Transmission Planning Process Discussion Joint PLWG / CMWG Meeting Presented by: Luminant Energy

Transmission Planning Process DiscussionJoint PLWG / CMWG MeetingPresented by: Luminant Energy June 14, 2011

Outline • Executive Summary • Background • Detailed Discussion • Rio Grande Valley Import • DFW Imports Southeast of Dallas • Long-Term Autotransformer Outage Risk • Next Steps 1

Executive Summary • ERCOT’s planning process is superior to most and continues to improve • The RPG process open to all interested stakeholders to develop & submit projects • ERCOT holds regular open meetings to discuss ongoing issues with the market • The ERCOT RPG process reviews a large number of project proposals • This culture of continuous improvement requires periodic review to see if amendments to the current planning criteria / process may be appropriate • Valley Import Situation in 2010/2011+ ($76 million) in the first 120 days after Nodal Go-Live1 • DFW Imports SE of Dallas – congestion observed in 2010 & aggravated by Valley mothballing • Long Term Auto Forced Outages – multiple outages (e.g. Bellaire, Cedar Hill, etc…) lasted several months, which could have caused market impacts if occurring over summer peak • TAC’s May 5th, 2011 directive to the PLWG / CMWG to develop processes that close the gaps between operations and planning provides for reevaluation of: • Specific ways to clarify the language of the existing planning criteria to ensure more consistent application by all parties going forward • Specific criteria / process changes, both reliability and economic, that are needed to better align transmission planning with operational realities We are hopeful that PLWG discussion of the examples discussed in this presentation will lead to a tighter integration of operational realities into the planning criteria / processes in order to mitigate much of the SCED irresolvable congestion outcomes. 1) Calculated from Shadow price congestion of the Valley Import interface, Lon Hill – NE Edinburg 345 kV Line, and Rio Hondo 345/138 kV auto. According to John Dumas’s May 5, 2011 TAC presentation, there were an additional 43+ hours of manual management of the Valley Import constraint that would have pushed this cost to $300+ million. 2

Executive Summary • Aspects of the planning criteria that need to be clarified / improved are: • Load Variability – lack of clarity regarding reasonable load variability in studies • Thermal Generation Availability – need a common way to recognize large area expected generation capacity unavailability • Ancillary Service – planning process assumes real time co-optimization of energy and A/S, to such a degree that outcomes of no RRS, URS, and NSRS capacity being carried in major load pockets (e.g. greater Houston & DFW) can occur1 • Transmission Maintenance Outages – need a consistent method of protecting the transmission and generation maintenance windows. • Wind – need a clear definition of study requirements for high / low wind • Dynamic Line Ratings – experience below nominal ratings when temperatures associated with above normal weather load are studied • Long Term Auto Outages – similar impact to system as loss of a large unit • Operational Safety Margin – Operations has found the need to utilize a safety margin in some cases (W-N, Valley import) that isn’t being utilized in planning Planning criteria / process guidance is needed with regard to these items to insure that consistent treatment / application of these issues occurs in all planning studies. 1) Potentially in conflict with the location considerations of Operation Guides Section 2.1 (1) (h), as well as, may be impractical to run multiple SASMs to accommodate. 3

Planning Criteria Review • Nodal Operating Guides - Section 5.3 (2) • The contingency studies will be performed for: • Reasonable variations of Load level, • Generation schedules, • Planned transmission line Maintenance Outages, and • Anticipated power transfers. • At a minimum, this should include projected Loads for the upcoming summer and winter seasons and a five-year planning horizon. • The TSPs involved should plan to resolve any unacceptable study results through the provision of: • Transmission Facilities, • The temporary alteration of operating procedures (i.e., RAPs), • Temporary Special Protection Systems (SPSs), or • Other means as appropriate. For over a decade1, ERCOT has relied on this language to make sure that studies reflected reasonable variations of system conditions and resolved unacceptable outcomes via the provision of RAPs, SPSs, and Transmission Facilities 5 1) This exact same Planning language has been utilized since at least 2/20/2001 and is believed to have been in use several years before that.

Planning Criteria Review • Nodal Operating Guides - Section 1 (Credible Single Contingency) • A single facility, comprised of transmission line, auto transformer, or other associated pieces of equipment… • The Forced Outage of a DCKT in excess of 0.5 miles in length… • Any Generation Resource: • A combined-cycle facility shall be considered a single Generation Resource; or • Each unit of a combined-cycle facility will be considered a single Generation Resource if the combustion turbine and the steam turbine can operate separately, as stated in the Resource Registration on the MIS Public Area Contingency studies are to be performed for single branch outages, DCKT lines in excess of 0.5 miles, and Generation Resource outages. 6

Planning Criteria Review • Nodal Operating Guides - Section 1 (Credible Single Contingency) • With any single Generation Resource unavailable, and with any other generation preemptively redispatched, the contingency loss of a single Transmission Facility… with all other facilities normal should not cause the following: • Cascading or uncontrolled Outages; • Instability of Generation Resources at multiple plant locations; or • Interruption of service to firm Demand or generation other than that isolated by the transmission facility, following the execution of all automatic operating actions such as relaying and SPSs. • Furthermore, the loss should result in no damage to or failure of equipment and, following the execution of specific non-automatic predefined operator-directed actions (i.e., RAPs) such as generation schedule changes or curtailment of interruptible Load, should not result in applicable voltage or thermal ratings being exceeded. Contingency studies are expected to consider generation redispatch, where possible, and ensure that no damage to equipment occurs following the execution of RAPs and SPSs while continuing to serve demand and generation 7

Detailed Discussion • Three specific project situations from across ERCOT have been chosen to facilitate a deeper discussion of the issues outline above: • Rio Grande Valley Import – Since Nodal Go-Live (~6 month period): • Has experienced severe congestion, irresolvable constraints, and firm load shed • DFW Imports Southeast of Dallas (Trinidad Area) • Experienced frequent congestion in summer 2010 • Studies indicate situation further stressed by mothballing of Valley generation • Long-term Autotransformer Outage Risk Review • Outages can last for several months, sometimes 12 months • Similar market impact in many cases to loss of a large generating unit • These recent & ongoing issues aren’t meant to single anyone/thing out, rather to facilitate a deeper discussion of how to avoid them going forward • Combination of existing / old studies utilized to illustrate key observations What is common about all three of these specific situations is that the current application of the planning criteria / processes in some cases has hamstrung ERCOT and/or the TDSPs from developing timely transmission projects to mitigate this risk 9

Rio Grande Valley - Overview • In the six months since Nodal Go-Live: • Experienced over ~$75 million in congestion1 • Has experienced 6.5 hours of SCED irresolvable congestion (reliability issue)2 • Transmission Watches Issued on 2/3/11 (17:35) and 2/10/11 (06:10)3 • Transmission Emergency Notice on 2/3/11 (22:29)3,4 • Initial situation had an entire CCCT train out of service for planned maintenance • Temperatures were ~28 degrees • At 21:47 approximately 486 MW of CCCT generation forced out • Voltage dropped to 0.91 pu (125 kV) • ~442 MW of firm load was shed to recover voltage to acceptable levels • All load was restored approximately 25 ½ hours later (2/4/11 at 23:28) • Only ~$20 million of the~ $75 million since Nodal go-live occurred during this Emergency • Excluding the period of the Transmission Emergency, it is the combination of a CCCT train being out-service for a planned outage and N-1 transmission security requirements that are associated with all this congestion. It is alarming that the current planning process indicates that no transmission upgrades are needed to remedy this situation. It is important to identify what is causing this gap between operational outcomes and planning studies 1) Calculated from Shadow price congestion of the Valley Import interface, Lon Hill – NE Edinburg 345 kV Line, and Rio Hondo 345/138 kV auto; 2) Calculated as when constraint shadow price hit the shadow price cap; 3) Based on February 2011 ERCOT Operations Report to ROS; 4) Temperatures from NOAA; 5) The December 2010 ERCOT Constraint and Needs Report indicates no expected congestion associated with Rio Grande Valley Imports, while the 2010 Five-Year Plan (issued March 2011) doesn’t recommend any projects to address this severe congestion either. 11

Rio Grande Valley - Overview Valley Interface 1. North Edinburg - Lon Hill 345 kV 2. Ajo – Rio Hondo 345 kV 3. Raymond 2 – MV Yutt 138 kV 4. North Edinburg - Rachal 138 kV 5. Roma Switch – Falcon Switch 138 kV The Rio Grande Valley is connected to the rest of ERCOT via two 345 kV lines and three 138 kV lines, contains 1,712 MW of generation located at four plants, and set an all time peak load of 2,734 MW on February 3, 2011 12

Rio Grande Valley – Summer Load Variability Implies ~10% annual load growth Scatter plat provided by AEP in presentation titled “Lower Rio Grande Valley (LRGV) Import” at March 2011 RPG Meeting. Approximate trend lines developed heuristically by Luminant The load growth in the Rio Grande Valley appears to be quite strong, year over year (~10% at 100o F). Materially different temperature outcomes experienced in 2009 (109o F) vs. 2010 (101o F). 13

Rio Grande Valley – Winter Load Variability Scatter plat provided by AEP in presentation titled “Lower Rio Grande Valley (LRGV) Import” at March 2011 RPG Meeting. Winter load growth is less clear from this graph. The peak 2011 load shown is potentially understated if it occurred during the firm load shed event on 2/3/2011. Materially different temperature outcomes experienced in 2010 (39o F) vs. 2011 (28o F)1. 14 1) Note the data shown on this slide excludes weekends and holidays. Ironically, the same 28 degree annual minimum temperature from 2011 also occurred in 2010, but on a Saturday.

Rio Grande Valley – Summer Load Variability 2010 peak hour temp were extremely mild. Only 1 year in the last 30 has had a lower peak hour temp (2007). Analysis conducted by Luminant Energy utilizing 30 years of NOAA weather data Across the prior 30 year period, the average (1 in 2 year outcome) summer maximum hour temperature for the year was 103.6o F. The 1 in 10 year peak temp was 107.1o F, while the 1 in 20 year peak temp was 108.6o F1. 15 1) Refer to Appendix 2 for summary of NOAA statistical details referenced above for McAllen.

Rio Grande Valley – Winter Load Variability While the 28o F temperature outcome on 2/3/2011 was cold, it only represents a 1 in 5 year minimum peak hour temperature for the McAllen area Analysis conducted by Luminant Energy utilizing 30 years of NOAA weather data Across the prior 31 year period, the average (1 in 2 year outcome) winter minimum peak hour temperature for the year was 32.1o F. The 1 in 10 year peak temp was 26.0o F, while the 1 in 20 year peak temp was 22.5o F1. 16 1) Refer to Appendix 2 for summary of NOAA statistical details referenced above for McAllen.

Rio Grande Valley – Load Variability Assessment • When combining the findings from the prior four slides, and compare them to the table showing the loads being used in planning it becomes clear: • That expected (1 in 2) loads are materially higher than what is being planned Table provided by ERCOT in presentation titled “Lower Rio Grande Valley (LRGV) Import” at March 2011 RPG Meeting. • When reasonable variations in max and min hour seasonal temperatures are applied to the prior scatter plots and even a 5%1 annual load growth is used: • 2011 Sum Peak – 1 in 2 year peak load of 2,415 MW, 1 in 10 year 2,520 MW • 2011 Winter Peak – 1 in 2 year peak load of 2,550 MW, 1 in 10 year >3,000 MW According to the scatter plots, winter hourly temps must stay at or above 40o F for Winter Peak loads to not exceed the 1,800 MW levels used in the planning cases. NOAA weather data indicates, that there hasn’t been a single year in the last 30 where minimum hour temps for the year haven’t been below this level. 17 1) Note that on Slide 14, it appears as if the 2010 over 2009 load growth was ~10%

Rio Grande Valley – Operational Safety Margin ERCOT Planning verbalized during discussion at the March 2011 RPG meeting that the thermal limit for Valley Import is ~1,400 MW However, ERCOT Operations normally utilizes a safety margin of ~300 MW and sets the operational limit at 1,100 MW unless conditions drive it lower In the 90-day period from 1/1/2011 through 3/31/2011, the Valley Import Limit has been set at or below 1,100 MW in either the DAM or RT during 70 days. This ~300 MW operational safety margin appears to be a material driver of why congestion occurs in operations that is not seen in planning. 18

Rio Grande Valley – Maintenance Outages • Discussion by AEP at the March 11, 2011 RPG meeting during the ERCOT presentation titled “Lower Rio Grande Valley (LRGV) Import”: • “A major operational issue is the difficulty to grant clearances for both transmission and generation maintenance outage in the LRGV.” • “There is a limited time windows to grant maintenance outages in the LRGV and any extreme weather during this period could require load shed.” • “There is a potential for severe weather (hurricanes) that could affect both the existing 345 kV lines sourced from the Corpus Christi area and significantly reduce the load serving capability in the LRGV.” • Nodal Operating Guide 5.3 (2) already states that contingency test must be performed for planned transmission line maintenance outages • However, this criteria doesn’t currently explicitly address the combination of area planned transmission and generation outages AEP has uncovered a very real issue, in that the current planning criteria doesn’t explicitly cover the need to insure that a combination of both necessary planned generation and transmission outages can be accommodated during the spring and fall maintenance windows 19

DFW Imports SE of Dallas - Overview • Between August 2nd and August 20th, 2010 • RPRS congestion was experience by ERCOT Operations 16 out of 18 days in a row bringing on between 1 and 16 generating units to resolve this congestion Source: ERCOT – Replacement Reserves Service – Daily Reports as posted on the ERCOT Website • ERCOT reported cost in the System Planning Report Source: September 2010 ERCOT System Planning Report for August 2010 Load growth in DFW revealed congestion here in 2010 that is expected to be worse in 2011 due to the retirements and mothballing at Permian and Valley. However, the 2010 Five-Year Plan (issued March 2010) doesn’t mention or recommend any projects to address this congestion 21

DFW Imports SE of Dallas - Overview • Constraining Elements • Richland – Trinidad 345 kV (both) ckts • Trinidad – Watermill 345 kV line • Trinidad – TriCorner 345 kV line • TriCorner – Seagoville 345 kV line • Big Brown – Venus 345 kV line The post summer 2010, mothballing / retirement of Permian and Valley increase the loading on these constraints. However, the current planning process indicates no transmission upgrades are needed in 2011.1 22 The 2010 Five-Year Plan (issued March 2011) doesn’t mention or recommend any projects to address this congestion.

DFW Imports SE of Dallas – Gen Availability Source: 2009 State of the Market Report (SOMR) for the ERCOT Wholesale Electricity Markets, page 56. A summary of 2003 – 2009 SOMR data is shown in Appendix 3. It is not possible to reflect ‘anticipated power transfers’ and expected ‘generation schedules’ without recognizing historical generator unavailability.1 Currently, reliability studies only analyze the loss of the single largest unit, while economic studies completely ignore generation outages all together. 23 1) The 2009 summer (June – Sept.) generator unavailability averaged ~13%. A review of a longer period of time (2003 –2009) suggests a range between 8% and 15%

Data from highest 5 peak load days in July and August 2010, looking at 3 peak hours (1600-1800). The QSE level data available makes it unclear in which zone this deployment took place DFW Imports SE of Dallas – A/S Consideration • During high load periods, there is a higher risk of congestion and reliance on generators that have a portion of their capacity committed to the ancillary services market • To demonstrate this, the most recent data available for the high load conditions of July – August 2010) has been summarized below1 • A review of historical generation schedules has shown that significant amounts of generation capacity is tied up carrying RRS, URS, and NSRS All of the reliability and economic studies performed currently assume a level of real time co-optimization of energy and A/S, to such a degree that outcomes of no RRS, URS, and NSRS capacity being carried in load pockets such as DFW and everything electrically north of Trinidad in the study mentioned above.2 24 1) Detailed results of this analysis are provided in Appendix 4; 2) Potentially in conflict with the location considerations of Operation Guides Section 2.1 (1) (h), as well as, may be impractical to run multiple SASMs accommodate.

DFW Imports SE of Dallas – Dynamic Line Ratings • A number of the major TDSPs have implemented dynamic line ratings over the past 5 years and ERCOT systems have been designed to reflect them • Has the effect of freeing up more of the transmission capacity for market use much of the year • However, the above normal temperatures (e.g. 1 in 10 or 1 in 20) that drive above normal loads, are 100% correlated with below nominal dynamic line ratings • The table below illustrates the relative temperature impact of dynamic line ratings on an Oncor 2-795 kcmil 345 kV line • For the DFW area expected peak hour annual temperatures, a: • 1 in 10 year value is 107.4 oF, which equates to ~a 1.9% below nominal line rating • 1 in 20 year value is 109.6 oF, which equates to ~a 3.8% below nominal line rating Given the widespread application of dynamic line ratings that has occurred since the planning criteria was written and the 100% correlation of above normal temperatures causing both higher than normal loads and lower than nominal dynamic line ratings, these two effects should be studied simultaneously 25

DFW Imports SE of Dallas – Wind Generation • The chart below depicts the top and bottom 4,000 intervals (1,000 hours) of load in the combined North and West Congestion Zones for 23 months (1/1/2009 – 11/30/2010) Low Load High Load SSWG 11Sum1 Case has 796 MW West and North Zone wind Given the high degree of variability of wind from hour to hour, the planning requirement to study ‘anticipated power transfers’ should require studies at both very high and very low wind generation conditions. Clearer definitions of study requirements are needed in the planning criteria to insure consistent study application. 26

DFW Imports SE of Dallas – Net Impact A Without varying system conditions to the degree required by the criteria, the contingency loading on this line can be as low as 77%. B B When considering the full set of impacts experienced on the line based during operations, the loading can easily be in the range of 110% - 120% A As a result of the lack of explanation in Section 5.3 (2) of the Operating Guides on how to apply ‘reasonable load variations, generation schedules…, and anticipated power transfers’; most planning studies simply ignore them. This is a key factor on why many constraints become quite impactful in actual operations long before they are identified for upgraded through the planning process. 27

Long-Term Autotransformer Outage Risk • Given their large size and potential long-term outage characteristics, autotransformer outages require careful planning consideration • Forced outages of large autotransformers can be equally impactful to the reliability and economics of the market as a generation plant • Recent history has provided several notable examples: • Bellaire 345/138 kV auto – long term outage in spring 2010 • Cedar Hill 345/138 kV auto – long term outage in summer 2010 • Paris 345/138 kV auto – long term outage in winter 2010 / 2011 • In each of the above examples, the potential for market impacts existed: • Bellaire & Cedar Hill – significant congestion impacts in the market • Paris – Monticello mining load was tripped by SPS #25, as well as, transmission related backdown at Monticello plant also occurred • Historically, some TDSPs (e.g. Oncor) planned the system to accommodate one auto xfmr as unavailable and maintain N-1 security Recent operational experience, pending NERC TPL standards, as well as, some historical TDSP planning criteria point out that long-term outages of auto transformers can be just as impactful to the system as a long-term generator outage and perhaps should be treated equivalently1 29 1) NERC Standard TPL-001-2, which is currently being targeted for implementation in 3Q11, addresses this same concern in Section R2 2.1.5

Conclusions and Next Steps • Conclusions • The enhancements proposed within, will allow the PLWG / CMWG to quickly respond to the TAC directive of finding ways of tightening up the planning process to mitigate continue SCED irresolvable constraint outcomes • Enhanced clarity of the study assumptions and performance requirements sought by the planning criteria / process, should result in a better match of planning studies to real time operational realities • These improvements are likely needed in both the Operational Planning and Long-Term Transmission planning studies • Next Steps • PLWG drafts a PGRR to incorporate revisions to the planning criteria that would codify these amendmants • Recommend PGRR to ROS and TAC at their Jul. 14th and Aug. 4th meetings • Hold joint PLWG / OWG meeting to review this material and the associated companion OGRR associated with Operational Planning 31

Appendix 32

Appendix 1) John Dumas’ TAC presentation “Valley Import Constraint” May 5, 2011 Congestion Cost Violated & overridden with 2 step limits $200 & $2000 MSP – One constraint at limit + margin with Max Shadow Price at $200 and another at limit with Max Shadow Price at $200 0

Appendix 2) Statistical Summary of NOAA Temp Data for McAllen 34

Appendix 3) Potomac Economics 2003 – 2009 annual report related to generator Short-Term Outages and Deratings Generating capacity that has historically been unavailable during summer months ranges from 7.9% to 14.8%, with an average of 10.2% Above: Source: 2009 State of the Market Report for the ERCOT Wholesale Electricity Markets, page 56. 35

Appendix 4) Details of Generation Capacity Committed to Ancillary Services Table Continued on next page 36

Appendix 4) Details of Generation Capacity Committed to Ancillary Services (continued) The generation capacity committed to ancillary services was approximated by collecting data for the five highest peak load days in July and August 2010, for Hours Ending 16-18. Additional sources include the Resource Plans, QSE AS Schedules, ERCOT dispatch instructions for AS and BES, and MCPE 37

Transmission Planning Process Discussion Joint PLWG / CMWG Meeting Presented by: Luminant Energy