ML18213A088
| ML18213A088 | |
| Person / Time | |
|---|---|
| Issue date: | 07/31/2018 |
| From: | Tammie Rivera NRC/RES/DRA/FRB, Electric Power Research Institute |
| To: | |
| Shared Package | |
| ML18213A072 | List: |
| References | |
| Download: ML18213A088 (33) | |
Text
Module III - Fire Analysis Detection and Suppression Appendix P Joint EPRI/NRC-RES Fire PRA Workshop August 6-10, 2018 A Collaboration of the Electric Power Research Institute (EPRI) & U.S. NRC Office of Nuclear Regulatory Research (RES)
Detection & Suppression Objectives The objectives of this module are:
Describe the process for calculating the non-suppression probability Describe the assumptions underlying the recommended approach for determining the non-suppression probability.
Related SR: FSS-C7 2
Detection & Suppression Generalities Time to target damage and non-suppression probabilities are independent calculations
- Its like a probabilistic horse race - will damage win or will suppression win?
- We calculate time to damage through fire modeling and use that as an input to detection/suppression analysis
- We then ask whats the probability that suppression succeeds before damage occurs?
Fire models cannot model the effects of all the different fire detection and suppression strategies available in NPP fire scenarios
- We do pretty well with simple things like smoke detection time, sprinkler head activation time
- We currently dont do things like water droplets interfering with fire physics (although there are folks working on those kinds of problems) 3
Detection & Suppression Crediting a Fire Det. or Supp. System Detection and suppression systems can and should be credited in the fire PRA if they are effective and available Effectiveness - Will the system detect/control the fire?
- Designed, installed and maintained according to the code of record and fire protection engineering judgment
- Based on the specific characteristics of the postulated fire scenario Availability - Probability of the system actually operating as designed upon demand 4
Detection & Suppression Fire Detection and Suppression Systems The following fire detection and suppression systems are considered in the recommended approach:
Fire Detection
- Prompt detection
- Automatic detection
- Delayed detection (by plant personnel)
- Incipient detection*
Fire Suppression
- Prompt suppression
- Automatic suppression
- Manual suppression
- covered in a separate presentation 5
Detection & Suppression Conceptual Relationship Tdet: time interval from start of fire up to time when fire is initially detected Tsupp: time interval from when fire is detected until it is suppressed Tfb: time from fire detection until fire brigade begins to apply suppressant agents 6
Availability of new data Similar to the ongoing improvement work with fire ignition frequencies, manual NSP data has also been updated since NUREG/CR-6850
- All data sources excluded fire suppression events prior to 1/1/1981 (pre Appendix R)
- Contains original methodology and original suppression curves FAQ 08-0050 (Supplement 1 Chapter 14)
- Updated methodology to eliminate delay time for brigade arrival EPRI 3002002936 / NUREG-2169
- Same methodology as FAQ
- Added additional event experience through 2009 7
FAQ 08-0050 Manual Non-Suppression Probability Issue:
- NUREG/CR-6850 / EPRI 1011989 assumes fire fighting begins after detection and brigade arrival on the scene
- Too little credit to manual fire suppression before the fire brigade arrives on the scene compared to experience Resolution
- Updated guidance for treatment of manual suppression and the fire brigade response
- Includes a process to adjust the non-suppression analysis for scenario-specific fire brigade responses.
Reference
- Supplement 1 to NUREG/CR-6850 (EPRI 1019259) Chapter 14 8
FAQ 08-0050: Solution How the Pns curves were re-calculated 6850/1011989 used manual suppression time if available, and fire duration if no suppression time is identified from fire event records
- Many cases deferred to duration FAQ uses the fire duration time for all events (time from detection to suppression)
- Fire duration time is either the same or longer than suppression time
- Result: the base Pns curves are slightly more conservative, but FAQ also assumes fire control and suppression activities start at the time of detection
- No time delay for brigade arrival
- Credits suppression by plant personnel other than fire brigade
- More than makes up for shift in curves FAQ presents new non-suppression (Pns) curves for all bins Includes method to adjust for above or below average fire brigade response time 9
Detection & Suppression FAQ 50 changes the calculation of Pns Original 6850/1011989 approach:
[
t damage (t detection + t brigade response )]
Pns = e Revised FAQ 50 approach:
[
t damage t detection ]
Pns = e Where is the suppression rate constant 10
Detection & Suppression Selection of a suppression curve The suppression curve should be selected based on the type of postulated fire. There are 13 suppression curves available (see next two slides)
For prompt suppression by a welding fire watch, use the welding suppression curve If the fire watch is not successful, an appropriate suppression curve should be selected depending on the combustibles ignited due to hot work activities 11
Detection & Suppression Selection of a suppression curve Transients - all transient fire event not specified as belonging to any of the following categories were used.
Welding - usually labeled transient fires caused by welding and cutting or cables fires caused by welding and cutting. This curve is used for calculating the probability of prompt suppression.
Electrical - events involving electrical cabinets, electric motors, indoor transformers, and junction boxes among other electrical equipment. Electrical events in the control room, transformer yard, and T/G excitor were excluded. HEAFs are also excluded form this curve.
- Note: events including overheating equipment, such as bearings due to lack of lubricating oil were included in this set.
Cable - events for cables in raceways. Records with insufficient details, such as simply reporting a cable/wiring fire were included in this data set. Does not include extension cord fires or cable/wire fires within electrical equipment.
Oil - includes events were a lubricating substances was ignited.
Flammable Gas - includes only events involving hydrogen fires. Fires in off-gas/h2 recombiners were also included.
Transformer Yard - includes events occurring in the transformer yard.
- Excludes events with fixed fire suppression and events where the fire brigade decided to let the fire burn out.
12
Detection & Suppression Selection of a suppression curve Containment PWR - for events occurring in PWR containment during at-power conditions. The fire brigade access to the building in the event of a fire is relatively different from other locations in the plant. Events in this data set include all fire events occurring in containment.
Containment (LPSD) - new suppression curve that is applied to LPSD only conditions. Applicable when containment is open to maintenance and operations staff when the plant is not at power. Primarily for hotwork, but can include electrical or other in-containment fires. Applicable to both PWR and BWR.
Control Room - include fires within the control room regardless of ignition source.
Turbine Generator - includes events labeled as T/G oil, T/G hydrogen, T/G excitor. Some T/G hydrogen events were excluded from this data set because they were suppressed with an automatic system.
HEAFs - includes events labeled as high-energy arcing faults.
All events - includes all events not considered as prompt suppression.
This generated a generic suppression time probability curve that may be used for those cases where the analyst cannot find a proper match from the preceding list of categories.
13
NSP Update From EPRI 3002002936 / NUREG-2169 14
Updated NSP Curve Plots (EPRI 3002002936 / NUREG-2169)
Probability vs. time to suppression 15
Updated Non-Suppression Curves EPRI 3002002936 / NUREG-2169 16
Detection & Suppression Event Tree format The event tree approach is one way to deal with suppression
- Works well
- Quantifies various outcomes (success/failure)
- Very flexible You may need to modify the event tree to suit application
- The generic event tree is a simple single-stage version
- We will show one example of a more complex tree For the simple tree only one target set
- Success means the fire is put out before damage occurs
- Failure means suppression is not timely and the target set is damaged Remember: all fires are put out eventually - it is a question of timing 17
Detection & Suppression Detection-Suppression Event Tree Fire Prompt Automatic Manual End Sequence Outcome Ignition Detection Suppression Detection Suppression Detection Suppression State Propmt detection and prompt A ND suppression succeed B ND Automatic suppression succeeds Prompt detection and manual C ND suppression succeed D DMG All means of suppression fail E ND Automatic suppression succeeds Automatic detection and successful F ND manual suppression G DMG All means of suppression fail Notes:
Failures are down paths, No auto detection but auto suppression H ND Success is up path succeeds Delayed manual detection and successful N/V - not a valid end state I ND manual suppresion ND - Suppression is timely Delayed manual detection but manual J DMG and prevents damage suppression fails DMG - Suppression is not Fire is not detected in time to prevent K DMG timely and damage occurs damage 18
Detection & Suppression Detection-Suppression Event Tree Fire Prompt Automatic Manual End Sequence Outcome Ignition Detection Suppression Detection Suppression Detection Suppression State Propmt detection and prompt A ND suppression succeed B ND Automatic suppression succeeds Prompt detection and manual C ND suppression succeed D DMG All means of suppression fail E ND Automatic suppression succeeds Automatic detection and successful F ND manual suppression G DMG All means of suppression fail Notes:
Failures are down paths, No auto detection but auto suppression H ND Success is up path succeeds Delayed manual detection and successful N/V - not a valid end state I ND manual suppresion ND - Suppression is timely Delayed manual detection but manual J DMG and prevents damage suppression fails DMG - Suppression is not Fire is not detected in time to prevent K DMG timely and damage occurs damage 19
Detection & Suppression Fire ignition event There are two ways to enter the event tree:
- Enter with 1.0 - a fire has ignited This makes it the outcome a conditional probability - given that a fire has been ignited
- Enter with a fire frequency (events/yr)
This makes the outcome a frequency for each end state Branch point probabilities have no units so frequency units come through 20
Detection & Suppression Detection-Suppression Event Tree Fire Prompt Automatic Manual End Sequence Outcome Ignition Detection Suppression Detection Suppression Detection Suppression State Propmt detection and prompt A ND suppression succeed B ND Automatic suppression succeeds Prompt detection and manual C ND suppression succeed D DMG All means of suppression fail E ND Automatic suppression succeeds Automatic detection and successful F ND manual suppression G DMG All means of suppression fail Notes:
Failures are down paths, No auto detection but auto suppression H ND Success is up path succeeds Delayed manual detection and successful N/V - not a valid end state I ND manual suppresion ND - Suppression is timely Delayed manual detection but manual J DMG and prevents damage suppression fails DMG - Suppression is not Fire is not detected in time to prevent K DMG timely and damage occurs damage 21
Detection & Suppression Prompt Detection and Suppression Prompt detection
- Assume 1.0 if a continuous fire watch is credited or in-cabinet detection is available for fires postulated inside cabinets
- Justify the use of 1.0 if an incipient fire detection system is available
- Assume 0 if automatic or delayed detection only are credited Prompt suppression
- Credit prompt suppression in hot work fire scenarios
- Probability is obtained from the welding suppression curve 22
Detection & Suppression Detection-Suppression Event Tree Fire Prompt Automatic Manual End Sequence Outcome Ignition Detection Suppression Detection Suppression Detection Suppression State Propmt detection and prompt A ND suppression succeed B ND Automatic suppression succeeds Prompt detection and manual C ND suppression succeed D DMG All means of suppression fail E ND Automatic suppression succeeds Automatic detection and successful F ND manual suppression G DMG All means of suppression fail Notes:
Failures are down paths, No auto detection but auto suppression H ND Success is up path succeeds Delayed manual detection and successful N/V - not a valid end state I ND manual suppresion ND - Suppression is timely Delayed manual detection but manual J DMG and prevents damage suppression fails DMG - Suppression is not Fire is not detected in time to prevent K DMG timely and damage occurs damage 23
Detection & Suppression Automatic Detection and Suppression Automatic detection
- Assume a probability of failure no larger than 0.05. This the unreliability for Halon systems reported in NSAC-179L
- Check for availability!
Automatic suppression (from NSAC-179L)
- Halon systems = 0.05
- CO2 systems = 0.04
- Wet pipe sprinklers = 0.02
- Deluge or pre-action = 0.05
- Check for availability!
24
Detection & Suppression Detection-Suppression Event Tree Fire Prompt Automatic Manual End Sequence Outcome Ignition Detection Suppression Detection Suppression Detection Suppression State Propmt detection and prompt A ND suppression succeed B ND Automatic suppression succeeds Prompt detection and manual C ND suppression succeed D DMG All means of suppression fail E ND Automatic suppression succeeds Automatic detection and successful F ND manual suppression G DMG All means of suppression fail Notes:
Failures are down paths, No auto detection but auto suppression H ND Success is up path succeeds Delayed manual detection and successful N/V - not a valid end state I ND manual suppresion ND - Suppression is timely Delayed manual detection but manual J DMG and prevents damage suppression fails DMG - Suppression is not Fire is not detected in time to prevent K DMG timely and damage occurs damage 25
Detection & Suppression Delayed Detection and Suppression Manual/delayed detection
- Assume 1.0 - All fires will eventually be detected (max = 15min)
- Compare time to target damage vs. time to detection and suppression Manual suppression
- Probability of manual suppression is obtained from the suppression curves we already discussed
- Can includes possibility of manual actuation of fixed fire suppression systems Credit here should include human reliability analysis and consideration of dependencies (e.g., failure of the fire water system) 26
Detection & Suppression Detection-Suppression Event Tree Fire Prompt Automatic Manual End Sequence Outcome Ignition Detection Suppression Detection Suppression Detection Suppression State Propmt detection and prompt A ND suppression succeed B ND Automatic suppression succeeds Prompt detection and manual C ND suppression succeed D DMG All means of suppression fail E ND Automatic suppression succeeds Automatic detection and successful F ND manual suppression G DMG All means of suppression fail Notes:
Failures are down paths, No auto detection but auto suppression H ND Success is up path succeeds Delayed manual detection and successful N/V - not a valid end state I ND manual suppresion ND - Suppression is timely Delayed manual detection but manual J DMG and prevents damage suppression fails DMG - Suppression is not Fire is not detected in time to prevent K DMG timely and damage occurs damage 27
Detection & Suppression End States The sequence number is just numbering the outcomes For the simple event tree, the end states are either no damage (ND), or loss of the target set (DMG)
- We will show a more complicated version with more possible outcomes The probability of each end state is simply the product of the branch probabilities leading to the end state 28
Detection & Suppression Dependencies The following dependencies in suppression analysis could be important:
Between automatic detection and suppression
- Example: control panel for a gaseous suppression system Between actuated barriers and fire suppression systems Between safe shutdown capabilities and automatic suppression
- Example: crediting fire fighting water for core injection, heat removal or secondary heat removal Between manual and automatic suppression 29
Detection & Suppression Conceptual Example The scenario consists of an MCC fire affecting a target in the hot gas layer.
The room is equipped with a smoke detection system and automatic sprinklers Using fire modeling
- Time to smoke detection = 1 min
- Time to sprinkler activation = 8 min
- Time to target damage = 15 min From fire drill records and/or plant procedures
- Time to delayed detection is assumed to be 15 min 30
Detection & Suppression Example for the 6850/1011989 approach No prompt detection Failure of auto. det.
Sequence Prompt Automatic Manual
- P = 0.05 Pr(non-Fire Detection Suppression Detection Suppression Detection Fire Brigade End State suppression) Failure of sprinklers:
FI PD PS AD AS MD FB 1 0 0 A Not valid
- P = 0.02
- Both require justification***
1 0.98 B Not valid Manual suppression:
0.02 0.78 C Not valid - Damage time: 15 min
- Auto detect: 1 min 0.22 D Not valid
- Time available for manual suppression:
1 0.95 0.98 E ND 15 - 1 = 14 min 0.02 0.76 F ND - Use electrical fire curve:
P = EXP(-0.10214) 0.24 G DMG 4.6E-03 P = 0.24 0.05 0.98 H ND Overall solution for this 0.02 0 I Not Valid scenario, DMG outcome:
1 J DMG 1.0E-03 Total 5.6E-03 Pns = G + J Note typos in your set at bottom of tree (yellow boxes) Pns = 5.6 E 3 31
Detection & Suppression Concluding remarks on general approach The non-suppression probability is credited in Task 11, detailed fire modeling Target damage is evaluated assuming no detection/suppression capabilities in the room The time to target damage is an input to the detection and suppression analysis.
The recommended approach includes an event tree capturing prompt, automatic, and delayed detection and suppression capabilities The event tree may need to be modified depending on the scenario 32
And now for a more complicated example When you assume a single target set, the whole set is lost when the first member of the set is lost The way fires really work is that damage spreads over time until the fire is suppressed
- The fire grows
- Spreads
- More targets become damaged A more realistic approach is to define multiple target sets based on spatial location and separation from the fire
- e.g., targets in the plume vs. targets in the hot gas layer
- Each target set is lost when first member is of the set is lost, but time to damage for the different sets may be quite different 33