IR 05000369/1986033

Insp Repts 50-369/86-33 & 50-370/86-33 on 861027-31.No Violations or Deviations Noted.Major Areas Inspected: Followup on Licensee Actions Required by IE Bulletin 85-003, Motor-Operated Valve Common Mode Failures..

ML20215C342
Person / Time
Site:	McGuire, Mcguire
Issue date:	11/26/1986
From:	Jape F, Schnebli G NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION II)
To:
Shared Package
ML20215C288	List: IR 05000369/1986033 ML20215C286
References
50-369-86-33, 50-370-86-33, IEB-85-003, IEB-85-3, NUDOCS 8612150108
Download: ML20215C342 (11)

Text

{{#Wiki_filter:y _ _

 . >  - - - - -     -
          ,

y-

        ,  _
' ' ' '

N, ,

      ,

      '

n'

 '
       '
'(
       -
          ,
          ,
-o  y #$  NUCLEAR R ULAT RY COMMISSION    -'
  [  o   ~ REGION li
 > >j'  101 MARIETTA STREET, ,
  ?  2  ATLANTA, GEORGI A 30323 y
  %, ._ . . . . f . ~
     -

_

          -
  .

h <

        ~
          ,
.
  -,

_ .

Report Nos.: 50-369/86-33 and 50-370/86-33-r Licensee:.l Duke-Power. Company * c-422 South Church Street - Char.1otte, NC. ~ 28242 -

;.c  Docket Nos.: 50-369 and 50-370   License Nos.: NPF-9 and NPF-17
 ' Facility. Name: McGuire 1 and 2
 ~ Inspection Conducted: October 27-31, 1986 - Management meeting at Duke Corporate
'

J0ffices conducted: ' November 5,1986. n: Telephone' conference calls ebetween. Duke Power Company and-NRC conducted: November.12!and 14,-198 " Inspector: mv - " 4 b // / 2d I G.-A chnebli g# Date Signe Approved'by: Ar& O cLA2r_ / /JUdd F. Jape, Section Chie .

      (f . * /  'Date Signed
          "
          . ._
 '
 .  ' Engineering Branch,     ,
  ,
   : Division of Reactor Safety     r
.
       '

, _ SUMMARY- .

      ~
<  Scope:- This routine, unannounced inspection was conducted in the' area of follow-- _    ,

up on licensee acti_ons' required by IE Bulletin 85-03, " Motor Operated Valve ' Common Mode Failures During Plant Transients Due to Improper Switch Settings."

, Results: No violations ~or deviations were: identifie s

.

y b^

      'r'.

t ( f 1 q

    "

8612150108 861126 PDR ADOCK 05000369 G PDR

       .
,

e "tp

     -

      -
- -
  -
       ,
 "

t Yn

' ~
.
 .
      .
      %

REPORT DETAILS 1. . Persons Contacted

 ' Licensee Employees
      "'-

T.. Cline, I&E Supervisor .

 #T. L.-Edwards, Design Engineer
 #N. Estep, NPD, Maintenance
 #D. H. Gabriel, NPD, Maintenance
 #G. Gilbert, NPD, Maintenance

_

 '#C. W. Graves, Nuclear Operations
 #D..G. Gwyn, Design Engineer
 #S. Hart, NPD, Maintenance
 #*N. McCraw, Ccroliance Engineer
 *T. L. McConne i Station Manager
 *D. J. Rains, _ erintendent of Maintenance
 .
 #R. 0. Sharpe, ticensing
 "*R. B. White, IAE Engineer-Other ' licensee employees contacted included engineers, technicians,' opera-tors, mechanics, and office personne Other Organizations
      ~

R. AEnold, Representative for Rotork Controls, In R. Kiessel, NRC, IE- , NRC Resident Inspectors

 #*W. Orders, Senior Resident Inspector
 #*S. Guenther, Resident Inspector
 #M. Lesser, Resident Inspector - Catawba
 * Attended exit interview on October 31, 198 # Attended Corporate meeting.on November 5, 198 . Exit Interview t

The inspection scope and findings were summarized on October 31, 1986, with those persons indicated in paragraph 1 above. The inspector described the areas inspected and discussed in detail the inspection findings. No d:s-senting comments were received from the licensee. The following new itam was identified during this inspectio (0 pen) Inspector Followup Item 369, 370/86-33-01, " Followup on Final Resolution to Rotork Torque Switch Issue" - paragraph 5.

. w * -- o,%- r- , ,_

_

 -

.

.

- 2 _ E" The licensee did not identify as proprietary any of the material provided to or reviewed by the inspector during this inspectio . Licensee Action on Previous Enforcement Matters " This subject was-not addressed in the inspection.

- " Unresolved Items , i i Unresolved items were not identified during the inspection.

! Followup on IE Bulletin 85-03(92703,25573) ) The major purpose of this inspection was to conduct a routine followup on the licensee's progress for the actions required by IEB 85-03. This bulle- [ tin was generated as a result of motor operated valve (M0V) comon mode failures during plant transients due to improper switch settings.

L When the inspector arrived onsite, October 27, 1986, he was informed by the F resident inspectors that a problem had been discovered with the torque = switch settings on all MOVs with actuators supplied by Rotork Controls.

-

'

The problem was initially identified at Catawba on October 24, 1986, when - DPC personnel were testing MOVs in response to IEB 85-03. The same problem

was verified to exist at both the Oconee and McGuire stations on October 25-26, 1986. However, the impact at McGuire was more severe due to the large number of Rotork actuators used in safety systems at McGuire in E

comparison to the other two site The problem identified during testing was that the actuator torque output _ for various torque switch settings did not correspond with expected values.

s Specifically, the torque switches installed in the Rotork actuators have a = variable torque switch setting with setpoints at positions of 1, 2, 3, 4, it and A setpoint of I corresponds to the minimum actuator output and a I setpoint of 5 corresponds to the maximum output. See Attachment for a graph of torque output versus switch setting. Duke Power Company (DPC) had i - previously assumed, due to information available at the time, that the E correlation between actuator torque output and torque switch setting was a s linear function. The licensee utilized a linear graph, which they stated " was provided by Rotork, showing that a torque switch of I corresponded to 40 percent of rated actuator output torque, and a setting of 5 corresponded [[ to 100 percent of rated actuator output. The actuator output torque values F for the remaining switch settings of 2, 3, and 4 could be detennined using g the linear graph.

F Actual bench testing by DPC showed this linear correlation was not tru .

       '

i Subsequently, DPC obtained additional data from Rotork that showed a wide " variation in actuator output for a specific torque switch setting. For " example, the data scatter for tests of over 40 separate model 14NA actuators - . revealed that the actual torque output for a switch setting of 1 varied

through a range of approximately 11 to 55 percent of maximum actuator torque a, output versus the previous accepted value of 40 percent. Additionally, a

= N { -

- - _ . _ _ _ _ . . _ _ . . .  .

~   . . - . . -  .
   -
 ,

e

,  ;' '
~
 .y

'

 ..  ,
-
: -
   ,

r y 3

.

switch setting of 3_showed_a variation;in. actual; torque output between 24

  ~

- n < _ _ f rand 76 percent versus-the previous _ accepted value.of 70 percent. The Q Emajority of the data points-were-on the_ low side of the-previously accepted linear graph',Lthat.is less than 40 percent for'a switch' setting of .1 and 11ess-than:70 percent for a switch setting.of 3. . Data supplied by the . vendor :

    -

indicated that a switch" setting of 5 provided 100 percent output torque with

,
 ' no data scatter..as this setting was: the maximum output of the actuato LThis'was confirmedLin a_ telephone conversation.with the vendor o '
       -

' October.29,11986;' however, subsequent bench testing proved that this was not - the'. case and will be discussed later in this repor The~ major concern with this problem is that without performing some type of

 . bench test ~or using specific output graphs ;for each actuator there .is no way to determine. actuator output. torque. The vendor stated (in the October 29,-
*

1986,' conference call) .that actuators supplied prior to 1978 were _ tested at the switch setting of 5 for maximum actuator output and then the torque switch.was ~ adjusted to the required setting that would produce the as-L, ordered output torque. Thus, the customer would receive an actuator for a

 ,
 ' specific application with the torque switch preset to some value between-1

, and 5. Any subsequent changes to the torque switch setting during ~ construction, preoperational testing, normal operation, or maintenance would

  'then require bench testing since the previously accepted linear relationship between torque switch setting and actuator output was not true. The vendor

. 'further stated that actuators supplied after 1978 were supplied with certi-fication papers .that_ provided actual actuator output torque, in percent, for ' the switch settings of 1, 3, and 5. Therefore, a graph could be drawn using these three data points and this graph could be used to deter-mine actuator output at the various setpoints between 1 and 5. Each graph:would be for a specific actuator due to the wide margin of data scatter at the different setpoints. The inspector initially felt that the

       '

graphs could solve.the problem for actuators supplied.after 1978, however, , , the vendor indicated that he would not rely on the data for settings at ' switch positions 1 and 3 due to the data scatter and method of test-L ing. He considered the switch setting at 5 to be the only accurate data poin The problem this poses for the industry is that once a torque switch setting L is changed (which a is common occurrence industry wide) a. bench test os reouired to ensure the actuator output is as require For example, using , [ the graph provided as Attachment I: The graph is for a 14NA actuator. Assume 100 percent actuator output torque of 100 ft-lbs at a switch setting of 5.

t

  * Assume, through an engineering evaluation, the licensee determines the actual output required for the application is only 70 ft-lbs. Using

' the linear graph, this corresponds to an actuator output of 70 percent and a torque switch setting of 3.

l l

-
-
 - - -

_

.     .
     ,
,

*

Using the ' data scatter provided by the vendor, the actual output could - be anywhere between 24 and.76 ft-lbs at a switch setting of 3 for this'

'particular actuato *

Thus', if the output is on the. low side (24 ft-lbs) the valve may not function when required;-or if on the high side (76 ft-lbs), the valve could become. damaged or stuck in the sea On October 28, 1986, a conference call was held between McGuire site manage-ment, the resident _and region-based inspectors, DPC corporate management,

   -

and NRC management .in. Region II. The purpose of the call was to ensure all parties concerned understood the issue with the torque switch settings and the inability to determine actual actuator output without bench testin DPC stated that approximately 328_. valves with the Rotork actuators were installed in safety systems at McGuire 1 and 2. They indicated that a random survey using the worst case graphs (generated by plotting the least conservative data supplied by the vendor for each model actuator, see Attachment I) showed that the valves _would perform their safety function during an accident. They proposed continued operations of both units based on their initial random survey while they performed the following: Evaluate all systems and prioritize the systems as required to function during an acciden Perform a design engineering evaluation on all valves identified to determine operability based on the least conservative graphs discussed abov * Valves with actuator torque switch settings of 5 would not be evaluated as this setpoint was equal to 100 percent of actuator outpu Valves which did not require routine cycling during normal operation , would be placed in their safety position and have power disconnected.

l Valves in this category would not require evaluation.

! i'

*

Valves which would not significantly affect plant response to an accident would not be evaluated.

l i

*

If any operability questions were identified during the evaluation the l problem would be immediately corrected or the particular valve in

question would be declared inoperable and the applicable Technical

! Specification (TS) Limiting Condition for Operation (LCO) would be l followe * Additionally, if a large number of valves required immediate corrective l actions to satisfy an operability question, the units would be shut ! down as this would indicate the initial random sample was in questio All groups involved in this conference call considered DPC proposals to be acceptable.

l l

. ,

g i s t

. - r
,'  -

 ,
  .

_ At~ 2100 on October 28,11986,iDPCdeclared.thecharging-headerisolation-valves NV244 and 245 on both. units inoperable.' This was due to the fact

      ~

that the indepth design evaluation had determined the actuators may not have

   -
   .
     :     .
         ,

s -adequate torque switch settings which could render the valves ~~ incapable of~

 - performing their safety function under accident conditions. . TS 3.03 was
 "

entered 1immediately and an orderly shutdown:of both units commenced. The

  - four valves in question'were closed and power was remove After the units were shut down, the torque switch settings were increased to'.
  . 5,-the maximum. When attempting to functionally test the. valves by
 . cycling, the first valve tested.became stuck in the seat after closing and
  '

burnt up the motor was damaged when attempting to open the valve.- This indicated that the design engineering evaluation, which allowed theiswitch

  . setting to be increased to the 5 position was also in question. Subse-quent. bench testing of. numerous: actuators confirmed: that the data scatter-previously noted at positions one and three was also exhibited at position-5.;

Althoup IEB 85-03 recognizes there is-an' industry wide problem.with M0V switch. settings, it-does not recognize the wide variance in-actuator output for'a_ specific torque switch setting. Therefore, IE Notice 86-93 was prepared to identify this problem. The notice was written with inputs from NRC, DPC, and the vendor and: issued on November 3, 198 The' inspector conducted an exit meeting with McGuire management personnel on

 : 0ctober 31, 1986. In this meeting, DPC was informed that this entire . issue
  .was an unresolved item pending further discussion with Region'II management,-

IE-and DPC managemen On November 5,1986, a meeting was conducted at DPC corporate offices in Charlotte, North Carolina. The purpose of the meeting was to discuss the current status of activity on this issue and was attended by DPC Design Engineering and management personnel, a Resident Inspector from Catawba, , both Resident Inspectors from McGuire, and this Regional-based inspecto DPC stated that the McGuire units would remain shut down until the following could be accomplished.

* A design evaluation would be conducted on all safety-related Rotork , actuators to determine what actions were necessary to assure operability r of the valves. These actions may include bench testing to set the

actuators for a specific torque value.

!- "

  -

Corrective actions would be made based on the Engineering evaluations ' determined above.

l . l

  *

The current torque switch settings on all safety-related Rotork 1 actuators would be field verified and documented. Any discrepancies l between the "as found" settings and the settings determined by the l- Design Engineering evaluation would be resolved and corrected.

. q P- *e5 -r *9e 3 9--l-mee--=*tw ww es eer-m*we-eT-+ -ww yr---s-* * *e: ---me=-- -ur- - w ---=-----=--'"-a -- - --- - *

.

6 M -

Additional items discussed at:the meeting included:

 * The current status o'f testing at McGuire showed that 27. Unit 2 valves

_j- had been bench tested. Readjusting the torque switch was required on 29 Unit 2 actuators and (at the time of the meeting) 24'of these had

 .been complete ,
 * DPC 1ndicated. all .their priorities were focused on Unit 2 as they wanted.to return that unit to service first. Unit 1 would be discussed gb  at_a later dat '

The licensee stated that the worst case curves, discussed previously, would continue to be used to-determine operability.. They.would be used to determine if the existing setting was sufficient to ensure the valve would operate in an accident condition and not damage the valve; or raise the switch setting to some higher i alue to ensure the sam The data base for the worst case curves would be increased as more data became available from the vendor and actual bench testing. They

  ~

indicated approximately 40 to 50 actuators would be bench teste Between November 10-14, 1986, several conference calls were held between DPC corporate personnel, site personnel, site Resident Inspectors, and Region II

'

personnel to discuss the current status and actions. remaining prior .to restarting the' units. During these conversations, DPC stated that actual data obtained during bench testing showed that some of the actuators would not achieve-100 percent output at a switch setting of 5. They indicated the actual output for these actuators was 25-30 percent low. However, even at this lower torque, analysis showed that the valve in question would still perform its safety function during accident condition Region II requested that DPC provide the NRC a list showing the various categories the actuators were assigned te for evaluation, the reason as-signed, and the total number for each category. The list was only required for McGuire ~ Unit 2 as this was the first McGuire unit to be placed in service. The same methodology was to ba used by DPC for Unit 1 prior to placing it in service. The following sumary was provided to Region II by DPC on November 14, 1986, and subsequently discussed in detail in a confer-ence call later that day: Valves Determined To Be " Acceptable As Is" Based On: Valves are already set a maximum torque. No change is required based on the fact that the actuator is set to provide maximum output which is a tested point (48 valves set at maximum). Valves for which an operability statement has been written by Design Engineering personnel. Valves in this category would not significantly affect plant response to an accident if they failed i to perfor (Twenty-six valves have operability statements. DPC i ~ .-.

 ,
*
.

 -further stated that of the 26 valves, 11 were bench tested prior to the operability statements being written). . Valves with a torque switch setting less than maximum.. Actual ,

required valve torque based on actual differential pressure was compared to design valve torque and then evaluated against the worst case curves which indicated no change.was required. (Twen-ty-one valves did not require readjustment). Some of the valves initially identified to be evaluated were later

  -

determined:to have no safety. function and were removed from the lis (Five valves had ro safety function). Nine valves were placed in their safety position and power was_- removed.

. Valves Wit.: Reset Torque Switche The actual required torque based on actual differential pressures encountered were compared to design torque and evaluated against the worst case curves. Evaluation indicated torque switch adjustment was required. DPC pointed out that in some cases _ the torque switch setting was decreased to further protect the valve from_ damage. Settings were decreased only when adequate margin was indicated on the worst case curve-to ensure the valve would perform its safety functio (Thirty valves required adjustment). Valves Requiring Tes These valves were judged to have limitations on maximum applied torque and required testing to a specific output value by the following methods: l-l A total of 52 valves were bench tested and set to a specific value on a Rotork test bench.

' Seven valves were tested in place by cycling under actual required differential pressure conditions. DPC stated that none of these valves required torque switch adjustment due to failure and that

. all operated as required.

f One valve was set in place using a strain gage tes The NRC expressed a concern over the fact that the licensee's evaluations were based on a switch setting of 5 corresponding to 100 percent actuator output, when bench testing proved this was not true in all cases. DPC l !

 - -.--,. - --- - -.

 *

8-stated they would reevaluate all actuators with switch settings between 3 and.5_as these would be_the only points affected. They felt confi-- dent that this would pose no problem due to.the large number of actuators that;were already bench tested at a switch setting of 5. The test data showed in all cases if the actuator output _was low, it was still sufficient - -for the valve to perform its safety function. A total of 85. actuators had been bench tested-(52 actuators required testing due to the engineering-evaluation, actuators were tested prior to operability statements being issued, five 11 actuators were tested prior to operability. statements being

     .

issues, and 5 new actuators obtained from the warehouse were tested, and-17' actuators from Catawba were tested). Of the 85 actuators. tested,-only 13 indicated a lower than maximum output at switch position 5 and 4 of these were less than 5 percent low. The remaining 9 varied from five to about 27 percent lo At the close of this conference call, DPC reassured the NRC this was an interim solution for this problem to allow continued operation. The final solution to this issue will be resolved at a future date. The resolution for valves-identified in response to IEB 85-03 will be verified correct as required by the bulletin. . The NRC considered DPC actions concerning this issue to be acceptable. This item was previously identified'at the exit meeting as an unresolved item,

 : however, due to the actions taken by DPC and the numerous conference calls to resolve the issue, it is now identified as Inspector' Followup Item 369,370/86-33-01, Followup on final resolution to Rotork torque switch issu No violations of deviations were identified in the areas inspecte Attachment:

' Torque Output and Setting _- , _ _ __ . _. _

-

ATTACH MEAIT I

..

10b'h-7- '/. k / / '

     /

8 075 .

    /  j    l Q    ,A  /
   '  /  /

h ,

       /

S 6($' / / M / /

$ V    l /

M / / k 1 / / n #k i B

    /
    /

b / ' f 2 03 , , J/

   -
  ,s'
 #   -

i'- - jii;ljg;;1> . 4 ,

        ; , , ,

0% .

   . ,
    .
     ,
     .
       ,
       .

r s I Z 3 '4 .5' 702QUE StAJ/TCH 3ETTlA/6 i SEE SHEET Z. FOR. EXPLAINATioN O F C_U R.VES A N D N OTE S . SKEET I of 2 _. - . - - - _ . - . - - - - . --. _.. . _ _ - - .. . . - - - _ .- . - . .

n-

,
*
~
. . .

ATTACHMENT

.I.. EXPLANATION'0F CURVES 1 . LINEAR CURVE - Generated from previously available informatio Ass'umed a torque switch setting of 1 corresponded to 40 percent of actuator output and.a setting of 5 corresponded to 100 percen WORST CASE CURVE' LOW OUTPUT - Generated from vendor supplied data and bench. test data. This curve shows the lowest data points for switch settings of'1 and 3. -This curve assumed setting 5 corresponded to 100
 . percen WORST CASE CURVE HIGH OUTPUT - Generated from vendor supplied data and bench test data. This curve shows the: highest data points for switch-settings of.I and 3. .This curve assumed setting 5 corresponded to 100 percen MODIFIED WORST CASE CURVE - Generated from data obtained during actual bench testing. Showed that a setting of 5 did not always correspond to 100 percen II. NOTES- These curves are provided for information only to show the magnitude of the problem. Curves were based on data available at the time of-the inspectio The curves were generated from the data provided for a model 14NA actuato Although switch settings 1, 3, and 5 only show 2 data points, these

. were the most extreme points. The data for over 40 actuators of this ' type were plotted to obtain these worst case points. The majority of ' the data points were on the low side (lease conservative side) of curve 1.

i Sheet 2 of 2

 . . . - . . - .- .
     . _ . . .-

}}