ML20101U792

Forwards Evaluation of Util Approach to Predicting Torque Loads for Facility Valves by Establishing Inlet Pressure to Valve at Various Positions,Per 811202 Audit Meeting.Valves Capable of Closing If Stated Provisions Accomplished

ML20101U792
Person / Time
Site:	Fermi
Issue date:	01/20/1982
From:	Restivo T BROOKHAVEN NATIONAL LABORATORY
To:	Haughey M Office of Nuclear Reactor Regulation
References
NUDOCS 8502080046
Download: ML20101U792 (8)

Text

56 ' byf ^ Poe-

SDMI cn T4' W n r, C'BL]u BROOKHAVEN NATIONAL LABORATORY

-(G

- aUd f E 3 ASSOCIATED UNIVERSITIES, INC.

Upton. Long Island, New York 11973 (516)282x 2863

' Department of Nuclear Energy FTS 666' t.

Jari.uary 20, 1982 Ms. Mary Haughey Equipment Qualification Branch.

• U.S. Nucler Regulatory Commissi']n Wasington, D.C. 20555 D'ea r _ Ma ry :

f]

v Brookhaven National Laboratory (BNL) has reviewed the Fermi 2 purge and vent valve operability demonstration information submitted in letters and presented 'at the December 2,1981 audit meeting.

Based on the information available, BNL's Engineering Analysis and Human Factors Group concludes that Detroit Edison can demonstrate that the subject isolation valves are capable of closure under the accident conditions post-ulated (by the licensee) provided the following is accomplished:

1. All valves are to be oriented so that the flat sides of the discs face the LOCA flow. -

2.- Applicable valves are to be oriented so that their shafts are "in-plane" with the plane of their associated upstream elbow.

n LU

3. A maximum allowable stroke time must be established for each valve for the " inservice test" conditions, i.e., no-load stroke timing.

The maxirrum allowable times established must be low enough to assure that the loads used in the stress analysis presented are not ex'-

ceeded.

'In response to your request concerning time comparisons, Attachment 1 is" presented.

Sincerely yours, ohk DR DOC -

PDR Thomas J. Restivo Engineering Analysis and $

Human Factors Group TJR:sd attachment cc.: R. Bari wo/att.

R. Ha11 W. Kato wo/att.

W. Luckas B. Miller e_

e c 3' , .

-r

1.0 REFERENCES

A. Detroit Edison (DE) letter #EF2-55,538 dated November 18, 1981.

B. . Meeting " Purge Valve Audit" at Detroit Edison's Fermi 2 site on December 2, 1981. ,,

C. Wyle Laboratories Report, 55210 - 18.0" Jamesbury Valve Test '(shown to staff at audit meeting).

D. Allis Chalmers test report VER-0209 dated December 17, 1979.

E. .Enrico Fermi Atomic Power Plant Unit 2, Final Safety Analysis Re-port, Figure 6.2-11 " Recirculation Line Break - Primary-Containment .

Ini,tial Pressure Transient" (handout at audit meeting).

F. Jamesbury Corporation letter dated November 12, 1981, B.C. Zannini

- -Janesbury to J. Green - Detroit Edison (handout at audit meeting).

" Aerodynamic Model Tests on Butterfly Valves," by Dr. Ing. C. Keller

~

G..

and Dr. Ing. F. Salzmann, published in ESCHER - Wyss News, Volume IX, No.1, January March 1936.

H. " Tests on Streamlined Butterfly Valves," by H. Netsch and F. Schulz published in The Engineers Digest, Volume II No. 8, August 1950 (From Maschinenbau und Warmewirtschaft, Volume 4, No. 9, September 1949). .

I. " Supportive Data Relating to Torque Coefficient Selection for

.Jamesbury Wafer-Sphere Valve with 90* Elbow Directly Upstream of Valve" (handout at audit meeting).

J. Allis Chalmers letter, April 30, 1981. R.H. Zeiders-(AC) to M..

Haughey' (NRC):

Subject:

Butterfly valves for ' containment isolation,

/~'t

'n 1-Allis Chalmers Valve Division Tests.

Detroit Edison letter EF?-55,980 dated January 4,1981' (with attach-

~

K.-

ments: a) " Combined' Loading Stress Analysis on Shaft for Purge Valves" b) " Seismic Qualification of 6" Purge Valve; based on Report JHA-76-34'(PI-2406)." .

e 6,.

e s .

{

4- . . . .

2.0 EVALUATION Detroit Edison's/Jamesbury's approach to predicting torque loads for Fermi 2's valves was to establish the inlet pressure to the valve at the var-ious valve positions from 90' (full open) to O' (full closed). Delay time (LOCA initiate to start of stroke), closure time under load (based on direc- -

' tion of net torque) and containment response (worse-case) are accounted for in their approach.

Based on the information available, BNL's evaluation is as follows:

The drywell pressure response curve in Reference E was given as the de-

. sign condition for which operability demonstration of all purge and vent iso- ~

lation valves (drywell and wetwell) is based. During the audit (Reference E)

E -

the ifcensee confirmed that this was the worse-case accident (highest rate of

-pressure increase) condition. The wetwell response also shown in Reference E is shown to be far less severe relative to valve loads.

~

ok- . Dynamic torque coefficients (C T ) used for the stress analysis of the Janesbury valves in Fermi 2 stem from an 18.0" Jamesbury valve test program conducted by Wyle Laboratories and reported in Reference C. Pressure and valve closure history data _ from this test was used in con,iuriction with data

-from an actuator - solenoid valve assembly vent test (same type used on test valve at Wyle) to establish Cy values at the various disc angles.

Based on Jamesbury's briefing at the audit meeting and information con-

- tained in submittal letters, the following can be summarized:

- -The test setup used at Wyle was. basically a straight pipe approach flow configuratrion.

- The Wyle test valve was installed and tested only with the discs flat side facing the flow.

. , ~

\j - The inlet test-pressure during the ifyle tests'was higher at all disc angles than the drywell pressure profile given in Reference E.

- The 18.0"' test valve at Wyle was of the same design (Wafer Sphere) as  !

the Fermi 2 valves. The aspect ratio of the 18.0", 20.0", and 24.0" ,

were said to be the same.

- _The operability case assumed in the load analysis is that one of the two in-series valves is failed in the full open (90') position.,

Jamesbury assumes.a.150% increase in the dynamic torque coef ficients developed from Wyle Tests to account for upstream piping elbow af- --

, fects.-

The test setup used at Wyle does not provide data to establish C T valves based on non-uniform approach flow conditions. Jamesbury has attempted to justify their 150% assumption by comparing their valves to the Allis Chalmers valves which were tested with elbow configurations as reported in Re-

'ference D. Jamesbury's comparisons shown in Reference I were on the basis of equivalent aspect ratios (t/D) where the Jamesbury CT values are shown to be significantly higher than the Allis Chalmers CT values. "

Jamesbury's comparison ofT C ' for their design to Allis Chalmer's CT 's on the basis of equivalent aspect ratios alone is inappropriate, particularly based on their approach to determining loads (determining con-tainment pressure vs. disc angle). The comparison does not reflect disc shape differences between the two designs. Information contained in Reference G.

Figure 3, indicate that discs having approximately the same aspect ratios but different shapes can have very different coefficients at given disc angles.

Some differences were significantly more than were indicated in the Jamesbury

- Allis Chalmers comparisons (Reference I).

For those valve installations with an " elbow upstream - shaft oriented out-of-plane configuration," disc closure rotation direction must also be con- g sidered when predicting dynamic torques. Allis Chalmer's test program indi-cated that closure rotation direction would significantly influence torque de-veloped. This aspect was not reported in Reference D, but was made known in Reference J. Jamesbury's analysis has not considered this aspect for the Fermi 2 valves.

Affect on dynamic torque due to flow pattern o.ff of an associated in-series upstream valves had not been addressed by D.E. in their evaluation.

Separation distances between each in-series valve pairs was observed (during walk-around inspection) to be small. These distances are nuch less than the 15D used in the Instrument Society of America's (ISA) Standard S39.4 for a similar configuration.

BNL believes that in the case of the Fermi 2 valves, elbows (where ap '

plicable) are in such close proximity to both of the in-series valves, the flow pattern from the elbow should provide the predominant affect on the ap-proach flow pattern as it affects dynamic torque. DE's 150% increase in h straight pipe established CT's to account for the upstream elbow effect is considered conservative for the valve to valve effect.

NOTE: Test results reported in Reference G (Figure 12) indicate that the af fect to increase dynamic torque developed in a downstream butterfly valve due to flow distribution off of a modulating upstream butterfly (close proximity) is minor at large angles of upstream valve opening (Figure 12 uses 0* as full open). At smaller upstream valve opening angles, the affect is to significantly reduce the dynamic torque developed in the downstream valve.

Detroit Edison has indicated that they plan to re-orient (where neces- ~

sary) the subject . isolation valves as follows:

s

a. Flat face of disc to face LOCA flow (This was presented at audit meeting, Reference B).

m

[ .

b.- Applicable valves-are to be re-oriented so that their shaf ts are "in-plane" with the plane of their associated upstream elbow (Re-ference I).

NOTE: Based on information available from submittals and from discus-sions with some butterfly valve manufacturers, the industry's general belief is that the "in-plane shaft-upstream elbow" configuration is equivalent to the ..

" straight upstream piping" configuration with respect to dynamic torques. No documentation (test or analytical) has-been made available by manufacturers or licensee's nor does BNL know of any documentation to support the industry's belief. At this time, BNL can only offer its opinion that the maqnitudes of the dynamic torques are likely to be equivalent for the aforementioned on-figurations.

Based on implementation of these plans, BNL finds that Detroit Edison can demonstrate that the subject valves are capable of closure against the accident condition (LOCA and seismic) postulated.

m NOTE: A maximum allowable stroke time must also be established for each Q valve for the " Inservice Test" conditions i.e., no-load stroke timing. The maximum allowable times must be _ low enough to assure that the loads used in the stress analysis are not exceeded.

Table 'l _ summarizes the torque and pressure loads used in the stress analysis as~ compared to the predicted torque loads at 0* and 90" and maximum ,

containment pressure potential (drywell and wetwell). Predicted torques at 90* are based on 150% of straight pipe determined Cy's and torques at 0* are from seating torque curves presented in Reference A.

. As can be'seen from Table 1, the torque values used in the stress an-alysis are significantly higher than the predicted torques at 0* and 90 for the 6.0",10.0", and 20.0" val ves.

The stress analysis' torque for the 24.0" valves ~is. shown to be lower than the predicted torque at 90*. The piping installations of_the 24.0"

( )

C tvalves are considered straight pipe (i.e. uniform approach flow) configured, and as such, the Jamesbury determined straight pipe C T 's would apply. The 3440 ft. Ibs. which is the straight pipe predicted torque times 1507. should be considered conservative for the 24.0" valves.

The pressure loads'used for the stress analysis are also shown to be

, conservative. Table I shows that'the pressure loading used in the stress analysis is higher for all valve sizes than the maximum containment pressures predicted. In fact, the pressures used for the 6.0",10.0" and 20.0" valve are more. than twice the maximum pressures predicted.

The licensee has shown that the stress values predicted are within the design allowables specified. Seismic loads have been included in the anal- -

ysis.

~

r

In Reference A, Detroit Edison presented the torque output curves for the various air-operated actuators used on the Fermi 2 valves. Based on these curves, it can be concluded that the operators can provide the torques neces-sary to close and seat the valves while operating from their 90" (full open) initial position. These same curves indicate that the operator torque ratings are not exceeded.

Detroit Edison also (Reference A) provided information concerning sizing '~

of the motor operator for the 24.0" valve. The information demonstrated that the operators are sized to stroke the valves under the leads postulated at 80%

reduced voltage.

Table 1 STRESS TOROUE PREDICTED STRESS MAXIMUM ACCIDENT ANALY. (LOCA) ANALY. PRESSURE SIZE LOCATION T0 ROUE O 90" PRESS. DRYWELL WETWELL VALVE ID in. DW or WW FT-LBS (FT-LBS (FT-LBS) (PSIG) (PSIG) (PSIG)

V4-2061 V4-2063 6.0 6.0 WW WW 315 315 160 160 80

.30 150 150 25 25 g

V4-2060 10.0 DW /30 690 345 150 58 --

VR3-3014 20.0 WW 2300 1700 2060 62 --

25 VR3-3016 20.0 WW 2300 1700 2060 62 --

25 VR3-3013 20.0 WW 2300 1700 2060 62 --

25 VR3-3015 20.0 WW 2300 1700 2060 62 --

25 VR3-3012 24.0 DW 3200 3200 3440 62 58 --

VR3-3023 24.0 DW 3200 3200 3440 62 58 --

VR3-3011 24.0 DW 3200 3200 3440 62 58 --

VR3-3024 24.0 DW 3200 3200 3440 62 58 --

VR3-3019 6.0 WW 260 160 80 62 --

25 VR3-3026 6.0 DW 260 160 80 62 58 --

O 4

h w.

- [ .. .

• The_ maximum potential drywell accident pressure is approximately 58 psig and the wetwell's approximately 25 psig. Both are less than the 68.5 psig AP

. capability based on TD . Valve closure time from a valve assembly structural

, integrity. standpoint does not appear to be critical for the 6.0" valves.

For 10.0" Valves a.UgecurveD{ elbow)whereCT = 0.0213

b. 00 = 1000 in

c. TD = 730 ft-lbs.

= 34.3 psi AP = 0.02 x 1000 The 10.0" valve is attached to the drywell. The drywell reaches 34.3 psic in approximately 2.0 sec from LOCA initiate based on the drywell response in Figure 1 (Reference A). This time is approximately three times the 0.65 o sec shown by the licensee in Figure 6 (Reference A).

In that the AP capability is shown to be less than the maximun drywell potential, it would appear that a limitation is required on the valve closure time allowable during the inservice tests.

For 20.0" Valves

a. UgecurveC{ elbow)whereCT = 0.0125

b. D3 = 8000 in

c. TD = 2300 ft-lbs.

= 23.0 psig

_.AP = 0.0125 x 8000 l)

V The 20.0" valves are connected to the wetwell. .The maximum potenti '

wetwell pressure is approximately 25 psig-(Reference E) which is slightly higher than the AP capability of 23 psig which is based on TD = 2300 ft-lbs.

Reference E shows that the wetwell-would require over 20 seconds to reach 23 psig. This compared to the 1.3 second time shown in Figure 7 (Reference A).

Valve closure. tine from a valve assembly structural standpoint does not appear to. be critical for the 20.0" valves'.

For 24.0" Valves-

a. ~Use curve A'(3 straight pipe), where CT = 0.009

.b. -D3 =.13824 in

c. TO = 3200 ft-lbs. >

~

AP = 0.009 3824

= 25.7 psig 9

_~-

The 24.0" valves are connected to the drywell. Figure 1 (Reference A) shows that the drywell would require approximately one second to reach 25 psig. Figure 8 (CTQ = A) (Reference A) shows valve closure in 1.25 seconds, where the drywell pressure reaches approximately 27 psig. This is higher than

'the 25.7 psi AP capability (based on TD ). As previously mentioned, the CT value assumed was the maximum value on Curve A (Figure 4) 1.e., at 90" (full open). By using a CT value of 0.005 which is at 75* (less at lower angles) 1 "-

a AP capability (based on dT ) of 46.3 psi is calculated. Fiqure 1 (Reference A) shows that the drywell would require approximately 4 seconds to reach 46 psig. This can be compared to the 1.25 second closure time shown by the licensee for the 24.0" valves.

In that the AP capability is shown to be less than the maximum drywell potential, it would appear that a limitation is required on the valve closure time allowable during the inservice tests.

9 O

b r