ML20059C846
| ML20059C846 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 08/19/1993 |
| From: | Rueger G PACIFIC GAS & ELECTRIC CO. |
| To: | |
| References | |
| OLA-2-I-MFP-101, NUDOCS 9401060129 | |
| Download: ML20059C846 (14) | |
Text
Cf pp flh3Dd/Atf(fog (
i Pacific Gas and Electric Company "EL: 57 5-;:', " : :;L g ].:. 3 Nih
[- ','{ j::r c:
g'y
. 2 r: e Gm::-
4 November 20, 1992 PG&E Letter No. DCL-92-257 U.S. Nuclear Regulatory Commission y
f ATTN:
Document Control Desk Washington, D.C.
20555 S.
Re:
Docket No. 50-275, OL-DPR-80 Docket No. 50-323, OL-DPR-82 Diablo Canyon Units 1 and 2
'/
Licensee Event Report 1-92-023-00' Containment Fan Cooler Backdraft Damper Blade Cracking Due to High-Cycle Fatigue Gentlemen:
Pursuant to 10 CFF. 50.73(a)(2)(v)(9) and 50.73(a)(Pi(ii)(B), PG&E is submitting the enclosed Licensee Event Report (LER) regarding containment fan cooler unit backdraft damper blade cracks that were identified during the Unit 1 fifth refueling outage.
This condition potent 1 ally placed the units in a condition that was outside the design basis of the plant.
Also, pursuant to 10 CFR 50.73(a)(2)(i)(B), PG&E is reporting that this LER was not submitted on November 16, 1992, in accordance with the required 30-day time period due to information discovered on November 13, that required revision to the LER safety analysis.
This condition did not adversely affect the health and safety of the public.
Sincerely,
[
p.v Gregc,ry M. Rueger cc:
Ann P. Hodgdon John B. Martin Mary H. Miller Nucun van cr;" -va Harry Rood lo Distributi INP0
/
DCO-92-MM-N007 Enclosure 9
~ !I 3
10695/85K/JCN/2246 r
- ~
9401060129 930819 PDR ADOCK 05000275 G
$LW &-
LICENSEE EVENT REPORT (LER)
F ACluTY N AME til DO C
- U 6 W9 fP '7'
' ACE m DIABLO CANYON UNIT 1 Ol5l0l0l0l?I 7l5 1 i 'l 13 CONTAINMENT FAN COOLER BACKDRAFT DAMPER BLADE CRACKING DUE TO HIGH-CYCLE FATIGUE mum EvtmT DArE isi i
ta muuam m PENpr D ATE m CT"ER FACIGTitS !WGa[0 (61 MON DAY YM YM S E QUE.N TI AL P EV15dO N MON DAY VM DOC RET huMBEM (Si muun EM w w ste 0
0 3
2 3
DIABLO CANYON UNIT 2 10 15 92 92 0l 2l 3 0l0 11 20 92 0
5 0
0 0
i ggA s
ra:s at9 Car is susarrito pues.Aur to T, t atw!PrutNrs or to cra 01) 6 ttIck v
10 CFR 50.73(al'Jliv)(0).
50.73(a)(?)(ii)(9; and 50. 73 ( a )( ? )( i )(B )
0l0l0 OTHER c spec o y in seTtract betos and in text, waC Form 3664)
UCENSFF CONT ACT FOR TMS LLM (121
.,,.m DAVID P. SISK, SENIOR REGULATORY COMPLIANCE ENGINEER
""^ :"
805 545-4420 COMPT.ETE ONE UNE FOR EAC H COMPONENT F AtUAE DEJCRISED IN THS REPORT t13)
CAJ5L LYS?tM LePUNENT MAN FAC.
&(PO A
L caw ${
Sf51(M CCN NN(NT
- .a f4 -
T. i rs B
BlK UlDlMlP Al 3l 4l 0 Y
l ll l ll l 1
lll lll l
Ill III sVPPtT#EN T AL PtWAf DPLC T ED 04)
C"'
E PECTED SUBMISSION I
l l YES (if yes. cceptete EXPECTED SUBMISSICN DATE) j X l WO A85I8ACT {16)
On October 15, 1992, at 1620 PDT, with Unit 1 in Mode 6 (Refueling) and Unit 2 in Mode 1 (Power Operation), PG&E determined that the cracks found in Unit 1 containment fan cooler unit (CFCU) backdraft damper (BD) blades were a condition potentially outside the design basis (DB) of the plant.
On October 15, 1992, at 1714 PDT, non-emergency reports were made to the NRC in accordance with 10 CFR 50.M for Unit 1 and Unit 2.
Some BD blades were found to have cracks during visual inspections.
Evaluation of the identitied cracks has concluded that the cracks were of a sufficient length to have potentially resulted in blade failures from forces created by a pressure wave postulated to result from a loss-of-coolant accident.
On November 13, 1992, a magnetic particle examination of 74 of the 80 BD blades removed from Unit 1 identified nine additional blades with longitudinal cracks.
The root cause of the CFCU BD blade cracking was determined to be high-cycle fatigue.
The Unit 1 CFCU BD blades were replaced during the Unit I fifth refueling outage with blades made of a material more resistant to fatigue cracking.
The Unit 2 BD blades are being removed and will be replaced during the Unit 2 fifth refueling outage.
This Licensee Event Report (LER) was not submitted in accordance with the required 30-day time period due to information discovered on Novemoer 13, 1992, that required revision to the LER safety analysis.
10695/85K
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 199 %
"a m uym
-.o -.a m
<mun - m DIABLO CANYON UNIT 1 0l 5l 0l Ul 0l 2 l 7l 5 92
- 0l2l3 0l0 2 l 'l 13 nn <m 1.
Plant Conditions Unit I was in Mode 6 (Refueling) and Unit 2 was in Mode 1 (Power Operation) at 100 percent puter.
II.
Description of Event A.
Summary:
On October 15, 1992, at 1620 PDT, PG&E determined that cracks found in containment fan cooler unit (CFCU)(BK)(FCU) backdraft damper (BD)(BK)(UDMP) blades resulted in a condition potentially outside the design basis (DB) of the plant.
On October 15, 1992, at 1714 PDT, a four-hour, non-emergency report was made to the NRC in accordance with 10 CFR 50.72(b)(2)(iii)(D) for Unit 1 and a one-hour, non-emergency report was made to the NRC in accordance with 10 CFR 50.72(b)(1)(ii)(B) for Unit 2.
Some BD blades associated with CFCUs 1-2, 1-4, 1-5, 2-1, and 2-2 were found to have cracks during visual inspections performed between September 25 and October 18, 1992.
Evaluation of the identified cracks has concluded that the cracks in CFCOs 1-2, 1-4, and 2-2 were of a sufficient length to have potentially resulted in blade failures during a DB loss-of-coolant accident (LOCA).
On November 13, 1992, a magnetic particle examination of 74 of the 80 BD blades removed from Unit I during the fifth refueling outage (IR5) identified nine additional blades with longitudinal cracks.
Eight of these cracks were of a length sufficient to potentially cause failure of a BD blade.
This Licensee Event Report (LER) was not,_bmitted in accordance with the required 30-day time period due to the magnetic particle examination results that became available on November 13, 1992.
These results necessitated revision to the LER safety analysis.
B.
Background:
Technical Specification 3.6.2.3 states:
The Containment Cooling System shall be OPERABLE with either:
a.
At least three electrically independent groups of containment fan cooler units with a minimum of one unit in each group, or 10695/85K i
1 LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 199$$$
,FACJt!TV NAME (1)
DOCKET **uMsER (2) tra wuweta is) paGE ( 3).
vsAa as wf as ses DIABLO CANYON UNIT 1 0l5l0l0lOl2l7l5 92
- 0l2l3 0l0 3 l 'l 13 7 EXT (17) b.
At least two electrically independent groups of containment fan cooler units with a minimum of two units in each group.
I With less than the required CFCOs available in Modes 1, 2 (Startup), 3 (Hot Standby), or 4 (Hot Shutdown), an operating unit must be shutdown within 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />.
The CFCus are designed to remove heat from the containment (NH) atmosphere during normal operation and in the event of an accident.
There are five CFCUs in each containment located on the 140 foot elevation between the crane wall and containment liner.
A CFCU consists of several components:
a fan (BK)(FAN), a motor (BK)(MO), a BD with two side-by-side banks of eight blades each, cooling coils (BK)(CLR), moisture separator filters (BK)(FLT), and a drain system (BK)(DRN).
Each fan is driven by a two-speed motor that operates at 1200 rpm for normal operation and 600 rpm for accident conditions.
The CFCUs draw in ambient air from above the 140 foot elevation through the moisture separator filters (during accident only) and cooling coils and discharge the air through the BDs to the common annular ring duct (BK)(DUCT) under tne floor.
The annular ring distributes air for component and space cooling.
The CFCUs remove heat to maintain containment ambient temperature during normal plant operating conditions.
Following a LOCA or steam line (SB)(PSP) break, the CFCUs, in conjunction with the containment spray system (BE), reduce the containment ambient temperature and pressure and help to promote mixing between the lower and upper containment areas.
For normal containment ambient temperature control, the number of CFCUs running depends on the amount of cooling needed in the containment.
Based on the cooling needed, three or four CFCUs may be running during full-power operation, while only one CFCU may be needed during Mode 5 (Cold Shutdown).
For normal operation, the CFCU fan motors run at fast speed (1200 rpm).
In the event of a LOCA, two of the five CFCUs are required to operate.
During normal operation, the CFCUs are manually started from the control room (NA).
The CFCUs are automatically started by a safety injection (SI)(BQ) signal following the LOCA.
In the accident mode, the CFCU fan motors run at low speed (600 rpm).
Each CFCU discharge has an individual BD. The BDs open by fan air flow and are closed by gravity, with counterweight assistance, when the fan is stopped.
The normal function of the dampers is to isolate a CFCU to minimize back flow from the downstream common ventilation ductwork when the corresponding CFCU fan is not running.
This prevents an inactive fan from 1069S/05K
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION gg
.-m u..;-
m ia,, - m mon -u. m samma emmena DIABLO CANYON UNIT 1 0l5l0l0l0l2l7l5 92
- 0l2l3 0l0 4 l 'l 13 YEAT (17) experiencing reverse rotation and minimizes cool air bypass of the intended ventilation ductwork. The BDs also provide isolation for the CFCUs from the ventilation distribution ductwork in the event of a DB LOCA.
C.
Event
Description:
During January through March 1992, there were several inspections of the Unit I and Unit 2 CFCUs.
Degraded BD linkage components were identified in the CFCUs and actions were taken to repair conditions in both units. These conditions were reported in LER 1-91-019-00.
Complete inspection and overhaul of the CFCUs was performed on September 12, 1992, during 1R5 and cracking of the CFCU BD blades was found.
On September 25 1992, one blade of the CFCU 1-5 BD was found to have a semi-circular crack about the center mounting bolt hole.
On September 27, 1992, one blade of the CFCU 1-2 BD was found to have two longitudinal cracks.
On October 3,1992, one blade of the CFCU 1-4 BD was found to have a partial semi-circular crack about the center mounting bolt hole.
On October 8,1992, a BD blade was sent to PG&E Technical and Ecological Services (TES) for evaluation of the nature of the cracking and materials testing.
On October 10, 1992, during a magnetic particle inspection of a sample of 23 BD blades, one additional blade of the CFCU 1-4 BD was found with a crack.
i On October 13, 1992, TES concluded that the cracking was the result of high-cycle fatigue and the "as-found" longitudinal cracks were of a sufficient length to have potentially resulted in blade. failures due to a postulated pressure wave resulting from a DB LOCA.
On October 15, 1992, at 1620 PDT, PG&E determined that the cracks found in the BD blades potentially were a condition outside the DB of the plant.
The Unit 2 CFCU BD blades are the same design as those in Unit 1.
As a result, the potential existed for cracks to be present in some of the Unit 2 BD blades.
On October 15,1992, at 1714 PDT, a four-hour, non-emergency report was made to the NRC in accordance with 10 CFR 50.72(b)(2)(iii)(D) for Unit I and a one-hour, non-emergency report was made to the NRC in accordance with 10 CFR 50.72(b)(1)(ii)(B) for Unit 2.
10695/85K 1
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 19$$98
- u. y <.,
.m
,m u n - m
-m
. m DIABLO CANYON UNIT 1 015l 0 l 0l 0l 2 l 7 l 5 92
- 0l2l3 -l0l0 5 l 'l 13 TEXT (17)
On October 17 and 18, 1992, visual inspections of the Unit 2 CFCU BD blades were performed and cracks were identified in one blade of the CFCU 2-1 BD and two blades of the CFCU 2-2 BD.
The crack in the blade of the CFCU 2-1 BD was semi-circular about the center mounting bolt hole.
The cracks in the blades of the CFCU 2-2 BD were longitudinal and of sufficient length to potentially fail from the pressure wave postulated to result from a DB LOCA.
On November 13, 1992, a magnetic particle examination of 74'df the 80 BD blades removed from Unit I during 1R5 identified nine additional blades with longitudinal cracks.
Eight of these cracks were of a length sufficient to potentially cause failure of a BD blade.
This LER was due to the NRC on November 16, 1992.
However, due to magnetic particle examination results received on November 13, 1992, that required revision to the LER safety analysis, this LER was not submitted in accordance with the required 30-day time period.
D.
Inoperable Structures, Components, or Systems that Contributed to the Event:
None.
E.
Dates and Approximate Times for Major Occurrences:
1.
September 25, 1992:
CFCU 1-5 BD crack was identified.
2.
September 27, 1992:
CFCU 1-2 BD cracks were identified.
3.
October 3, 1992:
CFCU 1-4 BD crack was identified.
4.
October 8, 1992:
A BD blade was sent to TES for analysis.
~
5.
October 10, 1992:
An additional CFCU 1-4 BD crack was identified.
6.
October 13, 1992:
PG&E evaluation determined that longitudinal cracks greater than 1/10-inch in length would result in BD blade failure during a DB LOCA.
7.
October 15, 1992; 1620 PDT:
Discovery / Event date.
PG&E determined that cracks found in the Unit 1 CFCU BD blades potentially were a condition outside the DB of the plant.
PG&E also determined that cracks could 10695/85K
)
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 1MB*?'Js f 4CILITY NAMC (3)
DOCKET NUMB [R (2)
((9 NVM$(9 I6)
PACE ())
vtAn L
m DIABLO CANYON utilT 1 0l5l0lC!0l2l7l5 92
- 0l2l3 0l0 6 l"l 13 ft37 (37) potentially exist in the Unit 2 CFCU BD blades.
8.
October 15, 1992; 1714 PDT:
A four-hour, non-emergency report was made to the NRC in accordance with 10 CFR 50.72(b)(2)(iii)(D) for Unit 1.
A one-hour, non-emergency report was made to the NRC in accordance with 10 CFR 50.72(b)(1)(ii)(B) for Udit 2.
9.
October 18, 1992:
Visual inspection of all Unit 2 CFCUs found cracks in BD blades of CFCU 2-1 and 2-2.
10.
November 13, 1992:
Magnetic particle examination found nine additional Unit 1 BD blades wich longitudinal cracks.
11.
November 17, 1992; 00:00 PDT: Event date.
LER not submitted to the NRC in accordance with 10 CFR 50.73(a)(2)(1)(B) within the required 30-day time period.
4 F.
Other Systems or Secondary Functions Affected:
None.
G.
Method of Discovery:
During 1R5 CFCU inspection and overhaul, performed as a corrective action in response to events reported in LE" l-91-019-6v, utility personnel identified cracking in Unit 1 CFCU BD blades.
H.
Operator Actions:
Operations placed all five Unit 2 CFCU fans in operation as a conservative measure until BD blade replacement can be performed to assure that post-accident fan auto-start will not be impaired.
1.
Safety System Responses:
None.
10695/85K
i 199598-LICENSEE EVENT REPORT (LER) TEXT CONTINUATION
' FACILITY ppMC (1)
DOCKET NUH5[R (2) t(R NUM9(R
'6) pag ( (3)
{
==
. y, j DIABLO CANYON UNIT 1 0l 5l 0l 0l 0l 2 l 7l 5 92
- 0l2l3 0l0 7 l 'l 13 TEXT (17)
III.
Cause of the Event A.
Immediate Cause:
1 1.
CFCU BD Blades The immediate cause of this event was cracking identified in BD blades during CFCU inspections.
2.
Late Submittal of LER This LER was submitted four days after its due date.
B.
Root Cause:
1.
CFCU BD Blades The root cause of the CFCU BD blade cracking has been determined to be high-cycle fatigue.
Initial evaluation has deternined air flow loading on the BD blades during normal CFCU operation caused the high-cycle fatigue.
Instrumentation is being installed on Unit 1 BD blades and CFCU ductwork to verify this evaluation.
2.
Late Submittal of LER This LER was due to the NRC on November 16, 1992.
However, due to information received on Novembc.- 13, 1902, that required revision to the LER safety analysis, this LER was not submitted in accordance with the required 30-day time period.
IV.
Analysis of the Event 1.
CFCU BD Blades The safety-related function of the BDs is to provide isolation of the CFCUs from the ventilation distribution ductwork in the event of a DB LOCA. This isolation feature protects each CFCU fan / motor from the effects of a postulated large-break LOCA pressure wave force resulting from an instantaneous reactor coolant system (RCS)(AB) loop piping double-ended guillotine rupture.
The BD blades are approximately 1/8-inch thick, 11 inches long, and 21 inches wide.
The blade material is SAE 1075 spherodized steel, heat treated to 180,000 psi minimum yield.
This is a very low ductility material that would exhibit a brittle fracture type of failure that has a high sensitivity to flaws.
Calculations show that the critical crack length at which a blade will fail under the FSAR Update DB LOCA 1069S/85K
INN LICENSEE EVENT REPORT (LER) TEXT CONTINUATION "a m m;-
m
-m -n. m
. mum m wuwam nuussa DIABLO CANYON UNIT 1 0l 5l 0l 0l 0l 2l 7l 5 92
- 0l2l3 0l0 8 l 'l 13 TEAT (11) pressure of seven pounds per square inch, differential (psid), is on the order of 1/10-inch. This seven psid could be caused by a postulated large-break LOCA pressure wave force created by an instantaneous RCS loop piping double-ended guillotine rupture. The FSAR Update defines the pressure transient as a seven psid pressure surge over a ten second period.
The pressure wave entry into a non-coerating CFCU could cause the fan / motor to experience unacceptable reverse rotation.
Excessive reverse rotation can impair the capability of the fan motor to restart and consequently impair the capability of the CFCU to perform its safety function. Testing of the blade material has verified the BD blade cracking would follow the direction of maximum stress and eventually result in a longitudinal crack the length of the blade and separation of a portion of the blade as one piece.
DCPP has not experienced any CFCU BD blade failures during normal operation as a result of cracks. However as stated above, the characterization and evaluation of the cracks and material, supported by laboratory testing, concludes that the identified longitudinal cracks in this BD blade material were of sufficient length to have resulted in BD blade failure if challenged by the postulated instantaneous DB LOCA pressure wave.
It should be noted that two blades in Unit 1 and one blade in Unit 2 were discovered with semi-circular cracks.
Semi-circular cracks would not lead to blade failure under the postulated loading. A total of ten BD blades in Unit I and two BD blades in Unit 2 have been identified to have longitudinal cracks that could have failed if challenged by a seven psid pressure wave.
Currently all olades with identified cracks have been removed from the dampers.
The FSAR Update identifies the LOCA differential preesure across a BD caused by a sudden RCS pressure release to be seven psid.
- However, the seven psid results from postulating an instantaneous double-ended guillotine RCS loop break. Analysis has been performed and submitted for NRC review under PG&E letter DCL-92-059, dated March 16, 1992, which demonstrates that RCS loop pipe failure at Diablo Canyon will not be instantaneous.
Instead, a crack may occur that results in some energy release, that will increase with time as the crack propagates.
The lower rate of energy release assures that large compartment differential pressures will not occur; therefore, it is not appropriate to postulate a significant differential pressure across the BDs.
PG&E identified and reviewed secondary piping breaks for the potential of creating a pressure differential across the BDs.
The steam lines and the feedwater lines (SJ)(PSP) pass through the containment structure annulus below the 140 foot elevation operating deck.
Pipe routing was reviewed, and the effects of a pipe break in those areas 10695/85K
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 19N
..u m u.;.
,m u n - m man -n m DIABLO CANYON UNIT 1 0l5l0l0l0l2l7l5 92
- 0l2l3 -l0l0 9 l *'l 13_
707 (m in containment were evaluated.
Due to the large upper containment volumes and the size of the vent paths between the lower and upper levels, it was concluded that the differential pressure that might occur across the BDs in the event of a secondary pipe failure is nsgligible, The previous discussion demonstrates that pressurization significant e e ugh to propagate the blad-racks will not occur and therefore the ability of the CFCUs to perform their design functions has been and will continue to be maintained.
Although an event that could have caused a back pressure across the BD blades has not occurred and is not considered credible, the pote.itial consequences to the BD and the CFCU have been evaluated.
If a longitudinal crack had resulted in a BD blade failure under the postulated large-break LOCA loads, the BD blade is anticipated to fail longitudinally along the blade bend and separate from its mounting as one piece.
The estimated piece dimensions are approximately 21 inches wide and nine inches long, and it would have weighed approximately seven pounds.
Since the BD blade tip is captured under the beam of the adjacent blade, the failure is expected to result in rotational motion imposed by the back pressure causing the blade to be directed away from the flow stream.
The blade tip is supported by an adjacent blade that acts as a pivot point causing the blade to rotate away from the opening it creates in the damper when it detaches from its mounting.
The blade is also anticipated to travel in the direction of the air flow through the opening in the BD damper resulting from the blade i
failure.
The air flow would expand to fill the large volume in the j
fan housing and thereby generate a flow pattern toward tP sides of the fan housing. Therefore, there is a high probability that a blade l
would have hit the sides of the duct and dissipated most of its energy there instead of directly hitting the fan rotating element.
Using this geometry as an input, a calculation has been performed to determine the trajectory of a failed blade.
The results show that 4
only 2 of the 16 damper blades per CFCU have the potential to directly impact the fan.
Other blades, because of the initial rotational motion imparted, would initially impact the side of the ductwork or j
the inner fan scroll (see Figure 1 for damper fan geometry) and dissipate most of their energy.
Because of the postulated failure mechanism (brittle failure with crack propagation speed equal to the speed of sound in the metal), it i
is expected that the flawed damper blade would experience a " clean" break along its shoulder (highest stress concentration point) and would have the tendency to proceed along its trajectory without turning on end.
For the blades that can directly impact the fan, the 1069S/85K
~
-,-,n
,,n
,,a-
-ne e-,
--,e,,.-~m,.
-,e
,,-,----,nn--,
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 199599 omn,- m
==.- c. m u.j;;
.m DIABLO CANYON UNIT 1 Ol 5l 0l Gl 0l 2 l 7l 5 92
- 0l2l3 0l0 10 l"l 13 stat 07) trajectory does not result in a direct hit on the fan impeller blades since the damper blade length is wider than the fan wheel.
Since the impact area is relatively large, minimal fan damage is expected.
However, although judged unlikely, there is a remote possibility that a damper blade can turn on its edge before impacting the fan, or that a blade.that has hit elsewhere could rebound and then go through the limited area that leads directly tc the fan (see Figure 1).
PG&E has therefore, analyzed the effect of a blade impacting the impeller.
The separated fan blade weighs approximately seven pounds; the fan impeller is of rugged construction and weighs approximately 1336 pounds.
The impeller is made of carbon steel with good ductile properties and is capable of absorbing. impact and possible local deformation without fracture. While it is difficult to accurately quantify the potential for damage to the fan, simulated test and.
analyses show that the fan casing and internal structure wouH not likely experience significant damage from an impact with a biade piece.
It is judged, based on these tests and analyses, that if the damper blade strikes the impeller, the impeller blade could be dented to a depth of approximately one inch. The damper blade would most likely bounce off the fan impeller after impact.
Calculations show that the fan imbalance caused by this material displacement on the impeller would be insignificant.
In the event that the failed BD blade pieces did come into contact with the fan impaller, the fan blade curvature (backward incline) with respect to fan rotation would not have caused the piece to be drawn into the impeller.
The likely tendency would have been for the impeller to rebound the piece away from the impeller.
The potential interaction of a loose BD blade piece with the fan impeller has been discussed with the manufadurer of the CFCU fan.
While they indicate that the potential fan damage due to impact with a BD blade fragment is difficult to accurately characterize, their review and judgement concurs with PG&E regarding the ruggedness of the fan blades.
The possible effects of a BD blade fragment becoming jammed in some manner that could impair fan operation was also evaluated.
Review and evaluation of the fan impeller / housing physical configuration led to the judgement that it is unlikely that a blade fragment would become jammed and prevent the fan from performing its function.
The failure of a BD blade would also provide an open flow area that could potentially degrade the isolation function of the damper.
Degradation of the BD isolation function could permit reverse air flow through a CFCU and impose reverse rotation forces on the CFCU fan.
Investigations for LER l-91-019-00 have shown that for the CFCUs with their fans running, a complete BD could fail open in the event of a 10695/85K
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION 1995 M ractutt kant p) oocati e sta (2) tra e arn si east (1)
"C7 "l%
DI O CANYON Uf4IT 1 0l 5l 0l 0l 0l 2 l 7l 5 92 012l3 0l0 11 l 'l 13 i: 01)
LOCA without the fan experiencing unacceptable reverse rotation or adversely affecting the fan structural integrity.
Fans that are not running are susceptible to higher levels of reverse rotation.
Studies showed that one BD blade could be open in this case without creating problems with motor restart. Also, the reverse rotation and failure to start would not cause damage to the component or switchgear.
The switchgear can be reset and the motor started after the fan's reverse rotation has diminished.
E Normal operation has not resulted in' blade failure.
Failurh during normal operation does not present a safety concern.
If a loose piece is generated, it would fall and come to rest in the downstream duct without adverse consequences.
The Unit 1 CFCU BD blades have been replaced.
Unit 2.is in power operation with approximately five montn; of operation until its scheduled refueling in March 1993.
As a conservative action, the operability of the Unit 2 CFCus will be maintained by implementing a design change to remove all the blades and by placing all the CFCUs in operation. An operating fan will not reach a reverse speed that will impair its capability to restart in the event of a DB LOCA.
A fan not operating will be declared inoperable as a conservative measure.
The design change includes a safety evaluation review under 10 CFR 50.59 that demonstrates there is no safety concern introduced by the modification.
This Unit 2 BD blade and fan operating configuration will be maintained until new blades are installed during the Unit 2 fiith refueling outage (2RS), at which time the normal damper function will be restored.
Based on the above, the health and safety of the public were not adversely affected by this condition.
2.
Late Submittal of LER This LER was due to the NRC on November 16, 1992.
However, due to information discovered on November 13, 1992, that required revision to the LER safety analysis, this LER was not submitted in accordance with the required 30-day time period.
Submittal of this LER four days late has not adversely affected the health i
and safety of the public.
1
{
V.
forrective Actions l
A.
Immediate Corrective Actions:
4 1.
CFCU BD Blades l
1069S/85K
k~
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION
-m fumm - m u y.,
um -m m DIABLO CANYON UNIT 1 0] 5l 01010l 2 l 715 92
- 0l2l3 0l0 12 \\"l 13 TEXT (17)
Unit 1 - Unit I was in a refueling outage at the time of discovery of this condition, therefore no immediate corrective actions were required.
Unit 2 - All BD blades are being removed and all CFCUs will be maintained in operation.
2.
Late Submittal of LER No corrective action was deemed necessary. The late sdbmittal was required to ensure technical accuracy of the LER.
B.
Corrective Actions to Prevent Recurrence:
1.
CFCU BD Blades a.
All Unit 1 CFCU BD blades have been replaced with blades fabricated from a material more resistent to fatigue cracking.
b.
All Unit 2 CFCU BD blades will be replaced with blades fabricated from a material more resistent to f atigue cracking during 2R5.
2.
Late Submittal of LER No corrective action was deemed necessary.
It is standard practice for PG&E to submit LERs on time as required by 10 CFR 50.73, NUREG 1022, and the Diablo Canyon Technical Specifications. However, information that became available on November 13, 1992, after the LER had been written, reviewed by PG&E management, and prepared for submittal made a subst ntial change in the analysis section of the LER.
Ratner than submit a document that was known to require revision, PG&E delayed submittal of the LER so that the analysis section of the LER could be revised to incorporate the new information.
VI.
Additional Information A.
Failed Components:
CFCU Discharge Backdraft Damper, manufactured by American Warming and Ventilation, model nunber SHW-P-7900-18.
B.
Previous Similar Events:
None.
10695/85K
LICENSEE EVENT REPORT (LER) TEXT CONTINUATION ggg g FACILITY M (1)
COCKET Nt79BEA (2)
((R NL74W(3 f4)
PAG ( ()} yI TEAR 3Gamstrtu nyvesog i
DIABLO CANYON UNIT 1 0l 5l 0l 0l Ol 2 l 7l 5 92
- 0l2l3 0l0 13 l 'l 13 ftsi (17)
FIGURE 1 i
CFCU FAN AND BACKDRAFT DAMPER CONFIGURATION I
l h
t
- {
l
^
.2.
)
c<
r.-
..1 p.
' I r*
e q p--(
I'----
\\j
~
~
,1],
._jg u.ie-r m--
i
.'... BJ ".: sL !
- q a
.,. _ -.l
.t g g.,
'~YfETYU~Y 2 J
/
$=.-
ELEVATION VIEW A-A sca vr.r-e oinume or 1'
FAM ROTOR
{
,- IA.-,4
/.._
g g
l oret
-s
{
f M
- h N
- Eenou 3.-a
/
7 g
sea vr.. -e
- f3UPPORTBARJ l
o L -. V-U
.Y4 PLAN VIEW
- tf"At t 1/2" e l'-(f" 0695/85K
.-