ML20035A678
| ML20035A678 | |
| Person / Time | |
|---|---|
| Site: | 05200001 |
| Issue date: | 03/25/1993 |
| From: | Kelly G Office of Nuclear Reactor Regulation |
| To: | Duncan J GENERAL ELECTRIC CO. |
| Shared Package | |
| ML20035A675 | List: |
| References | |
| NUDOCS 9303290150 | |
| Download: ML20035A678 (3) | |
Text
v
, g 4 Kf0 9 0
jo UNITED STATES g
f y
g NUCLEAR REGULATORY COMMISSION p
WASHINGTON, D. C. 20555 g
/
March 25, 1993 NOTE T0:
Ja D 'ncan, GE /
FROM:
Glen Q1yt S 05$
R
/
SUBJECT:
CLARIFICATION OF YOUR RESPONSE TO QUESTIONS ON THE ABWR PROBABILISTIC FLOODING ANALYSIS I have enclosed my comments on your response to questions (Kelly to Duncan, February 22, 1993) on the ABWR probabilistic flooding analysis that you faxed to me on March 4, 1993.
If you have any questions on my comments, please contact me.
Enclosure:
as stated 9303290150 930324 PDR ADOCK 05200001 A
PSR
i ENCLOSURE-Internal Floodino Clarifications (a) In Figures 19R.5-1 and -2 (Turbine Building Flooding (low and high UHS,
.l respectively)), credit was taken for the truck entrance doors not holding back any water. These figures also assumed that the door from the turbine building to the service building access tunnel is a watertight door and is closed.
Lastly, it is assumed that the probability of the reactor being brought to a safe condition is 1 L-8 given the flooding up to that point.
(i) The staff is not convinced that the truck entrance is not capable of retaining water.- It is quite possible that it could hold water to a level of one foot or higher before there was a catastrophic failure of the door that would open it enough to let all the flood waters out. GE should provide an evaluation that t
determines at what flood height a flood from the turbine building to the i
service building is a concern.
GE should look into what height a truck door actually can retain water or should assume that the door will retain some level of water.
(ii) GE should report to the staff what equipment in the control building would be failed by various levels of flooding from the support building.
(iii) GE should reconsider its IE-8 conditional core damage frequency in top event " Reactor Brought to a Safe Cond" given the equipment failed in "11" above. GE will be certifying in its final SSAR that the 1 E-8 conditional core damage frequency is appropriate and correct.
(iv)
The top event; in Figure ISR.5-1 does not appear to consider the potential for a siphoning effect.
To what extent is siphoning possible for a plant with a low ultimate heat sink? If it is possible, either modify the event tree or justify why it does not need to be considered.
(b)
Figure 19R.5-3 is the event tree for Control Building Flooding.
(i) In this figure there are three places where the event tree tonsiders the possibility of level sensors detecting the flood. The assumed failure probability of all sensors is E-9.
This value appears to be very unreasonable and does not appear to consider common cause failure.
Please discuss the values chosen and discuss how common cause failure (including such things as improper maintenance) was factored into the estimate.
(ii) Please provide a list of the equipment that would be flooded in the Control Building based on where the flood waters travel to and the height of the flood.
(c)
In Figure 19R.4-2 (Reactor Service Water Systens, it appears that-there are two valves in the service water system supply line to the control building, but there is only one isolation valve after the service water has cooled its associated equipment in the control room.
(i) Is there a corresponding isolation valve to M0 F014 (in the supply line) in the line down stream of M0 F005? (ii) How was the volume of water in the service water line downstream of M0 F005 treated in the internal flooding analysis? Is this volume included in the calculation of the 2000 meters maximum? (iii) If the RSW pumps trip but the isolation valves fail to close on a break in the control building, can a siphon effect continue to pull water into the control building, especially if the ultimate heat sink is at a similar height or even higher than the control building? (iv) If a siphon effect can occur, what would be the corrected value to the top event " Automatic Flooding Isolathn" in Figure 19R.5-37 (d)
In Table 19R.5-3 (Control Building Flooding), does "Div. 2 Power or
4.
+
Service Water Unavailable" mean that both ac and de power that supply Division 2 equipment are unavailable or does it only mean that ac power is lost?
If dc power is not considered lost, why not? When the table states "Div. 2 and 3 Power or Service Water Unavailable", does it mean that the ac and dc electrical power to both Divisions 2 and 3, or the Service Water to both Divisions 2 and 3 are unavailable?
(e) Figure 19R.4-1 is unreadable.
Please provide a readable copy.
t (f)
Let us assume that the CWS including its water source is at an elevation that is higher than the passage from the Turbine Building to the Service Building.
This is the case at a number of power plants today.
Let us also assume that the door between the Turbine Building and the Service Building is not a watertight door (This information was orally communicated to me recently-by GE). Now a break in the CWS pipe in the Turbine Building (4 E-2 per year) would flood the building.
Even if level switches detected the flood and tripped the CWS pumps, unless the three MOVs isolated, the water would continue to pour into the Turbine Building.
If we also assume that the truck entrance door can retain some height of water at its base without severely buckling, the fire door leading from the Service Building to the Control
+
Building would allow water to enter into the Service Building and on into the Control Building. This is about a 5 E-4 per year event.
If, as in Figure 19R.5-3, the conditional probability of a large flood in the Control Building preventing the reactor from being brought to a safe shutdown is 1 E-1, then the overall frequency of an internal flood leading to core damage is on the order of 5 E-5 per year.
Please discuss this possible event and why it should not be considered a potential vulnerability, if the CWS is sufficiently high.
(g)
Figure 19R.5-4 (Peactor Building Flooding in ECCS Room) has as its top event "(No) Water in the Corridor".
This water in the corridor is only considered if the sump level switches have failed to detect a flood. This does not seem to be appropriate.
It would appear that water could get into the corrider regardless of whether the sump level switches alarmed or not.
(i) Consider modifying the event tree to include consideration of water being in the corridor after an ECCS pipe break, since it makes no difference whether or not the sump level switches alarm.
(ii) The event tree does not consider the chance of a common cause failure failing all three ECCS doors.
- Instead, the doors are treated as independent, having a combined chance of 1 in a billion of being open or leaking.
Please modify the event tree to more realistically quantify the chances of a flood in an ECCS pump room flooding the other ECCS pump rooms.
(h)
In Figure 2 in your March 3, 1993 fax to the staff on the ABWR Probabilisite Flooding Analysis, you took credit for a 1 E-4 recovery factor.
(i) Please explain this factor.
(ii) Does it come from the Class II CET?
(iii) Exactly what is assumed to be recovered and in what time frame? Does it include recovery of feedwater pumps, condensate pumps, or fire water?
(iv)
What is meant by " Failure to Restore Normal Heat Removal"? (v) Why is the loss of service water frequency chosen as 1 per year? (vi) The value of 6.38 E-4 does not appear in any of the event trees. How is the loss of service water event tree used?