ML20213A460

Discusses 861212 Visit to Plant Re 861209 Piping Failure. Addition of Snubber or Piping Changes May Be Necessary.Addl Info from Util Needed

ML20213A460
Person / Time
Site:	Surry
Issue date:	12/15/1986
From:	Rodabaugh E E.C. RODABAUGH ASSOCIATES, INC., Advisory Committee on Reactor Safeguards
To:	Saffell B Battelle Memorial Institute, COLUMBUS LABORATORIES
References
RTR-NUREG-CR-4305 ACRS-CT-1876, NUDOCS 8702030289
Download: ML20213A460 (10)

Text

-

f..

[f-/87f

, E. G. cR ~ L Lmin cauoatata, Baa.

4625 CEMETERY ROAD e HILLIARD. OHIO 43026 gj j/g8/81 614/876-5719 -

December 15, 1986

~ .

Mr. Bernie Saffen Battelle-Columbus Laborratory 505 King Avenue ~

Columbus, Ohio 43201 Subj ect: Surry-Unit 2 Feedvater Piping Failure, December 9, 1986

Dear Bernie:

I went to the Surry plant on 12/12/86, along with NRC personnel.

In view of the failure occurring in, by NRC' classification, non.-essential piping, I was pleasantly surprised by the openness and seeming viningness of VEPCO to supply us with relevant data on the failure. Hopefuny, VEPCO vin continue to cooperate with NRC as the investigation proceeds. In turn, hope-fully, NRC vin be able to allot resources to the investigation so as to be able to thoroughly understand this failure and thus help to avoid future sini-lar failures. Following portions of this $etter implicitly assume that VEPCO cooperation and NRC involvement will continue.

Deseeiption of Failure Enclosure 1 is a schematic of the main feedvater system. Enclosure 2 is ,

the best description of the piping system that VEPCO had available. VEPCO agreed to provide a dimensioned isometric. Enclosure 3 is my to-scale sketch of the failbre area. At the time of my visit, VEPCO was not sure if they d all the pieces and had not tried to fit the pieces back into the elbow before-failure. Accordingly, my " guess" in Enclosare 3 any not be right.

DESIGNATED ORIGINAL 0702030289 861215

-18 PDR Cortified By G N_,/A.

9 '

2 P

s Heasurements of vall thicknesses (0.500" nominal origina11y), after the failure, ranged down to 0.1" away from edges; and perhaps down to 1/16" at edges. The internal pressure was estimated by VEPCO to be about 450 psi at I b ' th*

the time of failure; temperature about 375F. However, VEPCO was not sure that p, t.hlk s5 - ti,4 a pressure surge did not occur.at the time of failure. The ultimate strength /

of A106 Grade B material at 375F would be expected to be about 60 to 70 ksi.

Using the relationship, P = 2S gt/D (reasonably accurate for an elbow along the line marked L in Encl. 3) gives P = 667 psi for t assumed to be 0.100";

t = 0.0675" for P assumed to be 450 psi; both for S,= 60 kai.

Leak-Before-Break Implications Failure occurred almost instantaneously. However, based on eye-witness ac-counts related by VEPCO, there may have been a few seconds (maybe 5 or 10) of leakage before the break. Obviously, even if a 10 second leak did exist, it vann't sufficient to make the leak-before-break concept ,useful.

My present impression is that the failure vill be essentially explainable by considering it as what I termed,in NUREQ/CR-4305*, a "no crack" type of l

break. I dso pointed out that reduced vall thickness due to general corrosion and/or erosion is one of the ways to get to the "no crack" type of break. In-deed, for the "no crack" type of break, the unusual aspect of the Surry failure may be the ~10 second leak that perhaps existed prior to the break. The problem with the "no crack" break is that, once vall penetration occurs, there isn't any extra strength in the pressure boundary to pick up the load; so the cracks keep going until, the pressure drops off sufficiently.

" Comments on the Leak-Before-Break Concept for Nuclear Power Plant Piping Systems", E. C. Rodabaugh, August 1985

3 I presume that all of the current leak-before-break theories, when applied to the Surry failure and assuming any pre-existing crack was very small, will alsoarriveattheanswerthatbreakwouldoccursimulboouslywithleak.

Perhaps the Surry failure message, in relation to leak-before-break con-cepts, is that we shouldn't forget about the possibility of a "no crack" type of break. On p. 17 of NUREQ/CR-4305, I listed three possible reasons for the "no crack" type of break; there may be others.

Pipe Whip Restraints There were no pipe whip restraints. The broken-end moved to the point in-dicated by the dashed line on Encl. 2. The elbow identified as E6 was badly tvisted and collapsed; but I saw no evidence of pressure , boundary failure in it.

It was not possible to tell if any of the other elbows (E2-E5) had undergone any gross plastic deformation. There was a relatively flimsy, floor-mounted support for an instrument that was bent out of the path of the moving pipe. Otherwise, there was no apparent external damage.

In NUREG/Ch-4305, p. 37, I expressed th'e thought that installing pipe whip l restraints so they vill really restrain the pipe, in accor h ee with the design intent, isn't so simple. The piping system that failed is a good example of one of the aspects I had in mind. Suppose we applied SRP 3.6.2 to this piping.(It is

'high-energy piping). We would have to postulate DEGB's at the terminal ends; the restraints vould have to resist thrust in the X-direction at elbov El and thrust i in the Y-direction from the pump flange. Neither restraint would have restrict-i ed the pipe whip that occurred. SRP 3.6.2 formerly required a couple of inte d mediate DEGB's. But one can select two (or more) intermediate locations and 6

i 4

E i

still not have the needed restraint. A restraint at elbov E2 in the Z-direction is what would have kept the broken pipe from whipping., My point is that with DEGB's postulated under SRP 3.6.2, the pr' obability that a restraint vill be lo- .

cated at the right place to prevent pipe whip might be no better than 50%.

Of course, construction of the pipe whip restraints, whether effective or not, would entail a massive structure because>the header is about 20 or 25 feet above the floor. Based on present day methods of design pipe whip restraints, the structure would probably dwarf the piping system itself.

Along the lines of des'igning pipe whip restraints, I think the Surry-2 fail-ure would provide a field check on those design methods. To help accomplish this, the remaining pipe down to the pump flange should be thoroughly inspect-ed for evidence of gross plastic deformations. It would be helpful to index the flanged joint at the pump, then remove the piping to a more convenient loca-tion for inspection and storage. Eventually, coupons should be cut out for aa-terial tests. .

Is There an " Essential" Pipine Generic Eros' ion Problem?

The failure was in main feedwater piping in the turbine building. It is

, "Non-Essential" piping as NRC uses the term. (The auxiliary feedwater piping is " Essential"; I presume the piping failure did not impair the functioning of the auxiliary feedwater system.)

j At the time of our visit, VEPCO had shut down Unib1 and were getting ready to make ultrasonic vall thickness checks on the Unit-1 piping corresponding to the failed piping in Unit-2. Once those checks are available, we should htye

, a better feel for the generality of the erosion problem.

1 .

- - - - , , -- - - - , - - . - , - - , .e-- . . - - - - - - - - - - , - - - - - . .m ..

.. ---..:~

~

O 5

Surry-2startedcommercialoperationon5/1/73;Surry-1on12/22/72 There has been considerable down-time caused, in major part, by steam generator tu1>-

/

ing problems. - -

If the verbal infonnation I obtained, P = 450 psig, at 375F, represents normal operating conditions, then there is a fairly substantial margin against cavitation (bubbleformation). The saturation, pressure at T = 375 is about D 9 185 psia; the saturation temperature at P = 465 psia is about 460F. Could the erosion have occurred without cavitation? There are many details of operation procedures, ficv velocities, detailed pattern of erosion, details of piping, etc.,

that seemingly could be relevant to the erosion-cause question.

Mike Mayfield pointed out a detailed aspect of design which might have some bearing on the erosion. The branch connection was fabricated; hence, its in-side contour looked like as indicated in Enclosure 4. Most piping designers, for a 24x18 branch connection, would use a manufactured tee with inside corner more-or-less as indicated in Enclosure 4. Would the rounded inside corners of a manufactured tee reduce the amount of erosion?

What implication (if any) does the Surry-2 failure have on essential pip-ing? I think we need to thoroughly understand the cause of the Surry-2 erosion before ve really know. The auxiliary feedwater has to connect into the main a

feedwater lines (or, maybe, through another nozzle in the steam generator?).

But, regardless, it would be an infrequently used line. At this stage, I am a bit concerned about what we know or don't know about erosion rates (particular-ly in elbovs) in main steam lines. ASME Code Section XI, Inservice Inspection, concentrates almost entirely in looking for cracks in pressure boundary velds.

I assume that they also measure the vall thickness; but I don't know whether they record the vall thickness.Section XI doesn't require any inspection away

- - , --r- ,, -

, -u - ,-n--n-----.- - --- -- - - - - - - - - - -

~

9 6 from velds; e.g., at the mid-back-vall of elbows or more than a couple of incl >-

es from branch connection openings. I suggest that NRC should ask licensees to state what they know about the erosion rate in their mainsteam lines. This should be a relatively quick process; future NRC action would then depend upon what answers were received.

Pump Piping Load Capacity Pu=p manufacturers tend to give very low allowable piping loads on the pu=p nozzles. In some cases, it may be necessary to add snubbers or othen-vise change the piping system to meet these allovable loads. In the Surry-2 failure, very high loads were applied to the feedvater pump nozzle. I think we can reasonably approximate the magnitude of those loads. If VEPCO could furnish answers to:

(a)Was any repair of the pump needed prior to placing it back in service? .

(b)What is the description of the pump, its driver, the pump /driverconnectionandbedplate(ifused)?

then we might have one "real life" calibration of pump piping load capacity.

If the answer to (a) is "Yes", then there might be no point in pursuing (b).

Yours very truly, SW ECR;mr E. C. Roda W gh Encls: 1, 2, 3 and 4 cc: Paul Shevmon Mike Mayfield

.< ,e i .

rw E ATERS

! D ,g,-c= === a 1

i '

IM y . - ~

j e5 - _

g '

i

! SM k,  ! kt hiA h 4 kt A i, j

.. ry vs

,, ,,,,,, ~

.my i

falldrc .

v _

rousans 1 . w,o, ,, -

4 &m, -

u~O (~' ~ '- )

~

~'

is > >

is_ , u ^ '

em 1 X7 - p

_q d-I i SM kt e e

a e NA 'NvQ .Z _N -

f s

ki 33 ,

i 0 84 g

34 VN /\ '

z ..on - ,u-s v o M4ki

-is e B s

k A y  ;

, ==

c m, ...

su &

/N e- ~

' j l si j iM vs ,

{

v N

{

i RD n i  % D i t w CONDENSATE /FEEDWATER SYSTEM 43 N g

__ _ _ _ _ __- - - _ ~ . . . . .

~ .

l tap) f. 0 f

~ \

=z .

~/ . -

= '

. +

. . ,. /S , ,. ,

EI ' -

I...,'*

p' g

.~ .

. .- l n ,i, a - ,

e. ,s ,. o . w .

u :\

~" .-

~

57 I ' .

63

.))

y *, 1

.; 9 4 fadsve 2 -

1 *

~ . ,

.nesa d a e,

- z + ,,

q .,

U a

l PUMP (-A .

I a .A w

7, r,aus<, si,y ~6 . .

s .

hm[

/*

ns,p .. \

~

,,k / .

FF '

. V ,

p - '

i .. - . .- '

i 7: - -

/*7 .

m in - !

- , o a OAJ Slp'1 .

s 4.hg- j ! .

% ei Pp Sch'n- l} ,. .

'Q R l; r

, Qni

~

. / .

s E,edaara 5

, ece, n/ir/rs I m .

m - 3 Z Y ' g.O

, Woj3le : .=

A////

' ,, , - x o.us%u l'

) f ,

$ 00ft b itt N /TJo/t r _

, s

,,/, i.

y./

L i- ,,

,' I -

6

/,

Malsnal:M ,- on.r ,--

s ir SR 23 4 N PB .

&a-e as sca.r JAsol Crede k, ,, l

(

sea ,iw > .

i i

, IS 'x asou".aall

$ k!*1,Y N/erf &e hlf*/

= IE + flheto AleG C h B l

0 0

N .

1 I why katzw s

19 4

r

{

e Ddail> Elboa fl> 6,r el. 2 l

P

$/tah $3l0llC h&JCl Ort d' tl/ r>1/*ta/*

s rt fe FM a b's er i

I i

,a *

• b 1'

l l

1-I

i

• Ene/ssors Y etR, '/2/#/tt..

- ^~

FaA ikalte .

N< ld .

s f/ow -

\

, i .." ~ ~-1 it , }

Inilia Tjets Corner c- ,,

i

n. ./- _

+

I5" fa betcaled 2 Y x >a Bnoneh co < n e c ha n, ceIai/ / is.rth c o tra e e, lon,q:/sJwal pIane , Surey-2 zuis 3 >

/

. . l

/116.s% ,

l

/'

/

/

'( .

F7ow

\ .

)1omfoc luted 2h'It Trs ,- referrenh/we loside corr /ove, lo nyifdtmlp lans

. _ _ _ .