20220317IPC to Staff 29-43 Attachment 32.pdf

3Em.RECEIVET)
LISA D. NORDSTROTUT
Lead Councel
[e-mail]
2022 MAR 1/ Ptt'1 2:59
IDAHO PUBt IC
I]TII. ITIES COIVIN/I SSION
March 17,2022
VIA ELECTRONIC FILING
Jan Noriyuki, Secretary
ldaho Public Utilities Commission
11331 W. Chinden Blvd., Bldg 8,
Suite 201-A (837141
PO Box 83720
Boise, ldaho 83720-001 4
Re Case No. IPC-E-2142
!n the Matter of ldaho Power Company's Application forApproval of Special
Contract and Tariff Schedule 33 to Provide Electric Service to Brisbie, LLC's
Data Center Facility
Dear Ms. Noriyuki:
Attached for electronic filing, pursuant to Order No. 35058, is ldaho Power
Company's Confidential Response to the Confidential Second Production Request of the
Commission Staff in the above entitled matter.
Due to the voluminous nature of confidential and non-confidential information
provided in response to the data requests in this case, the Company is posting the
attachments to these requests to the secure FTP site to allow the parties to view the
requested information remotely. Because some of the attachments contain confidential
information, the FTP site is divided between confidentialand non-confidential information.
The login information for the non-confidential portion of the FTP site will be provided to
the parties in a separate email. The login information for the confidential portion of the
FTP site was provided to the parties on February 23,2022.
PIease feel free to contact me directly with any questions you might have about
this filing.
Very truly yours,
An EAiCORP COltlPanY
X; !.7("t-+r^,
LDN:sg
Enclosures
Lisa D. Nordstrom
CERTIFICATE OF ATTORNEY
ASSERTION THAT INFORMATION CONTAINED IN AN IDAHO PUBLIC UTILITIES
COMMISSION FILING IS PROTECTED FROM PUBLIC INSPECTION
Case No. IPC-E-2142
ldaho Power Company's Application for Approval of Special Contract and Tariff
Schedule 33 to Provide Electric Service to Brisbie, LLC's Data Center Facility
The undersigned attorney, in accordance with RP 67, believes that ldaho Power
Company's Confidential Response to the Confidential Second Production Request of
the Commission Staff dated March 17, 2022, and its aftachments, may contain
information that ldaho Power Company or a third party claims is confidential as
described in ldaho Code S 74-101, ef seg., and $ 48-801, ef seg., and as such is
exempt from public inspection, examination, or copying.
DATED this 17th day of March 2022
,4" 1.4^1"+,,.^,,
Lisa D. Nordstrom
Counselfor ldaho Power Company

Variable Energy Resource
(VER) Integration Analysis

Idaho Power Company
December, 2020
PUBLIC

ii | P a g e

PUBLIC

© 2020 Copyright. All Rights Reserved.
Energy and Environmental Economics, Inc.
44 Montgomery Street, Suite 1500
San Francisco, CA 94104
415.391.5100
www.ethree.com
Variable Energy Resource
(VER) Integration Analysis

Idaho Power Company
December 2020
PUBLIC

iv | P a g e

Executive Summary
Energy and Environmental Economics, Inc. (E3) was retained by Idaho Power to
investigate the integration cost of variable energy resources in Idaho Power’s
service territory. These costs are incurred due to increased dispatchable unit
cycling (from increased unit stops and starts; increased load following ramping)
and imperfect unit commitment and dispatch (resulting in higher average thermal
unit heat rates and/or lower net market earnings); and, in cases in which
economic VER curtailment is allowed, increased curtailment costs. E3’s analysis
calculates both average and incremental integration costs on a $/MWh basis,
using the proposed unit additions and retirements to Idaho Power’s 2023 system
as a base case.
The study examines eleven cases of potential future VER builds in Idaho Power
territory. These cases are illustrated below in Table ES1. These include high wind
and high solar builds; cases in which low, average and high annual hydro energy
budgets are simulated; cases in which there is solar plus investment tax credit
(ITC)-enabled storage; cases in which solar can be economically curtailed; and a
case in which a planned unit retirement at the Bridger coal plant is not in effect
in 2023. As can be seen in Table ES1, the overall incremental integration costs
were found to range from $0.64/MWh-$4.65/MWh. Generally, these costs are
lower than those in the 2018 Idaho Power VER Integration Analysis, although it is
PUBLIC

notable that the method of deriving integration costs was substantially different
in the last study.1
Table ES1: Case Description and Results Summary

E3 believes that the integration costs in this study are lower than previous studies
primarily due to four factors: 1) Reduced need for modeled ancillary services, 2)
The fact that the remaining 2023 coal fleet is modeled as must-run (i.e. its
commitment decisions are not affected by VER penetration), 3) Access to the
Energy Imbalance Market (EIM) makes it easier to use market transactions to

1https://docs.idahopower.com/pdfs/AboutUs/PlanningForFuture/wind/VariableEnergyResourceIntegrationAn
alysis.pdf
Case Description
First
Bridger
Unit
Proposed
Existing
2023
Wind
Capacity
(MW)
Hydro
Year
New
2023
Solar
Build
(MW)
New
2023
Wind
Build
(MW)
Total
Integration
Cost
1 Base 2023 Case Retired 561 728 Normal 0 0 No 0 $ 2.93
2
Base Case + First
Bridger Unit Online Online 561 728 Normal 0 0 No 0 $ 3.61
3 High Solar Retired 561 728 Normal 794 0 No 0 $ 3.86
4
High Solar, Low
Hydro Retired 561 728 Low 794 0 No 0 $ 4.55
5 High Wind Retired 561 728 Normal 0 669 No 0 $ 0.77
6
High Solar, High
Wind Retired 561 728 Normal 794 669 No 0 $ 2.46
7
Existing Solar Base
Case Retired 310 728 Normal 0 0 No 0 n/a
8
High Solar, High
Hydro Retired 561 728 High 794 0 No 0 $ 4.65
9
High Solar + 200
MW Storage Retired 561 728 Normal 794 0 No 200 $ 0.64
10
High Solar + 400
MW Storage Retired 561 728 Normal 794 0 No 400 $ 0.93
11 Curtailable Solar Retired 561 728 Normal 794 0 Yes 0 $ 3.13
Proposed
Existing
2023
Solar
Capacity
(MW)
Amount of New
VER Added to
Existing 2023
Builds
New
Solar-
Coupled
4-hr Li-
Ion
Battery
Build
(MW)
Can New
Solar be
Curtailed?
PUBLIC

vi | P a g e

integrate VERs (the EIM was not included in the previous study) and 4) Allowing
additional system flexibility, in some cases (e.g. from batteries).
The integration costs calculated in this current effort specifically do not consider
fuel savings or capacity contributions from, nor do they consider the capital costs
of new VERs. Therefore, this VER integration cost study serves as a valid basis for
calculating integration costs but may not include all economic and operational
factors required to integrate VERs on the Idaho Power system.
PUBLIC

Table of Contents
Executive Summary ...................................................................................... iv
1 Introduction ............................................................................................ 1
1.1 Motivation and Background ................................................................ 1
2 Methodology ........................................................................................... 2
2.1 Calculating VER Integration Costs .................................................... 2
2.2 Production Cost Modeling ................................................................... 6
2.3 Reserve Modeling ................................................................................. 8
3 Data Collection, Processing and Methods ....................................... 10
3.1 PLEXOS Modeling .............................................................................. 10
3.1.1 Load Profiles, VER Profiles and Dispatchable
Generation Fleet ................................................................. 10
3.1.2 External Market Representation ...................................... 13
3.2 RESERVE Modeling........................................................................... 16
3.2.1 Derivation of 2023 VER Profiles ...................................... 16
3.2.2 Deriving Reserves Components ...................................... 18
3.3 Case Matrix .......................................................................................... 19
4 Results ................................................................................................... 22
4.1 RESERVE Outputs ............................................................................. 22
4.1.1 Annual Average results ..................................................... 22
4.1.2 Detailed Reserve Results ................................................. 26
4.2 2019 PLEXOS to Historical Case Benchmarking ......................... 31
PUBLIC

viii | P a g e

4.3 2023 Case Result Summary ............................................................ 33
4.4 System Dispatch Results .................................................................. 35
4.4.1 Existing Solar, 2023 Base Case and Jim Bridger First
Unit Online Cases .............................................................. 35
4.4.2 High Solar, High Wind, and High Solar + Wind Cases 39
4.4.3 High Solar with Low, Average and High Hydro Budgets
............................................................................................... 42
4.4.4 High Solar With and Without Storage ............................ 45
4.4.5 High Must Take Solar and Curtailable Solar Cases .... 49
5 Discussion ............................................................................................ 53
5.1 Discussion of Current Study Results .............................................. 53
5.1.1 Effects of Binding Pmin Constraints on VER Integration
Costs ..................................................................................... 53
5.1.2 High Solar With Storage Cases ....................................... 56
5.2 Comparison to Data in Literature and 2018 Idaho Power VER
Study ..................................................................................................... 59
5.3 Methodological Differences between 2020 and 2018 Idaho Power
Company Variable Energy Resource Analysis ............................. 60
5.3.1 Overview .............................................................................. 60
5.3.2 Reserves .............................................................................. 61
5.3.3 Treatment of External Markets ........................................ 65
5.3.4 Multistage vs. Single Stage Model .................................. 65
6 Conclusions .......................................................................................... 67
6.1 Integration Costs ................................................................................. 67
7 Appendix 1: Process Document ........................................................ 68
PUBLIC

7.1 Introduction ........................................................................................... 68
7.2 Results Processing ............................................................................. 74

PUBLIC

x | P a g e

PUBLIC

P a g e | 1 |
Introduction
© 2010 Energy and Environmental Economics, Inc.
1 Introduction
1.1 Motivation and Background
In 2019, Idaho Power committed to using 100 percent clean energy by 2045.
While more than 50 percent of Idaho Power’s annual load was served by clean
resources in 2018 (primarily from hydroelectricity, with some additional wind and
solar resources), Idaho Power may potentially add significant amounts of variable
energy resources (VERs), such as wind and solar power, to achieve this 2045 goal.
Energy and Environmental Economics (E3) was retained by Idaho Power to
perform a study of the cost of integrating new VERs into Idaho Power’s system.
Idaho Power has periodically performed these studies and analyses to inform
regulatory proceedings, and to determine integration charges included in Public
Utility Regulatory Policies Act (PURPA) contracts. Idaho Power hired E3 to update
integration costs. E3 conducted this analysis by designing a suite of scenarios that
are relevant to the 2023 timeframe.
The following report details the modeling methodology, data collection and
assumptions, and results from E3’s 2020 investigation of VER integration costs for
Idaho Power.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 2 |
2 Methodology
2.1 Calculating VER Integration Costs
E3 used five metrics to estimate the total cost of VER integration to Idaho Power’s
system. These were:
 Start/Stop Costs: The costs resulting from changes in unit start and stop
counts due to differing patterns of net load fluctuations caused by higher
VER penetration
 Ramping Costs: The costs resulting from changes in unit ramping due to
differing patterns of net load fluctuations caused by higher VER
penetration
 Imperfect Unit Commitment and Dispatch Costs (Fuel Costs): The costs
resulting from holding a greater amount of committed dispatchable
resources operating at part load and lower efficiency. These resources
operate at part load to provide reserves necessary to manage increased
VER-induced forecast error and subhourly net load variability
 Imperfect Unit Commitment and Dispatch Costs (Net Import Costs): The
costs resulting from suboptimal market transactions due to holding more
headroom and footroom on generators to account for increased VER-
induced forecast error and subhourly net load variability
 Curtailment Costs: In all cases, VERs are assumed to be contracted on a
take-or-pay basis (i.e. all VER energy is paid for whether it is consumed or
not). However, in the case in which solar can be economically curtailed,
Idaho Power would incur a cost from no longer generating a renewable
PUBLIC

P a g e | 3 |
Methodology
© 2010 Energy and Environmental Economics, Inc.
energy credit (REC). This REC cost is factored into the integration cost for
that case.
The total VER integration cost for each case is calculated by summing each of
these costs.
To calculate these costs, E3 performed three model runs for each of the eleven
analyzed cases. In the first model run, E3 ran a 2023 “base case” model that
served as the reference point for each of the subsequent investigated cases. The
base case included potential unit additions and retirements, the relevant hydro
budget, as well as projected load growth from 2019 through 2023. Next, E3 ran
an intermediate “perfect foresight” case in which any new VER additions beyond
the 2023 base case have perfect foresight (i.e. no new forecast error reserves are
held vs. the base case), and the new VER profiles are “smoothed” on a subhourly
timescale (i.e. no new regulation reserves are held vs. the base case). This perfect
foresight case is designed specifically to look at the effect of forecast error and
subhourly variability from VERs on integration costs, not factoring in savings from
extra energy provided by new VER additions. Finally, E3 ran a case with higher
VER-induced regulation reserves and higher net load forecast error reserves. The
formulae for calculating integration costs from these three cases are provided
below. In the formulae, “Case j” refers to an individual case for which E3
calculated the VER integration costs. The “base case” is the 2023 base case
common to all but two of the evaluated cases. The remaining two cases are the
2023 base case and the base case with Bridger Unit 1 cases. These use the existing
solar case instead of the 2023 base case due to the need for an incremental VER
build to assess the integration costs in the equations provided below. The
resulting Total Integration Costs pursuant to this study are calculated in units of
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 4 |
$/MWh. The graphical depiction of this three-part process is also shown below in
Figure 1.
𝑰𝒏𝒄𝒓𝒆𝒎𝒆𝒏𝒕𝒂𝒍 𝑺𝒕𝒂𝒓𝒕 𝑪𝒐𝒔𝒕𝒔 𝒇𝒐𝒓 𝑪𝒂𝒔𝒆 𝒋
=𝛴𝐴𝑙𝑙 𝑈𝑛𝑖𝑡𝑠 𝑆𝑡𝑎𝑟𝑡 𝐶𝑜𝑠𝑡𝑈𝑛𝑖𝑡 𝑖∗(𝐴𝑛𝑛𝑢𝑎𝑙 𝑆𝑡𝑎𝑟𝑡 𝐶𝑜𝑢𝑛𝑡𝑈𝑛𝑖𝑡 𝑖,𝐶𝑎𝑠𝑒 𝑗
− 𝐴𝑛𝑛𝑢𝑎𝑙 𝑆𝑡𝑎𝑟𝑡 𝐶𝑜𝑢𝑛𝑡𝑈𝑛𝑖𝑡 𝑖,𝐵𝑎𝑠𝑒 𝐶𝑎𝑠𝑒)
𝑰𝒏𝒄𝒓𝒆𝒎𝒆𝒏𝒕𝒂𝒍 𝑹𝒂𝒎𝒑𝒊𝒏𝒈 𝑪𝒐𝒔𝒕𝒔 𝒇𝒐𝒓 𝑪𝒂𝒔𝒆 𝒋
=𝛴𝐴𝑙𝑙 𝑈𝑛𝑖𝑡𝑠 𝑅𝑎𝑚𝑝𝑖𝑛𝑔 𝐶𝑜𝑠𝑡𝑈𝑛𝑖𝑡 𝑖
∗(𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑅𝑇5 𝑀𝑊 𝑅𝑎𝑚𝑝𝑖𝑛𝑔𝑈𝑛𝑖𝑡 𝑖,𝐶𝑎𝑠𝑒 𝑗
−𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑅𝑇5 𝑀𝑊 𝑅𝑎𝑚𝑝𝑖𝑛𝑔𝑈𝑛𝑖𝑡 𝑖,𝐵𝑎𝑠𝑒 𝐶𝑎𝑠𝑒)
𝑰𝒏𝒄𝒓𝒆𝒎𝒆𝒏𝒕𝒂𝒍 𝑰𝒎𝒑𝒆𝒓𝒇𝒆𝒄𝒕 𝑼𝒏𝒊𝒕 𝑪𝒐𝒎𝒎𝒊𝒕𝒎𝒆𝒏𝒕 & 𝑫𝒊𝒔𝒑𝒂𝒕𝒄𝒉 𝑪𝒐𝒔𝒕 𝒇𝒐𝒓 𝑪𝒂𝒔𝒆 𝒋
=𝛴𝐴𝑙𝑙 𝑈𝑛𝑖𝑡𝑠 𝐹𝑢𝑒𝑙 𝐶𝑜𝑠𝑡𝑈𝑛𝑖𝑡 𝑖∗(𝐹𝑢𝑒𝑙 𝑈𝑠𝑒𝑈𝑛𝑖𝑡 𝑖,𝐶𝑎𝑠𝑒 𝑗
−𝐹𝑢𝑒𝑙 𝑈𝑠𝑒𝑈𝑛𝑖𝑡 𝑖,"Perfect Foresight" 𝐶𝑎𝑠𝑒𝑗)+(𝑁𝑒𝑡 𝐼𝑚𝑝𝑜𝑟𝑡 𝐶𝑜𝑠𝑡𝐶𝑎𝑠𝑒 𝑗
−𝑁𝑒𝑡 𝐼𝑚𝑝𝑜𝑟𝑡 𝐶𝑜𝑠𝑡"𝑃𝑒𝑟𝑓𝑒𝑐𝑡 𝐹𝑜𝑟𝑒𝑠𝑖𝑔ℎ𝑡" 𝐶𝑎𝑠𝑒 𝑗)
𝑰𝒏𝒄𝒓𝒆𝒎𝒆𝒏𝒕𝒂𝒍 𝑪𝒖𝒓𝒕𝒂𝒊𝒍𝒎𝒆𝒏𝒕 𝑪𝒐𝒔𝒕𝒔 𝒇𝒐𝒓 𝑪𝒂𝒔𝒆 𝒋
=𝛴𝐴𝑙𝑙 𝑈𝑛𝑖𝑡𝑠 𝐶𝑢𝑟𝑡𝑎𝑖𝑙𝑚𝑒𝑛𝑡 𝐶𝑜𝑠𝑡𝑈𝑛𝑖𝑡 𝑖
∗(𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑅𝑇5 𝑀𝑊 𝐶𝑢𝑟𝑡𝑎𝑖𝑙𝑚𝑒𝑛𝑡𝑈𝑛𝑖𝑡 𝑖,𝐶𝑎𝑠𝑒 𝑗
−𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑅𝑇5 𝑀𝑊 𝐶𝑢𝑟𝑡𝑎𝑖𝑙𝑚𝑒𝑛𝑡𝑈𝑛𝑖𝑡 𝑖,"𝑃𝑒𝑟𝑓𝑒𝑐𝑡 𝐹𝑜𝑟𝑒𝑠𝑖𝑔ℎ𝑡" 𝐶𝑎𝑠𝑒 𝑗)
𝑻𝒐𝒕𝒂𝒍 𝑰𝒏𝒕𝒆𝒈𝒓𝒂𝒕𝒊𝒐𝒏 𝑪𝒐𝒔𝒕𝑰𝒏𝒄.,𝑪𝒂𝒔𝒆 𝒋
=(𝐼𝑛𝑐.𝑆𝑡𝑎𝑟𝑡 𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗+𝐼𝑛𝑐.𝑅𝑎𝑚𝑝𝑖𝑛𝑔 𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗 +
𝐼𝑛𝑐𝑟𝑒𝑚𝑒𝑡𝑛𝑎𝑙 𝐼𝑚𝑝𝑒𝑟𝑓𝑒𝑐𝑡 𝑈𝑛𝑖𝑡 𝐶𝑜𝑚𝑚𝑖𝑡𝑚𝑒𝑛𝑡 𝑎𝑛𝑑 𝐷𝑖𝑠𝑝𝑎𝑡𝑐ℎ 𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗+ 𝐼𝑛𝑐.𝐶𝑢𝑟𝑡.𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗)

PUBLIC

P a g e | 5 |
Methodology
© 2010 Energy and Environmental Economics, Inc.
𝑻𝒐𝒕.𝑰𝒏𝒕𝒆𝒈𝒓𝒂𝒕𝒊𝒐𝒏 𝑪𝒐𝒔𝒕𝑰𝒏𝒄.,𝑪𝒂𝒔𝒆 𝒋=
(𝐼𝑛𝑐.𝑆𝑡𝑎𝑟𝑡 𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗+𝐼𝑛𝑐.𝑅𝑎𝑚𝑝𝑖𝑛𝑔 𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗 +
𝐼𝑛𝑐.𝐼𝑚𝑝𝑒𝑟𝑓𝑒𝑐𝑡 𝑈𝑛𝑖𝑡 𝐶𝑜𝑚𝑚.𝑎𝑛𝑑 𝐷𝑖𝑠𝑝.𝐶𝑜𝑠𝑡,𝐶𝑎𝑠𝑒 𝑗
𝑉𝐸𝑅 𝐸𝑛𝑒𝑟𝑔𝑦 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙,𝐶𝑎𝑠𝑒 𝑗 −𝑉𝐸𝑅 𝐸𝑛𝑒𝑟𝑔𝑦 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙,𝐵𝑎𝑠𝑒 𝐶𝑎𝑠𝑒

Figure 1: VER Integration Cost Calculation Methodology

This methodology for deriving VER integration costs does not calculate various
costs and benefits from the VER additions. Additionally, this method does not
calculate fuel cost savings due to VER deployment, nor the capacity value of new
VERs in offsetting the need for firm generation unit additions. This method also
does not calculate capital or PPA costs associated with contracting new VERs.
Therefore, the future use of these VER integration costs must be done with
knowledge and awareness of what costs and benefits they omit.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 6 |
2.2 Production Cost Modeling
E3 used Energy Exemplar’s PLEXOS 7.2 Software2 to calculate the total production
costs associated with each evaluated case. The model uses load, VER, generator,
fuel cost and external market data provided by Idaho Power and other data
sources to calculate annual production costs for Idaho Power under varying
scenarios, which are then used to calculate VER integration costs. This is shown
schematically below in Figure 2.
In order to perform this modeling, E3 used a four-stage PLEXOS model. For each
day, the model sequentially solves the day-ahead (DA), hour-ahead (HA), 15-
minute (RT15) and 5-minute (RT5) markets. In each stage, the model is solved
completely (i.e. all units and transmission committed and dispatched). Then, any
unit commitment or other model decisions with a lead time longer than the next
phase’s lead time to the real time are passed down to the next stage. In this
manner, the model approximates the actual unit commitment and dispatch
constraints associated with the longer commitment times of thermal and
transmission markets. This captures the effects of greater average forecast error
and higher average reserves in model stages that are farther from the real time
on the ability of Idaho Power to efficiently commit long start units. This daily
sequential model execution process is depicted in Figure 3.

2 https://energyexemplar.com/solutions/plexos/
PUBLIC

P a g e | 7 |
Methodology
© 2010 Energy and Environmental Economics, Inc.
Figure 2: Using PLEXOS to Calculate VER Integration Costs

Figure 3: PLEXOS Multistage Modeling

The change in start/stop cost, and the imperfect unit commitment costs are
calculated endogenously in PLEXOS. However, E3 used data from the 2013
National Renewable Energy Laboratory’s (NREL) Western Wind and Solar
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 8 |
Integration Study: Phase 23 to estimate $/MW ramping costs for Idaho Power’s
thermal units. The annual total ramping costs were calculated as a post-
processing step by calculating the total annual MW of ramping in the RT5 stage
for each thermal unit, and multiplying that by the per MW ramping cost from
NREL. The $/MW values that E3 used are shown in Table 2 below.
Table 2: Ramping Costs Used in Study (Sourced from NREL4)
Value Coal Gas GT Gas CCGT
Median Ramping Cost ($/MW) $3 $2 $1
2.3 Reserve Modeling
E3 used its RESERVE tool5 to model 2019 and 2023 levels of hourly reserves that
Idaho Power needs to hold in each PLEXOS interval. This is done to account for
the fact that Idaho Power needs to hold reserves to manage net load forecast
error and subhourly net load variations in its daily operations.
Idaho Power’s participation in the California Independent System Operator’s
(CAISO’s) Energy Imbalance Market (EIM) means that Idaho Power holds reserves
of CAISO’s Flexible Ramping Product6 (FRP). It must do this so that it can trade in
the RT15 and RT5 EIM markets. Additionally, Idaho Power holds amounts of
regulation reserves and contingency reserves dictated by the North American

3 https://www.nrel.gov/docs/fy13osti/55588.pdf
4 https://www.nrel.gov/docs/fy13osti/55588.pdf
5 https://www.ethree.com/tools/reserve-model/
6http://www.caiso.com/informed/Pages/StakeholderProcesses/CompletedClosedStakeholderInitiatives/Flexib
leRampingProduct.aspx
PUBLIC

P a g e | 9 |
Methodology
© 2010 Energy and Environmental Economics, Inc.
Electric Reliability Corporation (NERC) and the Western Electricity Coordinating
Council (WECC).
While the derivation of contingency reserves is standardized (calculated as 3
percent of load and 3 percent of generation total, with at least half held as for
spinning reserves and the rest as non-spinning reserves), Idaho Power’s future
CAISO FRP and regulation reserve needs are unknown. This is because future VER
additions and load growth will increase the level of net load forecast uncertainty
on Idaho Power’s system relative to current conditions. Therefore, E3 used its
RESERVE tool along with Idaho Power’s 2019 forecast and actual load and VER
data to simulate reserves that approximate the CAISO FRP and regulation needs.
E3 also used RESERVE to calculate CAISO FRP and regulation reserves in 2019 to
enable a consistent baseline for model comparisons.
These contingency, CAISO FRP and regulation reserves were input to the PLEXOS
model such that the reserves are held in all time intervals. Further information on
the derivation of the 2023 load and VER profiles for each analyzed case are
provided in subsequent sections of this report.

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 10 |
3 Data Collection, Processing
and Methods
3.1 PLEXOS Modeling
3.1.1 LOAD PROFILES, VER PROFILES AND DISPATCHABLE GENERATION
FLEET
E3 collected forecast and actual gross load, wind and solar profiles for 2019 from
Idaho Power for the DA, HA, RT15 and RT5 phases. The VER data was on a plant-
level basis and covered most of Idaho Power’s existing PURPA and Idaho Power-
owned facilities, with only a few small wind and solar plants omitted from the
data collection process due to their small effect on net load forecast error. Idaho
Power also provided the total 2019 wind and solar nameplate build in Idaho
Power territory so that E3 could use a correct baseline VER build in its analysis.
Idaho Power’s 2019 average load was 1,980 aMW. To estimate 2023 loads, E3
used load growth projections from Idaho Power to uniformly increase 2019 loads
by approximately 5 percent total to 2,081 aMW. The method for deriving new
2023 VER profiles is detailed below, but the 2019 historical VER profiles were used
in all cases to derive the 2023 VER profiles.
In all cases, E3 modeled existing and proposed solar, solar + storage and wind
plants as qualifying facilities (QF) operating under PURPA. This means that, under
all circumstances except for one case, these resources are treated as must take
facilities.
PUBLIC

P a g e | 11 |
Data Collection, Processing and Methods
© 2010 Energy and Environmental Economics, Inc.
E3 chose to use 2019 load and VER data to derive 2023 load and VER profiles in
order to capture the spatial and temporal correlations between load, wind and
solar production and forecast error, as well as the typical hourly and seasonal
distributions of load, and VER production. Most of Idaho Power’s existing solar
capacity is modern, single-axis tracking utility solar, meaning that future
installations were likely to have similar annual capacity factors as existing arrays.
Idaho Power’s solar and wind is mostly distributed across the Snake River Plain
and Eastern Oregon, as shown below in Figure 4, because this is where the
majority of existing Idaho Power transmission and load is, and it is also the best
solar resource in Idaho Power’s service territory. Idaho Power stated that they
are likely to continue to add new VER resources within the Snake River Plain.
Therefore, E3’s use of 2019 VER profiles to represent future profiles is reasonable.
Idaho Power proposed that, for the 2023 base case, it was reasonable to assume
that 251 MW of new solar was online in their service territory (131 MW of
unspecified PURPA contracts and 120 MW from the planned Jackpot Solar
facility). Idaho Power also proposed that the 2023 wind capacity remain the same
as that from 2019.
Idaho Power provided detailed information on each of its thermal (coal, natural
gas combustion turbine, natural gas combined cycle and diesel) plants, as well as
its hydroelectric fleet. Unit outages, heat rates, fuel prices and other relevant
data were collected. Coal plants are modeled as must-run units with seasonal
outages for Idaho Power’s North Valmy Generating Station. Combined Cycle
plants (Langley Gulch) are committed in the hour-ahead timeframe and the gas
combustion turbine fleet has subhourly commitment intervals.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 12 |
Figure 4: Existing Idaho Power VER Units for which E3 was Provided 2019 DA,
HA, RT15 and RT5 Profiles

Given the large share of hydroelectricity on Idaho Power’s system, E3 focused on
ensuring proper representation of the hydro fleet’s capacity, ramping capability,
daily energy budgets, hourly maximum and minimum power ratings and other
such data. Additionally, E3 considered three hydro years, comprising
representative “low,” “average,” and “high,” hydro years. These profiles were
determined by Idaho Power by choosing from historical data. The average daily
energy profiles for these low, average and high hydro years are shown in Figure
5.
Planned future coal unit retirements through 2023 were modeled per Idaho
Power input. The overall planned change in fleet composition from 2019 to 2023,
as well as the total unit capacities by generation type are provided in Table 3.
PUBLIC

P a g e | 13 |
Data Collection, Processing and Methods
© 2010 Energy and Environmental Economics, Inc.
Idaho Power’s projected base case load and resource balance is shown below in
Figure 6.
Table 3: 2019 and 2023 Base Case Unit Capacities by Generator and Resource
Type

Figure 5: Daily High, Average and Low Hydro Energy Budget Profiles for Idaho
Power

3.1.2 EXTERNAL MARKET REPRESENTATION
Idaho Power was modeled as being able to trade with external electricity markets
at the Palo Verde and Mid C hubs. In the DA and HA stages of the model, Idaho
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 14 |
Power can make bilateral trades with its neighbors, while incurring a hurdle rate
to do so.
Figure 6: Base Case Load and Resource Balance in Idaho Power through 2030

E3 determined historical 2019 bilateral energy prices, hurdle rates, and transfer
limits through discussions with Idaho Power. In the RT15 and RT5 stages, Idaho
Power can trade with its neighbors in a manner consistent with Idaho Power’s
participation in the CAISO EIM, i.e. there are no hurdle rates, but there are
transfer limits. In the RT15 and RT5, Idaho Power trades electricity at the RTPD
(RT15) and RTD (RT5) 2019 EIM prices for the DGAP_IPCO_APND node, which is
an aggregated node that averages Idaho Power prices. E3 benchmarked the 2019
DGACP_IPCO_APND node prices versus 2019 nodal prices for the Elkhorn, High
Mesa and Rockland plants and found that the aggregated node provided a
satisfactory representation of these various wind plants.
In Q1 of 2019, there was a natural gas pipeline outage in the Alberta Electricity
System Operator (AESO) service territory, which inflated market prices in the
Pacific Northwest. Accordingly, E3 replaced the Q1 2019 market prices with Q1
PUBLIC

P a g e | 15 |
Data Collection, Processing and Methods
© 2010 Energy and Environmental Economics, Inc.
2020 market prices for the DA, HA, RT15 and RT5 phases. Additionally, given the
2023 timeframe of the model, E3 used its AURORA Market Price forecasts to
create a month-hourly average basis differential between 2023 and 2019. This
was added to the historical market prices in order to reflect the effect of
anticipated growth of VERs and storage across the Western Interconnection from
2019 through 2023, among other changes.
E3 benchmarked the historical interaction of the Elkhorn, High Mesa and
Rockland wind plants with the EIM. E3 found its representation of Idaho Power’s
interactions with the EIM to be reasonable.
Finally, E3 combined Idaho Power’s multiple hydroelectric projects into two units
for modeling simplicity. One unit consisted of aggregated run-of-river (RoR)
plants, whose output is largely inflexible and in flat hourly blocks, and the other
consisted of the combined Hells Canyon Complex (HCC) units (consisting of the
Oxbow, Brownlee and Hell’s Canyon dams), whose output can be varied within
certain time windows. This division of Idaho Power’s hydroelectric assets into two
aggregated units was done to reflect the variation in flexibility, water storage and
dispatchability across Idaho Power’s hydro fleet.
Planned future coal unit retirements through 2023 were modeled per Idaho
Power input. The overall planned change in fleet composition from 2019 to 2023,
as well as the total unit capacities by generation type are provided in Table 3.
Idaho Power’s projected base case load and resource balance is shown in Figure
6.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 16 |
3.2 RESERVE Modeling
3.2.1 DERIVATION OF 2023 VER PROFILES
As new VER resources are added, the average forecast error and subhourly
variability of the aggregated fleet will decline on a per MW of installed resources
basis. This is due to well-known diversity effects (i.e. as solar and wind plants are
installed at different locations, the average forecast error and subhourly variation
across all units will tend to decline on a per MW basis). Additionally, based on
experience in other jurisdictions, E3 assumed that there will be improvements to
VER forecast error in the future.
In order to capture these effects while using the 2019 VER data, E3 assessed the
reduction in forecast error and subhourly variability that Idaho Power has
observed to date. A similar approach was taken in Idaho Power’s 2018 Variable
Energy Resource Analysis. E3 performed this through the following steps
 Randomly order the forecast and actual profiles for existing wind and
solar that Idaho Power provided to E3
 Sequentially add solar profiles or wind profiles, each time calculating the
average forecast error and regulation reserves of the aggregated solar or
wind profiles using RESERVE
 Fit a polynomial trend to the average reserves versus the total MW of
online VERs for the solar and wind profiles
PUBLIC

P a g e | 17 |
Data Collection, Processing and Methods
© 2010 Energy and Environmental Economics, Inc.
 From prior work in the CAISO Extended Day Ahead Market project7, E3
assumed a 2 percent per annum improvement in VER forecasting
(average mean average percentage error reduction)
 For each future VER build, linearly scale up the 2019 VER forecast and
actual profiles by the ratios of future VER build total online MW to 2019
online MW
 Reduce the forecast error equally in all intervals using the polynomial
trend fit to forecast error data and using the estimated 2 percent per
annum improvement in forecast error from 2019 to 2023
 Reduce the subhourly interval-to-interval variation by the amount
derived from the polynomial trend fit to the regulation error data
 Run RESERVE for this new set of profiles; and
 Input these new set of profiles to PLEXOS

Using this process, the average standalone (i.e. not net-load-based) HA forecast
error reserves and regulation reserves for wind and solar would decline as shown
below in Table 4. These data show the reduction in average forecast error and
regulation needs across all hours of the year, relative to a case with no diversity
benefits or forecast error improvements and the same VER unit additions.
As can be seen in Table 4, E3 projects that regulation reserves will drop more on
a percentage basis than CAISO FRP reserves needs will in the high solar and high
wind cases. This is consistent with the larger percentage increase in solar build
than wind build in the high solar versus high wind cases, respectively.

7 https://stakeholdercenter.caiso.com/StakeholderInitiatives/Extended-day-ahead-market
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 18 |

Table 4: Average Projected Improvement in Forecast Error and Regulation
Reserves from Diversity and Forecasting Improvements
Case Average CAISO FRP
Reserve Improvement
Average Regulation
Reserve
Improvement
Base 2023 Case Solar (251 MW
new solar added to 2019 build)
11.7 % 14.2 %
Base 2023 Wind Case (0 MW new
wind added to 2019 build)
7.8 % 0.0 %
2023 Hi Solar Case (794 MW new
solar added to 2019 build)
17.2 % 31.6 %
2023 Hi Wind Case (669 MW new
wind added to 2019 build)
13.2 % 19.1 %
3.2.2 DERIVING RESERVES COMPONENTS
The CAISO FRP’s reserves for each interval consist of an uncertainty component,
plus a net load change from the previous interval, minus a credit component
based on the lesser of either the EIM-wide average footprint diversity or the
Balancing Authority’s (BA) trading position-derived credit. E3 used the
information provided by Idaho Power on forecast and actual load, wind and solar
to derive uncertainty requirements for the CAISO FRP. Given E3’s simplified
representation of Idaho Power’s external market transactions, E3 assumed that
the credit component of the reserve created a 40 percent reduction versus the
uncertainty component alone. This 40 percent value is an approximate value, and
was calculated using average historically-observed EIM footprint diversity in
PUBLIC

P a g e | 19 |
Data Collection, Processing and Methods
© 2010 Energy and Environmental Economics, Inc.
2019.8 This derivation, and its differences from the 2018 Idaho Variable Energy
Resource Integration Study is further discussed in Section 5.3.2.
3.3 Case Matrix
E3 and Idaho Power worked together to derive a total of eleven 2023 cases to
examine, in addition to a 2019 base case, which are described below. Table 5
details the specifics of each case.
 Case 1 is the 2023 base case for Cases 3-6 and Cases 8-11, which has
proposed unit additions and retirements and also includes the known
2019 through 2023 load growth
 Case 2 explores the effect of not retiring one of the Bridger coal plant’s
units, but is otherwise identical to Case 1
 Case 3 builds on Case 1 by exploring the effect of adding enough new
solar (794 MW of new solar) such that 10 percent of the 2023 Idaho
Power average gross load is provided by this new solar build
 Case 4 extends the Case 3 analysis to a low, rather than average hydro
year
 Case 5 builds on Case 1 and explores the integration costs of a high wind
build. Case 5 assumes a new wind build that can supply 10 percent of the
annual 2023 Idaho Power gross load (669 MW of new wind)
 Case 6 builds on Case 3 and Case 5, including both high solar and high
wind builds (794 MW of new solar and 669 MW of new wind)

8 https://www.westerneim.com/Pages/About/QuarterlyBenefits.aspx
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 20 |
 Case 7 is identical to Case 1, except that none of proposed solar additions
come online from 2019 to 2023, resulting in 251 MW fewer of solar than
Case 1 and lower reserves needs
 Cases 8 extends the Case 3 analysis to a high, rather than average hydro
year
 Case 9 extends the Case 3 analysis to have 200 MW of 4-hour, Federal
Investment Tax Credit (ITC)-enabled Li-Ion battery storage
 Case 10 extends the Case 3 analysis to have 400 MW of 4-hour, ITC-
enabled Li-Ion battery storage
 Case 11 extends the Case 3 analysis to allow economic curtailment of the
794 MW of new solar resource, while the 561 MW of existing and
proposed solar remain must-take resources
PUBLIC

P a g e | 21 |
Data Collection, Processing and Methods
© 2010 Energy and Environmental Economics, Inc.

Table 5: Case Matrix for 2023 Cases

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 22 |
4 Results
The following section provides detailed results from this work. A discussion of the
implications of these detailed results on VER integration in Idaho Power’s system
is provided in Section 5.
4.1 RESERVE Outputs
4.1.1 ANNUAL AVERAGE RESULTS
The average annual reserves for each of the cases is shown below in Table 6. It
should be noted that actual reserves vary on an hourly or subhourly basis in all
stages. However, E3 provided these average annual reserves as a general
indicator of how reserves needs change from case to case. These same data are
displayed below for the hour-ahead forecast’s CAISO FRP, regulation and
contingency reserves on a percentage of average monthly load basis for each
unique combination of solar and wind in Table 7, Table 8, Table 9, Table 10 and
Table 11. As observed in Table 6, wind reserves have more forecast error (CAISO
FRP reserves), whereas solar reserves have more subhourly variability. This trend,
observed here, is also expressed elsewhere in the literature.

PUBLIC

P a g e | 23 |
Results
© 2010 Energy and Environmental Economics, Inc.
Table 6: Average 2023 Case Reserves Needs
Case Total
MW
Wind
(MW)
Total
MW
Solar
(MW)
Avg.
RT15
FRP
Up
(MW)
Avg.
RT15
FRP
Down
(MW)
Avg.
Reg.
Up
(MW)
Avg.
Reg.
Down
(MW)
Avg.
Conting.
Res.
(MW)
Avg.
Total
Res. Up
(Percent
of Avg.
Load)
Avg. Total
Reserves
Down
(Percent of
Avg. Load)
1. 2023
Base
Case 728 561 100 97 40 41 104 13 % 7 %
2. Jim
Bridger
Online 728 561 100 97 40 41 104 13 % 7 %
3. Hi
Solar 728 1,354 147 142 71 72 104 17 % 11 %
4. Hi
Solar,
Low
Hydro
728 1,354 147 142 71 72 104 17 % 11 %
5. Hi
Wind 1,396 561 152 147 50 52 104 16 % 10 %
6. Hi
Solar, Hi
Wind 1,396 1,354 193 186 79 81 104 19 % 13 %
7.
Existing
Solar
Case
728 561 87 86 32 33 104 11% 6%
8. Hi
Solar, Hi
Hydro 728 1,354 147 142 71 72 104 17 % 11 %
9. Hi
Solar,
200 MW
Battery 728 1,354 147 142 71 72 104 17 % 11 %
10. Hi
Solar,
400 MW
Battery 728 1,354 147 142 71 72 104 17 % 11 %
11.
Curtail.
Solar 728 1,354 147 142 71 72 104 17 % 11 %
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 24 |
Table 7: 2023 Monthly Average, Load Normalized CAISO FRP, Regulation and
Contingency Reserves, Base 2023 Cases (Case 1 and Case 2)

Table 8: 2023 Monthly Average, Load Normalized CAISO FRP, Regulation and
Contingency Reserves, Existing Solar 2023 Case (Case 7)

Month
Hour Ahead
FRP + Reg. +
Contingency
Headroom,
Total
(% of Load)
Hour
Ahead FRP
+ Reg.
Headroom,
Solar
(% of Load)
Hour
Ahead FRP
+ Reg.
Headroom,
Wind
(% of Load)
Hour Ahead
FRP + Reg. +
Contin.
Headroom,
Load
(% of Load)
Hour
Ahead FRP
+ Reg.
Footroom,
Total
(% of Load)
Hour
Ahead FRP
+ Reg.
Footroom,
Solar
(% of Load)
Hour
Ahead FRP
+ Reg.
Footroom,
Wind
(% of Load)
Hour
Ahead FRP
+ Reg.
Footroom,
Load
(% of Load)
1 11.6% 0.5% 2.9% 8.2% 5.1% 0.5% 2.8% 1.7%
2 11.2% 0.5% 2.5% 8.3% 5.8% 0.6% 3.5% 1.6%
3 12.8% 1.5% 3.0% 8.2% 6.2% 1.5% 3.0% 1.7%
4 13.3% 1.6% 3.5% 8.2% 8.0% 1.8% 4.6% 1.6%
5 12.4% 1.6% 2.7% 8.2% 7.4% 2.0% 3.8% 1.6%
6 12.1% 1.4% 2.6% 8.1% 4.8% 1.0% 2.2% 1.6%
7 10.6% 1.0% 1.4% 8.2% 3.9% 0.8% 1.7% 1.4%
8 10.7% 1.0% 1.5% 8.2% 4.1% 0.8% 1.8% 1.5%
9 12.3% 1.1% 2.7% 8.5% 5.5% 1.0% 2.7% 1.8%
10 12.2% 1.2% 2.8% 8.3% 7.2% 1.3% 4.3% 1.6%
11 12.1% 1.2% 2.5% 8.4% 6.7% 1.1% 3.8% 1.8%
12 10.9% 0.5% 2.3% 8.1% 6.3% 0.6% 4.1% 1.6%
Avg. 11.86% 1.1% 2.5% 8.2% 5.9% 1.1% 3.2% 1.6%
PUBLIC

P a g e | 25 |
Results
© 2010 Energy and Environmental Economics, Inc.
Table 9: 2023 Monthly Average, Load Normalized Regulation Reserves, High
Solar Cases (Cases 3, 4, 8-11)

Table 10: 2023 Monthly Average, Load Normalized Regulation Reserves, High
Wind Case (Case 3)

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 26 |
Table 11: 2023 Monthly Average, Load Normalized Regulation Reserves, High
Solar and High Wind Case (Case 6)

4.1.2 DETAILED RESERVE RESULTS
While additions of new solar and wind both cause a similar increase in average
reserves needs, the hours in which they increase reserves are very different. The
following discussion illustrates these differences.
As observed in Table 6, wind reserves have more forecast error (CAISO FRP
reserves), whereas solar reserves have more subhourly variability. This trend,
observed here, is also expressed elsewhere in the literature.9
Conversely, the incremental FRP needs from adding solar shown in Figure 11
indicate that CAISO FRP reserves increase primarily during solar hours. FRP
reserves do increase at night because caps on the level of uncertainty imposed

9 https://www.nrel.gov/docs/fy13osti/55588.pdf
PUBLIC

P a g e | 27 |
Results
© 2010 Energy and Environmental Economics, Inc.
by the CAISO FRP derivation10 (see further discussion in Section 5.3.2) also
increase. Similarly, solar regulation needs increase only during solar hours.
Because reserves can only be provided with dispatchable resources in the PLEXOS
model, it is important to compare the need for reserves with the availability of
dispatchable resources. Figure 13 and Figure 14 show month-hourly average
residual net load, calculated as load minus wind, solar, and RoR hydro for the High
Solar and High Wind cases. This residual net load is the average load that must be
met by dispatchable resources and imports. If the need for reserves is greater
than the residual net load, then the model must export power to the market to
be able to serve Idaho Power’s reserves needs while not violating minimum
generation setpoints for online units. As discussed below, this can result in
exports to the market at a loss.
As can be seen from Figure 13, in the High Solar case, in March, April, May and
October, the residual net load is very low during the midday hours in which there
is high demand on reserves. Alternatively, as can be seen in the high wind case
for Figure 10, the residual net load is significantly higher during those midday
hours, and as shown earlier, average reserves needs are not especially high
midday.

10 See https://bpmcm.caiso.com/Pages/BPMDetails.aspx?BPM=Market percent20Operations for a discussion of
these caps; E3 derives its own caps from P98 and P2 values of the seasonal forecast error.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 28 |
Figure 7: Average Month-Hourly CAISO FRR Headroom Needs for Base 2023
Case

Figure 8: Average Month-Hourly Regulation Reserves Headroom Needs for 2023
Base Case

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 54 72 54 103 82 83 99 84 73 95 98 95 103 103 103 103 103 103 100 103 54 69 75 93 88
2 71 52 35 63 75 57 43 50 67 95 103 103 103 103 103 103 103 103 80 84 103 103 103 103 84
3 68 81 80 92 65 80 84 80 102 110 127 127 71 124 127 127 127 127 127 73 64 78 75 71 95
4 56 47 58 74 80 97 83 67 84 127 127 127 105 105 127 127 127 127 127 113 67 90 93 67 96
5 67 90 84 63 68 71 67 86 112 127 127 127 89 127 127 127 127 127 127 121 104 71 72 78 99
6 71 78 130 151 151 151 151 151 151 151 151 151 151 151 151 128 101 151 151 151 116 99 101 80 132
7 57 63 53 50 41 44 59 67 151 142 151 128 115 151 144 122 147 151 151 151 120 103 87 53 104
8 25 50 63 59 61 57 54 53 113 151 142 151 151 117 134 144 151 151 151 151 98 85 103 61 103
9 71 72 76 66 80 72 92 108 129 129 129 129 129 129 129 129 129 129 129 122 122 83 69 68 105
10 76 69 73 56 53 63 53 54 61 113 129 129 129 129 125 129 129 129 74 50 89 74 54 78 88
11 56 54 65 58 57 75 80 78 93 122 118 129 118 109 129 129 129 95 110 87 70 59 59 56 89
12 71 55 65 63 66 80 70 67 51 78 91 103 103 103 94 71 71 66 47 50 61 68 87 87 74
62 65 70 75 73 77 78 79 99 120 124 125 114 121 124 120 120 122 115 105 89 82 81 74
Average Modeled CAISO FRR Headroom (MW)
Hour of Day
Month
Hour Average
Month
Average
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 24 24 24 24 24 25 25 27 27 32 39 40 41 48 47 53 51 41 26 26 25 24 24 24 32
2 25 25 25 25 25 25 26 28 27 33 39 41 41 48 46 55 51 42 27 26 26 25 25 25 33
3 21 21 21 22 23 23 23 33 49 53 50 74 74 85 99 109 99 90 67 34 22 22 21 21 48
4 21 21 20 21 22 23 23 34 48 49 46 58 65 71 82 85 91 83 66 35 24 24 22 22 44
5 20 21 21 21 21 22 22 38 43 42 42 53 58 62 67 77 86 74 63 34 24 23 22 21 41
6 25 24 23 23 23 23 28 45 60 60 40 37 41 44 44 62 68 71 59 54 35 34 29 27 41
7 29 27 25 24 24 25 29 57 74 73 49 42 45 45 44 68 76 81 59 52 35 37 38 34 46
8 26 24 24 23 23 24 27 56 72 61 48 44 46 47 49 61 66 77 83 53 35 36 33 29 45
9 23 23 23 22 23 23 24 24 35 54 51 47 49 53 61 78 77 57 36 26 27 26 25 24 38
10 22 21 21 21 23 23 24 24 35 58 51 56 51 61 69 84 83 57 36 23 23 24 23 22 39
11 23 22 22 22 22 22 23 23 34 57 54 57 62 68 78 87 82 57 36 23 22 22 23 23 40
12 23 23 23 23 23 23 24 24 25 30 37 40 40 50 50 57 53 43 25 24 24 24 23 23 31
23 23 23 23 23 23 25 34 44 50 46 49 51 57 62 73 74 64 49 34 27 27 26 24
Average Regulation Headroom - RMS Combined Load + Wind + Solar (MW)
Hour Average
Month
Hour of Day Month
Average
PUBLIC

P a g e | 29 |
Results
© 2010 Energy and Environmental Economics, Inc.

Figure 9: High Wind Minus Base Case CAISO FRR Headroom

Figure 10: High Wind Minus Base Case Regulation Headroom

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 16 17 17 16 16 17 17 16 15 14 11 11 11 9 10 9 9 11 15 15 15 16 16 16 14
2 17 17 17 17 17 17 17 16 16 14 13 12 12 11 11 10 10 12 16 16 16 16 16 17 15
3 13 13 13 13 13 13 13 10 8 7 7 5 5 4 4 3 4 4 5 9 12 12 12 13 9
4 16 17 17 16 15 15 14 11 8 8 8 7 7 7 6 6 6 6 7 11 15 15 16 17 11
5 14 13 13 14 13 13 11 8 7 7 7 6 6 6 6 6 5 6 7 11 14 14 14 13 10
6 15 14 14 14 14 13 11 8 6 7 8 8 8 8 8 8 7 7 8 9 13 13 14 14 10
7 14 14 15 14 14 13 12 7 6 6 6 7 7 7 8 6 6 7 8 9 13 13 12 13 10
8 13 13 13 13 13 12 11 6 5 6 7 6 6 6 6 6 6 5 5 8 12 11 12 13 9
9 15 15 15 14 14 14 13 13 10 7 7 8 7 8 7 6 5 7 10 13 14 14 14 15 11
10 15 15 15 14 14 14 15 14 11 8 8 8 9 7 6 5 5 8 12 15 15 15 15 15 12
11 13 12 13 13 13 12 13 12 9 6 6 6 5 5 4 4 4 6 10 13 13 13 13 13 10
12 15 14 14 14 14 14 13 13 12 11 9 8 8 7 7 6 7 8 12 13 13 14 14 15 11
15 15 15 14 14 14 13 11 9 8 8 8 7 7 7 6 6 7 10 12 14 14 14 14 11
Month
Hour Average
Difference, Hi Wind to Base Case, Average Regulation Headroom (MW)
Hour of Day Month
Average
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 30 |
Figure 11: High Solar Minus Base Case CAISO FRR Headroom

Figure 12: High Solar Minus Base Case Regulation Headroom

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 0 0 0 3 0 0 0 0 -1 19 47 61 60 60 60 60 60 58 2 6 0 0 0 0 21
2 0 0 0 0 0 0 0 1 7 40 60 60 60 60 60 60 60 60 7 0 42 21 60 44 29
3 0 0 0 0 0 0 0 12 28 75 114 114 63 103 114 114 114 114 104 23 0 0 0 0 45
4 0 0 0 0 0 1 10 36 52 114 114 114 124 116 114 114 114 114 114 56 4 0 0 0 55
5 0 0 0 0 0 2 24 41 129 114 114 114 119 114 114 114 114 114 114 88 12 0 0 0 60
6 0 0 0 9 45 99 89 116 124 124 124 124 124 124 124 147 67 124 124 124 43 1 0 0 77
7 0 0 0 0 0 11 31 72 124 133 124 113 106 124 116 136 126 124 124 124 27 1 0 0 67
8 0 0 0 0 0 2 9 70 135 124 132 124 101 69 72 104 124 124 124 104 4 0 0 0 59
9 0 0 0 0 0 0 3 20 85 111 111 111 111 111 111 111 111 111 111 23 0 0 0 0 52
10 0 0 0 0 0 0 4 19 40 127 111 111 111 109 95 111 111 111 30 0 0 0 0 0 45
11 0 0 0 0 0 0 0 7 15 28 58 82 81 71 111 111 111 37 0 0 0 0 0 0 30
12 0 0 0 0 0 0 0 0 3 33 62 60 60 60 68 49 35 6 0 0 0 0 0 0 18
0 0 0 1 4 10 14 33 62 87 98 99 93 93 97 103 96 91 71 46 11 2 5 4 47Hour Average
Difference, Hi Solar to Base Case, Average CAISO FRR Headroom (MW)
Hour of Day Month
Average
Month
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 0 0 0 0 0 0 0 0 0 19 37 39 42 55 53 65 60 41 0 0 0 0 0 0 17
2 0 0 0 0 0 0 0 0 0 20 35 40 41 53 49 60 59 42 0 0 0 0 0 0 17
3 0 0 0 0 0 0 0 27 58 64 60 97 97 106 131 144 134 122 87 33 0 0 0 0 48
4 0 0 0 0 0 0 0 30 56 57 54 71 82 91 108 112 121 106 85 32 0 0 0 0 42
5 0 0 0 0 0 0 1 38 48 46 46 63 72 77 85 99 107 91 78 28 0 0 0 0 37
6 0 0 0 0 0 0 12 45 71 63 26 16 27 32 33 55 66 73 70 51 3 0 0 0 27
7 0 0 0 0 0 0 5 62 92 74 30 21 32 36 37 77 93 100 85 52 3 0 0 0 33
8 0 0 0 0 0 0 5 65 87 66 34 19 34 38 44 60 77 88 105 52 3 0 0 0 32
9 0 0 0 0 0 0 0 1 31 59 56 46 56 59 70 99 99 67 30 0 0 0 0 0 28
10 0 0 0 0 0 0 0 0 30 68 60 65 61 73 86 117 110 71 35 0 0 0 0 0 32
11 0 0 0 0 0 0 0 0 30 72 65 71 80 89 105 118 111 71 35 0 0 0 0 0 35
12 0 0 0 0 0 0 0 0 0 19 36 42 42 59 60 71 64 45 0 0 0 0 0 0 18
0 0 0 0 0 0 2 22 42 52 45 49 55 64 72 90 92 76 51 21 1 0 0 0 31
Month
Hour Average
Difference, Hi Solar to Base Case, Average Regulation Headroom (MW)
Hour of Day Month
Average
PUBLIC

P a g e | 31 |
Results
© 2010 Energy and Environmental Economics, Inc.
Figure 13: Residual Net Load, High Solar Case 3

Figure 14: Residual Net Load, High Wind Case 5

4.2 2019 PLEXOS to Historical Case Benchmarking
E3 and Idaho Power performed rigorous benchmarking to ensure that the PLEXOS
model was able to reasonably replicate actual 2019 historical behavior. E3 and
Idaho Power verified that the following were in line with historical 2019 behavior:
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 32 |
 Hydro and thermal unit flexibility (ramping rate) and dispatch
(distribution of ramps);
 Total generation by unit and technology class;
 Market transaction behavior and external market prices;
 Average Idaho Power nodal energy prices;
 Unit capacities;
 Unit outages;
 Number of unit starts; and
 Average unit marginal operational cost
Particular attention was paid to the HCC to ensure its operation was reasonable.
This was critical because of the large amount of Idaho Power’s energy from
hydroelectricity in a typical year, as well as the crucial role that this unit has in
providing flexibility. Figure 15 below shows a sample of the verification of the
model wherein actual dispatch of the PLEXOS HCC is shown to be within the daily
maximum and minimum power output ranges, and the dispatch of the HCC
adheres to the input daily hydro budget.
Additionally, after initial results were analyzed, the Idaho Power team thought
that EIM transactions were unrealistically high in the PLEXOS model, given that
the model operates a price taker for market transactions. In reality, if Idaho
Power made particularly large sales or purchases in the EIM, prices would be
affected. Therefore, E3 and Idaho Power worked together to limit total net sales
and purchases in the EIM to +/- 300 MW in price taker mode. In instances in which
the model traded between +/- 300 MW up to the line limits in the real time, the
model paid a hurdle rate of $150/MW, which was implemented to approximate
PUBLIC

P a g e | 33 |
Results
© 2010 Energy and Environmental Economics, Inc.
“price setting” behavior. Overall, there were few hours in which the model
accessed this additional EIM flexibility.
Figure 15: PLEXOS HCC Dispatch vs. Historical Power and Hydro Budget Bounds

4.3 2023 Case Result Summary
The Incremental specific integration costs for each of the cases is provided below
in Table 12. These results are discussed in greater detail below in Chapter 5.

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 34 |

Table 12: Summary of Incremental VER Integration Costs
Case Inc. Start
Costs
($Million/ yr)
Inc.
Ramping
Costs
($ Million/ yr)
Total
Inc.
Imperf. Unit Commit. &
Dispatch
Costs
($Mill./yr)
Total
Curtail.
Costs
($Million/yr)
Total
Inc.
Integrat. Costs
($Million/ yr)
Total
Product.
Cost ($Million/ yr)
Total
Inc.
VER Gen.
(GWh /yr)
Total
Inc.
Specific Integrat. Costs
($/MWh)
1. 2023
Base Case -$0.15 $0.22 $1.62 $0.00 1.69 $181 577 $2.93
2. Jim Bridger
Online
-$0.17 $0.37 $1.88 $0.00 $2.08 $180 577 $3.61
3. Hi Solar $0.80 $0.45 $5.78 $0.00 $7.04 $146 1,824 $3.86
4. Hi Solar,
Low
Hydro
$0.60 $0.53 $7.16 $0.00 $8.29 $172 1,824 $4.55
5. Hi
Wind $0.35 -$0.07 $1.12 $0.00 $1.41 $143 1,823 $0.77
6. Hi
Solar +
Hi Wind
$1.63 $0.33 $7.01 $0.00 $8.96 $109 3,647 $2.46
7.
Existing
Solar
Base
n/a n/a n/a n/a n/a $193 0 n/a
8. Hi
Solar, Hi
Hydro
$2.41 $0.19 $5.87 $0.00 $8.47 $75 1,823 $4.65
9. Hi
Solar,
200 MW
Battery
$0.58 $0.02 $0.56 $0.00 $1.16 $144 1,823 $0.64
10. Hi Solar,
400 MW
Battery
$0.58 -$0.34 $1.46 $0.00 $1.69 $142 1,823 $0.93
11. Hi
Curtail.
Solar
$0.72 $0.39 $4.31 $0.29 $5.71 $147 1,823 $3.13
PUBLIC

P a g e | 35 |
Results
© 2010 Energy and Environmental Economics, Inc.
4.4 System Dispatch Results
In the following subsections, detailed day plots and other modeling results will be
used to illustrate how the Idaho Power system responds to adding different
capacities and kinds of VERs, and increasing or decreasing system flexibility. To
facilitate this, this study will examine the following case groupings:
 Existing Solar (Case 7), Base Case (Case 1) and Jim Bridger First Unit Online
(Case 2)
 High Solar (Case 3), High Wind (Case 5) and High Solar + Wind (Case 6)
 High Solar with Low (Case 4), Average (Case 3) and High (Case 8) Hydro
Budgets
 High Solar with (Cases 9 and 10) and without (Case 3) battery storage
 Hi Solar with (Case 11) and without the ability to economically curtail
solar (Case 3)
4.4.1 EXISTING SOLAR, 2023 BASE CASE AND JIM BRIDGER FIRST UNIT
ONLINE CASES
This case comparison illustrates the effect of adding successively more VERs, as
well as increasing the aggregate system thermal minimum power level (Pmin).
The salient differences between cases are outlined as follows
 Total online solar
o Existing Solar (Case 7): 310 MW
o 2023 Base Case (Case 1): 561 MW
o Jim Bridger Online Case (Case 2): 561 MW
 Jim Bridger Coal Plant Pmin/Pmax
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 36 |
o Existing Solar (Case 7): 89 MW / 533 MW
o 2023 Base Case (Case 1): 89 MW / 533 MW
o Jim Bridger Online Case (Case 2): 118 MW / 707 MW
In the modeled year of 2023, there will be periods during the daytime in the
spring and fall in which external electricity prices are low or negatively priced.
This is due to the growing penetration of solar across the WECC footprint and the
low net loads during these periods. Figure 16 illustrates the Idaho Power system
operation operating during a day (April 23, 2023) that exhibits these conditions.
Beginning with the “Existing Solar Case,” which models the Idaho Power system
with the 2019 levels of wind and solar, the model will choose to purchase power
from the market rather than generate its own power during these periods. This is
shown by the purchase of electricity 4 am through 8 pm MST in Figure 16.
In the 2023 base case, 561 MW of solar is assumed to be online, which increases
Idaho Power’s total VER Pmin during midday periods. This decreases Idaho
Power’s ability to purchase negatively priced electricity from the market. This is
shown in Figure 16, wherein purchases are now only made in the morning and
evening periods.
Per discussions with Idaho Power, the Jim Bridger coal plant is modeled as a must-
run unit. As such, in the first Jim Bridger unit online case, the aggregate thermal
Pmin increases during all hours by 29 MW. Having both more solar and Jim
Bridger’s first unit online further increases Idaho Power’s aggregate Pmin. In
Figure 16, this results the model no longer purchasing negatively priced electricity
in the afternoon.
PUBLIC

P a g e | 37 |
Results
© 2010 Energy and Environmental Economics, Inc.
Though not depicted here, during periods of high net load (e.g. during summer
peaking events), the addition of extra solar and the ability to dispatch more power
from Jim Bridger can prove beneficial in reducing system costs by displacing
expensive market purchases and/or natural gas combustion turbine (CT) and/or
combined cycle (CCGT) generation. Per Table 13, as more solar is added, and if a
Jim Bridger unit is not retired, total incremental specific VER integration costs rise
but total production costs fall.

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 38 |
Figure 16: Existing Solar vs. 2023 Base Case vs. First Bridger Unit Online Daily
Dispatch Plots

PUBLIC

P a g e | 39 |
Results
© 2010 Energy and Environmental Economics, Inc.
Table 13: Summary of Results for Existing Solar, Base Case Solar and Jim Bridger
Cases
Case Inc. Start
Costs
($Million/ yr)
Inc.
Ramping
Costs
($ Million/ yr)
Total
Inc.
Imperf. Unit Commit. &
Dispatch
Costs
($Million
/ yr)
Total
Curtail.
Costs
($Million/yr)
Total
Inc.
Integrat. Costs
($Million/ yr)
Total
Product.
Cost ($Million/ yr)
Total
Inc.
VER Gen.
(GWh /yr)
Total
Inc.
Specific Integrat. Costs
($/MWh)
1. 2023
Base
Case
-$0.15 $0.22 $1.62 $0.00 1.69 $181 577 $2.93
2. Jim
Bridger
Online
-$0.17 $0.37 $1.88 $0.00 $2.08 $180 577 $3.61
7.
Existing
Solar
Base
n/a n/a n/a n/a n/a $193 0 n/a
4.4.2 HIGH SOLAR, HIGH WIND, AND HIGH SOLAR + WIND CASES
This set of cases illustrates the difference in the ease of integrating equivalent
amounts of new VER energy from solar and wind. Additionally, the effects of
combining these solar and wind additions is shown.
The salient differences in VER capacities between these cases are as follows:
 Total Online Solar
o High Solar Case (Case 3): 1,355 MW
o High Wind Case (Case 5): 561 MW
o High Solar + High Wind Case (Case 6): 1,355 MW
 Total Online Wind
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 40 |
o High Solar Case (Case 3): 728 MW
o High Wind Case (Case 5): 1,397 MW
o High Solar + High Wind Case (Case 6): 1,397 MW
This case builds on the phenomena observed in Section 4.4.1, wherein adding
more VERs reduces the model’s ability to optimally perform market transactions
during low net load, springtime conditions. Figure 17 below depicts the high wind,
high solar, and high solar + high wind cases on the same low net load spring day
(April 27, 2023).
Starting with the high wind case, one observes that during periods of low net load,
the system is fairly balanced in terms of imports and exports, only exporting to
the low to negatively priced EIM market in the afternoon when wind generation
begins to climb. Additionally, the system is able to provide the required reserves
for carrying wind with only the coal and HCC units. This is due to the relatively
low level of reserves required to integrate wind, as shown in Figure 9 and Figure
10.
In the high solar case, the increased midday reserves needs shown in Figure 11
and Figure 12 coincide with high solar production. The increase in reserves needs
causes the model to start a CCGT unit, as the reserve can no longer just be
provided with hydro and coal. Bringing the CCGT unit online when there is high
solar production causes the model to make significant exports to the EIM market
during low and negatively priced hours. This, along with the start costs of the
CCGT, increases the costs of integrating solar relative to the costs of integrating
wind.
Finally, adding both high solar and high wind further exacerbates the issues that
arise during the high solar case. Due to the increase in production of wind during
the afternoon, the model must make further exports to a low and negatively
PUBLIC

P a g e | 41 |
Results
© 2010 Energy and Environmental Economics, Inc.
priced market. Additionally, the model turns on a CT instead of a CCGT to provide
the additional reserves required due to wind and solar.
Figure 17 presents daily operations from the imperfect foresight cases. However,
as described in Section 2.1, the difference in total market transactions and
generator costs for each case are calculated using the difference between each
case’s perfect and imperfect foresight cases. Though not shown here, on the day
shown in Figure 17, the model chooses to not start CCGTs or CTs in the respective
high solar and high wind + high solar cases in the perfect foresight cases. This is
due to the lower reserve need of the perfect foresight case.
Table 14: Summary of Results for High Solar, High Wind and High Solar + High
Wind Cases
Case Inc. Start
Costs
($Million/
yr)
Inc.
Ramping
Costs
($ Million/ yr)
Total
Inc.
Imperf.
Unit Commit. & Dispatch
Costs
($Million/
yr)
Total
Curtail.
Costs
($Million/yr)
Total
Inc.
Integrat.
Costs
($Million/ yr)
Total
Product.
Cost
($Million/ yr)
Total
Inc.
VER
Gen.
(GWh /yr)
Total
Inc.
Specific
Integrat. Costs
($/MWh)
3. Hi
Solar $0.80 $0.45 $5.78 $0.00 $7.04 $146 1,824 $3.86
5. Hi
Wind $0.35 -$0.07 $1.12 $0.00 $1.41 $143 1,823 $0.77
6. Hi
Solar +
Hi Wind
$1.63 $0.33 $7.01 $0.00 $8.96 $109 3,647 $2.46
As shown in Table 14, total incremental VER integration costs are highest in the
high solar + high wind case, followed by the high solar case and the high wind
case. However, the total specific incremental VER integration cost is lower for the
high wind + high solar than the high solar case because, while the total integration
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 42 |
cost rises with more VERs, there is also more total incremental VER generation in
the high wind + high solar case versus the high solar case.
Figure 17: High Wind vs. High Solar vs. High Solar + Hi Wind

4.4.3 HIGH SOLAR WITH LOW, AVERAGE AND HIGH HYDRO BUDGETS
This set of cases compares the effects of varying hydro budgets under high solar
conditions. On a typical year, Idaho Power derives the majority of their power
PUBLIC

P a g e | 43 |
Results
© 2010 Energy and Environmental Economics, Inc.
from their hydro fleet, but the total annual energy derived from hydro varies
considerably year-to-year. The simulated conditions considered in this set of
cases is depicted below in Figure 18.
Figure 18: Hydro Conditions in Low, Average and High Hydro Cases

In the model, RoR hydro is treated as an inflexible, must take resource, whereas
HCC is dispatchable. The high hydro budget case capacity factor shown in Figure
18 indicates that both HCC and RoR hydro must operate near their Pmax
throughout the year in order to not violate daily hydro energy budgets, which
greatly reduces hydro system flexibility. As shown in Figure 15, hydro conditions
are generally highest in the spring due to runoff from snow melt. Figure 19 below
compares a spring day (April 20, 2023) in which the combination of low electricity
market prices, hydro availability and VERs interact with one another.
Starting with the high hydro case, the model must sell HCC and RoR output to the
market all day, due to the high hydro budget. This includes sales during periods
of negative external market prices. Additionally, the model must start a CT to
provide solar reserves during midday. Conversely, during average hydro
conditions, this need to sell to the market at a loss is reduced, and the model
shifts HCC production to avoid selling hydro at a loss during the morning. The
model switches from using a CT to a CCGT to provide solar reserves. Finally, during
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 44 |
low hydro conditions, Idaho Power’s system can buy from the market during
negatively priced hours, but the model must run the CCGT more due to lower
hydro budgets.
Table 15: Summary of Results for High Solar with Low, Average and High Hydro
Budgets Cases
Case Inc. Start
Costs
($Million/
yr)
Inc.
Ramp
Costs
($
Million/
yr)
Total
Inc.
Imperf.
Unit
Commit.
&
Dispatch
Costs
($Million
/ yr)
Total
Curtail.
Costs
($Million
/yr)
Total
Inc.
Integrat.
Costs
($Million
/ yr)
Total
Product.
Cost
($Million
/ yr)
Total
Inc.
VER
Gen.
(GWh
/yr)
Total
Inc.
Specific
Integrat.
Costs
($/MWh)
3. Hi
Solar $0.80 $0.45 $5.78 $0.00 $7.04 $146 1,824 $3.86
4. Hi
Solar,
Low
Hydro
$0.60 $0.53 $7.16 $0.00 $8.29 $172 1,824 $4.55
8. Hi
Solar,
Hi
Hydro
$2.41 $0.19 $5.87 $0.00 $8.47 $75 1,823 $4.65

As shown in Table 15, total incremental specific VER integration costs are higher
in both the low and high hydro year cases. Moving from low to high hydro
conditions, market purchases and thermal generation decreases. This causes
production costs to decrease.
PUBLIC

P a g e | 45 |
Results
© 2010 Energy and Environmental Economics, Inc.
Figure 19: Low, Average and High Hydro Case Comparison

4.4.4 HIGH SOLAR WITH AND WITHOUT STORAGE

This set of cases compares the cost of integrating solar with and without battery
storage. Because Idaho Power is a vertically integrated utility, there is no ancillary
services market for these PURPA facilities. Therefore, batteries do not provide
reserves to the Idaho Power system in these cases. Additionally, the model treats
solar + storage systems having ITC-eligible battery storage. Per ITC regulations,
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 46 |
this requires storage to charge solely using solar power production. At the time
of this study’s completion, compensation rate methodologies had not been
finalized for PURPA solar + battery storage facilities pursuing contracts with Idaho
Power. Thus, the model used a simplified approach of allowing the battery to only
discharge between 4 pm and 10 pm daily. However, the model allowed the
battery dispatch to minimize total Idaho Power production costs when during the
permitted charging and discharging periods. Finally, as shown in Table 6, the
reserves needs are modeled as identical in each of these cases.
In all of these cases, the model uses a high solar build (1,355 MW of total solar),
but only the 794 MW of the solar (i.e. the incremental solar built vs. the 2023
Base Case) is coupled with an ITC-eligible battery. The differences in these cases
are as follows:
 Total Battery Capacity
o High Solar Case: 0 MW
o High Solar + 200 MW Battery Case: 200 MW, 4-hour (800 MWh)
Li-Ion Battery
o High Solar + 400 MW Battery Case: 400 MW, 4-hour (1,600 MWh)
Li-Ion Battery
As can be seen in Figure 20 and Figure 21, on a typical medium-load spring day
(5/10/2023), the battery is used to move solar energy from morning and evening
solar production hours to increase net sales to the market and reduce Idaho
Power coal generation.
PUBLIC

P a g e | 47 |
Results
© 2010 Energy and Environmental Economics, Inc.
Figure 20: High Solar vs. High Solar + 200 MW Battery, Medium Load Spring Day

Figure 21: High Solar vs. High Solar + 400 MW Battery, Medium Load Spring Day

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 48 |
The average month-hourly dispatch of charging and discharging for the ITC-
eligible storage is depicted in Figure 22. As can be seen in each of these figures,
having greater battery capacity does not fundamentally alter when charging and
discharging occur on a given day, or across the year.
Figure 22: Month-Hourly Average Battery Charge and Discharge Power for 200
MW and 400 MW ITC-Eligible Batteries

Table 16 shows the summary of results for these cases. The total production costs
are lowest for the 400 MW battery, increasing in the 200 MW battery case and
further increasing in the no battery cases. However, the total specific integration
costs are lowest for the 200 MW battery size. Both storage cases exhibit
dramatically lower VER integration costs than the high solar without storage case.
This is discussed in greater detail in Section 5 of this report.
PUBLIC

P a g e | 49 |
Results
© 2010 Energy and Environmental Economics, Inc.
Table 16: Summary of Results for High Solar with and without Storage
Case Inc. Start
Costs
($Million/
yr)
Inc.
Ramping
Costs
($
Million/
yr)
Total
Inc.
Imperf.
Unit
Commit.
&
Dispatch
Costs
($Million/
yr)
Total
Curtail.
Costs
($Million/
yr)
Total
Inc.
Integrat.
Costs
($Million/
yr)
Total
Product.
Cost
($Million/
yr)
Total
Inc.
VER
Gen.
(GWh
/yr)
Total
Inc.
Specific
Integrat.
Costs
($/MWh)
3. Hi
Solar $0.80 $0.45 $5.78 $0.00 $7.04 $146 1,824 $3.86
9. Hi
Solar,
200
MW
Battery
$0.58 $0.02 $0.56 $0.00 $1.16 $144 1,823 $0.64
10. Hi
Solar,
400
MW
Battery
$0.58 -$0.34 $1.46 $0.00 $1.69 $142 1,823 $0.93
4.4.5 HIGH MUST TAKE SOLAR AND CURTAILABLE SOLAR CASES
Idaho Power is not able to perform economic solar curtailment of PURPA
facilities. The high must take solar and high curtailable solar cases were therefore
implemented to show how being able to economically curtail PURPA solar would
change the cost of integrating VERs.
In the high solar case, the model can only perform reliability-based curtailment,
i.e. the model will curtail VERs only when the alternative is to have unserved
energy or face some other infeasibility. In the curtailable case, the model may
economically curtail power for the incremental 794 MW of solar installed vs. the
2023 base case. This allows the model to curtail power to reduce Idaho Power’s
total production costs. There would be no difference in short-run marginal energy
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 50 |
costs from economically curtailing PURPA solar, however Idaho Power may have
to pay for the lost renewable energy credit (REC) due to curtailing solar.
Therefore, the model assumes a $20/MWh curtailment penalty, which is a typical
REC price in WECC. Similarly to the solar with storage cases, the VER reserves
needs are modeled as identical between the must take and curtailable cases.
Figure 23 and Figure 24 respectively show the difference between the must take
and curtailable cases on a low net load spring day (4/21/2023) and a high net load
summer day (7/21/2023). In Figure 23, the model chooses to curtail power both
when the external market price is below the curtailment penalty (i.e. below
-$20/MWh), as well as during the middle of the day. The model chooses to curtail
power midday because, while the market price is not below -$20/MWh, the
model performs reliability curtailment of solar in the must take case as well. In
other words, this low net load day requires VER curtailment of some sort. Total
annual curtailment in the curtailable solar case is 3.8% of potential generation for
the 794 MW of new solar. This curtailment is largely confined to spring hours,
when the net load is very low.
Alternatively, Figure 24 shows that the model does not curtail solar when solar
helps reduce total production costs. This is because solar increases net sales to a
high-priced market.
PUBLIC

P a g e | 51 |
Results
© 2010 Energy and Environmental Economics, Inc.
Figure 23: High Must Take Solar and High Curtailable Solar, Low Load Day

Figure 24: High Must Take Solar vs. High Curtailable Solar, High Load Day

PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 52 |
Table 17 shows that while the total incremental specific integration cost is lower
in the curtailable solar case than the must take solar case, the total production
costs are essentially identical between the two cases.
Table 17: Summary of Results for High Must Take and Curtailable Solar
Case Inc. Start Costs
($Million/ yr)
Inc. Ramping Costs
($ Million/ yr)
Total Inc. Imperf. Unit Commit. & Dispatch Costs
($Million/
yr)
Total Curtail. Costs
($Million/yr)
Total Inc. Integrat. Costs
($Million/ yr)
Total Product. Cost ($Million/ yr)
Total Inc. VER Gen.
(GWh /yr)
Total Inc. Specific Integrat. Costs
($/MWh)
3. Hi
Solar $0.80 $0.45 $5.78 $0.00 $7.04 $146 1,824 $3.86
11. Hi
Curtail.
Solar
$0.72 $0.39 $4.31 $0.29 $5.71 $147 1,823 $3.13

PUBLIC

P a g e | 53 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
5 Discussion
5.1 Discussion of Current Study Results
E3’s results provide several high-level insights about integrating VERs:
 Integration costs are driven by the need for procuring system flexibility
on dispatchable generators during periods of low net load
 Integrating solar is more expensive that integrating new wind resources
 VER integration costs can be lowered by adding flexibility to the Idaho
Power system, such as battery storage, allowing economic curtailment
and reducing the must-run thermal Pmin of the system
 VER integration costs increase during abnormal hydro conditions (low or
high annual budgets)
 The integration costs found in this 2020 Idaho Power VER integration
study are lower than the 2018 Idaho Power Variable Energy Resource
Analysis
These results are discussed in more detail below.
5.1.1 EFFECTS OF BINDING PMIN CONSTRAINTS ON VER INTEGRATION
COSTS
As discussed in Section 3.2, as more VERs are added to Idaho Power’s system, the
aggregate reserve and flexibility needs tend to increase. Only HCC, coal, CTs and
CCGTs are modeled as eligible to provide reserves. Because all these generators
have a non-zero Pmin, the aggregate thermal Pmin grows when more generators
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 54 |
are brought online to provide reserves. Idaho Power has a large penetration of
PURPA VERs, which are treated as must take units by Idaho Power. When these
must take resources produce large amounts of power, the net load on Idaho
Power’s system can fall to very low values. In order to maintain supply-demand
equilibria on Idaho Power’s system, Idaho Power must export power to the
market when the aggregate system Pmin, plus the required system footroom, is
greater than the system net load. This is depicted schematically below in Figure
25.
Figure 25: Effects of Additional Solar on Unit Commitment and Market
Transactions

During these “binding Pmin” events, exporting power to the market does not by
itself cause VER integration costs to rise. However, due to the growing
penetration of solar across the EIM footprint, 2023 EIM market prices are
projected to be, on average, below typical marginal thermal unit generation costs
during daytime hours in the spring and fall, as shown in Figure 26. These periods
of low EIM prices are also when Idaho Power’s solar generators will be producing
enough power to significantly lower Idaho Power’s net load to binding Pmin
PUBLIC

P a g e | 55 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
levels. Therefore, under high solar builds, Idaho Power is often exporting power
at a financial loss to a low- or negative-priced EIM market. At other times, Idaho
Power may have to shift its hydro production to non-optimal hours (e.g. away
from times when hydro could earn the greatest amount of export revenues) in
order to provide sufficient flexibility on HCC while adhering to the HCC daily
energy budget.
Figure 26: Month-Hourly Average 2023 EIM Market Prices

As shown in Section 3.2, in contrast to the High Solar case, in the High Wind case,
the reserves profile is more uniform across time. Additionally, the period of
highest reserves needs do not necessarily coincide with low net loads resulting
from high Idaho Power wind production because Idaho Power wind production
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 56 |
tends to be highest during wintertime evenings. This results in fewer binding Pmin
intervals in the High Wind case that force suboptimal market transactions.
Not retiring a Bridger unit and high hydro conditions increases the cost of
integrating new solar. In these cases, having higher levels of must run coal or must
take hydro has the effect of decreasing the solar production level at which these
binding Pmin events take place.
As shown in
Table 12, the VER integration costs are typically dominated by the costs of
imperfect unit commitment and dispatch costs. Therefore, the reader can largely
focus on periods in which these binding Pmin events occur when seeking to
understand what drives integration costs for the different cases.
5.1.2 HIGH SOLAR WITH STORAGE CASES
A paradoxical finding of this analysis is that the total specific integration cost of solar
is lower for the High Solar + 200 MW Battery case than the High Solar + 400 MW
Battery case.
The reason for this is due to the way in which this study calculates VER integration
costs. As discussed in Section 2.1, the VER integration costs are calculated as the
sum of the ramping and start costs, plus the total imperfect unit commitment and
dispatch costs. The total imperfect unit commitment and dispatch cost is calculated
for each case as the difference of production costs for the imperfect foresight and
perfect foresight cases. The only difference between these cases is how much VER
forecast error, subhourly VER variability and reserves are carried for the
PUBLIC

P a g e | 57 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
incremental VER build. Due to its greater capacity, the larger 400 MW battery
allows for a greater production cost savings than the 200 MW battery when moving
from the imperfect foresight to the perfect foresight case. This larger savings is
added into the integration cost. Therefore, the apparent integration cost is higher
for the 400 MW battery than the 200 MW battery. However, there are limitations
to how this study was able to model a PURPA solar + ITC-enabled solar fleet in
PLEXOS. These limitations are discussed below.
The PLEXOS model used to calculate Idaho Power’s VER integration costs has
multiple stages that reflect different levels of uncertainty the DA, HA, RT15, and
RT5 time intervals. Storage dispatch can change between the stages due to
different grid conditions and solar forecasts. If storage provides more flexibility
ahead of real time, it can leave real-time dispatch with lower levels of flexibility, or
vice versa. The difference between storage dispatch in perfect and imperfect
foresight cases, propagated through multiple modeling time horizons, results in the
potential for small, unexpected swings in VER integration costs. Considerations
with respect to storage scheduling include:
 Storage scheduling between different commitment timeframes will
evolve as more storage is deployed. Currently, there is not a standard
practice for battery storage scheduling
 The scheduling of PURPA-contracted storage over multiple timeframes is
especially uncertain given the lack of experience with this type of
resource
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 58 |
 The scheduling of PURPA-contracted storage in a perfect foresight
counterfactual will never be known with any precision because grids are
not operated with perfect foresight.
The impact of storage sizing on unit commitment may be non-linear – a bigger
battery may cause a large Idaho Power unit to alter its commitment schedule
whereas a small battery would not be able to cause as big of an impact.
Additionally, the interaction between storage dispatch and Idaho Power market
revenues can create significant swings in the VER integration cost. The extent to
which Idaho Power has control over PURPA-contracted battery operations can
impact market revenues, especially during periods of extreme EIM prices.
The considerations above imply that there is uncertainty around future PURPA-
contracted storage dispatch and VER integration costs. E3 has included many of
the relevant dynamics in the PLEXOS model, and believes that the two integration
cost calculations for storage are within reasonable bounds of error given what is
known currently about PURPA-contracted storage. However, E3 believes it is
appropriate to use the results from these two cases to derive an average solar +
storage VER integration cost, rather than assign discrete values to different storage
capacities.
PUBLIC

P a g e | 59 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
5.2 Comparison to Data in Literature and 2018 Idaho
Power VER Study
In its Western Wind and Solar Integration Study: Phase 211, NREL calculated
integration costs for up to 33 percent penetration of wind and solar in the
Western Interconnection. The summary integration costs by scenario from the
NREL study, the 2018 Idaho Power VER integration study and this study are shown
below in Table 18, in 2020 dollars. Generally, it can be seen that the values from
this study vary considerably more than the values from the NREL study. The NREL
study integrated wind and solar across the Western Interconnection versus a
small individual balancing area, and did not use the same reserves derivation
process as this study. Modeling the entire Western Interconnection meant that
NREL did not assess the effects of suboptimal market trades on integration costs
at the interconnection footprint level. Additionally, the greater resource diversity
across the entire Western Interconnection likely reduces specific VER integration
costs. However, the general takeaway from this modeling is that VER integration
costs in the 2018 and 2020 Idaho Power VER integration studies are generally
higher than those from prior NREL work.

11 https://www.nrel.gov/docs/fy13osti/55588.pdf
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 60 |
Table 18: Comparison of 2020 Idaho Power VER Study Results to Other VER
Integration Cost Results
Case Total percent of
Annual Load
Supplied by VERs
(Total VER
Generation/Gross
Load)
Specific
Integration
Cost, Low
Bound
(2020$/MWh
VER)
NREL High Wind 33 % $0.25-0.75
NREL High Solar 33 % $0.22-0.56
NREL Mixed Resources 33 % $0.16-0.43
2020 Idaho Power VER Study High Solar Cases
(no storage or curtailment allowed) 28 % $3.86-4.65
2020 Idaho Power VER Study High Wind Case 28 % $0.77
2020 Idaho Power VER Study High Wind and
Solar Case 38 % $2.46
2018 Idaho Power VER Study 1,000 MW of
Wind Case 14 % $6.17
5.3 Methodological Differences between 2020 and
2018 Idaho Power Company Variable Energy
Resource Analysis
5.3.1 OVERVIEW
The incremental integration costs shown in this study are lower than those from
the 2018 Idaho Variable Energy Resource Analysis. While it was not in scope for
E3 to perform a detailed analysis of the 2018 study and how its methodology
differed from that of this analysis, several things stand out as important
differences between the two studies.
PUBLIC

P a g e | 61 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
5.3.2 RESERVES
The 2018 study calculates reserves in a very different manner than in the 2020
study. The resulting average reserves levels are higher in the 2018 study than
those investigated in the 2020 study. The 2020 study includes CAISO FRP reserves,
regulation reserves and contingency reserves. The 2018 study included regulation
reserves and contingency reserves, but the regulation reserves were calculated
differently.
In the 2020 study, to derive the CAISO FRP reserves, E3 used a method that
approximates the method used to derive the CAISO FRP within reasonable
bounds.12 The CAISO FRP has RT15 and RT5 stages. For the RT15 stage, E3
calculated the uncertainty component of the FRP using the difference between
2019 HA forecast net load and RT5 actual net load. Similarly to CAISO’s derivation
methodology, E3 then binned this net load forecast error by month-hour and
used a 95 percent confidence interval (as does CAISO) to determine headroom
and footroom components of the uncertainty reserves. After capping these net
load-based reserves using P98 and P2 values for footroom and headroom,
respectively, E3 assumes a 40 percent diversity credit to reduce the uncertainty
component by the same percentage in all hours, based on historical levels of EIM
footprint diversity. This 40 percent value approximates the caps and “credit”
system that the CAISO FRP uses.13 Finally, E3 calculates the RT5 CAISO FRP using

12 See, e.g. http://www.caiso.com/InitiativeDocuments/DMMResourceSufficiencyEvaluationPresentation-
EnergyImbalanceMarketofferRulesTechnicalWorkshop.pdf for a description of CAISO FRR components.
13 See, e.g. http://www.caiso.com/InitiativeDocuments/DMMResourceSufficiencyEvaluationPresentation-
EnergyImbalanceMarketofferRulesTechnicalWorkshop.pdf for a description of CAISO FRR components.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 62 |
historical data derived from the ratio of 2019 CAISO RT5 FRP uncertainty reserves
to the 2019 CAISO RT15 FRP uncertainty reserves.14
E3 calculates regulation reserves for the individual load, wind and solar profiles
using a persistence forecast of the 5-minute data. Solar data are then binned by
season, hour and percent output, whereas load and wind are binned by percent
of maximum observed load and output, respectively. A 95 percent confidence
interval is then used to derive headroom and footroom needs for these reserves,
and they are then combined using a root mean square, assuming that the load,
wind and solar regulation components have no covariance on this short
timescale. Finally, spinning contingency reserves are calculated at 3 percent of
load. This results in the average reserves shown below in Table 19.

Table 19: Reserves Summary for Different 2020 Idaho Power VER Integration
Cost Cases
Case Total MW Online Wind
(MW)
Total MW Online Solar
(MW)
Avg. RT15 FRP Up
(MW)
Avg. RT15 FRP Down
(MW)
Avg. Reg. Up
(MW)
Avg. Reg. Down
(MW)
Avg. Conting.
Res.
(MW)
Avg. Total Res. Up
(Percent of Avg. Load)
Avg. Total Reserves Down
(Percent of Avg.
Load)
1. 2023
Base
Case
728 561 100 97 40 41 104 13 % 7 %
2. Jim
Bridger
Online
728 561 100 97 40 41 104 13 % 7 %
3. Hi
Solar 728 1,354 147 142 71 72 104 17 % 11 %

14 http://oasis.caiso.com/mrioasis/logon.do
PUBLIC

P a g e | 63 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
4. Hi
Solar,
Low
Hydro
728 1,354 147 142 71 72 104 17 % 11 %
5. Hi
Wind 1,396 561 152 147 50 52 104 16 % 10 %
6. Hi
Solar,
Hi Wind
1,396 1,354 193 186 79 81 104 19 % 13 %
7.
Existing
Solar
Base
Case
728 561 87 86 32 33 104 11% 6%
8. Hi
Solar,
Hi
Hydro
728 1,354 147 142 71 72 104 17 % 11 %
9. Hi
Solar,
200
MW
Battery
728 1,354 147 142 71 72 104 17 % 11 %
10. Hi
Solar,
400
MW
Battery
728 1,354 147 142 71 72 104 17 % 11 %
11. Hi
Curtail.
Solar
728 1,354 147 142 71 72 104 17 % 11 %
In the 2018 study, Idaho Power calculated the regulation reserves using 2HA
forecasted wind and load, and 1-minute actual wind and load data. These data
were then binned by percentage of wind output or maximum load. It is not clear
from the study if confidence intervals are subsequently applied to this data, but
the resulting reserves, as a percentage of normalized load, are shown below as
Table 20 and Table 21. Spinning reserves are calculated as 3 % of the hourly load,
which is identical to the method E3 used.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 64 |
Table 20: 2018 Idaho Power VER Integration Study Wind Reserves
Winter Spring Summer Fall
Wind Quartile of Forec.
Output
Reg Up % of Avg Wind
Forec.
Reg Down % of Avg (Namplate – Forec.)
Reg Up % of Avg Wind
Forec.
Reg Down % of Avg (Namplate – Forec.)
Reg Up % of Avg Wind
Forec.
Reg Down % of Avg (Namplate – Forec.)
Reg Up % of Avg Wind
Forec.
Reg Down % of Avg (Namplate – Forec.)
1. 100% 28 % 100% 62 % 100 % 48 % 100 % 66 %
2. 86 % 51 % 94 % 79 % 93 % 75 % 80 % 65 %
3. 55 % 65 % 71 % 81 % 68 % 85 % 76 % 75 %
4. 49 % 34 % 43 % 69 % 59 % 82 % 39 % 43 %

As shown in Table 20 and Table 21, the 2018 study had much higher reserves than
the 2020 study, particularly for VERs. This likely results in higher costs for
integrating VERs in the 2018 study, due to the high reserves levels causing more
binding Pmin constraints for a given VER penetration level.
Table 21: 2018 Idaho Power VER Integration Study Load Reserves
Winter Spring Summer Fall
Load Quartile of Forecast
Maximum
Reg Up % of
Avg
Load
Reg Down
% of
Avg
Load
Reg Up % of
Avg
Load
Reg Down %
of Avg
Load
Reg Up % of
Avg
Load
Reg Down
% of
Avg
Load
Reg Up % of
Avg
Load
Reg Down %
of Avg
Load
1. 4.9 % 9.1 % 8.1 % 10.5 % 7.9 % 11.5 % 8.0 % 10.6 %
2. 9.3 % 6.8 % 6.8 % 11.3 % 8.1 % 6.0 % 7.5 % 8.9 %
3. 9.5 % 5.8 % 9.9 % 6.7 % 9.7 % 9.8 % 9.9 % 8.5 %
4. 7.9 % 6.9 % 8.3 % 7.0 % 6.2 % 13.3 % 7.3 % 7.1 %
E3 believes that its 2020 reserve derivation methodology is closer to standard
practice than the method used in the 2018 study. There was negligible observed
unserved energy in E3’s models. Similar normalized levels of reserves (MW per
PUBLIC

P a g e | 65 |
Discussion
© 2010 Energy and Environmental Economics, Inc.
MW of installed VERs) and confidence intervals of historical forecast error have
been used elsewhere.15 16 17
In both the 2018 study and the 2020 study, there were a significant number of
hours in which the AURORA and PLEXOS models were unable to hold sufficient
reserves to meet the requirements outlined above. In the PLEXOS model, the
reserve violation penalties were set up such that regulation reserves were
typically not met whereas CAISO FRP reserves and contingency reserves were
nearly always met.
5.3.3 TREATMENT OF EXTERNAL MARKETS
The 2020 study is modeled with an EIM market, whereas the 2018 study is not.
Because Idaho Power joined the EIM in Q2 2018, this omission was reasonable in
the 2018 study. In the 2020 study, the presence of the EIM market allows the
model to balance forecast error from the DA and HA intervals to the real time.
The 2018 model had less flexibility in its ability to trade, which likely reduces the
ability of Idaho Power’s system to buy and sell from the market to enable
procuring reserves relative to a scenario with the EIM.
5.3.4 MULTISTAGE VS. SINGLE STAGE MODEL
The 2020 study used a multistage PLEXOS model, which contains information
about typical net load forecast error and subhourly net load variability, whereas

15 Z. Zhou, T. Levin, G. Conzelmann, “Survey of U.S. Ancillary Services Markets.”
https://publications.anl.gov/anlpubs/2016/01/124217.pdf
16http://www.ercot.com/content/wcm/key_documents_lists/137978/9_2019_Methodology_for_Determining_Mini
mum_Ancillary_Service_Requirements.pdf
17 http://www.caiso.com/Documents/Addendum-DraftFinalTechnicalAppendix-FlexibleRampingProduct.pdf
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 66 |
the 2018 study used a single hourly stage AURORA model that did not reflect
forecast error. In executing its multistage PLEXOS model, E3 did not observe
significant levels of unserved energy. Therefore E3 believes its reserves derivation
method provides reasonable reserve levels.
PUBLIC

P a g e | 67 |
Conclusions
© 2010 Energy and Environmental Economics, Inc.
6 Conclusions
6.1 Integration Costs
Overall, it was found that integration costs for new VERs on Idaho Power’s system
vary from $0.64/MWh up to $4.65/MWh. Generally, solar integration costs are
significantly higher than those for new wind. Adding more must-run resources,
such as hydro operating at very high capacity factors, or keeping must run thermal
units online, increases VER integration costs. Increasing system flexibility, such as
by pairing solar with dispatchable storage, or by allowing solar to be economically
curtailed, reduces VER integration costs.
Additionally, the VER integration costs found herein are significantly lower than
those from the 2018 Idaho Power VER integration study. This is due to multiple
factors, but likely the single greatest cause is the reduction in growth in reserves
per unit of online wind and solar capacity in the 2020 study versus the 2018 study.
Finally, the results from this study are contingent upon VERs being must take; coal
units being committed as baseload, must run units; maintaining strategies for
deploying Idaho Power’s HCC hydroelectric resources; storage paired with solar
not being able to provide reserves; and other assumptions about current
practices that may change in the future.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 68 |
7 Appendix 1: Process
Document
7.1 Introduction
This Appendix is provided as a guide to further understand how E3 developed its
PLEXOS model for this study.
The production cost simulation software, PLEXOS, was used to calculate VER
integration costs in this study. This was done by using PLEXOS to generate the
outputs necessary to derive the VER integration costs: start/stop costs, ramping
cost, imperfect unit commitment and dispatch fuel costs, imperfect unit
commitment and dispatch net import costs and curtailment costs.
To yield results, PLEXOS requires various inputs into E3’s four stage model. The
inputs to the PLEXOS model were developed by E3, Idaho Power, and in some
instances in collaboration between Idaho Power and E3. These include:
 Load Profiles: The 2019 profiles were developed by Idaho Power and E3
and consist of 4 comma separated value (CSV) files to represent load
forecasts at the DA, HA, and RT15 stages with the RT5 profile seen as the
actual load profile, and these were scaled to 2023 load profiles by E3.
 Renewable Profiles: Solar and wind profiles were developed by E3 using
Idaho Power’s data and consist of 4 CSV files to represent generation
forecasts at the DA, HA, and RT15 stages with the RT5 profile seen as the
actual output.
PUBLIC

P a g e | 69 |
Appendix 1: Process Document
© 2010 Energy and Environmental Economics, Inc.
 Hydro Profiles: Daily hydro budgets were created by E3 using Idaho
Power’s historical hydro data, and Pmax/Pmin levels were derived using
Idaho Power input. These are fed into the model using separate CSVs for
daily HCC maximum power, daily HCC minimum power, daily HCC energy
budget and daily RoR power outputs
 Generator Characteristics: Generator characteristics were provided by
Idaho Power as E3’s part of the data collection process and include
properties such as maximum and minimum capacity, ramp rates, start-up
costs, VO&M, as well as any must-run flags or particular generating
patterns. These are input for each generator using the PLEXOS UI.
 Reserve Policies and Profiles: Reserve profiles for the “perfect foresight”
and “imperfect foresight” cases were developed using E3’s RESERVE tool,
along with the renewable and load profiles provided by E3. Each case has
its own set of reserve profiles, which are in the form of CSVs read in for
the flexible ramping requirement and the regulation needs. Contingency
reserves are enforced within the PLEXOS UI.
 Topology and Transmission: The transmission and zonal topology of the
model was created by E3 with input from Idaho Power towards
transmission capacity to the Mid C and PV market nodes. These limits and
the topology were input to the PLEXOS UI.
 Markets: Market transaction limits were provided by Idaho Power for the
two markets nodes, Mid C and PV, represented within this model.
Forward Q2-Q4 2019 and Q1 2020 market prices were provided to E3 by
Idaho Power, and E3 downloaded historical Q2-Q4 2019 and Q1 2020 EIM
market prices. These prices are then modified using E3’s in-house
AURORA price forecasts to adjust them to 2023 expected market prices.
These adjusted prices are fed into the model using CSVs for each market
and model stage.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 70 |
 Fuel Prices: Fuel prices were provided to E3 for each of the generators,
and are enforced within the PLEXOS UI.
When running a case within PLEXOS, it is important to ensure that the appropriate
renewable profiles are added as data files in the model. These are found in the
‘Wind Profiles’ and ‘Solar Profiles’ subfolders within the ‘Data’ directory and ‘Data
Files’ folder illustrated in Figure 27. In addition, if need be, updated reserve
profiles must also be added to the PLEXOS model. These data files are named to
correspond with the relevant case they will be used for and can be found under
the ‘Reserves Idaho Power’ subfolder in the ‘Data’ directory and within the ‘Data
Files’ folder. Daily hydro budget profiles can be added or adjusted within the
‘Hydro Budgets’ subfolder within the ‘Data Files’ folder.
Figure 27: PLEXOS Data Directory

Creating a new case or editing an existing case’s properties can be done within
the PLEXOS UI’s ‘Scenarios’ folder seen in Figure 28 under ‘Idaho Power Core
PUBLIC

P a g e | 71 |
Appendix 1: Process Document
© 2010 Energy and Environmental Economics, Inc.
Cases’. Each Scenario represents an individual case. The properties that are
tagged with this case ‘Scenario’ will only be used if this case is being run.
Figure 28: PLEXOS Scenario Directory

A specific case is only run if the ‘Scenario’ associated with it is included in the
‘Membership’ of each monthly stage model (DA, HA, RT15, RT5) and can be
identified as shown in Figure 29. Only one ‘Idaho Power Core Cases’ ‘Scenario’
can be linked to the models at any one time. If multiple case ‘Scenarios’ are
included in the model ‘Memberships’, errors may occur while attempting to
execute the full model or may yield incorrect results.
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 72 |
Figure 29: PLEXOS Membership view

To derive VER integration costs, the overall PLEXOS model is run twice for each
case, once using the perfect foresight profiles for the relevant VER resources and
reserves, and then once using the imperfect foresight reserve and VER profiles.
The individual cases are expressed as individual PLEXOS models with custom
modifications and, in some instances, CSV files. The primary differences between
the cases are described below.
 Case 1 is the 2023 base case for Cases 3-6 and Cases 8-11, which has all
known unit additions and retirements and also includes the known 2019
through 2023 load growth. The Solar and Wind objects are scaled to the
appropriate size for Case 1
PUBLIC

P a g e | 73 |
Appendix 1: Process Document
© 2010 Energy and Environmental Economics, Inc.
 Case 2 explores the effect of not retiring one of the Bridger coal plant’s
units, but is otherwise identical to Case 1. The Bridger coal plant Pmin
and Pmax are increased to reflect this change
 Case 3 builds on Case 1 by exploring the effect of adding enough new
solar (794 MW of new solar) such that 10 percent of the 2023 Idaho
Power average gross load is provided by this new solar build. This is done
using the existing aggregated solar plant from Case 1
 Case 4 extends the Case 3 analysis to a low, rather than average hydro
year. The hydro budgets and daily Pmin/Pmax levels are updated using
the CSVs fed into the model
 Case 5 builds on Case 1 and explores the integration costs of a high wind
build. Case 5 assumes a new wind build that can supply 10 percent of the
annual 2023 Idaho Power gross load (669 MW of new wind). This is
performed using the existing wind object from Case 1
 Case 6 builds on Case 3 and Case 5, including both high solar and high
wind builds (794 MW of new solar and 669 MW of new wind). This is done
using the existing solar and wind objects from Case 1
 Case 7 is identical to Case 1, except that none of proposed solar additions
come online from 2019 to 2023, resulting in 251 MW fewer of solar than
Case 1 and lower reserves needs. This is done using the existing solar
object from Case 1
 Cases 8 extends the Case 3 analysis to a high, rather than average hydro
year, and as in Case 4, this is accomplished by feeding in different CSVs
to adjust the energy budgets and Pmax/Pmin levels
 Case 9 builds on Case 3 by adding a 200 MW 4-hour Battery object with
a roundtrip efficiency of 85% and can only charge from the additional 794
MW of new solar
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 74 |
 Case 10 adds a 400 MW 4-hour Battery object with an 85% roundtrip
efficiency and is only able to charge from the additional 794 MW of new
solar
 Case 11 splits the solar object in Case 3 into two distinct generator
objects: an ‘Idaho Solar’ and ‘Idaho Solar Curtailable’. The ‘Idaho Solar’
resource is modeled as must-take, while the ‘Idaho Solar Curtailable’
object is allowed economically curtail
7.2 Results Processing
The results viewer enables us to display annual PLEXOS ST data in a more user-
friendly format and consists of several different tabs. Below, we explain how to
navigate and manipulate each tab in the order of their use when processing
results:
 Cover: this tab provides a high-level overview of the workbook and is not
of any practical use in processing results
 Params: The Params tab is used as a library that the embedded excel
macro will read and use to pull outputs from individual properties in the
PLEXOS solutions zip files. The ‘ParentClassName’ column corresponds to
the tabs within the PLEXOS UI either ‘System’ or ‘Simulation’ as seen in
Figure 29. The ‘ParentName’ is the system name within PLEXOS which is
given as ‘IPC’ in this model. ‘ChildClassName’ is the subfolder name
within any of the ‘Production’, ‘Transmission’, ‘Generic’, ‘Data’ folders.
For example, ‘Generators’ or ‘Lines’. The ‘PropertyName’ column is the
name of the property to be output to the results viewer. ‘ChildName’ is
the name of the object that the output property belongs to. If the
generation of a generator called ‘GEN1’ needed to be brought into the
PUBLIC

P a g e | 75 |
Appendix 1: Process Document
© 2010 Energy and Environmental Economics, Inc.
results viewer then the ‘PropertyName’ would be ‘Generation’ and the
‘ChildName’ would be ‘GEN1’.

Figure 30: PLEXOS UI

If pulling in individual object properties, the ‘AggregrationEnum_type’
column by default should be input as ‘AggregationEnum_None’ and the
‘agg_category’ column should be left blank; however if it is more
beneficial to load properties from all objects within a subfolder of the
‘ChildClassName’ folders such as ‘IPC Solar’ as seen in Figure 30, then it is
possible to do this by leaving the ‘ChildName’ column blank, changing the
‘AggregrationEnum_type’ column entry to ‘AggregationEnum_Category’,
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 76 |
and changing the ‘agg_category’ entry to ‘IPC Solar’. Finally, the ‘Units’
column should contain the units of the property that is being selected.
One should ensure that the properties that are being listed in the Params
tab in the results viewer are being output by the PLEXOS model. It is
possible to verify and, if need be, add the property to be output as part
of the PLEXOS solution zip file through the PLEXOS UI. As seen in Figure
31, by clicking on the ‘Simulation’ tab in the PLEXOS UI and double clicking
on the object within the ‘Reports’ subfolder, the ‘Field List’ tab will show
the entire list of possible outputs from the model.
Figure 31: PLEXOS Reports

Ensure that the desired outputs have the ‘Period’, or ‘Flat File’ boxes
checked. PLEXOS Help documentation is extremely thorough in providing
additional detail in understanding the full amount of available output
properties. This must be done before running the models to ensure that
the selected outputs are created in the PLEXOS solution zip files.
Control: Once the desired outputs are set in the ‘Params’ tab, the results
viewer can be run. The ‘Control’ tab contains a few cells that must be filled
PUBLIC

P a g e | 77 |
Appendix 1: Process Document
© 2010 Energy and Environmental Economics, Inc.
out before running the Macro. The ‘Start Solution Month’ and ‘End
Solution Month’ allows the flexibility to run the results viewer for one
month or a set of months if need be, though use caution as the results
viewer capacity factor calculations are set up to calculate over the whole
year so may yield incorrect results if not run over the whole year. In
addition, ensure that the ‘Stage Name’ and ‘Model Name Constant’ inputs
are aligned with the model names as seen in Figure 32, where the ‘Stage
Name’ is ‘RealTime5’ and the ‘Model Name Constant’ is ‘IPC’. The rest of
the values within the ‘Control’ tab should not be touched. Ensure
calculations within the workbook are set to manual and then click the ‘Do
all the PLEXOS things NOW!’ button to start the results viewer.
Figure 32 PLEXOS Model Naming Convention

 TimeSeries Data: Once the results viewer is finished compiling the
PLEXOS outputs these will all appear in the ‘TimeSeries Data’ tab.
 Plot: The ‘Plot’ tab provides dispatch plots, price plots, and market
transaction plots of a user-selected date. The day chosen can be toggled
between any days represented within the output data. The ‘Plot’ tab also
PUBLIC

2020 Idaho Power VER Integration Study
P a g e | 78 |
provides an annual look at capacity factor, cost, generation, number of
starts by generator and provides annual cost and generation figures
associated with market transactions to provide an overall production cost
for the system over the year.
 Month-Hour Summary: This tab converts the 5-minute data within the
‘TimeSeries Data’ tab to hourly average values which is then used to
create heat maps.
 Month-Hour: This tab is used as a data visualizing tool to display output
data as month-hour average heat maps. The data being shown in the heat
map can be toggled by the user via the dropdown menu.
 SummaryAll: The ‘SummaryAll’ tab offers a quick average value of each
of the properties listed in the ‘Params’ tab.
 Hydro Budget: This tab provides information on Hells Canyon Complex
hydro budgets.
 Conversion: This tab provides conversion figures within the workbook.

PUBLIC