ML20105A847

Rev 0 to Comanche Peak Steam Electric Station Unit 2 Engineering Bwip Check Valve 2AF-0083 Failure Investigation

ML20105A847
Person / Time
Site:	Comanche Peak
Issue date:	08/06/1992
From:	TEXAS UTILITIES ELECTRIC CO. (TU ELECTRIC)
To:
Shared Package
ML20105A845	List: ML20105A842 ML20105A847
References
PTR-32, PTR-32-R, PTR-32-R00, NUDOCS 9209180015
Download: ML20105A847 (14)

Text

.. -.

l l

TU ELECTRIC l

COMANCHE PEAK STEAM ELECTRIC STATION i

UNIT 2 ENGINEERING BWIP CHECK VALVE 2AF-0083 FAILURE INVESTIGATION PTR-32, Revision 0 i

AUGUST 1992 l

Prepared 8 $

Md

\\

c&-

%L LeairDistipi ginee r Date

$8 f5 Engineering Assurance Manager Datee Cfl b

A >

l Pr oc C fngi

, Jcoge S Djdte (

X J sadne'JLV shhz i

l.

<pt:t Engine' scope C

~

i

._ w Oer, j

Date f f

- 8 /7 /9 2_.

u

'fNI/WIE

• Yh'bL EngineeringManager, knit 2 i

920918001kQO 46 PDR ADOC PDR.

S

PTR-32, Rev. O Page 2 of 10 UNIT 2 ENGINEEMNG BWIP CHECK VALVE 2AF-0083 FAILURE INVESTIGATION 7-BACKGROUND AND PFtOSE

'this engineering report provides.:.

mmary of the investiga-tion into the cause of the failurr c1 valve 2AF-Obd3.

on July 3, 1992, Borg Warner International Products (BVIP) w

' heck valve ;.hF-0083 failed a baa.kflow test being performed

.cordance with Unit 2 pecoperational test procedure 2PT-3

.1.

The valve was subsequently radiographed, disassem-

[.;

e., and found to have a broken disk-stud assembly.

The F

.k vu locat?.d on the stud just above the disk to stud 9

.d (Rel'erence B).

a valve is located in the motor driven auxiliary reedwa-c-

purp (MDAFWP) 2-03 discharge line to steam generator (d;) number 2 and is relied upon to prevent backflow from the feedwater system when the associated MDAFWP is not l'

running.

T=.ven (7) other valves perform similar functions

)

v in the as iary feedwater system.

II.

IMWEDIATE ACTIONS TAKEN A.

,-.evier was performed of the opera.ional history of the 23Th+ (8) BWIP check valves.

The review indicates val'.or CAF-0075/83/93/101 associated with the MDAFWPs hc.a best, in service.

All four (4) valves were disas-sembled and visually examined.

Valvec 2AF-0075/93/101 had 2.;ar patterns on their backstops.

Valve 2AF-0083 had a major indentation on its backstop which was mapped by QC prior to reaesembly of the valve (Refer-ence Q).

B.

Because Unit 1 has performed maximum expected differe7-o tial tests (MEDP) tests and subsequent backflow checks, and no other stud failures were identified, Unit i valves were considered operable.

In addition, a pre-q liminary review determined that a modification nr backstop) had been added to the Unit 2 valves.

This modification appeared to increase stresses in the Unit O. valves.

9 PTR-32, Rev. O Page 3 of 10 C.

Since the cause of the failure was unknown at the time i

of occurrence, Unit 2 was administratively limited to a flow rate of 300 gpm through each of these valves, This flow rate corresponds to the maximum expected i

5.

flow rate during normal operation.

The valves were reworked and the systems placed.in service to support j

HFT (Reference N).

D.

A review of startup testing revealed that there was a i

flow of approximately 670 gpm'through 2AF-0083 on 5/25-

/92 and only 580 gpm when performing-the runout flow I

and MEDP tests on'7/2/92 (Reference-P). Tnis data indicated the valve had failed in the time period i

between these tests.

i III. INVESTIGATION In order to determine the cause of the failure of valve 2AF-0083, metallurgical examinations, transient analyseu, system and valve design reviews, and an operating history review were performed. In addition, a review was performad for generic implications.

A.

Metallurgical Examination (Reference A).

The report 4

stated:

t The chemical-composition,: physical proper-ties, metallurgical structure,,and appearance of the fracture surface _were evaluated tc determine the failure mechanism of the stud.

1

?

-The fracture occurred due to ctress overload at the convergence of_the stud heat-affected zone, a thread root, and the fillet weld toe extending from the disk.

Tha part was sub-

_jected to a high energy, complex load with components of axial tension and bending.

At the terminus of the fracture, the_ stud had.

significant cold work in the microsegregated microstructure. Due~to the high rate of load-ing, little evidence ofLductile stretching or necking was visible to the unaided eye....

NOTE:

Reference A is currently _under final review.

i B.

Transient. Analyses / Investigation Design basis transients have bean calculated (Reference

[

R) for pump start and a pump stop for a water solid j

~

i

.ondition The force.in the piping-segment containing-the check valves is very small with an order of magni-i b

e 2

+, -,,

r

~

>we.-e---a,

.e,me1 w+--se

&=

e- - - = - - -

m

.~r--

,e-,+e--

w e..es

-+%--

e

t 4

PTR-32, Rev. O Page 4 of 10 tude of about 20 lbf for pump start and 572 lbf for pump stop. These forces are not capable of causing the failure whict occurred.

4 A non-design basis transient which results in a slug fluid velocity of 29.48 ft/sec and a force of approxi-mately 15,000 lbf was determined using the Jaukowski equation, assuming a partially drained system (Attach-ment 3).

A walkdown of the system was performed as described in Reference E.

No damage to supports or evidence of large pipe movements was found.

The check valves are located just upstream o r anchors, and the forces trans-mitted would have been vertical inte 'Te valve bonnst.

Therefore, due to the general stiffnews of the piping configuration it was unlikely to have found any support damage or indications of large pipe movements.

An auxiliary feedwater system elevation diagram was developed (Attachrent 1).

This diagram shows that the system has a potential to drain down to the elevation of water (853'-4" to 834'-3") in the condensate storage tank (CST) which is between 8" to 19'-9" below the check valve which corresponds to between 29 linear ft to 70 linear ft.

This drain down could occur when the associated steam generator is not receiving flow through the preheater nozzle, the steam generator is at atmospheric pressure, and in addition if the check valve has any leakage.

The check valve in question had not been backflow tested prior to the failure.

C.

Stress s:alysis u

A dynamic strest. analysis was performed (Reference D) of the val"a $ith and without a backstop.

The analysis evaluated the mechanisms which could cause the failure of the disk-stud and estimated the lower bound of the force required to break the stud under a one-time event to be between 9.6 to 10.9 kips.

A disk velocity of 30 ft/sec was used, which corresponds to the upper bound fluid transient analysis velocity of 29.48 Ft/sec.

The corresponding reaction forces predicted.on the bennet backstop are consistent with a static test performed at bWIP which resulted in an indentation of 24 mils for a load of 8 kips.

The greater energy involved in a dynamic collision would probably cause a greater inden-tation.

The actual-measured indentation was 64 mils (Reference Q).

4 PTR-32, Rev. O Page 5 of 10 D.

system operation The auxiliary feedwater is normally operated to main-tain level in the steam generators during startup and shutdown ope, rations (modes 2 and 3) until the feedwater system is placed in operation. The system is also used to fill and maintain level in the steam generators. The system is operated in accordance with SOP-304-B (Refer-ence M).

The system is initially filled and vented up to the isolation valves. The flow control valves and isolation valves are then fully opened and the piping monitored for temperature to determine if the check valves are leaking. This check would not La meaningful with the steam generator at atmospheric / ambient condi-tions.

E.

System Design Review (Reference J)

The auxiliary feedwater systes is designed to provide the following flow rates through the subject check valves:

1.

Loss of feedwater 225 gpm minimum 2.

Feedwater line break 265 gpm minimum to non-faulted generator 70C ypm maximum to fault-ed. generator for MOAFWFs 680 gpm. maximum for TDAFWP 3.

Hot standby 235 gpm minimum 4.

Plant startup 300 gpm minimum 5.

Normal plant startup 40 to 275 gpm and shutdown The subject check valves are not relied upon to prevent backflow in the event of a feedwater line break to the faulted steam generator.

The flow control valves on the MDAFWP trains differ from the TDAFWP flow control valves in that they-are provided with an automated feature which drives them to a full open position on an auto MDAFP initiation signal.

The flow control valves

-* normally open when the system has been filled and

. ted and the pumps are not in operation. The motor operated isolation / block valves are normally open and fail as is.

PTR-32, Rev. O Page 6 of 10 F.

BWIP Check Valve Design Review (Reference I)

There are 25 pressure seal bonnet BWlP check valves in Unit 2 which were modified.

Bolted bonnet check valves were not modified and therefore were excluded from the review.

There are eight (8) 4" diamatar valves in the auxiliary feedwater system which have been modified by adding counter weights, backstops, and a height ad]ust-ment spacer. The corresponding eight (8) valves in Unit 1 have added counter weights.

No height adjustment spacer or backstop was added.

This allows the Unit i valves to swing further out of the flow' stream and significantly reduces tens.il stress on the stud during transients similar to the event discussed in this report.

The other seventeen (17) valves 17 Unit 2 have backstops and height adjustment spacers, but since the do not have counter weights the disk-studs were gener y ally shortened to allow the disk swing angle to be near to those in Unit 1 which do not have backstops (Refer-ence 1).

G.

Operating History Review Flow was provided thirteen (13) times to steam genera-tor Number 2 prior to discovery of the failed valve.

There were periods of up to six (6) to ten (10) days between pump operations, including one inadvertent auto start on 7/1/92 (Reference K).

The first two operations of this valve, on May 26, 1992, were at increasing flow rates up to maximum flow (670 gpm from HDAFWP 01 and 650 gpm from MDAFWP 2-02 to SG #2) to determine if test line' restriction orifice was the cause of pump vibration.

The third operation of the valve, on May 26, 1992, was to fill SG #2. No flow rate was recorded.

The fourth operatio

.i the valve, on June 2, 1992, no reasor. or flow rate wa. provided.

The tifth operation of the valve, on June 11, 1992, MDAFWP 2-01 was run on minimum flow and to fill SG #2.

No flow rate to SG #2 was provided.

The sixth operation of the valve, on Juna 12, 1992, was to verify the test line restriction orifice modifica-tion was acceptable.

MDAFWP 2-01 was run on minimum flow and through the test line, not ts the steam gener-ator.

ator.

PTR-32, Rev. O Page 7 of 10 The seventh operation of the valve, on June 21,

1992, MDAFWP 2-02 was run on minimum tiow and to feed SG #2.

No flow rate was recorded.

The eighth operation of the valve, on June 21, 1992, s

was a backflow test on the minimum flow check valves.

MDAFWP 2-01 was run on minimum flow, not to the steam generator.

The ninth and tenth operations of the valve, 1992, was to fill and vent.

on July 1, The eleventh operation of the valve, also on July 1, 1992, was an inadvertent auto start of MDAFWP 2-01, which is recorded in Reference K.

The flow control valve would have been in a full open position after the fill and vent.

It is also possible for drain down to the CST level to have occurred.

This appears to have been the most probable system configuration to develop the transient which damaged the valve.

i The twelfth and thirteenth operations of the valve, July 2, 1992, the MDAFWPs 2-01 and 2-02 were again on operated to fill, vent, and perform runout and MEDP testing in accordance with 2CP-PT-37-01.

l IV.

CONCLUSIONS A.

Cause i

Based on the evidence available to date, the fai)ure of 2AF-0083 was most likely dua_to a system fluid tran-sient which was the result of starting MDAFWP 2-01 into a partially drained system.with the flow control and isolation valves fully open.

This conclusion is based on the following:

1.

A potential for a partial-system drain down.in conjunction with atmcspherze pressure in the steam generators.

The potential for a similar occur-rence in the operating Unit is. low due to the administrative controls and physical barriers described in Attachment 2.

2.

Evidence of a rapid failure from the metallurgical examination. The failure was not due to incorrect and/or defective material / components.

s

i PTR-32, Rev. O Page 8 of 10 4

3.

The stress analysis is consistent with the tran-sient analysis and metallurgical examination.

B.

Contributing Factors 1.

The backstop installed in the tinit 2 valves allows a higher stress to be developed in the Unit,

disk-stud than if the backstop were not installed.

There is at least approximately 40% more stress in Unit 2 (due to the backstop) for the same tran-l sient.

2.

An orifice is installed upstream of each of the check valves to limit flow through the line in the event of a feedwater line break.

For normal oper-ating bases the orifice has little effect on the flow.

It is difficult to quantify the effect of the orifice at high flow and low backpressure conditions, and it may have contributed in a sig-nificant way to turbulence.

3.

The disk-stud is a S/8" diameter all thread rod with a 0.507" root diameter.

This provides both a stress riser and a " notch effect" where the fail-ure occurred.

V.

CChRECTIVE ACTIONS A.

The eight (8) Unit 2 check valves IAF-0075/78/83/86/93-

/98/101/105

. ill have their backstops removed.to make w

them less susceptible to failure prior to proceeding with high flow / low steam generator pressure testing.

i B.

SOP 304A and B, " Auxiliary Feedwater System," will be revised in accordance with Attachment 2.

C.

BWIP will be requested to provide replacement disk-sten assemblies with a shank instead of an all-thread stud for the eight (8) valves in each unit.

The new disk-stud assemblies will be installed as replacements for the disk-studs are required.

4 D.

7.he valves will be acoustically monitored during these conditions to determine what eftect the oritice has on the valves following 1:FT.

PTR-32, Rev. O Page 9 of 10 l

VI.

REFERCNCES A.

Failure Analysis of BWIP Swing Check Valve 2AF-0083 disk-stud, VL-10295 B.

TUE-92-5719 C.

CPSES-9224904 dated July 29, 1992, subject:

Unit 2 Valve Modifications.

J.

Calculation 0218-SQ-0096 E.

CPSES-92 224.: 7 dated July 15, 1992, subject:

Auxiliary Feedwater Pipe Support Walkdown Subsequent te Transient j

Event.

F.

Technical Evaluation TE-92-001445 i

G.

TU Electric office memorandum U20P-92252 dated July 27, 1992, subject:

Run Times of MDAFWPs H.

DCA-94663, Rev. 3 I.

CPSES-9225677 Interoffice memorandum JO/WO: 01531.02 dated July 22, 1992, subject:

Auxiliary Feedwater Check Valve Modification History.

J.

DDB-ME-206, Revision 6, Auxiliary Feedwater System i

K.

ONE Form 92-629 L.

Interoffice correspondence from Jim Sabin of S & W to Glenn Milley of S & W, subject:

CP2-Pump Start Empty Column.

M.

SOP-304A, Rev. 9, Auxiliary Feedwater System N.

CPSES-9221931 dated July 9, 1992, subject:

Justifica-tion for hot functional testing.

O.

Flow Diagram M2-206 Auxiliary Feedwater System i

P.

2LP-PT-37-02 Q.

SWP-20057 R.-

Calculation 1561600-F020, Rev.-1, " Water Hammer Analy-l sis for Auxiliary Feedwater" l

l I

m

PTR-32, Rev.~ 0 Page_10 Of 10 VII. ATTACwumuTS 1.

Elevation _ Sketch of-Auxiliary Feedwater System 2.

Interoffice-Memorandum, Jim Brau to John Roberts, dated August 4, 1992 3.

. Interoffice Correspondence from Jim Sabin to Glen Milley, dated July 24, 1992 a__-----axu_-u----a.aan------

--1.

n.wa.-a-.--.e.- - - - - -. -aa

4 Tee b

P a.e

1. 14 e,

rA a,

tw, he A

W

.m

~ _"mf l

Wl I

. I EP, [

i W

1r j

Ik 1

I 1

E '

I E

I h

E E

E L

Y W

' ~~

! m' M

i g : ' I -

E.

1 P '

. m FE -

M

-1 h,.

M M

sp w

WW M

I M

m I

M ~

h

' M I

~

Ig 1 -

1

! 4 e

1 4

L E

y I

m -. F I

E k _

a,

e i

X AI 7

_M.

I

. mI i

1 a

]

~ I '

- _. _ _E M

3 I

I '

I M

a.

I.

1

' 1 1

N 1

m.

i I

4 3

N' l '

I m

• I f6 5f a lN 0

i be e

k 1

I e

I i

M.

[

E E

_a y

E'-

'U d

'M.

E

'm s

b I

I

]

s m

__% w

'l

-' M 1

V E

1

'N J

r F

k U_

l I

j 1-m I

j, jh M

W. -

1 mp' FX min '

4 uk C'

'T m-I

[,

1 '

)

hh :

N

[

^7' W

..a my= ?

v j

i W

1 M

WM '

W lM l

L Mr :

,f" J

I

^

I I

' V'%n i

(

- I im,<

' i g

i m

L-b-----

.L R

' ~

k m.

amy p

ws1 :m h

y

. h I

M-

^

, em v m-

- h.

I, T'

y 4M

.' WM'

~ '

I m

1

[

ah.'m:

ik

.I I.

1a' E 2 ;

F" 1

-A_

1 m..

,m m

y I,

'a.

I W'

M M

'I 1

n l

F~

.1-EF av.

p

~ V An TET E

a l

l A

IA M

1.

Jc"

'm RW M

> m.

m ft 2

^ :

wn.

6

-mik O

e 1-W X m

.dI ;

w 4

E I'_

.w 9

A EM F

Li I

C Ir!

Em J

m A,

I n A =

2 i, __

H*

[

M M

I m-.

J M

r pas W

l

,r.,

1_

r i

i J 1 m

's r

H 4 w

i b

f R

i I

& F ' E _

^g^

(

(

• ". mL M j,

1 - _-.. -

m..

-y.

E ww m

=

a g-3 C._

iM' g

g-E i,

g 7

I' a-E I

R gm m

'M

- y y

RW' M a

E m.

EW E

-e-en.

.1 I

a e

h1. Em

=

"'1 4

M I N

1 y

3 g

.- 3 m M

s a e _

^-

1 -

ak

'UE A.

mW T

]

5 3

-y

+4 -

m=~

3 5 I

iiumd -

• ' l '?

I

.m

'sy -

I w

3

[

r ua w

am t

A w

Bad E

t i

meo a

'A I

a

'I

.L'.

E' M1 i

E' 9

l MI _-

m l

F1 1

M-

.Ruh.

TF ma ai

' A I

t N,$ $$.-&".

1-i 7-9-

_\\,_

y -

_A _w i

.) i M maie -ee s%

s eeswsff e g h -

. m.f v

A M

^

M u

r n

' 5E :

AJ m,

m i

m A

w-e--

j I

m.

g ;

y I

~ F m--.1 W 1 I

I Mi M'r w :

y M -

- W' y.

I u-h -

I ' I 33 E '

L e I

i -

w,(,

f, j

m-

~.

I PTR-32 Rev. 0 ATTACIDENT 2 Pcgs 1 of 2

!"!1UELECTRIC OFFICEMEMORANDUM

?

To John Aoberts August 4, 1992 subject: Damage to Unit 2 Adailisry Feedwater check Valve 1

i This letter is in response to your discussion with my staff on August 4, 1992.

The discussion addressed damage to check valve '!AF-0083 d,i to an inadvertent start of the Motor Driven Auxiliary Feedwater Pump. Questions were raised if adequate steps exist in sop-3045, Auxiliary Feedwater Pump, to prevent this reoccurrence.

1 An evaluation was performed on 800-3045 for a'll modes of operation. The results of the evaluation are as follows:

Modes 1

2. andJ In Mode 1, 2 and 3, OPT-2065 is performed to ensure eyeten operabilLty.

The

~

Motor Driven and Turbine Driven Auxiliary Feedwater Pump-flow control valves are required to be fully open per Technical specifier.tions.

In Mode 3, the i

fuses for the Main Foodwater Pump trip signal are removed which reduce the possibility of an inadvertent pump start. A manual start of the Auxiliary Peedwater Pumps with the control valves open would be a violation of procedure.

In addition, the piping downstreas of the check valves would be preswurized due to Auxiliary and Main.Paedwater System opa. ration.

This would reduce the differance in pressure across tha -heck valve which woald in turn reduce the forces applied to the valve.

Any back 2eakage of the check valves would be sensed by upstreata temperature monitors.

Recommendations i

A change to SOP-3043 for Modes 1, 2, mad 3 is not' W red.

Mode 4 In Mode 4 the Motor Driven and Turbine Driven Auxiliary Feedwater Pumps flow cor/.rol valves are fully open.

The fuses for the Main Feedwater Pumpe are J

removed in Mode 3 which reduces the possibility of an-inadvertent pump start.

A manual start of the Auxiliary Feedwater Pumpe with the flow control valves open would be in violation of the procedure.

Recommendation A change to 50r-3043 is not required.

Changing the sop to close the flow control valves in Mode 4 would misalign the Ana111ary Feedwater System prior 1-to the transition to Mode 3.

- The fuse r for the Main reedwater Pumpe are removed which reduces the pose'.nility of an inadvertent pump start. - In addition, the suggested chang could increase the potential for systeen misalignment at a time when multiple activities are in progrees during the i

Mode change.

i i

t i

w 1-

,,.y

. -,r w

3-,

+ - -

yi:-.

o,-

.w

-3

,rn

-w v---r we-

--,2+

PTR-32, Rov. 0 ATTACHhi.NT 2 Pago 2 of 2 4

11 Medes 5 and 6 In Modes 5 and 6, the Motor Driven 'and Turbitse Driven Auxiliary Feedwater Pumps flow control valves.are fully open.- The SSPS is placed in Mode 5/6 i

which removes all automatic start sirjnals from inadvertently starting the Auxiliary Feadwater Pumps. Due to a void that would exist upstream and-downstream of the chech valves, a manual start of the Auxiliary Foodwater Pump could result in check valve damage.

t Recommendations f

r:haage SOP-304E to close the flus control valves in the shutdown and standby operations while in Modes 5 and 6.

start from possibly damaging the check valves.This would prevent an inadvertent pump If you as.,m'any questions os extension 5443.

c-saments centact myself or Kit Wilson at

'[uM'%

i Jim Brau 011 5

Supervisor, Operations Support 4

4-4

+

d 4

4-4 s

l

m.

~

ATTACHMENT 3 Pago 1 of 1 INTEROFFICE CORRESPONDENCE to

-~

iI,oca E I / E m /2. No.

an Millev

/9 CP2 - % N % - W i m Jim Sabin 1145/9

-m As you requested, this ICC will document the data previously provided to you on the telephone regarding column rsjoining following the AF actor driven pump start.

Water ha m r pressure waves aretted by void collapse can be determined by the Joukowsky equation AP = p a AV and tJte impact forse caused by over-pressurisation after a voie colltpee is calculated.

by F = AP'A.

whores a

= jeessure wave speed (* 3715. fps) a consity of tua 11guia (A.ws slugs /rt')

=

eV = Ve_ocity of the liquid just prior-to impact (ft/sec).

(From the manufactures pump performance curve, the runout flow rate is a 9J0 gpa = 0.408(950/3.526 ) = 29 48 ft/sec - for a 48 5

sch 120 line)

AP = {(1.94) (3715) (29.48) }/144 = 1474. psi 3

T = 1474(ft(3.626 )/4) = 13,220, lbs 0

MM

/

$ vast suly 24, 2n2 DATE

//

SEDfAttME

)(

~h M

/

/

1 i

., _., -., _,... _..... _. -.. _, -,..,. -.,,