pdf
IPC-E-r4-r8
YOURKOWSKI, DI
ocToBER 23,20t4 13
1
2
3
4
5
6
7
8
9
10
1l
t2
l3
14
15
16
17
model again. The difference between the two production cost model runs would represent the
incremental netted balancing reserve requirement associated with solar generation.
This is how the Arizona Public Service Company did its Study. which ldaho Power
referenced in Mr. Devol's testimony: "The quantity of reserves for the "load only" case was
determined, and then the reserves required for "'load and solar" were calculated. The difTerence
between the reserves required for the load only case and the load with solar case represents the
incremental reserve requirements necessary needed for solar PV integration."s
Another approach Idaho Power could use to account for the netting effect is called the
Incremental Standard Deviation ("lSD") method. BPA uses the ISD methodology to determine
its integration costs associated with solar, wind, other generation and load fbllowing. ln BPA's
own words, "'[the ISD methodology] takes into account any diversity benefits that may exist
between the regulation signals for load, wind, solar, thermal. and hydro.... ... The result is a
method identifying the relative drivers behind the BPA Balancing Authority Area's need for
balancing reserve capacity and a reasonable methodology for assigning balancing reserve
capacity to the various uses of the system for the purpose of allocating costs and establishing
rates for different types of service."'
18 Q. Are there any other benefits associated with accurately modeling the impacts of the
19 netting effect?
20 A. Yes. Regardless of which of the above-described approaches ldaho Power uses. if the netting
21 effect was modeled not just for solar. but also for load. wind. and other resources, Idaho Power
n Exhibir 205 at pg 6 (highlighted text).
'Puyleart et al.. BP-14-E-BPA-22 at Section 6. pagc 26. lines 5-19. availablc
at https://rvrvw.hpa.gov/secure/RateCasekrpenfile.aspx?tileNanre:BP-l4-F.-BPA-
22.pdf&content'f y pe:appl ication%2fpd1.
IPC-E- l.{- l8
YOURKOWSKI, DI
ocToBER 23.2014
4
5
could determine the netting benefits associated with allof these sources of variability. For
example, if ldaho Power analyzed the impact that netting scheduling errors had on loads, it
would find that compared to its current methodology, it reduces the reserve requirement for load
and thus saves ratepayers money.
Q. In preparing your testimony, did you ask Idaho Power why they did not incorporate
7 this netting effect into their Study methodology?
8 A. Yes. ICL Data Request NO. 6, included at exhibit 206. asked ldaho Power if they recognized
9 the principle of netting and, if not, to please explain ''*hy." Idaho Power responds as fbllows:
10
11
t2
13
t4
l5
l6
17
18
t9
20
2l
22
23
24
25
26
27
28
29
30
Idaho Power does not net solar generation schedule errors with other generation sources,
other schedule errors. or load. Idaho Power designed the solar integration study to
identify the integration issues specifically associated with solar generation. The objective
of the solar integration study is to identifo the effects of the intermittent solar generation
in order to calculate the integration cost imposed by solar upon ldaho Power's existing
system. Idaho Power's system design includes the capability of system dispatchable
generators to manage variability in customer load and other generation. Intermittent solar
generation introduces new variability and uncertainty into system operations. Because of
the inherent differences and levels of confidence in load forecasts versus forecasts fbr
intermittent generation, such as wind and solar, load fbrecast errors are often auto
correlated. reflecting a tendency fbr fbrecast errors to persist in magnitude and direction
throughout the day, and are more readily addressed as the system is managed through to
real time. However, in order to maintain the reliable operation and stability of the
system. as well as to meet its various regulatory reliability criteria, the Company must
provide adequate reserves based upon the higher magnitude and nature of the forecast
error present in intermittent and variable wind and solar forecasts. Thus, the challenges in
fbrecasting wind and solar as compared to load for unit commitment are considerably
different. requiring the system to treat differently the possibility of errors in fbrecasting
these elements of load and resource balance.
Q. What is your reaction to Idaho Power's response?
3l A. A carefulreading of ldaho Power's response illustrates that it does not actually explain or
32 justily why ldaho Power does not net solar scheduling error with other generation and load
IPC-E- l4- l8
YOURKOWSKI, DI
ocToBER 23.20t4
1 scheduling error. I will expand on this point by examining each part of Idaho Power's response
2 (shown in italics):
3 l) "ldaho Power designed the solar integyation study to identfu the integration issue,s
4 specificalll,associatedwith solar generation. The ob.jective of the solar integration study is to
5 identifu the fficts o.f'the intermittent solar generation in order to calculate the integration cost
6 imposed by solar upon ldaho Power's existing system.;'
7 I agree with this statement and find it to be a statement of the obvious. All solar
8 integration studies "identifo the integration issues specifically associated with solar generation"
9 and, in doing so, all solar integration studies (other than Idaho Power's) account for the netting
10 effect before determining the amount of balancing reserves attributable to solar generating
1l resources. This statement from ldaho Power does not explain why they do not net solar
12 scheduling error with other generation and load scheduling error.
1 3 2) " ldaho Power's system design includes the capability o/' system dispatchable
14 generators to manage variability in custonrcr load and olher generation. Intermittent solar
l5 generation introduces new variabili4t and uncertainty into system operation,s."
l6 I don't know any experl that would disagree with this statement. What is also true is that
17 the new variability introduced by solar generation will net with the variability and associated
l8 scheduling error of other resources and loads, thereby reducing the reserve requirements fbr
19 loads and resources and offsetting some of the impact of the incremental variability attributable
20 to solar resources. This poftion of ldaho Power's data response also does not explain why they
21 do not net solar scheduling error with other generation and load scheduling error.
22 3) "Because of the inherent difibrences and levels of conJidence in loadforecast,s versus
23 forecasts./br intermittent generation, such aswind and solar,load./'orecost errors are o/ien auto
IPC-E- 14- 18
YOURKOWSKI, DI
ocToBER 23.2014 r6
I correlated, reflecting a tendency.for forecast eruors to persist in magnitude and direction
2 throughctut the day, and are ntore readily addres.sed as the system is managed throngh to real
3 time."
4 The premise of this sentence is f'alse. In statistics, autocorrelation refers to a time-series
5 data set that exhibits statistical properties whereby an earlier data point in the time series is a
6 statistically significant indicator of future data points in the same time series. Load forecast
7 errors may be autocorrelated because of persistent biases in the load forecasts or because of the
8 inherent pattems of weather and load. The statement that load forecast errors can be
9 autocorrelated and are "more readily addressed as the system is managed through real time" has
l0 no bearing on the appropriateness of netting the scheduling error of solar with other generation
1l and load. Regardless of the statistical properties of the fbrecast error (autocorrelated or not) and
12 regardless of when the scheduling error is more "readily managed," when you get to real-time, it
l3 is an electrical and mathematical fact that the scheduling errors will net.
14 4) "However, in order to maintain the reliable operation and stability of the system, as
15 well as to meet its various regulatory reliability criteria, the Company must provide adeqltate
l6 resen)es based upon the higher magnitude and nature of the.forecast error present in
17 intermitlent and variable wind and solar forecasts."
l8 The first part of this sentence is not controversial; Idaho Power must maintain adequate
19 reserves to operate a reliable and stable system. The explanation fails in the next part of the
20 sentence. Regardless of the magnitude and nature of the forecast error. the scheduling errors of
21 solar, other generation, and load will net. I don't disagree that Idaho Power needs to carry
22 adequate reserves to account fbr the observed/modeled magnitude of forecast error associated
IPC-E- 14- 18
YOURKOWSKI, DI
ocToBER 23,20t4
I with the variability of solar resources, the point is that without accounting for the netting effect,
2 Idaho Power will be systematically carrying excessive reserves for solar and load.
3 5 ) "Thus, the challenges in forecasting wind and solar as compared to load for unit
4 commitment are considerably di./ferent, requiring the system to treat dilferently the possibility of
5 errors in,fitrecasting these elements of load and resotu'ce balance. "
6 I have not suggested-nor would I suggest-that forecasting solar generation is exactly
7 the same exercise as fbrecasting load. I also have not suggested-nor would I suggest-that the
8 statisticalcharacteristics of the variability of load and solarare identical. No solar integration
9 study does. The fact is that once you have taken into account the inherent differences between
10 load and solar generation and have done your best to forecast both load and solar generation,
11 there will be netting of the inevitable remaining real-time scheduling errors. Nothing in Idaho
12 Power's data response. testimony. or Study demonstrates otherwise.
13
14 Q. Given the current status of Idaho Power's Solar Integration Study, what is your
15 opinion about the proposed rates?
l6 A. First and foremost, it is extremely important for ldaho Power to correct the flaws and improve
17 the analysis in its Solar Integration Study before any of the estimated rates are approved. As
l8 solar power continues to come on line in ldaho, accurately calculating the integration costs is the
19 only way to ensure solar projects pay their fair share: no more; no less. An accurate
20 methodology will also reveal the positive benefits of the eff-ect of netting solar (and wind)
21 reserves with the reserve requirement for loads. Accounting for this decrease in the reserve
22 requirement for load would reduce costs to ldaho Power's customers.
IPC-E- l4- r 8
YOURKOWSKI, DI
ocToBER 23,2014 l8
I Ultimately, I recommend that the Commission not adopt Idaho Power's proposed solar
2 integration rates until ldaho Power corrects the flaws in its solar integration analysis. At a
3 minimum, a corrected solar integration study should account for the netting effect and should
4 also include sensitivities around the scheduling leadtime and the confidence interval issues
5 described above.
6 In the meantime, until Idaho Power corrects the flaws in its current Solar Integration
7 Study, my recommendation is to allow PURPA solar projects and ldaho Power to negotiate
8 integration charges, consistent with what is my understanding of current practices.
9
10 Q. Does this conclude your direct testimony?
1l A. Yes. it does.
IPC-E-14-18
YOURKOWSKI, DI
ocToBER 23,2014 t9
Benjamin otto (lSB No. 8292)
710 N 6th Street
Boise, ID 83701
Ph: [phone] x 12
Fax: [phone]
botto@ idahoconservation.org
Attomey for the Idaho Conservation League
IN THE MATTER OF IDAHO )
POWER COMPANY'S )
APPLICATION TO IMPLEMENT )
SOLAR INTEGRATION RATES AND )
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
CASE NO. IPC-E-I4-18
IDAIIO CONSERVATION LEAGUECHARGES.
DIRECT TESTIMOIYY
CAMERON YOURKOWSKI
EXHIBIT 2OI
RENEWABLE NORTTTWEST COMMENTS TO IDAHO POWER ON THE DRAFT
SOLAR INTEGRATION STUDY
From Idaho Power Response to Sierra Club Production Request No. 12
.otlllar, Renewabte.@ Northwest
DATE:
TO:
FROM:
RE:
June 6,2014
Phillip DeVol, ldaho Power
S ubm itted V ia Emai I to t, De V o l(Zi idallSpoggru-Sll
Cameron Yourkowski, Senior Policy Manager
Idaho Power Solar lntegration Study Report (June 2014)
Renewable Northwest appreciates the opportunity to participate in the Technical Review Committee (TRC) for
Idaho Power's Solar Integration Study (June, 2014) and we look forward to working with tdaho Power (lDP) on
solar integration issues in the future.
The compressed timeline of this study process, especially at the end of the process and during the analysis phase,
has diminished my ability to comprehensively review the details of this study. Clearly, more thinking and
analysis will need to be done as Idaho Power and the region gain more experience with solar resources. That
said, given the compressed timeline that IDP determined to be appropriate for completing this study, I appreciate
the willingness of ldaho Power staff to try to make as much information available as possible. I hope that we can
continue the dialogue going forward.
The compressed timeline and the inability of the TRC to obtain and work with any of the actual analysis behind
this study makes it difficult to verify the accuracy of this study or to estimate the impact of the identified
shortcomings. Instead, I will comment on the aspects of the study that are clear at this time and compare the
results to other solar integration studies I have worked on.
I outline below what I view as the positive aspects of this study and also the areas where I believe funher work is
warranted. Based on my experience with other wind and solar integration studies, I observe that the estimated
integration cost of
S0.40/MWh (associated with a 100 MW solar buildout) is within the range of what has been calculated by other
studies. However, the issues identified below suggest that even this lower cost may be too high and that an
adjustment is warranted.
Given the opaqueness of this study and the issues identified below, my recommendation is that ldaho Power not
use the costs associated with the larger buildouts in any official manner at this time. Those scenarios are not
imminent and the estimated costs should not be relied on until further analysis is completed.
Exhibit 20 | . Page I of 7
IPC-E- t4- l8
C. Yourkowski. tCL
Source: Idaho Power Response to Sierra Club Request No l2
2.
J.
B$ i!iw- 4 cpe c tr-,o-f the!-ts-dr;
l. Solar Data: Based on my understanding of the process and effbrt prior to my joining the TRC, I think that the
study has done a good job collecting a quality set of solar data for IDP's service territory.
l.
Solar Diversity: The study has appropriately analyzed both concentrated and dispersed solar buildout scenarios.
The "diversity value" associated with dispersed buildouts significantly influences the need forbalancing reserves
and, based on the information available to me at this time, the study has captured this effect sufficiently.
Wavelet Transformation: The study's use of the Sandia National Lab's "wavelet-based variability model" to
transform the raw single-point irradiance data into what is always less volatile plant-level generation data is
appropriate. The wavelet model is the best approach to handling this data transformation issue that I am aware
of.
Hour-Ahead Forecast: The study incorporated a lot ofpositive work on developing an hour- ahead forecast for
solargeneration. lnconcept,theuseofapersistenceforecastadjustedfortheknowncurvatureoftheirradiance
curve, based on a "clear sky index" for that day of the year, is a sound approach. Unfortunately, due to the
compressed timeline and the lack of data made available to the TRC, I did not have time to scrutinize the details
of this approach as it was implemented for this study. See more below.
Sh o q!! o{_n LUCS_9f t he$qdy;
No Netting Effect: The study approach analyzes the incremental reserve requirement and associated production
costs for solar assuming that there is no netting effect befween the balancing reserves needed for load, wind and
solar. This assumption is inaccurate as the scheduling errors for load, wind and solar will often offset each other,
thereby reducing the total system reserve requirement and reducing the production costs attributable to all three
components, including solar. This is an important shortcoming of this study that has the effect of systematically
overestimating the reserves needed for solar and the associated production costs.
Hour-Ahead Forecast: While the study's conceptual approach to developing an hour-ahead forecast is sound, the
rushednatureofthisstudyhasnotallowedtbranyfine-tuningoranalysisofitsapplicationhere. Specifically,
the mean absolute error of this forecast has not been made available to date and would be invaluable for
assessing the accuracy, comparing it to other approaches, and improving its application within this study. This is
an important area for further analysis, as the forecast error is an extremely important driver for determining
balanci ng reserve requirements.
Schedule Lead-Tirne: Related to the forecast methodology is the use of a 45-minute lead- time assumption for
submitting solar schedules. As page l3 of the study identifies, an area fbr future study is looking at shofter lead-
times, such as 30 minutes. Schedules can be submitted up until 20 minutes before the hour. Leaving l0 minutes
to conduct trading activities, a 3O-minute lead-time is reasonable and is used in similar integration studies. The
amount of lead-time for submitting schedules greatly influences the accuracy of a persistence-based forecast and
thus the amount of incremental balancing reserves required. Similarly, as the region transitions to greater use of
I 5-minute schedu ling, adj usting solar schedules every I 5 m inutes will greatly reduce the incremental reserve
requirement and associated production cost.
Exhibit 201. Page 2 of7
IPC-E- l4- I 8
C--. Yourkorvski. ICL
Source: Idaho Porvel Response to Sierra Ctub Request No l 2
4.
2.
J.
4. Confidence Interval: The study's use of a 95% confidence interval is significantly higher than the 90%
confidence interval IDP used in its wind integration study.
Recommendation:
The region's experience with integrating solar energy is rapidly evolving and new tools and methodologies
are currently being implemented. This evolution, coupled with the rushed nature of this study and the
ambitiousness of a 300 to 700 MW solar buildout in IDP's service territory, suggests that additional analysis
should be completed before adopting the estimated costs associated with those larger buildouts of solar
($1.20lMWh-S2.50/MWh). The $0.40/MWh estimated cost associated with a 100 MW solar buildout is
more imminent and is within the range of similar solar integration studies. However, the shortcomings of
this study may also warrant reductions to the $0.40/MWh figure until more analysis can be completed.
Exhibit 201. Page 3 of7
IPC-E-t4-t8
C. Yourkowski. ICL
Source: Idaho Power Response to Sierra Club Request No t2
1)
2l
June 13,2014
Phil DeVol IRP Manager ldaho Power
PDeVol@ idahopower.com
Dear IDP Solar lntegration Study Team,
Thank you for the opportunity to participate on ldaho Power's Solar lntegration Study (Study) Technical
Review Committee (TRC). I found the exercise interesting, many of ldaho Power's methods innovative, and
the results informative. Although the Study's methods warrant additional review, ldaho Power's results
appear reasonable and comport with solar integration costs estimated throughout the region.
The Study development and TRC participation process was well meaning and appreciated, but ultimately
limited by a near term regulatory requirement to file the Study results at the Commission. Nonetheless,
the early phases of the Study process allowed for meaningful dialogue between the Study team and the
TRC. This dialogue allowed for a refinement of data sources and led to at least two important
enhancements to the Study methodology. Late phases of the TRC process were compromised by the
Study's accelerated schedule. The accelerated schedule did not leave time to consider improvements to
the reserve requirement methodology, to review the production cost simulation assumptions and models,
or to review the Study in light of preliminary results. Despite these limitations, ldaho Power has made a
strong effort to accurately measure solar integration costs. Subsequent efforts to estimate these costs will
provide the opportunity for additional review.
Four Areas Where this Study Excels:
The solar integration study stood out in four ways where ldaho Power made a strong effort to refine
assumptions and implement innovative ideas.
The Study used the difference between hour-ahead forecasts and actual generation to determine schedule
errors. While this Study assumption is not innovative, it ls a change from ldaho Power's earlier
perspectives regarding variable energy resource integration. Re- evaluating old assumptions can be
challenging, and ldaho Power's willingness to use hour-ahead schedule errors should be acknowledged as
evidence where the Company challenged itself to develop a methodology suitable for solar integration.
The Study made creative use of algorithms to develop hour-ahead forecasts whose accuracy was superior
to simple persistence forecasts. The function used to develop the hour-ahead forecasts recognizes the
time of day and season to predict how hourly solar generation will differ from the preceding hour. This
solution is impressive given the relatively few solar integration studies that exist nationally, many of which
lack this forecast intelligence.
Exhibit 20 l. Page .l ol7
tPC-E- t4- t8
C. Yourkou'ski. ICL
Source: Idaho Porver Response to Sien'a Club Request No 12
3) The Study applies an advanced "Wavelett-Based Variability Model" to gross up point source
irradiance data into simulated output from a large utility scaled solar array. As a TRC member as
lwasnotawareofthisanalyticapproachandfoundmyself learningthroughitsapplication. l'd
like to commend ldaho Power for using an innovative solution not widely known to the utility
industry.
4) The Study gathered data from diverse solar sites and went to great lengths to gather the
associated generation data. lt would have been far easier to use solar data from one particular
site, or to use a collection of synthesized data, but by using historical data from local AgriMet
towers ldaho Power's study was able to include real ground based measurements. By including
diverse resource sites, the Study well captures the beneficial effects of geographic diversity.
Areas Where Further Scrutiny is Merited:
The Study does include some assumptions and methodologies that warrant further review. Relatively
little time was made available for the TRC to discuss assumed reserve requirements or production cost
simulations. Asaresult,someofthesesuggestionsmayresultfrommyownmisunderstanding.
1) The Study assumes that a separate quantity of balancing reserves be held for load, wind, and solar
resources. These separate capacity requirements are inputted into the production cost simulation
model. However the total capacity of reserve requirements needed to integrate load, wind, and
solar should be less than the sum oftheir constituent needs because the schedule errors for load,
wind, and solar are generally non-correlated. The Study calculates the amount of reserves
required to meet 95% of solar's schedule error, but it would be more accurate to calculate the
reserves required to meet 95% of solar, wind and load combined schedule error. Then the amount
of reserves attributable to solar can be determined by subtracting the reserve requirements
for just load and wind. Making this adjustment would lower solar integration costs f
or all portfolios.
2) The incremental and decremental reserve requirements (RRs) calculation methodology would
benefit from additional review. The methodology is innovative and its ability to adapt to changing
solar conditions on an hourly basis is advanced and highly desirable from an operational
perspective. However, it's likely that the method to calculate RRs could be further refined to lower
the intensity of RR limit excursions. Presently, the RRs are based on a percentage above and
below the hourly forecast. This approach is counter-intuitive (but not necessarily wrong) because
thosehourswhichhavelowerforecastscanhavesomeofthehighestvariability. Forexample,the
most RRs should be required at mid-day with partial clouds, but because the hour ahead forecast
will be low (due to clouds) the calculated RRs will be lower than it would with a clear sky. lt's
possible that this negative result is diminished by the diverse solar sites used to generate the
forecast.
3) The Study used wind data that was not contemporaneous with the solar data used in the
production cost simulation. While it is understandable why ldaho Power used generation data
Exhibit20l. Page 5 ol7
tPC-E- 14- t 8
C. Yourkouski. ICL
Source: Idaho Porvel Response to Sierra Club Request No l2
from later years, the Study should try to determine whether wind and solar generation is
correlated, or whether the intra-hour variability of wind and solar generation is correlated. Such
follow up analysis would be helpful to understand what effect this data decision has upon results.
4l The Study used Solar Anywhere data at the Grand View site where AgirMet data was not available.
ldaho Power showed resourcefulness for gathering data from a wide array of sites. However, I am
concerned that the SolarAnywhere data would be difficult to time synchronize with the ground
based AgriMet sites. Furthermore, because SolarAnywhere is based on satellite observations of
clouds, and their approximated velocity, it seems that the satellite data would understate the
solar variability relative to ground based measurements. The Study should address these concerns
by comparing SolarAnywhere data to ground based AgriMet data of the some site. This
type of follow-up analysis would determine how comparable the datasets are and help answer
whether they are suitable for combination. lt's possible that the aggregate variability of the solar
resources is understated due to the under-represented correlation between SolarAnywhere's
Grand View data and the rest of the portfolio.
5) The TRC did not review the production cost simulations assumptions orthe internally developed
simulation model.
Recommended Additions for Subsequent Studies:
lntegration studies typically benefit from subsequent iterations v'ihen power costs are updated and
additional scenarios can be considered. The following suggestions may be appropriate for study when
that time comes.
1) Future studies should model solar portfolios with smaller projects. lt is possible that few, if any
QFs, are able to finance their projects at recently signed avoided costs. lt is also possible that
projects aggregate in one particular site and are less geographically diverse than assumed in this
study. Future studies could model a portfolio with less than 100MW capacity and a portfolio
with 300MW-500MW clustered near one particular site.
2l Future studies should include the sub-hourly scheduling periods. While sub-hourly markets are
not currently widely used, they will become more liquid as BPA and the California lntertie begin
offering 15-minute and 30-minute schedules. ldaho Power will likely want to know what type of
savings may be achieved by transacting in sub-hourly markets. Studying sub-hourly scheduling
in subsequent integration studies will be informative.
lnsummary,theresultsofthestudyarecomparableothersolarstudiesperformedintheregion. While
improvements remain, ldaho Power should be commended for demonstrating considerable ingenuity
Erhibit 201 . Page 6 ol7
IPC-E-I'l-18
C. Yourkoriski. IC[.
Source: ldaho Porvcl Response to Sierra Club Request No l2
and a commitment to accurately capturing solar integration costs. I appreciated the opportunity to
participate on the TRC and am always available for follow-up questions.
Sincerely,
Jimmy Lindsay
Exhibit 20l,Page 7 of ,7rPC-E-I4-t8
C. Yourkowski. ICL
Source: Idaho Power Response to Sierra Club Request No l2
Benjamin ono (lSB No. 8292)
710 N 6th Street
Boise, ID 83701
Ph: [phone] x 12
Fax: [phone]
botto@idahoconservation. org
Attomey for the ldaho Conservation League
IN THE MATTER OF IDAHO )
POWER COMPANY'S )
APPLICATION TO IMPLEMENT )
SOLAR INTEGRATION RATES AND )
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
CASE NO. IPC-E-14-18
IDAHO CONSERVATION LEAGUECHARGES.
DIRECT TESTIMONY
CAMERON YOURKOWSKI
EXHIBIT 202
NATIONAL RENEWABLE ENERGY LABORATORY AND THE UTILITY
VARIABLE-GENERATION INTEGRATION GROUP
PRINCIPLES FOR TECHNICAL REVIEW (TRC) INVOLVEMENT IN STUDIBS OF
VARIABLE GENERATION INTEGRATION INTO ELECTRIC POWBR SYSTEMS
From ldaho Power Response to ICL Production Request No. 2
'6-,l
qil'Il3*F","L**,u.n.,n,,-u*,o,o,,
Principles for Technical Review Committee (TRC) Involvement in Studies of
Variable Generation Integration
into Electric Power Systems
Exhibit 202. Page I of 2
IPC-E-t4-18
C. Yourkowski. ICL
Source: Idaho Power Response to ICL Request No 2
UVIG
What Will a TRC Provide?
A properly constituted TRC will assist the project sponsors in ensuring that the quality of the technical
work and the accuracy of results will be as high as possible. TRC participation will also enhance the
credibility and acceptance of the study results throughout the affected stakeholder communities. And
TRC members will be qualified to carry the key messages of the study to their respective sectors.
What Is a Properly Constituted TRC?
TRC membership should include individuals that collectively provide expertise in all of the technical
disciplines relevant to the study. A TRC facilitator should be selected from among the TRC membership.
Sponsorship and facilitation of the TRC should be independent from, but closely coordinated with, the
project sponsors and the team conducting the work. Observers from relevant government agencies and
other interested parties may attend TRC meetings and be included in TRC communication at the
discretion of the project sponsors. Alternatively, a separate stakeholder group can be considered in
order to update interested parties on study progress and key results.
What are the TRC's Functions and Requirements?
The TRC will
Review study objectives and approach, and offer suggestions when appropriate to strengthen
the study.
Help ensure that the study:
' Builds upon prior peer-reviewed variable generation integration studies and related
technical work;. Receives the benefit offindings from recent and current variable generation
integration study work;
' lncorporates broadly supported best practices for variable generation integration
studies;. ls developed with broad stakeholder input.
Engage actively in the project throughout its duration. ln general, project review meetings
should be held nominally on a quarterly basis; some meetings can be held telephonically, but
some should also occur face-to-face. A face-to-face kickoff meeting to establish and agree on
the general direction of the work is required.
Engender collegial discussions of methods and results among TRC members, the study team,
project sponsors and other interested parties. The aim of these discussions is to improve
accuracy, clarity and understanding of the work, and reach consensus resolution on issues
that arise.
Avoid public disclosure of meeting discussions and preliminary results. ln general, findings
should not be released until accepted and generally agreed upon by project sponsors, the
study team and the TRC. When advisable, possible and agreed to by all project participants,
interim progress reports can be provided to a broader stakeholder group.
Ensure that findings are based entirely on facts and accurate engineering and science.
Project sponsors need to embrace this aim so that the results and findings are objectively
developed and not skewed to support any desired outcome.
Document results of TRC meetings and distribute meeting presentations and minutes.
To carry out these functions, the TRC requires
Access to all relevant information needed to properly evaluate the work and the results.
When required, TRC members will enter into confidentiality agreements to protect this
information. ln no case can certain information needed by the TRC be declared "off-
limits."
Assurance that the study results will be made public through published documentation or
other suitable means, with the understanding that business- sensitive information will not
be made public.
Assurance that project sponsors will describe the project as having the benefit of expert
review by a TRC only if the TRC has clearly expressed its acceptance of and agreement with
the results of the study.
Assurance that, in the event agreement is not reached by the TRC and other project
participants, any reference to the TRC will be removed from the final report and any
associated documents or publicity.
How Can Project Sponsor(s) and a TRC Agree To Conduct A Study in Accordance With These Principles?
Each can sign below:
for the Project Sponsor(s)
for the Technical Review Committee
Uxhibit 202. Page I of2
IPC-F-- l.l- I 8
('. Yourkowski. lCt.
Source: ldaho Power Response to ICL Request No 2
Benjamin Otto (lSB No. 8292)
710 N 6th Street
Boise, ID 83701
Ph: [phone] x 12
Fax: [phone]
botto@ idahoconservation.org
Attorney for the ldaho Conservation League
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
IN THE MATTER OF'IDAIIO CASE NO. IPC.E-I4-I8
POWER COMPANY'S
APPLICATION TO
IMPLEMENT SOLAR IDAHO CONSERVATION
INTEGRATION RATES AND LEAGUE
CHARGES.
DIRECT TESTIMONY
CAMERON YOURKOWSKI
EXHIBIT 203
JUNE 5,2014 EMAIL FROM PHIL DEVOL,IDAHO POWER TO TRC MEMBERS
From Idaho Power Response to Sierra Club Production Request No. ll
Pankau, Kathy
From:
Sent:
To:
Cc:
Subject:
DeVol, Philip
Thursday, June 05, 20142:45PM
'Cameron Yourkowski'; Paul Woods; Brian K. Johnson; Myers, Kurt S; Jimmy Lindsay;
Rick Sterling; ANDRUS Brittany; CRIDER John
Stokes, Mark; Youngblood, Mike
IPC draft solar study report
Good afternoon,
I would like to take this opportunity to express to the entire TRC how much ldaho Power appreciates your
involvement and participation throughout this process. Specifically, I want to apologize if, in recent meetings or
correspondence, I have given the impression that the TRC comments are merely a formality. That is definitely not
thecase. Yourfeedback duringthecourseofthestudyhasbeenimportanttous,andcontinuestobeimportantover
the final stages of this first phaseof the study.
As you know, the recent interest surrounding potential PURPA solar development in our service territory, as well as
the IPUC'sdirectivesfromitsMay28Order,requirestheurgentcompletionofthestudy. Thestudyneedsto
continue to move forward towards a target completion date of mid-June. However, the Company does value the
participation and input from the TRC and other participants, and very much would like your review, comments, and
observations incorporatedintotheprocesspriortocompletionofthefinal reportandthisPhaseOneprocess. I
apologize that my prior comments may have given the impression to some of you that, "comments from the TRC
were not going to impact the final report" and that it would not be a "worthwhile investment of time to discuss the
report as a group." This is not the case, nor is that ldaho Power's view of the process and the role of the TRC.
As we've discussed before, the development of the appropriate integration costs for solar is a complicated process.
Not many utilities have done this previously, and we have all learned new concepts along the way. We understand
that some of you may feel the need to comment on what you may perceive as study shortcomings, at least for this
Phase One process. lt is, however, our hope and expectation that you will also comment on your perception
of thestrengthsof thestudy,sothatweall cangainthroughoutthisprocessandcontinuetoimprove. Myhopein
sending the draft report out on.June 2,2014 was to allow everyone to review and communicate comments about it
this week, and that we could meet to review, discuss, and comment sometime next week - with a goal of working
towards a final report by mid-June. I am happy to take any comments, etc... by email or phone, but would certainly
scheduleaTRCmeetingto discussfinalizationofthereportandphaseoneprocessifthatwasdesiredbytheTRC.
Please let me know right away if you would like me to schedule a TRC meeting for review, discussion, and comment
prior tofinalizingthereport-orifyouprefertosubmitcommentsandwrap-upthisphasewithoutanadditional
meeting. Either way, at this point the Company intends to move forward with a target completion of the Phase One
process on June 16,2OI4. The study will then be incorporated into a filing with the IPUC, which will not likely occur
on thesamedaythatthereportisfinal,butinsteadamatterof daysafterwards. Pleasefeel freetocontactmeif
you'd like to talk, and I look forward to your feedback and comments.
Many
thanks, Phil
Exhibit 203. Page I ol I
rPC-E-t4-18
C. Yourkou,ski. ICL
Source: Idaho Power Response to Sierra Club Request No 1 I
Benjamin Ono (lSB No. 8292)
710 N 6th Street
Boise,lD 83701
Ph: [phone] x t2
Fax: [phone]
botto@ idahoconservation.org
Attomey for the ldaho Conservation League
IN THE MATTER OF IDAHO
POWER COMPANY'S
APPLICATION TO IMPLEMENT
SOLAR INTEGRATION RATES AND
CHARGES.
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
CASE NO. IPC.E.14.I8
IDAHO CONSERVATION LEAGUE
DIRECT TESTIMOI\ry
CAMERON YOURI(OWSKI
EXHIBIT 204
TRC MEMBER PAUL WOODS COMMENTS TO IDAHO POWER ON THE
DRAFT STUDY
From Idaho Power Response to Sierra Club Production Request No. 12
Comments of Paul Woods, TRC Member
Idaho Power Solar Integration Study Report
fune 2Ol4
The following comments are provided on behalf of myself as a member of the Technical
Review Committee (TRC) to ldaho Power staff on the draft Solar lntegration Study Report. The
comments are intended to clarify in the report the limited role of the TRC and to accurately
reflect the study process.
General Comment
The concept of a TRC to provide input in completing a Solar lntegration Study had the
potential to provide real benefits to rate-payers, solar generation entlties and ldaho Power in
completion of the study. I appreciate the opportunity to serve on the TRC and commend
ldaho Power for inviting someone of the general public to serve on the Committee with
limited experience as compared to the knowledge and skills of my fellow members of the TRC.
The study started out in a very collegial manner and members of the TRC provided valuable
input on how to acquire and incorporate solar characteristic data and generation estimates
using wavelet techniques. I believe the initial phase of the Solar lntegration Study represented
a best practice in conducting such a study.
The study took a dramatic after the public meeting at which ldaho Power told all those present
that the process had just begun. lmmediately following the meeting ldaho Power filed with
the PUC regarding solar generation contracts and the Solar lntegration Study took a dramatic
turn and the exposure to major assumptions and methods of study ceased to be elements of
review by the TRC.
lnthereport,itisimportantthatthetextprovideanaccuraterepresentationofthe elements
of thestudythatwereavailableandreviewedbytheTRCandthosethat werenot. lstatedin
the last meeting of the TRC that this is not a study but rather it is a document presenting the
results of a single computer model run of an internal ldaho Power computer model that was
not reviewed by the TRC for accuracy, sensitivity and variability.
In conclusion, the items that are listed as "Phase ll" study tasks are the tasks that need
to be done now to consider the work an actual study. Absent that work, the work to date
does not constitute a "study", and therefore should be labeled more accurately as a single
computer model output run.
Exhibit 204. Page I ol3
l lPC-rr-14-18
C. Youlkorrski. [('1.
Source: Idaho Porrcr Response to Sierra Cluh ReqLrest No t2
My specific comments on each section of the report are as follows:
Acknowledgments
The second sentence ofthe opening paragraph states that the TRC provided "substantial"
guidance. This is not accurate and the word substantial should be deleted or replaced with
the word limited. ln addition I do not believe that the study methods were consistent with the
UVIG and NREL TRC study guidelines and the reference to this standard should be deleted.
ldaho Power did not actively engage the TRC on key elements of the study such as load
reserves, production cost simulation sensitivity analysis and the interpretation of the output of
one single model run. The limited opportunities for input and technical review of these
elementsofthestudymakethetextofthereportoverstatetheroleoftheTRCand imply
technical review in areas where none occurred.
Pege 4- 9-o-! e !: P,! a n-! -c,h-aneleruu cJ
The following sentence is not supported by data or analysis that was available to the TRC.
"While panel orientation and tracking capability are key factors in the determination of
avoided costs, these attributes are of lesser importance with respect to the variability and
uncertainty driving integration costs." This sentence should be deleted or the report should
produce data supporting this claim. The TRC did not review model runs of Idaho Power's
production cost simulation model, nor were there multiple model run results that would
support this statement. More information is required to support this statement in the report.
Page 5 Statistical-Based Analysis of Solar Characteristics
The top half of page 5 is a discussion of Idaho Power's participation in the Mid- Columbia
electric power market that presents assumptions and arguments for the hour-ahead framework
of the final study. The decision to use hour-ahead trading under the parameters discussed in
this section is an important element of the study outcome. The report should clearly note that
this analytical framework was not part of the TRC review.
ln the last TRC meeting prior to issuance of this report, ldaho Power staff presented a power
point showing the final report conclusions of integration charges based on a single Production
Cost Model Simulation. This information was presented with virtually no supporting
documentation or sensitivity analysis of the model assumptions and input variables. When
asked whether there would be additional Productions Cost model runs to understand the
sensitivity of the model to the various model parameters and assumptions, Idaho Power's
response to this request was no. How sensitive the model is to this assumed framework is a
key question that remains unanswered by the work to date and it is not clear what exposure
rate- payers or solar generation entities face from this absence of analysis.
Urhibit 204. I'age 2 ot'3
tP('-Ft- t4- 18
C. Yourkowski. I('l-
Sourcc: Idaho I)o*er Response to Sierra Club Request No 12
Pages 5 & 6 Hour-Ahead Solar Production Forecast
The framework and assumptions in this section were not an element of technical review by
the TRC and the study report should accurately state this fact. The sensitivity of the final
Production Cost Simulation as it relates to the required capacity to be held in reserves were
not determined through multiple model runs and sensitivity analysis. This work should be
done now in order to call the work a study.
Page 9 Simulation Model
The Production Cost Simulation Model was not part of the technical review of the TRC. More
importantly the whole study is based on a single model run without any sensitivity analysis.
The exposure of the TRC to the Production Cost Simulation Model was a single power-point
slideinameetingimmediatelypriortoissuanceofthereport. lnthe slide,themodel output
showed the need to dispatch natural gas peaker resources to meet load and reserve capacity on
a day in April up to 200 MW. A day later the model showed the spin up of a coal resource to
meet load reserves. Asked to explain the conditions that would dictate this type of dispatch
pattern,theresponseof staff wasthattheyjustgottheresultsinthelastdayandwerenotyet
abletodescribe anddefendtheresults. Oneexplanationofferedbystaffwasthatfloodcontrol
releases from the hydro system were perhaps driving the outcome which raised the question of
why that condition becomes limiting on reserves for other resources.
The point is that the Production Cost Simulation model and the model output did not receive
outcome technical review by the TRC and the few questions of the output by some TRC
members were not been addressed in any meaningful manner.
Pages 9 and 10 Reserves
One of the key elements driving the analysis is the amount of reserves needed to integrate
solarpower. AsamemberoftheTRC, lcameintothestudyprocess supportingtheconceptof
integration charges and looking forward to a robust discussion of the pollcy choices and
reliability needs that would drive the integration analysis. Unfortunately there were no
opportunities for the TRC to understand and provide input to ldaho Power on the types and
characteristics of reserve capacity or how the other generation sources within the ldaho
Power system work to minimize forecast errors and reliability and therefore reserve needs.
There should be language in the report clarifying the TRCs limited role in reviewing the reserve
needs.
Page 12 Further Study
Llxhibit 20.1. Page 3 ol3
IPC-E- l;l- l8
C. Yourkorvski. ICl.
Source: Idaho Porvcr Response to Sierra Club Requesl No 12
All of the elements listed as suggestion for further study are elements that should have been
performed in the initial study. Absent those elements, the title of the report should be Solar
lntegration Single Production Cost Model Results.
In conclusion, it is not often in today's age of analysis paralysis that I find myself recommending
further analysis, but in this case I cannot see how a single Productions Cost Simulation Model run
can be considered a "study" an how this provides the necessary information to insure that rate-
payer and solar generation interests are fairly represented in the "study" outcome.
Lastly, please simply list my name in the report as Paul Woods, member of the public. Please do not
refer to my Woods Consulting Group company status.
Exhibit 204. Page 4 of4
rPC-E-14-18
C. Yourkowski. ICL
Source: Idaho Power Response to Sierra Club Request No l2
Benjamin otro (lSB No. 8292)
710 N 6th Street
Boise, ID 83701
Ph: [phone] x 12
Fax: [phone]
botto@ idahoconservation.org
Attomey for the Idaho Conservation League
IN THE MATTER OF IDAHO
POWER COMPANY'S
APPLICATION TO IMPLEMENT
SOLAR INTEGRATION RATES AND
CHARGES.
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
CASE NO. IPC-E-14-18
IDAHO CONSERVATION LEAGUE
DIRECT TESTIMONY
CAMERON YOURKOWSKI
EXHIBIT 205
EXCERPTS FROM:
IEEE: THE EVOLUTION OF WIND POWER INTEGRATION STUDIES
&
NVENERGY: LARGE-SCALE PV INTBGRATION STTTDY
&
ARIZONA PUBLIC SERVICE COMPANY: SOLAR PHOTOVOLTAIC (Pv)
INTEGRATION COST STUDY
From ldaho Power Response to ICL Production Request No. 5
The Evolution of Wind Power Integration
Studies: Past, Present, and Future
Erik Ela. Menrber, IEEE, Michael Milligan, Mentber, IEEE,Brian Parsons, Menber, IEEE,Debra
Lew, Menber, IEEE, and David Corbus, Mentber, IEEE
Abstract-The rapid growth of wind power as a generation
resounce in the past decade has given many utilities and Regional
Transmission Organizations (RTO) concerrs due to its
unconventional characteristics. Because of these concerns, many
of these entities have initiated studies that evaluate the feasibility
of large amounts of wind power onto their system and the
operational impacts present. This paper will discuss some of the
past major studies, mostly focusing on the United States, and the
basic methodologies that were used during these studies. The
paper will also review many of the different results and
conclusions of the studies and discuss how they have helped the
power industry as a whole. Lastly, the authors will attempt to
share their ideas on some of the limitations of the current and
past integration studies, and some insight on how these may be
evolving in the future.
Index Terns- Power system economics, power system
operations, power system planning, power system reliability,
power systems, wind energy, and wind power generation
I. lN'rnoorrr:rloN
\f,/fNO power iu the United States and the workl hasv v experienced substantial growth in the past decade. Its
zero-cost fuel and emissions-free output has been a tavorahle
alternative to volatile-priced fbssil fuel generation in a
changing environmental climate. Because utility-scale wind is
such a new resoufce and is incleasing at such a rapid rate,
utilities and systern operators are hecouring concerned about
keeping up with the integration issues and costs that it brings.
For this reas()n, many of the utilities and system operators
have established wind power integration studies for their areas
tll-tl2l, [221. These inte-eration studies usually simulate a
power systern irr the future with a large penetration of wind,
and evaluate the system impacts on the grid and incrernental
operating costs tlrat are incurred [31. The impacts vary flom
study to study, region to region. but often show similar
conclusions.
E. Ela is tith thc National Renervablc F-ncrgv Laboratoru. Goldcn. ('O
80!101 tlSA (e-nuil : [e-mail];.
M. Milligun is rvith the National Renervrble Encrgy Laboratory. Goldcn.
CO 80401 USA (c-rrrail:rnichael rnillisan(4nrcl.sov).
B. Parsurs is rvith thc Nrtional Renewablc F.nergy l,aborttory. Golden. ('o
8040 I llSA i e-nui I : brirn_palsons @r nrcl.gor ).
D. Lew is with tlre National Reneu'able F.nersv Lahoraton. Golden. Co
8040| tISA (c-rmiI: debra [e-mail],).
D. Corbus is uith lhc National Reneuuble Fircrgy I-aboratory. Golden. ('o
8040 I LISA t c-nuil : drvid,corbus (dr nrel.-sor ).
97 8- l -1244-421 I -6109/525.00 02009 I El.-Ll
The results of these studies are used in many difterent
ways. For instance, the New York integration study of 2005
[], studied the impacts of 3300 MW ol wind (10%, oi peak
capacity) on its system and eventually incorporated that
number in its market services tariff fl41. ln other studies [4],
[6], integration costs that are fbund through the study are
considered Ibr use as a basis tbr an integration rate that the
utility charges incorning wind power. In these cases, the wind
integration cost may be used as part of a comparison of costs
of altenrative fbrms of power generation and considered in the
resource acquisition process. Many results are used for rnore
infbrmational purposes tll-t l2l,t l9l. The results and
conclusions that come out of these studies give operations and
system planners the infbrrnation they need to plan fbr
increasing wind penetrations on their systems, and are otien
ret'erenced frequently during stakeholder meetin-es and
working -eroups when deciding on new operational and market
rules.
The National Renewable Energy Laboratory (NREL),
under the sponsorship of the U.S. Department of Energy
(DOE), has initiated two large regional wirrd power
integration studies in 2008. These regional wind power
integration studies were initiated for many reasons including
to support the U.S. 20% wind ener-ey by 2030 vision oi DOE
[5]. The Western Wind and Solar lntegration Study
(WWSIS) includes the WestConnectr utilities in the states ol'
Wyorning, Colorado, New Mexico, Arizona, and Nevada.
lnside this tixrtprint, the study is evaluating penetrations of
30%, of total energy consumptiorl with wind ener-uy and 5%,
with solar energy.t The study is rnocleling the entire Western
Electricity Coordinating Council (WECC) outside of the
WWSIS study tixrtprint with 20% wind and 37c solar also, to
address concerns that variability would be exported out fiorn
the WestConnect rltilities to the rest ol WECC. The outcornes
of the study should show results from a re-rional scale on
integration costs, ancillary services needs, and other irnpacts
that a large rnix ol wind and solar have on the westem study
area. [n the Easteln Wind Integration and Transmission Study
(EWITS) project, most of the Easteln Interconnection is being
studied fbr sinrilar integration irnpacts. This study includes
Midwest lSO, PJM, SPP, TVA, Mid-Continent Area Power
Pool (MAPP), ISO New England, New York lSO, and other'
interested parties. EWITS will include both a wind irrtegration
I WestCtunrct Iitilities consists of rbout 1,1 utilities in souths'estenl [].S.
More irttbrnrution can be tbund on ws !v.\vestconncct.conlI wwSIS included solar penerrations ol J.5'ri dler,gv Concentrulir)g Solar
Pouer and 1.5(l Photovoltaic to rnake up the -ir.? solar in thc study'.
cornponent and a transmission expansion componetlt. The
study will evaluate the dift'erent integration irnpacts of 207o
wind as a percentage ol total energy consumption and the
different impacts of having large transmission build<tut to
send wind from high wind-potential areas to large load
ceuters. The WWSIS and EWITS studies are planned to be
complete in 2009 and tbllowing the completion of this paper
[25] and [26]. However, the authors will share some of the
experiences gained fiom these two studies throughout the
paper where appropriate.
Though many of these wind integration studies come up
with dittbrent results and are olietr perfbrmed filr different
purposes, the general procedures and methodologies used in
pertbrrning a wind power integration study usually fbllow a
somewhat consistent structure. This paper will discuss the in-
depth methodologies and conclusions resulting from these
wind power integration studies, and will show how they have
evolved in the past several years, including assumptions attd
observations from the authors on changes predicted in future
studies. Section ll will cover the introductory stage of the
integration study. This includes the methodolo-eies used in the
data gathering that is required fbr all study pieces. One of the
larger tasks of this part ol the study is the undertaking
involved with coming up with wind resource data that is
needed to model luture wind power output. The arrangement
of assumptions and scenario developments needed befbre the
analysis can be conducted are also discussed. Section III will
go into some of the detailed analysis that goes into an
integration study of this type. This will cover statistical
analyses, production cost simulations, and reliability-based
assessments. In section IV, we will discuss many of the
results and conclusions that come out of these studies. Sectkrn
V will introduce the authors' thoughts on certain limitations
that are present in some of the past studies, and where we
believe trends will continue and changes will occur in future
integration studies. Section VI will provide a conclusion to the
paper. lt is important to note that many ditfbrent studies
involving wind and renewable energy have heen conducted.
This paper will lbcus prirnarily on U.S. studies along with
results and observations of international studies where
appropriate. and will be limited in scope to studies involved in
analyzing power system operations of moderate to high
penetrations of wind power.
ll. Dern Geruentuc AND ScENARIo DEVELoPMENl
Wind power integration studies that have been pertbrmed
in the last several years in the United States have increased in
cornplexity, realism, and geographic scope. This evolutitx is
expected to continue fbr the fbreseeable future. For a typical
integration study, a significant et'fbrt is devoted to obtaining
sirnulated wind plant data that is physically cousistent with the
underlying weather driver and with load. To achieve this,
wind data fi'om the same year as load data is required.
Although actual wind plant data can be used, the fircus of
most studies has been to analyze the irnpact ol a future wind
scenario, representing wind plants that have not yet been built.
Typical integration studies analyze 3 years of wind and
load data in an attempt to capture the inter-annual variability
of the weather. Production or market simulations typically
require hourly wind energy estimates, but most studies also
use 5-minute or [0-minute wind data fbr statistical analysis.
These wind data are developed using a Numerical Weather
Prediction (NWP) model that recreates historical weather in
time and space. Wind speed data can be extracted tiorn the 3-
year model runs at surface heights that correspond to the wind
turbine hub height and converted to wind power using wind
plant power curves. The gridded outputs of the NWP are the
building blocks that can be aggregated to model hypothetical
wind plants of various sizes. As an exarnple, in WWSIS, each
grid cell in the NWP sirnulation is about 2 km x 2 krn,
yielding a wind capacity per grid point of about 30 MW. This
represents l0 3-MW turbines placed throughout the grid cell.
A 90 MW wind plant would consist of the surn of the output
of 3 grid cells. More detailed infbrrnation on the wind
resource modeling process can be fbund in [61.
Land use restrictions are applied so that wiud development
is excluded in urban areas, national parks and environnrentally
sensitive areas, and other unlikely developed areas. For wind
power integration studies involving only one state, utility or
RTO, scenarios may be designed rnanually, as promising wind
locations are often known locally. For large, re-sional wind
power integration studies such as WWSIS and EWITS,
scenario selection can be a difficult task because of the
extremely lar-Ee number of potential locations, and rnust be
partially automated. Wind scenarios are typically designed
using existing or planned wind plants, plus various site
selection algorithms tbr new sites which may include a
combination of the tbllowing selection criteria: wind plant
capacity thctor, load correlations, proximity to existing or
proposed transmission corridors, and geographic diversity
[231. A family of scenarios allows utilities to answer more
questions than simply the cost of wind integration. By
comparing dift'erent scenarios, the utility can examine issues
such as the advanta-ees or disadvantages ol local versus
remote wind resources and impacts of geographic diversity on
integration cost.
To realistically simulate power system operation, the
uncertainties associated with load forecast errors and wind
forecast effors are important. Because wind and load forecast
etrors are generally statistically independent, they do not add
arithmetically, and should be developed for the simulation in
as realistic a numner as possible. Utilities ofterr do uot save
old bad tirrecasts, but these fbrecasts provide valuable
inlbrmation tirr the integration study. Because wind power is
simulated for the integration study, it is also necessary to
simulate wind power tbrecasts. This is because the unit
commitment process is done to target the conrbined load and
wind lbrecast. whereas the fbrecast errors will not become
apparent until the operating hour. This approach is sirnilar to
what actually happens during power systern operations. and is
a valuable component ol any wind power integration study. At
' Lord con'clation irr this scrrsc is typically lmking rt ccrtain high-load hours
and comparing rvirrd tutput dtu'ing these hours.
NAvTcANT
EiIE*GY
LencE-Scnrn
Prepared for
PV IUTEGRATIox SrunY
NVEnergy
NV Energy
6226 1N est Sahara Avenue
Las Vegas, NV 891"45
Jurv 30,2011
Navigant Consulting, Inc.
77 South Bedford Street, Suite 400
Burlington, MA 0L803
Sandia National Laboratories
Albuquerque, NM 87185
Pacific Northwest National Laboratory
Rictrland, Washington 99354
Silda
tlailiorClturffiis kiftc Ncthtrtet
MIIOT.IAL LAAORATOR/
NAVI CANT @ffi*v-F,"k#,k**
1. Develop time-syndrronized PV minute-by-minute output profiles and day-ahead
forecasts of PV and DG hourly output. Analyze the results to quantify the impact on
grid operations of increasing amounts of large-scale PV generation and DG.
Identify the additional
regulation and load following requirements, including capacity and ramp rate,
associated with each case study. Determine whether existing generation fleet can meet
these requirements without and with generation redispatch. Quantify the impact of PV
variability on generation fleet cycling and movements in regulation and load following,
to establish a basis for the assessment of generator wear and tear.
Integrate the results of the balancing area studies and the PV and DG output profiles to
quantify the impact of increasing levels of large-scale PV and DG on NV Energy's
generation mix and production costs using hourly production simulation models.
Identify the increased fuel and operations and maintenance (O & M) costs for existing
generating units caused by higher operating reserves, increased cycling, higher heat
rates, and changes in dispatch schedules.
Evaluate the impact of incremental DG on NV Energy's bulk power transmission
system, including steady state and dynamic performance within NV Energy's southem
Nevada balancing area operations.
Identify mitigation strategies or upgrades required to accommodate variable generatiory
including those needed to satisfy NERC balancing area performance requirements.
Assumptions
The evaluation of large PV and DG is performed assuming existing conditions, including a
system grid configuration and generating resource mix for 2011.
Study assumptions include:
o The study assesses the ability of the system as it exists today, except that load and
weather data from 2007 are used. Solar data needed to predict lntermittenry is not
readily available for 2008 and beyond. 2007 is also the year when NV Energy
3.
4.
5.
11
soLAR PHOTOVOLTATC (PV)
INTEGRATION COST STUDY
B&V PROJECT NO. 174880
PREPARED FOR
Arizona Public Service Company
NOVEMBER 2012
Principal lnvestigators:
Tim Mason, Project Manager
Trevor Curry
Mon Hong
Benson Joe
Scott Olson
Mary Sprouse
Dan Wilson
BLACK&VEATCH
,.riia, I t" ,re:,.:l t(r Lrr:r, _.:...1 j'
l" euilaing a world of difference..
Arizona Public Service Company I PV) integration cost Study
4 Reserve Requirement Determination and Cost
This analysis seeks to quantify and value the incremental operating reserves needed in order to
integrate the anticipated PV capacity on the APS system in2020 and 2030. As discussed in Section
3, Black & Veatch used the current NERC CPS2 standard as a proxy to measure the reserves
required and as the basis to estimate the costs of future reserves. Specifically, Black & Veatch
considered the requirements and cost to achieve CPS2 compliance at the 90 percent, 95 percent,
and 99 percent monthly levels.
The methodology to calculate the quantity and costs for solar PV reserves was completed in a two
step process. The first step was to calculate the amount of incremental upward and downward
regulating reserves required to maintain a specified level of monthly CPS2 performance caused by
forecasting errors from the solar PV penetration levels in each case. The second step took the
amount of incremental regulating reserves calculated in the first step and modeled the cost impact
to the system using an electric system production simulation cost modele to capture the system
energy cost differential of providing the regulating energy margin.
4.T RESERVE CALCULATION METHODOLOGY
The incremental amount of reserve capacity required to maintain CPS2 compliance was calculated
as a difference from the reserve capacity required due to loads only and that due to the
combination of loads and solar. For the base case, it was assumed that APS would maintain a 99
percent CPS2 compliance fthe standard that APS has historically achieved). Sensitivity cases were
also generated to reflect different levels of CPS2 compliance (90 and 95 percentJ and using solar
profiles with greater variability.
To calculate reserve requirements for any given level of NERC CPS2 compliance, Black & Veatch
developed a spreadsheet model using Microso[t Excel. The spreadsheet requires forecasted hourly
loads and L0-minute expected generation inputs an entire year (the inputs used in the CPS2 model
are discussed in Section 2]. After the load and solar data are entered into the model the number of
CPS2 violations is calculated. Violations due to load only were first calculated, assuming perfect
solar forecasting. After the solar forecast variability was added, the incremental reserves required
beyond what was needed for load only during daylight hours was calculated. The net difference
between the actual load and solar generation and the forecasted load and solar generation is the
forecast error.
As discussed in Section 2, the approach taken to estimate the solar output profiles was
conservative, reflected in both the number of cloudy time periods forecast during daylight hours
and the level of variability seen in the actual solar output data used. While it is possible that the
level of variability (and hence CPS2 violations) could be lower, for planning purposes it is
appropriate to take a conservative approach. A sensitivity case was developed that includes
additional solar variability (more cloudy time periods and greater variation when there are clouds)
to assess the potential impact on the quantity of CPS2 violations. After completion of this revised
" ABB/Venty'x ProMod production cost model lvas used in this studr
Arizona Public Service Company | >O
dataset, the profiles were placed into the same CPS2 models as the base case, with the new level of
10-minute reserves calculated.
A number of assumptions were made in developing the CPS2 model for location of future projects,
technology type, level of variability in the output profile, and forecast approach. The model is
sensitive to changes in many of these inputs, impacting accuracy when estimating integration costs.
As actual projects are developed in the future and specific algorithms are used for estimating solar
output, the model should be revised. This will greatly increase the level of accuracy and confidence
in integration costs.
4.2 RESERVE REqUIREMENTS
APS' Lro in 2012 is 46 MW. In 2020 and 2030 the Lro is estimated to be 51 MW and 59 MW
respectively. Figure 4-1 depicts the calculated ACE for all the ten minute periods in a single day in
2020. Since the Lro is 51 MW for the year 2020, the ACE can stay at +/- St MW without a CPS2
violation for a ten minute period without having to deploy any additional resources, System
operators would require 10-minute reserves to bring the system back into tolerable L16 range to
avoid negatively impacting the system frequency. In instances where the system ACE experiences a
large deviation it is possible for system operators to use l-minute regulating reserves to bring the
system back into balance.
To maintain a 95 percent CPS2 compliance monthly average for the year in 2020 APS would need to
carry and deploy, on average, 8L MW of incremental regulation up and 81 MW of incremental
regulation down reserves during hours when the solar is potentially operating.
Figure 4-1 below depicts the interdependence of the load and solar forecasting error to the open
loopto 46t. In certain time periods the load and solar forecast error offset and keep the ACE low.
In other periods the ACE is high because the load and solar forecast errors are both moving in
directions that make the ACE worse.
"'Open loop ACE is the Area Control Error of the s1'stem befbre AGC dispatch signals are deploy'ed to correct fbr
ACE. Closed loop ACE is the Area Control Error after AGC dispatch signals have deplol'ed to correct fbr ACE.
Arizona Public Service Company I SoLAR PHoTOVoLTAIC (PV) integration cost Study
-
ls3fl Forecasting Error r$916p Forecasting Error
Figure 4-1 lncremental Regulating Reserves and ACE
4.2.1. Year 2020 Reserve Requirements
APS applied the methodology used by Black & Veatch to develop the incremental monthly reserve
requirements necessary to maintain the CPS2 standards at 90 percent, 95 percent and 99 percent.
Table 4-1 provides the monthly reserve requirements for different levels of CPS2 compliance
during daylight hours, while Figure 4-2 depicts the inter-temporal reserve requirements for 2020.
The incremental 10 minute reserve requirement applies only to hours when solar generating
output is available.
Arizona Public Service Company | j,i.) ,in ii!f ):/-.r\1ril--lri- iPV) integration cost Study
Table 4-1 2020 Monthly 10-minute PV Reserves (+/- MW)ry
Month
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
sep
Oct
Nov
Dec
Monthly
Avg
June-Sept
Avg
90% 95Yo
't17 136
95 94
137 140
126
50
31
-4
7
25
74
128
107
74
15
122
57
52
15
24
40
88
1',t1
91
81
33
99%
133
103
139
221
96
75
53
61
68
140
95
90
106
64
9OYo
119
97
141
133
55
35
-3
10
28
76
133
107
78
18
95%
133
92
141
145
62
62
20
25
41
85
116
89
84
37
99%
133
94
144
221
142
70
53
6'r
73
145
96
90
107
64
250
200
150
100
50
0
-50 Apr [/ny-lun lul Aug-lsp Oct-llo-v,Dec
-90%
Compliance
-95%
Compliance
-99%
Compliance
Figure 4-210-minute Reserve Requirement by Month in Year 2020 (+/-MW)
The amount of incremental regulating reserves required to integrate solar is less in the summer
compared to other seasons during the year. This is the result of two factors. First, in the absence
of incremental PV APS requires more reserves during the summer because loads are highest during
this time and have greater variability. The higher amounts of reserves used to balance out ACE
deviation caused by load forecasting errors also aids in solar integration. Second, there is less
cloud coverage during summer monthsrr which leads to less solar variability in the model, reducing
" Ma1'through September a\/erages 20 clear da1,s in Phoenix. while the rest of the ),ear avcrages 16. See
http://www^publ ic.asu.edu/..aunj s/C I i mateofPhoeni r/rvxpaft4.htm#sun3
Arizona Public Service Company I jOl-An Pr() fi:'"iO1:q I IPV) integration cost Study
the need for reserves. The algorithm used to estimate variability in the solar datasets used the
presence of clouds as the ftigger for whether or not to add variabiliry for a given ].0 minute period.
The net solar dataset shows the highest standard deviation between 1.0 minute periods from
January to April, with relatively low variability from May to August. Review of actual L0 minute
datasets provided by APS show mixed variability levels depending on location; since the model
algorithm did not distinguish between partially cloudy and full cloudy days, this leads to months
with full clear days (summer) having the least variation. The assumptions made in the model are
conservative, likely producing more variation in the non-summer months than may actually exist.
As can be seen from Table 4-L, the "High Variability" case did not lead to a significantly greater
amount of CPS2 violations in any of the cases. While the reasons are not entirely clear, this is likely
due to three factors. First, the anticipated geographic diversity of the PV resources greatly
smoothes out the aggregate generation profile. Referring back to Figure 2-3, this shows that for the
aggregated solar profile in 2020 in one of the most volatile months (fanuaryJ, the average change
from nameplate capacity over a 10 minute period is 2 to 3 percent. If the level of variation included
in the dataset increases by 50 percent (as was assumed in the High Variability case), the maximum
impact is modest (an increase in roughly 1 percent of nameplate, or 10 MWJ. However, this
maximum impact assumes that the variation in each solar project is in the same direction, which is
not the case. This leads to the second factor, that although individual project variation is now
greater, the geographic diversity leads to some projects varying in the up direction while others
vary downwards, eliminating some of the net impact. Finally, in modeling the High Variability solar
datasets, additional time periods were included where clouds, and hence variability, is present.
This does not necessarily lead to more CPS2 violations than the Base Case, since it is the level of
variation, not the number of time periods where variation is present that is most important when
calculating CPS2 violations. The level of reserves calculated in the Base Case may be sufficient to
handle the majority of new variation created in High Variability time periods where there originally
were none.
4.2.2 Year 2030 Reserve Requirements
The reserve requirements for 2030 are approximately 60 percent greater than the 2020
requirements at 99 percent CPS2 on an annual basis, consistent with the increase in solar PV. The
summer requirements do increase at a greater rate, but still represent a small portion of the total
installed PV capacity. Table 4-2 provides the monthly reserve requirements during daylight hours
for different levels of CPS2 compliance, while Figure 4-3 depicts the inter-temporal reserve
requirements throughout the year.
Benjamin otto (lSB No. 8292)
710 N 6th Street
Boise, ID 83701
Ph: [phone] x 12
Fax: [phone]
botto@ idahoconservation.org
Attorney for the ldaho Conservation League
IN THE MATTER OF IDAHO )
POWER COMPANY'S )
APPLICATION TO IMPLEMENT )
SOLAR INTEGRATION RATES AND )
BEFORE THE IDAHO PUBLIC UTILITIES COMMISSION
CASE NO. IPC.E-14-I8
IDAHO CONSERVATION LEAGUECHARGES.
DIRECT TESTIMONY
CAMERON YOI,RKOWSKI
EXHIBTT 206
IDAHO POWBR RESPONSE TO ICL DATA REQUEST NO.6
For Requests No. 6 through No. 1O please reference Exhibit 1 to the Direct Testimony of
Phil Devol [slc/, Haho Power Company Solar htegration Report (Solar Report).
REQUEST NO.6: Please reference pages 10-11of the Solar Report (pages t2- !3 of Exhibit 1):
a. When calculating the incremental reserve requirement for integrating solar generation,
does the Company net the decremental, incremental, and total capacity needs identified for solar
generation with the decremental, incremental, and total capacity held by the Company for managing
the variability of other generation and load? Please explain why or why not.
b. Does the Company acknowledge that the scheduling error associated with solar
generation nets with the scheduling error associated with other resources and load to form a total
system error signal that is statistically smaller than the sum of the errors of the individual parts? lf
the Company disagrees with this principle, please explain why.
C. How much decremental and incremental capacity (in MW) does the Company hold in
reserve to respond to wind scheduling errors?
d. Has the Company conducted an analysis of the effect of netting the scheduling
errors of solar, wind, other generation, and load on the total reserve requirement? f so, please
provide the resuhs of such analysis. lf not, please explain why the Company has not performed such
an analysis.
RESPONSE TO REQUEST NO.6:
a. ldaho Power does not net solar generation schedule errors with other generation
sources, other schedule errors, or bad. ldaho Power designed the solar
Exhibit 206. Page I ol3
IPC-E-1,1-18
C. Yourkowski, ICL
Souroe: Idaho Power Response to ICL Request No 6
integration study to identify the integration issues specifically associated with solar generation.
The objective of the solar integration study is to identify the effects of the intermittent solar
generation in order to calculate the integration cost imposed by solar upon ldaho Power's existing
system. daho Power's system design includes the capability of system dispatchable generators to
manage variability in customer load and other generation. htermittent solar generation introduces
new variability and uncertaintyintosystemoperations.
Because of the inherent differences and levels of confidence in load forecasts versus
forecasts for intermittent generation, such as wind and solar, load forecast errors are often auto
correlated, reflecting a tendency for forecast errors to persist in magnitude and dlrection throughout
the day, and are more readily addressed as the system is managed through to real time. However,
in order to maintain the reliable operation and stability of the system, as well as to meet its various
regulatory reliability criterla, the Company must provide adequate reserves based upon the higher
magnitude and nature of the forecast error present in intermittent and variable wind and solar
forecasts. Thus, the challenges in forecasting wind and solar as compared to load for unit commitment
are considerably different, requiring the system to treat differently the possibility of errors in
forecasting these elements of load and resource balance.
b. ldaho Power does not acknowledge that the schedullng error associated with solar
generation nets with the scheduling error associated with other resources and load to form a total
system error signal that is statistically smaller than the sum of the errors of the individual parts.
Please see the Company's response to 5.a above.
Exhibit 206. Page 2 of3
IPC-E- t4- 18
C. Yourkowski. ICL
Source: Idaho Power Response to ICL Request No 6
C. ldaho Power holds a minimum of .25 per megawatt ("MW") for up to 240 MW of
wind.or 60 MW incremental and 70 MW of decremental reserve for scheduling errors in real time;
however, this is only a minimum and in most cases the operator will determine what amount of
reserve is required beyond the minimum based on system conditions at that time. Additionally,
the Company is required to hold 5 percent contingency reserve for the amount of wind generation on
the system at any given time.
d. ldaho Power has not conducted an analysis of the effect of netting the
scheduling errors of solar, wind, other generation, and bad on the total reserve requirement.
Please see the Company's response to 5.a above.
The response to this Request is sponsored by Phil DeVol, Resource Planning Leader,
ldaho PowerCompany.
Exhibit 206. Page 3 of3
IPC-E-r4-I8
C. Yourkorvski. ICL
Source: Idaho Power Response to ICL Request No 6
CERTIFICATE OF SERVICE
I hereby certii/ that on this 23rd day of October 2014, I delivered true and correct copies
of the foregoing DIRECT TESTIMONY OF CAMERON YOURKOWSKI the following persons
via the method of service noted:
Hand deliverv:
Iean Iewell
Commission Secretary (Original and 7 copies provided)
Idaho Public Utilities Commission
427 W. Washington St.
Boise, ID 83702-5983
Electronic Mail:
Donovan E. Walker
Greg Said
Michael f. Youngblood
Regulatory Dockets
Idaho Power Company
1221 West Idaho Street
P.O. Box 70
Boise, lD 83707
dwalker@idahopower. com
[e-mail]
[e-mail]
dockets@idahopower. com
Kelsey J. Nunez
Ken Miller
Snake River Alliance
PO Box 1731
Boise,ID 83701
kmiller@snakeriveralliance.o rg
[e-mail]
Matt Vespa
Sierra Club
85 Second St., 2nd Floor
San Francisco, CA 94105
matt.vespa@sierraclub. org
Dean J. Miller
McDevitt & Miller LLP
420 W. Bannock St.
Boise,ID 83702
[e-mail] rkBenjamin I. Otto