Attachment 2.

10-1

Analysis of “Loss of Load Probability”

(LOLP) at various Planning Reserve
Margins

Prepared for: Prepared by:

Public Service Company of Ventyx
Colorado 2379 Gateway Oaks Drive, Suite 200
Sacramento, CA 95833
Date Submitted: (916) 569-0985
December 1, 2008 www.ventyx.com

Contact:
Richard Lauckhart, Senior Vice President
Chintamani Kulkarni, Consultant
916-609-7769
Attachment 2.10-1
 Attachment 2.10-1

Liability Note
Ventyx provides this document “as is” without warranty of any kind, either expressed or implied,
including, but not limited to, the implied warranty of merchantability or fitness for a particular
purpose. Ventyx may make changes or improvements in the equipment, software, or specifications
described in this document at any time and without notice.

Ventyx has made every reasonable effort to ensure the accuracy of this document; however, it may
contain technical inaccuracies or typographical errors. Ventyx disclaims all responsibility for any labor,
materials, or costs incurred by any person or party as a result of their use or reliance upon the content
of this document. Ventyx and its affiliated companies shall in no event be liable for any damages
(including, but not limited to, consequential, indirect or incidental, special damages or loss of profits,
use or data) arising out of or in connection with this document or its use, even if such damages were
foreseeable or Ventyx has been informed of their potential occurrence.

© 2008 by Ventyx. All rights reserved. No part of this document, or any software included with it, may
be reproduced, stored in a retrieval system, or transmitted in any form or by any means, including
photocopying, electronic, mechanical, recording or otherwise, without prior written consent of Ventyx.
This document contains the proprietary and confidential information of Ventyx. The disclosure of its
contents to any third party is strictly prohibited, without the prior written consent of Ventyx.
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Table of Contents

Section Page

EXECUTIVE SUMMARY .................................................................................................... 1

1 RECENTLY ACCEPTED APPROACHES.................................................................1-1
1.1 Previous Lolp Studies .........................................................................................1-1
2 PSCO FOCUSED ANALYSIS USING PORTFOLIO STATISTICAL ANALYSIS AND
EXPECTED ENERGY NOT SERVED (ENS)............................................................2-1
2.1 Overview Of Analysis Performed ........................................................................2-1
2.2 Test Year For Analysis........................................................................................2-1
2.3 Resources in the base year ................................................................................2-1
2.4 Year 2013 loads ..................................................................................................2-2
2.4.1 Load Stochastic Process And Volatility Parameters ..................................2-3
2.5 Modeling Of Wind Volatility .................................................................................2-4
2.6 Forced outage rates on supply resources ..........................................................2-5
2.7 Rocky Mountain Reserve Group Support ...........................................................2-6
2.7.1 Modeling the RMRG Support .....................................................................2-8
2.8 Transmission LIFELINE - Non Psco Imports ......................................................2-9
2.9 Using A Generic Gas Turbine As A Proxy For Increasing Planning Reserves At
The Margin ..........................................................................................................2-9
2.10 Determination Of Lolp .......................................................................................2-10
2.10.1 Calculating The Planning Reserve Margin...............................................2-10
2.11 Analysis Starting Point Operating Reserve Margin...........................................2-11
3 ANALYSIS RESULTS .................................................................................................. 1

List of Figures

Figure 1 PSCo Load Distribution - Confidence Intervals .................................................2-4

Figure 2 PSCo Stochastic Wind Data for a Location .......................................................2-5
Figure 3 Expected Hours of Energy Not Served................................................................. 1

List of Tables

Table 1 Public Service of Colorado expected 2013 Summer Resource Capacity .........2-2
Table 2 PSCo Load Stochastic Parameters ...................................................................2-3
Table 3 Public Service of Colorado Station Outage Rate...............................................2-5
Table 4 Rocky Mountain Reserve Group Response Matrix ...........................................2-7
Table 5 RMRG Contingency Assistance for PSCo Units Greater than 200 MW............2-8
Table 6 2013 PSCo Expected Reserve Margin ............................................................2-11
Table 7 LOLP Results Summary ........................................................................................ 1

| i
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

EXECUTIVE SUMMARY

At the request of Public Service Company of Colorado (PSCo), Ventyx performed a

stochastic analysis of the relationship between electric generating capacity reserve
margin (aka, planning reserves) and the ability of the PSCo system to reliably maintain
service to load. The analysis focused on the year 2013 and accounts for PSCo existing
and expected generation resources and the anticipated availability characteristics of
those resources. The analysis takes into consideration PSCo’s hourly customer electric
demands and the volatility of those demands due to weather. The analysis incorporates a
representation of the reliability support that PSCo can expect to receive from the Rocky
Mountain Reserve Group (RMRG) under single contingency events of 200 MW or
greater. The reserve margin study also incorporates PSCo’s obligation to carry
approximately 419 MW of operating reserves for year 2013 as part of its membership in
the RMRG. Additionally, the analysis considers the reliability contribution of transmission
lifeline capacity generally reserved for system emergencies.

Ventyx performed the analysis using the Market Analytic’s Planning & Risk Module
(PaR). The load, wind generation, and unit availability were treated stochastically. The
level of energy not served from the PaR modeling work was used to determine the
expected level of reliability for the system for different levels of capacity reserve margin.
The analysis indicates that a Planning Reserve Margin of 16.3% would provide an
expected probability that the PSCo system would be unable to serve firm load customers
approximately 1-day-in-10-years. This level of reliability is considered acceptable and
often used as a standard for reliable systems within the electric utility industry.

| ES-1
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

1 RECENTLY ACCEPTED APPROACHES

1.1 PREVIOUS LOLP STUDIES

In 2003 Resource Plan Filings with the California Public Utility Commission, three
different Investor Owned Utilities (PG&E, SCE, and SDG&E) all performed portfolio
stochastic analysis to assess appropriate levels of planning reserves. In these analyses,
the utilities selected an upcoming applicable year and tested the ability of their power
supply systems to meet customer loads in that year under different utility supply portfolios
that gave different levels of planning reserve. The methodology involved performing
hourly economic dispatch of resources against loads for each hour of the year. Because
of the uncertainties of unit forced outage and load level variations caused by weather,
multiple iterations of the year were performed. Under each iteration, Monte Carlo draws
were made daily that adjusted load levels either upward or downward. Further, Monte
Carlo draws were made to reflect possibilities of unit forced outage. The California PUC
accepted the methodology at that time, but more recently some utilities have indicated
that higher reserve margins should be required because of the possibility of non-
performance of PPAs, etc. The California PUC has therefore opened another proceeding
to discuss possible changes to reflect these matters.

It is typical to use a 1-day-in-10-year Loss of Load Probability (LOLP) when determining

the needed Planning Reserve Margin. This level of LOLP is equivalent to failing to serve
the energy requirements of the system for 2.4 hours each year or 24 hours during a 10-
year period.

| 1-1
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

2 PSCO FOCUSED ANALYSIS USING PORTFOLIO

STATISTICAL ANALYSIS AND EXPECTED ENERGY NOT
SERVED (ENS)

2.1 OVERVIEW OF ANALYSIS PERFORMED

Ventyx has performed a stochastic analysis of Loss of Load Probability on the PSCo
system in a manner similar to the analysis performed by California investor owned utilities
in the year 2003 and accepted by the California PUC as well as by PSCo in 2004 (filed
with PSCo’s 2003 LCP). In particular, Ventyx focused on PSCo existing and expected
generation resources and loads in year 2013. The analysis also reflects a PSCo
operating reserve of 419 MW, which represents PSCo’s expected operating reserve
obligation under the RMRG after the Comanche 3 unit becomes operational.

Ventyx utilized its regional Market Analytics software module, Planning & Risk, to perform
this stochastic reserve margin analysis of the PSCo system. The key factors represented
stochastically in this analysis are:

• Unit forced outages and maintenance,

• Weather related load volatility,
• Wind generation, and
• Transmission lifeline capacity.

Ventyx stochastically simulated the hourly dispatch of the PSCo system for year 2013,
where Monte Carlo draws were performed for 100 iterations in order to capture the
impact of uncertainties in these key factors.

2.2 TEST YEAR FOR ANALYSIS

Consistent with PSCo’s 2007 CRP, PSCo provided the portfolio of resources, wind
pattern, unit maintenance and forced outages and the hourly load forecast for the year
2013 for the purpose of this study.

2.3 RESOURCES IN THE BASE YEAR

PSCo generation resources in the year 2013 reflected in the analyses are listed in Table
1 below. The Comanche 3 facility was modeled at it full expected capacity of 784 MW
and the full load requirements of IREA and Holy Cross were included in the modeling of
customer demand (i.e., as opposed to modeling only PSCo’s share of Comanche 3 and
removing the portion of IREA and Holy Cross’s load that will be served by their ownership
share of Comanche 3).

| 2-1
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Table 1
Public Service of Colorado expected 2013 Summer Resource Capacity
Peak Peak Peak Peak
Resource Capacity Resource Capacity Resource Capacity Resource Capacity
MW MW MW MW
Alamosa 1 12.82 Comanche 1 325 Hayden 1 139 Sunshine Hydro 0.7
Alamosa 2 13.5 Comanche 2 335 Hayden 2 99 Tacoma Hydro 8.5
AMES HYDRO 3.75 Comanche 3 784 HillCrest Hydro 2.3 Thermo RS1 31CC 152
Arapahoe CC 479 Craig 1 41.6 Kohler Hydro 0.15 Tower04WT 42.12
Basin1 LRS2 50 Craig 2 41.6 LakeGeorge Hydro 0.23 Tower41WT 98.75
Basin1 LRS3 50 CT_129_A 258.6 Manchief CT 260.7 Tower49WT 41.37
Basin2 LRS2 37.5 Dillon Hydro 1.9 Maxwell Hydro 0.15 Tri2 Craig1 9.93
Basin2 LRS3 37.5 Foothills Hydro 2.3 On_Site Solar 11.89 Tri2 Craig2 9.93
Betasso Hydro 8.57 Fruita 15 Orodell Hydro 0.22 Tri2 Craig3 38.29
BioGas 75th ST 0.5 FSV CC 1x1 226 Ouray Hydro 0.5 Tri2 LRS2 19.18
BioMass 4 FSV CC 2x1 252 Palisade Hydro 1.7 Tri2 LRS3 19.18
Brush 13 75 FSV CC 3x1 230 Pawnee 1 505 Tri3 Craig1 2.49
Brush 4D CC2 133 FSV CT 270 PlainsEnd2 CC 224 Tri3 Craig2 2.49
Cabin Crk Gen1 105 Ft Lupton 1 44.7 Redlands Hydro 1.4 Tri3 Craig3 9.84
Cabin Crk Gen2 105 Ft Lupton 2 44.7 Roberts T Hydro 6.1 Tri3 LRS2 4.8
Central Solar 11.12 Georgetown Hydro 1.2 Rocky Mtn CC21 601 Tri3 LRS3 4.8
Cherokee 1 107 Gross Res Hydro 8.1 Salida Hydro 1.4 TST Brighton 132
Cherokee 2 106 Spindle_CT 269 Shoshone Hydro 15 TST Limon 66
Cherokee 3 152 SPS TieLine 101 Stagecoach Hydro 0.8 UNC Greeley EXT 68.86
Cherokee 4 352 Valmont 6 43 Strontia Hydro 1.2 Valmont 5 186
Cherokee Diesel 5.5 WM Landfill Gas 3.2 SunEdison Solar 2.87
(Wind contributed 12.5% of nameplate, Solar at 58% and Cabin Creek 210 MW)

In the analysis, the PSCo wind generation resources were lumped together into three
distinct geographic zones: Colorado/Wyoming border zone near the existing Ponnequin
facility, northeast zone near Peetz Table, and the southern zone near the Colorado
Green facility. The three wind zones provide geographic diversity for wind generation
based on the modeling techniques applied for stochastic wind generation discussed later
in this report. For the calculation of planning reserves, the wind capacity is counted at
12.5% of their nameplate capacity.

2.4 YEAR 2013 LOADS

The analysis applied Monte Carlo draws on load to reflect the likelihood that loads will be
higher or lower as a result of weather, than what is being forecast for year 2013. To
perform this type of Monte Carlo analysis, an hourly profile of PSCo loads for the year
2013 was developed. The forecasted peak demand for year 2013 is 7,310 MW, which is
comprised of the September 2007 peak demand of 7,094 MW and an additional 216 MW
of coincident peak demand from IREA and Holy Cross. As seen above Comanche 3 was
modeled at its full capacity to accommodate serving the full load requirements of IREA
and Holy Cross. While IREA and Holy Cross will have a 250 MW share of the Comanche
3 unit, it is expected that only 216 MW of load would be coincidental with the PSCo peak
demand and only that coincident amount was considered for the total 2013 PSCo peak

| 2-2
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

demand. As described above, PSCo’s portion of Comanche 3 and IREA’s and Holy
Cross’s portion of Comanche 3, totaling to 784 MW of capacity for Comanche 3, was also
included since IREA and Holy Cross wholesale load requirements were included as part
of the PSCo load.

2.4.1 Load Stochastic Process and Volatility Parameters

The stochastic model used to perform the stochastic draws on load is a two-factor model
in which one factor represents short-term or temporary deviations and the other factor
represents long-term or cumulative deviations. Long-term effects include trends such as
change in annual peak demand growth and other forces whose effects are of long
duration, which follow a random walk. In the short term, shocks may drive variables
away from their long-term equilibrium level, but adjustment processes tend to pull them
back to their equilibrium or expected level in the short term. In other words, short-term
shocks such as changes to load due to weather are mean reverting. The rate at which
the random variable tends to revert to the expected value is an input to the process. This
is referred to as the mean reversion rate. The two-factor model combines the short-term
mean reverting process with the long-term random walk process.

The volatility estimates for PSCo load in this study were developed from historical hourly
load data from 1996-2007. The estimated short-term stochastic parameters for PSCo
load, used as inputs into the Planning & Risk models stochastic analysis, are presented
in Table 2 below. Long-term stochastic parameters were not necessary since the study
period is a single year. As a result of these stochastic parameter inputs, a distribution of
load volatility is created.

Table 2
PSCo Load Stochastic Parameters
1 Load
Season
PSCo
2013 Alpha Sigma
Winter 0.275 0.014
Spring 0.266 0.015
Summer 0.195 0.016
Fall 0.276 0.019
Source: Ventyx.

Figure 1 illustrates the 5th, Average, and 95th confidence intervals of load distribution for
the year 2013.

1
Season definition: Winter = December-February; Spring = March-May; Summer = June-August;
Fall = September-November. Sigma is the volatility parameter and alpha is the mean reversion
parameter.

| 2-3
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Figure 1
PSCo Load Distribution - Confidence Intervals

PSCO Peak Load - 2013

9000

8000

7000

6000
Peak MW

5000

4000
5Percentile 95Percentile Expected Values
3000

2000

1000

13 13 -1
3
r-1
3
-1
3 13 -1
3 13 13 -1
3 13 13
n- b- ar ay n- ul g- p- ct v- c-
Ja Fe M Ap M Ju J Au Se O No De

2.5 MODELING OF WIND VOLATILITY

Using historical hourly wind generation from existing PSCo wind facilities, Ventyx created
100 different hourly wind patterns that reflect the unpredictable nature of the PSCo wind
resource. PSCo provided wind shapes for three wind zones: Colorado/Wyoming border,
northeast Colorado, and southeast Colorado. These three wind shapes were utilized to
model wind variability within the analysis.

The stochastic wind data was developed external to the Planning & Risk model, and
introduced during model simulation. The following method was used in creating the
stochastic wind data:

1. Hourly historical wind shapes for the three locations were developed and each
fluctuates differently due to their location and associated wind pattern.
2. To capture the randomness of wind generation, Ventyx used its Hourly Historical
Simulation Tool, which randomizes daily-hourly profiles within a month. This process
was repeated for each aggregated wind location. For example, in creating the 24-
hour by 100 iterations of data for January 1 for a location, the random number
generator picked which hourly day profile in January to choose. Since January has
31 days, the random number generator chose any one of the 31 days of January for
each of the 100 iterations for January 1. So for January 1, iteration 1 may use the
hourly profile of day 30 of January, iteration 2 may use the hourly profile of day 2 of
January and so on. This process was continued until all days of the year for each of
the 100 iterations was developed. Figure 2 shows the stochastic wind data for a
representative week in July.

| 2-4
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

3. The randomized wind data was then fed into Planning & Risk through XML
integration and included in the model simulation.
Figure 2
PSCo Stochastic Wind Data for a Location
350

300

250

200
MW

150

100

50

0
11

16

21

26

31

36

41

46

51

56

61

66

71

76

81

86

91

96

101

106

111

116

121

126

131

136

141

146

151

156

161

166
1

Hours of a Week

2.6 FORCED OUTAGE RATES ON SUPPLY RESOURCES

The expected level of forced outages for PSCo units (both owned and purchased) was
estimated from actual historical availability data. The model assumed the following
expected levels of forced outage rates on the following supplies.

Table 3
Public Service of Colorado Station Outage Rate
Station EFOR Station EFOR Station EFOR
Alamosa 1 0.10% Dillon Hydro 5.00% Rocky Mtn CC21 5.00%
Alamosa 2 0.10% Foothills Hydro 5.00% Salida Hydro 3.00%
AMES HYDRO 6.00% Fruita 7.30% Shoshone Hydro 1.00%
ArapCC 1.60% FSV CC 1x1 2.50% Spindle_CT 3.00%
Basin1 LRS2 3.00% FSV CC 2x1 2.50% SPS TieLine 0.50%
Basin1 LRS3 3.00% FSV CC 3x1 2.50% Stagecoach Hydro 5.00%
Basin2 LRS2 3.00% FSV CT 3.00% Strontia Hydro 5.00%
Basin2 LRS3 3.00% Ft Lupton 1 9.50% Sunshine Hydro 5.00%
Betasso Hydro 5.00% Ft Lupton 2 17.20% Tacoma Hydro 5.00%
BioGas 75th ST 3.00% Gen GT 3.60% Thermo RS1 31CC 3.00%
BioMass 10.00% Georgetown Hydro 2.00% Tri2 Craig1 4.80%
Brush 13 2.00% Gross Res Hydro 5.00% Tri2 Craig2 4.80%
Brush 4D CC2 2.00% Hayden 1 6.60% Tri2 Craig3 3.00%
Cabin Crk Gen1 6.00% Hayden 2 3.50% Tri2 LRS2 3.00%
Cabin Crk Gen2 6.00% HillCrest Hydro 5.00% Tri2 LRS3 3.00%

| 2-5
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Station EFOR Station EFOR Station EFOR

Cherokee 1 9.50% Kohler Hydro 5.00% Tri3 Craig1 4.80%
Table continued on next page.
Cherokee 2 12.40% LakeGeorge Hydro 5.00% Tri3 Craig2 4.80%
Cherokee 3 10.10% Manchief CT 5.00% Tri3 Craig3 3.00%
Cherokee 4 8.90% Maxwell Hydro 5.00% Tri3 LRS2 3.00%
Cherokee Diesel 9.40% Orodell Hydro 5.00% Tri3 LRS3 3.00%
Comanche 1 13.30% Ouray Hydro 5.00% TST Brighton 5.00%
Comanche 2 4.40% Palisade Hydro 3.00% TST Limon 5.00%
Comanche 3 6.30% Pawnee 1 8.40% UNC Greeley EXT 5.00%
Craig 1 4.80% PlainsEnd2 CC 1.50% Valmont 5 4.20%
Craig 2 4.80% Redlands Hydro 5.00% Valmont 6 9.90%
CT_129_A 1.00% Roberts T Hydro 5.00% WM Landfill Gas 5.00%

100 iterations of the model were run for year 2013. Monte Carlo draws determined if a
resource was on forced outage or not. In this case, the model was set up so that if a unit
was forced out in a week as a result of a Monte Carlo draw, the unit is assumed out for
the entire week. If a unit has an expected forced outage rate of, for example, 5%, then
the average outage hours for that unit over the 100 iterations is 5% of the time. However,
any individual iteration could have an outage rate for that iteration for the year of greater
or less than 5%. The Monte Carlo draws are designed such that over a large number of
random draws of unit outage, statistically one would expect the average hours of unit
being forced out during a year to be 5%. However, statistically it is possible that over 100
iterations the average outage rate is slightly above or below the 5% number.

2.7 ROCKY MOUNTAIN RESERVE GROUP SUPPORT

One key aspect of the analysis was to reflect the reliability support that PSCo receives
from neighboring electric systems. PSCo is a member of the Rocky Mountain Reserve
Group (RMRG) and thus has the right to call for support from the group under certain
qualifying contingency events. In accordance with the RMRG rules, PSCo must notify the
RMRG group and may request group support for outages of PSCo plants of 200 MW and
larger. For outage events of less than 200 MW, PSCo is not required to notify the RMRG
group and generally covers the event using its own reserves. For this analysis Ventyx
reflected RMRG support to PSCo for outages of plants of 200 MW or larger. Table 4
shows the RMRG Response Matrix and the contingency assistance provided to PSCo by
the RMRG Members. The RMRG support contained in Table 4 is based on the individual
members’ forecasts of load for year 2013.

The RMRG Response Matrix details the amount of contingency assistance provided to
PSCo at different megawatt levels of outages. The contingency assistance by RMRG
rules is available only for the hour of the event and the following hour for a total of 2
hours per outage event per month. If multiple units are out at the same time, the
contingency assistance is provided to the unit with largest capacity.

Based on the RMRG response matrix, Ventyx calculated the RMRG contingency
assistance provided by the participating surrounding utilities to PSCo for the PSCo units
above 200 MW. Table 5 summarizes the RMRG assistance available for each unit.

| 2-6
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Table 4
Rocky Mountain Reserve Group Response Matrix
B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14
RMRG responsibilty 784 759 734 709 684 659 634 609 584 559 534
EMERGENCY ASST * -> FOR PSCO 391 378 367 354 341 329 317 305 294 283 269
RRR Member response requirement
MEAN 0.011756 10 9 9 9 8 8 8 7 7 7 7
WMPA 0.002232 2 2 2 2 2 2 1 1 1 1 1
TRIS 0.108974 88 85 83 80 77 75 72 69 66 64 61
BHPL 0.035493 29 28 27 26 25 24 23 23 22 21 20
CSU 0.074984 61 59 57 55 53 51 49 48 46 44 42
FRPC 0.007988 6 6 6 6 6 5 5 5 5 5 4
WACM 0.062762 51 49 48 46 44 43 41 40 38 37 35
0 0 0 0 0 0 0 0 0 0 0
WALC 0.034187 28 27 26 25 24 23 23 22 21 20 19
0 0 0 0 0 0 0 0 0 0 0
PRPA 0.060373 49 47 46 44 43 41 40 38 37 35 34
WPEC 0.014808 12 12 11 11 10 10 10 9 9 9 8
PSCO 0.517850 419 406 393 380 367 354 341 328 315 302 289
BEPC 0.068591 55 54 52 50 49 47 45 43 42 40 38

GROUP 1.0000 810 784 760 734 708 683 658 633 609 585 558

WACM AGC offsets -297 -288 -282 -274 -265 -258 -249 -243 -235 -229 -220
After 15 minutes, change to: -247 -238 -232 -224 -215 -208 -199 -193 -185 -179 -170
PSCO AGC offsets -225 -215 -208 -199 -189 -181 -172 -165 -156 -149 -139
After 15 minutes, change to: -275 -265 -258 -249 -239 -231 -222 -215 -206 -199 -189

WACM AGC offsets -318 -312 -304 -295 -288 -279 -273 -265 -259 -250
After 10 minutes, change to: -288 -282 -274 -265 -258 -249 -243 -235 -229 -220
PSCO AGC offsets -245 -238 -229 -219 -211 -202 -195 -186 -179 -169
After 10 minutes, change to: -315 -308 -299 -289 -281 -272 -265 -256 -249 -239
WALC AGC offsets -28 -27 -26 -25 -24 -23 -23 -22 -21 -20 -19

B18 B19 B20 B21 B22 B23 B24 B25 B26 B27 B28 B29 B30 B31
RMRG responsibilty 434 409 384 359 334 309 284 259 234 209 184 159 134 109
EMERGENCY ASST * -> FOR 218 206 194 182 171 161 147 135 123 111 89 76 65 53
RRR Member response requirement
MEAN 0.011756 5 5 5 4 4 4 4 3 3 3 2 2 2 1
WMPA 0.002232 1 1 1 1 1 1 1 1 1 1 0 0 0 0
TRIS 0.108974 49 47 44 41 39 36 33 30 28 25 20 17 15 12
BHPL 0.035493 16 15 14 13 13 12 11 10 9 8 7 6 5 4
CSU 0.074984 34 32 30 28 27 25 23 21 19 17 14 12 10 8
FRPC 0.007988 4 3 3 3 3 3 2 2 2 2 1 1 1 1
WACM 0.062762 28 27 25 24 22 21 19 18 16 14 12 10 8 7
0.000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0
WALC 0.034187 16 15 14 13 12 11 10 10 9 8 6 5 5 4
0.000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0
PRPA 0.060373 27 26 24 23 21 20 18 17 15 14 11 10 8 7
WPEC 0.014808 7 6 6 6 5 5 5 4 4 3 3 2 2 2
PSCO 0.517850 235 222 209 196 195 178 157 144 132 119 95 82 69 56
BEPC 0.068591 31 29 28 26 24 23 21 19 17 16 13 11 9 7

GROUP 1.0000 453 428 403 378 366 339 304 279 255 230 184 158 134 109

WACM AGC offsets -137 -130 -122 -114 -109 -102 -93 -85 -78 -70 -56 -48 -41 -33

PSCO AGC offsets -153 -145 -136 -127 -121 -113 -103 -95 -87 -78 -62 -53 -46 -37

WALC AGC offsets -16 -15 -14 -13 -12 -11 -10 -10 -9 -8 -6 -5 -5 -4

| 2-7
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Table 5
RMRG Contingency Assistance for PSCo Units Greater than 200 MW
RMRG Contingency
PSCo Units > 200 MW Capacity MW Assistance MW
(shadow station)
Comanche3 784 391
RockyMontCC2 601 301
Pawnee1 505 253
Cherokee4 352 179
Comanche2 335 171
Comanche1 325 167
RockyMontCC1 259 135
FSV2 252 132
FSV3 230 121
FSV1 226 119
Plainsend2 224 118

2.7.1 Modeling the RMRG Support

For each PSCo units ≥ 200 MW, Table 5, Ventyx modeled a corresponding RMRG
support unit called a shadow station. The size of each RMRG shadow station was
determined by the actual plant size and the corresponding assistance available as
reported in Table 5.

To reflect the fact that each RMRG shadow station may only be called upon during its
parent station’s outage event, PaR Rules of Existence (Rule Groups) modeling was
utilized. Rule Group modeling included assigning each of the RMRG Shadow Units to a
Rule that tells PaR the RMRG unit can exist to help serve load only if the parent station is
on outage.

Since only the largest station during overlapping outages receives the RMRG
contingency assistance, a Rule Group hierarchy of RMRG shadow stations was
implemented to ensure only the largest contingency was called upon.

Under the terms of the RMRG, pool members are required to provide contingency
assistance to PSCo, if requested, for up to two hours for each qualifying contingency
event. To reflect this real-life constraint, Ventyx modeled the RMRG Shadow Units as
“limited energy” stations. For each of the RMRG shadow stations Ventyx input a weekly
energy limit equal to 2 times the MW rating of the shadow unit (i.e., 2-hours of full load
operation). Once the RMRG unit is been called upon in the modeling, it will not be
available again for contingency assistance until the next outage. As a limited energy
station, PaR will attempt to choose the best hours to run limited energy RMRG shadow
station based on dispatch economics. For instance, if a 300 MW PSCo plant is tripped
off-line at 12 am and is forced out for the week, PaR will not immediately activate the
RMRG shadow station but rather will attempt to save the limited energy from the shadow
unit for peak hours or for hours where energy not served exists. In other words, because
the RMRG shadow stations are modeled with such a high cost of operating (i.e. just
below the cost of ENS), the units will only be run when there would otherwise by ENS,
and the units will only run for two hours following each outage. This methodology allowed
the RMRG unit to be available to contribute generation assistance to PSCo after the

| 2-8
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

station goes on forced outage and only during an ENS event. This limited energy
methodology meets the two hour limitation of the RMRG but has a shortcoming in that it
provides the two hours of generation support during the highest marginal energy cost
hours. Given that high marginal energy costs typically occur during hours when system
load is at it’s highest, this means that the reliability contribution provided by the two hours
of RMRG support is likely somewhat overstated in the PaR modeling To understand the
potential magnitude by which the RMRG support might be overstated, a sensitivity was
performed in which the RMRG units were excluded from the analysis. The results of this
sensitivity showed the generation support provided by the RMRG acts to reduce the
Planning Reserve Margin from approximately 17.8% to 16.3% or 1.5%. From this we can
see that the limited energy methodology used to represent the RMRG support is likely to
be a small factor in the overall reserve margin level required for the system (i.e., it is
probably a small part of the 1.5% total impact of the RMRG support).

2.8 TRANSMISSION LIFELINE - NON PSCO IMPORTS

PSCo is interconnected with the Western Interconnect (WECC reliability council area)
and expects that in an emergency situation it can utilize these interconnections to import
additional power supplies into its system. The exact quantity of additional power supply is
dependent on the availability of unused transmission capacity. PSCo estimates that it will
have access to roughly 200 MW plus or minus 50 MW of unused transmission capacity
during peak load periods. The reliability benefit of this transmission import capability was
included in this analysis through the representation of an additional 200 MW of imports
with Monte Carlo draws around plus or minus 50 MW.

Model runs were also performed without this 200 MW of import capability. These runs
allowed PSCo to isolate the contribution that this 200 MW of import capability provides to
the system through a reduction in the required planning reserve. As reported in Section 3
below, from this sensitivity run it was found that the existence the 200 MW Transmission
LIFELINE allows reducing the Planning Reserve Margin from approximately 19.2% to
16.3% while maintaining the LOLP at 1-day-in-10-years.

2.9 USING A GENERIC GAS TURBINE AS A PROXY FOR INCREASING PLANNING

RESERVES AT THE MARGIN

In order to perform this study, it was necessary to run the stochastic analysis at several
different levels of planning reserve. For example if additional resources need to be
added to the model in order to move the Planning Reserve Margin level from 10 percent
to 12 percent and so on. The resource used to incrementally increase Planning Reserve
Margin needs to be (a) highly reliable as a supply source and (b) relatively low cost to
acquire since it will likely used at a very low capacity factor. While there are numerous
supply technologies available for increasing supply, the reasonable supply unit to use for
this purpose is a simple cycle GT.

| 2-9
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

2.10 DETERMINATION OF LOLP

The LOLP analysis methodology Ventyx applied in this study is a marked improvement
over traditional methods for determining LOLP. Where, in the past, company’s often
computed an annual LOLP index as the summation of daily probabilities (often termed
the “daily risks”) over the entire year being studied, Ventyx computes LOLP based on a
stochastic production cost model simulation where all relevant factors and uncertainties
are included in the simulation. The analysis predicts both the probability of not serving a
specific amount of load, and in addition provides insights into the dimension and amount
of energy that would not be served—referred to as unserved energy or expected un-
served energy (EUE). The Ventyx LOLP methodology calculates LOLP for each hour
where the LOLP is the probability that available generation capacity in a given hour is
less than the system load. The primary measurement used in accessing resource
adequacy in this analysis is Loss of Load Hours (LOLH), which is typically used in the
energy industry. Generally, if a utility’s loss of load hours is not greater than or equals 1-
day-in-10-years (or 2.4 hours in 1 year), it is seen as a reliable system. Unserved Energy
(aka Energy Not Served…ENS) results in the model if on a particular hour the model is
unable to find sufficient supply to meet the load plus the required operating reserve
margin. If that happens on an hour, then this is counted as one LOLH. For LOLH
counting purposes, there is a single LOLH if on an hour the load is not met. The counting
is the same if the unserved load is 1 MW or if it is, for example, 200 MW. Given multiple
iterations of the study year (with different Monte Carlo draws on loads and unit forced
outages, etc), the metric used for this LOLP study is the average number of hours of
LOLH over the 100 iterations. So if there are 99 iterations with zero LOLH and one
iteration with 100 LOLH, then the expected (average) LOLH for the 100 iterations for this
year is 1 LOLH. As indicated above, and average LOLH of 2.4 hours in the 1 year
analysis is considered to be 24 LOLH hours in 10 years or 1 day in ten years.

For purposes of this study, Ventyx analysis looks for that Planning Reserve Margin level
that will provide a 1-day-in-10-year LOLP.

2.10.1 Calculating The Planning Reserve Margin

A number of questions arise when the objective is calculating an accurate Planning
Reserve Margin for a system. The common method of calculating Planning Reserve
Margin is represented by the following equation:

[(Resources - Peak Load ]

(Peak Load )

Peak Load: Peak load is generally the needle peak load of the control area. In this
study, where PSCo is modeled as a single zone, the peak hour for the entire system
occurs in July.

Resources: The peak capacities of thermal and hydro stations that are in PSCo are
included in the calculation except for Cabin Creek Pumped Storage which is counted at
210 MW. Wind capacity is counted at 12.5% of nameplate rating. Interruptible loads and
demand side management programs are included as resources but for load and resource
balance purposes, they are subtracted from the peak load.

| 2-10
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Table 6
2013 PSCo Expected Reserve Margin
2013 L&R MW LOLH
Peak Load 50th percentile 7310
interruptible loads -401
Firm Peak Obligation 6909
Net Dependable Capacity from Table I above not including
7410
CT 129A and not including FSV CT
NET Planning Reserve Margin in 2013 without CT 129A 7.3%
Needed Operating Reserves 5.7%
Effective Starting Point Planning Reserve Margin 13.0% 69.8
Recommended Reserve Margin -- 1 day in 10 years 16.3% 24.0

The conclusion of this LOLH is that a 16.3% PRM is needed to provide a 1 day in 10 year
LOLP. This level is determined by performing analysis that does not interrupt load until
the operating reserve drops below zero.

Table 7 in section 2.11 below reflects the loads and resources in the year 2013 for PSCo
currently planned, but without the assumed generic CT 129A and without the new FSV
CT units. This was the starting point for the LOLP analysis in this report. The generic CT
129A and FSV CT units were removed to assure that the starting analysis results in a
LOLP that was greater than one-day-in-10 years. That staring point as indicated above
resulted in a LOLH of 69.8 hours. A one-day-in-10 years would have an LOLH of 24.0
hours. To achieve that, Ventyx then started adding gas turbines until it found the level of
Planning Reserve Margin that resulted in a LOLP of one-day-in-10 years.

2.11 ANALYSIS STARTING POINT OPERATING RESERVE MARGIN

For year 2013, PSCo estimates it will be required to maintain approximately 419 MW of
operating reserves as its portion of the RMRG reserve obligation. If operating reserves
fall below 419 MW, PSCo would likely curtail load if it cannot arrange for additional power
supplies. The PaR model used to perform this LOLP analysis, however, is not capable of
curtailing load (i.e., registering unserved energy) and enforcing an operating reserve
requirement. The model will only register unserved energy events in hours where the
sum of all generation resources operating at their full capability is less than the load on
the system and there is energy not served.

To account for this PaR model limitation, it is necessary to add “operating reserves” to
the “planning reserve” level included in the model run that produces a 1-day-in-10-year
level of reliability. Based on the 2013 peak load forecast of 7,310 MW, the 419 MW
operating reserve requirement represents 5.73% (419 MW/7,310 MW = 0.0573) that
must be added to the model results. As summarized in Section 3 below, the starting point
for the PSCo Planning Reserve Margin analysis is a 13.0% starting reserve level that
resulted in an LOLH of 69.8, which is 2.9-days-in-10-years (69.8 / 24 hours = 2.9 days)
as shown in the table above. To determine an expected LOLP of 1-day-in-10-year LOLP,
Ventyx added 210 MW of generic CT generation and found a LOLP of slightly higher than
1-day-in-10-years, or 26.9 hours. This level equates to a Planning Reserve Margin of
16%. Ventyx then added another 60 MW of CT capacity which is a Planning Reserve
Margin of 17% and found a LOLP of less than 1-day-in-10-years, or 17.7 hours.

| 2-11
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

Interpolating between these two LOLP values determines a Planning Reserve Margin of
16.3% equates to a target 1-day-in-10-years LOLH of 24.0 hours. This interpolation to a
one-day-in-10-years indicates PSCo’s Planning Reserve Margin should be 16.3%.

| 2-12
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

3 ANALYSIS RESULTS

The goal of this LOLP analysis was to determine the Planning Reserve Margin for the
PSCo system that would achieve an LOLP of 1 day in 10 years or, an Energy Not Served
(LOLH) of 24 hours in 10 years (or 2.4 hours in one year). Table 7 contains a summary
of the relationship between reserve margin and LOLH, both with and without 200 MW of
transmission lifelines (i.e., transmission capacity held for use in accessing additional
power supplies on short notice). All reserve margin values in Table 7 include the effects
of operating reserve requirements.

Table 7
LOLP Results Summary
Reserve Margin
Reserve Margin
(with 200 MW
(no transmission LOLH (hrs in 10 Years)
transmission
lifelines)
lifeline)
16% 13% 69.8
18% 15% 39.6
19% 16% 26.9
20% 17% 17.7
22% 19% 7.9
24% 21% 1.8

Figure 3 below is an illustration of the LOLH / Energy Not Served values provided in
Table 7 as a function of reserve margin level. By interpolation a reserve margin of 16.3%
(with 200 MW transmission lifeline) yields 1-day in 10 years level of LOLH

Figure 3
Expected Hours of Energy Not Served
80

70

60
Hrs of ENS in 10 years

50
Planning Reserve Margin for 1 day in 10
years is 16.32%
40

30

20

10

0
13% 15% 16% 17% 19% 21%
Planning Reserve Margin

| A-1
 Attachment 2.10-1
Analysis of LOLP at various Planning Reserve Margins

| 2