ML20155J704

Partially Withheld Results of Joint Ofc of Investigations/ NRR Evaluation of Aw Miller Re Leak Rate Testing Irregularities.Denial of Involvement or Knowledge of Irregularities Questioned.W/O Encl 2

ML20155J704
Person / Time
Site:	Three Mile Island
Issue date:	09/06/1985
From:	Russell W Office of Nuclear Reactor Regulation
To:	Harold Denton Office of Nuclear Reactor Regulation
Shared Package
ML20151L202	List: ML20151L198 ML20155J582 ML20155J591 ML20155J599 ML20155J625 ML20155J636 ML20155J641 ML20155J649 ML20155J653 ML20155J677 ML20155J684 ML20155J692 ML20155J704 ML20155J710 ML20155J713 ML20155J726 ML20155J745 ML20155J753 ML20155J770 ML20155J778 ML20155J792 ML20155J808 ML20155J825 ML20155J835
References
NUDOCS 8605270168
Download: ML20155J704 (24)

Text

-

~

" ndosure H Do., UNITED STATES

, NUCLEAR REGULATORY COMMISSION s wasmuovow,o.c. zones l

• t September 6, 1985 g*....,/

MEMORAN00M FOR: Harold R. Denton, Director Office of Nuclear Reacter Regulation FROM: William T. Russell, Acting Director Division of Human Factors Safety, NRR

SUBJECT:

RESULTS OF JOINT 01/NRR INVESTIGATION AND EVALUATION OF ADAM W. MILLER

Reference:

1. Memorandum from S. J. Chilk (SECY) to B. B. Hayes (01) and W. J. Dircks (ED0) dated April 2, 1984

Subject:

- Staff Requirements-Discussion of Pending Investigation-

~

TMI

2. MemorandumfromH.R.Denton(NRR)toB.B. Hayes (01) dated May 3, 1984

Subject:

NRR Review of OI Investi-gation Materials Concerning Hartman Allegations of Falsification of Leak Rate Data at TMI, Unit 2

3. Memorandum from W. T. Russell (DHFS) to H. R. Denton (NRR) dated January 24, 1985

Subject:

Follow-up O Action on Additional TMI-2 Operators The purpose of this memorandum is to document the results of the joint 01/NRR investigation and evaluation of Mr. Adam W. Miller, a licensed Senior Reactor Operator (SRO) at Three Mile Island, Unit 2 currently serving as the Manager ,

Plant Operations, and to provide a reconnendation regarding whether his  ;

current SRO license should be revoked, modified or suspended under 10 CFR 55.40 or other enforcement action taken under 10 CFR Part 55.50, due to his involvement in preaccident leak rate testing irregularities at THI-2.

Background

As the result of a Commission meeting on March 23, 1984 NRR was directed by Reference 1 to review OI investigative materials concerning falsification of reactor coolant system (RCS) leak rate tests at TMI-2 and refer back to 01 NOTE: This memorandum and Enclosure 1 discuss infonnation which is the subject of an ongoing OI investigation. This memorandum and Enclosure 2 dlscuss information that is maintained in the NRC's Privacy Act System of Pecords (NRC-16). This memorandum and enclosures may not be disseminated outside the NRC without coordination with NRR and the pefTnission of the EDO or the Director. 01. Internal access and l distribution should be on a "need to know" basis.

i t

O e B60527016e 86051630 DR, ADOCK O 29y

l i

September 6, 1985 I l

l l~

those matters which required further investigation. The results of NRR's review was provided in Reference 2. The review detemined that follow-up l investigation by 01 and further evaluation by NRR was needed in the case of seven currently licensed operators. Mr. Miller was not one of those seven s i individuals. At a follow-up Comission meeting on May 23, 1984 NRR proposed - i joint OT/NRR investigations and evaluations of these individuals.

Subsequently, NRR issues letters to these individuals under 10 CFR 55.10(b) requesting additional information regarding current performance. Based upon - '

the joint investigation into the individuals' past involvement in improper activities at TMI-2 and an evaluation of the individuals' subsequent performance, NRR would recomend what action, if any, should be taken against '

the identified operators. As a result of some of the early interviews with these individuals and further technical analysis, Reference 3 recomended that three additional currently licensed operators also be interviewed. Mr.

Miller was one of the operators identified in Reference 3. NRR issued  ;

similar letters to,these individuals under 10 CFR 55.10(b). Of the l ten individuals identified for investigation, seven are currently licensed on  ;

TMI-2, two are licensed on Waterford 3 and one is licensed on San Onofre 2.

and 3. l

, Past Involvement in TMI-2 Leak Rate Testing Irregularities  ;

On June 5,1985, a joint 01/NRR interview of Mr. Adam W. Miller was held at the Three Mile Island Training Facility, Middletown, Pennsylvania. The interview was conducted in the presence of Mr. Miller's personal attorneys.

The purpose of the interview was to detemine Mr. Miller's role, if any, in O improper activities associated with RCS leak rate surveillance testing at TMI-2 prior to the accident on March 28, 1979. A't that time, Mr. Miller was a licensed Senior Reactor Operator (SRO) and served as the Shift Foreman for Shift "D." A detailed evaluation of Mr. Miller's interview and a sumary of l the technical analysis of leak rate surveillance tests involving Mr. Miller l are provided in Enclosure 1 to this memorandum.

)

During the interview, most of Mr. Miller s responses to general questions i

, associated with RCS leak rate surveillance test problems at TMI-2 were consistent with statements made to the NRC by other licensed operators, including two of the three Control Room Operators (CR0s) on Mr. Miller's shift (Messrs. Mark Coleman and Lynn Wright). However, Mr. Miller denied any

personal involvement in or knowledge of leak rate test manipulation by the

operators conducting these tests on his shift. Based upon the results of NRR's technical evaluation and the testimony of Mr. Miller and the three CR0s i on his shift, the following findings and conclusions are drawn.

1. - TMI-2 Technical Specification 4.4.6.2 required that RCS leakages be demonstrated to be within limits by the performance of a RCS water inventory balance (leak rate test) at least once per 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> during steady state operation. THI-2 Technical Specification 3.4.6.2 limited unidentified leakage to 1 gpm. On Mr. Miller's shift, leak rate tests were routinely run at least once per shift, plant conditions and ,

evolutions permitting. ,

I

2. TMI Administrative Procedure 1012 " Shift Relief and Log Entries,"

required the title, number, start and completion times of all periodic tests required by the technical specifications (i.e., leak rate tests) be logged in the CRO's Log. Contrary to this requirement only l

l a_-_--_ _ . _ _ _ . . . _ _ _ . -

September 6, 1985 satisfactory (i.e., unidentified leakage less than 1 gpm) leak rate tests were logged in the CRO's Log by Shift "D" personnel. i l

3. Bec'ause Shift "D" performed leak rate tests more frequently than I requtred by the technical specifications, only satisfactory tests were considered surveillance tests. Tests with results in excess of the

_' j i

technical specification limits were not considered surveillance tests j and were discarded. This action was contrary to TMI-2 Technical - r i Specification 6.10.1.d " Record Retention," which required that records of surveillance activities required by the technical . 1 specifications be retained for a period of at least 5 years. Mr. l Miller was not sure how this practice originated; however, it had ,

existed for years and was common practice all shifts. l

4. Because leak rate tests were run so frequently, the CR0s on Shift "D" conducted these tests on their own as part of their shift routine. If tests results exceeded the limits of the technical specifications, the l CR0s frequently discarded tests on their own. If an unsatisfactory test was provided to Mr. Miller, he would personally discard the test and instruct the CR0s to perform another test.

5. In reviewing leak rate tests, Mr. Miller gave them a very cursory look before signing his approval. He essentially just looked at the bottom line numbers. As long as the results did not exceed the technical specifications, he would approve the test.

i O In general, Mr Miller believed that prior to the accident, the '

6.

general attitude of " don't worry about the paperwork just keep the hardware operating" was imparted to the operators by the Supervisor of Operations, Mr. James Floyd.

7. Surveillance Procedure 2301-3D1, " Reactor Coolant System Inventory,"

i required the operators to enter the action statement of Technical Specification 3.4.6.2, if unidentified leakage exceeded 1 gpm. The action statement required leakage to be reduced within limits within 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> or be in at least hot standby within the next 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> and in cold shutdown within the following 30 hours3.472222e-4 days <br />0.00833 hours <br />4.960317e-5 weeks <br />1.1415e-5 months <br />. Mr. Miller believed that

he and his CR0s were aware of these requirements at the time.

Nevertheless, as long as a satisfactory leak rate test was obtained within 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> of the last satisfactory test, the action statement was not invoked, regardless of how many unsatisfactory test results were ,

received in the interim. This policy of accepting good tests and ignorirg bad tests was also practiced by all shifts.

1 8. Mr. Miller stated that unidentified leakage of 1 gpm was net only the triggering point for entering the action statement of the technical specificaticns, but it also required notification to the.NRC and was ,

the triggering point for entering the Emergency Plan as an unsual i event. He believed that people may have tended to look the other way j and ignore tests that had results greater than 1 gpm because they did l not want to trigger those actions.

9. Although Mr. Miller had initialed the routing sheet for LER 78-62/1T, -

he did not have a preaccident recall of the incident. Contrary to

)

---

__y%-- -, * - - + - - - - - ,,.ye.,m-,-rwmw -w-w-m --gw,.y e,y,y -ecy9g7%77, q,p. 7 9 p y 3yw-9w ywy,- y, q r g e p --9 wf -w m,9 9 - 9 mm-

September 6, 1985 statements in the 1.ER that the event was caused by a misinterpretation of the requirements of the technical specifications and that operators would be instructed to invoke the action statement when the limiting condition for operation was exceeded, Mr. Miller did'not recall being .,

advised or instructed in this matter. Consequently, he neither changed _

his interpretation of when the action statement should,be invoked nor provided any revised guidance to the operators on his shift. He did not believe there were any changes made to leak rate testing practices at -

TMI-2 as a result of this incident.

10. Mr. Miller stated that he was not aware until after the accident that adding hydrogen to the MUT towards the end of a leak rate test could affect MUT level indication and consequently leak rate test results. He

. disagreed with Mr. Coleman's testimony that this phenomenon was comon l

knowledge among Shift Foremen and Shift Supervisors. Mr. Miller stated I he had no personal knowledge at the time that anyone on his shift manipulated le'ak rate test results by the addition of hydrogen.

11. Mr. Miller stated that he was not aware until after the accident that adding water to the MUT just prior to the end of a leak rate test could produce the same effect on MUT level and leak rate test results "

as adding hydrogen. After the accident, Mr. Miller had reviewed all of the leak rate tests performed by his shift. He could not see any legitimate operational reason why, day after day, his shift had to add water to the MUT five to ten minutes before the end of the test. Mr.

Miller agreed that if his operators were taking advantage of O instrument errors by adding water and using the number off the batch controller as the amount of the addition, knowing that the computer was going to read the water level off the MUT level instrument, they j were manipulating test results.

~

12. After reviewing Shift "D" leak rate tests, Mr. Miller stated that' while it appears fairly obvious now that his operators may have been taking advantage of perturbations in the MUT level transmitters to i influence the outcome of leak rate tests, he stated he had no knowledge of this activity at the time. He had no suspicions that

this was occurring on his shift or any other shift. He had no ,

, conversations with other Shift Foremen who indicated suspicion on their l l part that operators on their shifts were performing these actions.

13. Mr. Miller did not believe that if operators were manipulating test

results it was caused by pressure from their supervisors. He could

only speculate that it may have caused by some misguided pride, j

competition or one-upmanship between shifts.

In sumary, Mr. Miller agrees in retrospect that many of the actions involving leak rate surveillance testing at TMI-2 violated approved plant

, procedures 1 nd were contrary to the TMI-2 Technical Specifications. Mr.

Miller was aware that as Shift Foreman it was his responsibility to ensure his shift's compliance-with the technical specifications. - Mr. Miller also agrees that-there is strong evidence, including the testimony of Mr. Coleman, to indicate operators on his shift were perfoming actions that were contrary to intent and/or spirit of the leak rate, test surveillance procedure and that 1

September 6, 1985 O these actions were done with the intent of influencing or manipulating the outcome of leak rate surveillance tests. Nevertheless, Mr. Miller states that prior to the accident, he had little involvement in the performance of '

leak rate tests and had no involvement in or knowledge of operators on his shift intintionally manipulating the outcome of test results. _'

The results of the technical analysis indicate that of the six shifts standing watch during the last three months of operation of TMI-2, Shift "D" -

had the highest percentage (93%) of " questionable" leak rate tests. The vast majority of these questionable tests show water additions to the MUT during the last few minutes of the test resulting in a calculated leak rate that was smaller than the actual leak rate. Despite conflicting testimony by the three CR0s, as to why the water additions were made, the weight of the evidence would indicate that these additions were made to intentionally influence the outcome of leak rate surveillance tests. While Mr. Miller was not directly involved in the actual conduct of the tests and his statements that he did not or' der or direct the manipulation of tests are credible and are supported by the operators' testimony, it is not plausible that Mr.

Miller was unaware that leak rate test manipulations were taking place on his shift.

h Current Performance e

a O .

0

,, -,..,-w.--,.n ~---,-,1, - , , . -r,--. ,- n- y

. September 6, 1985 I

Conclusions

! During the period under investigation, while Mr. Miller served as the Shift Foreman at TMI-2 for Shift "D." he admitted he was involved in activities I associated with reactor coolant system leak rate testing that were in  !

violation of approved plant procedures and the TMI-2 Technical Specifications. l These actions included:

1. Failure to ensure that leak rate surveillance tests were conducted in accordance with the approved plant procedure;

2. Failure to properly review and approve the results of leak rate surveillance test requirements;

3. Failure to ensure that all leak rate surveillance test results were

! properly recorded in the CRO's Log; -

- 4. Failure to take or direct the required follow-up action when the limiting conditions for operation specified in the TMI-2 Technical Specifications were exceeded;

5. Personally throwing away or directing his CRO's to throw away leak rate

, test results which exceeded the limits of the TMI-2 Technical Specifications.

- Despite these actions, Mr. Miller denies that he was involved in or was  ;

knowledgeable of other actions on the part of his CRO's to intentionally J

influence or manipulate the outcome of leak rate test results. The weight of  ;

i available evidence, including statements by one of his CR0's, Mr. Mark '

Coleman, and the technical analysis, stro'ngly suggests that Mr. Miller was l either not truthful in answering questions regarding his knowledge of these l activities or he was grossly negligent in his duties as Shift Foreman with

respect to leak rate surveillance tests.

Based upon the testimony of the

majority of CR0s and other Shift Foremen interviewed to date, it is clear i that management's standards for procedural compliance were lax and permissive.  !

In my opinion, this was created through negligence on the part of managament. +

Lack of operator respect for the leak rate test procedure coupled with ,

inanagement's failure to correct test procedure problems and insti ll high  !

standards for procedural compliance led to conditions that motivated some operators to falsify leak rate test results. In this case, contrary to Mr.

Miller's denial, I believe' that Mr. Miller was personally responsible for the ,

gressureonoperatorstoproduceacceptableleakratetestresults. His bottom line" attitude, discarding of tests, admission of his knowledge of l required actions for exceeding limits and his desire to avoid initiating the emergency plan or a plant shutdown are strong evidence of his personal

! involvement in improper activities. I also believe manipulation of leak rate tests occurred on his shift with his knowledge.

J e

t

---__w-, ._ _ ._-- - , - , ,,- - _ , , . - - . , , _ , , . . . -

,,,,,__-,,,_.--.,,e,.- - -

,,--,,.,,,--,,,,,.._._._,._.,m.

e,. ,y.._,, ,,,,,-,...,,,.,_,,,,,,,.,_.,,y-,

l . September 6, 1985 l

l 4

3 _

i __ -

1 i

i l

i i

I

~

&jh =2??

i William T. Russell, Acting Director Division of Human Factors Safety Office of Nuclear Reactor Regulation

Enclosures:

i As stated l

! cc: B. Hayes K. Christopher ,

4 I

1 i

i

)

.l 1

i

'l i

i l

i

h .

' I NOTE: This enclosure discusses infonnation which is the subject of an t

i O _

ongoing OI investigation. This enclosure may not be disseminated outside the NRC without coordination with NRR and the permission of the EDO or the Director 01. Internal access and distribution should

,. be on a "need to know" basis. '

j g Enclosure 1

~

PAST INVOLVEMENT IN TMI-2 LEAK RATE TESTING IRREGULARITIES I. Background '

.I i

On June 5,1985, a. joint Office of Investigations (01)/0ffice of Nuclear Reactor Regulation (NRR) interview of Adam W. Miller was held in the Three i

Mile Island Training Far.ility, Middletown, Pennsylvania. The purpose of the l' interview was to determine Mr. Miller's role, if any, in improper activities associated with reactor coolant system (RCS) leak rate surveillance testing at Three Mile Island, Unit 2 (TMI-2) prior to the accident on March 28, 1979.

Present during the interview representing the NRC were: R. Keith Christopher, Director, Office of Investigations, Region I; William T. Russell, Acting Director, Division of Human Factors Safety; and Robert A. Capra, Senior.

Program Manager, Generic Requirements and Regional Operations Staff.

Representing Mr. Miller at the interview were his personal attorneys Mr.

Smith B. Ge Harrisburg,phart and Ms. Jane Pennsylvania. A copy G. Penny of theoftranscript the law fira of Killian & Gephart, Mr. diller's interviw ,

is included as Attachment 1 to this enclosure.

Mr. Miller is employed by General Public Utilities Nuclear Corporation (GPUN) and currently holds the position of Manager Plant Operations TMI-2 and is a '

licensed Senior Reactor Operator (SRO) on TMI-2. Mr. Miller started with Metropolitan-Edison Company (Met-Ed) in April 1973 as an Auxiliary Operator (AO). He served in that capacity for a 3

promoted to Control Room Operator (CRO)pproximately .

two years Mr. Miller served as an before on-shift being-CR0 at TMI-2 from mid-1975 to mid-1978 at which time he was promoted to Shift Foreman. He remained in that position at TMI-2 until the Fall of 1979 at which time he was transferred to the Training Department as an instructor. j In the Spring of 1981, he was transferred back to the Operations Department I as a Technical Analyst, serving as the assistant to the Manager Plant  !

Operations TMI-2. Mr. Miller was promoted to his present position 'in the l summer of 1983. As Manager Plant Operations TMI-2, Mr. Miller has four Group i

Managers who report to him

Chemistry, Radwaste Operations, Radwaste Support, and Plant Operations. Mr. Millar reports directly to the Site Operations

{. Director.  !

The interview with Mr. Miller concentrated on the period September 30, 1978-through March 28, 1979. During this time period, he served as the Shift Foreman of Shi.ft "D" at TMI-2.

A summary of the technical analysis of the leak rate tests involving Mr.

Miller is provided in Section II of this enclosure. A sumary of the ,

i' interview with Mr. Miller is provided asSection III. The sumary of l interview section includes citations to the page numbers of the transcript  !

from with the sumarized information was extracted. Overall findings and  !

conclusions regarding Mr. Miller's involvement in improper activities l associated with TMI-2 leak rate surveillance testing are presented in Section .

IV of this en' closure.

, , - - -,-.? , - ~ , - .m.-,,,-._, , . . , . . _ , , . , _ . . , . , _ - - , . . . . - - , . , . - , . . , - - , , . . , , . . , , , , ,n , _ , -, ---n .

I II. Summary of the Technical Analysis O In late 1983 and early 1984, a technical analysis of the leak rate surveillance tests conducted during the last six months of operation of TMI-2 was performed by the NRC. This analysis was done as technical support t$ the Department of Justice (D0J) in its criminal proceeding against Met-Ed. In l

1985 _NRR performed a reevaluation of that analysis by factoring in -

infonnation that was learned during interviews with former TMI-2 operators.

Mr. Miller was questioned on the updated 1985 analysis. While a copy of the-complete evaluation of leak rate tests at TMI-2 is provided as Attachment 2 to this enclosure, a discussion of the tests involving Mr. Miller is provided  ;

here to help the reader understand the basis for the questions posed to him '

during' his interview and the basis for the findings andconclusions shown in Section IV.

The vast majority of leak rate tests performed at TNI-2 were done by using the plant computer. The computer-generated leak rate surveillance test sheet i

would be signed by the operator performing the test and approved by an SRO.

In most cases, the Shift Foreman approved tests run on his shift. The test records show that of the 161 leak rate surveillance tests retained by the

licensee during the period under investigation, Mr. Miller was involved in 25 leak rate surveillance tests. The table below shows a breakdown of the overall conclusions regarding these tests.

Breakdown of Leak Rate Tests Approved by A. W. Miller i

Evaluation Category Number of Tests 1

Total number of tes ts on fi1 e. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 -

Tests with no apparent problems.................. 7 Unstabl e pl a nt condi ti ons . . . . . . . . . . . . . . . . . . . . . . . . . ..................

................. 1 Unstable or out-of-service makeup tank (MUT) level transmitter..... 5 i Water additions to MUT (not included in the calculation)........... 2 Water additions to MUT (partially included in the calculation).....11 Feed and bleed operations (not included in the calcula Hydrogen additions to MUT............................. tion)........

2 0

Unidentified. leakage greater than 1 gpm............................

........ I

_ Note: Four tests fall into more than one evaluation category.

Test #50 contains a water addition that was not included in the calculation and was run with an unstable MUT level transmitter providing input to the computer.

Test #70 was conducted during unstable plant conditions and was also run with an unstable MUT level transmitter providing input

to the computer.

_ Tests #132 & #141 were begun during feed and bleed operations and also contains water additions that were not fully accounted for in the calculation.

As can be seen from the table above, only seven of the 25 tests involving Mr.

Miller appear to have been conducted in accordance with the requirements and precautions of the surveillance procedure governing the conduct of leak rate tests (Sp 2301-3D1). The result of one test exceeded the limits of the technical specifications and the remaining 17 tests involve actions that violate the limits and precautions of the procedure, thus yielding j questionable or invalid results. For some tests, these actions may have been

\

taken deliberately in a~n attempt to manipulate or influence the outcome of O the test results. The basis for the evaluation of each test is presented below. _

The seven tests.that appear to have been conducted in accordance with the _'

surveillance procedure are tests #1 (09/30/78), #38 (11/06/78), #54 i 1

(12/10/78)r #57 (12/11/78), #60 (12/13/78), #88 (01/06/79) and #156 l

(03/28/79). Test #156 was actually performed by Shift "A" just prior to the ' l accident but was signed later by Mr. Miller . '

i Onetest,#70(12/21/78), was perfonned during a period when the plant was i not in a steady state condition. SP 2301-3D1 directs that the test be

! perfonned once per 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> during steady state operation when in Modes 1 through 4. The procedura cautions the operator to maintain the RCS and makeup system in a steady state condition during the test by avoiding. change

in valve line-ups, coolers-in-service, pumps-in-service, etc. Power level changes should be minirized and the operator should avoid additions or removal of water from tne RCS and makeup system during the test. For the most accurate determination of RCS leak rate, the initial and final

. conditions of reactor power, RCS temperature, pressure and pressurizer level should be identical. Test #70 was performed during a plant heat-up prior to criticality. The MUT strip chart shows significant esci11ations during the .

one hour test period. As will be discussed below, an unstable MUT level transmitter was also used during the test to provide input to the plant computer.

O At TMI-2 there are two level transmitters (LT-1 and LT-2) that provide MUT level indication. The output of one of the leve1' transmitters drives the MUT l 1evel strip chart recorder in the control room while the other level l transmitter provides MUT level indication to the plant computer. A selector switch in the control room allows the operator to switch level transmitters feeding the strip chart recorder and the plant computer. When the selector switch is positioned to LT-1, the strip chart recorder would be driven by the.

output of LT-1 and LT-2 would provide automatic input to the plant computer for leak rate test calculations. When the selector switch was changed to LT-2, the opposite would occur. .

Between October 30 and November 6,1978, operators experienced. problems with both level transmitters. At various times during this period the level transmitters were reading radically different levels in the MUT and both were taken out-of-service for maintenance work and recalibration. During this period, tests #33 (11/02/78) and #35 (11/03/78) were performed by Shift "D" i

operators and approved by Mr. Miller. During test #33 the level transmitter j providing input to the plant computer was reading approximately 14 to 15 inches higher than the level recorder providing input to the strip chart i reccrder. According to the strip chart, MUT level decreased 3 inches or 90 gallons during the test, the level transmitter providing input to the plant computer reflected only a .5 inch or 15 gallon decrease during the same period. During test #35, the MUT level transmitter providing level input-to the computer was out-of-service and was pegged high at 99 inches. -

Beginning in early December,.1978 and continuing through January 11, 1979,

' O the output of LT-1 became very erratic and unreliable. Because of the erratic nature of the output of LT-1 during this period, any leak rate tests i

perfonr.ed with the use of LT-1 providing input to the computer must be

• l

considered questionable or invalid. Of the 50 tests conducted by all shifts during this period only 16 were perfomed with the stable level transmitter O (LT-2) providing input to the computer during the test. During this period Mr. Miller approved seven tests; f0ur were run with LT-Z and three were run with LT-1 providing (input to the computer. The four tests using LT-2 were . '

1. 54 (12/10'/78), #57 12/11/78), #60 (12/13/78) and #88 (01/06/79). The three tests using LT-1 were: #50(12/08/78),#66(12/17/78)and#70(12/21/78).

For test #50, there does not appear to be a large difference between LT-1 and LT-2 for the initial and final values of MUT 1evel and therefore the use of LT-1 may not have produced a significant error. As will be discussed below, test #50 also contains a possible unaccounted for water addition. For tests

1. 66 and #70 rendering thethere testsare large differences between the outputs of LT-1 and LT-2, invalid. The Shift Foreman's Log, maintained by Mr.  !

Miller, records the completion of two additional leak rate tests during this '

period: 1045 on 12/03/78 and 1800 on 12/09/78. No surveillance test sheets were retained for these tests; however, a review of the MUT strip chart during these times indicates both of these tests were also run with unstable LT-1 providing input to the plant computer.

During the entire period of time LT-1 was essentially "out-of-consission,"

all six shifts, including Mr. Miller's, continued to run leak rate tests (34 out of 50) with LT-1 providing MUT level indication to the computer. These actions resulted in a meaningless calculation of unidentified leakage that could gpm.

not demonstrate conformance with the technical specification limit of 1 O Two tests, #50 (12/08/78) and #144 (03/10/79) include possible water additions to the MUT that were not accounted for in the leak rate test calculation.

In addition to using unstable LT-1 as input to the computer for test #50, from the offset in the MUT strip chart it also appears that approximately 60 gallons of water may have been added to the RCS near the beginning of the test. There is no log entry in the CRO's Log and no water addition is included in the leak rate calculation. Test #144 contains a possible jogged water addition to the MUT of approximately 100 gallons. There is no log entry in the CRO's Log and no water addition is included in the leak rate calculation. A jogged water addition is when water is added to the MUT in small increments over a period of time rather than adding water in a single large or batch addition. According to fomer Shift "E" CR0 Haro18 Hartman, jogged water additions were one of the niethods used by by operators to manipulate the results of leak rate tests. According to Mr.

Hartman jogged water additions were made to hide the fact water was being added during the test. Twst #144 was actually perfomed by Shift "E" operators, Messrs. Bocher and Blessing and logged in the Shift Foreman's Log by the Shift "E" Foreman, Ken Hoyt; however, the surveillance test sheet was '

signed by Mr. Miller after he relieved Mr. Hoyt.

i Onetest,#128(10/17/78), indicating an unidentified leak rate of 2.07 gpm, '

was perfomed and gigned by Mr. Coleman but was not signed by Mr. Miller.

Mr. Miller was the Shift Fareman on watch at the time the test was conducted.

This test'waTrun during a period when the plant was experiencing high leakage (gross leakage of approximately 4.5 gpm). This surveillance test was O not filed with the other surveillance tests maintained by the licensee..

Instead, this tests was filed with a set of tests associated with the l

{

J

O submission of LER 78-62/1T on November 1,1978.

Transcript pages 45 through 55).particular LER is discussed d - ~

s Two tests, #132 (02/27/79) and #141 (03/08/79) operations caTcuTation. that were not taken into account in the leak' rate testcontain fe Test #132 was begun during a feed and bleed operation with test.

approximately 122 gallons being bled from the RCS at the beginnin Two water additions, totaling 210 gallons, were made durin Only 162 gallons of the addition were included in the leak on; rate however, since the water bled from the system.was not included Similarly, Test #141 was also begun during operation. .

eed th

' calculation. Water bled from the system was not included in the leak rate minutes beforeTest #141ofalso the end contains a 250 gallon water addition three the test.

additions that are not fully accounted for er is disc

)

I net positive suction head for the RCS makeu was maintained in the MUT.

i i

i of the operating band, the CR0 would add hydrogen the addition of hydrogen should not have affected MUT level; how of the configuration and environment of the MUT level detection , because, O instrumentation (dry) reference leg ofsystem the levelat TMI-2, water could collect in the low-transmitters.

resultant water slug or " loop seal" could cause a temporary incr Under this conditions, the indicated actually MUT adding water level to thewhen tank. hydrogen pressure was increased out in Thus, the addition of hydrogen at the before the final data reading were taken) could in a nonconservative manner (e.g., the calculated leak rate would b s than the actual leak rate). e less MUT, water additions to the MUT shcrtly befor final data set would thus have the same result as a hydrogea s If an addit operator were aware of this cause and effect relationship leading to th .

" instrumentation inaccuracy," (without having to know it was

" loop seal" effect described above), he could take advantage of th written.

phenomenon to manipulate test results by f the increase last 510 minutes inches (30 galof/ inch);a leak rate test however. should if sufficient o cause water hadthe co in the loop increase 6 inchesseal, (180thegal).indicated level on the MUT stripected chart y might l

including the 6 inch rise in MUT 1evel, the operator would e i gallon water addition into the computer as an operator-caused . Thus, c, when the computer calculated the net uniden minutes).

l Mr. Mark Coleman stated during his joint r 14, er 01/NR ,

i i

I a

_________.__.,__,....?_._.__-,:.__,_,__-,.._._,.,_.._,____,-,._.-.,,,.,._,.~.n.*,._,._,.,

i near the end several leak rate tests in order to take advantage of this O phencmenon to help him get tests results with the limits of the technical specifications. As will be shown below, the technical analysis supports Mr.

Coleman's admission.

While Mr. Coleman stated he could not be sure that others en his shift also manipulated test results in this manner, after - s reviewing the leak rate test data, he stated it would be hard to believe that they did_not also use this method. The table below shows that the last 13 ,

leak rate tests performed by Shift "D" Control Room Operators all exhibit-this same characteristic. Mr. Miller approved 11 of these tests. Two tests

(#137 and #142) were approved by Mr. G. R. Hitz.

Computer MUT 5 trip Error Time before Test # Date Input Chart Produced End of Test CRO-Test CRO-Panel )

93 01/11 200 gal 240 gal 40 gal 23 mins Coleman Olson 122 02/16 204 qa1 210 gal 4 gal 13 mins Olson Coleman 129 02/25 150 gal 200 gal 50 gal 4 mins Coleman Olson 131 02/26 154 gal 180 gal 26 gal 17 mins Coleman Olson 132 02/27 162 gal 210 gal 48 gal 20 mins Coleman Coleman 133 02/28 100 gal 150 gal 50 gal 4 mins Olson Wright 03/03 152 gal 137 200 gal 48 gal 3 mins Olson Olson 138 03/04 238 gal 270 gal 32 gal 2 mins Coleman Coleman 139 03/05 128 gal 310 gal 182 gal 10 mins Olson 'Olson 140 03/06 180 gal 210 gal 30 gal 1 min Wright Olson 141 03/08 228 gal 250 gal 22 gal 3 mins Olson Wright 142 03/09 181 gal 240 gal 59 gal 3 mins Wright Olson 03/13 305 gal 4

O 146 330 gal 25 cal 5 mins-Olson-Key: Computer Input = amount of water included in test calculation by CRO.

- Coleman I

MUT Strip Chart = amount of addition indicated on MUT strip chart.

Error Produced = difference between MUT Strip Chart and Computer Input.

Time Before End of Test = time before end of test that water addition was duringmade the or time of last water addition if more than one occurred test.

CRO-Test = CR0 who performed and signed leak rate test.

CRO-Panel = CR0 who was on the panel during test and who signed the CRO's Log.

As stated earlier in this section, SP 2301-3D1 cautioned the operators to avoid the addition and removal of water from the reactor coolant and makeup systems during leak rate tests. A review of all leak rate tests, during the period under investigation, show that this pattern of "underrecorded water additions" near the end of the leak rate test is unique to Shift "D." Five other tests are also classified as underrecorded water additions; however, i

these tests were done by three different shifts and or.ly two show that,the water additions were made during the last 10 minutes u? the test.

~

l

' During interviews with the three CR0s on Shift "D." three different explanations were offered for why the water additions were made. As stated above,-Mr. Colemaft said water additions were made with the intent of influencing the outcome of the test by taking advantage of the MUT level instrument error. ~

1 .

l

1 l

> i O he said he was not aware of this phenomenon procedural cautions not to add water during the test, Mr. Wright said he Despite the desc addedapproximately MUT-to water near the thesame endlevel of the test inatorder as existed the start toofincre the te Wright technique.said he believed calibration errors could be minimized by tests involving While Mr.the IW ightinitial and were final within two values inchesof (60 MUT levelMrfor the f gallons),

- . Wright was different from the MUT strip chart value.could not exp Mr. Olson could not recall why the water was added, but assumed e additions were necessary to keep the MUT level in the normal op or they were required to move control rods from their outer limit. A review not approached during any of the 13 tests.of the tests snows other hand could be a legitimate explanation; however, since five ot

' shifts were operating the plant under the same conditions and with t i controlThis shifts. rod is limits, not the such case.a pattern should also be present for the other In addition to Mr. Hartman five of the i3 operators interviewed by O

- including during leak rate Messrs. tests. Coleman and Wright have admitted adding hydr ,

level indication and consequently leak rate test Wright results.Coleman while admitting adding hydrogen during tests, stated he added hydroge ensure the hydrogen overpressure at the beginning and end of the test ~

the same calculation. in order to minimized instrumentation errors in In light of these admissions by two of the Shift "D" operators, the NR evaluation additions among ofthe leak tests rate approved testsbyshould Mr. Miller.have identified at least some Miller's t'es t. tests show definite indications of hydrogen being added There are several possible explanations for this occurrence.

only the licensee. a small portion of the leak rate tests that were run First.were re Second, the effect hydrogen additions would have on leak rate tests were subject to several variables including:

w&ter inadded hydrogen the instrumentation lines to cause a loop seal; the amount ofthe test hydrogen w;as added.the rate of the addition; and the time before the end o Third, because their are two separate level two level transmitters diffe~rently (i.e., the LT pro may not be influenced). computer may be affected while the LT provitii In sunnar March 28,y, the technical analysis shows that between September 30, 1978 and 1979,18 of the 25 tests (72% involving Mr. Miller included actions that were contrary to the precau)tions, limitations and r of Surve1 Hance Procedure 2301-301.

These actions included: ignoring test O conducted during unstable plant conditions or with a out-of-comission sensor providing input to the plant computer; an properly accounted for in the leak rate During test calculation. the last

. te

. three months of operation, where identified leakage was high and

- , , . - - , , - - - . . - , . , . - . _ , , , , , , - . .. , .- ,-.,m_ ,, ,,,,,wy .m.o  %, w- -m .%-y_,,,,..w,w,.,,--e,---...ne%w-,

. leak rate tests were more difficult to obtain, 12 out of 14 (86%) exhibited these characteristics. Based upon the high number of questionable or invali-O. tests, it appears that the tests were accepted as valid as long as the results were less than the technical specification limit, regardless of the ~

• evolutions in progress during the tests. In some cases, these' evolutions - -

were performed with the intent of wrongfully influencing leak rate test results.:_

III. Interview Sumary l During the period under investigation, September 30, 1978 through March 28, i

1979 Mr. to assigned Miller Shiftindicated "D."

thet he was a Shift Foreman at TMI-2, primarily l The licensed members of his shift included: Shift Supervisor Greg Hitz, and CR0s Dennis Olson, Mark Coleman and Lynn Wright.

The CR0s reported directly to Mr. Miller and Mr. Miller reported directly to ,

l the Shift Supervisor. The Shift Supervisor was cross-licensed on both TMI-1 l and THI-2 and ran the shift for both units. See pages 5 & 6.

Prior to relieving the shift, the Mr. Miller would receive a briefing from  !

the off-going off-going Shift Foreman and the CR0s would receive a turnover from the CR0s. i' Depending upon what evolutions were scheduled during the shift, there would be a short meeting with the Shift Supervisor, Shift Foreman, CR0s and at least some of the A0s to discuss what had occurred on the their previous shift. shift and to assure that everyone knew what to expect during Mr. Miller stated that he could not recall any specific issue related to the meetings. performance of leak rate tests being discussed during those jeepages7&8.

Mr. Miller where leakalso rate did not recall problems wereparticipating discussed. in any management level meetings 1 He did not remember receiving any '

written correspondence or instructions from the Supervisor of Operations, or -

any higher level of management, pertaining to problems associated with getting acceptable laak rate surveillance tests or associated with management level decisions related to correcting any procedural problems with the leak rate test procedure. je_epages8&9.

According shift. As to Mr. Miller, the Shift Supervisor was the lead person on the shift.

a result, Mr. Hitz had a large part in moving things along on the Mr. Hitz would not physically watch what operators were doing;,

however, he objectives would were met. set the objectives for the shift and assure that those Since Unit I had been in operation for quite sometime, Mr.

page 9.. Miller stated that Mr. Hitz spent the majority of his time at Unit 2. 5,ee e

2 Mr.

evolutions Miller stated that the policy for shift activities, including approval of wu set by the Supervisor of Operations, Mr. James Floyd. Mr.

Miller was,-asked to describe Mr. Floyd's approach towards managing the Operations Department, particularly in terms of what he expected from his supervisors and andevolutiuns.

other plant operators in terms of carrying out surveillance requirements Mr. Miller stated that Mr. Floyd was very knowledgeable about the proper operation of the plant and how operation was O pile up and sort of get shuffled to the side."related general, the attitude of " don't worry about the paperwork, just keep the Mr. Miller agreed to safety; that in however, "

l hardware operating" was imparted to the operators. 5,e_e,pages 10-12. ,

---

,,---,--.--a ~< ,n.,e.--..,,-~-.,,,n.n, ,,m-,-,, ,_ ,_,,,--.v-n-g,,p..,,wwy _ --wmw,.--,ap,gn

In addition to the Supervisor of Operations. Mr. Miller was asked to describe his interface with other management personnel, such as the Superintendent of Techtlical Support (Jim Seelinger until December 1978 and then George Kunder),

or the TMI-2 Plant Manager (Joe Logan). Miller did not recall much contact with Mr. Seelinger. He stated that he associated him mainly with TMI-1. He i

alsoldtd not have occasion to interface direct 1/ with Joe Logan. He stated that Mr.-Logan was new to TMI. As a result, Logan was splitting his time _ e i between his job as Plant Manager and the training program to become licensed i

on Unit 2. Miller did not think he could do both tasks properly and therefore, Miller did not believe that Logan every really "got up to speed" on the actual operation of the unit prior to the accident. With respect to George Kunder, Miller stated that Kunder was much more aware of all the operations associated with the unit than either Logan or Seelinger.

Mr. Miller could not recall any particular suggestions, directions or recomendations that he received from any of these individuals that may have-impacted 12-15.

on the way leak rate surveillance tests were performed. S g pages After reviewing copies of TMI-2 Technical Specification 3/4.4.6.2, " Reactor Coolant. System Operational Leakage " and Surveillance Procedure 2301-3D1,

" Reactor Coolant System Inventory," Mr. Miller was asked to describe how leak rate tests were performe~d in 1978/1979 at TMI-2. Mr. Miller stated that first the operator would check the status of the plant. If it was relatively stable, the operator would go to the plant computer and input the required 3 command to start the test. The computer would then print a series of i

questions including the desired time interval that the leak rate test would o be run. This was usually one hour. Once the operator answered the questions, the computer.the test would begin and the initial set of data would be taken by After one hour, the computer would obtain its final set of t

- data and again ask a series of questions including whether there were any operator-caused changes, such as water additions. After responding to the questions. the computer would print out the initial and final values of key plant parameters and.the leakage values (e.g., gross leak rate, total identified leak rate and net unidentified leak rate). See page 16.

With three CR0s on watch, the panel operator would not usually be the individual conducting the leak rate test; however, he would normally be j

informed that the test was being conducted by the operator running the test.

If water additions were made to the MUT the panel operator would nonnally be the one to make these additions. While this was the general the operators could run the test or make the water additions. practice, In orderany to of e

properly compute leak rate, any water added to the MUT had to be accounted for prior to the computer performing its final calculations. If a water 1 addition was made Mr. Miller stated that the amount of the addition would be l

obtained was from the batch controller. He recalled that the batch controller very accurate.

If the batch controller was already reset to zero, the

. operator performing the test would either ask the operator who made the addition for the total or he could look in the CRO's Log. Generally all log entries were made by the panel operator. Se_e e pages 17-19.

Mr. Miller recalled even though the technical specifications required that a i

O leak rate test be run at 'least once every 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />, he recalled that they were actually run every shift. The operators did not need to be directed to perform the test; it was part of their shift routine. Because the~ tests were run much more frequently than were required, they were not considered a -

i

_ _ . _ _ _ _ _ _ _ _ _ _ _ ~ . _ _ - _ _ . _ _ _ _ _ _ _ _ . _ _ . _ __. -___4

surveillance test per se.

i O If a test was run and the result was within the allowable limits of the technical specification, it would be considered a valid test and the next test would not be required for another 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />. If the test did not meet the requirements of the technical specifications, M.

was dot considered a surveillance test as long as they had not exceed the 72 hour8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> yindow since the last good test. See pages 19-21.

1 ,

Mr. Miller did not recall how that interpretation of the technical specification came into existence. He recalled it being a fairly comon understanding in the Operations Department that had existed for years. While he did not know who in management know of this practice, he stated he would i

be practice.

surprised if people like Jim Floyd and George Kunder were not aware of the He did not know if other managers such as Jim Seelinger or Gary l

Miller (Station Manager) would be aware of this practice; however, since they met quite often with the Shift Supervisors, he believed they would have j plenty of opportunity to be aware of it. Sel pages 20-24.

1 As best he could recall, Mr. Miller did not remember any other surveillance tests being treated in this manner. He believe that leak rate tests were treated differently because they were considered a routine task rather than a

] technical specification surveillance.

Mr. Miller stated that other methods were used by the operators to check for leakage in the RCS such as

! containment particulate monitors and containment sump inventory and discharge; however, the inventory balance test was the only method that could

{ be used to quantify leakage. See,pages 24 & 25.

When asked if he thought the leak rate test was a valid indicatoi of leakage, he could recall some question in his mind about about the tolerance associated with the calculated number; however, between activities associated

' with startup, testing and early comercial operation coupled with his feeling that the test was a vary routine task, he never really pursued the issue. He See reca-11 did pages 25their

& 26.being a considerable amount of variability in test results.

Tests results essentially that exceeded the unidentified leakage limit of 1 gpm were ignored. He could not recall specific examples of his operators bringing him test results that exceeded the technical specifications, but he believed that it occurred. He stated that if that did happen, he would have probably given the operator instructions to run another test and discard the bad one. He did not remember discussing this practice with his Shift Supervisor, but believed whatever he did was probably very consistent with what Mr. Hitz would have done. Mr. Miller could not recall any of the operators on his crew expressing concern about the way leak rate surveillances were being handled. Se_e e pages 27 1'28.

In reviewing leak rate tests Mr. Miller believes he may have given them a very c'Trsory look before signing his approval. He felt that he may have -

essentially just looked at the bottom line numbers and nothing else. As long as it did not exceed the technical specification, he would approve it. If it exceedet the technical specification, he would throw it away and have the operators run another test.

O paperwork near the end of his Heshift.

recalled that he reviewed most of the He had several things, in addition to l leak rate tests, that required his review, including: Shift and Daily Checks, the CR0's Log. the normal hourly log readings and about four or five A0s' Logs.

l I

See pages 28-30.  ;

l Mr. Miller was advised, that Mr. Olson had testified that after an incident

'. in October 1978 (Note: this incident will be discussed later), he never threw ,

away a bad leak rate. Olson said he would run another test or tests until he i got a good result and then submit the results to his Shift Foreman as a s package (e.g., good test on top with the bad tests stapled to.it). Mr.

Miller _ did not remember Mr. Olson doing that. Mr. Miller was also informed--

i

' that Mr.:Coleman testified once he had turned in a leak rate test greater than I gpm to his Shift Foreman and that when the Foreman showed it to his- ,

Shift Supervisor, he received a very negative reaction. Mr. Coleman said it 4 was clear from that incident, that his Supervisor only wanted to see good results. Hr.' Miller did not recall such an incident. He did recall seeing tests that exceeded the technical specification and he would consider them t invalid. The only reason for considering them invalid was because they l exceeded the technica! specification limit. There was re conscious l j

rationalization applied, that was just the practice. Not'e: Leak rate test 12B, perfomed by Mr. Coleman, exceeded the technical specification limit and a

was retained by the licensee but not signed by Mr. Miller, jtepages30-37, Mr. Miller agreed that the technical specifications are related to safety and i that as a Shift Foreman, it was his responsibility to ensure compliance with l the technical specifications. He also agreed that since the leak rate {

surveillance test was the only method to quantify the amount of leakage, it was the only method that could be utilized to shown compliance with the l

1 technical specifications for leakage. He was then asked if he received a test result that indicated unidentified leakage was greater than 1 gpm, how he would know that leakage was not greater than the limit. Mr. Miller o stated: "Looking back on it today, there is no way. I don't know." jee, pages 37 & 38.

Mr. Miller was asked if he had any reason to conclude that plant management had made a conscious decision to ignore the surveillance test in order to keep the plant operating rather than having to enter the action statement and i

possibly shutdown the plant. Mr.. Miller stated that the only infomation he wasawareofalongthatlinewasaquoteherecalledhearingduringtheGrand

Jury investigation to the effect that Mr. Floyd considered the leak rate tests that were filed just as invalid as the ones that weren't filed." From that Mr. Miller assumed that Floyd considered the leak rate tests useless.

But he could not say that the tests were ignored by Floyd just to keep the plant running. je_e,pages38-40.

4 Mr. Miller was asked if the operators knew that Technical Specification 3.4.6.2 required that the leakage be reduced to within limits within four hours or they were required to commence shutting down, if the unidentified l

1eakage was greater than 1 gpe. Mr. Miller stated that he believe that was i

understood. He said the way he remembered it was that the 1 gpm was not only the triggering point for entering the action statement of the technical i

! specifications but it was the triggering point for entering the emergency plan as an unus,ual event and notifying the NRC that an action statement had .

j i .

.atered into. He believed that people may have tended to look the other x snd-ignore tests that had results greater than I gpm because they did not

. to trigger those actions. See pages 40-45.-

A copy of LER 78-62/1T along with an attached routing sheet with the typewritten names and handwritten initials of licensed operators was l

provided to Mr. Millar for his review. The routing sheet was used to -

1raicate which individuals had seen the LER. Although Mr. Miller's initials were on the routing sheet, he did not have a preaccident recall of  !

the incident described in the LER. See,pages 45 & 46.

f _

O Note:

During a routine inspection of TNI-2 operations on October 18 1978, and NRC inspector discovered several bad leak rate tests lying in the

_ ccontrol room and that TMI-2 had been operating for an extended perted

- of time with unidentified leakage exceeding the technical '

" specification limit. The incident resulted in the submittal of ~ '

Licensee Event Report (LER) 78-62/1T.

~

The LER states in part: "Thiseventwascausedbyamisinterpretatkon of the requirements of the technical specifications....The appropriatt personnel. will be instructed on the requirements of applicable sectior of the T.S. and the requirements to famediately invoke applicable actions statements when the provisions of the LCOs [ limiting conditions for cperation) are not met."

Mr. Miller stated that he did not recall being advised or instructed that they had been misinterpreting the technical specifications, consequently he did not provide any revised instructions to his CR0s. Mr. Miller did not recall ever changing his interpretation of when the action statement should be invoked. He did not believe there were any changes made to leak rate testing practices as a result of this LER. ge,pages45-50.

Mr. Miller was infomed that the testimony provided by Messrs. Coleman and Wright were similar to his; however, Mr. Olson testified that he did recall the incident and that he was instructed in the proper interpretation of the requirements of the technical specifications. Mr. Miller stated that it was

+

O week and may have learned of the incident fromMr.

said that if Mr. Olson had that knowledge. Olson had not discussed it with that shift.possible Miller also

, him. Miller did not recall Mr. Olson ever coming to him with a leak rate 4

test greater than 1 gpm and saying to Miller that based on his interpretatior.

of the technical specifications they had only four hours to identify the source of the leakage or else they were required to shut down. Mr. Miller did not ever recall invoking the action statement of Technical Specification 3.4.6.2. hepages49-51.

3 Mr. Miller was asked. if at some point after this LER was issued, it was emphasized to him by his Shift Supervisor or any member of plant management to instruct his operators to ensure that they got rid of bad leak rate tests so that the NRC would not see them. Mr. Miller did not believe that was the reason unacceptable leak rate tests were thrown'away. He said that they were discarded because they were not going to be doing anything with them in the future. It was something that had been going on for years. See pages 51-54.

The next portion of the interview with Mr. Miller centered around Hartman's allegations that leak rate tests results were manipulated by operators and that pressure was being applied to o'perators to get good leak rates. See page 55.-

Mr. Miller stated that until about a year or two 'after the accident, he was' not aware of discussions or of the possibility that you could add hydrogen te t the MUT towards the end of the test in order to affect the leak rate test results. He stated that he had never observed the additiorr of hydrogen having any effect on MUT level indiction. Mr. Miller was advised that Coleman had testified that this phenomenon 'was comon knowledge among Shift 1  % men and Shift Supervisors. Mr. Miller did not agree. He said ti. 'ere,

_ _ _ _ = _ _ _ _ _ _ _ ___

i may have been some who had this knowledge, but he did not believe it was a widely known phenomenon. He said that from the discussions he had with individuals after the accident, the majority of people he talked with were as surprised as he was that hydrogen additions would have that effect. '

He did nof have any discussion with anyone who said that they knew of this - '

phenomenon-prior to the accident. Mr. Miller stated he had no personal knowledge'at that time that anyone on his shift manipulated leak rate tests _ -

, with the addition of hydrogen. See pages 56-59.

Mr. Miller stated that prior to the accident he was neither aware of nor had any suspicions of the operators on his shift using any mechanism to assist them in getting good leak rate tests results. He was not aware until after the accident that adding water to the MUT could produce the same effect on MUT level indication as a hydrogen addition. He said when he first heard about it in 1980 or 1981, he was surprised that the effect was that pronounced. He was also surprised that Mr. Coleman had been aware of the effect and used that mathed to assist him in getting a. good leak rate test.

See pages 59-61.

The next portion of the interview concentrated on a detailed discussion of individual leak rate tests performed by Shift "D" and approved by Mr. Miller.

l It was pointed out to Mr. Miller that compared with the other five shifts.

Shift "D" had the highest percentage of questionable tests during the last three months of operation at TMI-2. This was the period when the plant

! experienced a large amount of identified leakage from the top of the pressurizer and that as a result, satisfactory leak rate test results were l

more difficult to obtain. During this period his shift was the only shift

' that day after day was able to.obtain a satisfactory leak rate test. In almost all cases this was accomplished by the addition of water to the MUT during the last few minutes of the test, thus taking advantage of the phenomenon discussad above. It also pointed out to Mr. Miller that this was

' riot only something that could be seen from a technical analysis of test  !

results, it was also confinned by one of his CRos-Mark Coleman. In response '

Mr. Miller said, "I can state with one hundred percent fact that I had no knowledge that changing -- addin manipulate the leak rate tests."g hydrogen S g pages or adding 61-67. water to the MUT would Mr. Miller stated that he had reviewed all of the leak rate tests performed by his, shift recently and he did not see any legitimate operational reason why, day after day, their shift had to add water five to ten minutes before the end of the test. Mr. Miller stated that if operators were taking advantage of instrument errors by adding water and using~ the number off the i

batch controller as the amount of the addition, knowing that the computer was going to read the water level off the MUT level instrument, they were manipulating test results. When asked if he felt that way because of the situation today, Miller responded by saying: -

Definitely for today's situation that is definitely the case, and as far as ! remember tt was just not my nature t6 dtr things like that intentionally. I would think that even back hen I would have considered that manipulation.

See pages 69 & 70. . .

! Mr. Miller also agreed that performing a test with an instrument that is '

i

._.,,,- ,, -,-+.----,r,.-, --,.._--,.-_,-.w.- r--,_ ._--.--_e_-.--.r--,,,---.~m-----,,,--y--,-.~rw,--.--,-.4,-mm---y--v

O inoperable or out-of-service to calculate a test result would also be considered manipulation and would provide a meaningless result. Mr. Miller was asked if he payed very much attention to the performance of leak rate tests.. He responded by saying: '

Well, like I said earlier with regard to reviewing this, it

' sounds like I was very -- like I wam't doing my job very _

well when you focus it in on one aspect like this. I don't remember really getting too involved ever in the conduct of a leak rate test. It may have been done during the shift and I would see that the sheet -- sometimes during the shift,

(

towards the end of the shift, it just seems to me like I was doing other things. I was out in the plant. I just don't have i

too much recollection on really being too involved with them.

See pages 70 & 71.

Mr. Miller was infomed that A

' he th'ought that prior to the accident ne was marv ensitive to procedural 71 & 72. compliance than most of the other supervisory people. je_epages Test #12B (10/17/78) was reviewed with Mr. Miller. It was pointed out that based upon the slope of the MUT strip chart it was. apparent that the plant was experiencing significant leakage. The test showed unidentified leakage of 2.07 gpm. After reviewing the data 3 Mr. Miller stated that looking at it today, he could see that there was no basis for considering the leak rate

! test invalid. See_ pages 73-77.

i Mr. Miller was asked if during the period October 1978 through March 1979 he was aware of any trouble with the MUT level transmitters. He said that based upon his recent review of the leak rate tests, he was aware of problems with LT-1 and he assumed that he would have been aware of.such problems at the time. He stated that if he was . aware of such problems, he would have informed his operators not to use the inoperable ' channel as an input to the computer for performing leak rate tests. Mr. Miller was shown several examples of leak rate tests perfonned by members of is shift, where both the unstable LT and the stable LT were used at differen ;imes to provide input to the computer. Mr. Millar agreed that test results using LT-1 could not be relied upon. See pages 77-87.

Several examples of tests involving water additions during the last few minutes of the test period were also discussed with Mr. Miller. Again, Mr.

' Miller _ stated that he could not see why these water additions were required and he agreed that the amount of water added according to the MUT level indicator was higher than the amount logged in the CR0's Log and manually included in the test calculation by the CR0s. See pages 87-92.

After reviewing these tests, Mr. Miller stated that while it appears fairly obvious now that his operators may have been taking advantage of

1 perturbations in the MUT level transmitters, he reiterated that he had no '

knowledge of these actions at the time. He had no suspicions that this was occurring on his shift or any other shift. He had no conversations with s c other-Shift Foremen who indicated suspicion on their part that operators-on 1

their shifts were doing this type of thing. See pages 93-96.

Mr. Miller was asked, if operators were not directed by their supervisors'to manipulate test results, why would they do it on their own. Mr. Miller could only speculate that it may have involved some misguided pride, competition or one-upmanship between shifts (e.g., they may have felt they could get a good leak rate test on their shift while the other shifts could not). S u pages97-100.

t j IV. Findings and Conclusions ~

During the six months prior to the accident at TMI-2, while Mr. Miller served i as Shift' Foreman for Shift "D." many of the practices associated with '

conducting RCS leak rate surveillance tdsts violated approved procedures and 4

the TMI-2 Technical Specifications. Based upon the NRR technical analysis of leak rate tests during this period coupled with the 01/NRR interviews of Mr.

l Miller and the three CRO's assigned to Shift "D" (Coleman, Wright and Olson),

the following findings and conclusions are drawn. t

, 1. TMI-2 Technical Specification 4.4.6.2 required that RCS leakages be demonstrated to be within limits b "O . . inventory balance (leak rate test)y the at performance

~ teady state operation. TMI-2 Technical Specification 3.4.6.2 limited s

least once perof72 a RCS hourswater during -

i unidentified leakage to 1 gpm. On Mr. Miller's shift, leak rate tests i were routinely run at least once per shift, plant conditions and

, evolutions permitting.

2. TMI Administrative Procedure 1012. " Shift Relief and Log Entries,"

required the title, number, start and completion times of all periodic tests required by the technical specifications (i.e., leak rate tests)

! be logged in the CRO's Log. Contrary to this requirement only i

- satisfactory (i.e., unidentified . leakage less than 1 gpm) leak rate 1 tests were logged in the CRO's Log by Shift "D" personnel.

1 l l 3. Because Shift "D" performed leak rate tests more frequently than '

i required by the technical specifications, only satisfactory tests were considered surveillance tests. Tests with results in excess of the technical specification limits were not considered surveillance tests i

and were discarded. This action was contrary to TMI-2 Technical

! Specification 6.10.1.d. " Record Retention," which required that -

'~

records of surveillance activities required by the technical

-specifications be retained for a period of at least 5 years. Mr.

Miller was not sure how this practice originated; however, it had i ,

existed for years and was comon practice on all shifts.

! . 4 Because leak rate tests were run so frequently 'the CRos on Shift "D" conducted these tests on their own as part of their shift routine. If i tests results exceeded the limits of the technical specifications, the e

CR0s frequently discardert tests on their own. If an unsatisfactory test was provided to Mr. Miller, he would personally discard the test j and instruct the CR0s to perform another test.

I

_- . -,.,-_-__-,.,-_________..,_.___,-_.-.-,_m_ _

_ - _ . - _ _ ~ - . . - , _ - . - . - _ - . _ . . _ -,,e

5.

O In reviewing before signingleakhis rate approval. tests, Mr. Miller gave them a very cursory loo line numbers. As long as the results did not exceed the technical He essentially just looked at the bottor specifications, he would apiprove the test.

~ s

~

61

=

In general, Mr. Miller believed that prior to the accident, the ~'

general attitude of " don't worry about the paperwork just keep the hardware operating" was imparted to the operators by the Supervisor et ' '

Operations, Mr. James Floyd.

7.

Surveillance Procedure 2301-3D1 " Reactor Coolant System Inventory,"

required the operators to enter the action statement of Technical Specification 3.4.6.2, if unidentified leakage exceeded 1 gym. The action statement required leakage to be reduced within limits within 4 ,

i hours or be in at least hot standby within the next 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> and in  !

cold shutdown within the following 30 hours3.472222e-4 days <br />0.00833 hours <br />4.960317e-5 weeks <br />1.1415e-5 months <br />. Mr. Miller believr.d that he and his CR0s were aware of these requirements at the time.

Nevertheless, as long as a satisfactory leak rate test was obtained

within 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> of the last satisfactory test, the action statement was not invoked, regardless of how many unsatisfactory test results were received in the interim. This policy of accepting good tests and ignoring bad tests was also practiced by all shifts. -

8.

hr. Miller stated that unidentified leakage of 1 gpm was not only the triggering point for entering the action statement of the technical specifications, but it also required notification to the NRC and was " ~ "

the event.

triggering point for entering the Emergency Plan as an unsual l and ignore He believed tests thatthat hadpeople results may have tended to look the other way not want to trigger those actions. greater than 1 gpm because they did I

9.

i Although Mr. Miller had initialed the routing sheet for LER 78-62/1T, he did not have a preaccident recall of the incident. Contrary to i statements in the LER that the event was caused by a misinterpretation i of the requirements of the technical specifications and that operators l would be instructed to invoke the action statement when the limiting condition for operation was exceeded. Mr. Miller did not recall being

- . advised or instructed in this matter. Consequently, he neither changed his interpretation of when the action statement should be ~

invoked nor provided any revised guidance to the operators on his shift. He did not believe there were any changes made to leak rate 3

4 testing practices at TMI-2 as a result of this incident.

10.

Mr. Miller stated that he was not aware until after the accident that i adding hydrogen to the MUT towards the end of a leak rate test could affect MUT level indication and consequently leak rate test results.

j He dfsagreed with Mr. Coleman's testimony that this phenomenon was comen knowledge among Shift Foremen and Shift Supervisors. Mr.

Miller stated he had no personal knowledge at the time that.anyone on his Oiift manipulated leak rate test results by the addition of hydrogen.

11.

Mr. Miller stated that he was not aware until after the accident that adding water to the MUT just prior.to the end of a leak rate test

4 could as produce adding the same effect on MUT level and leak rate test results hydrogen.

of the leak rate tests performed by his shift.After the accident, Mr. M He could not see any legitimate operational reason why, day after day, his shift had to ad water to the MUT five to ten minutes before the end of the test.

_ M411er agreed that if his operators were taking advantage of M r.'

~~ '

_ instrument errors by adding water and using the number off the batch controller as the amount of the addition, knowing that the computer __r weregoing was to read test manipulating the results.

water level off the MUT level instrument, they

12.

i After reviewing Shift "D" leak rate tests, Mr. Miller stated that while it appears fairly obvious now that his operators may have been taking advantage cf perturbations in the MUT level transmitters to influence the outcome of leak rate tests, he stated he had no  !

knowledge of this activity at the time. j this was occurring on his shift or any other shift.HeHe had no had no suspicions that i conversations with other Shift Foremen who indicated suspicion on their part that operators on their shifts were performing these actions.

13.

Mr. Miller did not believe that if operators were manipulating test results it was caused by pressure from their supervisors. He could only speculate that it may have caused by some misguided pride, competition or one-upmanship between shifts.

!~

[ In sunnary, Mr. Miller agrees in retrospect that many of the- actions involving leak rate surveillance testing at TMI-2 violated approved procedures and were contrary to the THI-2 Technical Specifications. plant Mr.

Miller was aware that as Shift Fcreman it was-his responsibility to ensure his shift's compliance with the technical specifications.

j Mr. Miller also i

ag.rees that there is strong evidence, including the testimony of Mr. Coleman, 1 to indicate operators un his shi(t were performing actions that were contra i

to intent ahd/or spirit of the leak rate test surveillance procedure and that I

these actions were done with the intent of influencing or manipulating the

outcome of leak rate surveillance tests. Nevertheless, Mr. Miller states

, that prior to the accident, he had little involvement in the perforinance of leak rate tests and had no involvement in or knowledge of operators on his t shift intentionally manipulating the outcome of test results.

i i, The results of the technical analysis indicate that of the six shifts

! had the highest percentage (g35) of " questionable" The vast leak r j majority of these questionable tests show water additions to the MUT durin ,

the last few minutes of the test, resulting in a calculated leak rate that was smaller than the actual leak rate.

three CR0s as to why the water additions were madeDespite conflicting testimony by the weight of the evidence wo,uld indicate that these additions were ma,de to intentioj influence the outcome of leak rate surveillance tests. While Mr. Miller was l

not directly involved in the actual conduct of the tests and his statements i that he did_not order or direct the manipulation of tests are credible and are supported by the operators.' testimony, it is not plausible that Mr.

Miller shift. was unaware that leak rate test manipulations were taking place on his l

l l -

l