ML20009G798

Safety Evaluation Re Resistance Temp Detector Response Time Characteristics,Based on NRC Review of Technical Repts.Loop Current Step Response Methods Are Acceptable

ML20009G798
Person / Time
Site:	Maine Yankee
Issue date:	11/30/1980
From:	Office of Nuclear Reactor Regulation
To:
Shared Package
ML20009G797	List: ML20009G796 ML20009G798
References
NUDOCS 8108040682
Download: ML20009G798 (45)

Text

- _ _ _ _ _ _ _ _

s' I

REVIEW OF RESISTANCE TEMPERATURE DETECTOR

[1.[RESPONSECHARACTERISTICS

[

SAFETY EVALUATICN BY U.S. NUCLEAR REGULATORY COMMISSION OFFICE OF NUCLEAR REACTOR REGutATICs DIVISI0tl CF SYSTEMS fitTEGRAT!CN

[NST.'.L."ENTATION AND CCNTROL SYSTE:iS 3 RANCH NovEMBEa 1980 b

8108040602 010724 genAcacxosooopg;,

5 4

AtSTRAC{

Historically Resistance Temperature Detector (RTO) time responses have been measured by the plunge test technique. For RTDs inst:11ed in nuclear plants the plunge test is inconvenient and very inaccurate sanetimes leading to errors as large as a factor of 3.

Recently EPRI has developed an in-situ method for measuring the RTD. time response called the Loop Current Step Response (LCSR) method. The LCSR method is convenient to perform and it produces results that are accurate to within about 10%. In addition. EPRI has developed two other in-situ methods which detett RTD degradation, but give no detailed information on the RTD time response. These methods are the Self Heating Index (SHI) method and the Noise Analysis (NA) method. We have examined the LCSR, SHI, and NA methouologies and find all three to be viable methods for monitoring RTD time response, cut we nave not :anducted a formal review of the SHI and NA metnoos. To date two vender time res;cnse topical reports have been submitted to the NRC one from Analysis and Measurement Services Corporation (AMS) and the g:her from Technology for Energy Corporation (TEC). Both vendor topicais propose only the use of the LCSR method. We have reviewed both the AMS and TEC LCSR topicals and find their methodologies acceptable for RTD time response measurement.

The extensive RTD testing done in conjunction with the LCSR development has revealed RTD time response degradation with ageing. In view of this degradation we a. e recommending increased surveillance testing of RTD time response.

t

- 1. -

_T_At_t_l _E o,f C.O_NI. I N.T,S

=

SfCTION SfCTION TillE NilHRER PAGE

. ! N1 R000CT 10N BACKGROUND. AND S UMMARY ----- ------------ --------------------- --------- 1.0 -------------- 4 RID TIME RESPONSE Cl!ARACTERIZATION AND HEASUREMENT

---

--------------------------- 2.0 -------------- 11 RID T l HE CON S T AN T CONC E P T ---------- -- ----- -- - - - - -- - - -- -- - -- --- - - - - - -- -- ------- --- 2.1 -- ----------- 1 1 LCSR HETil0D FOR HEASURING RID TlHE CONSTANT -------------------------------------- 2.2 ------------- 12 L C S R T E S T PR OC E DUR E - -- - - - - -- - - - - - - - - -- - - - - -- - - - - -- - - - - - - - - - - - - - - --- - - - - - - - - - - 2. 2.1 - - - -- - - - - - - 12 l ilE L C S R T RAN S FORMAT I ON -- --- - --- - --- --- - ---- - - - - - -- - - - -- - - - -- --- ----- -- - ---- - 2. 2. 2 ---------- -- 12 APPL I C AT I ON OF Tile LCSR T RANSIORMAT I ON --------------------------------------- 2.2. 3 ------------ 14 l

DE. MONSTRATION OF CONSERVATISH OF IllE LCSR TRAliSI ORMATION --------------------- 2.2.4 ------------ 14 n.

EPRI (AMS) HETil0D FOR CORRECTING FOR UNKil0WN lilGilER EIGENVALUES -------------- 2.2.5 ------------ 15 l

TEC HElliOD FOR CURRECTING FOR UNKNOWN lilGilER EIGEtiVAL UES --------------------- 2.2.6 ------------ 17 RI D D E GR A DA T I O N T E ST S -- --------- - - - - - -- -- - - - -- - - - - -- - - - -- - -- -- ---- - - ----- --- ------ -- 3. 0 - ---------- -- 18 RT D DEGRADAT I ON T EST S US I NG L CSR HElll0 D ------------------------------------------ 3.1 -------------- 18 RID DEGRADATION TEST S USING Tile SELF llEAT I NG Inut X (Sill ) ------------------------- 3.2 -------------- 22 RI D DEGRADAT I ON TES TS USI NG NO I SE ANALY S I S ( NA ) ---------------------------------- 3. 3 -------------- 25 PO T E NT I AL FOR RI D T lHE RESPONS E D E GRADA T I ON - ---------------------------------------- 4.0 -------------- 2 7 H0 DES Of RID TIME RESPONSE DEGRADATION ------------------------------------------- 4.1 -------------- 27 EVI DENCE OF RTD T lHE RESPONSE DEGRADAT ION ---------------------------------------- 4.2 -------------- 28 I

e

I a.

t I

e e

1 I

i e

s==

• tf*

W M

CIS m

C alf" k

M M

M M

M M

M

• ir air I

I I

6 8

8 I

I e

8 I

i 8

8 9

0 0

8 8

8 8

8 8

I t

9 I

4 1

0 0

t i

I f

8 I

8 0

I I

I I

i 8

8 I

8 9

I i

I 8

0 8

0 I

I I

a O

I I

I I

e i

I I

i 1

8 8

I I

I I

I f

8 I

8 8

I I

I i

8 8

8 I

O 9

I 9

I I

I I

e i

I I

e i

i I

t i

i I

I I

B I

I I

t I

8 C.

C.

C.

C.

ZM' C.

m.

N.

M.

4F CiW e

e== CC M

M M

M M

N CC C'l

>== {

U 8

I e

3 I

I I

t i

6 Z

I I

I I

I I

I I

8 M

I i

8 9

1 8

8 I

6 I

I I

I 8

8 8

e i

B I

8 8

1 0

I 8

I I

8 I

I I

I I

i 1

8 I

I I

8 I

B 8

I 8

8 I

t i

I I

I I

I B

4 I

0 8

8 I

I I

I I

e t

8 8

0 8

I e

8 I

I B

I 6

0 0

I B

I I

i B

I I

e t

8 I

I I

e 8

i e

I 1

I I

e 8

8 8

I I

I I

I I

I I

e I

I 8

8 I

I e

I e

a I

I I

I I

I I

e I

I 8

1 I

I e

I I

4 3

I 3

p 3

I e

1 I

I i

9 8

0 i

8 i

I I

t 8

9

^

I e

I I

I I

8 I

I C]

I I

I I

I 8

8 i

B LI 8

I a

3 8

I B

I 8

08 0

I I

i t

i I

I I

Z$

1 I

i 3

8 0

8 0

4 1

9 I

I e

I e

I I

B I

I I

I I

I I

e Z

e I

• • 8

• • I

• • 8 4

I i

i O.

I I

W 8 W t L.J 8 8

I l

4 Ut I

I U I U I U I I

I I

I

%e I

t

== I

== I

== 8 I

I I

d I

1

• == 4 h==

I

=== 8 4

I I

i M(

l 8

G B C I U l i

I I

e

= = =.

I a

a:l' e aC I aug e I

I t

M' O

8 I

% 8

% I

% I l

I I

I e

L L

L s a

a s

.i

.I

.I

.e 8

I e

Zi i

I Z l Z I Z l f

8 8

i C4 8

8

== 0

== 4

==

s 8

I I

f il I

I 6

8 M

4 8

8 I

a C e C CM s

I e

I L it a

W W I Wem W >=

t i

I e

C lj i

M M

MM MM i

i I

I I

Z CM C >== CW I

8 a

I W

e C

Z M

p=

I I

e a

8 L

MC M L.J M

I I

B B

CC 0

M Z==* ZM ZM I

I i

8 j

eC 0

W C >= C CW G

I I

I

>=[

I 3

= = = = === W== c 1

I e

I e

>= C >== C >= 4 9

4 I

i 8

W

== 3=== W== C i

I I

i E

CU Cw CW t

M I

e s

==

ZU Z Zw 4

>==

I 8

8

>=

C CM C I

M I

I U 3=

W >= UM I

W I

I M

C M

>=

8

>=

8 I

>==

CC CW CM i

I J

LN Z >= Z >= ZW W

C 8

8 C

== >

U I

t M

r=

>= 5 WZ >=

Z

• =*

1 i

W U

MC MC ME W

W 8

I 3

W W cc W== WC

=

I e

f

.w w ele *,

P= >= > = = = =

2 2

I 8

>==

1..

a

==

>=

W

=--

3 I

W W

4 W

WC W===

1 I

I W

MW MZ MC

>C 2

I

>=

>=

W Z Qll ZC ZZ W

C Z

0 cm ou CC o

5 a-i

> =.

W r~

c=

c= v C

M M

M ale l' WW M d

M

>==

I Z

Z W3 WW WW W

Z

==

I C

M I

C C

A XW 3m M V.=

==

==.

A 3

C e

3m

=

W

==

a

>=

M W

W M W C2: W llun

>=

a I

U W

zW zW m

O

.c W

g

= >

=M

.-= W W

>=

M M

>=

>==

>= M

>==

3 W

W

==

W C

U Z

M C

C% CU C

M

>=

Z i

aC

+==

>= G >== W Z

W aC W

i

-=

I

>=

4

%3 5W % >=

diC W

3 C

W C

M U

f3 W

I-=

E C2:

W W

K

• W.

Z 3

3 3-j

1.0 INTRODUCTION

, BACKGROUND, AND

SUMMARY

mus -

zums A Resistance Temperature Detector (RTD) is a type of thermometer in which the temperature in inferred fron. the electrical resistance of a piece of wire, wnich is called the elenent. RTDs are used extensively for monitoring water temperatures in nuclear reactor plants. The RTD elenent does not respond instantaneously to changes in water temperature, but rather there is a time delay before the element senses tt.e temoerature enange, and in nuclear reactors this delay must be factored into the computation of safety setpoints. For this reason it is necessary to have an accurate description of the RTD time response. This Safety Evaluation (SE) is a review of the currant state of the art of describing and measuring this time response.

Historically the 0 0 time resconse has been characterized by a single parameter

~ ~ ~

called the plunge time cons ant, or simply One Plunge

. The Plunge : is defined as the time recuirec for the RTO :: acnieve 53.2% of its final response after a ste: ts :erature tnange is imoressed On the surf ace of the RTD.

Sucn a tem =erature change can ce achieved by plunging the RTD into a heat sink, such as water, oil, sand, or molten metal. When ; is measured by this means the technique is called the plunge test metnod.

Until 1977 all testing of RTD time response was performed by means of the plunge test technicie.

In nuclear reactors, surveillance testing posed an in-Convenience in that the RTD had to be removed frem the reactor coolant pioing and shipped to a laboratory for testing. Nuclear reactor service conditions of 2235 psig and 540 CEGF are difficult to reprcduce in the laceratory, and hence all laboratory tests were performed at more benign condittons, and One -laboratory results were then extrapolated to service c:nditions. The comoination of manipulating the RTD and extrapolating the,

m

-t y

e v

r------=-

t-e we-t t+-+

v t-- v-iw

.h

-T--

-*M

laboratory results to service conditions lead to significant errors in the RTD time response, sometimes. ar high as a factor of 3.

Thus there was considerable incentive to find a better way to measure an RTD's time response.

With this impetus, in 1976 EPRI launched a research project at the University of Tennessee (U of T) to investigate other possible methods for measuring an RTD's time response. Two requirements for any method being developed were:

(1) tnat it could be performed in-situ, and (2) that it produce reasonably accurate results. The products of this investigation are described in three EPRI topical recasts, which are references 1, 2 and 3, wnich will henceforth be referred to as tne 1977, the 1978, and the 1980 EPRI topical reports.

This investigation produced three in-situ methods for testing the time response of RTDs, which are as follows:

1.

Looo Current Step Response (LCSR) Method.

In the LCSR Method the resistance element of the RTD is heated by in electric current, and the temperature transient in the element is recorded.

From this transient the response of the RTD to changes in external tenperature is inferred.

l 2.

Self Heating Index (SHI) Method.

In the SHI method a constant current is impressed through the element and the equilibrium change in resistance is recorded. The ratio of the elenent resistance change to the power dissipated is called the SH!. The SHI cannot be correlated with the Plunge r, but changes in the RTD SHI can be used as a means of detecting RTD degradation.

l

3.

Noise Analysis (NA) Method.

In the NA method the small fluctuations in RTD output under operating conditions are analyzed on line (or recorded for off'line analysis) using spectral density and/or auto regressive techniques. These fluctuations are the RTD response to fluctuations in the external temperature of the RTD.

If the pattern of fluctuations in the external tempert ture is known, then it is possible to deduce infornatica % cut the time response of the RTD. The NA method has been applied to obtain consistent results under optimum reactor conditions.for certain types of sensors; however, currently it has not been established in a statistically dependable manner that the NA method yields results comparable with deterministic methods. Thus. while in principle it should be possible to develop a viable deterministic method for measuring the Flunge : using NA, the realizationmf thts goahrM1 still require a-substantial amount of investigative work. However, at the present state-of-the-art the NA method could be useful for detecting RTD time response degradation.

Charactertstics of these-three in-situ methods and the plunge test method are summarized in tables 1.1,1.2 and 1.3.

All these methods have their purpose. However, for determining the RTD Plunge, the only currently viable method is the LCSR method.

Currently in-situ LCSR RTD measurement services and test equipment are available from two venders Analysis and Measurement Services Corporation (AMS) and Technology for Energy. Corporation (TEC). Both these vendors began operations before the final phases of the EPRI study were complete, and as a result developed somewhat different methodologies. The AMS methodology is identical to that described in the EPRI topicals. We have rev1ewed both the AMS and TEC LCSR methcdologies and find them both to be reliable and adequate to measure the RTD time constant to within 10"..

.s.

Table 1.1 Characteristics of Methods for Measuring RTD Time Response

., y Necessary Complexity

]!

Test f

Quality of Heasureinevit a

D Measur einen t

• {

of service n

Need to plunge test measures plun< e i directly, but measurement has poor quality for two reasons: 1) %nipulating 1 TID may change its time Plunge S

response and (2) Service conditions are usually not reproduced in the a

es 0 and Test lab. Lab results must be extrapolated to service conditions. The sh t

combined eilect of these two factors can result in errors up to a factor of 3.

i Test simple.

LCSR Yes Special test LCSR provides asi leidirect measure of 1.

Test e"

'd Results are generally accurate to within 10%.

e Test simple.

Sill can be measured quite accuratesy.

Uses simple

• • • "d #

I" 'I'd"9'S I"

I"

d"9" T t Yes electronic i5 test No good (orrelation between Plunge i and Sill exists, equipnen t.

A good deal of sophisticated work has gone into NA. Ilowever, NA Test simple.

measurements ut plunge 1 conducted to date have been in error by up 9

to a factor el S.

NA 1

P"'I' A siisaber of liivestigators are still endeavoring to No develop a vial le method for measurleig the Plusige i using NA, and it is Test

'?

Lest hoped that future work will lead to much improved agreement between 5

']" E theory arid experiniesit.

NA is still a useful tool for detecting HID degradation.

e

Table 1.2 Pract scal Aspects d Avallibility of RTD Time Response Testina Hethods p

R sit utility of AMS TEC Test 3

Test Proceduee Provides Provides Provide Test for RTD Yes Yes None Degradation Plunge Test i

Measure Poor -- Errors to Service only Plunge 1 a factor of 3 (Lab Tests)

Ye5 Y'S OK -- llowever if the utility buys equipnent for Equipnent Equipment Test for RID degradation test they might egradation and Training and Training LCSR as well buy equipuent for Test.

measuring Plunge i.

Service or Service or Measure Good Equipuent Equipment Plunge i 101 Accuracy and Training and Training Test for RTD Good -- No special test Training Training Degradation equipnent needed.

Sgg Test Measura Poor -- No good Plunge i correlation with i exists.

Good.

Eclutpnent Equipment Test for RID Need Special Test Equipment.

Degradation /

RfD need not be Tm

's~n

/

taken out of service.

NA initial attempts to measure plunge r produced poor results with errors up tu a factor of S.

Test Equipuent Measure Over a period of 2 years a limited number of and Plunge T careful NA measurements have produced results Training with 110% variatlosi. No systematic correlation of these results with deterministic measurements has been made.

1

Table 1.3 Modes of RTD Surveillance Testing 1.

Historical Method: Plunge Test.

Because of the inconvenience of removing the RTD for testing and the inaccuracy of the test results this method is being abandoned by a number of utilities. The NRC should take steps to encourage all utilities to abandon this method in a timely fashion.

2.

LCSR Method: Maximum utility Involvement.

The utility can purchase their own electronic equipment and have their own trained personnel perform the LCSR tests.

3.

LCSR Method: Moderate Utility Involvement.

The utility personnel can do regular degradation tests using either the SHI or NA methods.

If evidence of RID degracaticn is found then a consultant can be brought in to measure tne RTD time constant using the LCSR method.

l 4.

LCSR Method: Miniram Utility Involvement.

The utility can have the consultants measure the RTD time constants on their regular surveillance schedule.

I i

l,

I r

The current Stir:ard Technical Specifications require that one quarter of the safety channel RTDs be tested once every 18 ronths. The data on RT9 degradation collected to date is rather scant, but does appear to give positive evidence of RTD time constant degradation with service. A prudent interim regulatory position would be to require the utilities to either:

a. Perform a surveillance test of all their safety channel RTCs at least once every 18 months, and verify that the time response of the slowest RTD is at least as fast as that assumed in the safety analysis.

In additica perform a test of each newly installad RTD at operating conditions as soon as practical after fts installatten.

b. Continue with the present RTD surveillar.ce requirements and schedules in the Technical Saecifications, but in the safety analysis assume an RTD time constant ecual to the greater of:

I'ongest time constant measured in last surveillance tesi' I ',

• U"

[includinga10%allcwanceformeasurementuncertainty)_.

CE ----- Rosemont Model 104 RT3 ------ 12 sec.

W ------ Rosemont Model 176 RTD ----- 0.8 sec.

S&W ---- Rosemont Model 177 RTD ------ 12 sec.

The rationale for options (a) and (b) above are discussed in section 8.0 of this report.

• 2.0 RTO TIME RESPONSE CHARACTERIZATION AND MEASUREMENT 2.1 RTO TIME CONSTANT CONCEpi If an RTD were a first order system, the Laplace Transform of the sensing element's response to an external temperature change would be:

T(element) 1 3

T(external)

(1 + :s)

The response [T(element)] to a step function change in T(external) is T(element)

T(ext. final) - [T(ext final) - T(ext.initia')]

• exp(-t/:)

=

At ti.1e t=:

the element temperature has reached 100%/c = 53.2% of its final responst Fcr this reason the time required for tne RTD cut ut to attain 63.2% of its final res:ense has been namec 7e RTJ plunge time constant.

In fact, RTDs are not first order systems, but the historical definition of RTD time constant is still used and is still a useful concept.

In applications in nuclear plants the external temperature changes to an RTD are typically ramp functions, and the parameter of importance is the time by which the sensing element temperature lags the extern.1 RTD temperature.

This time is called the Ramo Delay Time (RDT). In the AMS Topical Report (Reference 5) pages 105-109 the relationship between the plunge : and the ROT is discussed, and it is shown that the plunge : is a* ways equal to or longer than the RDT, the maximum difference being about 2%. Thus the plunge : can -

be used as a conservative measure of the RDT, and in practice all Technical Specifications are written in terms of the Plunge : and herte all mearar; ment techniques are directed toward evaluating the Plunge t.

2.2 LCSR HETH00 FOR MEASURING RTO TIME CONSTANT 2.2.1 LCSR TEST PROCEDURE In the LCSR method a constant current is impressed on the RTD sensing element which heats the element and the whole of the RTD experiences a temperature transient. A time plot of either the hes!.ing of the element while the current is impressed or One cooling after the current is discontinued is recorded.

Frem this plot the RTD plunge time constant is inferred by means of the LCSR transformation, wnich is described in the next section.

The element tanperature is inferred # rom its electrical resistance wnich is measures by a bridge circuit. The required electronic test equipment is discussed in detail in the subject references, and this discussion will not be reiterated in this SE.

2.2.2 THE LCSR TRANSFORMATION The aiathematical theory for analyzing heat transfer in an RTD is developed in the subject references. Two different approaches are described in detail:

(1) a nodal approa:5 and (2) a continuum approacn. In the 1980 E?RI Topical Repcet, page 3-34 and Appendix B, nwnerical results of the two approaches are ca.: pared, and for the two cases cited the differencer are 1.5% and 1.1%

respectively. Thus for practical purposes the two approaches can be considered to be identical. -

It is shown that if:

(1) The RTD has cylindrical symmetry and (2) There is neglegible heat capacity inside the sensing element then the transfer fur.ction which describes the RTO's response to an external tenperature change is (AMS Topicai page 23)

T(el ement) 1 (2.1 )

T(external)

(::s - 1)(::s + 1)( 3s + 1 )..... (:n3

• I) n is finite if the nadal approach is used and infinite if the continuum approach is used. This difference is not significant in that the higher order factors contribute little to the solution.

The important feature of the above equation is ina the transfer func; ion contains poles, but no zeroes. As will soon bec:me evident, this fac: permits One inference of an RTO's response to an external temperature enange from

ne results of an LCSR :ransien:.

It is shown tha: the plunge time constant is given by (AMS Topical page 27)

(2.2)

=

3 :1) - In(1 - ru/ t).... 3

(1 - In(1 - ::/: ) - In(1 - : /

It is shown that the response of an RTD to a step change in element current (LCSR transient) is given by (1973 EPRI Topical page 49)

_I an **9("D#In)

(2.3)

T(elenent) - T

=

o n

n.also defined in page 49 of the 1978 EPRI Topical) are functions of T

where the a the poles and :eroes of the tra,sfer function.

• y n

e

Experimentally, the i can be determined by breaking the temperature response n

into a series of expchantials. Once the :, are determined they can be plugged into equation 2.2 to determine, the plunge time constant. Thus all A

the information required to evaluate the plunge time constant is contained in the LCSR transient.

2.2.3 APPLICATION OF THE LCSR TRANSFORMATION In an ideal world the LCSR transfomation could be used as follows:

(1) Conduct an LCSR test to obtain a plot of T(element).

(2) Resolve this plot into a series of exponentials according to equation (2.3). This give: numerical values fer the 1 9

[It is not necessary to evaluate the a ]

g (3) Plug these values of :9 into equation (2.2) to cb: sin the plunge :.

In practice step 2 is performed ei-her by exponential stripoing er a least squares fit. Using either method it is usually possible to find :: and ::.

In exceptionally good cases it is possible to find ::, :: and :3, and in bad cases it is possible to only find ::. If equation (2.2) is truncated after the : /: term the result can be nonconservative by as much as 20%, and if equation (2.2) is truncated to : = :: the result can be nonconservative by as much as 47%. AMS and TEC correct for this problem in different ways, which will be discussed in sections 2.2.5 and 2.2.6.

2.2.4 DEMONSTRATION OF CONSERVATISM OF THE LCSR TRANSFORMATION In reference 4 it is shcwn that if either the assumption of cylindrical e.

v symmetry is violated (say by

• crack in the RTD) of the assumption of having t\\

no heat capacity within the element is violated, then the transfer function l

(equation 2.1) would have zeroes as well as poles. If this were the case,.

l 1

i

~^

~_

'~

1

then the Plunge ; expression (equation 2.2) would contain tems with these poles. It is sanwn in reference 10 that these terms would decrease the computed value of :. and hence applying the LCSR method when the two assumptions for the LCSR mathematical develor. ment are violated leads to a conservative computed value of the Plunge :.

2.2.5 EPRI (3) METHOD FOR CORRECTING FOR UNKNOWN HIGHER EIGENVALUES After trying a number of correlation schames, the U of T investigators found that a very good approximation for the Plunge : is given by (2.4)

Plunge f(T2 ?1)

• TICI - In(I - T2/ 1)3 >

/

=-

I wnere f( ;/:t) is given by the emperical relationship of figure 2.1.

Figure 2.1 was constructed by mathematically c mputing the Plunge : (equation 2.2) and 2 :1)] for a number of different hypothetical RTDs and plotting

1[1 - In(1 - : /

the ratio of the two. The hypothetical RTDs had a variety of sized and gecmetries, which included both hollow core and central element RTDs. Thus the curve of figure 2.1 applies to any RTD which fulfills the two requirements of section 2.2.2.

The fact that this large variety of RTDs all enjoy the same f( 2/?1) is, on the surface, rather amazing. With such a good correlation, one would naturally be inclined to search for an underlying physical reason for all RTDs to display the same f(:2/T1). However, to date this underlying i

physical relationship has eluded us.

l l

i ;

9 9

9 O

O N

I 8

.I O' N...-#

5 6

m...

t e

e

.g

.g.g gg mgg

.M.

a W

N s=.

g t

e c e

a v n

M.

s'.=3 p

g y

.s p

.s

.- i.

g g e

,,. g c

l O. n g,,

4mt

=0 m

y, o 5 g 3

...M.,e-e T..

.. 4

+

p g

M p

,...M..M.

m.M m.......'

M.

4 ea.

,N ij m

g

• 3

_s s e.

o o

....... g.._...

2l' t.J e

1

• • t.

p, L......

g 3

p.

.c..

m a

e

.c

= = *

• u g

1. =.

e.

o y

.......i e

W

.4 4 6id

.6.

i.............. -

J t

C g.

p

-m..

O

--y M

C. '.

.a

=

g

.T p

M

.N m

O.

1 in...,g n,. m.

.=

g -. g N

).

O O.

o p.

.d O

w f"*

• J U

gg.

.gi.

1j

=

,o 4

1.

Y d

p m'**

4--.

m1

.J W

-Q n

l.

,.....s-l-

m p.,

g..

N m

4

+

• "1 l..

l e

M.a gy e.43 c.

g 9.4 i.,,

a i

6.

g

c

=

.c,,

4.'

m y

N

....*1.....

m g

• J O

O O,

~

a, y e,......

.a.

o

.... lc.

a O

g..

3

.L....-... +

f)

y g

J b

u's

. ~...

6 6-y 6

C g

a.

s...

W g

g o

,,,m,

.... as..

g m

A, O

W 3

a

.M.

,c.

o:s gg 6

....i..

w....g_....

• 3

=.

Q U

U-

... 1....

..........o y

a N

3

... _...... t 1,4 }..=

..=. ~...

.... i g

3 s

y c

w

-5 j

i 6

e a

i e

6 b

O c3 u3 as" N

O 33 43 58' N

O 3

"8 C

N.

==.

C.

C.

C.

C.

C.

X e.=

s=

6 b"d P'"

16 -

e w

g-

.v-v

2.2.6 TEC METHOD FOR CORRECTING FOR UNKNOWN HIGHER EIGHEVAt.UES The method used by TEC is the following:

(1) Assume a continuum =cdel for the RTD which geometrically consists of a themowel1 (pipe which houses the RTD), and air gap, a steel sheath, a ceramic layer, a platinum elemenc and a ceramic core.

(2) Assume realistic vdhes for the thermal properties of the thernewell and the RTD steel sheath.

(the element is so small that it can be ignored in the themal calculation)

[

(3) The thermal resistance the film between the thermowell and water and that of th'e air gap between the thermcwell and the sheath are not well known. These two thermal resistances are ccmbined into a single resistance R(film + gap) which is left unkncwn. The thermal resistance of the ceramic R(ceramic) is also left unknown.

(4) The RTD continuum equations are solved for t and :: using various values of R(film + gap) and R(ceramic). This procedure is iterated until the

~

values derived for :: ano :2 match those measured experimentally.

(5) The now kncwn values of R(film + gap) and R(ceramic) are used in the RTD continuum equation and the plunge is c =puted.

The TEC method has the advantage over the EPRI (AMS) method that it uses a recogni:able line of physical reasoning to att:in its result, whereas the EPRI method is emperical. The TEC method has *he disadvantage that it requires a detailed knowledge of the geometry of the RTD, which is not needed for the EPRI method. Hcwever both the EPRI and the TEC method produce about equally l

accurate results, and thus from a regulatory point of view must be considered l

l equally good.

i l

i I

l *

=

-.e

3.0 RTD DEG9ADATION TESTS

===a Although neither AMS or TEC have presented proposals to do degradation tests, the subject of degradation tests is discussed in the EPRI reports, and it seems worthwhile to summarize the status of these degradation tests here.

3.1 RTD DEGRADATION TESTS USING LCSR METHOD A simple application of tr.e LCSR method is a degradation test. For this test an LCSR transient is impressed on the RTD and the time required for the RTD to achieve 62.0% of its final response is measured. This time is called the LCSR :. An increase in the LCSR - is a sign of RTD degradation.

The U of 7 investigators attempted to correlate the Plunge with the LCSR In making this correlation tne time restonse of the R*D was variec by adding

ace or ruccer insulation around :ne RTD anc measuring to:n :ne ?lange - anc ne LCSR. Two such correlations are shcwn in figures 3.1 and 3.2.

An obvious difficulty with this method is the following: This correlation was formed by altering the thermal resistance on the surface of the RTD. When an RTD degrades, it is most likely due to increases in the thermal resistance of the RTD internals or the RTD-thermowell gap. Therefore one would expect to find a different correlation for normal degradation than that determined by adding insulation to the surface of the RTD. For this reason we do not, at present, consider the correlations of figures 3.1 and 3.2 to be sufficiently well substantiated to be used in the determination of the Plunge.

18 -

r ee Wm-

i While not providing an accurate means of ecmputing the Plunge :. these correlations are useful for the degradation test. If in a degradation test the LCSR : is found to increase, then frcm the correlation the approximate increase in the Plunge : can be determined. If the Plunge deternined in this way is near the value assumed in the safety analysis, this would indicate that it is necessary to measure the Plunge via the usual LCSR procedure.

Using the LCSR technique to detect detector degradation is a rather wasteful use of the LCSR electronic equipment. With the addition of one microprocessor the degradation test equipment can be used to measure the Plunge : ss described in section 2.2.1.

i

24- -

22 +

9 20 - -

18 16-y 3

14..

12 E

g E-7,5*

"^

/h Emperical Correlation Curve 3

Y s

7 ed?

2.

.=

Emperical Ca:a 2

g)

LCSR : (sec) 0 1

2 3

A 5

Figure 3.1 Emperical Correlation Curve for Plunge : versus LCSR :

g Rosemont RTO Model 102AFC.

(Combustion Encineerinc 470)

(Reproduced frem Ff gure 6.4 of the 1978 EPPI Topical Report] '

~

w pw-

-mn m

+

-r m

0 3.0 -

2.5 -

7 o

8 2.0 - !

o i

?

.3*

o

1. 5 -.

o

=.

leperie'LCarreLition Cucve.

1.0 -

a =

1 ::erical Cata 0.5 -

0 0.0 LCSR : (sec)

O.0 0.1 0.2 0.3 0.4 0.5 0.6 l

Figure 3.2 E.moerical Correlation Curve for Plunge : versus LCSR :

for Roscront RTD Model 176KF.

(Westingnouse RTD)

(Reproduced from Figure 5.5 of the 1978 EPRI Topical Report].

W mnO@%

,7_

3.2 RTO DEGRADATION TESTS USING *HE SELF HEATING INDEX (SHI) 1 In the SHI test, a constant current is impressed through the RTD element and the steady state change in element resistance is measured. This test is performed at several different currents, and a plot is made of power dissipated by the element versus increase in element resistance. Emperically this has always been found to be a straight line, and the slope of this line (ohms / watt) is called the SHI.

An increase in SHI is a poritive indication of RTO degradation.

As with the LCSR :. the U of T investigators attempted to correlate the SHI With the Plunge t.

Again, as with the LCSR : measurement, the RTD time resocnse was varied by acding insulation to the surface of tne RTD, and plots of Pitnge : versus SHI were constructad. Two suca plots are shown in figurss 3.3 and 3.t.

'hese correlations suf f r the same proclem as tne Plunge : versus the LCSR :

correlations, ard thus we do not, at present, accept them as viatle means for computing the Plunge. However, like the Plunge versus LCSR : correlation, the Plunge : versus SHI correlations would be useful in a degradation test.

-- =~

w

.+.%

--go

i I

i as 12 to '

q 18 <

is.,7 8

~

e 14.$

,,, [

[l Emperical p

Correlation Curve t:

Gc E.moerical Data s

4

.. g ' Extrapolated to Zero SHI (ohms / watt) 5 6

7 8

9 Fiqure 3.3 Emoirical Correlation Cur're for Plunce e versus Sni, for Rosement RTD Model 104AFC. (Cembustion Encineerine i Q)

(Reproduced fecm Figure 6.7 of the 1978 EPRI Topical Report]

23 -

+-

-. ~

,e

3.0- -

2. 5- -

g Emperical Data --m-o

~

2. 0-

~

E 8

8

-c-a 1*:~--

Emperical

/

Correlation Curve 0

1.C

)

/

/

/

0..'.

a s/

SHI (ohms / watt) 0.G -

5 5

7 8

9 10 Figure M Emcerical Correlation Curve for plur.ce - versus SHI for Rosemont RTD Model 176KF. (Westinchouse RTD1 i

[ Reproduced from Figure 6.8 of the 1973 EPRI Topical Deport]

l L -

I v <

-n

l

~

3.3 RTO DEGRADATION TESTS USING NOISE ANALYSIS (NA)

NA tests are perfomed by carrying out statistical (spectral, correlation, zero crossing rate and/or auto regressive) analysis of normal fluctuations of the RTD output signal during nomal steady state reactor operation. These fluctuations are the RTD's response to the fluctuations in the reactor coolant temperature.

The statistical methods referred to above will not be discussed here, and the reader is referred to the three EPRI topical reports for a discussion of these methods.

In the application of the NA method, assumptions must ce made regarding the statistical properties of the coolant temperature fluctuations. If some minimum ret of assumptions, such as stationarity and repeatability are met, the NA method is a valid degradation method since any change in the utput fluctuations can be directly attributad to the RTD itself. If, in addition to stationarity and repeatabili:y, the coolant temperature fluctuations are *wnite" (having fluctuations whose Fourier representation displays constant energy per unit band width at every frequency in the range of interest), NA can be used to determine a Plunge :.

The initial theoretical work in NA done by EPRI was directed toward developing a detennin1stic method for measuring the Plunge : and.this work produced some very sophisticated physical and mathematical developments. Fowever, when the theory was applied to experiment, it was found that NA predictions of the Plunge

were seriously in error, sometimes by as much as a factor o' 5.

The EPRI resear:hers concluded that their principal problem was that the reactor coolant fluctuaticns were not white, as they has immed. Having no other reasonable model for reacter coolant fluctuations, EPRI has, at least for the time being, abandoned efforts to perfor n a deterministic measurement of the Plt.nge :

using NA.

25 -

L.

Researchers at TEC are still pursuing a deterministic method for measuring the Plunge : using NA. Over a period cf 2 years TEC has demonstrated that for certain types of sensors and certain reproducable reactor coolant conditions, I

careful NA measurements of the various statistical parameters have produced results with 110% variation. However, it has been established that coolant temperature fluctuations do not meet the requirements for a Plunge detennination under all reactor conditions for all sensors. To date TEC has not succeeded in developing a systematic correlation between the measured statistical parameters and detenninistic measurements of the Plunge. but there are reasons to believe that such a correlation can be derived for certain sensors under certain verifiable reactor conditions.

As was just stated, the conditions for the coofint temoerature fluctuations for -

an RTD degradation test are less restrictive than those for a deterministic Plunge : measurement.

It has been established that the measured statistical parameters wnich can be extracted from NA of RTDs under verifiable reactor conditions are highly reproducable and changes in these parameters can be used tc infer changes in the RTD Plunge :. Therefore NA methods can be used for RTD degradation measurements suoject to the statistical accuracy of the measurement.

• 4.0 POTENTIAL FOR RTD TIME RESPONSE DEGRACATION

mmes

mmes amme emner _

4.1 MODES OF RTD TIME RESPONSE DEGRADATION The U of T investigatars have evaluated various modes of RTO degradation in 1

section 2.5.3.1 of the 1978 EPRI report and part II, chapter 7 and part V of the 1950 EPRI report. Their conclusion 1: that the main modes of RTD degradation are due to deterioration of the PBX cement used to hold the RTD element in place and deterioration of NEVER-SEEZ, a substance used to increase the thtrmal conductivity between the thermowell and the RTD.

Most of the deterioration in the PBX and NEVER-SEEZ 1s due to high temparatures and takes place fairly soon after the elevated temperature is reachec. Thus the RTDs are expccted to show a marked degradation shortly after they are put in service, and afterward degrade more gradually. If future data bears out this trend, then a reasonable surveillance schedule would requtre frequent testing of the newer RTDs and less frequent tasting of the older ones. Mcwevcc, with the data currently available, this point is inconclusive.

In the TEC topical report it is suggested that RTD time response degradation may be caused by fouling of the thermowell by crud and cracking of the certmic insulatcr in ths RTD. While these are plausable modes of degrtdation, there is no evidence that either of these mechanisms is active in the observed time respor.se degradations.

l l

f.

27 -

---.__.-_l--

t

(

4.2 EVIDENCE OF RTD TIME RESPONSE DEGRADATION l

Records of measured RTD time constants for var tous reactors are presented l

in tables 4.1 and 4.2.

The AMS data from Millstone 2 indicates a systematic degradation of RTDs with service. However most of the other data does not show.this consistent trend. A prudent regulatory position for the present would be to increase tne required surveillance at all plants until enough data is collected to determine if a consistant trend in RTD degradation doas exist.

Table 4.1 Ccmaarf son of In-Flant LCSR and SHI Time Resoonse Tests Conducted py, M

[Taken from Table 11.1 of the AMS Topical Report and Reference 4 3 Time Resconse Test Results for Rosemont Model 104 RTDs t.t Millstone Unit 2 For the Millstaae tests, judging from either the Plunge r or the SHI test, almost all detectors degraded and a few remained unaffected by service. None improved.

August December August December 1977 1978 1977 1978 EIU Plunge :*

Plunge :*

SHI SHI

. umber d

(sec)

(sec)

(ohms / watt)

(chas/ watt)

A7770 3.2 5.2 5.6 7.a A7765 2.8 3.2 4.5 4.8 75313 4.7 5.6 6.2 6.5 A7774 3.3 4.3 5.3 6.2 75294 3.7 4.4 6.0 6.4 75299 5.5 9.3 8.6 9.1 75310 2.5 4.9 6.2 5.5 75300 4.6 4.7 6.5 6.5 75297 3.6 3.6 4.7 4.9 80364 4.0 4.4 5.6 6.1 75709 4.0 4.7 5.5 5.8 A77(9 3.1 3.5 4.3 5.0 l

t Time Resconse Test Results _for Rosemont Medel 176 RTDs at Farley Unit 1 In thase tests there was no evidence of time response degradation.

l October January October January 1978-90 1978 1980 RTD l

Plunge :

3Tunge :

SHI-SHI

""*D'"

(sec)

(sec)

(ohms / watt)

(ohms / watt) 4123 0.10 0.11 7.5 7.4 412C 0.12 0.12 5.3 5.7

• Since the correction factor had not been developed at the time of the August 1977 measurements,.all time constants shown here are uncorrectea value:.

l

~

l 1

e Table 4.2 Ccmoarison of In-Plant LCSR Time Resoonse Test Results Conducted bv TEC on Rosiment Model 104 RTOs at Saint Lucie Unit 1 (Taken from Peferences 7 and 8]

In these tests there is no evidence of time response degradation.

January May October March 1978 1978 1978 1979 Plunge :

Plunge :

Plunge :

Plunge :

(sec)

(sec)

(sec)

(sec)

TE-lil2CA 4.0 + 0.2 4.2 + 0.4 4.0 + 0.4 4.1 + 1.2/-0.7 TE-1112HA 5.2 1 0.5 4.4 + 0.3 4.4 1 0.2 4.5 + 0.3 TE-li22CA 5.5 1 0.2 5.7 + 0.3 5.0 1 0.5 5.0 1 0.7 TE-ll 22HA 5.0 + 0.5 5.5 1 0.3 5.3 1 0.5 5.7 + 0.7/-0.5 TE-lll2C3 5.0 1 0.5 4.3 + 0.5/-0.4 TE-lll2HB 5.0 1 0.9 5.3 2 0.5 TE-il22C3 5.9 1 0.3 5.41-0.2 TE-ll22H3 5.3 + 0.3 5.5 + 0.2 TE-1112CC 4.5 1 0.7 4.3 + 0.3/-0.5 TE-lll2HC 5.4 1 0.4 5.4 + 0.7/-0.5 TE-ll22CC 5.4 1 0.3 5.7 1 0.5 TE-il22HC 5.4 1 0.4 5.0 + 0.7/-0.5 TE-1112CD 4.8 1 0.3 4.9 1 0.5 TE-1112ND 4.9 1 0.5 5.7 + 1.0/-0.7 TE-1122CD 5.7 1 0.5 5.5 + 0.9/-0.7 TE-1122HD 4.3 1 0.5 4.8 + 1.5/-0.9

5.0 ATD TIME RESPONSE TEST RESULTS

===s

==us 5.1 PARAMETERS THAT AFFECT RTD TIME RESPONSE The time response is not only a function of the RTD itself, but depends as well on the properties of the ther=cwell and the thermal characteristics of the medium in which the thermowell or RTD is immersed. The thermal properties of all these components change with temperature and the heat transfer properties of the medium (water) change with flow velocity. The match between the RTD and the thermowell affects the time response, and even the slight change in match that occurs when an RTD is removed from a thermowell and placed back in the same well can significantly change the time resconse. Thus it is important to simulate strvice +..ditions as closely as possible when testing the RTD time response.

As statec earlier, historically the time response of RTOs nas been measured by a plunge test in the laboratory. Normal service conditions of 2235 psig and 540 DEGF are difficult to reproduce in the laboratory. For this reason, in the past most laboratory tests were pcrformed at more benign conditions and the results extrapolated to service conditions. With the advent of the LCSR method, the plunge tast methodology has been re-examined, ard it was found that the historical plunge test procedure often produced results wht:h were grossly in error, sometimes by as much as a factor of 3.

I I l y

-a~

-~

a One of the first suggestions for achieving 540 DEGF without elaborate laboratory equipment was to use hot oil or sand as the medium, rather than water. This was soon demonstrated to be unsatisfactory. The reason is that the heat conduction properties of oil and sand are so different from water that a test in oil or sand gives no indication of what would happen in water. In numerical terms, the thermal match between the medium and the RTD is given by a quantity called the 31ot modulus, which is defined as the ratio of the film thermal conductance to the internal conductance of the RTD [ Mare specifically, Biot modulus = hR/k, where h is the film coefficient, R is the RTD radius, and k is the thermal conductivity of the RTD]. When the Biot modulus is less than about D.1 the thermal resistance is dcminated by the film resistance, and when it is greater than about 10 the thermal resistance is dominated by the RTD internal resistance. The response of an RTD in one heat transfer regime indicates very little about how the RTD will respond in a different heat transfer regime. Values for the Stat modulus for several cases are given in table 5.1.

l l

i 32 -

l w-m g,

w 7-w-v-w--9

.g.

wy---_y,,-~m.,,

..m-9y,-.

,-e, m-m

,--.e, y

Table 5.1 Variation of Biot Modulus iLg,2 m Of fferent Film Coefficients Associated with Different Testing Conditions

[Taken from Reference 9]

Rosemont Rosemont RTD 104 176 Testing (Combustion (Westinghouse)

Conditions Engineering)

Reactor Service 300 3.8 Conditions 3 N sec 27 0.34-180 DEGF Water 1 q.

115 1.5 00DEu(see

2 r Solder a2 I

500 CEGF Oil 0.3 0.02 a

m 3

l j

500 DEGF Sand 0.4 0.01 x

5 Internal No available resistance laboratory test dcminates for both condition water and solder simulates service Connents tesi.s. Good conditions well.

service condition simulation is possible in laboratory tests.

J

• 5.2 RTO TIME RESPONSE TESTING CONDITIONS USED IN PRACTICE: ROOM TEMPERATURE LABORATORY CONDITIONS While room temperature tests do not indicate much about the RTO's behavior at service conditions, room temperature tests are a good way to compare various measurement methodologies. The main testing criteria for :omparing methodologies is that all methodologies are compared under identical conditions, whether these be service conditions or room temperature laboratory conditions.

In fact, all of the development work for the LCSR methodology was done under room temperature laboratory conditions. Results of the roem temperature tests are given in tables 5.2 and 5.3.

With the development work on the LCSR methodology complete, it seemed worthwhile to test the LCSR method ve42as the plunge method at simulated service conditions. The next two sections describe how this was accomplished.

i 5.3 RTO TIME RESPONSE TESTING CONDITIONS USED IN PRACTICE:

EPRI SERVICE CONDITION TESTS [EQE TESTS]

In order to test the LCSR method at service conditions, the U of T investtgators in conjunction with Electricita de France (EDF), performed tests on a simulated reactor coolant test loop constructed by EDF. This loop operates at reactor service conditions of temperature, pressure and flow, and has special valves to induce a step change in temperature for the purposes of simulating a plunge test. The results of this test are shown in table 5.2.

It can be seen that the agreement between the LCSR test and the plunge test is excellent. ~

e, n.-,

,w-r, e-

,-s-.n ww, r--,-

-5

(

Table 5.2 Results of LCSR and Plunce Testino done by the U of T (Taken from Table 10.1 of 1978 EPRI Report and Tables 7-1 17-3 of 1980 EPRI Report]

Room Temperature Tests at U of T Thennemetry Laboratory Measured Plunge r Inferred fran LCSR ercent RTD Plunge : Without Higher With Higher Model (sec)

Mode Correction Mode Correction Rosemont 175KF 0.38 0.39 0.41

+7.9 Rossnont 104ADA 3.1 2.9 3.1 0.0 (without thermowell)

Rosemont 104ADA 7.1 5.9 7.2

+1.4 (with thermowell)

Rosanont 104VC 2.3 1.7 2.1

-8.7 (without thermowell)

Rosemont 104VC 5.3 4.5 5.5

+3.8 (with thermowell)

Rosemont 177GY 5.8 5.1 5.2

+5.9 Rosemont 1773Y 5.1 5.2 5.3

-3.3 Sostman 3505 2.0 1.7 2.1

+5.0 5.2

-1.9 Rosemont IC4AFC 5.3 (air in well) 3.9 0.0 Rosemont 10aAFC 3.9 (NEVER-SEEZ in well) 12.3

+5.1 Rosement 177HW 11.7 0. 41

-2.4 Rosemont 175KF 0.42 Service Condition Tests aji EDF Test Looo Measured Plunge : Inferred RTD lunge :

from LCSR Test Percent Model (sec)

(sec)

Error Rosemont 104AFC 5.2 5.9

-4.8 (Air in well)

Rosemont 104AFC 4.1 3.7

-9.8 (NEVER-SEEZ in well)

Rosemont 177HW 8.8 8.4

-4.5 Rosemont 175KF 0.14 0.13

-7.1 _

,-n

~

M RTD TIME RESPONSE TESTING CONDITIONS USED IN PRACTICE: TEC SERVICE C_0NDITION TESTS [SOLDERTESTS]

TEC has gotten around the problem of getting service condition temperatures by using molten solder, rather than pressurized water, as was done in the EPRI-EDF tests. As can be seen in table 5.1, for the Rosemont 104 RTD tne molten solder provides a very good simulation of service conditions. For the Rosemont 176 RTD the simulation is rather poor.

The TEC comparison of plunge tests and LCSR tests is shown in table 5.3.

As wjth the EPRI tests, the agreement is excellent.

" ~

• _ Table 5.3 Results of LCSR and Plunge Testing done g TEC g Rosemont Model 12 RTOs

[Taken from Taoles 3.1 and 3.2 of Reference 11]

Room Temeeratug Tests Measured Plunge r Inferred Percent Thermo RTD Plunge

• from LCSR Tests **

Error Well Number (sec)

(sec) 60 57161 5.910.2 5.610.3

-5.1 60 57165 5.910.2 6.010.3

+1. 7 60 A8994 6.810.5 6.710.3

-1.5 60 B5642 8.310.7 7.210.6

-13.3 540 CEGF Solder Tests Measured Plunge : Inferred Percent Thermo RTD Plunge :*

frcm LCSR Tests-

g770,

.j,; 7

,.tumee r (sec)

(sec) 50 57147 5.910.2 6.010.4 v1. 7 60 57151 5.010.2 6.010.4 0.0 60 57161 5.0_4. 2 4.810.3

-4.0 60 57165 6.910.2 6.510.4

-5.3 60 57170 5.410.2 5.210.2

-3.7 60 A8934 6.710.2 7.010.4

+4.5 60 35630 5.610.2 5.810.4

+3.6 60 35642 6'.8_M.2 6.910.4

+1.5 66 57161 5.410.2 6.010.2

+i. 1 66 57165 5.910.2 5.310.5

-10.2 66 A8994 6.210.2 7.010.5

+12.9 66 35642 5.910.2 5.710.3

-3.4

'"Jncertainty = le baced on historical uncertainty in reproducibility of plunge tests.

• • Uncertainty = upper and icwer bounds of all variables with uncertainty in them. Uncertainties c::mbined additively.

6.0 AMS AND TEC FIELD EXPERIENCE

meer ammu

AMS has performed LCSR measurements at the following plants:

Millstone Unit 2 ------- Aug 1977, Cec 1978. June 1979, July 1980 AN01 Unit 2 ------------ Nov 1978 North Anna Unit 1 ------ Aug 1979 Farley Unit 1 ---------- Oct 1978, Jan 1980 Farley Unit 2 ---------- May 1980 AMS has sold testing equipment to North Anna, Farley, V.C. Summer, San Cnofre, LOFT, and DRNL.

In addition Millstone plans to purchase AMS test equipment in :ne near future.

TEC has cerformed LCSR measurements at the following plants:

Saint Lucie Unit 1 ----- Jan 1978, May 1978, Oct 1978, Mar 1979 LOFT ------------------- Mar 1979 Sequoya ---------------- May 1979 Zion ------------------- Aug 1979 TEC has sold LCSR testing equipment to Saint Lucie m

-c~

y a

p

-w--

7.0 NRC RESERVATIONS FOR IN-SITU TESTS Most of the reservations we have with in-situ tests have been iterated in other sections of this SE. 'de are listing them here in order to have a compact list for reference. These are:

(1) Using the Plunge : versus the LCSR r ctrrelation to infer the Plunge r from a measurement of the LCSR r (Section 3.1).

(2) Using the Plunge r versus SHI correlation to infer the Plunge r from a measurement of the SHI (Section 3.2).

(3) Using the NA method for measuring the Plunge t (Section 3.3).

P)[1 + (1 - 1)]2 to estimate the higher (4) Using the expression P

=

9 poles o f the transfe: function. [This appears on page 29 of the 1977 EPRI Topical Report.

It is demonstrated to be a poor apcroximation on page 42 of the same report.]

(5) On page 46 of the 1978 EPRI Topical Report it is stated that if only one eigenvalue, ri, can be found, then an upper limit for the Plunge : is 1.4 * :1 This should be 1.47 *:1 which for practical purposas can be rounded to 1.5

• rt.

The first four of these techniques were triginally described in the EPRI Topical Reports at a time when they were still in the experimental stage, and there was hope that these techniques would be proved viable. Since then the U of T investigators have conceded that these are not viable techniques. The disclaimers for these techniques appear on page 42 of the 1977 EPRI Topical Report and page 140 of the 1978 EPRI Topical Report..

• 8.0 REGULATORY p0SITION (1) The LCSR method has been demonstrated to be the only reliable method for measuring the time response of RTDs in nuclear plants. We should take a pos: tion that would favor the universal adoption of the LCSR method in a timely fashion.

(2) The historical plunge test has been demonstrated to be inadequate for measuring the time response of RTDs in nuclear plants. We should cease putting credance in RTD time constanta which have been measured by a plunge test.

(3) Both the AMS and TEC LCSR measurement procedures have been demonstrated to consistently predict the plunge : to within 10%. The number of ccmparisons done to date is inadequate to form a basis for any sophisticated statistical model, and the best procedure to account for uncertainties would be to simply add 10% to the measured plunge : and use this as the measured upper bound. (in some cases (e.g. the EDF data on table 5.2) One errors appear to be composed of a substantial bias plus a random fluctuation.

In this case simply adding a 10% uncertainty to the best estimate plunge :

l is a reasonable procedure.]

I (4) While the RTD degradation tests are discussed in scme detail both here and in the E?RI Topical Reports, neither AMS nor TEC nor any other vender /

consultant / utility has submitted a proposal to employ degradation tests.

Degradation tests should'not be permitted as a' substitute for LCSR tes'ts until such a proposal has been submitted, reviewed, and approved by us.

l Once degradation tests are approved they may be used by utilities instead of LCSR tests to detect RTD degradation, and t. ben only those-RTDs which l

show degradation would need to be tested via the LCSR procedure. *

.m

(5) The extensive RTD time response testing done recently has revealed that the RTDs 1.' operating reactors are suffering time response degradation as they age. Current Technical Specification surveillance schedules pernit such deficiencies in RTDs to go undetected for several years. Ccnsequently the RTD time lags assumed by utilities in their RPS setpoint cceputation may in some instances be unrealistically short.

In these cases the computed RPS setpoints will be nonconservative, and this situation should be corrected.

Fortunately, the transients against which RTDs provide protection ars all rather slow. Assuming a slightly slower RTD time response in the safaty analysis would change the RPS setpoints only a very small amount, and would not present severe restrictions on reactor operations.

In order to guarantee that all utilities are using conservative RTD time lags in their safety analyses, we recemr.end that they" comp' y with one of the Yo11owing options:

l

a. Perform a surveillance test of all their safety channel RTDs at least once every 18 months, and verify that the time res:ense of the sicwest RTD is at least as fast as that assumed in tne safety analysis.

In addition perform a test of each newly installed RTD at operating conditions l

as soon as practical after its installation. If this option is chosen the Technical Specifications must be modified to match the new surveillance schedule.

(As mentioned previously, most current Technical Specifications require that a quarter of the RfDs be tested every 18 months.)

6. Continue with the present RTD surveillance requirements and schedules in the Technical Specifications, but in the safety analysis assume an l

RTD time constant equal to the greater of:

l l

l [

o O

longest time constant measured in last surveillance test I2*

0" (including a 10% allowance for measurement uncertainty)

CE ----- Rosemont Model 104 RTD ------ 12 sec.

W ------ Rosemont Model 176 RTD ----- 0.8 sec.

B&W ---- Rosemont Model 177 RTD ------ 12 sec.

A few words are in order to explain the rationale for options (a) and (b) above.

The present Technical Specirication RTD surveillance schedule was formulated before any evidence of RTD time response degradation appeared, and it wts thought that an occasional spot check would be adequate to assure that no cegradation was taking place. However, with the testing done recently, it nas become apparent that RTD degradation is widespread, and we must take steos to assure that in every instance it occurs it is soon detected, and c:rrective ceasures taken.

For utili-ies anica have precured LC3R test scuipment, opticn (a) is decicecly preferable both frcm NRC's and the utilities point of view. Frem the NRC point of view the frequent and thorough surveillance testing would assure us that conservative values for RTD lags were being uset in the safety analyses.

From the utilities point of view, the accurately measured time lags of :neir RTOs, without any extra conservatism far. tors being added, would be direct input data to their safety analysis. This woula give them the most relaxed RPS setacints possib!c, which would add to their operating flexibility.

~

l In most instances utilities without LCSR equipment remove the RTDs from their reactors and send them to the Rosemont laboratories for surveillance testing. For these utilities having option (a) imposed upon them in a short time frame would represent a severe and unnecessary hardship.

For this reason we are recommending option (b) for those utilities which cannot easily comply with option (a). The time constants of 12 seconds and 0.8 seconds in option (b) are the longest time constants coserved to date for the RTDs in question.

It would not be prudent to assume any faster resconse for an RTD which has not been tested in several years. 'Ahile we do not anticipate measuring time constants greater than 12 seconds and 0.8 csconds, if this should occur, then the longest measured time constant, with an appropriate conservatism factor added should be used as the RTD time constant input into the safety analysis, e

4 g m -

4

g e

I j;g REFERENCES 1.

EPRI NP-459. IN SITU RESPONSE TIME TESTING OF PLATINUM RESISTANCE THERMOMETERS. Kerlin, Miller, Nott, Upadhyaya. Hashemian, Arendt, January 1977.(Herein called the 1977 EPRI Topical Report]

2.

EPRI NO-834. IN SITU RESPONSE TIME TESTING OF PLATINUM RESISTANCE THERM 0 METERS, Kerlin, Miller Hashemian, Poore, July 1978.

[Herein called the 1978 EPRI Topical Report) 3.

EPRI Report (To be 'sublished) TEMPERATURE SENSOR RESPONSE CHARACTERIZATION, Xerlin, Miller, Hashemian, Poore, Skorska, Connault. Upadhyaya, Jacquet.

[Herein cassed the 1980 EPRI Report]

4 Material extracted frcm a papar in preparation entitled ACCURACY OF LOOP CURRENT STEP RESPONSE TEST RESULTS, T.W.Xerlin, April 22,1980 5.

RESPONSE TIME CUALIFICATION OF RESISTANCE THERMOMETERS IN NUCLEAR POWER PLANT SAFETY SYSTEMS, Northeast Utilities Topical Report prepared by Dr.T.W.Kerlin of Analysis and Measurement Services Corporation (AMS),

November 1979. [Herein called the AMS Topical Report]

Gr--RESPONSE TIME OF PLATIN RESISTNACE THERMOMETERS USING THE LOOP CURRENT STEP RESPONSE TECHNIQUE, Mott, Robinson, Jones, Mathis, Fisher, Technology for Energy Corporation (TEC), April 1978.

[Herein called the TEC Topical Report]

7.

RTD TIME CCNSTANT SURVEILLANCE REPORT, Latter from Robert E. Uhrig (FPL) to accert W. Reid (NRC), January 3,1979.

8.

RTD TIME CCNSTANT SURVEILLANCE REPORT, Letter from Robert E. Uhrig (FPL) to Robert W. Reid (NRC), May 1,1979.

9.

TEC handout at NRC meeting entitled REVIEW OF TEMPERATURE SENSOR RESPONSE TIME USING LOOP CURRENT STEP RESPONSE TECHNIQUE, Ackermann,& Mott, August 16, 1978.

10. Letter, T.W.Kerlin (AMS) to P.S.Kapo (NRC), April 28, 1980

11. TEC LCSR METHOD TEST RESULTS, Letter from R.E.Uhrig (FPL) to R.W.Reid (NRC).

December 4,1979.

i.

-