ML20100B683

Intervenor Exhibit I-MOSBA-227,consisting of Correspondence Re Pneumatic Control Component Testing

ML20100B683
Person / Time
Site:	Vogtle
Issue date:	10/06/1995
From:	AFFILIATION NOT ASSIGNED
To:
References
OLA-3-I-MOS-227, NUDOCS 9601190227
Download: ML20100B683 (9)

Text

r ['/NObN5 ~ 2 Z' 7 Reciprocating Products Division 1351 Harbor Bay Parkway, Suite 1 Alameda, CA 94502-6541 00CKETED 95 OCT 20 P5:15 0FricE Coopdf EfriFrgy ' '

Services 00CKElif ,

CORRESPONDENCE RELATING TO PNEUMATIC CONTROL COMPONENT TESTING Co i..- .<a n 3,

r :: J . .. .

~ l y.

Of f.W a,,:. . ;t.% &,

f<

a ,,

;y ,r.g y : : 7. y 5 ji:@ E y ;.pl < fg &

~ '

NUCLEAR REGULATORY COMMISSION Docket No. 50-424/425 OLA-3 EXHIBIT NO. T ' DM

' in the enatter of peorain Pwor Co. et al., Vootle Units 1 & 2 D K O 24 G PDR O Staff O Applic t G(ircervonor O Other O ldentifie4 elved O Rejected Reporter CP De d * / 6 /9 r witns,.s

~

- i i

June 5,1990 Georgia Power Company i

40 Inverness Center Parkway i P.O. Box 1295 t Birmingham, AL 35201

Attention: Mr. Kenneth Burr '

i

Subject:

Vogtle Electric Generating Plant Enterprise Diesel Generators S/N 76021/22 Pneumatic Control Component Testing .

1

Reference:

Georgia Power Purchase Order No. 6003110-Cooper Job No. S88859  ;

Gentlemen:  ;

1 This letter documents our findings of pneumatic control component testing .

conducted at the San Leandro, CA facility of Cooper Industries, May 30-  :

1 31,1990. This testing was commissioned by Georgia Power Company as part ,

of the on-going investigation into the cause of events at Vogtle,3/20/90, and the control system related issues which have been raised since that time. '

The components addressed by this report consist of (4) California Controls' Pressure Sensors, Enterprise Part Number F 573-156, and (1) Enterpnse  ;

Pneumatic Logic Board, Enterprise Part Number 1A 7055. .

i Testing was conducted under the guidelines of Enterprise Procedure No.

i S88859 700 REV.1, dated 5/11/90 (copy attached). The following details our  :

findings on an item by-item basis

. Pressure Switch P/N F-573-156 Georria Power VEGP I.D. No. IPSL4903 +

This pressure switch was installed in the engine control panel and is ,

designated P3 on the control schematics. Its purpose is to detect a decrease

• in air pressure to engine mounted shutdown sensors and affect a fast venting i of the logic to overcome line volume and pressure drops. Georp'a Power  !

records indicate that this device was origm, ally calibrated on 3/16/90. The  !

4 device was installed in the control panel of engine S/N 76022, VEGP Unit IB.

The pressure switch became suspect of malfunctioning on 3/25/90 during the '

course of troubleshooting the controls to determine the cause of

. unannunciated engine shutdowns. The device was replaced on 3/26/90, ,

though this had no affect on the control s The pressure sensor was tested-by Georgia Power's I&ystem problems.C personnel three time  ;

i satisfactory results before bemg placed in storage on 3/26/90. The purpose of F

6 i

i

j Georgia Power Company

. June 5,1990 ,

Page 2 j I

4 j testing this device in San Leandro was for confirmation of the previous findings.  ;

L

< The pressure sensor is a California Controls' model B4400B, a two way, i normally open, pilot actuated valve. The nameplate indicates a September

• 1988 manufacturing date. Two regulated air supplies fed the device. One, 1 connected to the sensor port, was varied to cause the pressure switch to trip J

l and reset, the pressure was indicated on a Danton 0 to 160 psig gauge, 4 Enterprise M&TE No. C 7611, calibrated 5/10/90, due 8/10/90. The second i regulated supply was maintained at 60 psig and went through a .028" parallel orince/ check valve oriented so as to restrict flow to the supply port of the pressure sensor, air )ressure was indicated on kn Aschcroft 0 to 160 psig gauge, Enterprise M&T3 No. B 2902, calibrated 5/08/90, due 8/08/90. The  !

sensor port pressure was adjusted between 0 and 60 psig for each valve cycle; .

the results were.

Test No. Trio Point Reset Point 1 44.5 psif(falling) 49.5 psir (rising) ,

2 45 psif 49.5 psir

3 44.5 psif 49.5 psir 4 44.5 psif 49.5 psir This data is consistent with calibration records from 3/26/90 and indicates '

satisfactory performance of the pressure sensor. The specified pressures for this application are 45 psif to trip and a reset dead bank of 1 to 8 psi.

i An interesting observation made during this test came about when the .028" orifice / check was inadvertently installed backwards causing an unrestricted 4

60 psi air source to be fed to the supply port. This resulted in the same trip point but it increased the pressure switch's reset dead band from 5 psi to

' 26 psi. The relevance of this observation is to point out the importance of using the .028" orifice to the supply port when performing calibration (the

.028" oriSce duplicates the restricted atr supply coming from the logic board).

i Pressure Switch P/N F-573-156 Georria Power VEGP I.D. No.1PS4749C 3 i

This pressure switch was installed on engine S/N 76021, VEGP Unit 1A, for the purpose oflow lube oil pressure protection. It is designated as Trip Low

, Press. Lube Oil No. 3, Line E-100 on the control schematics. It is one of  :

three pressure switches monitoring lube oil pressure; the control system logic '

requires that two of the three must vent to affect an engine shutdown. These sensors are active in both the normal and emergency engine operational modes. - t l

Georgia Power Company'

• June 5,1990 Pace 3  ;

i Georgia Power records indicate that this device was originally calibrated on 10/21/88. It was removed under MWO 19001433, on 3/30/90, for testing following reports of a lube oil pressure sensor malfunction alarm during operation on 3/20/90. Georgia Power's I&C personnel found this switch to be satisfactory, but it was placed in storage and replaced by a new sensor from "

stores.

'i This pressure sensor is a California Controls' model B4400. It is the same as the P3 valve described earlier except that it does not incorporate the "B" modiEcation (to be described later in this report). The nameplate did not indicate a manufacturing date, however, the valve was still covered by factory engine paint which dates it as pre-August '81; engine S/N 76021 shipped from Oakland 8/31/81. A 1/4" Swagelok 900 ell was still attached to the sensor port. Upon removal we noted that this was a special orificed fitting of the type used on the fuel return side of the Bendix fuel injection  ;

pump. This fitting does not affect the Endings of this report, nor is it ,

suspected to have relevance to the events of 3/20/90, but it is a misapplication '

of a special fitting which maintenance and I&C personnel should be made aware of. .

This pressure sensor was tested by the same method as described for the P3 valve though a different pressure gauge was used to indicate sensor port pressure. A Delaval 20 to 0 to 100 In. Hg. Test Gauge, Enterprise M&TE No. B 4246, calibrated 5/08/90, due 8/08/90 was installed on the sensor port. .

The gauge substitution was made to improve resolution, the 0 to 100 7 si gauges specified by the procedure were not available at the time of t:21s '

testing. The sensor port pressure was adjusted between 0 and 100 In. Hg. for each valve cycle. The results were:

Test No. Trio Point Reset Point I 1 61.5 In. Hg. (30.1 psif) 71 In. Hg.

j (34.8 psir)

2 62 In. Hg. (30.4 psif) 70.75 In. Hg.

(34.7 psir) 3 62 In. Hg. (30.4 psif) 70.75 In. Hg.

(34.7 psir)

. 4 62.25 In. Hg. (30.5 psif) 70.75 In. Hg. l (34.7 psir) '

't i i

Georgin Power Comp ny June 5,19S0 Page 4 l

This data is consistent with Vogtle's calibration records from 3/30/90 and indicates satisfactory performance of the pressure sensor. The specified pressures for this application are 30 psif to trip and a reset dead band of 1 to 8 psi.

Pressure Switch P/N F-573-156 Georria Power VEGP I.D. No.1PS4749E This pressure switch was installed on engine S/N 76021 as the No. 2 Low Press. Lube Oil Trip on pneumatic line E-10B. It too was removed on 3/30/90 for the same reasons as described for the No. 3 device and when tested, was

found to be satisfactory. This pressure switch was also covered by factory engine paint, with no manufacturing date. It is assumed to be of the same

vintage as the "C" device. The results were:

Test No. Trio Point Reset Point 1 60.5 In. Hg. (29.7 psif) 74.5 In. Hg. (36.5 4

psir) ,

l 2 62 In. Hg. (39.4 psif) 74.5 In. Hg. (36.5 i psir) 3 62 In. Hg. (30.4 psif) 75 In. Hg. (36.8 :

psir) l 4 62 In.Hg. (30.4 psif) 75 In. Hg. (36.8 i psir)  !

This data again indicates satisfactory performance of the pressure sensor.

Pressure Switch P/N F-573-156 Georria Power VEGP I.D. No. IPS4749A This pressure switch was installed on engine S/N 76021 as the No.1 Low Press. Lube Oil Trip on pneumatic line E-10A. This switch was removed at the same time as the other two, however, when this one was checked by Vogtle I&C, they found that it would not reset. A failure of the device to reset would explain the sensor malfunction alarm reported on 3/20/90.

4

7 Georgia Power Company June 5.1990  ;

Page 5 l

This switch was again covered by factory engine paint though this one was l

, stamped with assembly. code No. 8103 to sipiify manufacture in March of i 1981. Georgia Power's records indicate initia. VEGP calibration on 10/21/88.

• We tested this device in San Leandro by the same methods and apparatus as  !

the others. Our findings confirmed VEGP's in that the device remained at a .;

full vent and would not reset with a sensor pressure of up to 100 In. Hg.  !

(49 psig), representing approximately normal lubricating oil pressure. We i

. proceeded to disassemble the pressure switch to determine the cause of  ;

malfunction. When the pressure head was removed we found that the  !

diaphragm was stretched and fully set solid against the head. The affect of this is a substantial reduction in the area in which sensor pressure can act to ,

counter the spring forces transmitted through the pressure plate. The ,

reduction in area requires a proportionalincrease in pressure to achieve the j same force necessary to reset the pressure switch.

This condition was identified as a potential problem by our 10CFR21 report t No.145, addendum 1, dated May 12,1988 (copy attached). The cause of this  !

problem was found to be the result of a manufacturing tolerance stack-up  ;

which allowed the pressure plate to push the diaphragm up squarely against i the head. Over time, the diaphragm would stretch to conform to the head t and reduce the sensor port area. To remedy the problem Calcon has developed the B4400B model which incorporates a counterbore in the head to ensure clearance between the diaphragm and head: existing model B4400 '

pressure sensors can be reworked to the B4400B configuration.  ;

The pressure switch was completely disassembled for examination to ven'fy that the diaphragm problem was the only cause of malfunction. One small piece of metallic debris and some dirt was found within the spring chamber i where it would not impact the switch's function. Evidence of moisture tracks and a rusty spring were also observed but again they are not considered as relevant to the reset problem. The remainder of components were found to be l in good condition.

To further verify that the diaphragm was the sole cause of malfunction we switched the diaphragm and pressure head assembly with the one installed on the P3 valve which incorporates the B4400B modtfication. The valve was tested without disturbing the adjusting screw's original position and the )

following results were obtained: (

Test No. Trio Point Reset Point 1 29 psif 33 psir l 2 29 psif 32.5 pair l 3 29 psif 32.5 psir l 4 29 psif 32.5 psir l The pressure switch was then reassembled with its original diaphragm and  :

, pressure head without disturbing the set of the diaphragm against the head. 1

't The Danton 0 to 160 psig pressure gauge, No. C-7611, was connected to-the  ;

sensor port and the pressure was increased to determine where the switch  ;

i 1

l 1

l l'

7_

Georgis Power Company June 5.1990 Page 6 -

T i

i would reset. The diaphragm broke free and the switch reset at 118 psir.

With the diaphragm free the device was tested for its trip point with the 4 following results:

V Test No. Trio Point Reset Point l

4 1 28.5 psif 33.5 psir I j 2 28.5 psif 33 psir j 3 28.5 psif 33 psir  ;

4 4 28.5 psif 33 psir  !

j These results clearly indicate that the diaphragm's set against the pressure  !

! head was the sole cause of sensor malfunct2on.

l t

i

! i i

i Shutdown Loric Board P/N 1 A-7054 i A shutdown logic board malfunction occurred on 3/26/90 during functional f

testing of the en gine control panel for 76022, VEGP Unit IB. In a simulated operational condition, one of the engine shutdown sensors was very slowly  ;

4 vented by loosening the tubing connection to that sensor at the control panel  !

bulkhead. If the venting was performed slowly enough the logic board would i

! " lock up" and fail to produce output to the shutdown cylinder. It should be  :

noted that actual venting of the on engine shutdown devices would consistently produce the proper shutdown response. Due to schedule  :

1 constraints at the time of functional testing, the logic board was replaced '

l with an assembly from stores which remedied the problem.  ;

The logic board was connected to the factory simulator and tested in '

i accordance with the procedure on drawing 61560-7055 REV.D. All of the f

test steps produced satisfactory results. The test fixture was then modified i in an attempt to simulate site conditions by installing a check valve, Enterprise P/N F 573127, and a Whitey two way shutoff valve at port 11 of l the logic board, the group II shutdown circuit. The check valve was the same i' 1

^ as that used to isolate engine mounted shutdowns at Vogtle; its orientation for this testing allowed flow from port 11 of the logic board to the Whitey

, valve and subsequently to vent to atmosphere. The check valve was used to 4

simulate the pressure drop in the control panel and the Whitey valve  :

provided a means to slowly vent the logic circuitry. Group I pressure was i j monitored by a Danton 0 to 160 psi gauge, Enterpnse No. C-7611, calibrated  ;

5/10/90, due 8/10/90. Group II pressure was monitored on an Aschcroft t J

0 to 160 psi gauge, Enterprise No. B-2717, calibrated 5/08/90, due 8/08/90. .

^

The following tests were conducted (a copy of the logic board schematic and  !

functional description is attached for reference): g

( .

[

i 4

4 m = - . - . _ --

Georgin Power Company June 5,1990 .

Page 7 l

A) Charge the logic board at 60 psig control airi pressu 4

(

running state and slowly bleed Group Il p 7 i

Group II at 8.5 psif.

l i B) pressure Fully drops: open the Whitey valve on port 11 to me l

1/4 psig. ,

l l 1

C) Increase control air pressure to 65 p >

Group II at 9.5 psif. i 1

D) Open the Whitey valve fully with control air at 65  !

l measure system pressure drops: l i Group 11 was 1.5 psig.

l E) Return control air pressure to 60 psig and replacl l l

4 with a new element; repeat Step A: occurred with 1 l i l

F) The logic board was removed 16, and 17, 18, all ol This consisted of looking at element Nos. 5, 7,13,14,d the boarl 20,22,23 and 24. No aanormalities were observed an i

, was reassembled.

l G) The board was retested after assembly in accorda 61 560 7055 i I 2

procedure on drawingSteps A&B were repeated.to determine if  ;

4 satisfactory results. reassembly affected the pressure d

initial test. l i

j i

CONCLUSIONS AND RECOMMENDATIONS <

The logic board could not be causeditto f l f l

which caused malfunction at Vogtle were not truly representativ l i

engine operation and a problem may nevl I

materials and returned to stock.

l The low pressure lube oil trip malfunctioned due to' manufa b l which alowed the diaphragm to seat against the pressure hel reducing forces generated on the sensing side of the dia i

( has or replacement.

been addressed by 10CFR21 notiScation

\

i

, , , _ _ . _ . _ _ _ _ . _- _.. - . - _ _ . . _ , ____.- __-~ _ _......____ ___ .__. . . _ _ _-... . _ . ._.

Georgia Powzr Company l

. Juns 5,1990 Page 8 E l

switches at Vogtle be reworked to ensure against the reoccurrence of this problem. l Report By: ,

i

l i

i l Robert Johnston  !

Project Engineer RJ:djl j i

l l t

F t

I i

2 i

d I

J J

i j

1 o

I e

e-m