ML022410004
| ML022410004 | |
| Person / Time | |
|---|---|
| Issue date: | 08/28/2002 |
| From: | Beckner W NRC/NRR/DRIP/RORP |
| To: | |
| Sanders S | |
| References | |
| OMB 3150-0195 RIS-02-014 | |
| Download: ML022410004 (37) | |
See also: RIS 2002-14
Text
Attachment 1
Attachment 1, Section 2.2, Mitigating Systems Cornerstone, of NEI 99-02, Regulatory
Assessment Performance Indicator Guideline (Draft)
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1
2 MITIGATING SYSTEM PERFORMANCE INDEX
3 Purpose
4 The purpose of the mitigating system performance index is to monitor the performance of
5 selected systems based on their ability to perform risk-significant functions as defined herein. It
6 is comprised of two elements - system unavailability and system unreliability. The index is used
7 to determine the significance of performance issues for single demand failures and accumulated
8 unavailability. Due to the limitations of the index, the following conditions will rely upon the
9 inspection process for determining the significance of performance issues:
10
11 1. Multiple concurrent failures of components
12 2. Common cause failures
13 3. Conditions not capable of being discovered during normal surveillance tests
14 4. Failures of non-active components
15
16 Indicator Definition
17 Mitigating System Performance Index (MSPI) is the sum of changes in a simplified core damage
18 frequency evaluation resulting from changes in unavailability and unreliability relative to
19 baseline values.
20
21 Unavailability is the ratio of the hours the train/system was unavailable to perform its risk-
22 significant functions due to planned and unplanned maintenance or test on active and non-active
23 components during the previous 12 quarters while critical to the number of critical hours during
24 the previous 12 quarters. (Fault exposure hours are not included; unavailable hours are counted
25 only for the time required to recover the trains risk-significant functions.)
26
27 Unreliability is the probability that the system would not perform its risk-significant functions
28 when called upon during the previous 12 quarters.
29
30 Baseline values are the values for unavailability and unreliability against which current changes
31 in unavailability and unreliability are measured. See Appendix F for further details.
32
33 The MSPI is calculated separately for each of the following five systems for each reactor type.
34
35 BWRs
36 * emergency AC power system
37 * high pressure injection systems (high pressure coolant injection, high pressure core spray, or
38 feedwater coolant injection)
39 * heat removal systems (reactor core isolation cooling)
40 * residual heat removal system (or their equivalent function as described in the Additional
41 Guidance for Specific Systems section.)
1
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 * cooling water support system (includes risk significant direct cooling functions provided by
2 service water and component cooling water or their cooling water equivalents for the above
3 four monitored systems)
4
5 PWRs
6 * emergency AC power system
7 * high pressure safety injection system
8 * auxiliary feedwater system
9 * residual heat removal system (or their equivalent function as described in the Additional
10 Guidance for Specific Systems section.)
11 * cooling water support system (includes risk significant direct cooling functions provided by
12 service water and component cooling water or their cooling water equivalents for the above
13 four monitored systems)
14
15 Data Reporting Elements
16 The following data elements are reported for each system
17
18 * Unavailability Index (UAI) due to unavailability for each monitored system
19 * Unreliability Index (URI) due to unreliability for each monitored system
20
21 During the pilot, the additional data elements necessary to calculate UAI and URI will be
22 reported monthly for each system on an Excel spreadsheet. See Appendix F.
23
24
25 Calculation
26 The MSPI for each system is the sum of the UAI due to unavailability for the system plus URI
27 due to unreliability for the system during the previous twelve quarters.
28
30
31 See Appendix F for the calculational methodology for UAI due to system unavailability and URI
32 due to system unreliability.
33
34 Definition of Terms
35 A train consists of a group of components that together provide the risk significant functions of
36 the system as explained in the additional guidance for specific mitigating systems. Fulfilling the
37 risk-significant function of the system may require one or more trains of a system to operate
38 simultaneously. The number of trains in a system is generally determined as follows:
39
40 * for systems that provide cooling of fluids, the number of trains is determined by the number
41 of parallel heat exchangers, or the number of parallel pumps, or the minimum number of
42 parallel flow paths, whichever is fewer.
43
2
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 * for emergency AC power systems the number of trains is the number of class 1E emergency
2 (diesel, gas turbine, or hydroelectric) generators at the station that are installed to power
3 shutdown loads in the event of a loss of off-site power. (This does not include the diesel
4 generator dedicated to the BWR HPCS system, which is included in the scope of the HPCS
5 system.)
6
7 Risk Significant Functions: those at power functions, described in the Additional Guidance for
8 Specific Systems, that were determined to be risk-significant in accordance with NUMARC 93-
9 01, or NRC approved equivalents (e.g., the STP exemption request.) The system functions
10 described in the Additional Guidance for Specific Systems must be modeled in the plants
11 PRA/PSA. of risk-significant SSCs as modeled in the plant-specific PRA. Risk metrics for
12 identifying risk-significant functions are:
13
14 Risk Achievement Worth > 2.0, or
15 Risk Reduction Worth >0.005, or
16 PRA cutsets that account for 90% of core damage frequency90% of core damage
17 frequency accounted for.
18
19 Risk-Significant Mission Times: The mission time modeled in the PRA for satisfying the risk-
20 significant function of reaching a stable plant condition where normal shutdown cooling is
21 sufficient. Note that PRA models typically analyze an event for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, which may exceed the
22 time needed for the risk-significant function captured in the MSPI. However, other intervals as
23 justified by analyses and modeled in the PRA may be used.
24
25 Success criteria are the plant specific values of parameters the train/system is required to achieve
26 to perform its risk-significant function. Default values of those parameters are the plants design
27 bases values unless other values are modeled in the PRA.
28
29 Clarifying Notes
30 Documentation
31
32 Each licensee will have the system boundaries, active components, risk-significant functions and
33 success criteria readily available for NRC inspection on site. Additionally, plant-specific
34 information used in Appendix F should also be readily available for inspection.
35
36 Success Criteria
37
38 Individual component capability must be evaluated against train/system level success criteria
39 (e.g., a valve stroke time may exceed an ASME requirement, but if the valve still strokes in time
40 to meet the PRA success criteria for the train/system, the component has not failed for the
41 purposes of this indicator because the risk-significant train/system function is still satisfied).
42 Important plant specific performance factors that can be used to identify the required capability
43 of the train/system to meet the risk-significant functions include, but are not limited to:
44 * Actuation
45 o Time
3
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 o Auto/manual
2 o Multiple or sequential
3 * Success requirements
4 o Numbers of components or trains
5 o Flows
6 o Pressures
7 o Heat exchange rates
8 o Temperatures
9 o Tank water level
10 * Other mission requirements
11 o Run time
12 o State/configuration changes during mission
13 * Accident environment from internal events
14 o Pressure, temperature, humidity
15 * Operational factors
16 o Procedures
17 o Human actions
18 o Training
19 o Available externalities (e.g., power supplies, special equipment, etc.)
20
21
22
23 System/Component Interface Boundaries
24
25 For active components that are supported by other components from both monitored and
26 unmonitored systems, the following general rules apply:
27
28 * For control and motive power, only the last relay, breaker or contactor necessary to
29 power or control the component is included in the active component boundary. For
30 example, if an ESFAS signal actuates a MOV, only the relay that receives the ESFAS
31 signal in the control circuitry for the MOV is in the MOV boundary. No other portions
32 of the ESFAS are included.
33
34 * For water connections from systems that provide cooling water to an active component,
35 only the final active connecting valve is included in the boundary. For example, for
36 service water that provides cooling to support an AFW pump, only the final active valve
37 in the service water system that supplies the cooling water to the AFW system is
38 included in the AFW system scope. This same valve is not included in the cooling water
39 support system scope.
40
41 Water Sources and Inventory
42
43 Water tanks are not considered to be active components. As such, they do not contribute to URI.
44 However, periods of insufficient water inventory contribute to UAI if they result in loss of the
45 risk-significant train function for the required mission time. Water inventory can include
46 operator recovery actions for water make-up provided the actions can be taken in time to meet
4
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 the mission times and are modeled in the PRA. If additional water sources are required to satisfy
2 train mission times, only the connecting active valve from the additional water source is
3 considered as an active component for calculating URI. If there are valves in the primary water
4 source that must change state to permit use of the additional water source, these valves are
5 considered active and should be included in URI for the system.
6
7 Monitored Systems
8
9 Systems have been generically selected for this indicator based on their importance in preventing
10 reactor core damage. The systems include the principal systems needed for maintaining reactor
11 coolant inventory following a loss of coolant accident, for decay heat removal following a
12 reactor trip or loss of main feedwater, and for providing emergency AC power following a loss
13 of plant off-site power. One risk-significant support function (cooling water support system) is
14 also monitored. The cooling water support system monitors the risk significant cooling functions
15 provided by service water and component cooling water, or their direct cooling water
16 equivalents, for the four front-line monitored systems. No support systems are to be cascaded
17 onto the monitored systems, e.g., HVAC room coolers, DC power, instrument air, etc.
18
19 Diverse Systems
20
21 Except as specifically stated in the indicator definition and reporting guidance, no credit is given
22 for the achievement of a risk-significant function by an unmonitored system in determining
23 unavailability or unreliability of the monitored systems.
24
25 Common Components
26
27 Some components in a system may be common to more than one train or system, in which case
28 the unavailability/unreliability of a common component is included in all affected trains or
29 systems. (However, see Additional Guidance for Specific Systems for exceptions; for example,
30 the PWR High Pressure Safety Injection System.)
31
32 Short Duration Unavailability
33
34 Trains are generally considered to be available during periodic system or equipment
35 realignments to swap components or flow paths as part of normal operations. Evolutions or
36 surveillance tests that result in less than 15 minutes of unavailable hours per train at a time need
37 not be counted as unavailable hours. Licensees should compile a list of surveillances/evolutions
38 that meet this criterion and have it available for inspector review. In addition, equipment
39 misalignment or mispositioning which is corrected in less than 15 minutes need not be counted
40 as unavailable hours. The intent is to minimize unnecessary burden of data collection,
41 documentation, and verification because these short durations have insignificant risk impact.
42
43 If a licensee is required to take a component out of service for evaluation and corrective actions
44 for greater than 15 minutes (for example, related to a Part 21 Notification), the unavailable hours
45 must be included.
46
5
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 Treatment of Demand /Run Failures and Degraded Conditions
2
3 1. Treatment of Demand and Run Failures
4 Failures of active components (see Appendix F) on demand or failures to run, either
5 actual or test, while critical, are included in unreliability. Failures on demand or failures
6 to run at any other timewith the reactor shutdown must be evaluated to determine if the
7 failure would have resulted in the train not being able to perform its risk-significant at
8 power functions, and must therefore be included in unreliability. Unavailable hours are
9 included only for the time required to recover the trains risk-significant functions and
10 only when the reactor is critical.
11
12 2. Treatment of Degraded Conditions
13
14 a) Capable of Being Discovered By Normal Surveillance Tests
15 Normal surveillance tests are those tests that are performed at a frequency of a
16 refueling cycle or more frequently.
17
18 Degraded conditions, even ifwhere no actual demand existed, that render an
19 active component incapable of performing its risk-significant functions are
20 included in unreliability as a demand and a failure. The appropriate failure mode
21 must be accounted for. For example, for valves, a demand and a demand failure
22 would be assumed and included in URI. For pumps and diesels, if the degraded
23 condition would have prevented a successful start demand, a demand and a failure
24 is included in URI, but there would be no run time hours or run failures. If it was
25 determined that the pump/diesel would start and load run, but would fail
26 sometime during the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> run test or its surveillance test equivalent, the
27 evaluated failure time would be included in run hours and a run failure would be
28 assumed. A start demand and start failure would not be included. If a running
29 component is secured from operation due to observed degraded performance, but
30 prior to failure, then a run failure shall be counted unless evaluation of the
31 condition shows that the component would have continued to operate for the risk-
32 significant mission time starting from the time the component was secured.
33 Unavailable hours are included for the time required to recover the risk-
34 significant function(s).
35
36 Degraded conditions, or actual unavailability due to mispositioning of non-active
37 components that render a train incapable of performing its risk-significant
38 functions are only included in unavailability for the time required to recover the
39 risk-significant function(s).
40
41 Loss of risk significant function(s) is assumed to have occurred if the established
42 success criteria has not been met. If subsequent analysis identifies additional
43 margin for the success criterion, future impacts on URI or UAI for degraded
44 conditions may be determined based on the new criterion. However, URI and
45 UAI must be based on the success criteria of record at the time the degraded
46 condition is discovered. If the degraded condition is not addressed by any of the
6
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 pre-defined success criteria, an engineering evaluation to determine the impact of
2 the degraded condition on the risk-significant function(s) should be completed
3 and documented. The use of component failure analysis, circuit analysis, or event
4 investigations is acceptable. Engineering judgment may be used in conjunction
5 with analytical techniques to determine the impact of the degraded condition on
6 the risk-significant function. The engineering evaluation should be completed as
7 soon as practicable. If it cannot be completed in time to support submission of the
8 PI report for the current quarter, the comment field shall note that an evaluation is
9 pending. The evaluation must be completed in time to accurately account for
10 unavailability/unreliability in the next quarterly report. Exceptions to this
11 guidance are expected to be rare and will be treated on a case-by-case basis.
12 Licensees should identify these situations to the resident inspector.
13
14 b) Not Capable of Being Discovered by Normal Surveillance Tests
15 These failures or conditions are usually of longer exposure time. Since these
16 failure modes have not been tested on a regular basis, it is inappropriate to include
17 them in the performance index statistics. These failures or conditions are subject
18 to evaluation through the inspection process. Examples of this type are failures
19 due to pressure locking/thermal binding of isolation valves, blockages in lines not
20 regularly tested, or inadequate component sizing/settings under accident
21 conditions (not under normal test conditions). While not included in the
22 calculation of the index, they should be reported in the comment field of the PI
23 data submittal.
24
25
26 Credit for Operator Recovery Actions to Restore the Risk-Significant Function
27
28 1. During testing or operational alignment:
29 Unavailability of a risk-significant function during testing or operational alignment need not
30 be included if the test configuration is automatically overridden by a valid starting signal, or
31 the function can be promptly restored either by an operator in the control room or by a
32 designated operator1 stationed locally for that purpose. Restoration actions must be
33 contained in a written procedure2, must be uncomplicated (a single action or a few simple
34 actions), must be capable of being restored in time to satisfy PRA success criteria and must
35 not require diagnosis or repair. Credit for a designated local operator can be taken only if
36 (s)he is positioned at the proper location throughout the duration of the test for the purpose of
37 restoration of the train should a valid demand occur. The intent of this paragraph is to allow
38 licensees to take credit for restoration actions that are virtually certain to be successful (i.e.,
39 probability nearly equal to 1) during accident conditions.
40
1 Operator in this circumstance refers to any plant personnel qualified and designated to perform
the restoration function.
2 Including restoration steps in an approved test procedure.
7
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 The individual performing the restoration function can be the person conducting the test and
2 must be in communication with the control room. Credit can also be taken for an operator in
3 the main control room provided (s)he is in close proximity to restore the equipment when
4 needed. Normal staffing for the test may satisfy the requirement for a dedicated operator,
5 depending on work assignments. In all cases, the staffing must be considered in advance and
6 an operator identified to perform the restoration actions independent of other control room
7 actions that may be required.
8
9 Under stressful, chaotic conditions, otherwise simple multiple actions may not be
10 accomplished with the virtual certainty called for by the guidance (e.g., lifting test leads and
11 landing wires; or clearing tags). In addition, some manual operations of systems designed to
12 operate automatically, such as manually controlling HPCI turbine to establish and control
13 injection flow, are not virtually certain to be successful. These situations should be resolved
14 on a case-by-case basis through the FAQ process.
15
16 2. During Maintenance
17 Unavailability of a risk-significant function during maintenance need not be included if the
18 risk-significant function can be promptly restored either by an operator in the control room or
19 by a designated operator3 stationed locally for that purpose. Restoration actions must be
20 contained in a written procedure4, must be uncomplicated (a single action or a few simple
21 actions), must be capable of being restored in time to satisfy PRA success criteria and must
22 not require diagnosis or repair. Credit for a designated local operator can be taken only if
23 (s)he is positioned at a proper location throughout the duration of the maintenance activity
24 for the purpose of restoration of the train should a valid demand occur. The intent of this
25 paragraph is to allow licensees to take credit for restoration of risk-significant functions that
26 are virtually certain to be successful (i.e., probability nearly equal to 1). The individual
27 performing the restoration function can be the person performing the maintenance and must
28 be in communication with the control room. Credit can also be taken for an operator in the
29 main control room provided (s)he is in close proximity to restore the equipment when
30 needed. Under stressful chaotic conditions otherwise simple multiple actions may not be
31 accomplished with the virtual certainty called for by the guidance (e.g., lifting test leads and
32 landing wires, or clearing tags). These situations should be resolved on a case-by-case basis
33 through the FAQ process.
34
35 3. Satisfying PRA success criteriaRisk Significant Mission Times
36 Risk significant operator actions to satisfy pre-determined train/system risk-significant
37 mission times can only be credited if they are modeled in the PRA.
38
39 Swing trains and components shared between units
40
3 Operator in this circumstance refers to any plant personnel qualified and designated to perform the
restoration function.
4 Including restoration steps in an approved test procedure.
8
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 Swing trains/components are trains/components that can be aligned to any unit. To be credited
2 as such, their swing capability should be modeled in the PRA to provide an appropriate Fussell-
3 Vesely value.
4
5 Unit Cross Tie Capability
6
7 Components that cross tie monitored systems between units should be considered active
8 components if they are modeled in the PRA and meet the active component criteria in Appendix
9 F. Such active components are counted in each units performance indicators.
10
11 Maintenance Trains and Installed Spares
12
13 Some power plants have systems with extra trains to allow preventive maintenance to be carried
14 out with the unit at power without impacting the risk-significant function of the system. That is,
15 one of the remaining trains may fail, but the system can still perform its risk significant function.
16 To be a maintenance train, a train must not be needed to perform the systems risk significant
17 function.
18
19 An "installed spare" is a component (or set of components) that is used as a replacement for other
20 equipment to allow for the removal of equipment from service for preventive or corrective
21 maintenance without impacting the risk-significant function of the system. To be an "installed
22 spare," a component must not be needed for the system to perform the risk significant function.
23
24
25 For unreliability, spare active components are included if they are modeled in the PRA.
26 Unavailability of the spare component/train is only counted in the index if the spare is substituted
27 for a primary train/component. Unavailability is not monitored for a component/train when that
28 component/train has been replaced by an installed spare or maintenance train.
29
30 Use of Plant-Specific PRA and SPAR Models
31
32 The MSPI is an approximation using some information from a plants actual PRA and is
33 intended as an indicator of system performance. Plant-specific PRAs and SPAR models cannot
34 be used to question the outcome of the PIs computed in accordance with this guideline.
35
36 Maintenance Rule Performance Monitoring
37
38 It is the intent that NUMARC 93-01 be revised to require consistent unavailability and
39 unreliability data gathering as required by this guideline.
40
9
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 ADDITIONAL GUIDANCE FOR SPECIFIC SYSTEMS
2 This guidance provides typical system scopes. Individual plants should include those systems
3 employed at their plant that are necessary to satisfy the specific risk-significant functions
4 described below and reflected in their PRAs.
5 Emergency AC Power Systems
6 Scope
7 The function monitored for the emergency AC power system is the ability of the emergency
8 generators to provide AC power to the class 1E buses upon a loss of off-site power while the
9 reactor is critical, including post-accident conditions. The emergency AC power system is
10 typically comprised of two or more independent emergency generators that provide AC power to
11 class 1E buses following a loss of off-site power. The emergency generator dedicated to
12 providing AC power to the high pressure core spray system in BWRs is not within the scope of
13 emergency AC power.
14
15 The electrical circuit breaker(s) that connect(s) an emergency generator to the class lE buses that
16 are normally served by that emergency generator are considered to be part of the emergency
17 generator train.
18
19 Emergency generators that are not safety grade, or that serve a backup role only (e.g., an
20 alternate AC power source), are not included in the performance reporting.
21
22 Train Determination
23 The number of emergency AC power system trains for a unit is equal to the number of class 1E
24 emergency generators that are available to power safe-shutdown loads in the event of a loss of
25 off-site power for that unit. There are three typical configurations for EDGs at a multi-unit
26 station:
27
28 1. EDGs dedicated to only one unit.
29 2. One or more EDGs are available to swing to either unit
30 3. All EDGs can supply all units
31
32 For configuration 1, the number of trains for a unit is equal to the number of EDGs dedicated to
33 the unit. For configuration 2, the number of trains for a unit is equal to the number of dedicated
34 EDGs for that unit plus the number of swing EDGs available to that unit (i.e., The swing
35 EDGs are included in the train count for each unit). For configuration 3, the number of trains is
36 equal to the number of EDGs.
37
38 Clarifying Notes
39 The emergency diesel generators are not considered to be available during the following portions
40 of periodic surveillance tests unless recovery from the test configuration during accident
41 conditions is virtually certain, as described in Credit for operator recovery actions during
10
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 testing, can be satisfied; or the duration of the condition is less than fifteen minutes per train at
2 one time:
3
4 * Load-run testing
5 * Barring
6
7 An EDG is not considered to have failed due to any of the following events:
8
9 * spurious operation of a trip that would be bypassed in a loss of offsite power event
10 * malfunction of equipment that is not required to operate during a loss of offsite power event
11 (e.g., circuitry used to synchronize the EDG with off-site power sources)
12 * failure to start because a redundant portion of the starting system was intentionally disabled
13 for test purposes, if followed by a successful start with the starting system in its normal
14 alignment
15 Air compressors are not part of the EDG boundary. However, air receivers that provide starting
16 air for the diesel are included in the EDG boundary.
17
18 If an EDG has a dedicated battery independent of the stations normal DC distribution system,
19 the dedicated battery is included in the EDG system boundary.
20
21 If the EDG day tank is not sufficient to meet the EDG mission time, the fuel transfer function
22 should be modeled in the PRA. However, the fuel transfer pumps are not considered to be an
23 active component in the EDG system because they are considered to be a support system.
24
25
26
27 BWR High Pressure Injection Systems
28 (High Pressure Coolant Injection, High Pressure Core Spray, and Feedwater Coolant
29 Injection)
30
31 Scope
32 These systems function at high pressure to maintain reactor coolant inventory and to remove
33 decay heat following a small-break Loss of Coolant Accident (LOCA) event or a loss of main
34 feedwater event.
35
36 The function monitored for the indicator is the ability of the monitored system to take suction
37 from the suppression pool (and from the condensate storage tank, if credited in the plants
38 accident analysis) and inject into the reactor vessel.
39
40 Plants should monitor either the high-pressure coolant injection (HPCI), the high-pressure core
41 spray (HPCS), or the feedwater coolant injection (FWCI) system, whichever is installed. The
42 turbine and governor (or motor-driven FWCI pumps), and associated piping and valves for
43 turbine steam supply and exhaust are within the scope of these systems. Valves in the feedwater
44 line are not considered within the scope of these systems. The emergency generator dedicated to
11
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 providing AC power to the high-pressure core spray system is included in the scope of the
2 HPCS. The HPCS system typically includes a "water leg" pump to prevent water hammer in the
3 HPCS piping to the reactor vessel. The "water leg" pump and valves in the "water leg" pump
4 flow path are ancillary components and are not included in the scope of the HPCS system.
5 Unavailability is not included while critical if the system is below steam pressure specified in
6 technical specifications at which the system can be operated.
7
8 Train Determination
9 The HPCI and HPCS systems are considered single-train systems. The booster pump and other
10 small pumps are ancillary components not used in determining the number of trains. The effect
11 of these pumps on system performance is included in the system indicator to the extent their
12 failure detracts from the ability of the system to perform its risk-significant function. For the
13 FWCI system, the number of trains is determined by the number of feedwater pumps. The
14 number of condensate and feedwater booster pumps are not used to determine the number of
15 trains.
16
17 BWR Heat Removal Systems
18 (Reactor Core Isolation Cooling or Isolation Condenser)
19
20 Scope
21 This system functions at high pressure to remove decay heat following a loss of main feedwater
22 event. The RCIC system also functions to maintain reactor coolant inventory following a very
23 small LOCA event.
24
25 The function monitored for the indicator is the ability of the RCIC system to cool the reactor
26 vessel core and provide makeup water by taking a suction from either the condensate storage
27 tank or the suppression pool and injecting at rated pressure and flow into the reactor vessel.
28
29 The Reactor Core Isolation Cooling (RCIC) system turbine, governor, and associated piping and
30 valves for steam supply and exhaust are within the scope of the RCIC system. Valves in the
31 feedwater line are not considered within the scope of the RCIC system. The Isolation Condenser
32 and inlet valves are within the scope of Isolation Condenser system. Unavailability is not
33 included while critical if the system is below steam pressure specified in technical specifications
34 at which the system can be operated.
35
36
37 Train Determination
38 The RCIC system is considered a single-train system. The condensate and vacuum pumps are
39 ancillary components not used in determining the number of trains. The effect of these pumps on
40 RCIC performance is included in the system indicator to the extent that a component failure
41 results in an inability of the system to perform its risk-significant function.
12
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1
2 BWR Residual Heat Removal Systems
3 Scope
4 The functions monitored for the BWR residual heat removal (RHR) system are the ability of the
5 RHR system to remove heat from the suppression pool, provide low pressure coolant injection,
6 and provide post-accident decay heat removal. The pumps, heat exchangers, and associated
7 piping and valves for those functions are included in the scope of the RHR system.
8
9 Train Determination
10 The number of trains in the RHR system is determined by the number of parallel RHR heat
11 exchangers.
12
13 PWR High Pressure Safety Injection Systems
14 Scope
15 These systems are used primarily to maintain reactor coolant inventory at high pressures
16 following a loss of reactor coolant. HPSI system operation following a small-break LOCA
17 involves transferring an initial supply of water from the refueling water storage tank (RWST) to
18 cold leg piping of the reactor coolant system. Once the RWST inventory is depleted,
19 recirculation of water from the reactor building emergency sump is required. The function
20 monitored for HPSI is the ability of a HPSI train to take a suction from the primary water source
21 (typically, a borated water tank), or from the containment emergency sump, and inject into the
22 reactor coolant system at rated flow and pressure.
23
24 The scope includes the pumps and associated piping and valves from both the refueling water
25 storage tank and from the containment sump to the pumps, and from the pumps into the reactor
26 coolant system piping. For plants where the high-pressure injection pump takes suction from the
27 residual heat removal pumps, the residual heat removal pump discharge header isolation valve to
28 the HPSI pump suction is included in the scope of HPSI system. Some components may be
29 included in the scope of more than one train. For example, cold-leg injection lines may be fed
30 from a common header that is supplied by both HPSI trains. In these cases, the effects of testing
31 or component failures in an injection line should be reported in both trains.
32
33 Train Determination
34
35 In general, the number of HPSI system trains is defined by the number of high head injection
36 paths that provide cold-leg and/or hot-leg injection capability, as applicable.
37
38 For Babcock and Wilcox (B&W) reactors, the design features centrifugal pumps used for high
39 pressure injection (about 2,500 psig) and no hot-leg injection path. Recirculation from the
40 containment sump requires operation of pumps in the residual heat removal system. They are
13
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 typically a two-train system, with an installed spare pump (depending on plant-specific design)
2 that can be aligned to either train.
3
4 For two-loop Westinghouse plants, the pumps operate at a lower pressure (about 1600 psig) and
5 there may be a hot-leg injection path in addition to a cold-leg injection path (both are included as
6 a part of the train).
7
8 For Combustion Engineering (CE) plants, the design features three centrifugal pumps that
9 operate at intermediate pressure (about 1300 psig) and provide flow to two cold-leg injection
10 paths or two hot-leg injection paths. In most designs, the HPSI pumps take suction directly from
11 the containment sump for recirculation. In these cases, the sump suction valves are included
12 within the scope of the HPSI system. This is a two-train system (two trains of combined cold-leg
13 and hot-leg injection capability). One of the three pumps is typically an installed spare that can
14 be aligned to either train or only to one of the trains (depending on plant-specific design).
15
16 For Westinghouse three-loop plants, the design features three centrifugal pumps that operate at
17 high pressure (about 2500 psig), a cold-leg injection path through the BIT (with two trains of
18 redundant valves), an alternate cold-leg injection path, and two hot-leg injection paths. One of
19 the pumps is considered an installed spare. Recirculation is provided by taking suction from the
20 RHR pump discharges. A train consists of a pump, the pump suction valves and boron injection
21 tank (BIT) injection line valves electrically associated with the pump, and the associated hot-leg
22 injection path. The alternate cold-leg injection path is required for recirculation, and should be
23 included in the train with which its isolation valve is electrically associated. This represents a
24 two-train HPSI system.
25
26 For Four-loop Westinghouse plants, the design features two centrifugal pumps that operate at
27 high pressure (about 2500 psig), two centrifugal pumps that operate at an intermediate pressure
28 (about 1600 psig), a BIT injection path (with two trains of injection valves), a cold-leg safety
29 injection path, and two hot-leg injection paths. Recirculation is provided by taking suction from
30 the RHR pump discharges. Each of two high pressure trains is comprised of a high pressure
31 centrifugal pump, the pump suction valves and BIT valves that are electrically associated with
32 the pump. Each of two intermediate pressure trains is comprised of the safety injection pump, the
33 suction valves and the hot-leg injection valves electrically associated with the pump. The cold-
34 leg safety injection path can be fed with either safety injection pump, thus it should be associated
35 with both intermediate pressure trains. This HPSI system is considered a four-train system for
36 monitoring purposes.
37
38
39
40 PWR Auxiliary Feedwater Systems
41 Scope
42 The AFW system provides decay heat removal via the steam generators to cool down and
43 depressurize the reactor coolant system following a reactor trip. The AFW system is assumed to
44 be required for an extended period of operation during which the initial supply of water from the
45 condensate storage tank is depleted and water from an alternative water source (e.g., the service
46 water system) is required. Therefore components in the flow paths from both of these water
14
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 sources are included; however, the alternative water source (e.g., service water system) is not
2 included.
3
4 The function monitored for the indicator is the ability of the AFW system to take a suction from
5 the primary water source (typically, the condensate storage tank) or, if required, from an
6 emergency source (typically, a lake or river via the service water system) and inject into at least
7 one steam generator at rated flow and pressure.
8
9 The scope of the auxiliary feedwater (AFW) or emergency feedwater (EFW) systems includes
10 the pumps and the components in the flow paths from the condensate storage tank and, if
11 required, the valve(s) that connect the alternative water source to the auxiliary feedwater system.
12 Startup feedwater pumps are not included in the scope of this indicator.
13
14 Train Determination
15 The number of trains is determined primarily by the number of parallel pumps. For example, a
16 system with three pumps is defined as a three-train system, whether it feeds two, three, or four
17 injection lines, and regardless of the flow capacity of the pumps. Some components may be
18 included in the scope of more than one train. For example, one set of flow regulating valves and
19 isolation valves in a three-pump, two-steam generator system are included in the motor-driven
20 pump train with which they are electrically associated, but they are also included (along with the
21 redundant set of valves) in the turbine-driven pump train. In these instances, the effects of testing
22 or failure of the valves should be reported in both affected trains. Similarly, when two trains
23 provide flow to a common header, the effect of isolation or flow regulating valve failures in
24 paths connected to the header should be considered in both trains.
25
26 PWR Residual Heat Removal System
27 Scope
28 The functions monitored for the PWR residual heat removal (RHR) system are those that are
29 required to be available when the reactor is critical. These typically include the low-pressure
30 injection function (if risk-significant) and the post-accident recirculation mode used to cool and
31 recirculate water from the containment sump following depletion of RWST inventory to provide
32 post-accident decay heat removal. The pumps, heat exchangers, and associated piping and valves
33 for those functions are included in the scope of the RHR system. Containment spray function
34 should be included if it is identified in the PRA as a risk-significant post accident decay heat
35 removal function. Containment spray systems that only provide containment pressure control are
36 not included.
37
38
39
40 Train Determination
41 The number of trains in the RHR system is determined by the number of parallel RHR heat
42 exchangers. Some components are used to provide more than one function of RHR. If a
43 component cannot perform as designed, rendering its associated train incapable of meeting one
15
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 of the risk-significant functions, then the train is considered to be failed. Unavailable hours
2 would be reported as a result of the component failure.
3 Cooling Water Support System
4 Scope
5 The function of the cooling water support system is to provide for direct cooling of the
6 components in the other monitored systems. It does not include indirect cooling provided by
7 room coolers or other HVAC features.
8
9 Systems that provide this function typically include service water and component cooling water
10 or their cooling water equivalents. Pumps, valves, heat exchangers and line segments that are
11 necessary to provide cooling to the other monitored systems are included in the system scope up
12 to, but not including, the last valve that connects the cooling water support system to the other
13 monitored systems. This last valve is included in the other monitored system boundary.
14
15 Valves in the cooling water support system that must close to ensure sufficient cooling to the
16 other monitored system components to meet risk significant functions are included in the system
17 boundary.
18
19
20
21 Train Determination
22 The number of trains in the Cooling Water Support System will vary considerably from plant to
23 plant. The way these functions are modeled in the plant-specific PRA will determine a logical
24 approach for train determination. For example, if the PRA modeled separate pump and line
25 segments, then the number of pumps and line segments would be the number of trains.
26
27 Clarifying Notes
28 Service water pump strainers and traveling screens are not considered to be active components
29 and are therefore not part of URI. However, clogging of strainers and screens due to expected or
30 routinely predictable environmental conditions that render the train unavailable to perform its
31 risk significant cooling function (which includes the risk-significant mission times)are included
32 in UAI.
33
34 Unpredictable extreme environmental conditions that render the train unavailable to perform its
35 risk significant cooling function should be addressed through the FAQ process to determine if
36 resulting unavailability should be included in UAI.
37
16
Attachment 2
NEI 99-02, Appendix F, Methodologies For Computing the Unavailability Index, the
Unreliability Index and Determining Performance Index Validity (Draft).
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 APPENDIX F
2
3 METHODOLOGIES FOR COMPUTING THE UNAVAILABILITY
4 INDEX, THE UNRELIABILITY INDEX AND DETERMINING
5 PERFORMANCE INDEX VALIDITY
6 This appendix provides the details of three calculations, calculation of the System
7 Unavailability Index, the System Unreliability Index, and the criteria for determining
8 when the Mitigating System Performance Index is unsuitable for use as a performance
9 index.
10 System Unavailability Index (UAI) Due to Changes in Train Unavailability
11 Calculation of System UAI due to changes in train unavailability is as follows:
n
12 UAI = UAItj Eq. 1
j =1
13 where the summation is over the number of trains (n) and UAIt is the unavailability index
14 for a train.
15 Calculation of UAIt for each train due to changes in train unavailability is as follows:
é FVUAp
16 UAIt = CDFp ú (UAt - UABLt ) , Eq. 2
UAp û max
17 where:
18 CDFp is the plant-specific, internal events, at power Core Damage Frequency,
19 FVUAp is the train-specific Fussell-Vesely value for unavailability,
20 UAP is the plant-specific PRA value of unavailability for the train,
21 UAt is the actual unavailability of train t, defined as:
Unavailable hours during the previous 12 quarters while critical
22 UAt =
Critical hours during the previous 12 quarters
23 and,
24 UABLt is the historical baseline unavailability value for the train determined
25 as described below.
26 UABLt is the sum of two elements: planned and unplanned unavailability. Planned
27 unavailability is the actual, plant-specific three-year total planned unavailability
28 for the train for the years 1999 through 2001 (see clarifying notes for details).
29 This period is chosen as the most representative of how the plant intends to
30 perform routine maintenance and surveillances at power. Unplanned
31 unavailability is the historical industry average for unplanned unavailability for
F-1
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 the years 1999 through 2001. See Table 1 for historical train values for
2 unplanned unavailability.
3 Calculation of the quantity inside the square bracket in equation 2 will be discussed at the
4 end of the next section. See clarifying notes for calculation of UAI for cooling water
5 support system.
6
7 System Unreliability Index (URI) Due to Changes in Component Unreliability
8 Unreliability is monitored at the component level and calculated at the system level.
9 Calculation of system URI due to changes in component unreliability is as follows:
m é FVURcj
10 URI = CDFp ú (URBcj - URBLcj ) Eq. 3
j =1 URpcj û max
11 Where the summation is over the number of active components (m) in the system, and:
12 CDFp is the plant-specific internal events, at power, core damage frequency,
13 FVURc is the component-specific Fussell-Vesely value for unreliability,
14 URPc is the plant-specific PRA value of component unreliability,
15 URBc is the Bayesian corrected component unreliability for the previous 12
16 quarters,
17 and
18 URBLc is the historical industry baseline calculated from unreliability mean values
19 for each monitored component in the system. The calculation is performed in a
20 manner similar to equation 4 below using the industry average values in Table 2.
21 Calculation of the quantity inside the square bracket in equation 3 will be discussed at the
22 end of this section.
23 Component unreliability is calculated as follows.
24 URBc = P D + lTm Eq 4
25 where:
26 PD is the component failure on demand probability calculated based on data
27 collected during the previous 12 quarters,
28 l is the component failure rate (per hour) for failure to run calculated based on
29 data collected during the previous 12 quarters,
30 and
31 Tm is the risk-significant mission time for the component based on plant specific
32 PRA model assumptions. Add acceptable methodologies for determining mission
33 time.
34 F-2
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 NOTE:
2 For valves only the PD term applies
3 For pumps PD + l Tm applies
4 For diesels PD start + PD load run + l Tm applies
5
6 The first term on the right side of equation 4 is calculated as follows.1
(Nd + a)
7 PD = Eq. 5
(a + b + D)
8 where:
9 Nd is the total number of failures on demand during the previous 12 quarters,
10 D is the total number of demands during the previous 12 quarters (actual ESF
11 demands plus estimated test and estimated operational/alignment demands. An
12 update to the estimated demands is required if a change to the basis for the
13 estimated demands results in a >25% change in the estimate),
14 and
15 a and b are parameters of the industry prior, derived from industry experience (see
16 Table 2).
17 In the calculation of equation 5 the numbers of demands and failures is the sum of all
18 demands and failures for similar components within each system. Do not sum across
19 units for a multi-unit plant. For example, for a plant with two trains of Emergency Diesel
20 Generators, the demands and failures for both trains would be added together for one
21 evaluation of PD which would be used for both trains of EDGs.
22 In the second term on the right side of equation 4, l is calculated as follows.
(Nr + a)
23 l= Eq. 6
(T r + b)
24 where:
25 Nr is the total number of failures to run during the previous 12 quarters,
26 Tr is the total number of run hours during the previous 12 quarters (actual ESF run
27 hours plus estimated test and estimated operational/alignment run hours. An
28 update to the estimated run hours is required if a change to the basis for the
29 estimated hours results in a >25% change in the estimate).
30 and
1 Atwood, Corwin L., Constrained noninformative priors in risk assessment, Reliability
Engineering and System Safety, 53 (1996; 37-46)
F-3
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 a and b are parameters of the industry prior, derived from industry experience (see
2 Table 2).
3 In the calculation of equation 6 the numbers of demands and run hours is the sum of all
4 run hours and failures for similar components within each system. Do not sum across
5 units for a multi-unit plant. For example, a plant with two trains of Emergency Diesel
6 Generators, the run hours and failures for both trains would be added together for one
7 evaluation of l which would be used for both trains of EDGs.
8 Fussell-Vesely, Unavailability and Unreliability
9 Equations 2 and 3 include a term that is the ratio of a Fussell-Vesely importance value
10 divided by the related unreliability or unavailability. Calculation of these quantities is
11 generally complex, but in the specific application used here, can be greatly simplified.
12 The simplifying feature of this application is that only those components (or the
13 associated basic events) that can fail a train are included in the performance index.
14 Components within a train that can each fail the train are logically equivalent and the
15 ratio FV/UR is a constant value for any basic event in that train. It can also be shown that
16 for a given component or train represented by multiple basic events, the ratio of the two
17 values for the component or train is equal to the ratio of values for any basic event within
18 the train. Or:
FVbe FVURc FVt
19 = = = Constant
URbe URPc URt
20 and
FVbe FVUAp
21 = = Constant
UAbe UAp
22 Note that the constant value may be different for the unreliability ratio and the
23 unavailability ratio because the two types of events are frequently not logically
24 equivalent. For example recovery actions may be modeled in the PRA for one but not the
25 other.
26 Thus, the process for determining the value of this ratio for any component or train is to
27 identify a basic event that fails the component or train, determine the failure probability
28 or unavailability for the event, determine the associated FV value for the event and then
29 calculate the ratio. Use the basic event in the component or train with the largest failure
30 probability (hence the maximum notation on the bracket) to minimize the effects of
31 truncation on the calculation. Exclude common cause events, which are not within the
32 scope of this performance index
33 Some systems have multiple modes of operation, such as PWR HPSI systems that operate
34 in injection as well as recirculation modes. In these systems all active components are not
35 logically equivalent, unavailability of the pump fails all operating modes while
36 unavailability of the sump suction valves only fails the recirculation mode. In cases such
F-4
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 as these, if unavailability events exist separately for the components within a train, the
2 appropriate ratio to use is the maximum.
3 Determination of systems for which the performance index is not valid
4 The performance index relies on the existing testing programs as the source of the data
5 that is input to the calculations. Thus, the number of demands in the monitoring period is
6 based on the frequency of testing required by the current test programs. In most cases this
7 will provide a sufficient number of demands to result in a valid statistical result.
8 However, in some cases, the number of demands will be insufficient to resolve the
9 change in the performance index (1.0x10-6) that corresponds to movement from a green
10 performance to a white performance level. In these cases, one failure is the difference
11 between baseline performance and performance in the white performance band. The
12 performance index is not suitable for monitoring such systems and monitoring is
13 performed through the inspection process.
14 This section will define the method to be used to identify systems for which the
15 performance index is not valid, and will not be used.
16 The criteria to be used to identify an invalid performance index is:
17 If, for any failure mode for any component in a system, the risk increase
18 (DCDF) associated with the change in unreliability resulting from single
19 failure is larger than 1.0x10-6, then the performance index will be
20 considered invalid for that system.
21 The increase in risk associated with a component failure is the sum of the contribution
22 from the decrease in calculated reliability as a result of the failure and the decrease in
23 availability resulting from the time required to affect the repair of the failed component.
24 The change in CDF that results from a demand type failure is given by:
25
FVURc 1 ü
MSPI = CDFp ' í
N similar comp URpc
' ý
a + b + D
26 Eq. 7
FVUAp TMean Repair
+ CDFp ' '
UAp TCR
27
28 Likewise, the change in CDF per run type failure is given by:
29
FVURc Tm ü
MSPI = CDFp ' í
N similar comp URpc
' ý
b + Tr
30 Eq. 8
FVUAp TMean Repair
+ CDFp ' '
UAp TCR
F-5
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 In these expressions, the variables are as defined earlier and additionally
2 TMR is the mean time to repair for the component
3 and
4 TCR is the number of critical hours in the monitoring period.
5 The summation in the equations is taken over all similar components within a system.
6 With multiple components of a given type in one system, the impact of the failure on
7 CDF is included in the increased unavailability of all components of that type due to
8 pooling the demand and failure data.
9 The mean time to repair can be estimate as one-half the Technical Specification Allowed
10 Outage Time for the component and the number of critical hours should correspond to the
11 1999 - 2001 actual number of critical hours.
12 These equations are be used for all failure modes for each component in a system. If the
13 resulting value of DCDF is greater than 1.0x10-6 for any failure mode of any component,
14 then the performance index for that system is not considered valid.
15
16 Definitions
17
18 Train Unavailability: Train unavailability is the ratio of the hours the train was
19 unavailable to perform its risk-significant functions due to planned or unplanned
20 maintenance or test during the previous 12 quarters while critical to the number of critical
21 hours2.430556e-4 days <br />0.00583 hours <br />3.472222e-5 weeks <br />7.9905e-6 months <br /> during the previous 12 quarters. (Fault exposure hours are not included;
22 unavailable hours are counted only for the time required to recover the trains risk-
23 significant functions.)
24 Train unavailable hours: The hours the train was not able to perform its risk significant
25 function due to maintenance, testing, equipment modification, electively removed from
26 service, corrective maintenance, or the elapsed time between the discovery and the
27 restoration to service of an equipment failure or human error that makes the train
28 unavailable (such as a misalignment) while the reactor is critical.
29 Fussell-Vesely (FV) Importance:
30 The Fussell-Vesely importance for a feature (component, sub-system, train, etc.) of a
31 system is representative of the fractional contribution that feature makes to the to the total
32 risk of the system.
33 The Fussell-Vesely importance of a basic event or group of basic events that represent a
34 feature of a system is represented by:
Ri
35 FV = 1 -
R0
F-6
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 Where:
2 R0 is the base (reference) case overall model risk,
3 Ri is the decreased risk level with feature i completely reliable.
4 In this expression, the second term on the right represents the fraction of the reference
5 risk remaining assuming the feature of interest is perfect. Thus 1 minus the second term is
6 the fraction of the reference risk attributed to the feature of interest.
7 The Fussell-Vesely importance is calculated according to the following equation:
j =1, n
i j
8 FV = 1 - ,
j =1, m
0j
9 where the denominator represents the union of m minimal cutsets C0 generated with the
10 reference (baseline) model, and the numerator represents the union of n minimal cutsets
11 Ci generated assuming events related to the feature are perfectly reliable, or their failure
12 probability is False.
13 Critical hours: The number of hours the reactor was critical during a specified period of
14 time.
15 Component Unreliability: Component unreliability is the probability that the component
16 would not perform its risk-significant functions when called upon during the previous 12
17 quarters.
18 Active Component: A component whose failure to change state renders the train incapable
19 of performing its risk-significant functions. In addition, all pumps and diesels in the
20 monitored systems are included as active components. (See clarifying notes.)
21 Manual Valve: A valve that can only be operated by a person. An MOV or AOV that is
22 remotely operated by a person may be an active component.
23 Start demand: Any demand for the component to successfully start to perform its risk-
24 significant functions, actual or test. (Exclude post maintenance tests, unless in case of a
25 failure the cause of failure was independent of the maintenance performed.)
26 Post maintenance tests: Tests performed following maintenance but prior to declaring the
27 train/component operable, consistent with Maintenance Rule implementation.
28 Run demand: Any demand for the component, given that it has successfully started, to
29 run/operate for its mission time to perform its risk-significant functions. (Exclude post
30 maintenance tests, unless in case of a failure the cause of failure was independent of the
31 maintenance performed.)
32 EDG failure to start: A failure to start includes those failures up to the point the EDG has
33 achieved rated speed and voltage. (Exclude post maintenance tests, unless the cause of
34 failure was independent of the maintenance performed.)
F-7
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 EDG failure to load/run: Given that it has successfully started, a failure of the EDG
2 output breaker to close, loads successfully sequence and to run/operate for one hour to
3 perform its risk-significant functions. This failure mode is treated as a demand failure for
4 calculation purposes. (Exclude post maintenance tests, unless the cause of failure was
5 independent of the maintenance performed.)
6 EDG failure to run: Given that it has successfully started and loaded and run for an hour,
7 a failure of an EDG to run/operate. for its mission time to perform its risk-significant
8 functions. (Exclude post maintenance tests, unless the cause of failure was independent of
9 the maintenance performed.)
10 Pump failure on demand: A failure to start and run for at least one hour is counted as
11 failure on demand. (Exclude post maintenance tests, unless the cause of failure was
12 independent of the maintenance performed.)
13 Pump failure to run: Given that it has successfully started and run for an hour, a failure of
14 a pump to run/operate. for its mission time to perform its risk-significant functions.
15 (Exclude post maintenance tests, unless the cause of failure was independent of the
16 maintenance performed.)
17 Valve failure on demand: A failure to open or close is counted as failure on demand.
18 (Exclude post maintenance tests, unless the cause of failure was independent of the
19 maintenance performed.)
20 Clarifying Notes
21 Train Boundaries and Unavailable Hours
22 Include all components that are required to satisfy the risk-significant function of the
23 train. For example, high-pressure injection may have both an injection mode with
24 suction from the refueling water storage tank and a recirculation mode with suction from
25 the containment sump. Some components may be included in the scope of more than one
26 train. For example, one set of flow regulating valves and isolation valves in a three-pump,
27 two-steam generator system are included in the motor-driven pump train with which they
28 are electrically associated, but they are also included (along with the redundant set of
29 valves) in the turbine-driven pump train. In these instances, the effects of unavailability
30 of the valves should be reported in both affected trains. Similarly, when two trains
31 provide flow to a common header, the effect of isolation or flow regulating valve failures
32 in paths connected to the header should be considered in both trains
33 Cooling Water Support System Trains
34 The number of trains in the Cooling Water Support System will vary considerably from
35 plant to plant. The way these functions are modeled in the plant-specific PRA will
36 determine a logical approach for train determination. For example, if the PRA modeled
37 separate pump and line segments, then the number of pumps and line segments would be
38 the number of trains. A separate value for UAI and URI will be calculated for each of the
39 systems in this indicator and then they will be added together to calculate the MSPI.
F-8
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1
2 Active Components
3 For unreliability, use the following criteria for determining those components that should
4 be monitored:
5 * Components that are normally running or have to change state to achieve the risk
6 significant function will be included in the performance index. Active failures of
7 check valves and manual valves are excluded from the performance index and will be
8 evaluated in the NRC inspection program.
9 * Redundant valves within a train are not included in the performance index. Only
10 those valves whose failure alone can fail a train will be included. The PRA success
11 criteria are to be used to identify these valves.
12 * Redundant valves within a multi-train system, whether in series or parallel, where the
13 failure of both valves would prevent all trains in the system from performing a risk-
14 significant function are included. (See Figure F-5)
15 * All pumps and diesels are included in the performance index
16 Table 3 defines the boundaries of components, and Figures F-1, F-2, F-3 and F-4 provide
17 examples of typical component boundaries as described in Table 3. Each plant will
18 determine their system boundaries, active components, and support components, and
19 have them available for NRC inspection.
20 Failures of Non-Active Components
21 Failures of SSCs that are not included in the performance index will not be counted as a
22 failure or a demand. Failures of SSCs that cause an SSC within the scope of the
23 performance index to fail will not be counted as a failure or demand. An example could
24 be a manual suction isolation valve left closed which causes a pump to fail. This would
25 not be counted as a failure of the pump. Any mispositioning of the valve that caused the
26 train to be unavailable would be counted as unavailability from the time of discovery.
27 The significance of the mispositioned valve prior to discovery would be addressed
28 through the inspection process.
29
30 Baseline Values
31 The baseline values for unreliability are contained in Table 2 and remain fixed.
32 The baseline values for unavailability include both plant-specific planned unavailability
33 values and unplanned unavailability values. The unplanned unavailability values are
34 contained in Table 1 and remain fixed. They are based on ROP PI industry data from
35 1999 through 2001. (Most baseline data used in PIs come from the 1995-1997 time
36 period. However, in this case, the 1999-2001 ROP data are preferable, because the ROP
37 data breaks out systems separately (some of the industry 1995-1997 INPO data combine
F-9
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 systems, such as HPCI and RCIC, and do not include PWR RHR). It is important to note
2 that the data for the two periods is very similar.)
3 Support cooling baseline data is based on plant specific maintenance rule unplanned and
4 planned unavailability for years 1999 to 2001. (Maintenance rule data does not
5 distinguish between planned and unplanned unavailability. There is no ROP support
6 cooling data.)
7 The baseline planned unavailability is based on actual plant-specific values for the period
8 1999 through 2001. These values are expected to remain fixed unless the plant
9 maintenance philosophy is substantially changed with respect to on-line maintenance or
10 preventive maintenance. In these cases, the planned unavailability baseline value can be
11 adjusted. A comment should be placed in the comment field of the quarterly report to
12 identify a substantial change in planned unavailability. To determine the planned
13 unavailability:
14 1. Record the total train unavailable hours reported under the Reactor Oversight Process
15 for 1999 through 2001.
16 2. Subtract any fault exposure hours still included in the 1999-2001 period.
17 3. Subtract unplanned unavailable hours
18 4. Add any on-line overhaul hours and any other planned unavailability excluded in
19 accordance with NEI 99-02. 2
20 5. Add any planned unavailable hours for functions monitored under MSPI which were
21 not monitored under SSU in NEI 99-02.
22 6. Subtract any unavailable hours reported when the reactor was not critical.
23 7. Subtract hours cascaded onto monitored systems by support systems.
24 8. Divide the hours derived from steps 1-6 above by the total critical hours during 1999-
25 2001. This is the baseline planned unavailability
26 Baseline unavailability is the sum of planned unavailability from step 7 and unplanned
27 unavailability from Table 1.
28
29
2 Note: The plant-specific PRA should model significant on-line overhaul hours.
F-10
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 Table 1. Historical Unplanned Maintenance Unavailability Train Values
2 (Based on ROP Industrywide Data for 1999 through 2001)
3
4
SYSTEM UNPLANNED UNAVAILABILITY/TRAIN
EAC 1.7 E-03
Support Cooling No Data Available Use plant specific Maintenance
Rule data for 1999-2001
5
F-11
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 Table 2. Industry Priors and Parameters for Unreliability
2
3
Component Failure aa ba Industry Source(s)
Mode Mean
Value b
Motor-operated Fail to open 5.0E-1 2.4E+2 2.1E-3 NUREG/CR-5500, Vol.
valve (or close) 4,7,8,9
Air-operated Fail to open 5.0E-1 2.5E+2 2.0E-3 NUREG/CR-4550, Vol. 1
valve (or close)
Motor-driven Fail to start 5.0E-1 2.4E+2 2.1E-3 NUREG/CR-5500, Vol.
pump, standby 1,8,9
Fail to run 5.0E-1 5.0E+3h 1.0E-4/h NUREG/CR-5500, Vol.
1,8,9
Motor-driven Fail to start 4.9E-1 1.6E+2 3.0E-3 NUREG/CR-4550, Vol. 1
pump, running
or alternating Fail to run 5.0E-1 1.7E+4h 3.0E-5/h NUREG/CR-4550, Vol. 1
Turbine-driven Fail to start 4.7E-1 2.4E+1 1.9E-2 NUREG/CR-5500, Vol. 1
pump, AFWS
Fail to run 5.0E-1 3.1E+2 1.6E-3/h NUREG/CR-5500, Vol. 1
Turbine-driven Fail to start 4.6E-1 1.7E+1 2.7E-2 NUREG/CR-5500, Vol.
pump, HPCI or 4,7
Fail to run 5.0E-1 3.1E+2h 1.6E-3/h NUREG/CR-5500, Vol.
1,4,7
Diesel-driven Fail to start 4.7E-1 2.4E+1 1.9E-2 NUREG/CR-5500, Vol. 1
pump, AFWS
Fail to run 5.0E-1 6.3E+2h 8.0E-4/h NUREG/CR-4550, Vol. 1
Emergency Fail to start 4.8E-1 4.3E+1 1.1E-2 NUREG/CR-5500, Vol. 5
diesel generator
Fail to 5.0E-1 2.9E+2 1.7E-3 c NUREG/CR-5500, Vol. 5
load/run
Fail to run 5.0E-1 2.2E+3h 2.3E-4/h NUREG/CR-5500, Vol. 5
4
5
F-12
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1 a. A constrained, non-informative prior is assumed. For failure to run events, a = 0.5 and
2 b = (a)/(mean rate). For failure upon demand events, a is a function of the mean
3 probability:
4
5 Mean Probability a
6 0.0 to 0.0025 0.50
7 >0.0025 to 0.010 0.49
8 >0.010 to 0.016 0.48
9 >0.016 to 0.023 0.47
10 >0.023 to 0.027 0.46
11
12 Then b = (a)(1.0 - mean probability)/(mean probability).
13
14 b. Failure to run events occurring within the first hour of operation are included within
15 the fail to start failure mode. Failure to run events occurring after the first hour of
16 operation are included within the fail to run failure mode. Unless otherwise noted, the
17 mean failure probabilities and rates include the probability of non-recovery. Types of
18 allowable recovery are outlined in the clarifying notes, under Credit for Recovery
19 Actions.
20
21 c. Fail to load and run for one hour was calculated from the failure to run data in the
22 report indicated. The failure rate for 0.0 to 0.5 hour5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> (3.3E-3/h) multiplied by 0.5 hour5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />,
23 was added to the failure rate for 0.5 to 14 hours1.62037e-4 days <br />0.00389 hours <br />2.314815e-5 weeks <br />5.327e-6 months <br /> (2.3E-4/h) multiplied by 0.5 hour5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />.
F-13
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
Table 3. Component Boundary Definition
Component Component boundary
Diesel The diesel generator boundary includes the generator body, generator
Generators actuator, lubrication system (local), fuel system (local), cooling components
(local), startup air system receiver, exhaust and combustion air system,
dedicated diesel battery (which is not part of the normal DC distribution
system), individual diesel generator control system, circuit breaker for supply
to safeguard buses and their associated local control circuit (coil, auxiliary
contacts, wiring and control circuit contacts, and breaker closure interlocks) .
Motor-Driven The pump boundary includes the pump body, motor/actuator, lubrication
Pumps system cooling components of the pump seals, the voltage supply breaker,
and its associated local control circuit (coil, auxiliary contacts, wiring and
control circuit contacts).
Turbine- The turbine-driven pump boundary includes the pump body, turbine/actuator,
Driven Pumps lubrication system (including pump), extractions, turbo-pump seal, cooling
components, and local turbine control system (speed).
Motor- The valve boundary inc1udes the valve body, motor/actuator, the voltage
Operated supply breaker (both motive and control power) and its associated local
Valves open/close circuit (open/close switches, auxiliary and switch contacts, and
wiring and switch energization contacts).
Air-Operated The valve boundary includes the valve body, the air operator, associated
Valves solenoid-operated valve, the power supply breaker or fuse for the solenoid
valve, and its associated control circuit (open/close switches and local
auxiliary and switch contacts).
1
F-14
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1ESFAS/Sequencer
2
DC Power Class 1E Bus
3
4
5
6
7 EDG
Breaker
8 Lubrication Governor and Exhaust
System Control System System
9
10
Control and
11 Protection System
12 Diesel Engine Generator
13 Starting Air
System Receiver
14
Jacket Fuel Oil Exciter and
15 Water System Voltage
Combustion Air
16 System and Regulator
Supply Fuel Oil Day
17 Isol. Tank
Valve
18 EDG Boundary
19
20
Cooling Water Fuel Storage and
21 Transfer System
22
23 Figure F-1
F-15
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1
Controls ESFAS
Breaker
Motor Operator
Pump
Motor Driven Pump Boundary
2
3 Figure F-2
4
5
F-16
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
Controls ESFAS
Breaker
Motor Operator
MOV Boundary
1 Figure F-3
2
F-17
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
1
Controls ESFAS
Turbine
Pump
Turbine Driven Pump Boundary
2
3 Figure F-4
4
F-18
DRAFT NEI 99-02 MSPI 8/28/20028/23/20028/9/2002
Non-active
1 Components
Active
Active Components
Components
T
A
N
K
Figure F-5
F-19