ML20086T648

Responds to Info That Substantial Safety Hazard May Have Occurred at Unit 3 Re Rosemount Pressure Transmitters.Feels That Causes of Failures Have Been Eliminated

ML20086T648
Person / Time
Site:	Millstone
Issue date:	04/07/1988
From:	Lasell D ROSEMOUNT, INC.
To:	Russell R NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION I)
Shared Package
ML20086T599	List: ML20086T648
References
FOIA-94-310, REF-PT21-95 NUDOCS 9508030113
Download: ML20086T648 (47)

Text

I.

y P*

MOSEMOUNT INC.

• • • • ~ :.~.nology Drive Ecen Prairie, MN 55344 U.S.A.

(s'? F -5560 rA?.: 43*0012 or 4310001 Rosemount April 7, 1988 Regional Administrator, Region F U.S.

Nuclear Regulatory Commission "631 Park Avenue King of Prussia PA 19406 i

Attention:

Mr. William T. Russell

Subject:

Millstone Nuclear Power Station Unit No 3 letter of March 25, 1988

Dear Mr. Russell:

Northeast Utilities recently provided information that a Substantial Safety Hazard (SSH) may have occurred in their Millstone Unit No. 3 plant relating to Rosemount's pressure transmitters.

To Assist the NRC in evaluating this concern Rosemount wants to provide further background information.

Northeast reported that five transmitters failed with symptoms that could simulate loss of fill fluid (oil) from the module or sensing cell on transmitters with final assembly serial nurhers that were similar.

When the module (or sensing cell.) serial numbers were checked they were determined to come from two separate lots, No other failures of this type have been reported from either of these two lots of sensing cells supplied to other customers.

When Rosemount has evaluated other 1-oss of oil failures we find various unrelated leakage paths including:

i

-cracked fill tube

-leakage at fill tube to glass seal

-damage to isolator diaphragm

-leakage at cell cup to glass seal

-isolating diaphragm weld failure 1

Our research shows that even if there are one or two reported failures in a lot there is tio increased chance that other transmitters in that lot will fail.

/fA CE NO. 4-90-009 EXHIBlJ A pad 001285 l-Page 9500030113 950727 310 PDR

/

. k I

lW? D fHC

a:

1 i

B Whe.;.3:sc::u;u's repair and failure tracking system indicated that there were loss of oil reports we investigated possible causes and instituted changes in our manufacturing to prevent further failures from happening.

i The changes we made were tightening our inspection criteria and instituting additional operator training.

These were i

done in May of 1986.

1 In looking specifically at Northeasts situation we do not I

have full details on leakage paths since only two of'the five transmitters have been returned to us for evaluation.

A review of the traceability does show that different manufacturing lots and materials are involved.

l Rosemount has evaluated and researched this failure type and j

we have concluded that it is of a random nature.

We also.

feel that the causes of these failures have been eliminated.

If you have further questions on this or want to discuss l

this in further detail please do not hesitate to call Jane

~

Sandstrom at (612) 828-3286.

i Sincerely, Dick LaSell

~

cc:

Tom Shaffer Northeast Utilities I

t i

i j

EXHIBIT /P CASE NO.

4-90-009 ease A of A pa.es D001286 m

-,,-4

k"'

099990 s a-s:s s ' ( As v.

12/ss)

PGandE FOR IrfRA-COMPANY ORRM t-From Region or NECS - PROJECT Department ENGINEERING To Region or Department DCPP FILE NO.

137.101/4.73

'RE: LETTER OF SUBJECT Rosemount Transmitters NCR #DCO-86-TI-N152 i

l July 31, 1987 MR. R. C. THORNBERRY:

j Attention:

Mr. M. W. Stephens This is the complete response.to the Engineering portion of NCR

1. DCO-66-TI-N152 originated by DCPP. The NCR indicates that a problem of generic nature may exist on Rosemount Model 1153 transmitters.

Engineering's portion of the resolution is to determine if the problem is reportable under 10CFR21.

The subject NCR identified six Rosemount transmitters which i abowed sluggish response to system pressures.

These six transmitters plus two more transmitters which were suspected of having the same failure were returned to Rosemount for evaluation and repair or replacement.

Five out of the eight returned transmitters were found by Rosemount to be damaged during installation' They were used to monitor reactor coolant flow and steam generator steam flow.

The remaining three transmitters showed indications of loss of oil through seepage, thus causing tha slow response to the pressure changes.

These three-transmitters were used to monitor the pressuriser level.

The five transmitters which showed field handling damages like punctured isolating diaphragas and broken diodes due to reversed polaritr, came from a group of transmitters which were field modified to obtain the desired pressurerrange to match the application. During that time the immediate shipment of complete transmitters from Rosereount was not possible.

Transmitters were therefore assembled from component parts to prcvide the required ranges.

In the process, some transmitters were inadvertently and unknowingly damaged.

y The three transmitters that were not field damaged, LT-460 and two LT-461's, showed evidence of oil seepage.

They were from a shipment of transmitters that were manufactured before 1984 when Rosemount adopted stringent quality control procedures in the manufseture, inspection and approval for shipment of the transmitters.

E M BIL CASE NL 4-90-009 D001291 PAGE

/ OF k PAGE(S)-

ti.> v pg

j m of:.

099999 P!!3 Mr. R. C. Thornberry Page 2 July 31, 1987

'1 Rosemount's Mr. R. C. LaSell has stated that in the last two years only a small percentage (approximately 0.4%) of shipped transmitters, reported to be in the thousands, have had failures This percentage represents all failures, from field handling caused damage to failure caused by manufacturing defects.

1984 purchases except for twenty ope transmitters wh The with these 21 transmitters when the routine mainte t

and calibrations were done.

3 Further problems of sluggish satisfactorily for nearly 2 years. response are not expected since they i

t NECs I&C has evaluated the DCPP problems and the Rosemount analyses and data on failed transmitters.

4 The problem experienced did not constitute a substantial safety hasard i

problem or a reduction in the degree of protection pr the instrument channel.

mode failure due to this issue.There is not a problem with with common to the conclusion that the problem described in NCRTherefore, Engineering has

1. DCO-86-TI-N152 is not considered reportable under 10CFR21 G h M R. Y M. R. THESLER ELHerman/EN7aleriano ( 8 -1405 ) : Is 1 co:

JCCarroll, 77/1485 W0Crockett, 104/5/518, DCPP i

GIMoore 45/10/C29 RPPowers,, 104/5/16B, DCPP TRGrebel, 104/5/536 DCPP MJJacobson. 45/9/B32 Data Disk 3:

5:RCT1

~

l l

t

)

EXHIBIT b

PAGE / OF _b_ PAGE(S) 4 90-009 CASEMO.

1 D001292

~1 s

\\.

naamannamer suc, 12001 Tecnnoengy Orwe l

Eden prene, MN $5344 u1A.

(613 941-5500 TwX:4310012 or 4310024 FAX:(Gia 828 2008 i

-December 12, 1988 i

Rosemount I

Subject:

Rosemount Nuclear Qualified Transmitters j

Dear Purchasinci Agent:

Rosemount has built a strong reputation for superior quality, high reliability, and dependability of our products.

This has enabled Rosemount to become a leader supplying safety related pressure seasurement instruments to the Nuclear Power Industry.

To maintain this position of leadership and respect, we continually monitor an.

improve the performance of our products and we maintain open r

i communication with our customers.

\\

As part of our colemunication effort, we want to inform you of a condition that Rosemount has become aware of over the past few years.

A small number of pressure transmitters (Model 1153 and 1154) have been found to malfunction due to loss of oil from the sensing cell.

Under such a condition, the transmitter will respond sluggishly-to input changes or may drift outside it's normal operating specifications.

It will, after a period of two to three months, be unable to hold calibration.

Our investigation of field returned units has revealed that the leakage of oil may occur in one of several locations in tho cellt such as glass to metal interface, fill tube, and isolator diaphragm (due to damage during field repair).

The failure modes we have identified prove to be random occurrences in.

cause and in time.

Becauce the leak occurs randomly in the cells a~nd tends tc show up under severe conditions (high static pressure, 1000 psi or greater), the potential for failures of this kind are low and the potential of simultaneous occurrence on a redundant loop is miniscule.

Although stringent standards were in effect prior to. these reports, Rosemount has increased acceptance criterion for the manufacturing a.nd testing of the sensor cell andr we believe that the criterion now being used has essentially ell:sinated the pot. ncial for failures of this type. =Among the most pertinent changes include a high pressure aging that subjects the sensing call to conditions in excess of those which exist under normal operating conditions.

We believe this added conservatism in our manufacturing process assures that a reliable product is providers.

EXHIBIT

//

cmm.

4 - 9 0 - 0 0 9 D000391 PAGE / ~OF bt TG

> /

^

ff, yv f/g

~

V

\\

tracember 12, 1988

) g',

Paga 2 s

Rosemount has thousanda of qualified transmitters in service around the world.

We have received reports of occurrences on less than.5%

of our installed bawe.

There have been no reports on transmitter cells built in the past 3 years.

Our nuclec.r qualified pressure transmitters have proven over the yearn to be a highly reliable product.

Rosemount is striving to continuously improve both the product and our service to the nuclear industry.

have experienced an occurrence as described or have any questionsIf you feel you ma about this informational letter or about oar products, please feel Neil Lien or Jane Sandstrom. free to call either your local Rosemount sales off Sincerely, Richard C.

LaSall i

Nuclear Products Group Manager RCL/kkh P

P EXHIBIT PAGE OF hAQC(S)

E CASE ND. 4-90-009 D000392

. ' ~

11AY.12 *89 12:49 ROSEt10LJiT ItC. E.P. 612-828-3145 e.== " w. s a-

['

RottMOUNT INC.

12001 Technology Detve Eden Prairie. MN 65344 U S A.

(612) 041 5560 TWX: 4310012 or 4310024 FAX:(612N28-3068 Rosemount

' data \\waddress. doc May 10,1989 Company address 1 8ddress2 city, state zip Rosemount has served the Nuclear Power Industry for many years and has a strong commitment to continue serving the industry.

Attached is the first in a series of technical bulletins to be issued by Rosemount.

This first butistin addresses our current understanding of the performance of the Model 1153 and 1154 series transmitters under loss of fill fluid conditions, it also j

outlines the results of testing which Rosemount is conducting to develop techniques to monitor and identify transmitters that have lost oil.

We strive to provide the best possible products and support to our customers in the Nuclear industry. Technical ana sls and product improvement efforts will continue to be focused on this problem to nsure a comprehensive solution.

Sinceroty, ROSEMOUNT INC.

Steve Wanek V.P. Operations attachments EXHIBIT CASEll0. 4-90-009

/ 07 gPAcE(s) pygg l htL/C/ fl3

n g ga TECHNICAL BULLETIN No.1 May 10,1989 i

This bulletin addresses the loss of fill fluid which has been Identified on Model 1153 and 1154 Pressure Transmitters. This fill fluid loss occurs as a j

result of an Intemal leak in the sensing module. No fill fluid is lost to the

]

process. When this condition occurs, transmitter performance can be reduced. The result may be increased time response and/or an inability of the transmitter to measure input accurately.

Rosemount's emphasis for field diagnostics has t;een directed toward developing techniques to identify and monitor suspect transmitters.

Rosemount has developed an analysis technique to slmelate fill fluid loss by artificla!!y varying the amount of fill fluid in a test trarsmitter. This is cono using a specially built sensing module with adjustable fill fluid volume.

' The following inbrmation details this analysis for the Range 5 (0 750 inches H 0).

Pre!!minary information on the 2

Range 9 (0-3000 psig) is also provided. Analysis has been focused on these pressure ranges because they exhibit the greatest potential to fail.

As a sensing module loses fill fluid, it's performance characteristics change. These clianges depend on whether the f!!! fluid is lost from the high or low pressure side of the sensor. The main ' characteristic of concem is the response t!me to a step change in pressure.

Other characteristics include span and zero shifts.

A basic knowledge of the sensor design is helpful to understand the impilcations of fill fluid loss. Please refer to the sensor cross sectional view (Diagram A). Under normal operations the amount of fill fluid under the isolating diaphragm is sufficient to insure that fut scale deflection of the center diaphragm occurs without permitting the isolating diaphragm to make contact with the convolution plate.

Even in an overpressure condition, the center diaphragm will contact the glass capacitor plate with sufficient fill fluid under the isolating diaphragm to avoid contact with the convolution plate.

EXHlbi'.'

OF M PAGGS)

PAGE CASE N0. 4-00-009 Page 1

. m 4c o [4c wo nv w m. ii. c..r.

o4c oco-o*-a

- - ' - ~

e As a sensor loses M fluid, a point will be reaehed where the isolating diap.hrsom will contact the convolution plate. When this condition occurs, the transmitter will fall to function properly since the center diaphragm will not move correctly in relation to the applied pressure. Nominally, the distance between the isolating diaphragm and convolution plate is approximately.014 inches in a normal sensor with no differential pressure applied. This A!! fluid volume (between convolution plate and isolating diaphragm) corresponds to 0% M fluid loss. Total M fluid loss (100%) is defined as the point at which the isolating diaphragm is in contact with the convolution plate, agaln with no differential pressure applied.

The loss of M fluid is believed to be pressure dependent. Thus differential transmitters that are prone to leakage and subjected to high s:stic pressure on both the high pressure and low pressure ports may lose fill fluid from either side of the sensor.

Gage and absolute pressure transmitters would not be expected to lose fill fluid on the low side since no pressure is applied to this side.

l There are significant differences between Range 5 and 9 sensor performanca characteristics as 2 fluid is lost. They are discussed t

separately for this reason.

EXHIBIT-PAGE 3 OF k PAGE(S) eggs.

4 - 9 0 - 0 0 9 Page 2

HANON 5 ANALYSIS

\\

Range 5 transmitters are typically used in differential applications. For this reason, analysis was conducted for simulated fill fluid loss on both the high and low pressure alde independently. Measurements were mede throughout the fill fluid loss simulation to evaluate the following.

1.

Calibration shift

2. Time response for a step increase of 750 in H O in pressure 2

3. Time response for a step decrease of 750 in H 0in pressure 2

4.

Overpressure to high pressure side This testing has been completed on one modified unit.

RANGE 5 HIGH PRESSURE SIDE FILL FLUID LOSS The measurements made during the high side fill flu!d loss simulation are plotted vs. fill fluid volume lost expressed in percent (see Figure 1). Refer -

to the paragraph defining fill fluid loss on page 2.

Following are i

comments on each of these measurements:

1.

CAUBRATION SHIFT As soon as fm fluid loss begins, the calibration of the sensing module begins to change. There is a negative shift in zero that is fairfy linear over the fm fluid loss range under study. Initla!!y there is very little span shift, but once fili f!uld loss exceeds 54%

the span also begins to shift in the negative direction. The zero

{

shifts.8% of Upper Range Umit (URL) before the time response of the transmitter is affected. However the time constant (up) increases significantly at the upper end of this range. It should be noted that this testing was done with the transmitter at full range catibration (0 to 750 inches of water.).

The zero shift would be amplified by the range down factor of the transtnitter for different calibrations. Thus if a Range 5 were callbrated 0 to 125 inches of water, the zero shift would ba expected to be amplified by a range down factor of 6.

EXHIBIT._ /Y CASENO. 4-90-009

[ or _/3_ p;,9;g p,C,3 pace

'811'819

f..

J ' - my.ta see 12:51 noSEMOUNT INC. E.P. 613-82e-3145 i

At fun range down, the zero wm shift 4.8% of calibrated span t

before the time response of the sensing module is effected.

The zero shift will be apparent at any. operating point, as there is little offsetting span shift. Accurate measurement of zero shift le most typically accomplished by reposted. monitoring of transmitter output with both high and Iow sides vented to atmosphere or both high and low sides at the same static

[

pressure.

l

2. TIME RESPONSE FOR A STEP INCREASE IN PRESSURE TO.

HIGH SIDE

{

The time censtant for a step increase (750 in. H 0) in pressure i

a begins to increase between 38% and 46% fill fluid loss. For Range 5 tho time constant is normally 100me.

1 Time constant for fill fluid loss exceeding 57% was not plotted because of the magnitude.

As a point of reference, total response time at 62% All flu!d loss exceeded 10 minutes.

3. TIME RESPONSE FOR A STEP DECREASE IN PRESSURE TO HIGH SIDE Over the range studied there was no change in response to a j

step decrease (750 in. H 0) in pressure. Funher testing will be 2

done to determine if this is true throughout additional flu fluid loss.

4.

OVERPRESSURE TO HIGH PRESSUi?E SIDE When a normal sensing module calibrated at fan range is overpressured to 140% of the upper range Irr4 (or more), the

~

sensing diaphragm will be forced against the low side glass capacitor plate and the output will saturate n approximately 26 ma.

As the available fit fluid is decreased, the isolating diaphrsom will bottom out before the sensing diaphragm. Thus

-l the output will not go to 28 ma (saturate). Our data shows that when 54% of the fill fluid has been lost the sensor can no longer saturate when overpressure is applied.

Thus the overpressure output will be less than 28 ma.

/

CME tto. * *: 90-009 EXHIBIT P

I OF [1 Pt.GE(S) age 4 PAGE

,,w.E.. '

my,12 *e912:51 RoSEMMT INC. E.P. 612-029-3145 P.012/019

[

3. '

1

~

We now believe that the overpressure test identified above can be used to determine if a sensing module has lost enough fill fluid on the high -

pressure side to affect response time.

in conducting this overpressure test, power supply voltage is critical.

There must be adequate voltage to power the transmitter at 28 ma (14 volts required) plus enough voltage to cover load at 28 ma. For ETpie, if the total load for the loop le 500 ohms, then minimum power supply '

would be 28 volts,14 volts for the transmitter and 14 volts for the load. 'If less than 28 volts is applied, the output cannot saturate at 28 ma.

RANGE 6 LOW PRESSURE SIDE FILL FLUID LOSS The measurements made during the low pressure side fill fluid loss simulation are plotted vs. fill fluid volume lost expressed in percent (see Figure 2). Following art comments on each of these measurements:

1.

CAUBRATION SHIFT As soon as fill fluid loss begins, tile calibration of the sensing module begins to change. There is a positive shift in zero that

-is fairly linear over the fill fluid loss range under study. The span shift is also positive over the entire range under study.

Once again, ranged down calibration would amplify the zero effects.

The full scale output shift would exceed 2.5% of calibrated span before the time response of the unit begina to degrade for a fth range calibration. For a fully ranged down calibration, the shift would be expected to exceed 10% of calibrated span before any change in time response occars.

2. TIME RESPONSE FOR A STEP INCREASE IN PRESSURE TO HIGH SIDE.

In this testing,92% of the fill fluid volume wns removed. Over this range there was no degradation in response to a step increase (750 in. H O) in pressure. F;rt 1er testing will be done 2

to observe the effect through total fill fluid loss.

CASE NO. 4-90-009 EXHIBIT _

Page 5 PAGE 5 0F d PE(S) w-e

-a

,q

,.,,_,,,,,,e,

,,_,,.,-_1,.

,,,.,_,,,,.,,,.,,,,n.m,,,,n-n._,a,>

,.,.,ung.,.

e-

mY.12 '9912:52 ROSEM0LNT INC. E.P. 612-829-3145 P.013/019 i

3. TIME RESPONSE FOR A STEP DECREASE IN PRESSURE TO HIGH SIDE.

Between 85% and 92% M f!uld loss, the time response to a step decrease in pressure (750 in. H 0) begins to increase. Again, 2

further testing will be done to determine the effect through total fill fluid loss.

4.

OVERPRESSURE TO LOW PRESSURE SIDE.

This test cannot be used to determine loss of fill fluid with standard Model 1153 and 1154 electronics.,

1 i

EXHIBIT -

/M PAGE 2 OF /_J__, PAGE(S) i cmEm.

4 - 9 0 - 0 0 9 Page 6 l

~

'MAY.12 '89.12:52 ROSEMOLNr INC. E.P. 612-929-3145

• P.014/019

~

j.-

RANGE 9 ANALYSIS The Range 9 transmitter is a gage pressure transmitter. Therefbre, the low pressure side is not pressurized and is not believed to be subject to All fluid loss. In fact, no low side fa. lures of gage pressure transmitters have been reported. For this reason high pressure side All fluid loss only was.

studied. Since this work is not complete, graphs of the data are not included at this time.

I Preliminary testing shows that the Range 9 begins to show increasing time response to a step increase (3000 ps0 in pressure when 40%.of the available fill fluid has been lost. By the time 53% of the fill fluid has been i

lost the total response time to a step increase exceeds 4 rninutes. When 93% of the available fill fluid was lost the total response time exceeds 8 minutes.

The range 9 does not exhibit any appreciable calibration shift during fill fluid loss.

4

'J# '

~

EXHlBIT PAGE OF/k.,Pf.GSS)

Page 7

L r%Y.12 '89 12:53 ROSEMOUNT DC. E.P. 612-829-3145

"*d1 #dA#

~

SUMMAhtY Normal Range 5 and 9 sensing modufes have a very fast time response to a step increase or decrease in pressure. Time constant is generally loss than 100 ms.

Generally, as a sensing module loses fill fluid, the response to a step l

pressure change degrades.

Range 5 response to a step pressure l

Increase or decrease can be affected depending on which side of the l

sensor loses ffU fluid. This Range 5 behavior should be true for all differential transmitters regardless of range. The range 9 degradation is only for a stop pressure increase. Range 9 response to a step pressure decrease should not be affected by fill fluid loss. Furthermore, this Range 9 behavior should be true for all gage pressure transmitters regardless of range.

Instrumentation personnel should be alerted to note any pressure transmitter sluggishness or slowness of response. Transmitters should have total response to a step pressure change of less than i second (Range 3 and 4 are slower, please see instruction manual). Any slower response Indicates possible fill fluid loss. Beyond the initlet stages of failure, the time response degradation is very obvious during calibration.

Substantial calibration shifts could also be an Indication of low fill fluid. As a result of testing to date, the Range 5 shows considerable calibration / operating point drift as flu fluid is lost. The Range 9 shows almost no calibration shift as flu fluid la lost. However, we believe that in advanced fill fluid loss, the Range 9 will have substantial negative span -

shift.

If sufficient fill' fluid loss has occurred to affect the time response, Range 5 transmitter that is zero based and calibrated at full range will not Indicate saturated output (28 ma) when 140% (or more) of URL pressure is appUed to the high side.

Again, the transmitter must be zero based and ca!Ibrated at full range for this to be meaningful. When evaluating Range 6 transmitters, we consider this test to be a conclusive method of finding a high side flu fluid loss condition. We believe this situation occurs at or.,. wi t'w r,od W uhich time response begins to degrade.

Over-pressuring the low side is not meaningful. !g CASENO. 4-90-009 EXHlBlT

(

pAmt1OF M Pf,GC(Sf*E'

' " ' " ~

................ ~ -..... -... - - - - - - - -

Analysh of the RInge 9 with regard to this overpressure / saturation effect has not been completed at this time.

FURTHER ANALYFdi Additional analysis is currently being done to provide more comprehensive diagnostic and monitoring methods. We will communicate this information when confirmed results are avallable.

The scope of current programa include:

Frequency response analysis Long term pressure testing to dete;mine leak rates Continued characterization of the failure mode to determine simple monitoring techniques (including testing for different pressure ranges Rosemount is continuing to monitor this situation closely. We ask your assistance by reporting directly to us if you have transmitters with the i

symptoms identifled in our correspondence to you. If you know of such problems, please provide us with the fo!!owing information:

Transmitter serialnumber Modelnumber Exact symptoms and trouble shooting performed The time in service and the application parameters (such as cal!bration and static pressure)

It is important that Rosamount receive this Information, and that fa!!ed units to be returned in as timely a fashion as possible. Rosemount needs these units in order to support our current work on the degradation rate so that information on this subject can be provided to our customers. Units will be submitted to failure analysis after this valuable information is obtained.

Please catt Nell Uen (612) 82B3100 Tim Layer (612) 828-3540 or Jane Sandstrom (612) 828 3286, if you have any information or questions on this situation.

EXHIBIT _.!__

_ f cases.

4 - o 0 - 0 0 9 PAGE M or /.,1.P?.GMS) p g, g

mv.ta 'es 22:54 RCTRO UC. E.P. 612-829-3145 e.uiuuas

[

SdNSOR CROS8 SECT 10NAL VIEW DIAGRAM A i

Tech Busetan No.1 1

\\

NN

3 ;;A f ljj ///

d s'

/!i x

r. d 9 a

4 !!A 4~i :,,.,5 4

\\

h ehfk)

\\

s' f

- CENTER Mm b

$f' L'M' M,,5 A

{

m

~

V

\\

w:

7, OIL FILL

,9

3 5,',3/

I

~ISOLAT6MG DIAPHRAGil

\\N

! t e

s 3

5

,'?

/ /s s*e.

\\

\\

l6l4l ss o,1 I'$s,i$5.'

y

's, gf',f, sI ff;;

• CONVOLUTION PLATE GLAS$

s FfE,h -75 d k

NW/

N

// h EXHIBIT-.

cm30.

4 90-009 PAGE,!/ OF 2f_ P3.0-(0) l Page 10

e l

Time Constant (irated spa)n Seconds Zero Shift car

~

Span Shift 6 caBbrated span dddEooeww ww bab6Lbbbbbb 88888888888 3

• G E

F;;

f R

5 da n

1 h3

• I D O

=9 ag 8m m

%n l 8

f>

a E

a g

e 8I:s 88~

E

=*

4 h

n e ag g o

]

p o

a

]

4>

l VSE b

/~

"3E EXHIBIT' 4 PAGE /P F iPAGE(S)

O CASENO. 4-90-009 Page 11

M M6 6 0 0 - 0 o - t wasva

-(S)30Va p.go p 3nyg l

N *"*',

.o 8

}

i i

o

l =

• h m

0 ;:

6 hO$

o f

oQ o g l

W O

i ao

\\

E @N6 l

aw o

o'

• i g

W ho ;

g o

u E

I o90 :

l A

D a w e-EE El a O

g g

m m 9

~

o g

88.88888888 essseene a

yy7 nnaw

.4 o o o d$ pole >ctleo$Q mg utd3 j

unds pelsJ SW nlNS cuez (spucco lus;suog e mi.1, g

G$

4 t

F-eze-27 a 3 7x mwasgsizi ce, ayuu 9

5' "a'"a'S

.1 ROSEteoWNT 10$C.

12001 West 78th Street foon Praires. Mesmeeone 55344 U S.A.

Tot. (612) 941 5560 rwx 910-576 3103 TELEX 29-0183 March 6, 1986 ye Rosemount-

~

FAILURE ANALYSIS Model Number:

ll53GD9PA Serial Number:

408466 Customer:

VEPCO (Surry)

Customer Problem:

Slow response time ROSEMOUNT EVALUATION: Nuclear Technicians Slow response time was verified, low oil problem suspected on the high side sensor cell half.

ROSEMOUNT FAILURE ANALYSIS:

Evaluation Laboratory Sensor Module The sensor module (S/N 957435) was. cut open and the sensor cell removed.

There was oil present on the inside of the sensor module housing. To determine the leakage path the sensor cell was mounted on a test fixture and pressure was applied.

An oil leak was observed from a crack in the weld between the isolatoi diaphram and the sensor cell body.

CONCLUSION:

The failure mode was oil leakage from the weld between the sensor cell body and the isolator diaphram.

The probable cause is an inadequate weld joint, which progressively fatigued and eventually failed.

ROSEMOUNT REWORK TO CORRECT PROBLEM:

No rework porsible due to the destructive nature of the failure analysis.

ACTION TO PREVENT RBCURRE18CE:

This type of fa'. lure is a rare occurrence.

Appropriate manufacturing and quality personnel have been made sware.of this failure.

I Robert E Reinertsen Nuclear Quality Assurance EXHIBIT Page / of /

Pa es a

GERO. 4-90-009 D001217

/

// ) v

(

j

ReeM00MNT 30c, seen PseMe, Menneems $5W U.SA is.sta ses.seso ptCA U' N "

nexeiosm-am inu moin

! -0.

April 28, 1966 h

James A. Fitzpatrick Station

'adncTs om.

New York Power Authority

m Attn: Mr. Radfoni J. Converse Plant knager P.O. a n 41 Lycoming, N1' 13093

Dear Mr. Converse:

I first wish to thank you for the time your personnel spent with Bob asch, George Poole and rnyself on our April plant visit. We have taken the infonnation we gathered from the visit and have been reviewing it to come up with answers to the events at your plant.

As I mentioned before, we have seen very few of t.he low oil failures before, at a rate of approxinutely 6 per every $000 units according to Rosemount Quality personnel. 'Diese low oil failures have not shom conmonality, as the two returned by you illustrated (one with a weld leak and one with a fill tube leak).

Rosenount has taken action to prevent future pmblems by revising the in-process testing for leaks and tightening the acceptance criteria for both the pressure tests and oil level tests in our manufacturing.

One of the other transmitters at Fitzpatrick had experienced inter-mittant failure irxiicated by low output (about 3 Stsdc). We discovered while out there you used snubbers on that piping. Rosemount recom-mends you remove these as we have found similar " failures" have been eliminated by doing this. The snubbers can be casily plugged arti can cause this type failure, and the snubber should not be necessary for proper operation of the transitter.

As to the occurance of zero shifts at your plant, we can not, at this time, give you a definative answer.

I can say that this is a. problem we have not seen in the field Wre it was not traced-to an applica-tion or calibration phenomenon (such as an overpressuro).

I would encourage you to review your calibration procedures and monitor those applications where you have seen the shifts.

EXHIBIT N Page

/ of-V Pa.es CASE ND. 4-90-009 D001293

/

?.

re g

... > c.:..:.

O o

New York Power Authority April 28, 1986 P30e'bn Please keep us informed of the perfomance as we ane anxious to help solve the problem, but quite honestly don't have an answer based on cummt data and evaluation.

Roserrount has continued to honor the warranty on the units you sent in even though the warranty period has expired. We also will replace at no charge the unit you noted with two zero shifts (S/N 410050).

You should retum that unit to us for evaluation. We are very inter-ested in testing and disecting this unit. Should any of the other unir.s that you noted as having shifted exhibit a second shift, we also will replace those at no charge.

Ron %tthews had requested infonration on our solid teminal block screws.

With your no-cost purchase onjer, we will set you six of these kits.

I hope by these actions, Rosemount has demonstrated our concern and let me assure you we will continue to support you to resolve these issues.

If there is anything you wish to discuss or need.

l please feel free to call me at 612-828-3286.

Sincerely, A

J Sandstrom Sr. %rketing F/tgineer cc: George Poole - # 17 Ibb Bach - A8 Dick LaSell - A10 I

Bob VandenBoom - A10 i

elbb Volsted - Allgp Vic Walz - New York Power Auth.

Joe Flaherty - ik.* York Power Auth.

i 1

N EXHIBIT Page M of P Pa es canno. 4-90-009 s

D001294 1

MOGGA00000i880C.

T 6

12001 West Feti Street Eden PreMe, Mkinesses 55344 U.S.A.

Tal,18121941 5600 i

'~

2 TWX 910-676-3103 TELEX 29-0183

~ W.:

July 10, 1986 g*

i i

Pacific Gas & Electric Co.

Dept. General Constru I & C P.O. Box 117 Ba1Mug 202 Avila Beach, CA 93424 Attn: Mr. Bill Hicks V Subj: Return Transmitter Evaluation J

3 Ref:

Pacific Gas & Elec. P.O. GM-167027 RMr H.O. 209481

Dear Mr. Hicks:

We have completed our evaluation of the two tmnsmitters returned to us on the above referenced ortier and detennined that both units will requim sensing module replacement. 'Ihe module replace will be done under terranty and at no cost to Pacific Gas & Elec Both units am curmntly scheduled for shipment to you on July 31 1986. The lengthy tum around is due to thc. current module si in this range. If we are able to impmve on t.his comnitment, you will be advised.

Both units were returned to us for having a slow'or sluggish response to pressum changes. 'lhis symptom is typically associated with low oil in the sensing module resulting fmm an oil leak. Oil leaks can occur in mom than one location in the sensor cell such as in welds, fill tubes or glassing to housing interface areas arri the returned units both leaked at the glass to rgetal interface.

Being sensitive to module. leaks, we have continued to change our in-pmcess acceptance criteria so that currently nothing with any potential t for failum could be shipped. Although stringent controls were inforced when the two redules were manufactured in 1984, we are confident our pmgress has impmved the pmduct.

Should you have any questions concerning the evaluation, please contact the undersigned directly. I can be reached at A/C 612-828-3473.

Very truly, GLV660ticL I R.L. Volsted Supervisor Nuclear Projects EXHl81T b Y

/kp Page l of I

CASENO. 4-90-009 g*'5 D001229 gjg e

s. n

E Rosemohdt*lNTRACOMPANYMEMO 6

'~

Distribution

.2 Sta:

Jerry Anderson BS o..

Each BS

.. p:

i.

F.'.0.*: :

Bil; rscch Bob Volsted BS Dick LaSell BS

SUBJECT:

Meeting Notice - Nuclear Review Cocnittee DATE:

Friday, July 11, 1986 TIME:

9:00 to 10:00 a.m.

LOCATION:

Officers Conference Room

,3..."

TOPIC OF CONCERN:

Low 011 in Cells of Nuclear Returned Transmitters

1

. g c::

R a n %./ a i l e e s - U s b, n j l.'n c e,2

g. /l c lmi.us d e

_' O A 4

i EXHIBIT

[)n0.* '152 Page d

7 N.es CASENO. 4-40-009 Form No. 60035 Rw. a l

• A

f. -

f

4' ROSE 0800NT Itec.

12001 West 78m Street Eden Prswee. Mennesota 55344 U.S.A.

Tel (612) 941 5560 TWM 910 576 3103 TELEX 29-0183 l

September 29, 1986 David Holliday James A.

Fitzpatrick Plant New York Power Authority P.O.

Box 41 Lycoming.

NY 13093

Dear Dave,

This letter is to follow-up our meeting earlier this month and to pass on further information on the responses you have had to the Network Bulletin you sent out.

We have checked further into lot 154 which is of concern and have confirmed that there are not any other loss of oil modules from that lot to date.

The two Utilities (Commonwealth Edison and Kansas Gas & Electric) who reported these types of symptoms to you over the network are not from i

this lot to the best of our knowledge, nlthough Commonwealth's transmitters were installed around the same i

time as yours they are from earlier lots (1982 vintage).

We are further verifying this information since at this time they have returned only two transmitters to us.

Kansas Gas

& Electric transmitters while also installed the same year 3

as yours are actually hand-me-down transmitters from the canceled Marble Hill plant and therefore were manufactured in an earlier time period.also.

I am getting the serial numbers on these to add to our data base.

Since these transmitters were disposed of due to contamination we cannot i

verify the loss-of-oil or the potential for mishandling.

I Dave, I will be following through on these and let you know of the results.

If you have further questions please feel j

free to call me.

sincerely, f

A - - -

N%

Jann 84Q@htrom Sr. m-thet:!ng Engineer Qff TM

/jka EXHIBIT 1

Page / at /

Pa.es cc:

Phil Mercandetti #33 cutno.

4 - 9 0 - 0 0 9 D001280 p/g/

. :.r :

i e

GPU Nuclear Corporation I,

g3 g{

Post Office Box 480 b

W Route 441 South Middletown, Pennsylvania 17057 0191 717 944 7621 TELEX 84 2386 Writer's Direct Olal Number:

l (717) 948-8151 1

3340-86-0015 File: Maintenance -

Corrective October 9, 1986 Rosemount, Inc.

12001 West 78th Street Eden Prairie, Minnesota 55344 Attention:

Robert VandenBoom/ Timothy Jennek.e

SUBJECT:

1153 Transmitter Failures RMA #N-209648 Enclosed please find supporting information on the two 1153 transmitter failures.

As was discussed earlier, the MU-42-DPT transmitter is installed to measure deal injection flow to the reactor coolant pumps (0-80 GPM) while the MS-PT-950 & 951 transmitters are installed to indicate steam generator pressure (0-1200 PSI).

Included in this package are:

(1) maintenance history files on the subject transmitters and (2) copies of the " Failure Mode Question Sheets" with supporting information.

This second failure in the MU-42-DPT application was quite similar to the first -

occurring after only four (4) months of operation. Output from the transmitter failed low and was erratic. Again, an attempt at calibration indicated slow response time (approximately 1 minute) at the upper 1/3 of its dP range.

Failures of the MS-PT-950 and 951 transmitters have been concurrent with steam generator pressure transients following a reactor trip.

PT-950 failed high and remained high. Attempts at venting the transmitter proved unsuccessful. Not until a vacuum was placed on the sensing module diaphragm did output return to normal.

PT-951, on the other hand, has failed and stuck low on several occasions.

Isolation and venting of the transmitter allowed the diaphragm to " snap" back to normal.

We are again requesting Rosemount to perform a failure analysis on the returned transmitters. However, as stated previously, we were not fully satisfied with l

the former failure analysis on MU-42-DPT (S/N 411633).

It was stated that failure EXHIBIT N Page I of S Pa.,es l

g g). 4-90-009 D001536 GPU Nuclear Corporation is a subsidiary of the General Public Utilities Corporation u

F/9

4 i

?'

8 I

Rosemount, Inc.

. October 9, 1986 I

Robert VandenBoom/ Timothy Jenneke 3340-86-0015 l

was due to a manufacturing defect and that "this type of failure is a rare occurrence." Yet, as originally indicated, we have experienced two similar failures now in the same application. Also, if failure truly was due to manu-i facturing defect, should not Rosemount be responsible for cost of repair? We ask that further evaluation be performed to determine root cause and _ that l

Rosemount provide recommended installation, calibration, or maintenance i

instructions to prevent such failures in the future.

l Sincerply,f i

/

,7 1

[

4~$

h V. P. Orlandi Lead I6C Engineer TMI-1

]

JAD/VP0/gh l

Attachments l

cc:

Engineer I, TMI-1, J. A. Dullinger EP&I Engineer, D. Smith i

Instrumentation Manager, Tech. Functions, A. P. Agarwal l

Lead I&C Maintenance Foreman, TMI-1, L. E. Eberly Manager, Plant Engineering, TMI-1, C. E. Hartman i

Plant Engineering Director, TMI-1, J. J. Colitz J. A. Dullinger Writer's File l

1&C Letters l

CARIRS 1

4 k

i I

3 I

EXHIBl{ 7M Page dM Pa.es cAsao.

4 - 9 0 - 0 0 9 D?O3.537

1

, 3 p-AOKWOUNT INC.

12001 West 78th Street 6

Eden Prairee, Menosota $5344 U S A Tei1612)941 5560 i

TWx 910 576 3103 TELEX 29-0183 October 23, 1986 Rosemount-Dick Patch James A.

Fitzpatrick Plant P.O.

Box 41

Lycoming, NY 13093

Dear Dick,

Please find enclosed a copy of our reliability study we committed to provide you during your recent visit.

As i

you and I discussed in our telecon of October 21 there are a few specific points I would like to emphasize.

Our study is based on a statistical analysis of o

a limited sample size of transmitter failures.

Although not conclusive the results indicate e

this aberration is near its end and that by February _1987 99.8% of all failures should have occurred.

We are reevaluating this study in light of one more confirmed loss of oil failure but at.this time it is our opinion it will not significantly change the results.

I hope this study will be useful to your company not only for the information it contains but also for its value as another gesture of support by Rosemount.

If you have any

]

questions about the report or anything else please feel free to call me at 612-828-3286.

Sincerely,

/

. W M

~

Jane Sandstrom Sr. Marketing Engineer 1

/jka 73 m'

EXHlBIT 1

cc: Bob Volsted 85 Page

/ rd /

l Pa.es 1

CASENo. 4-90-009 D001525 1

Y//r

ROSEMOUNT. lMC.

12001 Technology Dnve Eden Praine. MN 55344 U S.A.

(612) 941 5560 TWX: 4310012 or 431%24 FAX: (612) 828 3088 November 6, 1986 Davi Shih c/o Leon Guaquil New York Power Authority 123 Main Street White Plains, NY 10601

Dear Mr. Shih:

As per our recent telephone conversation I have summarized the trips to the Fitzpatrick plant by Rosemount Personnel:

April 8th 1986 - All day Plant visit by:

Jane Sandstrom Sr. Marketing Engineer, Bob Bach Sr.

Manufacturing Engineer, George,Poole Sales Engineer.

Actions:

Reviewed installation, installation drawings, bench testing by technicians.

May 29, 1986 - All day Plant visit by:

John McMillan Service Representative. Actions:

j reviewed transmitter checks, calibration procedures, error calculations.

)

I checked with Dave Holliday at the plant to see if the reliability study had arrived.

He said that Dick Patch will be forwarding you a copy.

If you have further questions on this or on any other aspects of Rosemount please do not hesitate to call.

Sincerely, Ow &

v

^

Ja e Sandstrom Sr. Marketing Engineer

/jka cc:

Dick Patch - New York Power Authority 7

Dave Holliday - New York Power Authority EXHIBlJf /

'i Pa.es Page o

CASE NO. 4-90-009 D001282 f///

i

Mail Distnbution Sta.

INTRACOMPANY MEMO Deb Enzenauer B5 Date:

November 6, 1986 To:

Harry Savage #4 From:

Jane Sandstrom

Subject:

Please have them complete the first page especially the "Brief description of problem" section.

Detailed symptoms would be best.

We will replace at no charge if symptoms are a loss of oil indication.

I told Bill Loomis on November 5, 1986 we could dar a delivery the week of January 5,1987 and he seemed somewhat content with that.

Give him a bad time about not having more spares. We must have a no-cost P.O. in by November 15 to hold this delivery date!

EXHI Page at /

F%es j

CASE NO. 4-90-009 D001336

<!l 0"

h}

t

N M INC.

12001 West 70ei Street Eden Proene. Wennesons $5344 U.S.A.

Tel-(612)941 5560 TWX 910-576-3103 TELE.X 29-0183 December 22, 1986 Dan Kosak Yankee Rowe Project Yankee Atomic Electric 1691 Worcester Road Framingham, MA 01701

Dear Mr. Kosak:

In our recent conversation we discussed the low-oil failure you recently had at your plant.

I will summarize our information on this situation here.

Rosemount normally sees 6-9 of these failures a year out of the tens of thousands in the installed base of our Nuclear transmitters.

New York Power Authority for some reason saw an unusually large number this past year (7-8).

If their numbers are taken out the actual return rate for low oils over the past 24 months is actually less then normally seen.

The leakage paths are evenly distributed over these four places:

Cracked fill tube Cracked isolator diaphragm weld Glass to fill tube seal Cell cup to glass seal Please refer to the attached diagram (taken from the Operation Section of our Instruction Manual) for details on these locations.

Please note the fill tubes are labeled Leadwires since they perform that function once they are sealed off.

New York Power Authority's failures were spread over all the potential leakage paths and no commonality was found between them (such as similar lot numbers or weld dates).

These different leakage paths are not related either (i.e. welding and glassing are two separate and unrelated manufacturing processes).

Rosemount has taken steps (effective in June 1986) to tighten up our inspection criteria to prevent any further problems.

We feel under our 200% testing and inspection we have eliminated any potential for future occurrences.

77 D001.338 CASENO. 4-90-009 EXHIBlJ d g as es Page i n u g

i Yankee Rowe Project December 22, 1986 Page 2 If you have further questions on this information please feel free to call me.

I will talk to our local Salesperson, Harry Savage (617) 849-0146 to let him know of your interest in seeing the cell manufacture display.

Sincerely, f

^

g___--

pgv Ja e Sandstrom Sr. Marketing Engineer cc:

Harry Savage #4 1

77 EXHIBl 4-90-009 Page of Pa.es CASENO.

D00133S j

l 091776 PACIFIC GAS AND E LE C T RI C C O M PANY l

77 BEALT STREET. SAN F R ANCISCO. C ALIF OR NI A 9J106 * (415) 781-4211

• TWX 910 372 6587

-+--

February 20, 1987 w...e ~,...mouo

%.7.UIE PGandE Lecter No.: 87000361 r

Mr. Richard C. LaSell Product Manager Rosemount, Inc.

12001 West 78th Street Eden Prairie, MN 55344

Dear Mr. LaSell:

Rosemount Model 1153 Series D Transmitter Failures Pacific Gas and Electric has become increasingly concerned with the number of Model 1153 Series D transmitters which have failed at our Diablo Canyon nuclear facilities.

In addition to the two instruments evaluated by Rosemount in July 1986, PGandE has had four other instruments exhibit identical slow or sluggish response to pressure changes.

After contacting other nuclear facilities, it is apparent that PGandE is not the only utility to report similar failures, thus warranting further action by Rosemount to identify suspect

~

sensing modules.

To assist us with evaluating the magnitude of the problem, we request the folloing actions be taken:

1.

Perform a complete failure analysis on PGandE-returned instruments and present an in-depth evaluation detailing failure mechanisms and proposed steps to prevent recurrence.

Considering other facility failures, the evaluation should also detail Rosemount's actions to determine reportablity requirements as outlined under Title 10, Chapter 1, Code of Federal Regulations, Part 21 (10 CFR Part 21), " Reporting of Defects and Noncompliance."

j 2.

Supply PG&E with serial numbers of transmitters and sensing modules susceptible to failure.

Rosemount's manufacturing record review shall cover, as a minimum, the following PGandE purchase orders:

4R-5680B 4R-69502 4R-70544 4R-70744 4R-5680C 4R-67818 4R-70714 PO 653785 In order to proceed with evaluating and monitoring remaining equipment, a reply is requested by March 13, 1987.

Should you have any questions, please contact Mr. Robert C. Washington at (415) 972-7023.

N EXHIBIT Sincerely, p,

/j CASE NO. 4-90-009 D001423

~~

WAR:Imt

/

)

e4; b'

ROSEMOUNT INC.

12001 Technotony Ottve =

Eden Pra.<ie, MN 55344 U S A '

(612) 941 5560 7WJO 4310012 or 4310024 FAX: (612) 828 3068 Rosemount-February 20, 1987 e

Pacific Gas & Electric Company 45 Fremont San Francisco, CA 94106 Attn:

Mr. Ed valeriano Subj:

Rosemount Transmitters

Dear Mr. Valeriano:

Our local field representative (Mr. Bill Gilchrist) indicates that Pacific Gas & Electric has expressed concern regarding " sluggish response" of several Rosemount Model 1153 pressure transmitters at the Diablo Canyon site.

The purpose of this letter is to respond to your concerns and to work with you to a firm resolution of this issue.

" Sluggish response" is a failure symptom caused by various transmitter as well as application problems.

This symptom can be attributed to neveral different failure mechanisms.

Our. concern, as a supplier, is that failures with similar symptoms may be' grouped'together.

This appears to be the case with the symptom " sluggish response".

Our service records show that a total of eight 1153 transmitters have been returned from Diablo Canyon for evaluation.

We understand an additional 2 units are being sent to Rosemount.

Of the eight transmitters evaluated, 5 were attributed to installation and/or handling caused failures and not manufacturing problems.

A brief summary of the evaluation findings is enclosed as for your review.

An extensive evaluation of the remaining 3 units indicated that the cells had lost oil through seepage.

Evaluations performed on failures exhibiting this same symptom showed that four different mechanisms in the cell have caused this mode of failure.

The areas of concern include the glass to metal interfaces within the cell I7 EXHIBIT

"*8'

' N Pa.es CASEHO.

4 00-009 D^0^9P ",

/

/l* l'/4 U

'Q

m

! M.4 e

=

cup, the welds, and the fill tube.

The loss of oil is extremely slow.

It appears to be seepage rather than a

' leak.

It is also interesting that the failures are occuring primarily in pressure range code 5 units.

Consequently, we would appreciate as much service data as you have available on the transmitter that demonstrated the " sluggish response".

While the number of transmitters exhibiting this symptom is very small, we are concerned and we will continue to study our manufacturing processes to determine if further changes (improvements).can be initiated.

About a year ago we intensified our inspection criteria due to heightened awareness of the above listed areas of Concern.

Upon our completion of the evaluations of the two transmitters coming back to Rosemount, we would like to get together with the cognizant individuals at Pacific Gas & Electric to discuss our findings.

We sincerely appreciate your cooperation in working through resolution of this issue.

Sincerely ikVY it R.J andenBoom M r eting Engineer R

emount Nuclear Products Enc:' Attachment 1 cc:

Bill Gilchrist - # 21

. Dick -LaSall~-

- B5d' Don Malone

- Pacific Gas & Electric Bill Reed

1. 35 Bob Volsted

- B5

/kp EXHIBl{ 8 7 Page d

Nees ens.

4 - 9 0 - 0 0 9 D"09M

~

-e t

b NITACHT M 1

^

m E.

I

?.0.

H.O.

flodel s/r; Problem e

o a

2 :540 209L81 1153HD5 L10172 low oil in module.

O i:.R-705LL )

Oo L -56S )

1153HD5 413045 low oil in module.

'i-72553 209494 1153DDL 410729 Punctures in isolatir.c dir.:. :,7 1153DDL 410731 Punctures in isola.ir.g diar 1153DDh 410730 Invalid rejection. could r.c_

duplicate problem.

I;o 1 css cf oil.

LR-71082 208h35 1153HD4 4120L2 Customer caused - defectice dicde problem.

Reversed polari.y.

1153DD3 412269 Customer caused - defecti se di:de problem.

Reversed polari _y.

LR-7054 760200 1153HD5 L13066 low oil in module.

)

$5 o

I 3

W

D

~4

,9. Ds W

a' a

w o n, -- m-. -

x.

f-l ' ?9 -

ROSEMOUNT INC.

12001 Technology Drwe Eden Prairie, MN 55344 U.S A.

(613 S41-55 %

TWX 431C012 or 4310024 j

FAX: f 612) 828-3088 February 25, 1987 Mr. Warren A.

Raymond (Room 1453) r Assistant to Vice President h

Nuclear Power Generation Pacific Gas and Electric Company 77 Beale Street San Francisco, CA 94106

Subject:

Rosemount Model 1153 Series D transmitter failures

Reference:

Your letter 87000361 of February 20, 1987

Dear Sir:

We share your concern for the failure as demonstrated by i

our letter o." 2/20 to Mr. Valeriano which obviously passed in the mail, copy attached.

I will attempt to address your specific concerns. We are aware of the two units you reference in the letter, S/N 410172 and 413045.

We are unsure of the additional four units referenced, we were told that there were two additional units to be sent to us.

To effectively address specific failures of specific units, we need to reference the S/N and date surrounding the failure such as servica conditions and operating life.

i As indicated in our letter of February 20 we have seen a few additional failed transmitters that demonstrate

" sluggish response".

We are unaware of the data you possess from other sites, however our data indicates that while the failure symptom is slow response the failure mechanism's that cause the symptom are multiple.

There are application oriented causes such as plugged lines or snubbers that could cause this phenomena as well as at least 5 identified unique causes within the transmitter.

These are:

Field handling caused damage (i.e. ruptured diaphragms), glass to metal cell cup seal failures, glass to fill tube seal failure, fill tube cracks or breaks, or a leak in the diaphragm seal weld.

We have evaluated failures from all sites and found each of the above failure modes to have occurred.

The total number of i

failures is small (approx. 0.4% confirmed transmitter caused failures in the past 2 years based on shipments of all nuclear qualified transmitters in the same time period.) and is represented by several different mechanisms.

Therefore, while we feel concern if only one EXHIBIT g/P -

i

/ of Pa.es Page CASENO. 4-90-009 n o n. n cu.,e

• /

o @

L

~ c

-,C i

unit fails, we do feel that this failure is of a random nature and is subject to remedy under our normal warranty.

The failures even at this small incident are high and concern is compounded by the fact that a few sites experience higher rates (while most have no failures). Rosemount has therefore had several engineering evaluation meetings to address this issue.

As a result we have tightened our inspection criteria considerably to assure all potential failures are rejected in manufacture and have ongoing engineering evaluation of means of improving the product.

Our evaluation to date of all the data received from all I

sources, does not indicate that the failure mode or modes are tied to a specific manufacture lot or lots.

The data is incomplete as to correlation to a specific t

application.

In addition, while it is easy to detect when the failure is caused by a problem in the impulse

{

line and simple inspection will detect a mishandling caused ruptured isolator diaphragm, a failure of one of the four manufacturing defects cannot be detected without destructive evaluation.

Therefore, we know of no single test or inspection to verify the potential for failure of an existing unit.

As indicated on our previous letter, we plan upon completion of evaluation of the units still in your possession, to visit your facility to discuss the failures, and application data.

To assist us in the evaluation, we would appreciate a run down on the total 3

number of units installed, the service conditions, and operational life.

We would anticipate the meeting within 3 to 4 weeks of receipt of the units.

If we can assist further or there are additional questions prior to the meeting, please call us.

Very truly yours, 0

,d/

R.C.

LaSell Manager

Enclosure:

Letter of 2/20/87 cc:

Bill Reid #35 Bill Gilchrist'#21 87

'"MI EXHIB)IA Di.y Pa.es Page CASENO. 4-90-009

ggggg, i

ROSEMOUNT INC.

12001 Technology Onve Eden Praine. MN 55344 U.S A (612) 941-5560 TWX: 4310012 or 4310024 FAX: (612) 828 3088 Rosemount E61 LUBE _6NeLYSIS Date :

April 6.,

1987 Customer :

Pacific Gas and Electric Model : 1153HDSRC S/N 410033 Problem :

Slow response time EEBE9BMeUGE_EYeLUeIION1_ Nuclear _Iesboicinos The returned unit was mounted on a test fixture and pressure cycled.

The transmitter responded extremely slowly to pressure changes, confirming the customer reported condition.

The transmitter was then disassembled and the sensor module removed.

Inspection of the isolating diaphragms revealed a low oil condition in the low pressure side of the sensor cell.

EeILUBE_eNeLYSIS1_Evaluc11on_ Laboratory The sensor module (S/N 1053052) was dissected and the sensor cell removed.

Oil residue was found on the outside of the cell near the low pressure side fill tube hole.

The cell was then flanged and connected to a pressure source.

High predsbre was applied and the fill tube hole was observed under a microscope.

Oil was seen leaking from the cell out of the fill tube to glass interface.

EeILUBE_GeUSE1 The oil leak path was caused by an inadequate glass to fill tube seal.

EEEeIB_eQIION1 i

Replace sensor module.

d

. Y O!

"'#8%

Robert E.

Reinertsen

/

Nuclear Quality Assurance EXHIBIT i

d---f Page CASE NO. 4-90-009 0001225 l

- b t{

s x&

\\')'f\\V\\

{},6k44-Sp NM INTRACOMPANY MEMO

/'f Date : April 29, 1987 3"

Bob Bach 805 Rob Lind.

BOS To-

Jerry Anderson Dick LaSell 805

=> B e t ti Iwanski C07 From t4 Bob Reinertseni Del Burmeister C15 Subj : Glassing Experiment We ran an experiment today to study the cause of the recent bubble problem in nuclear sensor cells.

The problem is a

" beer foam" continuum of small bubbles starting at the top of the inside cup wall near the fill tub? hole, and extending down the wall and out the fil4 tuta hole.

The experiment consisted of 27 parts divided into 9 groups of 3.

The groups were built per standard procedure except as below:

1) 3 all nuclear production parts without fill tubes installed 2) 3 all with a machining oil wipe over a small portion of the cup bottom 3) 3 all with an extra fill tube laying on the bottom of the cup 4) 3 all with an alkinox wipe over a small portion of the bottom of the cup 5) 3 with 1151 floor stock glass and all else nuclear 6) 3 with 1151 fill tubes and all else nuclear

,,, q '

,9 7) 3 with 1151 cell cups and all else nuclear 8) 3 with 1151 ceramic and all else nuclear 9) 3 normal nuclear u ve r e. i:.t n oosc ro n Prwr s w i'-~- /

d ~ ' "^' " 0 'hM U

e r o ra c i,rs r u r r.+1xc u m ~ G s re e. ' V

' u)

S uuru w,ri.s in c i<

The most notable result is that all of the groups had bubbles on the cup wall near the fill tube hole with the exception of group 7.

UQU:GeUSES_QE_QUBBLES We have examined the cell cups and believe that these bubbles are 091 caused by:

E111_Iubes The bubbles were present in cell cup group 1.

Group 3 did not show bubbles near the extra fill tubes on the bottom of the cup.

Group 6 was the same as the other groups with bubbles.

/Cf EXHIBIT 3

Pa,es D001206 CASDio. 4-90-009 Page _ /_._of e

g%/a jg

/' ', *f.

0 i /1/ V rnem Nn son a,

W G1035 Group 5 did not show a discernible difference from the rest of the groups with bubbles.

CetDmit Group 8 did not show a discernible difference from the rest of the groups with bubbles.

dacbicios_cil_on_1be_ bottom _of_1be_ cell _tue Group 2 did not have any. of these bubbles on the bottom of the cup near the wipe point.

elhicoE_deletsdot_on_1be_ bottom _of_1be_ cell _ cue Group 4 did not have any of these bubbles on the bottom of the cup near the wipe point.

CeUSES The oubbles ate caused by some condition in or on the ni-span.

The experiment indicates that two interrelating factors, glass flow rate and ni-span characteristics, combine to cause the bubbles.

Ilme:Iemeetatute:Geomettr In the cups with bubbles, the places in the cup where they are most common are the places where the glass gets to last in the flowing process.

This means that both the glass and ni-span are at different temperature when they contact at the place with bubbles than at the places without bubbles.

Looking at a cell cup you can see that if there is a large number of these bubbles in the fill tube hole, there is also a higher density on the cup wall above the fill tube hole.

Looking at the same cell cup you can see that when the aforementioned conditions exist there is also a tendency for more of these same bubbles to be located on the upper cup wall in between the positioning points of the glass slugs, with the wall near the slug positioning points being relatively free of bubbles.

Therefore, the points in the cup that tend to have bubbles are exposed to the oven atmosphere longer, are at a different temperature when the cup and glass meet.

Qxide_ Level Group 7, which had en 1151 cup and all else nuclear, was by far the best in the group.

Other than very few bubbles, this group had one other significant difference.

The color of the oxide on the cup was more greenish than the nuclear cups.

Note that these cups were run in the oven in the middle of the batch, within inches of the other parts in the experiment.

CASE NO. 4-00-009 EXHIBIT /C9/

Page b _a_ ?% es D001207

z ma 3900681 4

Although this sample size'is inadequate to " prove" something-statistically, the evidence solidly points toward the ni-span as the' culprit..Something in or on the ni-span is-causing the problem.

All of the other parts and procedures in the glassing area do not seem to affect the problem.

It appears to be an oxide issue.

The fact that the problem is worse-where the glass and ni-span meet last may be an important clue.

This problem is related to cell cup storage, ni-span composition, or some variable in the machine shop cell cup manufacturing process.

-t i

L t

a i

r i

4-90-009*

W lBIT NY CASE NO*

Page 34 3 Pass D001208 i

t

A Mail' D

1. gs,qghrson Boga.

Rosemount. INTRACOMPANY MEMO

2. R. Lind BOS

3. B. ReinertsenB05 4.

D. Zawacki B05 ay 87 Date:

l' i

Dick LaSell g

Bob Bach From:

Subject:

Ac ion on OH Maks Following is a summary of actions currently underway to address the Nuclear sensor oil leak problem.

1.

Stress Analysis I spoke to Ross Hoffmann, who is our Stress Analysis specialist, about performing a study of the 1151/1153 sensor to determine the stresses induced on different ranges as a result of assembly and operation.

This would be a worthwhile effort; the knowledge gained would_ help us in our current effort and in solving future problems as well.

Ross indicated that this would require 2-3 weeks of dedicated engineering work--time which he cannot provide immediately as he is working full time on Mt.

Everest.

He can, however, support such a project if the bulk of the labor is provided by someone else.

Mike Dean, a member of the IT Product Engineering group, has done some strcas analysis work and would be a good candidate for this effort.

I will contact Bob Diesch to establish priority for Mike on this project.

2.

Assembly Stresses As an empirical adjunct to stress analysis, we have initiated the assembly of a group of range 5 sensors into transmitters.

These will then be disassembled and tested for oil leaks in our HP/ Aging test and in life testing.

The purpose of this test is to determine whether assembly stresses do in fact induce leaks in range 5 sensors.

Bob Reinertsen is coordinating this project.

l 3.

Cell Cup Material j

An experiment run by Bob Reinertsen in the glassing ovens on 4/29/87 showed that the bubbles in the glassed cell l

cups are caused by some characteristic of the material or l

processes used to make the Nispan cell cup--not by the glassing process.

Samples of 1151 and 1153 cell cup material are being analyzed by an outside source; results

{01 CASE NO. 4-90-009 E

b a.es P

D001234 d

F/s20 n-C Form No. 60035 S. Rev. 8

~

l 3

1

'Ek (Action on'011 _anko May'1, 1987

,. C Page 2 l

1 are expected next week.

Betti Iwanski.of the IT _

]

Manufacturing Engineering group will pursue this issue.

4.= Life Testing-i the have run. one group of sensors through' pressure cycling on the MTS system in IRD to simulate accelerated field conditions.

The purpose of this' test.is to correlate our-i accept / reject criterion at HP/ Aging test with field performance,_ ensuring that we are in fact. screening out the. sensors most susceptible ~to oil loss..Results of the first group. tested indicated a weak correlation, but the sample size (ten sensors) was too small to draw a i

meaningful conclusion.

Bob Reinertsen is testing i

additional sensors to increase the data base.

i 5.

Bubbles Production of 1153 glassed cell cups has been. stopped'for about a month because of bubbles in the glass.

While I

small numoers of minute bubbles have been accepted in-

,i glassed cell cups before,-production parts. built in the

.{

last month have unacceptable numbers of bubbles.

While.

-l the bubbles are undesirable because'they, indicate poor glass flow, their presence has never been tiedito' oil leaks or other problems.

Betti Iwanski has initiated an i

SFO to build sensors'using cells with large numbers of l

bubbles and test those sensors for. oil leaks and other i

performance characteristics.

We will nat1 accept parts with large numbers of bubbles unless this' test proves that.the bubbles are~not related to loss of performance.

6.

Production-I Recent samples indicate that production yields of Nuclear i

sensors in glassing will be about.50%;'that is, half of the sensors built contain bubbles in quantities that are i

acceptable.

Since a 50% yield is~ preferable to no production at all, production'of Nuclear cell cups is i

being restarted this week.

Until the cause of the i

bubbles is identified and eliminated, the parts will be started more frequently in smaller lots.

Data on the inputs to each lot will be collected and compared against the occurrence of bubbles.

Bob Reinertsen and~ Rob Lind will examine'and sort each lot of parts as it comes out i

of the glassing oven.

s I

(07 cam. 4-90-009'

(,

d A Pa.n D001235 i

n,,

Betti-Iwanski C07 M

INTRACOMPANY MEMO Del Burmeister CIS Bob Bach 805 Date : May 5, 1987 Rob Lind BOS Dick LaSell B05 To

Jerry Anderson Sue Bjerke C10 From : Bob Reinertsen

\\.

Subj : Huntington and Carpenter Nispan A glassing experiment run last week indicated that the recent bubble problem in nuclear cell halves is related to some condition in or on the nispan.

This prompted an investigation into the elemental composition of the " good" cups and the " bad" cups.

The good cups were 1151 production parts from Carpenter heat 97640.

The bad parts were 1153 cups from Huntington Heat HH2848DK12.

A list of the element percentages for both of the heats used in the experiment is attached, as well as other heats used in production recently (from National Spectrographic Laboratories).

A notable difference in the the composition between the test parts is the lower Mn content of the 1153 parts.

Another difference is the lower sulpher content of the 1153 parts.

Huntington parts seem to be consistently different from Carpenter parts on these parameters.

These factors may or may not mean anything, but some other as yet unmeasured characteristic may be as consistently associated with the Huntington parts.

One thing that is known for sure is that the recent problems with nuclear cell halves, both at glassing, pressure cart, and at nuclear HP/ Aging, came at the same time as the Huntington parts.

Eat _about_a_ rent _etiot_to_tbis_we_wete_

UEiD9_CateeQ1SC_CQll_ Cues.

Huntington heat HH3076DK was used to make the nuclear Lot 21 cell cups.

These were Dred for an oxide discoloration, but were signed off UAI.

These same cell cups had a high failure rate at pressure cart, and had a 30% to 80% fallout at nuclear HP and aging WW 23, 24, and 25.

Cell cup lots 23 (1000 parts) and 26 (750 parts) were Huntington lots, and were the lots scrapped for bubbles.

Lots 24 and 25 were Carpenter lots and were good.

While these correlations are sketchy and need more proving out, in the short term we may want to use only Carpenter Nispan on nuclear cell cups.

This should be a no-risk action, but with some potential benefits in quality and yield rates.

HD EXHIBIT/ of /

Pa.es D001233 CASE N0.

4-90-009 Page

,.._... -.~

t f/') l i

j'

\\

6.

ROSEMOUNT INC.

12001 Technology Onve Eden Praene. MN 55344 U S A (612) 941 5560 TWX 4310012 or 4310024 s.

F A X' (612) 828 3088 June 2, 1987 GPU - Nuclear Corporation Post Office Box 480 Route 441 South Middletown, PA 17057-0191 Attention:

V.P.

Orlandi - Lead I & C Engineer

Subject:

1153 Transmitter Failures at TMI - Unit 1

Reference:

GPU Letter No. 3340-86-0015 Gentlemen:

Thank you for the supporting information and follow-up regarding the four Model 1153 failures that have occurred over the past 2 years at TMI Unit 1.

The information that you provided has been extremely valuable in helping us to evaluate root causes for the off scale failures on your Steam Generator Pressure application.

The primary purpose of this letter is to address your concerns as communicated in the above referenced letter and to inform you that corrective actiQas have been implemented at Rosemount to eliminate the potential for similar failures in the future.

In accordance with your request, we have performed a failure analysis on the transmitters returned to Rosemount.

Copies of the failure analysis are included herein.

At the time of the initial failure on your MU-42-DPT tag location, the failure characterized by sluggish response was considered rare.

The fact that you were initially charged for our manufacturing rclated problem was due to internal miscommunications at Rosemount.

We have taken actions to correct this situation.

We do, however, apologize for any inconvenience that may have resulted.

Some brief information regarding both failure types described in your letter to Rosemount referenced above is included on Attachments 1 and 2.

We hope that this information will satisfy your requirements.

/

EXHIBIT 4

a,es Page

/d Pe 4;-90-009 D001141 I

(

k,

, c.

GPU Nuclear June 2, 1987 Page 2 We value your support for our products and we hope that you will continue to maintain your confidence in Rosemount instrumentation.

If you have any additional questions or concerns regarding our products, please do not hesitate to contact me at (612),828-3540.

Regards b

.'VandenBoom

\\.

R keting Engineer emednt Nuclear Products

Enclosure:

& 2 Failure Analysis Report cc:

Dick LaSell BOS Tim Jenneke BOS Joe Steelman #17 Steve Siegrist H.O.

209648-P.O.

TP-047395 i

a O

D001142

s

. e.

Off Scale Failure " SYNDROME" - (PT-950, 951)

The intermittent offscale failures are caused by a combination of both the application and the transmitter.

The most prevalent application parameters that trigger this failure mode include vibration,. process noise, or a combination of both, acting simultaneously.

Recent testing on suspect units has confirmed that extremely high frequency vibration will accelerate the failure mechanisms and may consequently result in an offscale indication.

These high frequency vibration levels are well beyond anything that has been specified to-us and beyond any level previously tested.

The transmitter failure mechanism requires microscopic conductive particles in the sensor module fill fluid to act as a short circuit between the center diaphragm of the transmitter and a capacitor plate on either the high or low side of the sensor cell.

In the past year, Rosemount has implemented several process changes to reduce and/or eliminate any and all potential sources of conductive particles in the sensor cell.

As a result of our efforts, newer units subjected to the same high frequency vibration levels that had caused suspect units to fail did not fail. These process changes include:

Ultraviolet ozone cleaning, Electro-Chemical deburring, 100%

i Microscopic inspection of all center diaphragms, high pressure washes, etc.

I Since the offscale " Syndrome" has affected less than 1% of our installed base and specific tag locations seem to have a much higher susceptibility to a repeat failure, it is j

apparent that only a few tag locations are suspect.

i Rosemount has, and will continue to replace all units exhibiting this, failure syndrome regardless of warranty.

Never units placed in service at suspect tag locations have not experienced this failure " mode" and we are confident that we have corrected the problem regardless of the application dependency.

We would ask that the recently shipped replacement i

transmitters be placed in service in the tag location which j

had previously experienced a problem.

]

EXHIB

/

""8' CASENO. 4-90-009 D001143 i

wt

(

(

Low-Oil Failures - (MU-42-DPT)

Approximately one year ago, our review of our field return evaluations showed a higher than normal incidence of failure due to " Sluggish Response".

This review initiated a study into the problem where we discovered an inspection process that could potentially allow a few units to pass with extremely slow leak rates.

Over time, these units could eventually seep oil'into the sensor module housing and consequently the transmitter would not respond adequately to changes in the input pressure.

Corrective action.in the form of a revised inspection procedure was immediately. implemented for all production units.

All units built and shipped under this revised procedure and placed in service in applications that had previously experienced this failure have not failed.

1 In order to reduce our scrap rate and to improve the 1

reliability of the transmitter, we have also continued our study of the cell as it relates to this problem to identify j

any further corrective actions that may be beneficial.

In the meantime, the few units in the field that remain susceptible to this problem will be replaced, free of I l charge, regardless of warranty, if the units are returned to Rosemount and our evaluation confirms this failure mode.

k 4-90-009 '

EXHIBIT //7 M of Pases CASEHO.

Page D001144