Inactive State • What is its purpose? • When is it used? Examples • If there were no equipment issues going into an Inactive state, why are repairs required coming out of an inactive state? • IEEE 762 defines an inactive state where the unit is unavailable. Unavailable is an outage state with no equipment issues? • GADS adds “After some repairs” should this be “After some maintenance”? Repair implies an outage. • GADS Inactive Reserve (IA) states that the unit must be in Reserve Shutdown (RS) which is an Active State and available in a moments notice. How can a unit be available at a moments notice if some repairs are required? • GADS Mothball (MB) indicates that the units may have experienced mechanical problems and must be in outage (PO,FO,MO) for 60 days before going into MB. Does this negate the “not equipment problem” in the opening statement. • Can MB be use without a waiting period if there are no equipment problems? • Who determines which state (IA or MB)? • When does an investigation end and a repair begin? How does IEEE 760 define Inactive State?

Inactive State is called “Deactivated Shutdown” in IEEE 762 and is defined as “The state in which a group or individual WTG is unavailable for service for an extended period of time for reasons not related to the equipment.” GADS interprets this to include the following.

Inactive Reserve (IR) – IR is defined by IEEE 762 and GADS as “The State in which a group is unavailable for service but can be brought back into service after some repairs in a relatively short duration of time, typically measured in days.”

Mothballed (MB) – MB is defined by IEEE 762 and GADS as “The State in which a group or individual WTG is unavailable for service but can be brought back into service after some repairs with appropriate amount of notification, typically weeks or months.”

If they were not broken when they went into an Inactive State, why do they require repairs? How does GADS define Inactive State

GADS added “after some repairs” and defines this statement to mean that some action may be needed to prepare the group for service because it had been sitting idle for a period of time and some equipment parts have deteriorated or need replacing before the group can be operated.

Repairs imply an Outage state. Outage state implies an Active State that is Unavailable.

Should “Maintenance” be used instead of “Repairs”? How does GADS define Inactive Reserve?

Inactive Reserve: The group should be operable at the time the IR begins. This does not include groups that may be idle because of a failure and dispatch did not call for operation. A group that is not operable or is not capable of operation at a moment’s notice should be on a forced, maintenance, or planned outage and remain on that outage until the proper repairs are completed and the group is able to operate. The group must be on RS (Reserve Shutdown) a minimum of 60 days before it can move to IR status.

IR is defined as an unavailable condition. RS is an available condition.

Should “Maintenance” be used instead of “Repairs”? How. does GADS define Mothball?

Mothball (MB): GADS added “after some repairs” and defines this statement to mean that some action may be needed to prepare the group for service because it had been sitting idle for a period of time and some equipment parts may have deteriorated or need replacing before the group can be operated. The group may have also experienced a series of mechanical problems for which management may wish to wait for a period of time to determine if the group should be repaired or retired.

A group that is not operable or is not capable of operation at a moment’s notice must be on a forced, maintenance, or planned outage and remain on that outage for at least 60 days before it can be moved to the MB state.

1. The lead in section indicates that the unit cannot be down for equipment problems. The above in red seems to be in conflict with this condition. 2. If the unit is not in a outage state, can it be placed into MB without a 60 day waiting period?

How can a unit be MB and Outage at the same time? Is outage in conflict with “not an equipment problem”? How. does GADS define Mothball?

Mothball (MB) Continued: If repairs are being made on the group in order to restore the group to operating status before the 60‐day period expires, then the outage must remain a forced, maintenance, or planned outage and not MB.

What is defined as repairs? (Evaluating cost, ordering parts, investigating damage, evaluating equipment availability and lead times, and etc.

If group repairs for restoring the group to operation are made after the 60‐day period then the first 60 days must be a forced, maintenance, or planned outage and the time after the 60 days, including the repair time on the group up to operation, shall be the MB event. Agenda Item 9a.i.b Planning Committee Meeting December 13-14, 2011

Generating Availability Data System Working Group (GADSWG) Scope

Purpose The Generating Availability Data System Working Group (GADSWG) is to implement a uniform approach to reporting and measuring North American generating plant availability, performance and other related reliability data.

Activities To accomplish its purpose, the GADSWG will perform the following activities: 1. Review and recommend new generation availability data that should be subject to mandatory collection by NERC. 2. Review additions and changes to the GADS Data Reporting Instructions (DRI) and GADS Wind Data Reporting Instructions. 3. Analyze, assess and report on trends and risks to reliability from generator availability and performance.

Annual Deliverables The GADSWG will contribute to the PAS’s annual Performance Analysis State of Reliability Report.

Membership The GADSWG will consist of: • Chair • Vice Chair • NERC staff coordinator(s) • Eight (8) At least one Generation Owner members from each NERC region • One Canadian member • One North American Generation Forum (NAGF) member • One Supply Association (EPSA) member • One additional “at large” member • Chairperson(s) of subgroup(s) • Other members, as requested by the NERC staff coordinator • Additional Members can be added: o At the request of the Planning Committee sector representatives, or o As needed by the NERC coordinator.

The subcommittee working group chair and vice chair are appointed by the chairs of the NERC Operating Committee and Planning Committee for one, two-year term. The vice chair should be available to succeed to the chair. The Operating Committee members shall be appointed by the chair of the Operating Committee for a two-year term.

Order of Business In general, the desired, normal tone of GADSWG business is to strive for constructive technically sound solutions which also achieve consensus. On the relatively few occasions where that desired outcome cannot be achieved, the GADSWG will defer to a vote by the Planning Committee to settle the issue. If any strong minority opinions develop, those opinions may be documented as desired by the minority and forwarded to the PAS Chair and PC Chair for future meeting consideration.

NERC staff advice should be about what the ERO needs to be successful. The above normal tone of PAS to seek a technically sound consensus is very important. NERC staff and Observers are also expected to strive for constructive technically sound solutions and seek consensus.

Reporting The GADSWG reports to the Performance Analysis Subcommittee (PAS). Reports and recommendations developed by the GADSWG require approval by PAS and the Planning Committee.

Meetings Four to six open meetings per year, or as needed.

Approved by the NERC Planning Committee: _, 20112016

Generating Availability Data System Working Group Scope 2 Is It Time to Deprecate the EFORd Formula That Uses the Markov Approach? Deprecation means that although something is available or allowed, it is not recommended or that, in the case where something must be used, to say it is deprecated means that its failings are recognized.

Ron Fluegge GADS Open Source September 27, 2016 The peaking unit only operates (in demand) 5 hours per day. A 48-hour forced outage (U1) overlaps two of the 5-hour demand periods (10 hours of demand for the unit). If the unit is expected to be available to meet the expected daily 5-hour , it does not matter whether the unit is available in reserve shutdown or in a scheduled or forced outage during the off peak hours as long as it is available to operate during the 5 hours of demand each day.

Intuitively, the risk of not being available to serve load during the demand periods should be the ratio of the forced outage hours during the demand periods divided by the sum of the forced outage hours during demand periods plus the service hours during the demand periods. EFORd: is the unit available (i.e., not forced out) to meet the demand you expect it to serve however you define the demand period? Equivalent Forced Outage Rate demand (EFORd) Input Data • Forced Outage Hours – FOH • Service Hours - SH • Equivalent Forced Derated Hours - EFDH • Equivalent Forced Derated Hours During Reserve Shutdowns - EFDHRS • Available Hours - AH • Reserve Shutdown Hours - RSH • # of FO occurrences • # of unit actual starts Where’s the “demand” data? • # of unit attempted starts IEEE Std 762-2006

• 1. Overview - The term demand applied to a rate, as in EFORd, indicates that the probability of an occurrence has been estimated for periods when the unit is in demand to generate • 3.2 demand forced outage rate (FORd): A measure of the probability that a generating unit will not be available due to forced outages when there is demand on the unit to generate. (NOTE - See 8.16.2) • 4.1.1.1 In service - The in-service state is where a unit is electrically connected to the system and performing generation function. • 4.1.1.1.1 In-service nongenerating mode - The in-service nongenerating mode state is where a unit is electrically connected to the system and performing nongenerating functions. NOTE 1 - Certain types of units may be performing functions other than generating while being in service and exposed to failure: A pumped storage unit can be in pumping mode, an electrochemical unit (battery) can be in charging mode, and a combustion turbine or a hydro unit can be in synchronous condensing mode. NOTE 2 - Certain types of generating units may be kept on line at minimum output when there is no demand on the unit to reduce the number of starts. IEEE Std 762-2006 (continued)

• 8.16.2 Demand forced outage rate (FORd)

• FORd = FOHd / (FOHd + SH) • Where FOHd is as defined in 6.10.2

• NOTE - When FOHd is determined directly from recorded periods of demand as noted in 6.10.2, service hours (SH) in the above equation should include only those under the specified demand condition. IEEE Std 762-2006 (continued)

• 6.10.1 Demand factor (f) - The demand factor is used to estimate forced outage hours overlapping the period of demand for the unit to operate. • D = Average demand time (duty cycle time) • D = SH / Number of demand occurrences • The number of demand occurrences is presumably equal to the number of attempted starts to generate, but if this is not available, the following approximation may be used: • D = SH / Number of successful starts to generate • This method is documented in the IEEE paper by Ringlee [B9]. In this paper, it was proposed to estimate T from the following: • (T + D) = AH / number of attempted starts IEEE Std 762-2006 (continued)

• 6.10.2 - Forced outage hours overlapping the period of demand for the unit to operate (FOHd) • The FOHd is the number of hours a unit was in a Class 0, Class 1, Class 2, or Class 3 unplanned outage state AND the unit would have operated had it been available. • If periods of demand are not recorded, FOHd may be estimated using the demand factor defined in 6.10.1. The demand factor is applicable to traditional demand for economic or reliable system operation. • NOTE - FOHd can be determined directly if periods of demand are recorded. Demand can be defined as the traditional demand for the generating unit for economic or reliable operation of the system, or it can be any other user-defined condition, such as specific weather condition, load level, or energy price. IEEE Std 762-2006 (continued)

6.18.3 Equivalent forced derated hours overlapping period of demand for the unit to generate (EFDHd) EFDHd is the in-service forced derated hours (see 6.15.1) converted to equivalent hours in accordance with 6.18.

EFDHd = EFDH − ERSFDH

NOTE - Accurately determining EFDHd requires collecting data so that in-service deratings are separated from reserve shutdown deratings. Demand can be defined as the traditional demand for the generating unit for economic or reliable operation of the system, or it can be any other user-defined condition, such as specific weather condition, load level, or energy price. The conditional probability of a unit not being available given a demand occasion is developed using a four-state Markov model. The authoring Task Group recommends that the developed conditional probability be used in place of Forced Outage Rate for capacity planning studies, especially when the application is for units in peaking or cycling service. The Forced Outage Rate parameter has been recognized to be unsuitable as a measure of outage risk when unit annual service hours are low. (circa 1971)

The alternative approach proposed is empirical requiring additional data on start times and stop times of the units to define the demand period. Collection of these additional data could be a difficult task in some companies. (circa 1978) Example 1

• A combustion turbine runs as follows: • Must serve the daily “demand hours” as defined by PJM … for 2016, PJM defines peak hours as between hour-ending 0700 to 2200, Monday through Friday excluding holidays. So essentially, the CT must run between 0600 and 2200 (16 hours) only on week days (excluding holidays). • Each and every peak day the CT starts successfully at the beginning of the peak period (0600) and runs for 16 hours (State 2). • However, every day that it runs it “trips unexpectedly” at 2200 as a result of various equipment problems that occurred: a U1 forced outage. • It takes the first half of the off-peak hours (State 1) to fix the cause of the U1 trip. During the remaining off-peak hours, the unit is in RS (State 0) until needed to start on the next peak day. • There are no startup failures: attempted starts = actual (successful) starts. • The unit started once per peak day. • The unit tripped once per peak day. • There are no deratings. • There are only two GADS event types: U1 and RS … this matches the 4-state Markov model By most reasonable operations definitions, the unit has served the peak hours (demand period) without any failure to serve load 0600 and 2200. All failures occurred during “off-peak” hours (tripped at 2200); therefore, one would assume EFORd = 0%.

However, as calculated for the expected entire year of 2016 with this historic operation, the “gross approximation” EFORd formula in the GADS DRI and IEEE 762 calculates a 30.91% for this unit. Example 2

• A combustion turbine runs as follows: • Must serve the daily demand hours as defined by PJM … for 2016, PJM defines peak hours as between hour-ending 0700 to 2200, Monday through Friday excluding holidays. • Each and every peak day it starts successfully at the beginning of the peak period (0600) and runs. However, every day that it runs it trips unexpectedly at 1800: a U1 forced outage 4 hours before the end of the demand period (state 3). Therefore, of the 16 hours of demand, it only serves load 12 of the 16 hours (state 2). • In addition to the peak demand hours between 1800 and 2200 each peak day, it takes the first half of the off-peak hours to fix the cause of the U1 trip (state 1). During the remaining off-peak hours, the unit is in RS until needed to start on the next peak day (state 0). • There are no startup failures. There are no deratings. • There are only two GADS event types: U1 and RS … this matches the 4- state Markov model. FOH = 3372 RSH = 2352 SH = 3060 # of FO occurrences = 255 # of unit actual starts = 255 By most reasonable definitions, the unit has # of unit attempted starts = 255 served the peak hours (demand period) for 12 of the 16 hours. All failures occurred

during the last 4 hours of the 16 “on-peak” r = 13.224 hours; therefore, one would assume an D = 12.000 EFORd = 25% T = 9.224

However, as calculated for the expected entire f = 0.688327 year of 2016, the EFORd formula calculates a fp = 0.56541 43.13% for this unit. EFDHd = 0 FOHd = 2321.04

EFORd = 43.13%

CT 1:

Of the 168 designated peak periods, the unit was in a scheduled outage (MO or PO) during 23 of the periods (141.16 peak period outage hours) and in a single forced outage event overlapping two consecutive peak periods (12.62 peak period hours). The unit was on-line for 3.73 hours during the peak hours

During the calendar year, the unit was unavailable (MO, PO, U1) for 760.03 hours that includes 86.62 forced outage hours; otherwise, the unit was available either in reserve shutdown (8,011.7 hours) or in service (12.27 hours). There were no deratings. There were reported 3 actual (successful) starts and 3 attempted starts

The single U1 occurred on a Thursday in June at 15:23. The Summer Peak Period for this unit is from 12:00 to 20:00. The cause of the U1 was repaired and completed at 06:00 the next week Monday.

The unit ran 3 times:

1. Wednesday during Winter Peak Period from 9:19 to 10:15 for a test run. The Winter Peak Period for this unit is from 04:00 to 10:00. 2. Friday during Summer Peak Period from 16:27 to 19:30 for system request (i.e., demanded). The Summer Peak Period for this unit is from 12:00 to 20:00. 3. Wednesday during non-peak Fall period from 7:22 to 15:39 for RATA testing.

GADS EFORd = 21.69%

EFOR during Company’s 168 demand periods only = 76.28% CT 2:

Of the 168 designated peak periods, the unit was in scheduled outages (MO) during 13 of the periods (63.62 peak period outage hours). There were no forced outage events during or overlapping any peak periods (0.0 peak period hours). The unit was on-line for 6.30 hours during the peak hours

During the calendar year, the unit was unavailable (MO, U1) for 1,553.12 hours that includes 4.95 forced outage hours; otherwise, the unit was available either in reserve shutdown (7,211.3 hours) or in service (19.58 hours). There were no deratings. There were reported 3 actual (successful) starts and 3 attempted starts

The single U1 occurred on a Wednesday in May during a non-peak period at 11:27. The cause of the U1 was repaired and completed at 16:24 the same day.

The unit ran 3 times:

1. Wednesday during a non-peak period from 16:24 to 21:19 for a test run. 2. Friday during Summer Peak Period from 13:36 to 19:54 and the next day, Saturday, from 12:23 to 20:45 for system request (i.e., demanded). There was a new weekend peak record on that weekend. The Summer Peak Period for this unit is from 12:00 to 20:00.

GADS EFORd = 12.98%

EFOR during Company’s 168 demand periods only = 0.00%

The data is from the same generating company so they are reported “consistently” based on corporate policy. While the above CTs illustrate the two extreme differences, they also have few service hours and starts. Therefore, we have chosen 11 more simple cycle CTs that have between 650 and 1300 service hours and between 77 and 138 attempted starts for the study year. There are no derated hours.

The differences, while not as dramatic, do exist. Table 1 – ILLUSTRATION OF THE METHOD from Reference 5 shows the results for both a “Low Demand Peaking Unit” and a “Typical Peaking Unit” (See below) Conclusion

• While this study has used “standardized” peak periods, the methodology works equally well with actual historical peak demand hours (e.g., on-peak hours, seasonal peak hours, or shortage hours including certain periods of system-wide emergency operations or actions) as defined by the intended use: for example, considering wind and solar renewables in the system, the peak LOLP hours. • As the industry can now calculate peak periods only, the EFORd approximation is no longer needed since it really does NOT know whether or not the unit really was able to serve load during demand as discussed above. • EFORd should be replaced by a more powerful alternative method to calculate EFOR during demand. The latter provides better calculated results, supports more unit types, and integrates better with the collected unit reliability, availability and maintainability data (e.g., NERC GADS). • I recommend that the GADS Working Group indicate in the DRI that the EFORd equation has been deprecated. Salvatore A. DellaVilla Jr. August/September 2016

Mine Data for the Gems of Knowledge….1 IEEE 762 – RAM Performance Metrics

Factors vs Rates…

What do they mean???

2 IEEE 762 – Some Caveats…

3 IEEE 762 – Some Important Basics…

It all starts here…What is a Unit?

4 IEEE 762 – Some Important Basics (cont’d)

So…Factors are time based…they are additive

Rates are measures of probability…

How do they compare?

5 So what is a unit?

HD-Simple Cycle Plant – Cold End Drive – 1 Unit HD-Single Shaft Combined Cycle Plant – Cold End Drive – 1 Unit

Condenser GEN GT Dump HRSG

GEN GT SSS Clutch ST

Balance of Plant (Station Equipment) Balance of Plant (Station Equipment)

HD-Multi Shaft Combined Cycle Plant – Cold End Drive – 4 Units Fossil Steam Turbine Plant – 2 Units

These are plants, prime movers…units

6 Let’s start with some conclusions

. Factors have a higher intrinsic value when compared with Rates (including Equivalent Factors & Rates)…This is a game changer in the market

. Service Hours strongly influence the value & meaning of an average Rate  Low Service Factor units experience low Service Hours on some time basis, hence they have high outage rates  As Service Hours increase, outage Factors & Rates converge  Raises strong concern about using Rates as a probability or likelihood of occurrence

. The probability or likelihood of a Forced Outage during a period of demand is important to “characterize” reliability performance  Drives pro forma plant economics… expectation & justification  Service Hours, Service Factor, Service Hours per Start are strong proxies for demand. Therefore, Forced Outage Factors are better indicators of performance & the probability or likelihood of occurrence

Note: Comments apply to Equivalent Factors/Rates

7 Let’s start with some conclusions

. As an industry, we are beyond the need to approximate unit demand… Demand data is available empirically from the plant DCS, data highway, or historian  Demand is very important…Duty is a strong proxy  The IEEE 762 approximation of demand is not adequate…Data available from control/historian

. IEEE 762 requires a review and update to account for technology changes & advance, evolving energy economics, & the need for a “level playing field” in evaluating the performance of generation alternatives.

The committee is in place and the process is underway!

Note: Comments apply to Equivalent Factors/Rates

8 A view of operating states

This sets up the “Accounting” buckets

9 Outage classifications

IEEE NERC GADS (DRI) ORAP® Forced Outage – Automatic Trip: Class 1 unplanned outage (immediate) U1-Unplanned (Forced) Outage – Forced Outage – While the unit was operating a component failure or immediate Automatic Trip (FOA) other condition occurred which caused the unit to be With Amplification Code = T1 shut down automatically by the control system. (Tripped/shutdown grid separation)

Forced Outage Manual Shutdown: Class 1 unplanned outage (immediate) U1-Unplanned (Forced) Outage – Forced Outage – Manual While the unit was operating, a component failure or Class 2 unplanned outage (delayed) immediate Shutdown other condition resulted in a decision or action by plant Class 3 unplanned outage (postponed) With Amplification Code = T2 (FOM) personnel that immediately removed the unit from service (Tripped/shutdown grid separation - by either an emergency shutdown or a normal shutdown. -manual) U2-Unplanned (Forced) Outage – Delayed U3-Unplanned (Forced) Outage – Postponed

Failure to Start: A signal was given to start the unit Class 0 unplanned outage SF – Startup failure Failure to start (FS) but there was a failure in the start sequence prior to (starting failure) achieving an in service state within a predetermined time period. For power generation units, the in service state is considered to be breaker closure. For mechanical drive units, the in service state is the point at which stable operation of the driven equipment is achieved. Sequential failures to start due to a single cause are to be counted as one failure to start event, unless corrective action is performed or a successful start is achieved in the interim.

Forced Unavailability: 1. The unit was available in the Class 1 unplanned outage (immediate) U1-Unplanned (Forced) Outage – Forced Unavailability Reserve Shutdown (standby) state, but a component immediate (FU) failure or other condition caused it to be reclassified as Extended planned outage with Amplification Code = Any other “Unavailable”. 2. An extension of a planned Extended maintenance outage code than T1 maintenance action due to additional component (Tripped/shutdown grid separation) failure/repair. or T2 (Tripped/shutdown grid separation – manual) PE – Planned outage extension ME – Maintenance Outage Extension

10 Outage classifications

IEEE NERC GADS (DRI) ORAP®

Maintenance - Unscheduled: Maintenance that is required, Maintenance outage MO – Maintenance Outage Maintenance – but has not been specified in the maintenance plan. This Unscheduled (MU) outage type can be a result of a unit shutdown, when the unit is not required to serve load, to facilitate repairs to the unit. Or an extension of a Scheduled Maintenance outage to repair a discovered condition, minor in nature, that does not affect the units ability to meet its next mission.

Maintenance - Scheduled: Maintenance activity that is Planned Outage PO – Planned Outage Maintenance – Scheduled planned well in advance of the outage, often as part of (MS) an annual maintenance plan and typically consists of major maintenance (i.e., inspections and overhauls)

Derating Unplanned: A component failure or limitation Class 1 unplanned derating (immediate) D1 – Unplanned (forced) Derating Derating Unplanned causes a decrease in the output of the unit. -- Immediate (DRU) Class 2 unplanned derating (delayed) D2 – Unplanned (forced) Derating Class 3 unplanned derating (postponed) -- Delayed Maintenance derating D3 – Unplanned (forced) Derating Extended maintenance derating – Postponed

Derating Planned: An equipment limitation or external forces Planned derating PD – Planned derating Derating Planned which result in a decrease in the output of the unit which is (DRP) Extended planned derating D4 – Maintenance derating scheduled well in advance of its occurrence. DP – Planned derating extension DM – Maintenance derating extension

11 Outage classifications

IEEE NERC GADS (DRI) ORAP® Non-Curtailing Event: A redundant component fails, but None None Non-curtailing Event – NC does not impact the intended operation of the equipment. Or maintenance that is performed on a specific component, but does not impact the unit’s ability to operate in a standard manner or its operational readiness.

Concurrent Maintenance: Maintenance is None None. These events are listed Concurrent Maintenance performed while downtown is charged to another under their associated primary (CM) component or piece of equipment. event in the NERC event card.

Reserve Shutdown: The unit was available for Reserve shutdown Reserve shutdown – RS Reserve shutdown – RS service, but due to lack of demand was not required to operate.

Note: Each Reserve Shutdown must be reported to support developing an approximation of a period of demand – more on this later…

12 A simple view of time

A day has 24 hours… A week has 168 hours… A month has…depends on the number of days— 30 days = 720 hours 31 days = 744 hours 29 days = 696 hours 28 days = 672 hours A year has…8760 hours (typically)

It is the correct allocation of time that makes the IEEE Standard work!

13 Let’s start…

8.3 Forced outage factor (FOF)

퐹푂퐻 퐹푂퐹 = 푥 100 푃퐻

8.16 Forced outage rate (FOR)

퐹푂퐻 퐹푂푅 = 푥 100 퐹푂퐻+푆퐻

8.16.1 Forced outage rate total—

generating or other functions (FORT)

퐹푂퐻 퐹푂푅 = 푥 100 푇 퐹푂퐻+푆퐻+푆퐻푁퐺

Where: FOH = Forced Outage Hours PH = Period Hours SH = Service Hours SHNG = In Service Non-Generating Mode (e.g. Synchronous Condensing)

14 Example 1…

Let’s look at a 24 hour day, where: PH = 24 FOH = 7 SH = 10 … 7 hours left over

8.3 Forced outage factor (FOF) 8.16 Forced outage rate (FOR)

퐹푂퐻 퐹푂퐻 퐹푂퐹 = 푥 100 퐹푂푅 = 푥 100 푃퐻 퐹푂퐻+푆퐻

7 7 FOF = x 100 FOR = x 100 24 7 + 10

7 FOF = .2917 x 100 FOR = x 100 = .4118 x 100 17

FOF = 29.2% FOR = 41.2%

Is the probability of a Forced Outage 41%?

15 Example 2…

Let’s look at another 24 hour day, where: PH = 24 FOH = 7 SH = 16 … 1 hour left over

8.3 Forced outage factor (FOF) 8.16 Forced outage rate (FOR)

퐹푂퐻 퐹푂퐻 퐹푂퐹 = 푥 100 퐹푂푅 = 푥 100 푃퐻 퐹푂퐻+푆퐻

7 7 FOF = x 100 FOR = x 100 24 7 + 16

7 FOF = .2917 x 100 FOR = x 100 = .3043 23

FOF = 29.2% FOR = 30.4%

As Service Hours (SH) increase, FOF & FOR Converge

16 Example 3 … One more example with made up data…

Peaking Cycling Baseload

PH = 8,760 PH = 8,760 PH = 8,760 FOH = 1,000 FOH = 1,000 FOH = 1,000 SH = 720 SH = 3,600 SH = 6,600 RSB = 7,040 RSB = 4,160 RSB = 1,160

1,000 1,000 1,000 FOF = x 100 FOF = x 100 FOF = x 100 8.3 Forced outage factor (FOF) 8,760 8,760 8,760 퐹푂퐻 퐹푂퐹 = 푥 100 FOF = .1142 x 100 FOF = .1142 x 100 FOF = .1142 x 100 푃퐻 FOF = ퟏퟏ. ퟒ% FOF = ퟏퟏ. ퟒ% FOF = ퟏퟏ. ퟒ%

1,000 1,000 8.16 Forced outage rate (FOR) FOR = 푥 100 1,000 FOR = 푥 100 1,000 + 720 FOR = 푥 100 1,000 + 6,600 퐹푂퐻 1,000 + 3,600

퐹푂푅 = 푥 100 1,000 1,000 1,000 퐹푂퐻+푆퐻 FOR = 푥 100 FOR = 푥 100 FOR = 푥 100 1,720 4,600 7,600

FOR = .5814 푥 100 FOR = .2174 푥 100 FOR = .1315 푥 100

퐅퐎퐑 = 58.1% 퐅퐎퐑 = 21.7% 퐅퐎퐑 = 13.2%

What is the probability of a forced outage during a period of demand?

17 Example 4 … Some real ORAP® Data (Technology Focus)

2015 "Vintage" Class Gas Turbines (US Mkt) - Simple Cycle Plant Statistics 2015 "E" Class Gas Turbines (US Mkt) - Simple Cycle Plant Statistics 100 100 # Units SF 80 80 SH/ST FOF FOR 60 60 % FOR (%) % FOR (%) 40 FOF (%) 40 FOF (%) # Units SF 20 20 SH/ST FOF 0 0 FOR Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec FOR (%) 29.56 22.37 22.21 25.40 26.48 24.42 35.21 31.21 24.64 22.52 17.98 21.62 FOR (%) 7.82 9.98 11.13 6.67 3.13 7.16 6.56 9.23 13.60 19.70 21.13 17.60 FOF (%) 5.40 4.29 3.94 4.55 5.23 5.21 9.08 7.14 4.95 3.76 2.94 3.77 FOF (%) 0.92 1.43 1.62 0.90 0.53 1.42 1.39 1.86 2.51 3.35 3.73 2.64 2015 Total 2015 "F" Class Gas Turbines (US Mkt) - Simple Cycle Plant Statistics 100 Vintage "E" "F" # Units 190 272 259 80 SF 14.43 14.83 60.88 60 % FOR (%) SH/ST 63.23 18.58 95.90 40 FOF (%) FOF 5.04 1.84 1.52 20 FOR 25.88 11.03 2.44 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec EFOF 5.15 2.21 1.62 FOR (%) 1.63 1.35 1.58 2.59 2.09 1.89 0.78 1.70 2.09 5.02 6.02 2.69 FOF (%) 0.93 0.74 0.86 1.47 1.20 1.30 0.56 1.22 1.45 3.17 3.68 1.68 EFOR 26.32 12.96 2.59 EFORd 7.04 2.74 1.69 EFORd is made to converge with FOF or EFOF

18 Example 4 … Some real ORAP® Data (Duty Cycle Focus)

2015 "F" Class Gas Turbines (US Mkt) - Peaking Duty - Simple Cycle Plant Statistics 2015 "F" Class Gas Turbines (US Mkt) - Cycling Duty - Simple Cycle Plant Statistics 100 100

80 80

60 60 % FOR (%) % FOR (%) 40 FOF (%) 40 FOF (%)

20 20

0 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec FOR (%) 67.13 21.33 71.24 0.71 5.48 22.27 4.55 6.57 39.47 53.94 22.86 8.92 FOR (%) 4.40 9.34 0.51 13.62 1.41 1.86 1.35 5.54 15.46 23.29 23.60 2.59 FOF (%) 2.92 1.44 3.52 0.02 0.29 1.77 0.34 0.44 2.61 3.77 2.44 0.11 FOF (%) 0.33 1.24 0.08 3.51 0.33 0.58 0.42 1.89 5.00 7.70 7.79 0.38

2015 Total - "F" Class 2015 "F" Class Gas Turbines (US Mkt) - Baseload Duty - Simple Cycle Plant Statistics 100 Peaking Cycling Baseload

80 # Units 30 42 187 SF 4.37 22.51 78.12 60 % FOR (%) SH/ST 9.11 20.24 139.84 40 FOF (%) FOF 1.64 2.43 1.31 20 FOR 27.27 9.76 1.65

0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec EFOF 1.65 2.52 1.42 FOR (%) 0.98 0.74 0.86 1.79 2.10 1.59 0.69 1.33 0.55 2.64 4.03 2.68 FOF (%) 0.74 0.52 0.60 1.28 1.54 1.39 0.62 1.21 0.48 2.08 3.03 2.18 EFOR 27.39 10.08 1.79 EFORd 5.29 3.00 1.58 Duty cycle is a proxy for demand!

19 Measuring Demand

Either:

. Empirically… or . Approximation

IEEE Provides an Approach

But today, the approach is in question

20 Forcing EFORd to Work…

Numerous Equations and Factors reviewed to flatten EFORd curve… Converges with FOF or EFOF

21 Another view

Abstract:

“It is now recognized by many experts that the traditional NERC GADS generator reliability rates and factors do not adequately reflect the complexity of today’s market condition. EAF, EUOR, EFORd and such do not accurately measure the localized demand variability embedded in the Locational Marginal Pricing of energy. Key questions about the profitability of being available and the economic losses of being unavailable are not answered. Commercial Availability (CA) is emerging to be a better indicator of Generator Reliability and Performance when economic impact Date: April 2016 is attached to them.”

22 Let’s end with some additional conclusions

. Factors have a higher intrinsic value when compared with Rates (including Equivalent Factors & Rates)…This is a game changer in the market

. Market expectations for product performance drives the OEM design…  Fast start capability  Load following  Improved efficiency & higher output  Environmental friendliness

. OEM design influences owner/operator O&M practice…  Parts life  Ancillary services  O&M cost

23 Let’s end with some additional conclusions

. Economics drives:  Dispatch profile  Profitability  Reserves  Competitive generation: Merchants, Renewables

. Off-Grid alternatives are becoming a reality…we need to measure performance

. IEEE Metrics have a strong value!  Basis for comparing performance of the evolving generation mix  Standard metrics – Reviewed & improved based on current market needs

24 QUESTIONS?

Connect with us spsinc.com 25 Conventional GADS Training

Donna Pratt, Performance Analysis Manager – Data Analytics GADSWG Meeting September 27, 2016 Agenda

•Summary •Training Survey Responses •Plans for Additional Training

2 RELIABILITY | ACCOUNTABILITY Summary

•April 2016 . Atlanta – NERC Offices . Number of Attendees: 26 •June 2016 . Salt Lake City – WECC Offices . Number of Attendees: 70

3 RELIABILITY | ACCOUNTABILITY Training Survey Responses

Apr-16 Jun-16 Difference The objectives of the training were clearly defined. 4.32 3.73 -0.59 The content was organized and easy to follow. 4.18 3.58 -0.60 Participation and interaction were encouraged. 4.73 3.90 -0.83 The topics covered were relevant to me. 4.36 3.85 -0.51 This training experience will be useful in my work. 4.50 3.94 -0.56 The trainer was knowledgeable about the training topics. 4.77 4.06 -0.71 The trainer was well prepared. 4.79 3.85 -0.94 The training objectives were met. 4.42 3.64 -0.78 The time allotted for the training was sufficient. 4.47 3.84 -0.63 The meeting room and facilities were adequate and comfortable. 4.84 4.58 -0.26 Overall Average 4.54 3.90 -0.64

• Results are out of a maximum score of 5 • Reduction of 14% in Overall Average score between April and June sessions . More diversity in the experience level of attendees in June . About 1/3 of June class were GADS users with less than a year of experience

4 RELIABILITY | ACCOUNTABILITY Open Ended Question 1

What did you like most about this training?

Apr-16 Jun-16 Relevant subject matter. Factors/Rules Examples on how to apply codes in the calculation for different scenarios discussion on day 2. were very helpful Discussions about interpretations, amplification The introduction to what GADs was, and what people expect to get out of codes GADs The trainers, interaction of teachers w/students, meeting peers Good explanation & questions/answer session very helpful Knowledgeable instructors and fantastic platform for interaction with Covered topics that were helpful other users Explanation of factor calculations

5 RELIABILITY | ACCOUNTABILITY Open Ended Question 2

What aspects of the training could be improved?

Apr-16 Jun-16 question 9 comment) No time spent on actually entering plant data into GADS system. Question 10 comment) The large wood posts obstruct the screen(and laser pointer. Spend more time explaining the "why". Why is this reporting required? why is this level of granularity necessary? why is this applied equally to 2,500 MW plants and 25 MW plants? Why does Slides either less crowded or better paced NERC need this level of data? More time spent on data entry, more time spent on finding navigating NERC website, have training material posted online before presmtation at least the day of the presentation (aim before class starts), too much detail webE-GADS ran too late into the day in last presentation. Day 1 & 2 need to be provided as separate training sessions. Day 2 for Separate beginner GADS and Advanced GADS more advanced . add some additional backgrounds as to way we have to training do GADS what is NERC's reason? Printouts of all presentations Handouts of the slides so we could make notes more easily. More info on navigating NERC website Allow more time for Q & A during course presentations

6 RELIABILITY | ACCOUNTABILITY Open Ended Question 3

How do you hope to change your practice as a result of this training?

Apr-16 Jun-16 I'm more prepared to answer my customers questions and also help Better understanding validate their data entry Pay more attention to cause codes now! More efficient and clearer direction in reporting unit outages. More regional involvement, now that we understand the degree to which judgment is needed in choosing outage types, cause codes and

I will dip into the NERC Website more frequently.

7 RELIABILITY | ACCOUNTABILITY Open Ended Question 4

Please share other comments or expand on previous responses here:

Apr-16 Jun-16

Because of the diverse experience in the room, it would help to breakout they days into 1)beginner, 2) advanced. In beginner day more history and how/when GADS was formed and approved, what it's used for, etc. Also, a brief overview of OATI too would be helpful. Advance day could then be a Would like more hands-on examples using deeper dive majority GADS. Then attendees could choose to attend all or software/application just one of the days based on their specific needs. Allow more detailed instructions on how OATI looks at data and how to correct deficiencies during uploading into OATI

Working groups question & answer session was really helpful (what if's...helpful direction...how do others do it (similar circumstances) industry standards clarification, FERC/NERC/WECC PEAK RC-which group's compliance reporting guides the others?-DRIC uses forced-while the other doesn't, one views moth balled as a problem for licensing, the other group recommends it. Clarifying confusing requirements would help a lot. 8 RELIABILITY | ACCOUNTABILITY Plans for Additional Training

• Additional training . Interest in a webinar-based session for those who are unable to travel o Webinar would be recorded and then posted on NERC’s website as modules o Schedule dependent on trainer availability . Requests for separate beginner and advanced training •Requests for a 2017 schedule o Would allow users more time to plan to participate

9 RELIABILITY | ACCOUNTABILITY 10 RELIABILITY | ACCOUNTABILITY GADS Wind Update

Donna Pratt, Performance Analysis Manager – Data Analytics GADSWG Meeting September 27, 2016 Agenda

• Wind Data Reporting Instructions (DRI) Changes • Implementation Update . Project . Application • Training . Approach . Topics . Schedule . Call for Volunteers

2 RELIABILITY | ACCOUNTABILITY Wind DRI Changes

• Wind DRI Changes approved by NERC’s Planning Committee . September 13, 2016 • Revised Wind DRI will be posted on NERC’s website by end of September . GADS Wind page . Revision history within the document provides summary of changes by section . Notice to GADSWG and announcement on GADS page when available • Red-line of changes from earlier version . Will be posted under Archives section on GADS Wind page

3 RELIABILITY | ACCOUNTABILITY Implementation Update - Project

•Project status . Development: underway . Test: October through November . Rollout: December through February o Rollout to include industry outreach, training, and registration to access the reporting application – Industry outreach: webinars and conference presentations to various wind industry groups – Additional outreach: via NERC-issued announcements, NERC Regions, and GADSWG •Dedicated e-mail: [email protected]

4 RELIABILITY | ACCOUNTABILITY Implementation Update - Application

•GADS Wind Reporting application will support data submission, views, and reports/exports •Samples of the preliminary screen designs for the GADS Wind Reporting application follow . Application is still in development, therefore some changes to screen layouts may occur before release o Final screens will be included in the training materials

5 RELIABILITY | ACCOUNTABILITY Application Preview: Main Page

6 RELIABILITY | ACCOUNTABILITY Application Preview: Submission Page

Company 1 Company 2

7 RELIABILITY | ACCOUNTABILITY Application Preview: File Validation Results

Company 1 Company 2

8 RELIABILITY | ACCOUNTABILITY Application Preview: File Validation Results

Primary Validation: Utility ID, user, and file format

Secondary Validation: record content

9 RELIABILITY | ACCOUNTABILITY Application Preview: Reports/Exports

Company 1 Company 2

10 RELIABILITY | ACCOUNTABILITY Application Preview: Request Voluntary ID

11 RELIABILITY | ACCOUNTABILITY Application Preview: Contact NERC

12 RELIABILITY | ACCOUNTABILITY Training - Approach

• Three-pronged . Approach: o Process: how to register, get access to the system, request Sub-Group IDs, ownership transfer, what to report, NERC website resources o Subject matter: terminology, concepts, how to categorize and report operations o Application: how to submit data through the GADS Wind reporting application . Delivery methods: o In-person, narrated presentations, and videos

13 RELIABILITY | ACCOUNTABILITY Training Topics - Process

. Registration and Utility IDs . Sub-Group Ownership management o Requesting Sub-Group IDs o Ownership Transfers and related reporting obligations . Reports/exports/data retrieval . Reporting periods . NERC web site: GADS and GADSWG, DRI, training materials, reference documents

14 RELIABILITY | ACCOUNTABILITY Training Topics - Subject Matter

• Data Design • Practical Application . Plants, Groups and Sub-Group . Outage definitions and (coding): . Sub-Group configuration o Forced (FO), . Turbine hours o Maintenance (MO), . Performance records o Planned (PO) and Outside Management Control . Component records o (OMC) . Column definitions . Parallel Events . Hourly roll-up – Flow charts . Coding challenges . Data quality control . Pooling concepts . Component Codes . Derating Balance of Plant versus Turbine o . Reserve Shutdown . Frequently Asked Questions - FAQs

15 RELIABILITY | ACCOUNTABILITY Training Topics - Application

• Accessing the GADS Wind Reporting application • Requesting a Voluntary Reporting ID • Data Submission • Views, Reports, and Exports

16 RELIABILITY | ACCOUNTABILITY Training – Preliminary Schedule

Purpose Audience Delivery method Timeframe Testing of Process Regional contacts In-person, instructor-led Dec. 2016 Training Material PowerPoint Testing of Subject GADSWG Webinar-based, instructor- Dec./Jan. 2017 Matter Training led PowerPoint Material Revised Process and Potential GADS Wind In-person and webinar- Feb. and Mar. Subject Matter users and Regions based instructor-led 2017, plus Oct. Training PowerPoint, including and Nov. 2017 recorded webinar Testing of Regional contacts and Announcement with link to Jan. 2017 Application Tool GADSWG draft training videos on Training Material NERC’s website with request for feedback Application Tool Potential GADS Wind Videos available on NERC’s Feb. 2017 Training videos users and Regions website Train-the-Trainer Interested organizations In-person, instructor-led Q2 or Q3 2017 PowerPoint

17 RELIABILITY | ACCOUNTABILITY Training – Volunteers Needed

• Looking for volunteers to: . Assist in development of content and/or . Provide review and comments on draft training modules • If interested in participating, please send an e-mail by October 15 to [email protected]

18 RELIABILITY | ACCOUNTABILITY 19 RELIABILITY | ACCOUNTABILITY

Appendix C – Analysis of Generation Data

Introduction GADS began as a voluntary reporting system in 1982. In 2012, GADS data collection became mandatory as part of NERC’s reliability program. Mandatory reporting was phased in; only units 50MW and larger were required to report their operating data to GADS in 2012, all others were voluntary. Beginning in 2013, all units 20MW and larger were required to report their data. In addition, some smaller units report into GADS on a voluntary basis. Except where noted, the analysis for this report includes only active units with a mandatory reporting obligation.93 Data used in the analysis includes information reported into GADS through the end of the 2015 reporting year.

Currently, GADS does not include wind, solar or other renewable technology generating assets. Wind performance data reporting requirements have been developed and a phased in reporting process will begin in 2017–2020. Reporting data requirements for solar have been initiated with a target goal of beginning data submittal by 2021.

GADS collects and stores unit operating information on a quarterly basis. By pooling individual unit information, overall generating unit availability performance and other metrics are calculated. The information supports equipment reliability, availability analyses, and risk-informed decision making to industry. Reports and information resulting from the data collected through GADS are used by industry for benchmarking and analyzing electric power plants. Table C.1 shows some key characteristics of the population in GADS.

Table C.1: Key Characteristics of GADS Metric/year 2012 2013 2014 2015 Number of Reporting Units =>20 MW 4,343 6,033 6,169 6,236 Average Age of the Fleet (Years) 28.8 33.2 33.6 34.2 Average Age of Units (Years) 40.2 41.0 42.0 43.0 Average Age of Gas Units (Years) 19.2 21.9 22.4 22.9 Average Age of Nuclear Units (Years) 33.0 34.0 35.0 36.1

The age of the generating fleet is a particularly revealing statistic derived from GADS, because an aging fleet could potentially see increasing outages. However, with proper maintenance and equipment replacement, older units may perform comparably to newer units. Figure C.1 uses GADS data to plot fleet capacity by age and fuel type. Figure C.1 shows two characteristics of the fleet reported to GADS: (1) there is an age bubble around 36–45 years, driven by coal and some gas units; and (2) there is a significant age bubble around 11–19 years comprised almost exclusively of gas units. The data shows a clear shift toward gas-fired unit additions, and the overall age of the fleet across North America is almost 10 years younger than the age of the coal-fired baseload plants that have been the backbone of power supply for many years. This trend is projected to continue given current forecasts around price and availability of as a power generation fuel, as well as regulatory impetus.

93 In 2015, fewer than 100 MW of units had a voluntary reporting status in GADS. In addition, differences between historical data reported in this report and the 2015 report are due to this change in the analysis. Units that retired in 2015 are also excluded from the analysis.

NERC | State of Reliability | May 2016 115 Appendix C – Analysis of Generation Data

Figure C.1: Fleet Capacity by Age and Fuel Type as of January 1, 2016

Generator Fleet Reliability GADS contains information that can be used to compute a number of reliability measures such as EFOR and EFORd. EFORd is a metric that measures the probability that a unit will not deliver its full capacity during demand periods due to forced outages or deratings. These reliability measures are or have been used by various ISOs/RTOs for conducting resource adequacy planning and/or system operations assessments.

Figure C.2 presents the monthly megawatt-weighted EFORd94 across the NERC footprint for the five-year period 2009–201495. The mean outage rate over that period is 4.3 percent. EFORd has been fairly stable with only a few significant excursions, as indicated by the highlighted bars in the figure.

94 The use of the weighted EFORd allows the comparison of units that vary by size. 95 The reporting year covers January 1 through December 31 with a reporting deadline that occurs in mid-February of the following year. Performance analysis for calculating the megawatt-weighted EFORd of a reporting year is completed in a NERC system that requires additional validation and processing of the GADS data that continues beyond the preparation period of this report. Therefore, the megawatt-weighted EFORd in this report is based on unit performance in 2014.

NERC | State of Reliability | May 2016 116 Appendix C – Analysis of Generation Data

Figure C.2: Monthly Capacity Weighted EFORd 2009–2014

Forced Outage Causes To better understand the causes of forced outages of generators, the annual and top-10 forced outage causes for the summer and winter seasons were analyzed for the period of 2012–2015. This analysis is focused on forced outage causes measured in terms of megawatt hours lost, to reflect both the amount of capacity affected and the duration of the outages.

The levels of forced outages reported into the GADS database are presented in Figure C.3 and Table C.2, providing detail on the MWh lost due to forced outages for the period 2012–2015 by season.

Figure C.3: Total MWh Lost Due to Forced Outages 2012–2015

NERC | State of Reliability | May 2016 117 Appendix C – Analysis of Generation Data

Table C.2: Total MWh Lost Due to Forced Outages, by Season 2012–2015 NERC Total Annual MWh Summer MWh Winter MWh Spring/Fall MWh 2012 214,867,802 62,890,135 72,191,101 79,786,567 2013 651,511,562 129,920,201 363,617,775 157,973,586 2014 422,713,436 97,264,944 162,009,409 163,439,083 2015 450,958,972 129,703,616 204,677,109 116,578,248

Based on the four years of available data since GADS reporting became mandatory, the following observations can be made: x Between 2012 and 2013, the number of units with a mandatory reporting obligation increased by 39 percent. This increase in the number of units reporting is the primary reason for the increase in forced outage MWh reported in 2012 and 2013. x Severe storms in the last quarter of 2012, such as Hurricane Sandy, resulted in an increase in the forced outage MWh reported for winter96 2013 and 2014. ƒ For this analysis, the season of a forced outage is associated with the season in which the start date of the event was reported in that year; when an event continues into the next year, a new event record is created in January. This results in the event being categorized as occurring in the winter for the continuation event. x Between 2012 and 2014, the shoulder months of spring/fall have higher forced outage MWh than the summer period.

Further analysis into the causes of forced outages considered the impact of weather. Figure C.4 presents the percentage of MWh lost due to weather-related forced outage cause codes reported each year. This indicates that while weather does cause major headlines, the overall effect on the fleet is minimal. The real impacts of weather- related events are localized impacts and of relatively short duration.

96 Winter includes the months of January, February and December. When analysis is performed on a calendar year basis, as for this report, these three months are included from the same calendar year. Summer includes May through September; all other months are categorized as spring/fall.

NERC | State of Reliability | May 2016 118 Appendix C – Analysis of Generation Data

Figure C.4: Contribution of Weather-Related Causes to Annual Total MWh Lost Due to Forced Outages 2012–2015

To gain additional insight into the drivers for the reported megawatt hours lost due to forced outages, the top-10 forced outage causes were examined to determine the impact these top-10 forced outage causes have on the annual total of MWh lost. The top-10 forced outage causes represent one percent of the types of forced outages reported annually; Table C.4 lists the top-10 outage causes for each year in the analysis period. Figure C.5 shows the contribution of the top-10 forced outage causes on a NERC-wide basis over the period 2012–2015.

NERC | State of Reliability | May 2016 119 Appendix C – Analysis of Generation Data

Figure C.5: Contribution of Top-10 Causes to Annual Total MWh Lost Due to Forced Outages 2012–2015

Table C.3 provides a comparison of the top-10 causes to the corresponding annual total of MWh lost due to forced outages. The contribution from the top-10 causes to the annual total megawatt hours lost averages 33.8 percent, with the highest percentage of megawatt hours lost due to the top-10 causes occurring in 2013. The average is only slightly higher than the contribution of top-10 causes for 2014 and 2015.

Table C.3: Percentage of Top 10 Forced Outage Cause MWh by Season to Annual Total MWh Lost Due to Forced Outages 2012–2015 NERC Total Annual MWh Summer MWh Winter MWh Spring/Fall MWh 2012 30% 6% 12% 12% 2013 41% 5% 27% 9% 2014 32% 6% 14% 13% 2015 32% 7% 21% 5%

The top-10 causes vary annually and the contribution from each of the top-10 causes to the total megawatt hours lost varies as well. Figure C.6 shows the contribution from each of the individual top-10 causes that accumulate by year to the top-10 annual lost MWh shown in Figure C.3.

NERC | State of Reliability | May 2016 120 Appendix C – Analysis of Generation Data

Figure C.6: Contribution of the Individual Top-10 Cause Codes to Top-10 Annual MWh Lost Due to Forced Outages

Table C.4 lists the top-10 forced outage causes on an annual basis. The list is ordered from the most impactful cause to the least, based on annual MWh lost.

Table C.4: Top 10 Cause Codes on Annual Basis (MWh) Rank 2012 2013 2014 2015 Waterwall (Furnace Waterwall (Furnace 1 Rotor - General Main wall) wall) Emergency Generator Stator Windings, 2 Rotor - General Main Transformer Trip Devices Bushings, and Terminals Transmission System Stator Windings, 3 Problems other than Bushings, and Flood Generator Vibration Catastrophes Terminals Other Miscellaneous Waterwall (Furnace 4 Main Transformer Main Transformer Generator Problems wall) Other Boiler Lack of Fuel 5 Instrumentation and Stator - General (interruptible supply of Stator Core Iron Control Problems fuel)

NERC | State of Reliability | May 2016 121 Appendix C – Analysis of Generation Data

Table C.4: Top 10 Cause Codes on Annual Basis (MWh) Rank 2012 2013 2014 2015 Regulatory Other Low Pressure Major Turbine Overhaul 6 Second Superheater Proceedings and Turbine Problems (720 Hrs. Or Longer) Hearings Generator Output AC Conductors and 7 Rotor Windings Other Exciter Problems Breaker Buses Other Switchyard or Stator Windings, 8 Hurricane Flood High Voltage System Bushings, and Terminals Problems Regulatory Waterwall (Furnace Major Turbine Overhaul Other High Pressure 9 Proceedings and wall) (720 Hrs. Or Longer) Turbine Problems Hearings Air Supply Duct Miscellaneous 10 First Reheater AC Protection Devices Expansion Joints Regulatory

Several outage causes appear in the top 10 more often than others. Weather-related outages in 2012 due to Hurricane Sandy resulted in flooding which impacted a number of units that continued to report forced outages into 2013 and 2014. Table C.5 lists the recurring cause codes and number of years that the cause code appears in the top 10.

Table C.5: Recurring Top 10 Cause Codes Number of Years in Top 10 Code Description Causes 1000 Waterwall (Furnace wall) 4 3620 Main Transformer 4 4520 Stator Windings, Bushings, and Terminals 3 9000 Flood 2 9500 Regulatory Proceedings and Hearings 2 4400 Major Turbine Overhaul (720 Hrs. Or Longer) 2 4511 Rotor - General 2

The waterwall outages would generally be expected given the amount of steam generation in the fleet. These failures are not an uncommon occurrence in normal operations. Main Transformer outages are also high on the list. This is likely a result of the long lead time to replace a failed generator step up transformer. While the failure rate is very low, the impact is high for main .

NERC | State of Reliability | May 2016 122

GADS Work Group Some information Gleaned from gathered from the 2016 State of Reliability report for consideration as we develop some ideas for the GADS contribution to the 2017 version A. From the report, the following key findings were identified: 1. Key Finding 1: Protection System Misoperations Decline; Top Causes Remain Unchanged 2. Key Finding 2: BPS Resiliency to Severe Weather Improved 3. Key Finding 3: Human Error Has Decreased 4. Key Finding 5: Modeling Improvements Led to Improved Blackout Risk Assessments 5. Key Finding 6: Essential Reliability Services Trend is Stable; Faces Potential Challenges a. The prospect of a changing resource mix presents a potential challenge to ERSs, in particular frequency and voltage support. b. Stable frequency is a key ALR performance outcome. c. Additional concerns exist relative to BPS voltage support 6. Key Finding 7: No Load Loss Due to Cybersecurity Events

B. Highlights Identified in the Report that affect GADS

1. Winter Preparedness and Performance Review a. This was one of the highlights. Should we highlight more weather related outages?

C. Overview of Severity Risk Analysis (edited version)

This is to understand NERC’s calculation and how generation fits into the equation: Observations. From the 2016 SOR:

The 2015 daily SRI has shown improved performance from the 2014 SRI as expressed by the mean and stand deviation. For each component of the SRI, the following observations can be made:

• Generation Component: The generation loss component of the SRI indicates 2011 was the benchmark year for the generation fleet; however, this time period pre-dates the mandatory generation reporting requirements so it is inconclusive whether that year should be the measure against which subsequent years should be compared.

• Transmission Component: With regard to the transmission component of the SRI, a statistically significant improvement has been observed as measured by mean and standard deviation between the two three-year periods of 2010–2012 and 2013–2015.

• Load Loss Component: The load loss component of SRI exhibits a non- statistically significant trend with the mean remaining at an improved level, complemented by a reduction in the variation on a daily basis, as measured by the standard deviation.

Background to the Calculation

Since the inception of the State of Reliability Report, the industry has developed a metric, named SRI,16 which serves to measure the effect of BPS performance on a daily basis. The metric is a composite, weighting transmission system forced outages for voltages 200 kV+, generation system unplanned outages, and distribution load lost as a result of events upstream of the distribution system. Each of these components is weighted at a level recommended by the OC and Planning Committee (PC),17 dating back to the 2011 time frame. Generation capacity lost is divided by the total generation fleet for the year being evaluated and factored at 10 percent of the SRI score. Transmission line outages are weighted with an assumed average capacity based upon their voltage level and the daily outages divided by the total inventory’s average capacity and factored at 30 percent of the SRI score. Load lost due to performance upstream of the distribution system is calculated based upon outage frequency for the day, which is divided by system peak loading, and is factored at 60 percent of the SRI score.

D. Outages shown in reporting data in SOR From the report:

Figure 3.4 [below] breaks down the 2015 cumulative performance by BPS segment. The components are generation, transmission, and load loss, in that order. In Figure 3.4, the load loss component shows day-to-day load-loss events. The transmission loss component improves at the beginning of August, indicated by the change in slope. The unplanned generation unavailability component is typically the largest contributor to cumulative SRI.

E. Contributing Causes

The information below may be a way GADS can be mined to show generating outages in these categories and a similar graphic display? Chapter 3 – Severity Risk Assessment and Availability Data Systems Figure 3.4: NERC Cumulative SRI by Component for 2015

Figure 6.2: The Percentage of Contributing Causes by Major Category Identification of these large areas of concern allow for the ability to prioritize and search for actionable threats to reliability. When this data was turned over to the AC Substation Equipment Task Force (ACSETF) to further investigate, it was determined that while the initial data pointed to the potential problem areas, the data was not detailed enough to analyze any specific problem areas. As a result, and following recommendations from the ACSETF report, an addendum for the types of information needed to support the Event Analysis process when failed equipment is identified was developed.56 Issue Unsourced

Term "EDH" is undefined, Appendinx L-1-18 "Sample Data Summary"

Review/revise EFORd equation

Add summary tables to end of Event Reporting and Performance Reporting (Sections 3 & 4)

GADSWG Meeting Notes - March 8-9 2016 (Word Doc)

Inclusion of FAQs document in DRI

FAQ - Add a table of contents and group questions by topic

Full copy of DRI should be posted

Suggested that reporting obligation ends after retire date of a unit, rather than until the end of the year

Appears to be a conflict between DRI & IEEE 762 for synchronous condensing and pumping

No mechanism in GADS to capture Pumping & Condensing Service Hours, all reported as Service Hours

Possible alternative to equation 25 calculation

Review ownership transitions process

GADSWG Face-to-Face Meeting-March 2016-Final For Posting (Powerpoint)

Year in the footer needs updated from 2015 to 2016

Correct MO-Maintenance Outage definition, Section 3 Event Reporting Page III-7

Consider clarification of commented item

Reporting of 0 service hour units

How to record transition from synchronous condensing mode to generating mode without an intervening event Scenario #6 Issue 1: Remove "or ME" from second to last sentence in MO- Maintenance Outage. Page III-7 Scenario #6 Issue 2: Correct example relating to MO to MO transition. Page III-10

Consider revision of unit state transitions

Clarify lack of fuel events for various situations

Email (Email subject listed, description in comment)

Maintenance Outage - Derating Transition

GADS Data Reporting Instgructions, Appendix F

NERC vs ISO "differences" - DRI Appendix M

GADS DRI: Cause Codes cause code addition request

upstream cause code for

Download cause code

CO2 stack exceedance

Carbon sequestration Proposed Change (As available)

IEEE 762 Equivalent Derated Hours Use slide 2, 3 from Ron's slides and 4,5 with different methods noted; Ed to provide suggested language

Insert Ed's tables from the training? Check with Jack.

Hold off on FAQs; put in separate doc

Lee to work within NERC to resolve

OATI system constraint, would be a change order

Not an issue

Not an issue

Refer to Ron's proposed changes

Maggie and Lee to work

Lee to address Make proposed correction. "If the inspection reveals damage that prevents… Forced Outage (U1, U2, or U3).

Table

Table

Update DRI to indicate this -Steve Wenke and Tyler Brun. Send text to Steve and Tyler.

Use this text: "If the damage found during the inspection… could be considered MO or ME." Use this text: "Extending a Planned/Maintenance Outage when work is not part of the original scope of work."

Change; put in DRI. Lee to verify with OATI no change request is needed.

Re: Lines 23 and 25. Will need to be a change order for OATI for the DRI 2018.

For 2018: Stenghten training …allow transition from planned outage to maint outage or other outages

Lee to work Table: Canadian Electric Association issue Look at Wing Cheng's training

Update OATI text exists for steam not gas turbine, add new misc code for gas turbines. Re:steam turbines 4460

Ed to provide cause code description, diagram, and number. NERC / Ed to follow up off line

Ed to provide cause code description, diagram, and number. NERC / Ed to follow up off line

Ed to provide cause code description, diagram, and number. NERC / Ed to follow up off line

Ed to provide cause code description, diagram, and number. NERC / Ed to follow up off line Information/Comments

Ron Fluegge stated alternative was crude and IEEE recommended against its use

Maggie Peacock offered to assist with improvements

Transition from MO to U2 or U3 requires unit synchronization. The U2 and U3 event types can only occur from service. Large suggestion, see slides 76-83 from "GADSWG Face-to-Face Meeting- March 2016-final for posting"

Slides 76&83