09/10/2015 U.S. Nuclear Reeulatorv Commission Operations Center Event Renort Paop I Part 21 (PAR) Event# 51309. Rep Org: ROTORK CONTROLS Notification Date / Time: 08/11/2015 09:55 (EDT) Supplier: ROTORK CONTROLS Event Date / Time: 07/28/2015 (EDT) Last Modification: 09/10/2015 Region: 1 Docket #: City: ROCHESTER Agreement State: Yes County: License #: State: NY NRC Notified by: PATRICK SHAW Notifications: RAY POWELL Ri1DO HQ Ops Officer: MARK ABRAMOVITZ ANTHONY MASTERS R2DO Emergency Class: NON EMERGENCY ROBERT ORLIKOWSKI R30O 10 CFR Section: BOB HAGAR R4DO 21.21(a)(2) INTERIM EVAL OF DEVIATION PART 21/50.55 REACTORS EMAIL PART-21 NOTIFICATION - MICRO SWITCH INTERMITTENT VARIATION IN RESISTANCE The following report was received via fax: 'On June 4, 2015, Rotork Controls Inc. opened a formal Part 21 [10 CFR 21.21] investigation into a self-identified anomaly relating to a basic micro switch - Pt No N69-921, description 'V12'. The anomaly is intermittent variation in electrical contact resistance and was first observed during the factory acceptance test of a Rotork safety related NA Range Electric Actuator; also referred to as an electric Valve Operator. "Rotork and the switch manufacturer are currently characterizing switch population contact resistance to establish whether an unsafe condition could exist as defined under 10 CFR 21. * ** UPDATE AT 0958 EDT ON 9/10/2015 FROM PATRICK SHAW TO MARK ABRAMOVITZ ** * The following report was received via e-mail: "Rotork has concluded the investigation and based on test data determine an unsafe condition can exist as defined under 10OCFR21. Testing has established .that a small percentage of the switches supplied against Rotork purchase order P0116932 may exhibit an open circuit condition when operated. A slightly larger percentage of the population may exhibit high contact resistance ranging from 0.5 Ohm to open circuit. The anomaly affects the normally open and normally closed contacts equally. Contact resistance anomalies are random, not permanent, and appear to be unaffected by accumulated cycles. Additionally when a switch is oPerated its electrical state remains constant until the next operation. "Probabilities have been established by resistance measurements and by dynamic testing using voltages
Transcript
09/10/2015 U.S. Nuclear Reeulatorv Commission Operations Center Event Renort Paop I
Part 21 (PAR) Event# 51309.
Rep Org: ROTORK CONTROLS Notification Date / Time: 08/11/2015 09:55 (EDT)Supplier: ROTORK CONTROLS Event Date / Time: 07/28/2015 (EDT)
NRC Notified by: PATRICK SHAW Notifications: RAY POWELL Ri1DO
HQ Ops Officer: MARK ABRAMOVITZ ANTHONY MASTERS R2DOEmergency Class: NON EMERGENCY ROBERT ORLIKOWSKI R30O
10 CFR Section: BOB HAGAR R4DO21.21(a)(2) INTERIM EVAL OF DEVIATION PART 21/50.55 REACTORS EMAIL
PART-21 NOTIFICATION - MICRO SWITCH INTERMITTENT VARIATION IN RESISTANCE
The following report was received via fax:
'On June 4, 2015, Rotork Controls Inc. opened a formal Part 21 [10 CFR 21.21] investigation into a self-identifiedanomaly relating to a basic micro switch - Pt No N69-921, description 'V12'. The anomaly is intermittent variationin electrical contact resistance and was first observed during the factory acceptance test of a Rotork safety relatedNA Range Electric Actuator; also referred to as an electric Valve Operator.
"Rotork and the switch manufacturer are currently characterizing switch population contact resistance to establishwhether an unsafe condition could exist as defined under 10 CFR 21.
* ** UPDATE AT 0958 EDT ON 9/10/2015 FROM PATRICK SHAW TO MARK ABRAMOVITZ ** *
The following report was received via e-mail:
"Rotork has concluded the investigation and based on test data determine an unsafe condition can exist asdefined under 10OCFR21. Testing has established .that a small percentage of the switches supplied against Rotorkpurchase order P0116932 may exhibit an open circuit condition when operated. A slightly larger percentage of thepopulation may exhibit high contact resistance ranging from 0.5 Ohm to open circuit. The anomaly affects thenormally open and normally closed contacts equally. Contact resistance anomalies are random, not permanent,and appear to be unaffected by accumulated cycles. Additionally when a switch is oPerated its electrical stateremains constant until the next operation.
"Probabilities have been established by resistance measurements and by dynamic testing using voltages
09/10/2015 U.S. Nuclear Regulatory Commission Operations Center Event Report Page 2•representative of customer applications. The dynamic evaluation predicts an open circuit probability of 0.0008%per operation. To date, no switch failures have been reported from any customer sites. Below are listed all ordersprovided to customers containing V12 switches from the defect batch. Report ER791 provides full details ofRotork's technical investigation."
Notified the R1DO (Gray), R2DO (Shaeffer), R3DO (Pelke), R4DO (Drake), and Part-21 Group (via e-mail).
rotor wCotro~s
Rotork Controls, Inc.
675 Mile Crossing Blvd.
Rochester, New York
14624
U.S. Nuclcar Regulatory Commission, tel: +1 585 247 2304Washington, DC 20555-0001 lax: +1 585 247 2308
www.rotork.comnFrom: Patrick A. Shaw, P.E if~ookcr
To: NRC Operations CenterDate: September 9(11, 2015
Subject: Rotork Controls Inc. Part 21 Notification, V12 safety related micro switch, Pt No N69-921
Dear Sir/Madame,
On June 4, 2015 Rotork Controls Inc. opened a formal Pt 2l investigation into a self-identified anomalyrelating to a Basic micro switch - Pt No N69-921, description "V12". The anomaly is intermittentvariation in electrical contact resistance and was first observed during a factory acceptance test of aRotork safety related NA range electric actuator; also referred to as an electric Valve Operator.
On July 28 2015 Rotork Controls Inc. wrote to the U.S.NRC requesting a 40 day extension to the 60day investigation period permitted under 10CFR2 1. The extension was granted and the investigationexpiration date was extended to September 12 th 2015:
Rotork have concluded the investigation and based on test data determine an unsafe condition can existas defined under I10CFR21. Testing has established that a small percentage of the switches suppliedagainst Rotork purchase order P0116932 may exhibit an open circuit condition when operated. Aslightly larger percentage of the population may exhibit high contact resistance ranging from 0.5 Ohmto open circuit. The anomaly affects the normally open and normally closed contacts equally. Contactresistance anomalies are random, not permanent, and appear to be unaffected by accumulated cycles.Additionally when a switch is operated its electrical state remains constant until the next operation.
Probabilities have been established by resistance measurements and by dynamic testing using voltagesrepresentative of customer applications. The dynamic evaluation predicts an open circuit probabilityof 0.0008% per operation. To date, no switch failures have been reported from any customer sites.Below are listed all orders provided to customers containing V12 switches from the defect batch.Report ER791 provides full details of Rotork's technical investigation.
Page 1 of 5
rotornControls
Rotork Controls Inc. have provided ER791I so that U.S. Licensees may evaluate each application whercRotork equipment is affected by this Part 21 notification.
Below this letter is a summary pursuant to 10CFR Part 21 .21(d)(4), a list of all actuators & sparecomponents containing V12 switches from P0116932, and a copy of ER791.
Sincerely • /
Patrick A. Shaw, P.EwQuality Assurance ManagerTel (585) 770-1019Patrick.Shaw @ rotork.com
President
Page 2 of 5
rotorwControls
Required informlation as per 10CFR Part 21.21(d)(4) follows:
(i) Name and address of the individual or individuals informing the Commission.
Patrick A. Shaw, P.E.Rotork Controls Inc,675 Mile Crossing Blvd,Rochester, New York14624
(ii) Identification of the facility, the activity, or the basic component supplied for such facilityor such activity within the United States which fails to comply or contains a defect.
Rotork part number N69-921I micro-switch V 12/3252 (RSI104) manufactured by JohnsonElectric (formerly Burgess) during 2013.Rotork Controls Limited P0116932
(iii) Identification of the firm constructing the facility or supplying the basic component whichfails to comply or contains a defect
(iv) Nature of the defect or failure to comply and the safety hazard which is created or couldbe created by such defect or failure to comply.
In 2013 Johnson Electric changed V12 micro-switches assembly locations. The first batch(P0116932) from the new location was found to contain small particles of plastic casematerial that are loose within the switch housing and may lodge on the contacts. Switchesmanufactured prior to 2013 are not affected and due to batch control traceability withinRotork, switches from the affected batch have been traced.The manufacturer's acceptance criteria for the switch contact resistance is less than 25milliohms. Contact resistance has been found to randomly vary between compliant (lessthan 25 milliohms) to non-compliant (greater than 25 milliohms to occasionally opencircuit).From a sample of 1613 switches, the manufacturer established a 0.11% probability that aswitch may exhibit an open circuit condition when operated. Rotork in-house testing withinactuators, using voltages representative of in-service levels, established a 0.0008%probability that a switch may exhibit an open circuit condition when operated. Loss ofcontinuity is random, not permanent, and low in probability. To date, no V12 switch in-service faults have been reported by the industry.
Page 3 of 5
rotorwControls
(v) The date on which the information of such defect or failure to comply was obtained.
7"h September 2015
(vi) In the case of a basic component which contains a defect or fails to comply, the numberand location of these components in use at, supplied for, being supplied for, or may besupplied for, manufactured, or being manufactured for one or more facilities or activitiessubject to the regulations in this part.
(vii) The corrective act ion. which has beeni, is being, or will be taken; the name of the individual.or organization responsible for the action; and the length of lime that has been or will betaken to complete the action.
By this notification Rotork is notifying all utilities and listing affected customer orders.The manufacturer Johnson Electric has introduced cleanliness checks at applicable stagesof switch assembly. Rotork has already obtained replacement switches and further batchesarc now being manufactured.
(viii) Any advice related to the defect or failure to comply about the facility, activity, or basiccomponent that has been, is being, or will be given to purchasers or licensees.
U.S. Licensees with installed or in storage actuators, switch mechanisms or Add-On-Pak(AOP) spares modules containing VI2 switches identified in section (vi) should evaluatethe impact of a high contact resistance or occasionally open circuit condition on safetyrelated systems.
(ix) In th~e case of an early site permit, the entities to whom an early site permit was
transferred
Not Applicable
Page 5 of 5
rotorkRotork Controls Limited
Nuclear Engineering Report No.
Page 1 of 20
Approved by Engineering
Date: 9 th September 2015ER791 Issue 1
REVISION SHEETv
ORIGIN
DATE OF ISSUE
TITLE
Privang Jadav
09/09/2015Part 21 Investigation..Intermittent operation of V12 Micro-Switch N69-921
Prepared Checked Approved
Priyang Jadav Stuart Adam Kevin SweetProduct Engineer - Senior Engineer - Engineering Manager -
Nuclear Nuclear Nuclear
Issue No Date Revision By
1 09/09/15 Original PUJ
rotorwRotork Controls Limited Approved by Engineering
This report documents work performed to support a Part 21 investigation opened by Rotork ControlsInc. ref. tracking number NCR164927. Measurements and tests were performed to establish the
operational reliability of a Rotork safety related micro-switch containing a manufacturing defect. Non-
compliant switches were identified from a batch of 4997 switches supplied to Rotork in 2013 againstpurchase order P0116932. The investigation has confirmed that the potential defect is only attributed
to the aforementioned 4997 switches and that switches orders prior to this are not affected.
1281 switches were supplied to customers prior to detecting the non-compliance. A percentage were
supplied to North American customers by Rotork Controls Inc. 3716 switches in stock at Rotork Ltd.
were returned to the supplier for analysis. From the 3716 switches, 1613 switches were tested by the
supplier, who established a 0.11% probability that a switch may exhibit loss of continuity whenoperated. However, in-house testing in actuators using voltages representative of in-service levels,establish a 0.0008% probability that a switch may exhibit loss of continuity when operated. Test data
indicate loss of continuity is random, not permanent, and low in probability. To date, no V12 switch
in-service faults have been reported by the industry.
2. Deviation Description
On June 4, 2015 Rotork Controls Inc. opened a formal Part 21 investigation into a self-identifiedanomaly relating to a Basic micro-switch - Part No. N69-921, description "V12". The anomaly isintermittent variation in electrical contact resistance and was first observed during the FactoryAcceptance Test of a Rotork Safety Related NA Range electric Actuator; also referred to as an electric
Valve Operator.
3. Micro-Switch Function
The micro-switch has four functions:
* Rotork actuators can be set to operate to a maximum torque level of opening or closing a
valve. When the torque sensing mechanism in the actuator registers this maximum torque,
the V12 micro-switch will be tripped to turn off the motor.
* Rotork actuators can also be set to operate to a maximum travel limit. The mechanism inside
the actuator will register when the maximum travel in the open or close direction is reached
and the V12 micro-switch will be tripped to turn off the motor.
*In the Add-on-Pak (AOP) the switch is intended for indication purposes but can also be used
for interlocks and permissives to start other equipment such as pumps and valves. AOPswitches can be set to trip at any point during valve travel. The AOP can also be used for torqueswitch bypass. If the circuit is "open" then the MOV could stop before achieving "end of
travel".
* Switches must be "closed" to initiate travel.
rotorweP eoRotork Controls Limited Approved by Engineering
Defect: Intermittent and/or high contact resistance of the N69-921 V12 Micro-Switch.
Why - micro-switch did not function as intendedWhy - micro-switch contacts were open circuitWhy - microscopic plastic particles lodged between the contacts created an air gapWhy- the action of the switch plunger can potentially disturb plastic particles within the switch
housing and cause them to become lodged between the electrical contacts
Why - inadequate de-flash and cleaning prior to switch assembly
Following the moulding operation housings require de-flash and cleaning operations. In 2013, for
P0116932, these operations were transferred from the housing manufacturer to Johnson Electric. Aninvestigation by Rotork QA and Johnson Electric determined the intermittent switching was attributedto inadequate de-flashing and cleaning of the plastic housing at Johnson Electric. All switches supplied
against P0116932 are thus suspect. Microscopic plastic particles may randomly lodge between thecontacts and cause the resistance to exceed the Johnson Electric 25mf) acceptable limit, and possiblyloss of continuity.
Figure 2: Uncleaned flash inside of ho using
Figure 3: Plastic particles around the contact
rotorwaeoRotork Controls Limited Approved by Engineering
1281 switches have been supplied against customer orders worldwide. The remaining stock of 3716switches from the defect batch was rejected back to the supplier for further analysis. The aims of
analysis were:
* To investigate whether a method could be established by which non-compliant switches can
be identified and segregated from compliant switches. Readings exceeding themanufacturer's 25m0 end-of-line acceptance criteria were classed as non-compliant.
* To determine whether high resistance or open circuit condition occurs on the normally open
(NO) contact, normally closed (NC) contact, or both.
* To determine whether the number of switch operations has an influence on contact
resistance.
* To collect data which can be used to calculate probability of switch failure i.e. open circuit
condition.
Test Method
* Each switch was operated 10 times. At each operation the NC and NO contact resistance was
recorded with a milliohm meter.(This gives 10 NC and 10 NO resistance readings)
* Test voltage = 9VDCTest current = 5mA
*The switch was then operated a further 50 times to check if the number of operations
influenced the contact resistance value. The NC and NO contact resistance was recorded at
every tenth operation.(This gives 5 NC and 5 NO resistance readings)
Total no. of operations per switch = 60Total no. of resistance readings per switch = 30
Test Results
Of the 3716 switches, 1520 were tested to above test method.1520 x 30 readings per switch = 45600 readings
A further 93 switches were operated only up to the point where high resistance/intermittentoperation was recorded. The total number of readings across the 93 switches were 911.
Total number of switches testedTotal recorded resistance readings
Table 1 characterises the resistance test results from this data sample. The milliohm meter can registerreadings up to 2000 maximum. Readings over this limit are classed as open circuit. The column 'No.
of Readings' is presented graphically in Figure 4.
Resistance Range No. of No. of Readings as Cumulative FrequencyReadings % of Total above 25mC)
_____________Readings manufacturer acceptance0 < R < 25m0 46148 99.220% ___________
25 < R < l00mO 240 0.5216% 240100 < R < 200m0 30 0.065% 270200 < R < 300m0 27 0.058% 297300 <R•<400m0 4 0.009% 301
400 < R _< 500mO 5 0.011% 306500m0 < R < 10 1 0.002% 307
The majority of non-compliant readings are < 500mO. In the range 50 < R < 200 there was one readingat value 21.90. The effect of this in a 48V, l0mA circuit and a 120V, 30mA circuit is calculated below:
V=IxR48 =0.0l0 xRR =48 /0.010 =48000
V=IxR120 = 0.030 x RR = 120 /0.030 =40000
If circuit resistance increased due to high switch contact resistance of 21.90:
4800 + 21.9 = 4821.9048 = L x 4821.90I = 48 / 4821.90 = 9.95mA
4000 + 21.9 = 4021.90120 = I x 4021.90I = 120 / 4021.90 = 29.84mA
The calculations above show that:* In a 48V, l0mA circuit, a contact resistance of 21.90 would reduce the current to 9.95mA.
* In a 120V, 30mA circuit, a contact resistance of 21.90 would reduce the current to 29.84mA.
The high resistance values captured thus have negligible effect on the circuit current in aboveexamples. The primary concern is thus the open circuit condition. Table 1 and Figure 4 show 46148compliant operations and 51 open circuit operations. The probability of a switch displaying an opencircuit condition is thus:
P(A) = 51 / 46511 = 0.0011 or 0.11% (operation based data)
Although Rotork are treating the whole batch of switches as suspect, Johnson Electric test data shows165 switches exceeded the 25m0 acceptance criteria. The probability of selecting a non-compliant
switch is thus:
P(B) = 165 / 1613 = 0.1023, or 10.23% (population based data)
Summary
Resistance testing demonstrates the occurrence of high resistance or open circuit condition is random.The anomaly occurs on the NO or NC contact, or both. Once the anomaly has occurred, correct switchoperation may revert on the next operation, or it may revert after a number of operations. Data showsno trend between the number of operations and the number of anomalies. Some switches displayedan anomaly on the first operation while others displayed an anomaly on the final (6 0th) operation.Most switches passed the resistance test. However, due to the random nature of the anomaly it ispossible that on the 6 1St operation an anomaly could occur. The segregation method cannot provideabsolute assurance.
Contact resistances in the order of tens of Ohms are considered to have negligible effect and mostanomalous readings were below this value. However, the utility must evaluate this point. The opencircuit condition is a concern. Test data based on resistance testing indicate a 0.11% probability peroperation that any switch may exhibit an open circuit condition.
Resistance testing is performed at 9VDC. Rotork extended the investigation to establish whethersupply voltage level influenced switch operation reliability. ERA, an independent test laboratory,
performed the evaluation. The results are discussed in the next section.
5 switches that lost continuity during Johnson Electric test were sent to ERA to establish a voltage levelabove which the switches would operate reliably. The conjecture for testing was that switches in opencircuit condition have particles lodged between the contacts which create an air gap that may bebridged by the formation of an arc if the voltage potential is sufficiently high.
At ERA the open circuit condition could only be replicated on 2 of the 5 switches. On one switch a DCvoltage ramp was applied while in open circuit mode, and on the other an AC voltage ramp.
The other 3 switches, which had shown open circuit mode during Johnson Electric test, showed no
fault at ERA.
Test Method
* Similar to Johnson Electric test, the switches were operated 25 times. At each operation theNC and NO contact resistance was recorded with a milliohm meter.(This gives 25 NC and 25 NO resistance readings)
* Test voltage < 5VDCTest current < 10mA
* The switches were then operated a further 50 times. The NC and NO contact resistance was
recorded at every tenth operation.(This gives 5 NC and 5 NO resistance readings)
* During operations if loss of continuity was captured, the switch was kept in the fault modeand a voltage ramp applied.
For DC ramp test:Voltage representative of in-service levels = 48VDCCurrent representative of in-service levels = 10mARequired level of resistance during ramp = 48 / 0.010 = 48000
For AC ramp test:Voltage representative of in-service levelsCurrent representative of in-service levelsRequired level of resistance during ramp
= 120VAC= 30mA- 120 /0.030 = 40000
rotorl( Page 1of 2
Rotork Controls Limited Approved by Engineering
Date: 9th September 2015Nuclear Engineering Report No. ER791 Issue 1
DC Test Results
A contact resistance reading of 60k0 was captured on the NO contact. A voltage ramp up to 48VDCwas applied to the switch. The switch started to conduct at 15V, stopped conducting at 23V, andstarted to conduct again at 32V (Figure 5).
Figure 5: Voltage ramp up to 48VDC when 60kD resistance was captured
On the next operation of the same switch, a contact resistance reading above 2M0 (milliohm meterlimit) was captured on the NO contact and a voltage ramp was again applied to the switch. This timethe switch started to conduct at 10V, stopped conducting at 32V, and started to conduct again whenthe voltage was further increased to 64VDC (Figure 6).
AC voltage tests were carried out on a different switch. A contact resistance reading above 2M0Q(milliohm meter limit) was captured on the NO contact. A voltage ramp up to 120VAC was applied tothe switch. The switch started to conduct at 104V (Figure 7). However, with the switch in the same
state, a subsequent ramp was applied (Figure 8) and the switch failed to conduct.
Tests have revealed that when a voltage ramp is applied to a switch that is already in open circuitmode, the switch will not always operate reliably. A safe DC or AC operating voltage could thus not be
established.
However, when the voltage ramps were applied, the switch was static and fault condition already
established. It was thought that in dynamic operation, mechanical impact of the contacts and the heatfrom the arc may both act to clean the electrode contacts. Based on this the investigation was
extended to study the likelihood of an open circuit condition arising when switches are being used inan actuator at voltage levels representative of the application. The testing and results are discussed
Rotork in-house tests were conducted to test suspect switches in the most representative manner ofhow they would be utilised in service. The reason for testing in a manner that represents theapplication is based on the hypothesis that mechanical impact of the contacts and the heat from the
arc may both act to clean the electrode contacts thus switch reliability in application may be betterthan Johnson Electric resistance measurements suggest.
A batch of switches including those that passed and those that were found non-compliant during tests
at Johnson Electric were called back to Rotork for further testing. Switches were assembled in theSwitch Mechanism, Figure 11, and Add-on-Pak (AOP), Figure 12, of 2 nuclear actuators.
The actuators completed a number of life tests to at least 4000 cycles Open-Close-Open. In betweeneach life test the switches in the Switch Mechanism and AOP were removed and replaced with moresuspect switches.
rotorwRotork Controls Limited
Nuclear Engineering Report No.
Page 15 of 20
Approved by Engineering
Date: 9 th September 2015ER791 Issue 1
Open auxiliaryswitch (OAS2)
Open auxiliary _____
switch (OAS1)
Open torque/limit -
switch (OTS)
AOP indication switches
IASi - IAS6
~Close auxiliaryswitch (CAS2)
~Close auxiliary
switch (CASi)
•----Close torque/limit
switch (CTS)
AOP indication switches
IAS7 - IA512
Figure 12: Photo of Add-on-Pack (AOP)
rotorI Page 1 of 20Rotork Controls Limited Approved by Engineering
Switches OTS and CTS were operated at 24VDC, 2.6mA average.Switches OAS1-2 and IASI-IAS6 were operated at 48VDC, 16.9mA average.Switches CAS1-2 and IAS7-IAS12 were operated at 120VAC, 12.5mA average.A reduced voltage life test was also completed using 38.8VDC, 13.7mA and 80VAC, 6.6mA.Voltages above 24VDC were applied through relays as illustrated in Figure 13.
•,POWER SUPPLY
V12 TEST •.jVMODULE:SWITCH I•- eod
Figure 13: Typical schematic showing V12 switch test
The operation of each switch was monitored during test. The data captured showed which switchesbehaved intermittently and when. The testing is summarised in Table 2.
A total of 84 different switches were tested during tests 1 - 9. Of these:* 28 were switches that showed no fault during Johnson Electric tests. These also showed no
fault during representative testing in an actuator.
* 51 were switches that intermittently exceeded 25m0 during Johnson Electric tests, having
readings in the range 26m03 to 2.8903.* 5 were switches that showed open circuit condition. This includes the 2 switches that were
used for DC and AC voltage ramp test at ERA. The 2 were used in Test 9 and no faults were
reported.
In Test 1, the AOP was fitted with switches that showed no fault during Johnson Electric tests. The
actuator completed 5537 cycles after which, in Test 2, all switches other than OTS and CTS werereplaced by more switches that showed no fault. In both Test 1 and 2 the switches that passed JohnsonElectric resistance test did not display open circuit behaviour during actuator operating life.
Between Test 3 - 7 the same actuator was used with the same switches. Switches OTS and CTS werethe same as Test 1 and 2. All other switches were those that showed high resistance during JohnsonElectric tests. The total actuator cycles in Tests 3 - 7 were 15686. The purpose of these tests was tosee what effect the voltage levels had on switch failure. In Test 3 there were 3 intermittent operationsrecorded and in Test 7 there were 4, both tests conducted at low voltage 38.8VDC and 80VAC (limitedby the minimum voltage capability of the relays). In Test 8 more high resistance switches were testedand there were 13 fault occurrences at nominal voltage 48VDC and 120VAC. These tests show thatthe fault can potentially occur at nominal and reduced voltage.
rotorl( ae1 of 2Rotork Controls Limited Approved by Engineering
Note 1: Switch at OAS2 (at 38.8VDC) failed to close circuit during cycle 865.Switch at IAS8 (at 80VAC) failed to close circuit during cycle 28 and cycle 954.
Note 2: Switch at OAS2 (at 38.8VDC) failed to close circuit for 4 consecutive cycles from cycle
42.
Note 3: Switch at OAS2 (at 48VDC) failed to close circuit during the following cycles; 279, 280,
281, 283.Switch at IAS2 (at 48VDC) failed to close circuit during the following cycles; 276, 3647,
3651, 3655.Switch at IAS9 (at 120VAC) failed to close circuit during the following cycles; 6, 43, 45,
47, 158.
All switches that failed (Note 1 - 3) were those that were found to have intermittent high resistance.There were no failures of switches that had previously exhibited open circuit.
For the calculation of failure probability over actuator life only the first 4000 operations of each switch
are relevant. A switch operates once per actuator cycle. Based on Table 2, the total number of relevant
operations are:
Test 1 4000 operations of 18 switches.Test2 400 peraion of16 sitces. Demonstrates failure is random and the 25m£0
Test 3 4000 operations of 16 switches. tetdecsfu.
Test4 -7 Swich oeraionsfro tes 47 With Test 1 - 3 demonstrates that failure isare not included in the
probbilty alcuatinsrandom and not affected by accumulated cycles.
Test 8 4000 operations of 16 switches. Demonstrates failure is random and the 25mC)
Test 9 4000 operations of 18 switches. test detects fault.
Table 3: Relevant operations for calculation of switch failure probability
It should be noted that the switches used in representative testing were not randomly selected. Theselected switches were pre-identified as being compliant or non-compliant as per Johnson Electric
test.
rotorl( ae1 f2Rotork Controls Limited Approved by Engineering
Probability of a switch exhibiting open circuit in application is given by:P(C) = P(B) x P(D)
Where:P(B) = Probability of selecting a non-compliant switch i.e. >25m02. Calculated in section 6.P(D) = Probability of a non-compliant switch displaying open circuit when installed in an actuator
The probability of a non-compliant switch displaying open circuit, P(D), is determined using data from
Rotork in-actuator representative tests 3, 8, and 9. The total number of switch operations across thesethree tests are computed below and displayed in Table 3:
2(4000 x 16) + (4000 x 18) = 200,000 (total operations of tested switches classified non-compliant)
The total number of observed open circuit operations in these three tests are:
3 + 13 = 16 (see Table 2)
P(D) = (16 / 200,000) = 0.00008, or 0.008% (operation based data)
P(C) = 0.1023 x 0.00008 = 8 x 10.6, or 0.0008% (per switch operation)
In some applications when a valve is initially stroked two switches are required to function correctly,the torque/limit switch and the torque bypass switch. The probability that either of these goes open
circuit is:P(2 Switches) = 2 x P(B) = 0.0016%100 - 0.0016 = 99.998% probability of success
The probability that any one of the 18 switches in an actuator fitted with a switch mechanism and 12switch AOP could go open circuit is:P(18 Switches) = 18 x P(B) -0.0144%100 - 0.0144 = 99.986% probability of success
Summary
The probability P(C) of a switch displaying open circuit in application is considerably lower than the
probability P(A) of a switch displaying open circuit during Johnson Electric test. These result suggestthat switch operation is more reliable than resistance evaluation indicates, It also supports why no in-
service faults are reported by industry and why Rotork Controls Inc. eventually observed the anomaly
based on 100% resistance inspection of all V12 switches fitted to actuators.
rotorI(Pge2 of 2Rotork Controls Limited Approved by Engineering
Testing indicates that occurrence of high resistance or open circuit condition is random with a lowprobability of a switch exhibiting an open circuit condition. Based on Johnson Electric testing of 1613switches, there is 0.11% probability that a switch may exhibit open circuit when operated. Rotork's in-
house tests show that in application there is a 0.0008% probability that a switch operation may exhibitopen circuit when operated.
Thermal & radiation effects could not be evaluated but are considered not pertinent given thecontamination nature. The housing material is qualified for high temperature and radiation levels thusthe material properties of contaminating particles are thought to be unaffected by these factors.To date there have been no in-service faults reported.
Rotork cannot evaluate application risk. Utilities must evaluate their in service switches and determine
if a high contact resistance or open circuit conditions could potentially affect safety related functions.
