ML20207Q284

Proposed Findings of Fact & Conclusions of Law of Numerous Employees Vol One.*

ML20207Q284
Person / Time
Site:	Three Mile Island
Issue date:	01/23/1987
From:	Mcbride M GENERAL PUBLIC UTILITIES CORP., LEBOEUF, LAMB, LEIBY & MACRAE
To:
Shared Package
ML20207Q285	List: ML20207Q284 ML20207Q373
References
CON-#187-2285 LRP, NUDOCS 8701270123
Download: ML20207Q284 (287)

Text

- ______ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ - _ _ _

Yk W I "4"'

'87 JAV 23 p4 :26 UNITED STATES OF AMERICA bcL ,

NUCLEAR REGULATORY COMMISSION '

BEFORE THE PRESIDING BOARD I In the Matter of

)

INQUIRY INTO THREE MILE ) Docket No. LRP ISLAND UNIT 2 LEAK RATE )

DATA FALSIFICATION )

I )

PROPOSED FINDINGS OF FACT AND CONCLUSIONS OF LAW I OF Tite NUMEROUS EMPLOYEES VOLUME ONE Of Counnel: LeDOEUF, LAMD, LEIBY & MacRAE Suite 1100 IIARRY 11. VOIGT 1333 Now llampshiro Avo., N.W.

MICilAEL F. McBRIDE Washington, DC 20036 ROBERT ST. JOllN ROPER (202) 457-7500 I MOLLY S. BOAST JAMES W. MOELLER MARLENE L. STEIN C. CilRISTOPitER SPRAGUE SMITil B. GEPilART KILLIAN & GEPl{ ART JANE 0. PENNY 216-18 Pino Stroot TERRENCE 0. McGOWAN lla r risbu rg , PA 17108 (717) 232-1851 I Attornoys for the Numorous Employoen January 23, 1987 I

0701270123 f170123 I PDR 0 ADUCK 0D000:120 PDH I ,

I .

I I

I UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION BEFORE THE PRESIDING BOARD I In the Matter of

)

)

I INQUIRY INTO TilREE MILE ISLAND UNIT 2 LEAK RATE

)

)

)

Docket No. LRP I DATA FALSIFICATION

_)

)

PROPOSED FINDINGS OF FACT I AND CONCLUSIONS OF LAW OF Tile NUMEROUS EMPt.0 VEES VOLUME ONE Of CounnoI: LoBOEUF, LAMD, LEIIlY 1. MacRAE I llARRY l1. VolGT MICliAEL F. mci 1 RIDE Suite 1100 1333 llow flampnhiro Ave., N.W.

Wanhington, DC 20036 RollERT ST. JOllN ROPER (202) 457-7500 MOLLY S. Il0AST I JAMES W. MOELLER lif S bP L SPRAGUE SMITil 11. GEPifART KILLIAN f, JEPilART JAllE G. PENNY 216-111 Pino Stroot I TERRENCE G. McGOWAN litt r r i n bu rg , PA (ill) JJA-LUbl 1710 tl I A_t t o r noyn_f ontipL unp limp,loyou n N r ou n January 23, 19117 I

I I

I TABLE OF CONTENTS PAGE I Volume I I Introduction............................................. 1 Background and Summary of this Recommended Decision...... 4 I. llistory......................................... 11 A. Genesis of the Leak Rate Issue.............. 11 i 1. Ilartman Allegations..................... il

2. I&E Investigation....................... 20 I 3. Paogro & Benson Investigation...........

4. Grand Jury Investigationn...............

5. Indictment of Metropolitan Edison 21 22 I Company...............................

6. Plea Agreement..........................

7. NUREG-0680, Supp. No. 5.................

22 24 26

8. OI Investigation........................ 29 I 9. NRR Investigation.......................

10. Stier Investigations....................

a. TMI-l Investigation.................

30 33 33

b. TMI-2 Investigatlod................. 34 11 . Inntitution of Thin Proconding.............. 36

1. Origin in TMI-L (Rostart) Procunding.... 36 I 2. I.ogislativo Naturn of Inquiry...........

3. The pruniding floard In To Make 37 Findinga of Fact and Conclunlonn I of Law, Not to Mako Rocommendationn for Future Procoodin95................

4. Fut.uro procoodings May llo Conductod 3ll I Under Cortain Circumntancon...........

5. The lloard Can Exonorato Individualn and Will Da 50........................

39 40

6. Identity of Partin3..................... 40 I 7. Identity of Other Witnennon.............

11 . Doncription at proconding and 42 Ifoarin9n.............................. 44

!!. Doncription of tho Evidenco..................... 47 A. Inventlyativo Reportn....................... 50 i 1. Exhibit it tillor Report................

2. Exhibl* 2: Fangro h llonnon Roport......

50 51

3. Exhlblt 5: Hitit floport.................. 52

4. Exhibit 6: O ! It u p o r t. . . . . . . . . . . . . . . . . . . . 53 3 5. Exhibit 7: NU R EG -0 6110, Supp. No. 5..... 51 I

I B. Other Exhibits............................s. 54

1. Exhibit 3: Portion of GPU Assessment Panel Findings........................ 54 I 2. Exhibit 4: TMI-2 Computer Log (October 27, 1978)....................

3. Exhibit 9: Pictures of the TMI-2 54 I Control Room..........................

4. Exhibit 10: Examplos of RO and SRO Licenses..............................

54 55

5. Exhibit 11: J. Moore Charts:

I Effects of Oscillations...............

6. Exhibit 12: Administrativo Proceduto 1036, Revisions 0 and L...............

55 55

7. Exhibit 15. Curricula Vitae of Dr.

6 Harrison and Mr. Colo................. 56

8. Exhibit 16: Training Department I Administrativo Memorandum.............

9. Exhibit 17: Excorpts from Instru-mont Out-of-Service Log...............

56 56

10. Exhiblt 18: RCIU Test of Fobruary 5 15, 1979 with CR0 Log and MUT Lovut Strip Chart..................... 57

11. Exhibit 19: TMI-2 Control Room I Layout Diagram........................

12. Exh1 bit 2L: Hottonhausen Affidavit.....

13. Exhibit 22: TMI-2 FSAR Excorptn........

57 57 58

14. Exhibit 23: TMI-2 FSAR Excorpts........ 58 5 15. Exhibit 24: Stier Lottor (and Attachments) to Hoard Dated September 30, 1986................ 58 I 16. Exhlhlt 25: mph I. utter to Stior, Dated September 30, 1986.......... 59 17 Exhibit 26 I flovembe r 3, 1986 ilRC Rogion !

Innpoction Report for Rochontor dan and Electric Corporation'n Ginna plant.... 59 I 18. Exhibit 27: 1.ottor Crom J. Mooro to W. Rockwoll, Dated October 31, 1986.... f3 0 C. Tontimonial Evidenco (flot Othatwino Included in Investigativo Ropartn)........ 60

!!!. Tho Standard to llo Applied to the Factual I lanuon in Thin procoodiny In Clear and Convincing Evidenco............................. 61 IV. Findinga On the Technical Innuon................ 69 A. Statutory and Rogulatory Hackg:ound......... 69

1. Tho Atomic Enorgy Act of 1954, An Amonded............................... 69 I .n.

I

I

2. NRC Regulations......................... 69

3. NRC Regulatory Guide 1.45 Was Explicitly Incorporated Into the liases For The I 4.

TMI-2 Technical Specifications........

NRC Licenses............................

73 77

a. TMI-2 Facility Licenne.............. 78 I b. TMI-2 Reactor Operator Licenses.....

c. TMI-2 Senior Reactor Operator Licenses..........................

78 79 I .

5. TMI-2 Technical Specifications Applicable to RCS Leak Rate Testing............................... 79

a. The Technical Specif: cations Required Performance of a Leak Rate Test at Least Once por 72 Ilours, During " Steady State" I Operation, in Certain Modes.......

b. The Applicable Procedure Did Not Prohibit Water or flydrogen Additionn 79 I c.

to the Reactor Coo:, ant System During a Leak Rato Test...........

Thu Loak Rate Technical 81 Specification Wan Ambiguous

' I as to the Required Frequency of Performance of the Tont........

d. Tho Applicablo Survotllanco 83 I Proceduro Wan Ambiguoun with Respect to Invalidation of A Loak Rato Tout.................. 84

6. Mot. Ed Adminintrative Proceduron Woro I Not Legal Risquiremontn Imponod on Mot Ed or The the Liconned Operatorn... 86 I H. The 1 Upm 1.C0 for Unidentiflod Loakago llad No Safety Slynificanco............. 91 I C. Performanco of the Leak Rato Tont...........

1. Frequoney of Tonts......................

2. Tont performanco Proceduto..............

91 93 93

3. Cntry of Idontifled Loakago and Othor I Information...........................

4. Control Room Organlaation...............

95 96

5. Logging of To4t Porforminco and Tont I 6.

Ronulta...............................

Itoview and Approvil of Tont Roaultn.....

7. Runtattinq of tho "72-Ilour Ctock".......

97 97 9H E D. Thoro Woro Numoroun Dofoetn in tho I, oak 3 Rato Tout................................ 98

1. f.ack of RCDT Donnity Componnation...... 99

2. Lack of Donnity Componnatton for 1

Additionn to the MUT................. 101 1 5 ai k(=

I

I

3. RCS Temperature Was Not Correctly Entered If Temperature Exceeded 582*F. 102

4. RCS Pressure Differences Were Not Accurately Taken into Account........ 104

5. Instrument Errors Necessarily Produced Inaccuracies in Test Results......... 105 I 6.

7.

RCS Oscillations Significantly Affected the Results of the Leak Rate Test....

TMI-2, Unlike other Babcock & Wilcox 107 I Nuclear Plants, Did Not Have an Evaporative Loss Factor in its Leak Rato Test Procedure.................. 109

8. A Possible " Loop Seal" Effect on the MUT Level Sensor Could Pave Produced i Erroneous Test Results............... 110

9. The Loak Rate Test Procedure Failed to I 10.

Use The Correct Reactor Coolant Pressure Boundary....................

The Leak Rate Test Procedure Failed to 114 I 11.

Account Accurately for Pressurizor Lovel Changos........................

The Leak Rato Test Proceduro Failed to 115 State the Correct Reactor Coolant I 12.

Drain Tank Volume....................

The Humorous Defects in the Leak Rato Test Proceduro Causod Test Results 115 Indicating Hogativo Unidentified i I,oakago.............................. 116

13. Conclun100............................. I17 E. Unidentified Loakago -- Actual Amounta vs. Test Results.......................... 118 I F. Technical Analysin of !. oak Hato Test 1.

Falsification............................

Moann by Which falsification May H,ive 124 I Occurred ............................

a. Unrecorded Water Additions to the MUT..............................

124 124

b. "Underrocotded" Watur Additions to I the MUT..........................

c. MUT Hytdogon Additionn.............

132 138

d. " Food and Blood" Uporationn........ 145 I o. MUT !.ovol Tranumitter Swltching....

2. The Gonoric Dufocts or Inadoquacion in tho Analynon Provent Advotno Conclu-146 nlons from lloing Mado Concorniny I Individual Honponnibility............ 148

a. Control Room Organir.ation Wan Such an to Procludo a Conclunion an I to the Pornon lionponniblo f or a Ulvon Tont..................... 148

-iv-I

I

b. Control Room Communications Were Not Always Such as to Permit the Conclusion that Operators I Knew of Plant Manipulations During Leak Rate Tests, or Knew that Leak Rate Tests Were Being Performed I c.

During Plant Manipulations.......

TMI-2 Was a New Plant with Ongoing Plant Oscillations That Affected 150 I d.

Leak Rate Test Results...........

MUT Level Strip Chart Time Error Precludes a Conclusion that 152 Operator Action Occurred During I e.

Certain Tests...................

MUT Level Transmitters Were At Times Inacurate but, When Not 153 I Tagged Out-of-Service, Could Have Been Used for Leak Rate Testing and Were Not Used to I f.

Falsify Results.................

Water Additions to the RCS Were Permitted by the Procedute and were 155 Often Required for Safe Operation I g.

or Maintenance of Necessary Power Levels.....................

Hydrogen Additions to the MUT Were 158 I Normal Plant Evolutions, Were Not Prohibited by the Leak Rate Procedures, and Were Not I

Shown to Affect the Test Results.......................... 165

h. It Is Difficult to Determine Accurately the Amount of Water I Additions or MUT Level Changes from the MUT Level Strip Chart... 168 V. Training of TMI-2 Operators On Leak Rate I Testing Was Woefully Inadequate; it Largely Consisted of Minimal On-The-Job Training..... 173 I A. Classroom Training Was Virtually Non-Existent............................. 173 I B. On-The-Job Training Essentially Consisted of a Demonstration of the Manner in Which to Perform the Leak Rate Test, I with Virtually No Guidance as to the Significance of the Test or the Actions to Be Taken If Test Results Were Excessive................................ 175 E ...

I - -

I VI. The " Inaccurate and Meaningless" Leak Rate Procedure Precipitated Operator Indifference To Test Results.............................. 179 VII. The TMI-2 Operators Never Adopted the Haverkamp/

Seelinger Interpretation of the Leak Rate Surveillance Requirement..................... 185 A. Leak Rate Tests Between October 15-19, 1978, 185 I B.

C.

Discovery of the Problem On October 18, 1978.....................................

NRC Inspector Haverkamp's Response.........

189 189 D. Mr. Seelinger's Response to the I E.

F.

Conversation with Mr. Haverkamp..........

Mr. Floyd's Response.......................

192 195 Mr. Fels' Actions.......................... 197 G. The Plant Operations Review Committee's I Actions.................................. 197

1. The October 19, 1978 Reportable I 2.

Occurrence Letter....................

The November 1, 1978 Licensing Event Report...............................

197 198

3. The PORC " Action Item" Instruction to the TMI-2 Operators and Supervisors.. 201 VIII. The NRC Exacerbated tne Problem in TMI-2 Leak I Rate Testing And Failed to Make Clear to TMI-2 Operators and Supervisors the Actions Required of Them..................... 204 A. The TMI-2 Leak Rate Technical Specifi-cations and Surveillance Requirement Were the Product of NRC Licensing Actions.................................. 204 B. Lack of Clarity in the Technical Specifi-I cations and Leak Rate Surveillance Requirements Confused the Operators....... 207 C.

I The NRC Caused the 1 Gpm LCO on Unidentified Leakage at TMI-2 to Be, in Reality, 0.72 Gpm................. 209 I D. Frequent NRC Inspections Not Only Failed to Correct Improper Leak Rate Test Procedures and Practices, They Perpetuated and Compounded Them.......... 211 5

I l -vi-I

I I Volume II I IX. The Parties' Involvement in Leak Rate Testing......................................

Introduction...................................

A. Charles D. Adams...........................

I B. Raymond R. Booher..........................

C. John A. Brummer............................

D. Joseph J. Chwastyk.........................

I E. Mark S. Coleman............................

F.

G.

William T. Conaway, II.....................

Joseph R. Congdon..........................

I H. Craig C. Faust.............................

I.

J.

James R. Floyd.............................

Edward R. Frederick........................

K. Leonard P. Germer..........................

I L.

M.

Carl L. Guthrie............................

John G. Herbein........ ...................

N. Gregory R. Hitz, Sr........................

I O. Kenneth P. Hoyt............................

P. Theodore F. Illjes.........................

Q. George A. Kunder...........................

R. Walter J. Marshall.........................

I S. Hugh A. McGovern...........................

T. Brian A. Mehler........................ ...

U. Charles F. Mell............................

I V.

W.

X.

Adam W. Miller.............................

Gary P. Miller.............................

Frederick J. Scheimann, Jr.................

I Y.

Z.

Bernard G. Smith...........................

William H. Zewe............................

X. The Non-Parties' Involvement In Leak Rate Testing......................................

I

A. John R. Blessing...........................

I B.

C.

D.

Kenneth P. Bryan...........................

Martin V. Cooper...........................

William J. Fels............................

l I E.

F.

G.

Harold W. Hartman, Jr......................

Earl D. Hemmila............................

John M. Kidwell............................

l g H. Joseph B. Logan............................

g I. Thomas D. Morck............................

J. Dennis I. Olson............................

I I

-V11-lI

I K. Mark D. Phillippe..........................

L. James L. Seelinger.........................

M. Lynn O. Wright.............................

XI. Answers to the Commission's Questions..........

A. Technical Specification Interpretation I and Implementation.....................

B. Difficulties Experienced by Operators With Leak Rate Test....................

C. Discarding of Leak Rate Tests..........

I D. Alleged Leak Rate Test Falsification...

Appendix A -- List of Exhibits..........................

Appendix B -- Table of Corresponding NRR and Stier Test Numbers for TMI-2 Leak Rate Surveillance I Tests Performed Between September 30, 1978 and March 28, 1979........................................

I Appendix C -- Indictment in United States v. Metropolitan Edison Company, Criminal No. 83-00188 (M.D. Pa.,

November 7, 1983).....................................

I Appendix D -- Excerpts from Change of plea and Sentencing of Metropolitan Edison Company in Crim.

No. 83-00188 (M.D. Pa., February 28 and 29, 1984).....

I I

I I

I I

I I

l

-viii-I

I UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION l BEFORE THE PRESIDING BOARD 1I

)

In the Matter of )

I INQUIRY INTO THREE MILE

)

) Docket No. LRP ISLAND UNIT 2 LEAK RATE )

DATA FALSIFICATION )

I PROPOSED FINDINGS OF FACT I AND CONCLUSIONS OF LAW OF THE NUMEROUS EMPLOYEES VOLUME ONE I nt roduc t- ion

1. This proceeding was commenced on December 18, 1985 by the Commission to develop the facts surrounding alleged reactor coolant system ("RCS") leak rate data falsification at Three Mile Island, Unit 2 ("TMI-2") prior to the March 28, 1979 accident. Inquiry Into Three Mile Island Unit 2 Leak Rate Data Falsification, CLI-85-18, 22 N.R.C. 877, (1985). The proceeding was intended to develop the evidence in sufficient detail to determine the ultimate status of those likely involved, which include those segregated from TMI-l and those now working at other facilities. In its order, the Commission explained that this proceeding was instituted as a result of its determination in the TMI-l (Restart) proceeding to resolve issues involving alleged TMI-2 leak rate falsification separately from the Restart proceeding. Id.

~

I I

I

2. The Commission decided to depart from its customary practice of holding adjudicatory hearings, and empowered this Board to conduct a legislative inquiry and to make findings of facts and conclusions of law in the form of a recommended decision. The Commission reserved for itself the decision as to what actions, if any, to be taken against involved individuals. Id. at 883-84.

3. This Recommended Decision is issued in response to the Commission's order. As set forth herein, we have compiled a sufficient record to make findings of fact and conclusions of law with respect to the issuns surrounding the performance of leak rate tests at TMI-2. We have exhaustively analyzed the technical aspects of the TMI-2 leak rate test, have heard from nearly all the affected individuals, and have carefully reviewed the evidence. This Recommended Decision answers each of the questions set forth in the Commission's order, id. at 880-81, to the extent that the documentary evidence, and the I memories of those involved, permit.

4. Before setting forth the background of the leak  ;

rate issue, describing this proceeding, and making findings on the technical issues and with respect to the potentially involved individuals, we must observe that this has been an extremely difficult inquiry. The passage of time since the events at issue has blurred the memories of those TMI-2' employees whose " knowledge of the facts [was] vital to the I

I

I development of the record." Memorandum and Order, Docket No.

LRP, p. 2 (February 14, 1986). The passage of time similarly affected the recollections of other witnesses as well.

5. In this respect, we share the frustration of Mr. Stier (the independent investigator hired by GPU Nuclear Corporation) in reconstructing events which took place eight or nine years ago and that, even in the best of circumstances, I would have been hard to determine. He stated:

I "One of the problems that we faced in this investigation was the difficulty in reconstructing events that occurred in 1978 and 1979 through witnesses who could not always I recall specific details of those events. The long period that has elapsed since that time, the traumatic effect of the TMI-2 accident on the I lives of the witnesses we interviewed, and the numerous intervening investigations, have made it difficult to obtain reliable testimony. For that I reason,. we placed heavy emphasis on the reconstruction of critical events from an examination of documents and engineering analysis of data obtained from plant records."

Exhibit 1-A (hereinafter referred to as "Stier Report"),

Vol. I, p. 8.

6. In a number of instances, these same factors have prevented the Board from making a finding that we might have been (ble to make at a time nearer the events in question.

I Where significant doubt exists, we have resolved the issue in favor of the affected individual. In these circumstances, where there are significant and realistic risks of actual injury to any individual about whom we make an adverse finding, I.

I

I see e.g. Memorandum and Order, Docket No. LRP, at 2, 11 (February 14, 1986), we believe that fairness requires no less.

I Background and Summary of this Recommended Decision

7. Mr. Harold W. Hartman, Jr., a control room operator ("CRO") at TMI-2 prior to the March 28, 1979 accident, alleged that reactor coolant system ("RCS") inventory balance tests (generally referred to herein as " leak rate" tests), were at times purposely manipulated and records of unacceptable tests discarded at TMI-2 prior to the accident to cover up the l fact that, over an extended period of time, the results of the

=

leak rate tests exceeded the limit for unidentified leakage.

CLI-85-18, 22 N.R.C. at 878.

8. More particulary, Mr. Hartman alleged that the computer program for calculating leak rates was unreliable, frequently yielding unrealistic results. This made it more difficult to get " good" leak rates (the word employed at TMI-2 to connote a result below the Technical Specification " limiting condition for operation" ("LCO") of 1 gpm for unidentified leakage). Mr. Hartman further alleged that the operators at TMI-2 sometimes manipulated the leak rate test results by inputting wrong data into the computer, adding hydrogen gas to the makeup tank ("MUT") during leak rate tests, adding water to the MUT during a leak rate test and not inputting the addition into the computer calculation, or " jogging" water (i.e., adding I

I I

I it in small amounts) into the MUT tank while performing operations involving other tasks. Mr. Hartman specifically alleged that shift supervision was aware of such improper conduct. He did not, and has not, alleged that TMI personnel outside the Operations Department knew about these alleged practices. After a preliminary investigation into those allegations, the Commission, in April 1980, referred the matter to the Department of Justice for criminal investigation. Id.

9. After years of federal grand jury investigations, Metropolitan Edison Company (" Met Ed"), the TMI-2 licensee at the time of the accident, was indicted on November 7, 1983. On February 29, 1984, Met Ed entered into a plea agreement with the United States which ended the criminal prosecution. Id. at 878-79. Met Ed pleaded guilty to one count of the indictment charging it with failure to establish, implement, and maintain an accurate and meaningful RCS inventory balance procedure to demonstrate that unidentified leakage was within the allowable limits. Id. at 879. Met Ed also pleaded no contest to six other counts of the indictment, including those which charged Met Ed with improper manipulation of TMI-2 leak rate tests to generate results which would fulfill the Company's license requirements. Id. The United States Attorney represented to the United States District Court for the Middle District of Pennsylvania that the evidence developed by the grand jury inquiry did not indicate that any of the directors and officers I

I

I of GPU Nuclear Corporation from its inception in 1982 (as successor to Met Ed), or any director of Met Ed, to the date of the indictment, " participated in, directed, condoned, or was aware of the acts or admissions that are the subject of the indictment." Id.

10. Following the plea agreement, the NRC (through the Justice Department) moved for access to the record of the I grand jury proceedings. The Court denied the motion. 594 F.

Supp. 117 (M.D. Pa. 1984).

11. The evidence adduced in this proceeding confirms that, because the leak rate test used at TMI-2 was highly inaccurate, leak rate tests depicting unrealistic values for unidentified leakage were quite common. TH 184-236, infra.

The evidence also shows that acceptable test results for unidentified leakage became more difficult to get in the two months or so before the March 28, 1979 accident, because of a major error in the TMI-2 leak rate test procedure that failed to take into account the actual temperature of water that was part of the RCS inventory balance but was not at RCS temperature (approximately 581*F). 1H 186-91, infra. That error was present throughout the time that leak rate tests were run at TMI-2, but the problem was much more significant during February-March 1979 because identified leakage was relatively

, high during that period. H 254, inera.

I I

12. The evidence of record failed to substantiate Mr. Hartman's allegations that leak rate tests were falsified by entering erroneous data into the computer or by adding water to the MUT during leak rate tests and failing to account for it. 4H 238-54, infra; Stier Report, Vol. I, pp. 83-87. The evidence does indicate that, in some cases, especially in February-March 1979, hydrogen gas was added to the MUT, or water added to the MUT and accounted for in the computer calculation, to affect leak rate test results. However, there was no prohibition against adding hydrogen to the MUT, or adding water to the MUT (if accounted for), during a leak rate test. Id., pp. 87-90; SH 150-55, 256, infra. Also, hydrogen and water additions (especially during February-March 1979),

because of their tendency to offset errors in the test, did tend in nearly all cases to make leak rate test results more accurate. TH 249-51, infra. Therefore, although adding water (if accounted for) or hydrogen to the MUT during leak rate tests may not have been commendable, those additions did not constitute " falsification".

13. The evidence shows that training on leak rate testing was virtually non-existent. Tr. 2225-26. Neither classroom training nor on-the-job training provided any substantive guidance on leak rate testing, 19 315-26, infra, leading to justifiable skepticism about the utility of the test. See 1 327-39, infra.

I

14. The evidence shows that there were numerous defects in the TMI-2 leak rate test procedure. See HH 184-224, infra. The TMI-2 test procedure was " inaccurate and meaningless", as the grand jury alleged and Met Ed conceded in its guilty plea. T 43, infra. There is no evidence that anyone knew of these specific problems with the test in 1978-79 prior to the TMI-2 accident (HH 315-26, infra), but the TMI-2 operators and supervisors did, by experience and common sense, draw the conclusion that the TMI-2 leak rate test was a " random number generator" rather than an accurate measure of unidentified leakage. Tr. 641. This conclusion, which virtually all of them shared, led many of them to shortcut the procedures that were applicable to the leak rate test and to run them successively during the 72-hour period contained in the Technical Specifications until they obtained a test depicting unidentified leakage below 1 gpm. They should have invoked the applicable " Action Statement" of the Technical I Specifications if unidentified leakage really was over 1 gpm, but it is unclear whether it ever was in excess of 1 gpm (except for the period October 16-18, 1978, which was duly reported to the NRC in a Licensee Event Report). See 1H 225-35, infra; Stier Report, Vol. V(C), Tab 29. Moreover, the procedures which the operators and supervisors violated I ""'" "^ *'"'" """" "' """'"""' ""' " '" " ' '"" '

I I

requirements imposed on Met Ed or the licensed operators and supervisors. E.g., Exhibits 10-A, 10-B; see 1 162-70s infra.

15. The NRC must bear some of the responsibility for this unfortunate episode. 11 369-89, infra. The applicable Technical Specifications were drafted by the NRC and are unclear and inconsistent with the regulatory documents which give rise to them. Critical phrases were left imprecise. For example, the term " steady state", which was meant to identify the status of the plant when leak rate tests should have been run, was not defined. See Faegre & Benson Report, Vol.

Three A, Exhibit 3.

16. Further, the leak rate test was established to measure "intersystem leakage", rather than unidentified leakage, by NRC Regulatory Guide 1.45, which is the genesis of the test procedure. 1H 140-44, infra. The RCS temperature compensation problem in the TMI-2 leak rate test should have been known to the NRC (because the NRC participated in the adoption of the applicable temperature to which the 1 gpm limit applied) and yet neither the NRC nor Met Ed licensing personnel evidently were aware that procedure measured water added or lost at a different temperature than the temperature of the RCS inventory itself, thus equating " gallons" of substantially different densities. (On March 16, 1979, a Temporary Change Notice ("TCN") to the leak rate procedure was adopted to correct the problem insofar as leakage to reactor coolant drain I

i tank ("RCDT") was concerned, but that was only a part of the problem. Kirkpatrick & Wermiel Prep. St., p. 21, ff. Tr. 376.)

17. The evidence also shows that the NRC Inspector at TMI in 1978-79, Mr. Haverkamp, suggested that Met Ed "round off" unidentified leakage values, then later rescinded the suggestion when overruled by his supervisors. 347-48, infra. While not a major mistake in and of itself, the suggestion to round-off, and its rescission, started a sequence of events which may have caused the cancellation of the only, even somewhat clear, guidance the TMI-2 operators and supervisocs ever received with respect to the need to invoke the " Action Statement" when the LCO for unidentified leakage was exceeded. HH 354-44, infra.

18. In view of all the evidence of record, we recommend that virtually all TMI-2 personnel involved in leak rate testing be exonerated of falsification of such tests. For those who may have been involved in questionable test practices, we recommend that they be treated leniently. It is understandable to this Board why most operators were so skeptical of such an inaccurate and meaningless test. It is also plain why they treated the computer-generated test results so casually; they simply did not have the right tool to do their job. Many years have passed since the events in issue, and we find that the involved TMI personnel have been taught a

I most valuable, if searing, lesson. It is now time to close the book on this endlessly-investigated subject.

I. HISTORY.

A. Genesis of the Issue.

1. Hartman Allegations.

19. Harold W. Hartman, Jr. became a licensed CRO at TMI-2 in October 1977. Tr. 2249. Shortly after the March 28, 1979 accident, an " agitated" Mr. Hartman appeared at the site and expressed concerns about TMI-2 that he had harbored, including an assertion that Met Ed had "put the reactor into commercial operation before it was ready." Exhibit 2 (hereinafter cited as "Faegre & Benson Report"), Vol. One, pp. 3, 5. Soon thereafter, Mr. Hartman informed Mr. James R.

Floyd, the TMI-2 Supervisor of Operations, that he intended to resign from his position as a CRO at TMI-2. Id., p. 5; see Tr.

2288. A few days later, on April 4, 1979, Mr. Hartman met with Met Ed management personnel and announced that he had reconsidered his planned resignation and wished to return to work. Faegre & Benson Report, Vol. One, p. 5; Tr. 2288. On April 5, 1979, Mr. Hartman nevertheless submitted his written resignation. Stier Report, Vol. VI(E), Tab H, 3/26/80 NRC I&E Interview, pp. 57-58. Subsequently seeking re-employment, Mr. Hartman then underwent psychological evaluation at the request of Met Ed, but as a result of that evaluation, Met Ed I I

I refused to rehire him. Faegre & Benson Report, Vol. One, p. 6; Tr. 2288.

20. A year after the 1979 TMI-2 accident, Mr. Hartman was interviewed about the accident on a television program and made allegations of serious misconduct on the part of TMI-2 licensed operators, including himself. See Stier Report, Vol.

VI(E), Tab H, 3/24/80 Transcript of TV Program, "What's Happening America". Among other things, he alleged that TMI-2 "was a lemon" and that he and other CROs had done "certain things" to manipulate leak rate tests so that the results would satisfy the 1 gpm LCO for unidentified leakage. An implicit theme of the telecast, although not necessarily of Mr.

Hartman's allegations, was that " deliberate tampering" with leak rate tests by the CRos contributed to the TMI-2 accident.

See id., 3/26/80 NRC I&E interview, pp. 31-32. (No investigation of leak rate testing practices at TMI-2 has concluded that those practices contributed to the TMI-2 accident, and the Government conceded at the time of Met Ed's plea that those practices did not contribute to the accident.

Stier Report, Vol. V(A), Tab 2, p. 2.)

21. The broadcast triggered an NRC investigation of TMI-2 leak rate tests. See 22 N.R.C. at 878-79. Within 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> of the telecast, Mr. Hartman, who had already been interviewed on many occasions in connection with the TMI-2 accident, see e.g., Stier Report, Vol. VI(E), Tab H, was I

I I

reinterviewed by NRC investigators. See % 35, infra. In that interview, and in a contemporaneous signed statement, Mr.

Hartman disavowed and attempted to clarify several of the statements he had made on the television program. Stier Report, Vol. VI(E), Tab H, 3/26/80 NRC I&E Interview, pp. 33, 40-47, 66-68. Nevertheless, he reiterated both his specific admissions as to how he had " fudged" leak rate tests by adding water to the MUT without accounting for it and by adding hydrogen to the MUT, and his general allegation that he had seen other CROs also do such things. d Id., pp. 19-30. However, he refused to name any other CROs alleged to have "actually tried some of these tricks . . . . Id., p. 27.

I 22. The telecast also precipitated numerous other investigations of Mr. Hartman's allegations, see 14 36-63, infra, including, of course, this proceeding. 22 N.R.C.

878-79. In interviews in connection with these other investigations, Mr. Hartman maintained that his allegations had I been distorted by the telecast, but that it was a fact that he and other CROs had falsified leak rate test results. For example, during the investigation by Faegre & Benson, see

'l 36-27, infra, Mr. Hartman testified one month after the program was telecast that it tended to " exaggerate' and

" glorify" his allegations, and took them out of context.

Faegre & Benson Report, Vol. Four, Tab 6, pp. 136-37. However, he nevertheless insisted that, because the TMI-2 leak rate test I

I I

I was considered erratic and unreliable by control room personnel, he and other unnamed CROs typically discarded test results in excess of 1 gpm, and commonly manipulated the tests by adding either hydrogen or unaccounted-for water to the makeup tank, all within the knowledge of unnamed supervisors.

M., pp. 143-71.

23. Mr. Hartman was also questioned about his allegations during a deposition taken in connection with a civil law suit in 1982. See Stier Raport, Vol. I, p. 9; M.,

Vol. VI(F), 7/16/82 and 8/18/82 GpU v. B & W Depo.3ition. He reiterated that it was common practice at TMI-2 for CROs to discard test results depicting unidentified leakage in excess I of 1 gpm which were believed to be invalid, and to manipulate the test by either adding water to the MUT without accounting for it or by adding hydrogen to the MUT. See, ea., M., pp.

82, 86-88, 90-91, 107, 161. However, in that deposition, Mr.

Hartman identified only one CRO who, in addition to himself, had added water and hydrogen to the MUT to affect test results.

Id., pp. 109, 139-41. Moreover, as to the only occasion when he actually saw this CRO add water to the MUT, Mr. Hartman was unable to say that the CRO did not propetly account for it in the test calculations. I_d . , pp. 261-63. Similarly, with respect to his statement that this same CRO had added hydrogen to affect the leak rate test result, Mr. Hartman testified that he could not recall actually having seen that CRO, or anyone I

I else, do it. I_d . , pp. 274-76. Furthermore, Mr. Hartman did not know whether other CROs on his shift besides himself knew that hydrogen additions were made to the MUT to affect test result s. I_ d . , p. 275.

24. Mr. Hartman admitted that he first informed the NRC of his allegations concerning leak rate test falsification only after he became aware that Met Ed would not rehire him in a job with unescorted access to the plant. Tr. 2289; see Tr.

5024 (no complaint to Mr. Floyd by Mr. Hartman about leak rate tests immediately prior to TMI-2 accident). Thus, Mr. Hartman's allegations concerning leak rate test f alsification at TMI-'2 did not surface until after the TMI-2 accident -- at a time when TMI-2 was not in operation and thus when no leak rate tests were being run at that Unit (or, likely, ever would be again). It is logical, therefore, to infer that Mr. Hartman's early, sweeping allegations concerning TMI-2 operators were motivated by anger at his ex-employer, Met Ed, and must be viewed skeptically in light of the evidence.

See Tr. 5022-23 (Floyd).

25. Although Mr. Hartman has retreated from his sweeping allegations, certain of his statements concerning general conditions and practices at TMI-2 prior to the accident are supported in the record. Consistent with the testimony of several operators, Mr. Hartman remembered that leak rate tests were most often performed on the mid-shift (i.e., the shift I

I from 11:00 p.m. to 7:00 a.m.). Tr. 2280; accord, Frederick Prep. St., p. 5, ff. Tr. 2447. It was also Mr. Hartman's experience that leak rate test results were erratic, and frequently did not reflect'actua1 p' nt leakage. Tr. 2281; accord, Conaway Prep. St., pp. 2-3, ff. Tr. 3097. Thus, over the course of several leak rate tests, some results indicated negative unidentified leakage, soma results depicted unidentified leakage of between zero and 1 gpm and other results depicted unidentified leakage exceeding 1 gpm. Tr.

2281; accord, Conaway Prep. St., pp. 2-3, ff. Tr. 3097. Mr.

Hartman recounted a practice of responding to an unsatisfactory leak rate test result by running another leak rate test, and discarding the unsatisfactory result. Tr. 2281-82; accord, Coleman Prep. St., p. 2, ff. Tr. 2579.

26. Mr. Hartman alleged that it was difficult to get a " good" unidentified leak rate of less than 1 gpm during 1978-79. Faegre & Benson Report, Vol. One, P. 7; see Tr.

2244-45. The testimony of numerous operators supports that assertion. E.g., Faust Prep. St., p. 2, ff. Tr. 2511.

27. The difficulty of obtaining satisfactory leak rate test results had several causes, according to Mr. Hartman.

During most of 1978-79, problems with the computer program used to run leak rate tests as well as plant oscillations resulted in leak rate test results that were highly variable. Tr. 2282; accord, Phillippe Prep. St., p. 2, ff. Tr. 4432. Like many I

I -1e.

I

I TMI-2 operators, Mr. Hartman believed that efforts were being made to enhance the accuracy of the leak rate test by correcting deficiencies in the computer program. Tr. 2282; accord, Cooper prep. St., pp. 5-6, ff. Tr. 2835. In addition, during the two-month period preceding the accident, Mr. Hartman observed that the leaking pressurizer valves made it difficult to get a satisfactory leak rate test result. Faegre & Benson I Report, Vol. One, p. 7. That belief of Mr. Hartman has been confirmed by the technical analyses that are part of the record in this proceeding. E.g., id., p. 26.

28. Mr. Hartman recalled that during the last several months prior to the accident, as many operators corroborated, it was impossible to add hydrogen to the MUT from the TMI-2 control room. Tr. 2285; accord, Hemmila Prep. St., p. 5, ff.

Tr. 4039; see Faegre & Benson Report, Vol. Two, Chapter Four, pp. 74-75. Rather, hydrogen had to be added manually by an auxiliary operator ("AO"). Tr. 2285. Mr. Hartman was not aware, during 1978-79, of any prohibition against adding hydrogen to the MUT during a leak rate test. Tr. 2286; accord, Cooper Prep. St., p. 7, ff. Tr. 2835,

29. In other areas of his testimony, however, Mr.

Hartman has continued to qualify and tone down his original allegations. In 1979-80, Mr. Hartman depicted leak ratt test falsification as a widespread practice involving many TMI-2 operators and many leak rate tests. As to the frequency with I

I which leak rate tests were falsified, Mr. Hartman asserted in an early interview that "[tlhat leak rate had to be fudged everytime we got, just about everytime that we got it, we had to do something to make it right." Stier Report, Vol. Vl(E),

Tab H, 5/22/79 NRC I & E Deposition, p. 15.

30. One month later, Mr. Hartman retreated from his initial allegation by stating that only "one" shift foreman, I "one" shift supervisor and "about five" CROs were aware of the difficulty of obtaining satisfactory leak rate test results and the methods by which acceptable results were obtained. Stier Report, Vol. VI(E), Tab H, 4/29/80 Faegre & Benson Interview, pp. 143-44.

31. Mr. Hartman's allegation as of the date of the hearing was that he had observed only one operator (Mr. Booher)

" jogging" water (i.e., adding it in small amounts) for the purpose of falsifying the leak rate test result. Tr. 2242, 2282. However, Mr. Hartman does not recall whether Mr. Booher accounted for the " jogged" water additions that Mr. Booher allegedly made in the test calculation (Tr. 2282), and hence there is considerable doubt as to whether Mr. Booher intended to falsify leak rate test results.

32. In 1980, Mr. Hartman asserted that adding hydrogen to the MUT to affect leak rate test results was a

" fairly common" practice among all CROs. Stier Report, Vol.

VI(E), Tab H, 4/27/80 Faegre & Benson Interview, p. 47. Mr. Hartman I

I noted that the addition of hydrogen to the MUT was the technique most often used an his shift to manipulate leak rate test results. I_d., p. 58. Mr. Hartman also made the specific allegation that there were "probably eight" such hydrogen additions during leak rate tests made by members of his shift in 1978-79. Id., p. 66.

33. Two years later, when asked whether he was aware of other CROs adding hydrogen to the MUT to affect leak rate test results, Mr. Hartman responded " Ray Booher, for sure. I don't knoy about anybody else." Stier Report, Vol. VI(F),

7/16/82 B&W Transcript, p. 109. He directly contradicted the statement he made in the Faegre & Benson interview by admitting I that "No, I cion' t know that [the addition of hydrogen to the MUT] was common knowledge among all operators, but I do know that there were several that I talked to that had used this method." Id., p. 121. Before this Board, Mr. Hartman reverted back to his original allegation that the use of hydrogen additions to affect the leak rate test result was common knowledge among TMI-2 operators. Tr. 2240.

34. In summary, Mr. Hartman began with a sweeping set of allegations that he himself admitted were distorted when presented on television. Some of his statements have been verified or corroborated by subsequent investigations. In other areas, he has retreated from or qualified his early allegations. At this point, the principal significance of I

I

I Mr. Hartman's 1980 allegations is that they served to focus attention on possible leak rate falsification at TMI-2 and precipitated various investigations thereof, including this proceeding. Our mandate from the Commission is not to determine whether Mr. Hartman was right or wrong, but rather to find the facts as best we can based on all of the evidence in the record. Accordingly, we make no ultimate findings concerning Mr. Hartman or the accuracy of each of his statements. Where rele rant, we have considered his testimony in our discussion of, and findings concerning, the individual responsibility of other TMI-2 employees.

2. I &E Investigation.

35. Immediately following a telecast of Mr. Hartman's allegations in March 1980, the NRC's Inspection and Enforcement

("I & E") Branch began an investigation. In March and April 1980, I & E conducted several interviews of TMI-2 operating personnel, including Mr. Hartman. See Stier Report, Vol. VI(E),

Tab H, 3/26/80 NRC I & E Interview. However, the Commission referred the matter to the United States Department of Justice and the latter assumed control of the investigation on April 28, 1980. See 14 38-46, infra. At the Justice Department's request, I & E suspended its own investigation and began to provide assistance to Justice's investigation. I & E I I

I never ccmpleted its investigation because of the referral.

Tr. 172; Stier Report, Vol. I, pp. 2-4.

I 3. Faegre & Benson Investigation.

36. At about the same time as the I &E investigation, on April 16, 1980, Met Ed, then the operator of TMI-2, retained the law firm of Faegre & Benson to conduct an investigation of Mr. Hartman's allegations. The firm's report was issued four months later, and it is part of this record.

See T 98, infra.

37. Faegre & Benson was charged by Met Ed to conduct an independent investigation, and Winthrop A. Rockwell, Esq.,

formerly Associate Chief Counsel to the President's Commission on the Accident at Threa Mile Island, was selected to head the investigative team. Faegre & Benson Report, Vol. One, pp. 12, 66; M., Vol. Four, Tab 6, pp. 3, 5-6. Relying primarily on plant records and other technical data, as well as some interviews with employees, Faegre & Benson found, in summary, that the TMI-2 leak rate test was unreliable because of a combination of " peculiarities" in the Technical Specifications, errors in the procedure and instrumentation, and system oscillation, M., Vol. One, pp. 17-26; that unidentified leakage probably did not exceed 1 gpm in fact for any significant length of time, id., pp. 31, 36; that invalid leak rate test results were apparently discarded, id., p. 20; and I I

I that evidence to support Mr. Hartman's allegations of widespread manipulation of leak rate tests by other Met Ed employees was inconclusive. d Id., pp. 37-49.

4. Grand Jury Investigations.

38. As part of the Justice Department's investigation of Mr. Hartman's allegations, see 1 35, supra, a Federal grand jury was convened in May, 1980 in Harrisburg, Pennsylvania to gather evidence of possible criminal violations involving

" falsification of certain safety tests at the Three Mile Island nuclear facility". Stier Report, Vol. I, p. 3; see M.,

Vol. V(A), Tab 3. However, no indictments were returned by the May 1980 grand jury.

39. In March 1983, four years after the accident at TMI-2 had shut the plant down, a accond federal grand jury was empaneled to continue the investigation of TMI-2 leak rate testing. Eight months later, Met Ed (but none of its employees) was charged by the March 1983 grand jury in an eleven-count indictment with criminal offenses allegedly arising out of leak rate testing practices and procedures at TMI-2 in 1978 and 1979. See M., Vol. V(A), Tabs 2 & 3.

5. Ind_ictment of Metropolitan Edison CompanL

40. The indictment charged Met Ed in Counts 1 and 2 with illegally " establish [ing], implement [ing) and I maintain (ing]" an " inaccurate and meaningless (leak rate test]

I I

I proceddre in an effort to generate results which appeared to establish that reactor coolant leakage was within allowable limits." Counts 3 and 4 alleged that Met Ed had " intentionally manipulated the [ leak rate tests] by the addition of water and hydrogen to the RCS during the course of such tests and in this manner generated test results which appeared to, but do not in fact, fulfill (Met Ed's] license requirements. In Counts 5 and 8, Met Ed was charged with unlawfully " fail [ing] to enter the Action Statement when the (leak rate testl procedure indicated that reactor coolant leakage from unidentified sources exceeded one gallon per minute []" and "fai1[ing] to notify the NRC that it had not" done so.

41. Met Ed was charged in Counts 6 and 7 with having illegally " destroyed, discarded, and failed to retain records relating to the (leak rate test] procedure." Counts 9 and 10 alleged that Met Ed " failed to make and maintain adequate operating logs relating to the conduct of the [1eak rate test],

including the starting and completion times of such testing and the results thereof," and "having failed to make entire identifying deficiencies and non-conformities encountered with its [ leak rate test] procedure on an Exception and Deficiency test . . . . Finally, in count 11, Met Ed was alleged to have illegally " falsified, concealed and covered up" the fact that the leak rate test procedure "did not accurately measure the amount of reactor coolant leakage from unidentifled sources at" I

Ti1I - 2 . (See attached copy of the indictment in Appendix C, of which we take official notice.)

6. Plea Agreement.

42. Met Ed entered into an agreement with the Government to withdraw its not guilty pleas to certain counts of the indictment in return for the dismissal of the remaining charges. Met Ed also made some collateral promises to induce the Government to agree to this disposition of the case. See Stier Report, Vol. V(A), Tabs 2 and 3.

43. Met Ed agreed to plead guilty to Count 2, which charged that the TMI-2's leak rate test procedure was

" inaccurate and meaningless" and that Met Ed, with notice of the test's defects, used it in violation of NRC regulations in an attempt to show that RCS leakage was within allowable limits. Id.

44. Met Ed also agreed to enter pleas of nolo contendere (i.e., no contest) to six additional charges: that Met Ed " intentionally manipulated . . . leak rate tests by the addition of water and hydrogen" in violation of NRC regulations (Count 4 of the indictment); that it " failed to enter into the Action Statement ( i . e _._ , to reduce unidentified leakage to allowable limits or shut the plant down within specified times) when the" leak rate test indicated unidentified leakage in excess of allowable limits, in violation of license I

requirements (Count 5); that it discarded leak rate test records, in violation of NRC regulations (Count 7); that it

" failed to notify the NRC that [the company] had not entered the Action Statement" when required to do so, in violation of license requirements (Count 8); that it failed to maintain adequate leak rate test operating logs, in violation of NRC regulations (Count 9); and, finally, that it " failed to make entries identifying deficiencies and non-conformances encountered with its [ leak rate test] procedures on an Exception and Deficiency list", in violation of NRC regulations (Count 10). I_d .

45. At the time of Met Ed's plea and sentencing, the U.S. Attorney specified that Count 2 was considered the

" pivotal" charge of the indictment. Appendix D, p. 65. He alleged that the fact that the TMI-2 leak rate test did not accurately measure RCS leakage from unidentified sources was common knowledge at the plant and led to the several other

, violations of law charged in the indictment and admitted to by i

Met Ed:

"[W]hether one is jogging water, adding hydrogen, destroying leak rate tests, all of that becomes relevant and necessary . . . to perform only if there is a leak rate test that doesn't work.

, Because, if that leak rate test did work, then l all of the shenanigans contained in the balance E of the indictment would have been utterly unnecessary."

I_ d . at 65-66. The Government conceded that the indictment did not allege a nexus between leak rate testing and the TMI-2 l

l l

l

[

I accident. Stier Report, Vol. V(A), Tab 2, p. 2.

46. After deliberating overnight on whether to accept I Met Ed's pleas pursuant to the agreement with the Government, see Appendix D, pp. 74-75, the Court accepted the plea. M.,

p. 82. The Court agreed with the Government's assessment that the charge to which Met Ed was pleading guilty was " pivotal";

Met Ed's guilty plea was, in the Court's words, "both a meaningful and significant admission." I_d . , p. 81. The Court also agreed with the Government and Met Ed that even a full trial would not resolve the facts surrounding the allegations in the indictment:

"[T]he Court is of the opinion that the facts surrounding the issues in this indictment will in all likelihood continue to be clouded by the contradictory assertions of the scientific I experts. Even after a trial in this matter, it is highly doubtful that the issues will be much clearer or that convincing conclusions will be reached."

Id., pp. 80-81.

I 7. NUREG-0680, Supp. No. 5.

47. In connection with the Commission's TMI-l Restart proceeding, the NRC's Office of Nuclear Reactor Regulation (NRR) issued a report in July, 1984 on the management integrity of GPU Nuclear as it related to the future operation of TMI-1.

One section of the report, in which NRR discussed the results I of a previous investigation by the NRC's Office of I

I I

I Investigations ("OI") into TMI-l leak rate test practices, was made a part of the record in this proceeding. Exhibit 7: U.S.

Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation, TMI-l Restart: "An Evaluation of the Licensee's Management Integrity as It Affects Restart of Three Mile Island Nuclear Station Unit 1, Docket No. 50-289" (July 1984)

(NUREG-0680, Supp. No. 5, Section 4.0) (hereinafter referred to as "NUREG-0680, Supp. No. 5"). It was admitted for the purpose of providing the Board with some background on the genesis of leak rate practices at TMI-2. Tr. 311-21; see T 102, infra.

48. In NUREG-0680, Supp. No. 5, NRR summarized and independently evaluated information gathered by OI in its investigation. OI's investigation included a surprise inspection of TMI-l in July 1983 which was conducted to uncover any records indicating whether, prior to the TMI-2 accident in March 1979, TMI-1 leak rate tests involved practices similar to the allegedly deficient practices at TMI-2. NUREG-0680, Supp.

No. 5, p. 4-1. It also included sworn interviews with all individuals who conducted leak rate tests at TMI-l during the relevant time period, as well as a large number of other TMI-l staff personnel, site management and corporate officials. I_ d . ,

p. 4-3.

49. Based on the information provided in OI's investigative reports, NRR concluded "that the evidence does not support a finding that leak rate surveillance tests at I

l l

l I

TMI-l were intentionally or systematically falsified during the period investigated." Id., pp. 4-17. However, among other things, and of particular relevance to this proceeding, NRR also found that (1) hydrogen was added to the TMI-l RCS MUT during several TMI-l leak rate tests, M. , pp. 4-5 to 4-6; (2) unrecorded water additions to the MUT were made on several occasions (including some instances where water was added gradually, or " jogged") when TMI-l leak rate tests were underway, and the additions were not accounted for in the test calculations, M., pp. 4-7 to 4-8; (3) unrecorded " feed and bleed" operations were conducted during several TMI-1 leak rate tests and not accounted for, id., pp. 4-8 to 4-10; (4) leak rate tests were performed at TMI-l more frequently, i.e., every shift, than required by the Technical Specifications, M.,

p. 4-10; (5) in violation of Technical Specifications and approved procedure -- but with the knowledge and acceptance of TMI-l supervisory and management personnel -- it was I long-standing and common practice at TMI-l to discard leak rate tests that were deemed to be " invalid", and to neither log such tests nor record the reasons why a test was considered invalid on the data sheet nor record problems with an invalid test on an " Exception and Deficiency" report. _T d . , pp. 4-10 to 4-13.

50. Thus, it is apparent that the leak rate test practices at TMI-2, which were the basis of Met Ed's indictment and conviction, did not originate at that Unit. Rather, as the I

I I

I Stier Report later confirmed, sy TV 62-63, infra, the source of sloppy leak rate test practices at TMI 2 -

rate test practices was the casual leak innocently adopted first at TMI -1.

The practices at TMI-l were simply carried e.g., Floyd Prep. St., p. 2, over to TMI-2. See, ff. Tr. 4894.

8. OI Investigation.

P 51.

L Commencing in March of 1984, at {

the NRC, a joint the direction of

) investigation was conducted by OI and NRR into the involvement of ten licensed inCROs TMI-2 Reports on the investigation were s b i leak rate tests.

and NRR two years later. u m tted separately by OI 1986 and released on June 1986.

4,OI's report was completed on March ,

(June 6, 1986). It See Order, Docket No. LRP Exhibit 6: Office ofhas been made a part of this record .

Investigations Report entitled:

Mile Island-2: Investigation of "Three Concerning Individual Operator Actions the Falsification of Leak Rate s" Test (hereinafter referred to as "OI Report");

NRR's see Tr. 389-90; t 101, infra.

report was completed on April 1, 16, 1986. 1986 and issued on May Memorandum and Order, e

s_ee 4H 55-59, infra. Docket No. LRP (May 22, 1986);

52.

In the OI Report, Region I Investigations R. Director of Keith Christopher which TMI-2 identified several ways by CROs: (1) leak rate tests could have been pulated maniby switching between MUT level transmitte rs during leak rate tests when one or both transmitter s were operating

_ _ _ . _ - - ~ '- _ _ _ _ --- -

I Stier Report later confirmed, see 1 62-63, infra, the source of sloppy leak rate test practices at TMI-2 was the casual leak rate test practices innocently adopted first at TMI-1. The practices at TMI-l were simply carried over to TMI-2. See, e.g., Floyd Prep. St., p. 2, ff. Tr. 4894.

8. OI Investigation.

51. Commencing in March of 1984, at the direction of the NRC, a joint investigation was conducted by OI and NRR into the involvement of ten licensed CROs in TMI-2 leak rate tests.

Reports on the investigation were submitted separately by OI and NRR two years later. OI's report was completed on March 6, 1986 and released on June 4, 1986. See Order, Docket No. LRP (June 6, 1986). It has been made a part of this record.

Exhibit 6: Office of Investigations Report entitled: "Three Mile Island-2: Investigation of Individual Operator Actions Concerning the Falsification of Leak Rate Tests" (hereinafter referred to as "OI Report"); see Tr. 389-90; H 101, infra.

NRR's report was completed on April 1, 1986 and issued on May 16, 1986. Memorandum and Order, Docket No. LRP (May 22, 1986);

see 41 55-59, infra.

52. In the OI Report, Region I Director of Investigations R. Keith Christopher identified several ways by which TMI-2 leak rate tests could have been manipulated by CROs: (1) switching between MUT level transmitters during leak I rate tests when one or both transmitters were operating l l

I I erratically; (2) adding hydrogen to the MUT; (3) adding water to the MUT, and failing to include the amount added in the test calculations; (4) conducting " feed and bleed" operations. OI Report, pp. 13-16.

53. Based on interviews with the ten CROs, and others, and technical analysis of TMI-2 leak rate test records, Mr. Christopher was of the opinion that four of the CROs had been involved in or aware of the intentional manipulation of TMI-2 leak rate tests. I_dl c . , p p . 16-24. As to the other six CROs under investigation, Mr. Christopher concluded that there was insufficient evidence that they were involved in or had knowledge of intentional manipulations of leak rate tests at TMI-2. I_ d . , pp. 25-40.

54. Mr. Christopher concluded that, generally, CROs considered the leak rate test to be unreliable and little more than burdensome paperwork. As a result, he found that leak rate tests were typically run in a careless manner, or sometimes deliberately manipulated. I_ d . , p. 1.

I 9. NRR Investigation.

55. NRR's separate report on its joint investigation with OI has been made part of this record. Exhibit 5-A:

"Results of NRR Investigation and Evaluation of Ten Licensed Operators Involved In TMI-2 Preaccident Leak Rate Testing

.g i

5 Irregularities" (April 1, 1986); see also Exhibit 5-B:

" Memorandum from William T. Russell to Harold R. Denton, dated l 3l 1 tI

I September 20, 1985" (comparing technical evaluations in NRR Report and Stier Report on TMI-2 leak rate testing). See HT 99-100, infra.

56. The NRR Report had two purposes. First, it presented NRR's general findings and conclusions about TMI-2 -

leak rate test practices during the six months immediately preceding the March 1979 accident. Exhibit 5-A, Enclosure 1.

In summary, NRR found that leak rate tests were routinely performed every shift, although they were required to be performed only once every 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />; because errors in the procedure caused erratic and variable test results, it was common practice to run several tests per shift until an acceptable test was obtained; of the numerous tests run during a shift, acceptable test results would be retained and unacceptable tests would be discarded; tests were not logged nor were deficiency reports completed when appropriate; some CROs had added hydrogen or water during leak rate tests, or had used a faulty MUT level trancmitter to manipulate the results; and finally, " feed and bleed" operations, unstable plant conditions, and inaccurate and unreliable instrumentation invalidated many tests, but there was insufficient proof that these were used to intentionally manipulate test results. Id.,

8 pp. 9-11.

57. NRR concluded this section of its report by saying, in part, that I

I I

I

"[blecause of variability of the test results, most operators did not consider the test to be a true measure of actual plant leakage. Based on I postaccident analysis of the leak rate test procedure, the operators were correct. The staff believes that many instances occurred where leak l

l I rate tests were in excess of the technical specification limit, yet, the actual unidentified leakage did not exceed 1 gan)."

Id., p. 11 (emphases added).

58. The second purpose of the NRR Report was to provide the Commission with findings and conclusions on the ten TMI-2 CROs allegedly involved in leak rate testing irregularities. Exhibit 5-A, Enclosures 2-17. Additionally, l

NRR was to recommepd what actions, if any, should be taken as to these CROs. I_ d . , pp. 1-2. As to the latter, in commencing E this proceeding, the NRC ordered NRR to make its recommendations within sixty days after we have rendered our decision. 22 N.R.C. at 883. Accxdingly, the NRR Report merely recommended that the Commission take no action towards any individual CRO until after this Board has issued its findings. Exhibit 5-A, p. 4.

59. In its Report, NRR did make findings regarding the involvement of the 10 CROs in manipulating leak rate tests at TMI-2, which were generally consistent with Mr.

Christopher's findings in the OI Report. NRR found that the same four CROs, in varying degrees, were involved in or had knowledge of leak rate test manipulations. See Exhibit 5-A, Enclosures 3-12. Further, NRR found, as did Mr. Christopher, I

I

I that all ten operators had little regard for the utility of the leak rate tests. Id.,

I

p. 3.

10. Stiqr Investigations.

a. TMI-l Investigation.

60. In January of 1984, GPU Nuclear (the successor to Met Ed as licensed operator of TMI-2), retained Edwin H. Stier, Esq. to conduct another independent investigation of the leak rate testing practices at TMI-l and TMI-2. The purpose of this factual investigation was to assist GPU Nuclear in its understanding of these practices, any deficiencies in them, and the effectiveness of GPU Nuclear's subsequent measures to correct these deficiencies. This investigation, analysis and report were to be under Mr. Stier's sole control. Stier Report, Vol. V(A), Tab 1.

61. Mr. Stier's June 1984 report on TMI-l was submitted to the NRC, but has not been made a separate part of the record in this proceading; this Board previously ruled that Section 4.0 of NUREG 0680, Supp. No. 5 (see VH 47-50, supra) would constitute the only part of the record that specifically addresses TMI-l leak rate tests, and that TMI-l leak rate testing would not be a subject of inquiry in this pr 'eeding except to the extent that it shed light on our understanding of leak rate test practices at TMI-2. Tr. 320-21.

I I

I I

b. TMI-2 Investigation.

62. Mr. Stier's separate report on leak rate testing at TMI-2 was submitted to GPU Nuclear in September 1985. It was also filed with the NRC and has been made a part of this record. Exhibits 1-A, 1-B and 1-C: " Reactor Coolant Inventory Balance Testing, Prepared for GPU Nuclear Corp., by Edwin H.

Stier," September 5, 1985; " Review of NRR and OI Reports on TMI-2 Reactor Coolant Inventory Balance Testing, Prepared for GPU Nuclear Corp. by Edwin H. Stier," August 28, 1986, and September 2, 1986 letter from Mr. Stier to Mr. Phillip R.

Clark; see Tr. 388, 391, 569, 570; 1T 96-97, infra. Mr. Stier ,

made findings concerning leak rate test procedural requirements and leak rate test practices at TMI-2 in his Report wnich, as we anticipated, has contributed significantly to the record of this inquiry. See e.g., Declaratory Order and Certification Of Questions To The Commission, Docket No. LRP (May 30, 1986). In general, Mr. Stier found that:

" Practices were developed within the Operations Department at TMI-2 that controlled I the day-to-day performance of leak rate tests.

Among the practices was the discarding of leak rate test records for which the calculated I unidentified leak rate exceeded 1.0 gpm. It is clear that this practice was carried over from TMI-l where it had been established in September 1974. However, there were significant I differences between the two plants in the performance of leak rate tests.

I "At TMI-1, our investigation found that operators discarded and failed to document a small percentage of leak rate test results I

I '

I

I indicating excessive unidentified leakage. We concluded that this practice was contrary to the spirit of the procedural scheme for leak rate I testing. However, we also concluded that operators conducted a meaningful evaluation of leak rate test results and did in fact use the test to measure reactor coolant system leakage.

"At TMI-2, no similar evaluation of leak I rate test results took place. Personnel at TMI-2 had greater difficulty obtaining test results for unidentified leakage within the 1.0 gpm acceptance criterion than their counterparts at TMI-1, and they discarded the results of more than 50 percent of the tests they performed.

This was the case even though actual unidentified I leakage remained under 1.0 gpm during most of the time the plant was in operation, i "There were several major causes of the difficulty experienced at TMI-2. First, the TMI-l leak rate calculation permitted the subtraction of 0.5 gpm from gross leakage to I account for " evaporative losses." At TMI-1, this helped offset the variability of test results caused by such factors as instrument error and I oscillations within the reactor coolant system.

The TMI-2 procedure contained no similar provision, thereby increasing the frequency with I which calculated unidentified leakage exceeded 1.0 gpm due to inherent test variability.

"Second, a significant calculation error I existed in the TMI-2 procedure that was not present at TMI-1. The TMI-2 calculation incorrectly considered reactor coolant drain tank I collection to be at room temperature rather than at reactor coolant system temperature. This error caused calculated unidentified leakage to increase automatically as reactor coolant drain I tank collection increased. At times, it was virtually impossible to obtain a leak rate test result under 1.0 gpm, even though unidentified leakage in fact remained under that limit.

" Third, certain conditions also found at I TMI-1 caused difficulty in leak rate testing at TMI-2. These included systematic malfunction in makeup tank level instrumentation that at times I

overstated the indicated drop in makeup tank level, and a failure in the leak rate calculation I to consider operator identified leakage at reactor coolant system temperature."

Stier Report, Vol. I, pp. 20-22 (emphasis in original).

63. The Stier Report is the most comprehensive analysis of TMI-2 leak rate practices, and is discussed and substantially relied on by the Board throughout this e Recommended Decision. The Stier Report also includes an individual performance of TMI operators and supervisors wno were involved in leak rate testing at TMI-2 and were still employed by GpU or its subsidiaries. Id., Vol. II. We have given little or no weight to these assessments, because it is our obligation and prerogative to determine individual responsibility.

B. Institution of This proceeding.

1. Origin in TMI-1 (Restart) Proceeding.

64. In Metropolitan Edison Co. (Three Mile Island Nuclear Station, Unit 1) (Restart), 21 N.R.C. 282 (1985), the Commission considered whether to reopen the record in the TMI-l restart proceeding to consider the Hartman allegations of leak rate test falsification at TMI-2 and, if it were not reopened, whether there should be a separate proceeding to allow the matter to be fully aired. Id. at 285. The specific issue before the Commission was whether the Hartman allegations raised any significant safety issue regarding TMI-2 leak rate I

I I

I practices that had a bearing on the TMI-l restart proceeding.

The Commission found that none of those individuals who might be implicated in leak rate test falsification at TMI-2 were employed in responsible management or operational positions at TMI-1; further, the Commission found that the personnel, procedures and organization in place at TMI-l provided reasonable assurance that similar procedural deficiencies would not recur. I_d . at 297-304. Accordingly, the Commission decided not to reopen the record in the TMI-l restart proceeding to consider the Hartman allegations. I_d . at 305.

65. The Commission did decide, however, to institute a hearing separate from the restart proceeding to develop the facts surrounding the alleged leak rate falsifications at TMI-2 in sufficient detail to determine the involvement of any individual who may now work, or in the future desire to work, at a nuclear facility. M. at 305. The Commission ruled that the separate hearing would not address certain individuals specifically exonerated by the United States Attorney in his I statement at the sentencing hearing of Met Ed. I_d .

2. Legislative Nature of Inquiry.

66. In starting this separate proceeding to investigate the Hartman allegations, the Commission specified that our hearings were to have a legislative format " designed solely to gather information." 22 N.R.C. at 878. Thus, the I

I

I Commission provided that, subject to certain limited exceptions, its rules of practice for domestic licensing adjudications were not applicable. I_d. at 882.

67. The hearing procedures set out by the Commission defined the legislative, rather than adjudicatory nature of the inquiry before the Presiding Board. See id, at 881-84. The Commission plainly intended the hearing itself to serve as the I fact-finding mechanism, and thus ruled that no discovery would be permitted. M. at 882. Further, the Commission provided that only the Presiding Board would be able to call witnesses and question them. I_d . No right of cross-examination was provided to the parties. d Id.; 23 N.R.C. 51, 52 (1986) (denying Numerous Employees' Motion for Reconsideration). Finally, the Board was instructed that it could not deviate from the procedures set out in the order without " prior authorization from the Commission." 22 N.R.C. at 883.

I 3. The Presiding Board Is to Make Findings of Fact and Conclusions of Law, Not to Make Recommendations for Future Proceeding _s.

68. The Presiding Board has been charged with issuing a Recommended Decision which sets forth its findings on who participated in, had knowledge of and condoned, or by their dereliction or culpable neglect allowed the leak rate falsifications, and the facts surrounding any such involvement in sufficient detail to determine the involvement of any I

I individual who may now work, or in the future work, at a nuclear facility. Id. at 883. The Board is to address the specific issues set out by the Commission in its order. M. at 880-81. However, the Presiding Board is not to make any recommendations regarding whether action should be taken against any individual. Id. at 883.

I 4. Future Proceedings May be Conducted Under Certain Circumstances.

69. On the basis of the Presiding Board's Recommended Decision, and taking into account other information which the NRC Staff believes is appropriate for Commission consideration, the NRC Staff is charged with making recommendations to the Commission regarding what actions, if any, should be taken with respect to any involved individual. Id. After reviewing the Board's Recommended Decision and the NRC Staff's recommendations, the Commission will decide what further steps, if any, need to be taken with respect to the involved individuals. I_d . This will include consideration of whether to remove TMI-1 employment constraints and whether to initiate formal enforcement action or take any licensing action with respect to involved individuals. M. at 884. If, as a result of its review, the Commission institutes a formal enforcement proceeding or takes any licensing action, the facts found by the Presiding Board and Commission in the legislative hearing will not be binding in the subsequent enforcement or licensing I

I

I proceeding. I_d . The Commission also emphasized that the proceeding before the Presiding Board will not be used to deprive any individual of any statutory hearing rights he may have if formal enforcement or licensing action is subsequently initiated. 23 N.R.C. at 52.

I 5. The Board Can Exonerate Individuals and Will Do So.

70. The Presiding Board is charged with making findings as to which individuals, if any, participated or acquiesced in leak rate test falsification at TMI-2. 22 N.R.C.

at 883. As we stated before our hearing began, "this proceeding does and should have as one of its objectives to exculpate individuals whose names have arisen in connection with the subject matter of falsified leak rate testing at TMI-2." Memorandum and Order, Docket No. LRP, at 15 (July 16, 1986). The Board thus concludes that, while it is not authorized to recommend actions to be taken against individuals, it is empowered to exonerate individuals about whom there is insufficient evidence of involvement in leak rate test falsification, and does so in Sections IX and X of this Recommended Decision.

6. Identity of Parties.

71. On December 31, 1985, this Board sent letters to a group of about 120 present and former employees of GPU Nuclear and Met Ed who worked at TMI-2 between February 2, 1978 I

I I and the date of the accident. That group represented all those who were involved in, or were likely to have had knowledge of, reactor coolant inventory balance testing at TMI-2. The letter described the hearing to be conducted and invited each addressee to file a petition to intervene with the Board. The letter noted that the Board might subpoena as a witness any individual who elected not to participate as a party.

Contemporaneous with the Board's letter, the Commission promulgated a notice to the public, inviting "any person who has an interest which may be affected by this hearing" to petition to intervene. 50 Fed. Reg. 52,388 (1985).

72. In response to the Board's letter, numerous 1978-79 employees (" Numerous Employees") of Met Ed (Messrs. Adams, Booher, Brummer, Bryan, Chwastyk, Coleman, Conaway, Congdon, Faust, Floyd, Frederick, Germer, Guthrie, Hitz, Hoyt, Illjes, hunder, Marshall, McGovern, Mehler, Mell, Adam Miller, Scheimann, Smith, and Zewe) petitioned the Board for leave to intervene in the leak rate inquiry on January 31, 1986. Another former TMI-2 operator, Mr. John M. Kidwell, petitioned the Presiding Board for leave to intervene on February 4, 1986. The Presiding Board also received petitions to intervene from Mr. Marvin Lewis, GPU Nuclear Corporation, Mr. Gary Miller, Mr. John Herbein, and Mrs. Marjorie Aamodt.

73. By Memorandum and Order, dated February 14, 1986, the Presiding Beard granted the petitions to intervene of all I

I of the aforementioned petitioners except Mr. Lewis and Mrs. Aamodt. The Board deferred its ruling on the Lewis and Aamodt petitions until the prehearing conference that it set for March 7, 1986. By Memorandum and Order, dated March 26, 1986, the Board granted Mrs. Aamodt's petition to intervene (admitting both Marjorie Aamodt and Norman Aamodt as parties) and denied Mr. Lewis' petition to intervenc. See 1 125, infra.

74. By Notice of Withdrawal dated May 7, 1986, Messrs. Bryan and Kidwell withdrew as parties to the proceeding.

7. Identity of Other Witnesses.

75. By Memorandum and Order, dated March 26, 1986, the Presiding Board set out, inter alia, a list of proposed witnesses, and solicited objections or proposed additions to that list. The list included the following technical witnesses who were ultimately added as witnesses by the Board: Messrs.

Jared S. Wermiel and Donald C. Kirkpatrick (NRC Staff witnesses on the technical aspects of leak rate testing), Mr. Rockwell, Mr. Stier and Messrs. William T. Russell and Robert A. Capra (authors of the NRR Reports). See Memorandum and Order, Docket No. LRP (May 22, 1986).

76. In response to the Board's order, the " Numerous Employees", which generally includes all Met Ed employees discussed herein except Messrs. Bezilla, Blessing, Boltz, I Hartman, Herbein, Lawyer, Logan, Gary Miller, Morck, O'Hanlon,

I and Stair, proposed the following additional witnesses who ultimately testified at hearing: (a) Mr. Donald R. Haverkamp, a regionally-based NRC inspector whose inspection at TMI-2 during October 16-20, 1978 culminated in the filing of Licensee Event Report 78-62/lT concerning TMI-2 leak rate testing, (b) Mr. James p. Moore, Jr., an engineer employed by GpU Service Company who coauthored the Faegre & Benson Report, and (c) Mr. Hartman. Memorandum and Order, Docket No. LRp (May 22, 1986). Mr. Christopher, the principal author of the OI Report, was also proposed as a witness in the Board's May 22, 1986 Order, and a decision was made to call him as a witness at the August 25, 1986 prehearing conference.

77. Certain present or former TMI-2 employees were subpoenaed by the Board prior to commencement of the hearing.

Those individuals were Messrs. Blessing, Bryan, Cooper, Hartman, Hemmila, Logan, Olson, Seelinger, and Wright.

Subsequent to the commencement of the proceeding, Mr. Kidwell was subpoenaed to appear before the Board and Messrs. Boltz, phillippe ,and Fels agreed to appear voluntarily as witnesses.

78. Mr. Blessing never responded to his subpoena, and thus the Board did not hear his testimony. See Tr. 4332. The Board chose not to require Messrs. Brummer and Germer to appear based on their prior statements and prefiled testimony which indicated that Mr. Brummer had no involvement in leak rate testing and that Mr. Germer, who was a trainee, had very little I

I

I involvement and is now living a considerable distance from the Washington, D.C. area. Tr. 4168, 4541.

79. The Staff proposed that the Board hear the testimony of Messrs. Morck (who prepared a change to the TMI-2 leak rate procedure in March 1979) and Lawrence Lawyer. Tr.

4948-52, 5225. The Board decided not to call either Mr. Morck or Mr. Lawyer, because neither seemed likely to have I information to aid our inquiry. Tr. 5038, 5341. The Board also decided not to call Messrs. Bezilla and Stair, who were involved in the preparation of a Met Ed Licensee Event Report on TMI-2 leak rate testing, because the Board believed that it had the information it needed on that subject. Tr. 4168-69, 4948.

8. Description of Proceeding and Hearings.

80. The initial prehearing conference in this proceeding was held on March 7, 1986, and addressed hearing schedule, location of the hearing, documentary evidence, prefiled testimony of witnesses, questions to be propounded to the witnesses, sequestration of witnesses, role of counsel, role of the NRC Staff, and ex parte communications.

81. These issues were also addressed in subsequent orders by the Presiding Board. In its Memorandum and Order of March 26, 1986, the Board posited, inter alia, a " List of proposed Board Witnesses" and a " List of Documents Proposed for the Initial Record", and solicited objections or proposed I

additions to those lists from the parties. Responses were received from the Numerous Employees, GPU Nuclear Corp , Gary Miller, and Mrs. Aamodt. In its Memorandum and Order of April 3, 1986, the Board proposed that the hearing be divided into two separate phases -- an initial phase exploring the technical background of leak rate testing and a subsequent phase focusing on issues of individual responsibility for invalid leak rate tests.

82. Subsequently, the Board slightly altered its original notion of conducting a separate hearing on the technical issues. The hearing was ultimately conducted so that, irmnediately af ter testimony on the technical issues was I heard, we heard the testimony of Messrs. Boltz, Haverkamp, Hartman and the various TMI-2 employees having knowledge of leak rate testing.

83. The Presiding Board's Memorandum and Order of May 22, 1986 addressed issues raised by the Board's orders of March 26, 1986 and April 3, 1986, and the parties' responses thereto. The May 22 order proposed, inter alia, that the NRR and OI reports be included in the record and that Mr.

Christopher, the principal author of the OI Report, testify concerning that report. The NRR and OI Reports were ultimately admitted into the record. Memorandum and Order, Docket No. LRP (August 1, 1986).

I I

84. The Numerous Employees submitted proposed technical questions for Messrs. Kirkpatrick, Wermiel, Rockwell, Russell, Capra, and Stier on July 21, 1986, the majority of which were ultimately propounded to those witnesses by the Board. Proposed questions on individual responsibility for Messrs. Christopher, Russell, and Capra were submitted by the Numerous Employees shortly thereafter, i

85. By Order dated August 26, 1986, the Presiding Board was reconstituted to appoint Administrative Judge James H. Carpenter in place of Administrative Judge Jerry R.

Kline.

86. The hearing commenced on September 8, 1986 with the testimony of a panel of technical witnesses, including the NRC Staff witnesses (Messrs. Russell, Capra, Kirkpatrick, Wermiel), Mr. Rockwell and Mr. Moore, and Mr. Stier (accompanied by Dr. Harrison and Mr. Cole of MPR Associates, an engineering consulting firm that assisted Mr. Stier). That I panel was convened as such to respond to the technical questions common to all panel members. Panel members were excused once they had responded to both the common questions as well as questions specific to their report.

87. After the technical phase of the hearing, testimony was heard from Mr. Haverkamp, Mr. Dennis Boltz (who served as a training instructor assigned to the licensed operator training group at TMI-2 during 1978-79), and I

I

I Mr. Hartman. Following that, testimony on individual responsibility issues was heard from Messrs. Russell, Capra, and Christopher. Finally, testimony was heard from the numerous control room employees and their immediate supervisors (i.e., CROS, shift foremen and shift supervisors), followed by testimony from several management employees. Midway through the hearing, Chairman Kelley visited the site to gain an appreciation of the environment in which the TMI-2 personnel operated, and to observe the instrumentation described in the testimony. See Tr. 4192.

88. Throughout the hearings, the Board asked numerous questions submitted by the parties as well as questions of its Il own. Objections were also allowed, but generally the Staff was permitted to object only to questions put to its own witnesses.

89. The hearings were concluded on November 12, 1986.

The parties filed proposed findings of fact, conclusions of law, legal memoranda, and proposed reply findings, in January and February 1987. We have relied extensively on the parties' proposed findings and conclusions in preparing this Recommended Decision.

I II. DESCRIPTION OF THE EVIDENCE.

90. The Commission's mandate to the Board was to develop detailed facts surrounding the involvement of individuals in leak rate test falsification at TMI-2. 22 N.R.C. at 880. The evidentiary record amassed by the Board was I

shaped, to a great extent, by the basic procedural instructions adopted by the Commission for conducting a " legislative format hearing designed solely to gather information". Id. at 878.

91. These procedures included the criteria that must be met to participate as a party: not only must there be a showing of an interest that might be affected, but also a showing that the person "will likely contribute to an adequate record." I_d . at 881-82. Moreover, the Commission determined that the NRC Staff would not participate as a party but, rather, was to make all relevant documents available to the Board and to provide whatever assistance we requested "to ensure that the hearing record is fully developed." I_d. at 882. The Staff fulfilled the Commission's mandate.

92. The Commission further instructed the Board that, in conducting its hearing, only non-repetitive, relevant, material and reliable evidence was to be admitted; no discovery was to be permitted; no direct or cross-examination of witnesses by the parties were to be permitted; and only the Board could call and question witnesses, although the Board, in its discretion, could call witnesses and ask questions suggested by a party. d

[Id.; 23 N.R.C. at 52.

93. With this basic framework as its starting point, I the Board held several prehearing conferences, and issued several prehearing orders to implement the Commission's order.

See, e.g., Memorandum and Order, Docket No. LRP (filed I

I

I I February 18, 1986) (" February 14 Order"); Memorandum and Order, Docket LRP (May 22, 1986) ("May 22 Order"). The Board recognized that, over the course of the seven years since the accident at TMI-2, a voluminous amount of interview transcripts, technical reports, and other documents concerning TMI-2 had been collected by various public and private entities. See February 14 Order, supra, p. 7; Memorandum and Order, Docket LRP, p. 9 (March 26, 1986) (" March 27 Order").

Constrained by practical considerations, and in the absence of discovery, see Tr. 184, the Board sought to include in the record only those documents it believed were "particularly significant to the inquiry (,)" and to exclude those it believed were " cumulative or only marginally relevant to the issues in this inquiry." March 26 Order, supra, p. 9; se_e_ May 22 Order, supra, pp. 7-8; February 14 Order, sunra, pp. 6-7.

94. The Board anticipated that, although the TMI-2

" employees knowledge of the facts (would) be vital to the development of the record", February 14 Order, s uar a_ , p. 2, several of the prior investigative reports, particularly the Stler Report, Facqre Fs Benson Report, and the reports by HRR and OI, also would be especially helpful. Declaratory Order and Certification of Questions To The Conmisssion (" Declaratory Order"), Docket No. LRP (May 30, 1986); see hg., May 22 Order, supra, pp. 5-6; February 14 Order, supra, pp. 6-7. The Board intended to use these reports as, inter alia, a means of I

I  :

I--

I checking the thoroughness and accuracy of its own investigation, and not as a substitute for its own judgment I regarding individual responsibility. E.g., May 22 Order, p. 6.

Rather, the Board has " base [d] our conclusions on the best available direct evidence about leak rate falsifications." Id.

(emphasis added).

95. The laborious process of winnowing out the most useful of the scores of documents available regarding leak rate testing at TMI-2 finally yielded the Exhibits listed in Appendix A hereto. A chronology of the Exhibits comprising investigative evidence is set forth above in HH 35-63, supra.

A summary description of all the Exhibits is set forth below.

A. Investigative Reports.

1. Exhibit 1: Stier Report.

96. Exhibit 1 actually has three parts: Exhibit 1-A was Mr. Stier's excellent, 32-volume report of his comprehensive investigation of leak rate test practices of TMI-2, including a meticulous and detailed technical analysis of plant records by MpR Associates, Inc., an independent engineering firm. See HH 62-63, supra. Exhibits 1-B and 1-C are related technical analyses by MpR of the NRR technical analysis, and Mr. Stier's determination, after thoroughly assessing these subsequent investigative reports and comparing their results to his and those of MpR in Exhibit 1-A, that I

I .

I nothing in the tipR and OI reports caused him to alter his original findings.

97. As we anticipated at the outset of this hearing, the Stier Report and Mr. Stier's related testimony have contributed significantly to thi.s record. See Tr. 2034.

Except to the extent that the Stier Report contains Mr. Stier's conclusions about individual responsibility (to which we attach little weight) we have heavily relied upon it, especially upon the technical analysis by MpR.

2. Exhibit 2: Faegre & Benson Rego r t .

98. Exhibit 2 is the four-volume report of the investigation in 1980 of Mr. Hartman's allegations by the law firm of Faegre & Benson. Transcript 28-29. The investigation was headed by Mr. Rockwell. The Faegre & Benson Report has the advantage of being the most contemporaneous independent investigation of the circumstances surrounding leak rate test practices at TMI-2 in 1978 and 1979, id., p. 29; see $t 36-37, r .

s up_rit As with the Stier Report, we anticipated that the Faegre & Benson Report and Mr. Rockwell's related testimony would provide us with significant input into our inquiry. Jqe Declaratory Order, p. 2. Similarly, we have relied heavily on the Faegre & Benson Report (except Volume One, Section Four and Volume Two, Chapter Nine which were excluded from our considerations, Tr. 384) in making those findings.

I I

I

3. Exhibit 5: NRR Report.

99. Exhibit 5 is in two parts: Exhibit 5-A is the separate report of NRR's joint investigation with OI of the general practices and procedures for leak rate testing at TMI-2, and an evaluation of 10 licensed individuals alleged to have been involved in leak rate test irregularities. (We had the NRC Staff exclude from Exhibit 5-A, NRR's evaluation of the ten licensed individuals' current performance and NRR's recommendations about them, because those subjects exceeded the scope of our mandate from the Commission. Tr.1227-28.) See H 55-59, supra. Exhibit 5-B is NRR's comparative analysis of its technical evaluation of TMI-2 leak rate tests with that of I the Stier Report. See El.

100. We have found the NRR Report and the related testimony of Messrs. Capra and Russell to be of value to our inquiry, but not to the extent that we anticipated. Compare March 26 Order, p. 9 and Attachments A and B (although NRR Report not then publicly available, Board believed that at least major portion of it would be particularly significant),

with Tr. 1472-75, 4971-72 (Board indicates some question,about NRR's classification of particular tests as " questionable").

We have not relied upon the NRR Report as extensively as we have relied upon the Stier Report and the Faegre & Benson I "*" "-

I I

I

4. Exhibit 6: OI Report.

101. Exhibit 6 is OI's separate report of its investigation jointly with NRR into the alleged involvement of 10 individuals in leak rate test falsification at TMI-2. See Section I.A.8, supra. This report and the related testimony of its author, Mr. Christopher, were helpful to the Board in the instant proceeding but, as with all the other investigative reports in the record, we have not relied on any conclusions regarding the involvement of individuals in leak rate test result falsifications. The Commission ordered this Board to make those findings based on the evidence of record, and we I

have done so.

5. Exhibit 7: NUREG-0680, Supp. No. 5.

102. Exhibit 7 is Section 4 of NUREG-0680, Supp. No.

5. Tr. 390. It documents the fact that many, if not all, of the leak rate practices in TMI-2 under investigation here originated at TMI-1. See 1[1f 4 7-50, supra. We admitted Exhibit 7 without objection from any party because several TMI-2 W operators and supervisors testified that TMI-2 leak rate I testing practices originated at TMI-1. Therefore, we believed that Exhibit 7 would provide useful background to our assessment of TMI-2 leak rate testing practices.

I I

I I

B. Other Exhibits.

1.

I Exhibit 3: Portion of GPU Assessment Panel F_indings.

103. One page from a report entitled "GPU Assessment Panel for Individuals In TMI-2 Leak Rate Testing (in 1978-1979)," dated January 6, 1986, was admitted as Exhibit 3.

Tr. 389. The panel was composed of the presidents of GPU and three of its subsidiaries, Met Ed, Pennsylvania Electric Company and Jersey Central Power and Light Company; it was formed by GPU to take disciplinary action against GPU employees, where appropriate, based on the Stier Report.

Exhibit 3 sets out the panel's assessment of Mr. Herbein's involvement in leak rate testing at TMI-2.

I 2. Exhibit 4:

1978).

TMI-2 Computer Log (October 27, 104. Exhibit 4 is a photocopy of the TMI-2 computer log for October 27, 1978, recording the fact that the practice of rounding off the value for unidentified leakage on TMI-2 leak rate tests was stopped that day. (The practice had begun 9 days earlier.) See 14 347-48, 354-56, infra.

3. Exhibit 9: Pictures of the TMI-2 Control Room.

l 105. Exhibits 9A-9F consist of six different photographs of the TMI-2 control room and instrumentation therein.

B g lI

E

4. Exhibit 10: Examples of RO and SRO

[ Licenses.

106. Exhibit 10-A is an example of an NRC Senior Reactor Operator ("SRO") license authorizing the licensee "to r direct the licensed activities of licensed operators at, and to L

manipulate all controls of" TMI-l and TMI-2. Exhibit 10-A is

{ the type of license held by the shift foremen and shift supervisors. Exhibit 10-B is an example of an NRC Reactor Operator ("RO") license authorizing the licensee "to manipulate all controls of" TMI-2. Exhibit 10-B is the type of license held by a CRO. Tr. 1025-27.

^

5. Exhibit 11: J. Moore Charts: Effects of Oscillations.

107. Exhibit 11 consists of two charts, Exhibit ll-A

("TMI-2 Final Average Temperature") and Exhibit 11-B ("TMI-2 RCS Leakage Rate"), which were visual aids to the testimony of Mr. J. Moore, a GPU Nuclear engineer who participated in the preparation of the Faegre & Benson Report and who accompanied Mr. Rockwell. Tr. 373. These charts represent graphic illustrations of a computer model done by Faegre & Benson which demonstrated that oscillating plant conditions were producing

{ random leak rate test results at TMI-2. Tr. 776-86.

6. Exhibit 12: Administrative Procedure 1036, Revisions 0 and 1.

108. Exhibit 12 consists of two revisions of Met Ed's written procedures for documenting out-of-service instruments

I I and readout devices (meters, gauges, amplifiers, and recorders)

I at TMI-2. Tr. 1354-56.

7. Exhibit 15: Curricula Vitae of Dr. Harrison and Mr. Cole.

109. Dr. Harrison and Mr. Cole, from MPR, accompanied Mr. Stier to testify about Exhibit 1, the Stier Report. Tr.

373. Exhibit 15 consists of their curricula vitae, demonstrating their substantial engineering and technical expertise.

I 8. Exhibit 16: Training Department Administrative Memorandum.

110. Exhibit 16 was admitted during the testimony of Mr. Boltz, who was an instructor in the TMI-2 training department in 1978 and 1979. It was the only training document at TMI-2 which contained any reference whatever to leak rate testing, and confirmed Mr. Boltz's testimony as to the lack of virtually any classroom training concerning this subject. See I TH 315-18, supra.

I 9. Exhibit 17: Excerpts from Instrument Out-of-Service Log.

111. Exhibit 17 is comprised of copies of four pages from the TMI-2 "out of service log" referenced by Mr. Capra during his testimony concerning malfunctioning of MUT level I """*'""'"-'- " " " " " - "

I

10. Exhibit 18: RCIB Test of February 15, 1979 l with CRO Log and MUT Level Strip Chart.

112. Exhibit 18 consists of three pages. The first L page is a copy of the computer printout, or leak rate 7 surveillance sheet, for a test begun at 2026 on February 15, L

1979; the second page is a copy of the CRO log entries between J

l 1100 and 2200 on that same date; and the last page is a copy of the MUT level strip chart for the leak rate test beginning at 20:26 that day. Tr. 3461-70. A hydrogen addition to the MUT apparently occurred during that Test. See Exhibit 18, (CRO log L

entry at 2100).

11. Exhibit 19: TMI-2 Control Room Layout Diagram.

- 113. Exhibit 19 is a diagram of the layout of the

- TMI-2 control room admitted to aid the board in understanding L

the spatial relationships between the various equipment and instruments depicted in Exhibits 9-A to 9-F. Tr. 3500-01.

{

12. Exhibit 21: Bettenhausen Affidavit.

114. Mr. Bettenhausen was the Chief of the NRC's Operations Branch of the Division of Reactor Safety, Region 1.

Exhibit 21 is a four-page affidavit originally submitted to the Commission in connection with the TMI-l (Restart) proceeding.

In Exhibit 21, Mr. Bettenhausen describes, inter alia, the fact that the leak rate test will variously yield positive and negative results, even in a leak-free plant. He also discusses

I I the inclusion of an evaporative loss factor in the test calculation, which was used at TMI-1, but. not at TMI-2, see Tr.

4305-30, 4332-36, and concludes that the incorporation of an evaporative loss factor in the TMI-l leak rate procedure had no safety significance.

13. Exhibit 22: TMI-2 FSAR Excerpts.

115. The Board admitted FSAR pages 5.2-16 through 5.2-27 and 9.3-26 through 9.3-30 as Exhibit 22 because those pages were cited in the NRC Staff testimony as the technical basis for the 1 gpm unidentified leakage limit. See Tr.

4334-35, 4428-31.

I 14. Exhibit 23: TMI-2 FSAR Excerpts.

116. FSAR pages 5.2-10 and 5.2-11 and Table 5.2-12 were admitted along with Exhibit 22. They explain, inter alia, the reasons for adding hydrogen and other substances to the RCS, as well as the limitations thereon.

I See Tr. 4428-31.

15. Exhibit 24: Stier Letter (and Attachments) to Board, Dated September 30, 1986.

117. In Exhibit 24, Mr. Stier provided the Board with materials we requested of him during his previous testimony.

l l See Tr. 2028-30. Exhibit 24 includes an illustration of the differences between the " slope offset" and " peak-to-peak" methods of estimating water additions to the MUT; copies of TMI-2 MUT level strip charts showing alleged hydrogen additions lI l

lI

I when leak rate tests were not being performed; and copies of three leak rate tests from October 18, 1978 showing the full handwritten notations of Mr. Floyd on those tests. In Exhibit 24, Mr. Stier also made some minor corrections to the text of Exhibit 1-A.

I 16. Exhibit 25: MPR Letter to Stier, Dated September 30, 1986.

I 118. In Exhibit 25, MPR provided the approximate percentage of hours it performed services for GPU Corporation in the last ten years.

I 17. Exhibit 26: November 3, 1986 NRC Region I Inspection Report for Rochester Gas and Electric Corp.'s Ginna Plant.

119. During the testimony of NRC Inspector Haverkamp, see Tr. 5258-65, the Numerous Employees proposed a recent inspection report for another nuclear facility as an Exhibit.

The purpose of this proposed Exhibit, and three related questions to be asked of Mr. Haverkamp, was to demonstrate that the NRC now conducts far more thorough and comprehensive inspections of nuclear power facilities then it did at TMI-2 in 1978 and 1979; it was proffered, further, to show that, had the NRC conducted better investigations at TMI-2, the defects and deficiencies in leak rate test practices and procedures would have been identified and corrected sooner. Tr. 4258-65.

120. The Board rejected this Exhibit for two reasons.

First, it appeared to be beyond the scope of Mr. Haverkamp's I

I I

I I testimony. Second, it appeared to the Board that the proffer was untimely. Tr. 5265-67. On reflection, however, we now believe that Exhibit 26 should have been admitted, and hereby do so, because the subject of the Exhibit is material to the issues herein and Mr. Haverkamp's testimony (see 4 309, infra),

and the Exhibit was dated only nine days before it was offered.

16. Exhibit 27: Letter from J. Moore to W.

Rockwell, Dated October 31, 1986.

121. Exhibit 27 was admitted after the hearings were concluded. Order, Docket No. LRP (November 19, 1986). It was submitted to the Board by Mr. Rockwell in response to its request. Mr. Moore advised that GPU Nuclear Instrumentation and Control ("I & C") personnel agreed with MPR's conclusion that, on January 4-5, 1979, the reactimeter and the MUT level strip chart recorder were derived from the same MUT level  !

transmitter.

C. Testimonial Evidence (Not Otherwise Included in Investigative Reports).

122. We received prepared testimony from most of the witnesses other than the investigators whose reports were previously admitted. Mr. Haverkamp and Messrs. Kirkpatrick and Wermiel submitted prepared testimony (that of Messrs.

Kirkpatrick and Wermeil was a joint statement). All of the TMI-2 employees who appeared submitted prepared statements except Messrs. Boltz, Fels, Hartman, and Logan; the prepared I

I I

I I statements are bound into the transcript of the proceeding (except for that of Mr. Haverkamp, which was not bound into the transcript through an oversight, and was made Exhibit 20). The prepared testimony of Messrs. Brummer and Germer was bound into the transcript as their stipulated testimony although they were not required to appear. Mr. Wright's prepared testimony was not adopted by him because of his lack of recollection of the underlying events, but was bound into the record as a useful summary of his prior testimony. See Appendix A hereto.

I III. THE STANDARD TO BE APPLIED TO THE FACTUAL ISdUES IN THIS PROCEEDING IS CLEAR AND CONVINCING EVIDENCE.

123. In proceedings before federal administrative bodies, the proponents of an order usually bear the burden of producing sufficient evidence to prove their position, and the sufficiency of proof is typically "a preponderance of evidence." E.g., Davis, Administrative Law, S 16.9 (2d ed.

1980); McCormick On Evidence, S 357 (3d ed. 1984). We conclude, however, that the traditional burden-of-proof rules do not apply to this inquiry; further, the Board concludes that only " clear and convincing" evidence is sufficient to support adverse findings against any individual TMI-2 control room personnel.

124. As we have already noted, TH 66-70, supra, this is not the usual or typical administrative proceeding.

Instead, the Commission ordered this Board to conduct a factual I

I I inquiry in the format of a legislative hearing. Moreover, the Commission instructed that the vast majority of its general procedural regulations for adjudicatory hearings, which includes the rule that "the proponent of an order has the burden of proof" 10 C.F.R. 2.732, do not apply to this proceeding. 22 N.R.C. at 882. Thus, this is not the type of proceeding to which the typical burden of proof procedural rules generally apply. Compare Tr. 61 (this inquiry is discretionary and not required to be held on the record) with Attorney General's Manual on the Administrative Procedure Act, pp. 40-41 (1947) (APA procedural rules in Chapter 5, which include burden of proof requirements, apply to adjudications; investigations, no matter how formal, are not adjudications.

Further, Chapter 5 does not apply to hearings not required to be on the record). Accordingly, the parties to this proceeding have not been required to produce any evidence to support their respective contentions; indeed, only the Board was empowered by I the Commission to call witnesses and question them. 22 N.R.C.

at 882.

125. Certainly, it would be inappropriate to require the TMI-2 employees whose conduct is being questioned to bear l the burden of proof; nor can we draw adverse inferences from the failure or refusal of any of the TMI-2 employees to provide evidence that he was not involved in leak rate test practices.

See Consumers power Company, LBP-75-39, 2 N.R.C. 29, 45 (1975)

I 1

l lI

I I ("In this country, persons or entities accused of criminal or tortious conduct do not have the burden of proving a negative; i.e., that no such misconduct exists.") Nor do we think that either the NRC Staff, or the authors of the various investigative reports in the record, or the Aamodts bear the burden of proof in this proceeding, although there is authority for the proposition that, in a different context, "the burden of proof rests with the party accusing another of unlawful behavior." Id. Neither the Staff nor the authors of the various investigations were parties to this pr(ceeding.

Instead, Staff was ordered by the Commission to provide the Board with testimony or other assistance at the Board's discretion. 22 N.R.C. at 882. Similarly, the authors of the various reports were called as impartial witnesses by the Board to explain their respective findings. See, e.g., Declaratory Order, Docket No. LRP, pp. 2-3 (May 30, 1986). In like fashion, we granted leave to the Aamodts to intervene because we were persuaded that they met the Commission's criterion of not only having an interest that might be affected but, more significantly, that "they will likely contribute to the devlopment of an adequate record." 22 N.R.C. at 882; see Memorandum and Order, Docket No. LRP, pp. 4-5 (March 26, 1986);

cf., id., p. 6 (denying Mr. Lewis' intervention petition under same standard). This did not imply that they bear the burden l

I I I

I of proof. Thus, we conclude that assigning the burden of proof to anyone involved in this inquiry is unwarranted.

126. Because the Commission did not specify what evidentiary standard the Board should use in lieu of the normal administrative rule on the burden of persuasion, we must decide what standard of proof to use in measuring the evidence of a specific individual's involvement in leak rate test falsification. See Tr. 119-20 (the rules for this proceeding are not clear and will have to be established consistent with fairness to the parties as the hearing progresses); see generally, Steadman v. SEC, 450 U.S. 91, 95-96 (1981) (where Congress has not pr4 scribed degree or proof necessary to carry burden of persuasion, court may do so.) The Board concludes that the appropriate quantum is " clear and convincing" evidence.

127. The Supreme Court has explained the purposes and differences between the varying standards of proof available to the finder of fact. In relevant part, the Court has said:

I "The function of a standard of proof, as that concept is embodied in the Due Process Clause and in the realm of factfinding, is to I

• instruct the factfinder concerning the degree of confidence our society thinks he should have in the correctness of factual conclusions for a I particular type of adjudication.' The standard serves . . . to indicate the relative importance attached to the ultimate decision.

I " Generally speaking, the evolution of this area of the law has produced across a continuum three standards or levels of proof for different I

I I .

I types of cases. At one end of the spectrum is the typical civil case involving a monetary dispute between private parties. Since society I has a minimal concern with the outcome of such private suits, plaintiff's burden of proof is a mere preponderance of the evidence. . . .

"In a criminal case, on the other hand, the interests of the defendant are of such magnitude I that historically and without any explicit constitutional requirement they have been protected by standards of proof designed to I exclude as nearly as possible the likelihood of an erroneous judgment. In the administration of criminal justice, our society imposes almost the entire risk of error upon itself. . . .

"The intermediate standard, which usually employs some combination of the words ' clear,'

I ' cogent,' ' unequivocal,' and ' convincing,' is less commonly used, but nonetheless 'is no stranger to the civil law.' One typical use of I the standard is in civil cases involving allegations of fraud or some other quasi-criminal wrongdoing by the defendant. The interests at stake in those cases are deemed to be more I substantial than mere loss of money and some jurisdictions accordingly reduce the risk to the defendant of having his reputation tarnished

, erroneously by increasing plaintiff's burden of proof."

Addington v. Texas, 441 U.S. 418, 423-24 (1979) (citations and footnotes omitted).

128. Although the Court's explanation occurred in the context of an adversarial proceeding, its discussion of the appropriateness of the clear and convincing evidence standard in resolving factual questions of misconduct is relevant here.

i I Early in this proceeding, the Board recognized the significant adverse social consequences that would befall any employee we I

I _ _ - _ _ _ _ _ _ _ _ _ _ _

I found had falsified leak rate tests. We observed that, notwithstanding the Commission's preservation of the statutory rights of the individual employees, there is a realistic possibility of actual, practical injury to the reputation and professional status of any employee whom we find has committed wrongful acts involving leak rate tests. Memorandum and Order, Docket No. LRp, pp. 2, 11 (March 26, 1986); Tr. 45. This is I the type of " deprivation" and " stigma" that is "more important than the loss of money" and which the Supreme Court has indicated we must scrupulously guard against erroneously casting upon any individual employee. For these reasons, we conclude that the Board can find an employee guilty of leak rate test falsification or dereliction or culpable neglect at TMI-2 only where we are convinced of his misconduct by clear and convincing evidence. See Sea Island Broadcasting Corp. of S.C. v. FCC, 627 F.2d 240, 243 (D.C. Cir. 1980), cert. denied, 449 U.S. 834 (1980) (FCC must use clear and convincing evidence in lieu of traditional preponderance of evidence standard in broadcast license revocation proceeding); see also Motor and Equipment Manufacturers Ass'n v. EPA, 627 F.2d 1095, 1122 (D.C.

Cir. 1979), cert. denied, 446 U.S. 952 (1980) (EPA applied clear and convincing evidence standard required by statute in determining whether state motor vehicle emissions regulations would sufficiently protect public heath).

I I

I 129. Our decision to apply this intermediate standard is, in our judgment, also compelled by the fact that so many years have elapsed since the events at issue in this proceeding. As the Commission noted:

"Because the leak rate falsification events I to be addressed in this Board hearing are than 5 years old, the 5-year statute of limitations set forth in 28 U.S.C. S 2462 more may bar I the NRC from subsequently instituting an enforcement proceeding for involvement in that are the subject of this hearing."

events 22 N.R.C. at 884 n.3. The purpose of having a statute of limitations is to prevent the bringing of legal actions after so much time has elapsed that it may be fundamentally unfair to require the party against whom the charges have been made to defend himself. For example, in American pipe and Construction Co. v. State of Utah, 414 U.S. 538, 554 (1974), the Supreme Court stated:

" Statutory limitation periods are designed to promote justice by preventing surprises through the revival of claims that have been allowed to slumber until evidence has been lost, memories have faded, and witnesses have disappeared. The theory is that even if one has a just claim it is I unjust not to put the adversary on notice to defend within the period of limitation and that the right to be free of stale claims in time comes to prevail over the right to prosecute them."

130. Here, of course, the leak rate tests at issue were routine tasks, typically performed at least once every 8-hour shift. Stier Report, Vol. I, p. 51. Not surprisingly, I

1 I the TMI-2 operators typically could not recall performing specific leak rate tests or the conditions then prevailing at the TMI-2 reactor. Id., p. 8; see, e_. g . , Tr. 1650, 2452, 2459, 2523-24, 2632. Under these circumstances, it would be unfair to expect a TMI-2 operator to recall the circumstances of a specific leak rate test, or the conditions prevalent in the TMI-2 control room at that time. For example, there seems to be no way to reconstruct whether hydrogen was needed in the RCS at any given time (e.g., Tr. 2881); whether boron concentration required adjustment by water addition to the RCS at any given time (e.g., Tr. 3037-44, 3090-94); whether an operator may have thought water was needed for inventory control while a leak rate test was occurring (e.g., Tr. 2540-41); or whether one can be sure that a water addition occurred during a leak rate test (Stier Report, Vol. IV(A),Section VI, pp. VI.13-15). It would be unfair to find, in a close case, that an operator did something improper, when neither he nor his fellow operators could recall the circumstances extant at that time. We therefore conclude that the standard to be applied to the factual issues in this proceeding is one of clear and i convincing evidence.

1 I

I I IV. FINDINGS ON THE TECHNICAL ISSUES.

A. Statutory and Regulatory Background.

1. The Atomic Energy Act of 1954, As Amended.

131. The Atomic Energy Act sets forth the statutory authority that the Commission had to issue licenses to Met Ed and the TMI-2 operators and supervisors in 1977-79. 42 U.S.C.

SS 2133(a), 2137. Furthermore, 42 U.S.C. S 2232(a) requires that applicants for facility licenses have such technical specifications as are necessary to assure the public health and safety and provides that such technical specifications shall be part of any license issued by the Commission. 42 U.S.C. S 2236 f provides that a facility must be operated in accordance with its technical specifications, and that a license may be revoked for violations thereof.

132. Taken together, these provisions of the Atomic j Energy Act define the legal obligations of Met Ed and the licensed operators (including senior reactor operators) at TMI-2. The statute makes clear that the central requirement imposed on Met Ed and the TMI-2 operators was to comply with l their licenses, which incorporated the Technical Specifications for TMI-2, and the Commission's regulations. See Exhibits 10-A I

I and 10-B.

2. NRC Regulations.

133. The regulations of the Commission relevant to this proceeding are 10 C.F.R. S 50.36 and certain provisions of I I

I I 10 C.F.R., Part 50, Appendix A. Section 50.36(b) (1979) stated that each operating license "will include technical specifications. The technical specifications will be derived from the analyses and evaluation included in the safety analysis report, and amendment _s thereto, submitted pursuant to S 50.34. The Commission I may include such additional technical specifications as the Commission finds appropriate."

Section 50.36(c) provides that technical specifications shall include " limiting conditions for operation" and " surveillance requirements" with which the licensee *must comply.

134. Appendix A to 10 C.F.R. Part 50 sets forth

" General Design Criteria for Nuclear Power Plants". Two criteria are applicable to reactor coolant leakage. With respect to pressure boundary leakage (for which the limit was zeto), Criterion 14 states:

I "Reactory coolant pressure boundary.

reactor coolant pressure boundary shall be The designed, fabricated, erected, and tested so as to have an extremely low probability of abnormal I leakage, of rapidly propagating failure, and of gross rupture."

Criterion 30 states:

" Quality of reactor coolant pressure I boundary. Components whicn are part of the reactor coolant pressure boundary shall be designed, fabricated, erected, and tested to the highest quality standards practical. Means shall I be provided for detecting and, to the extent practical, identifying the location of the source of reactor coolant leakage."

The Commission's regulations make clear that, in the context of this proceeding, the obligation imposed on Met Ed and the TMI-2 I

I I licensed operators (including supervisors) was to meet the Technical Specification requirements for RCS leakage. E.g.,

Tr. 5069 (G. Miller) (" tech specs were the law"); see generally Stier Report, Vol. I, pp. 32-33. Those requirements, in turn, required that TMI-2 maintain, inter alia, identified leakage below 10 gpm, unidentified leakage below 1 gpm, and no pressure boundary leakage, in Modes 1, 2, 3, and 4. Stier Report, Vol. V(B), Tab 14. The issues in this case solely concern the 1 gpm LCO for unidentified leakage. CLI-85-18, 22 N.R.C. at 135. If the unidentified leakage exceeded 1 gpm, the

" Action Statement" required that it be reduced to less than 1 gpm "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 />." Stier Report, Vol. V(B), Tab 14, p. 3/4 4-15.

Importantly, however, a satisfactory leak rate test per se is not required by the Technical Specifications themselves; what is required is that unidentified leakage be below 1 gpm during Modes 1, 2, 3, and 4. See Faegre & Benson Report, Vol. One, p.

20 (questioning whether the leak rate test should have been used measure RC3 unidentified leakage); Stier Report, Vol. I, pp. 36-39; id., Vol. V(B), Tab 12 (Reg. Guide 1.45). It was the 1 gpm LCO, not the test result, which was the regulatory requirement imposed on Met Ed. The Commission's regulations did not require that Met Ed report an excessive leak rate test I

I I

I result; rather, they required that Met Ed report a violation of the TMI-2 LCO of 1 gpm for unidentified leakage. 10 C.F.R.

S 50.36(c)(7) (1978).

136. Four different methods were specified for determining whether RCS leakage was within the applicable limiting conditions for operation. Stier Report, Vol. I, pp. 38-39; I_d . , Vol. V(B), Tab 14. The provision for four different methods for determining RCS leakage led to confusion on the part of the TMI-2 operators, as Mr. Stier described:

"No guidance is provided in the Technical I Specifications on how to reconcile differences between leak rate test results and other sources of information about leakage that were available to plant personnel.

"In addition, the Technical Specifications do not define the responsibilities of plant personnel to respond to indications of leakage during the periods between leak rate calculations. These issues became quite significant at TMI-2 because I of the emphasis placed upon leak rate testing and satisfying the administrative requirements for periodic recording of test results at that plant."

Id., Vol. I, pp. 40-41; see Faegre & Benson Report, Vol. One, pp. 19-20.

137. The issues presented in this proceeding arose because the TMI-2 operators had little confidence in the leak rate test, but generally felt, based on other indicators of leakage, that the LCO for unidentified leakage was being met.

E_.S., Adams Prep. St., pp. 2-3, ff. Tr. 3776. Their conclusion that the leak rate test was not accurate has been borne out by I

I I

I I the technical analyses that are part of this record. See, e.g., Stier Report, Vol. IV; Kirkpatrick & Wermeil prep. St.,

ff. Tr. 376. Significantly, Mr. Stier's technical experts used reactor building sump collection data as the most reliable source of data on unidentified leakage. Stier Report, Vol.

IV(A),Section IV, pp. IV .4.- 2 . And, as Faegre & Benson concluded, there is a substantial question whether the leak rate test should even have been used for determining unidentified leakage, rather than for determining intersystem leakage. Faegre & Benson Report, Vol. One, p. 20.

3. NRC Regulatory Guide 1.45 Was Explicitly Incorporated Into The Bases for the TMI-2 I Technical Specifications. -

138. Regulatory Guides are issued by the Commission to discuss, at greater length, the general requirements contained in the Code of Federal Regulations. Regulatory Guide 1.45 specifically deals with the regulatory requirements applicable to RCS leakage. Stier Report, Vol. V(B), Tab 12.

It emphasizes the importance of creating the means to locate and quantify sources of leakage so that leakage from unidentified sources can be closely monitored. d Id., Vol. I,

p. 33.

139. " Identified leakage" is defined in Reg.

Guide 1.45 as " leakage into closed systems", for example, pump seal or valve packing leaks that are I

I I

I I " captured, flow-metered, and conducted to a sump or collecting tank . . . or leakage into containment atmosphere from sources that are both I specifically located and known either not to interfere with the operation of unidentified leakage monitoring systems or not to be a flaw in the RCpB [ reactor coolant pressure boundary)."

Stiet Report, Vol. V(B), Tab 12, p. 1.45-1. Reg Guide 1.45 explains that

" leakage to the primary containment from I identified sources should be collected or otherwise isolated so that (a) the flow rates are monitored separately from unidentified leakage I and (b) the total flow rate can be established and monitored."

Id., p. 1.45-3.

140. With respect to unidentified leakage and the reasons for the 1 gpm LCO, Reg. Guide 1.45 states:

" Uncollected leakage to the containmer.t I atmosphere from sources such as valve stem packing glands and other sources that are not collected increases the humidity of the I containment. The moisture removed from the atmosphere by air coolers together with any associated liquid leakage to the containment is l known as ' unidentified leakage' and should be collected in tanks or sumps where the flow rate

can be established and monitored during plant l operation. A small amount of unidentified l l leakage may be impractical to eliminate, but it

. E should be reduced to a small flow rate, l

preferably less than one gallon per minute (gpm),

'3 to permit the leakage detection systems to detect g positively and rapidly a small increase in flow rate. Thus a small unidentified leakage rate that is of concern will not be masked by a larger i acceptable identified leakage rate."

Id., p. 1.45-1.

141. With respect to "intersystem leakage", Reg.

lg Guide 1.45 states:

ig l

E I

I

" Substantial intersystem leakage from the RCPB to other systems across passive barriers or I valves is not expected. However, should such leakage occur, it may not be detectable through the above-mentioned detection systems, and other alarm and detection systems should be employed.

I For example, steam generator leakage in pressurized water reactors (PWR's) should be monitored to detect tube or tube sheet leaks." '

Id., pp. 1-2.

142. The leak rate test is referred to only once in Reg. Guide 1.45, and that reference makes clear that it was intended as a measure of intersystem leakage, not unidentified leakage. See Faegre & Benson Report, Vol. One, pp. 19-20. The Reg. Guide states:

I "Since intersystem leakage does not release reactor coolant to the containment atmosphere, detection methods should include monitoring of water radioactivity in the connected systems where the systems flows [ sic] through the containment boundary, and monitoring of airborne i radioactivity where such systems are vented i g outside the containment boundary. Another fg important method of obtaining indications of uncontrolled or undesirable intersystem flows would be the us_e of a water inventory balance, designed to provide appropriate information such as abnormal water levels in tanks and abnormal water flow rates."

l l5 Stier Report, Vol. V(5), Tab 12 (Reg. Guide 1.45), p. 1.45-2 (emphasis added).

143. The " Bases" for TMI-2 Technical Specification 3.4.6.1 (" Leakage Detection Systems") state:

"The RCS leakage detection systems required by this specification are provided to detect and monitor leakage from the Reactor Coolant Pressure Boundary. These detection systems are consistent with the recommendations of Regulatory Guide I

I I

I 1.45, Reactor Coolant Pressure Boundary Leakage Detection Systems, May 1973."

Id., Tab 14. Mr. Stier observed that:

"The basis of this Technical Specification I explains that it is intended to be consistent with Reg. Guide 1.45. However, this Technical Specification does not specify the leakage limits I that it is intended to monitor, nor does it require a leak rate test."

M., Vol. I, pp. 36-37 (footnote omitted).

144. Although, as noted above, the " Bases" of Technical Specification 3.4.6.1 state that "these detection systems are consistent with the recommendations of Reg. Guide 1.45. . .", there are significant differences between them. As Mr. Stier explained:

"There are two significant differences between TMI-2 Technical Specifications and Reg. Guide I 1.45. First, the leak rate test is included in the Technical Specifications as a surveillance requirement without limiting it to the I measurement of any particular form of leakage.

As noted above, Reg. Guide 1.45 only makes reference to the leak rate test as a means of monitoring intersystem leakage. Second, in I contrast to Reg. Guide 1.45 that emphasizes the continuous nonitoring of leakage, the Technical Specifications establish maximum time intervals I for periodic monitoring of the leakage detection systems to demonstrate compliance with limiting conditions for operation."

M., Vol. I, p. 40 (emphasis in original).

145. Therefore, there is a fundamental inconsistency I between various provisions of the TMI-2 Technical Specifications. By incorporating Reg. Guide 1.45, which contemplates use of a leak rate test as a measure of I

I I

E y intersystem leakage, and also referring to such a test as one L

of several measures of " reactor coolant system leakages" (T.S.

[ 3.4.6.2), the TMI-2 Technical Specifications contained a L

crucial ambiguity. See Faegre & Benson Report, Vol. One, pp.

r L 19-20 (RCIs test should have been used to detect intersystem leakage, not unidentified leakage). Therein, lay the seeds of the problem, because "it was not clear to the personnel involved as to which set of data came within the T.S.

requirements and when the time requirements of the action statement were applicable."

Stier Report, Vol. V(C), Tab 29 (Narrative to Met Ed's November 1, 1978 Licensee Event Report No. 78-62/1T). That Narrative, about which we will have more to say later in this Decision, describes accurately the confusion that TMI-2 operators had about their responsibilities with respect to the leak rate test. See Id., Vol. I, pp. 40-41.

4. NRC Licenses.

146. The NRC licenses applicable to the leak rate issue were the Met Ed facility license for TMI-2; the reactor operator ("RO") licenses held by the CROs; and the senior reactor operator ("SRO") licenses, held by TMI-2 shift foremen, shift supervisors, the Supervisor of Operations, and certain other TMI-2 management personnel.

E

[

i

a. TMI-2 Facility License.

147. The TMI-2 facility license was, of course, made subject to compliance with the TMI-2 Technical Specifications.

I_d., Vol. I, p. 36. Thus, Met Ed had a duty to comply with the TMI-2 Technical Specifications concerning RCS leakage, including the 1 gpm LCO for unidentified leakage, the surveillance requirement for performance of a 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," and the provisions of Reg. Guide 1.45. Id., Vol. V(B), Tab 14.

" Administrative Procedures" were not explicitly incorporated into the TMI-2 Technical Specifications or the TMI-2 facility license. Tr. 1166-68, 1176-77.

b. TMI-2 Reactor Operator Licenses.

148. Each CRO held an RO license for TMI-2 which required that he " manipulate all controls" of TMI-2 subject to 10 C.F.R. S 55.31 (which in turn required that he

" observe . . . all applicable rules, regulations and orders of the Commission") and that he:

" observe the operating procedures and other conditions specified in the facility license I which authorizes operation of the facility or facilities and shall comply with the following conditions [e.g., the wearing of eyeglasses]."

Exhibit 10-B. Importantly, the obligation on the CROs was that they comply with "the operating procedures and other conditions specified in the facility license", which did not explicitly I

I incorporate the TMI-2 Administrative Procedures. Tr. 1166-i,,

1176-77, I

c. TMI-2 Senior Reactor Operator Licenses.

149. The obligations imposed on the holders of TMI-2 SRO licenses were identical to the obligations imposed on the holders of RO licenses and, in addition, they were authorized to " direct [] the licensed activities of licensed operators".

Exhibit 10-A. This additional responsibility was important because the leak rate test, as part of the " Surveillance Testing Program", was required to be approved by an SRO, which typically was the shift foreman. See, e.g., id,, Vol. V(A),

Tab 6 (" Position Description" for shift foreman); Stier Report, Vol. I, p. 48. On occasion, however, as a review of the retained leak rate tests shows, shift supervisors approved the I leak rate test. See, e.g., NRR Test No. 13.

5. TMI-2 Technical Specifications Applicable to RCS Leak Rate Testing.

a. The Technical Specifications Required Performance of a Leak Rate Test At Least Once Per 72 Hours, During " Steady State" Operation, In Certain Modes.

150. The " Surveillance Requirements" in TMI-2 Technical Specification 4.4.6.2 provided that:

" Reactor Coolant System leakages shall be demonstrated to be within each of the above I limito by;

a. Monitoring the containment atmosphere particulate radioactivity monitor at least once per 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />.

I

b. Monitoring the containment sump inventory and discharge at least once per 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />.

Measurement of the CONTROLLED LEAKAGE from I

c.

the reactor coolant pump seals when the Reactor Coolant System preesure is at 21551 50 psig at least once per 31 days.

d. performance of a Reactor Coolant System water inventory balance 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."

Stier Report, Vol. V(B), Tab 14, pp. 3/4 4-15 to 4-16. These requirements were applicable in Modes 1, 2, 3, and 4. M.,

Vol. I, p. 42. "According to the Technical Specification definition of modes, this required the performance of a leak rate test at times when reactor coolant system temperature averaged greater [than] 200*F." M. This requirement was confusing because there was no guidance in the Technical Specifications to the TMI-2 operators as to their obligation if the results of the means of detecting RCS leakage produced inconsistent results. TV 136-145, supra. Also, the above-quoted language was unclear in that it failed to specify what the applicability of the leak rate test result was if, the reactor was not in " steady state" during the test. Lack of

" steady state" invalidated a leak rate test, particularly in light of the guidance the operators were given to disregard observed plant indicators if the operators had good reason for doing so. Stier Report, Vol. V(C), Tab 35; Floyd Prep. St.,

pp. 3-4, ff. Tr. 4894; see, e.g., Tr. 3057.

l I

I L

b. The Applicable Procedure Did Not Prohibit Water or Hydrogen Additions to

{ the Reactor Coolant System During a Leak Rate Test.

151. Although TMI-2 Surveillance Procedure 2301-3D1 states, in its " Limits and Precautions" (Section 3.1), that additions and removals of water from the RCS and make up F systems should be " avoid [ed]", and that " chemical addition [s]"

L "should not be conducted during this test" (Faegre & Benson Report. Volume Three A, Exhibit 18), the same Procedure states (in Section 6.3):

"If changes to the RCS inventory must be made during the performance of this test, they must be accounted for using Data Sheet 4.

[ Operations such as adding water to the Make-up Tank or sampling the RCS may be accounted for in this manner, however, these should be avoided if

[ at all possible."

152. Thus, the leak rate test procedure did not

[ prohibit water or hydrogen additions to the RCS during a test.

As Mr. Stier stated:

{

"Neither the addition of hydrogen nor water to the makeup tank was specifically prohibited by

[ the test procedure. Although water additions were discouraged by the test procedure, theoretically they should not have adversely affected the accuracy of the test if they were

{ accounted for in the test calculation.

"The procedure made no reference to hydrogen

[ additions. Although ' chemical

• additions are prohibited during the course of leak rate tests, there is no indication in the procedure that it

( was intended to bar a change in hydrogen overpressure."

[

[

I Stier Report, Vol. I, p. 88; see Tr. 634-35; 627-29, 1481. The leak rate test sheet printed out by the TMI-2 computer asked the operator to enter the amount of water from "DS [ Data Sheet]

4", which was the means by which water additions during the test were recorded. See, e.g., Stier Report, Vol IV(C), Test No. 1; Faegre & Benson Report, Vol. III-A, Exhibit 18 (Section 6.3). Therefore, provided the water additions were accounted for in the test calculation, the operators responded correctly.

153. The TMI-2 operators and supervisors received only on-the-job training concerning the leak rate test. See, 11 315-26, infra. It was thus reasonable for the operators and supervisors to assume that water additions to the MUT during a leak rate test were permissible, because the computer called for an entry corresponding to the amount of the water addition without cautioning against such an addition.

154. In contrast, the leak rate test sheet from the computer did not ask the operators whether hydrogen was added to the MUT during a test. See, e.g., Stier Report, Vol. IV(C),

Test No. 1. Therefore, because the operators were not precluded from adding hydrogen during leak rate tests by the test procedure, and because the computer did not ask whether hydrogen was added to the MUT during a test, the operators were entitled to assume (and were correct in so assuming) that hydrogen additions to the MUT during leak rate tests did not violate the leak rate test Surveillance Procedure.

I I

I I 155. Accordingly, we cannot find that water or hydrogen additions during leak rate tests were, per se, violations of the TMI-2 leak rate Surveillance Procedure,

c. The Leak Rate Technical Specification Was Ambiguous As to the Required Frequency of Performance of the Test.

156. Technical Specification 4.4.6.2 provided that a leak rate test was to be performed "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." Stier Report, Vol. V(B), Tab

14. However, pursuant to a Met Ed-imposed requirement, TMI-2 operators typically performed at least one leak rate test every shift. Id., Vol. I, p. 51.

157. The frequency of performance of leak rate tests originated with the same practice at TMI-1. Id., Vol. I, pp.

51-52; Floyd Prep. St., p. 2, ff. Tr. 4894; see Faust Prep.

St., p. 2, ff. Tr. 2511 (" company requirement"). The increased frequency of performance of the leak rate test beyond that required led to a general belief on the part of the operators that they could invalidate leak rate tests over 1 gpm and remain in compliance with the Technical Specifications. Stier Report, Vol. I, p. 121; Floyd Prep. St., pp. 3-4, ff. Tr. 4894; Faust Prep. St., p. 2, ff. Tr. 2511. Thus, as the LER filed by i Met Ed with the NRC on November 1, 1978 stated:

"At 1000 hours0.0116 days <br />0.278 hours <br />0.00165 weeks <br />3.805e-4 months <br /> on October 19, 1978, while performing Surveillance Procedure 2301-3DL, it was determined that data obtained subsequent to the last recorded acceptable surveillance

~

performance at 1935 on 10-16-78 showed that

I unidentified leakage during the interim period exceeded the limits specified in the Technical Specifications (T.S. 3.4.6.2) and that the required action statement was not invoked. The largest unidentified leakage during this period was 2.6 gpm.

"This event was caused by misinterpretation of the requirements of the T.S. Since the actual I frequency of performance of the surveillance procedure was greater than that required by the T.S., it was not clear to the personnel involved as to which set of data came within the T.S.

I requirements and when the time requirements of the action statement were applicable. . . .

Stier Report, Vol. V(C), Tab 29 (emphasis added). However, as we explain in 1 340-68 infra, the appropriate instruction to the TMI-2 operators did not alter their erroneous reading of the Technical Specifications.

d. The Applicable Surveillance Procedure Was Ambiguous With Respect to Invalidation of a Leak Rate Test.

158. Surveillance Procedure 2301-3D1 (Stier Report, Vol. V(C), Tab 19) provided that the leak rate test was to be performed at least once every 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> "during steady state operation" in Modes 1-4. " Steady state" was not defined in TMI-2 Technical Specifications (see Faegre & Benson Report, Volume Three A, Exhibit 3), nor was its meaning clear to the TMI-2 operators (Floyd Prep. St., pp. 3-4, ff. Tr. 4894). Not surprisingly, therefore, it was not clear whether a leak rate test which was performed during a period of significant change in key parameters could be invalidated due to lack of " steady I

I I

state". This uncertainty was what gave rise to most of the leak rate test problems at TMI-2. Tr. 4675-77.

159. The Surveillance Procedure goes on to state:

I "3.2 The RCS and make up system should be maintained in a steady state condition during this test. Changes in valve line-ups, I coolers-in-service, pumps-in-service, etc. should be avoided.

"3.3 For the most accurate determination of the RCS leak rate, the initial and final conditions of reactor power, RCS temperature, pressure and pressurizer level should be identical.

l "3.4 The same sources should be used when recording initial and final RCS temperature, I pressurizer level, make up tank level and RCDT level. Differences in sources could be misinterpreted as RCS leakage when comparing successive readings.

I "3.5 Minimize power level variations during this test."

Stier Report, Vol. V(C), Tab 19; see id., Vol. I, pp. 42-43.

Each of the above-quoted instructions to the operators was reasonably read to mean that the leak rate test results were likely to be inaccurate (and, hence, could be invalidated) if reactor conditions changed during the test. See, e.g., Tr.

1357-61, 2206-09 (Capra); Floyd Prep. St., pp. 3-4, ff. Tr.

4894; Adams Prep. St., pp. 2-3, ff. Tr. 3776.

160. Surveillance Procedure 2301-3D1 clearly stated that leak rate tests could be rendered inaccurate if important plant parameters changed during the test; that is what the TMI-2 operators and supervisors observed. E.g., Adams Prep.

St., pp. 2-3, ff. Tr. 3776.

I Obviously, TMI-2 was a new plant I

in 1978-79 (it went commercial on December 31, 1978); it was not only to be expected that plant oscillations would be more pronounced during such a period, but in fact they were. See, e.g., Exhibits ll-A and ll-B; Faegre & Benson Report, Vol.

Three B, Exhibits 37 and 38 (graphs depicting plant oscillations); Tr. 1120-21.

161. Mr. Floyd confirmed that leak rate tests were discarded when they were considered invalid:

I "Similarly, throwing out leak rates was not by direction but something that grew out of Operations and was carried over from Unit 1 to Unit 2 by me, the supervisors, and the foremen.

i I am now aware that the control room operators believed that I thought it permissible to throw invalid leak rates away. I, however, have no I specific recollection of discussing with any CRO whether leak rates should be discarded. It was my opinion that blatantly bad leak rates (for I example, excessively negative ones) had no connection with reality; consequently, it was permissible to discard them. The shift I supervisor had the ultimate responsibility for throwing out leak rates. I never knew that the NRC was unaware that we were discarding leak rates. I now believe that discarding leak rates I was inappropriate, but I did not have that opinion at the time, for the reasons I have stated."

Floyd Prep. St., pp. 2-3, ff. Tr. 4894. Mr. Capra confirmed that it was not appropriate to invoke the Action Statement when a leak rate test was obviously invalid. Tr. 2206-09.

I 6. Met Ed Administrative Procedures Were Not Legal Requirements Imposed on Met Ed or the Licensed Operators.

162. The licenses issued to Met Ed and the TMI-2 operators incorporated the TMI-2 Technical Specifications and I I

I NRC regulations, but did not explicitly incorporate the TMI-2 Administrative Procedures. Exhibits 10-A and 10-B; see Tr. 1177. Therefore, failure to observe TMI-2 Administrative Procedures did not constitute violations of law. Rather, those failures merely violated Met Ed-imposed requirements, unless a given Administrative Procedure was specifically incorporated into the Technical Specifications. See, e.g., Surveillance Procedure 2301-3D1 (Stier Report, Vol. V(C), Tab 19, Section 6.4.4 (Paragraph 4)) (incorporating Administrative Procedure 1014).

163. Mr. Russell nonetheless took the position that the TMI-2 Administrative Procedures were a part of the legal requirements imposed on the TMI-2 operators. Tr. 1166. His reasoning was that those Administrative Procedures were incorporated by reference in Technical Specification 6.8:

"TMI-2 technical specification 6.8 entitled

' Procedures

• requires written procedures be established, implemented and maintained in areas of surveillance and test activities of safety-related equipment and specifically includes those procedures identified in Appendix A of Regulatory Guide 1.33, November 1972."

Tr. 1166-68. The Board disagrees with this Staff position, for several reasons.

164. The fact that the licenses issued by the Commission imposed on the operators a duty to comply with technical specifications does not, in our judgment, mean that every " recommendation" in every Regulatory Guide referred to in I

! l I

the technical specifications somehow becomes a legal obligation imposed on the license holder. That is too great a stretch.

If the Commission intends that company administrative procedures be legal requirements, it should either explicitly refer to them in the licenses it issues, or state in plain English in those technical specifications that the company's administrative procedures are incorporated by reference.

Neither was done in this instance, and it is not legally appropriate to adopt such a policy ex post facto.

165. In this instance, " Regulatory Guide 1.33" (which is actually entitled " Safety Guide 33" (Tr. 1173)) merely recommends (id.) that there be administrative procedures, but provided no guidance to Met Ed or the TMI-2 operators as to what procedures were required. As Mr. Russell conceded:

"All [the operator] would get out of reading Reg. Guide 1.33 or, in this case, Safety Guide I 33, would be that there must be a procedure entitled, for instance, ' Shift Relief and Turnover.' It would give him no substance as to I what he is required to do as a part of Shift relief and turnover. He must, in fact, go to the facility procedures to find out what it is he has to do."

Tr. 1176-77. This is precisely the problem. Mr. Russell conceded that it would have been "very unusual" for the TMI-2 operators to have even seen Reg. Guide 1.33, let alone know what was in it, because Regulatory Guides are:

" generally a licensing document used by a licensing engineer or the group that is I

l 1

I responsible for drafting procedures. Not necessarily for training.

Tr. 1178.

166. Before a person can be held to have violated a law, it must be clear what the law is. papachristou v. City of Jacksonville, 405 U.S. 156, 162 (1972).

167. When asked whether TMI-2 facility license included any explicit reference to the TMI-2 administrative procedures, Mr. Russell further admitted:

". . . in general, the facility license itself references the regulations in effect and the I regulations at 50.36 require administrative procedures. The license, however, does not reference, or the technical specifications, specific administrative procedures of the facility. It only requires that they exist. The numbering, for example, would not be in the license or the technical specifications."

Tr. 1177. Mr. Russell's testimony demonstrates that the TMI-2 operators were not on notice that Met Ed Administrative procedures were legal requirements as part of the TMI-2 license or Technical Specifications. Mr. Russell admitted that Regulatory Guides are intended for the licensing employees of facility license holdera such as Met Ed, not for plant operators and supervisors. Tr. 1178-79. If the operators never even saw Safety Guide 33, it is impossible to conclude that they could have known what it supposedly required of them, or known that the mere reference to " Regulatory Guide 1.33" in the TMI-2 Technical Specifications made the TMI-2 I

g I

l l

Administrative Procedures legal requirements imposed on the operators.

168. The Board concludes that, to find that a violation of law occurred, it must look to the operators' and supervisors' licenses, and in turn to NRC regulations and TMI-2 Technical Specifications, not to Met Ed's Administrative Procedures.

169. Even if Mr. Russell's view were correct that the single, obscure reference to Reg. Guide 1.33 in TMI-2 Technical Specification 6.8.la does constitute the wholesale incorporation of the TMI-2 Administrative Procedures into the Technical Specifications, we think his subsequent explanation of the reality of the situation is illuminating. After thinking overnight about his testimony, Mr. Russell clarified his views. He stated:

"Let me start by rereading question 269:

Were the administrative procedures applicable to TMI-2 included in the legal requirements imposed on the TMI-2 operators?

"I answered that in a narrow sense of what I was the chain of responsibility to the individual operators. I believe it is also very significant and relevant that I believe the operators did not I understand that. And I base that upon my interviews of the operators, where I specifically queried them on their understanding of administrative procedures, their legal obligations, et cetera.

"This was a part of the proceeding that was I not included in the Staff's report because it related to current knowledge, current understanding. But in the process of getting information on current knowledge, current I 1

r L

F L

, understanding, it also became obvious that at the

{ time of the accident they didn't have that.

. . . So the response I gave was a narrow P one and did not address their understanding at b that time.

m "I felt that was rather significant and the L record ought to reflect that."

Tr. 1224-26. The Board agrees that, even if the TMI-2 Administrative Procedures were legal requirements, it is "very p significant and relevant" that the TMI-2 operators and L.

supervisors did not understand their obligations. The

[ Commission and facility license holders must do a better job than was done here to explain legal obligations to operators and supervisors.

170. Accordingly, the Board finds that, although it would have been better practice for the operators and supervisors to have followed the TMI-2 Administrative Procedures, they were not legally obligated to do so.

Therefore, we do not find any TMI-2 operator or supervisor to have violated his license, NRC regulations, or the TMI-2 Technical Specifications because he failed to comply (in whole or in part) with Administrative Procedures 1010 (Technical Specification Surveillance Program), 1012 (Shift Relief and Log Entries), and 1036 (Instrumentation Out of Service Control).

B. The 1 Gpm LCO for Unidentified Leakage Had No Safety Significance.

{ 171. Exhibit 22 includes selected pages from Met Ed's Final Safety Analysis Report ("FSAR"). The FSAR analysis

[

U

I I concluded that the lowest RCS leak of safety significance was 9 gpm. After a detailed engineering analysis of the amount of leakage expected from a critical crack in RCS piping, the FSAR stated:

" Based upon the above analysis the critical I parameter would be a crack in the cold leg piping of approximately 9 gpm. In establishing a maximum unidentified leakage, the following criteria are considered:

I 1. The magnitude of the leakage should be well below the leakage associated with a crack of critical size.

2. The magnitude should be well within the I capability of the normal makeup system.

3. The magnitude should be sufficiently large I 4.

to allow for ease of detection within a reasonable period of time.

Offsite releases should be within 10 CFR 20 limits.

Accordingly, a 1 gpm leak was selected as the maximum allowable unidentified leakage rate.

This value is well below the leakage associated I with a crack of critical size. It can be detected within a reasonable period of time as discussed previously. . . ."

Exhibit 22, p. 5.2-25; see also Stier Report, Vol. V(B), Tab 12 Regulatory Guide 1.45, pp. 1.45-1 to 1.45-2 (basis for selection of 1 gpm (instead of some higher number less than 9 gpm) as the LCO for unidentified leakage is " industry practice"). In other words, the selection of 1 gpm was arbitrary. The engineering analysis confirms, as quoted above, that critical crack size was associated with a 9 gpm leak, not a 1 gpm leak.

172. Most prominent among the considerations underlying the choice of the 1 gpm LCO for unidentified leakage I

I

I L

was supposed ease and capability of detection, rather than just l

L safety considerations, as Mr. Stier confirms. Stier Report, p Vol. I, p. 34. As Mr. Stier stated at the hearings: "the b important thing about Reg. Guide 1.45 is that it indicated that industry experience was that you could measure that amount of leakage. . . . Tr. 416. However, as we find in 4H 184-224, -

supra, the numerous defects in the leak rate test procedure rendered it incapable of detecting 1 gpm of unidentified L

leakage.

C. Performance of the Leak Rate Test.

173. The manner in which the TMI-2 operators actually performed the leak rate test was, generally, as follows:

1. Frequency of Tests.

174. The TMI-2 operators typically performed a leak rate test on each eight-hour shift. Stier Report, Vol. I, p.

51; Exhibit 5-A, Enclosure 1, p. 9. A test was performed by every shift if the computer was available and if the performance of other shift evolutions and responsibilities did not interfere. The practice of performing a leak rate test on every eight-hour shift was established at TMI-1 and was subsequently implemented at TMI-2. Stier Report, Vol. I, p.

52.

2. Test Performance Procedure.

175. TMI-2 Surveillance Procedure 2301-3D1 governed the performance of the leak rate test and is set out at Stier

[

a

I Report, Vol. V(C), Tab 19. The actual performance of a leak rate test was rather routine and uncomplicated. It was often performed (although not always) by the " switching and tagging" CRO, thac is, the CRO who was not manipulating the controls of the plant (i.e., " assigned to the panel"). If there were three CROs or more on shift, one would be assigned to the panels, one to " switching and tagging", and one to surveillance tests.

I E.g., Olson Prep. St., pp. 2-3, ff. Tr. 3911. The CRO performing the test would instruct the computer, which was located in the TMI-2 control room, to calculate the RCS leakage by entering the appropriate code ("RCSL"). E.g., Cooper Prep.

St., pp. 3-4, ff. Tr. 2835. At the same time, the CRO would I specify the time interval in which the test was to be performed (which, in practice, was always one hour). The computer would then conduct the leak rate test and at the end of the hour would request certain information of the CRO to complete the test. The CRO was asked to enter any known (or identified) leakage and any operator-caused change to the volumes of the RCDT or RCS. With that information, the computer would print the initial and final values of the parameters used in the leak rate test and the calculated values for the gross leakage, identified leakage, and unidentified leakage. Stier Report, Vol. 1, pp. 52-53; Exhibit 5-A, Enclosure 1, p 2.

176. Generally, although not always, the CRO who performed the test signed the sheet printed out by the I

I I

I I computer; on occasion, however, a licensed operator other than the CRO who performed the test signed it. Stier Report, Vol.

I, p. 54; e.g. NRR Test Nos. 96, 98, 108, 154, Tr. 3986-87.

The test was then approved by an SRO, usually the shift foreman on duty during the performance of the leak rate test. Exhibit 5-A, Enclosure 1, p.2.

3. Entry of Identified Leakage and Other Information.

177. Surveillance Procedure 2301-3D1 asked the operator, at the end of the test interval, to enter any known identified leakage into the computer. Stier Report, Vol. I, p.

53; Exhibit 5-A, Enclosure 1, p 2. However, most of the leak rate tests do not reflect any entry of identified leakage. Of the 222 leak rate tests examined by Mr. Stier, only 34 (or approximately 15 percent) contain an entered value for identified leakage. Stier Report, Vol. I, p. 53. Of course, the operators were not required to take credit for identified leakage if it was not needed to produce an acceptable test result. Tr. 655-57 (Russell).

178. The operators generally entered any additions to the RCS inventory. There were, according to Mr. Stier, thirty-one water additions during the performance of the 222 leak rate tests that he reviewed. Stier Report, Vol. IV(A),

I Table VI-l (water additions during RCIB tests at TMI-2).

Twenty-four of these water additions were accounted for in the tests during which they were made. Moreover, an additional I

I I

I I three water additions were documented in the control room log.

Thus, only four water additions during leak rate tests went unrecorded in both the leak rate test and the control room log over the one year of TMI-2's operation. Id., Vol. I, pp. 53-54.

4. Control Room Organization.

179. As we noted above in H 179, supra, either the

" switching and tagging" CRO or, if there were three CROs, the CRO assigned to perform surveillance tests, was ordinarily responsible for the performance of the leak rate test.

However, in actual practice, any CRO might conduct some part, or all of, a leak rate test. I_ d . , Vol. 1, p. 49 ("in practice, the division of responsibilities among CRO's was not rigidly enforced and their duties overlapped somewhat. The switching and tagging CRO might make entries in the control room log and the CRO assigned to the control panel might, on occasion perform a leak rate test."); see Tr. 3995.

180. Generally, the CRO performing the leak rate test would inform the CRO assigned to the panel that the computer was being instructed to conduct a leak rate test. On occasion, however, the CRO performing the test failed to so alert the CRO at the panel. This lack of communication led to occasional mistakes in the performance of the test. Stier Report, Vol. I,

p. 52.

I I I

L r

9

, 5. Loqqing of Test Performance and Test Results.

1 L 181. There are no log entries indicating the starting p times of the 222 tests examined by Mr. Stier. Stier Report, L

Vol. I, p. 55. Ninety-two of the 222 tests are logged in both the CRO log and the shift foreman log (even though the shift foremen were not required to log leak rate tests); an additional 76 tests were logged by the shift foremen but not by the CROs; finally, two tests were logged by the CROs but not by the shift foremen. Id., Vol. IV(A),Section V, Table V-3.

Fifty-two of the 222 tests were recorded neither by the CROs nor by the shift foremen, including eight tests that were signed by neither the operator nor by the shift foreman. Id., Vol. I,

p. 54; id., Vol. IV(A),Section V, Table V-3.

6. Review and Approval of Test Results.

182. At the conclusion of the leak rate test interval and the entry into the computer of tl.e required information (identified leakage, any change in the RCDT, and any change in the RCS volume), the computer would print out the calculated RCS leakage including the unidentified leakage. Typically, if the leak rate test depicted unidentified leakage below 1 gpm (except, in some cases, if it was negative) a CRO would sign the test and give it to the shift foreman or shift supervisor.

The shift foreman or shift supervisor would examine the leak rate test sheet to confirm that the calculated unidentified RCS

{ 1eakage was below 1 gpm. If he was satisified with the result,

[

he would approve the test sheet and sign it. Stier Report, Vol. I, p. 54. The test sheet was thereupon submitted to the GMS coordinator for filing. Id., p. 55.

I 7. Restarting of the "72-Hour Clock".

183. Although a satisfactory leak rate test was not required by the Technical Specifications, operators and supervisors used the test as a means of determining unidentified leakage. Stier Report, Vol. I, p. 57. The test was to be performed at least once every 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> during " steady state" operation, in Modes 1-4. I_ d . , p. 50. In fact, that was accomplished. See Stier Report, Vol. III(A), Table 1. Upon obtaining a satisfactory test, the "72-hour clock" would restart. E.g., Faust Prep. St., p .2, ff. Tr. 2511.

D. There Were Numerous Defects in the Leak Rate Test.

184. This Board recognized early in this proceeding that "[tlhe overall accuracy or inaccuracy of the test will be sort of a bottom-line question." Tr. 776. The TMI-2 leak rate test procedure was intended to measure, at the beginning and at the end of the test interval, the temperature in the two RCS hot legs, the temperature in two of the four RCS cold legs, the pressurizer level, the MUT level, and the RCDT level. Stier Report, Vol. IV(B), Appendix A, pp. A.6-7. However, the leak rate test procedure had numerous defects.

I I I

I I 185. Some of the defects in the TMI-2 leak rate test procedure produced errors in the test results that were independent of the errors produced by other defects in the test procedure; others produced errors that " amplified" the errors produced by other defects. For example, the leak rate test error produced by RCS oscillations amplified the leak rate test error produced by the failure of the test to account for changes in RCS pressure. Some defects produced errors that, under certain circumstances, " cancelled out" the errors produced by other defects. Tr. 804-08; 849; 888-89. We discuss each significant defect below. This discussion includes all but one of the errors identified by MpR (see Stier Report, Vol. IV(A),Section III, especially Table III-2).

Other errors, such as that occurring during an RCDT pumpdown (Faegre & Benson Report, Vol. One, p. 23), the possiblility that the average of the RCS hot leg and cold leg temperatures would not equal RCS mean temperature (Stier Report, Vol. IV(A),

Sectin III, p. III.4), and the slight misstatement of RCS volume (Kirkpatrick & Wermiel prep. St., Attachment 4, ff. Tr.

376), were inapplicable to most tests or had only a negligible effect on the test results, and accordingly are not discussed.

1. Lack of RCDT Density Compensation.

186. The leak rate test procedure failed to correct for a difference in density that existed between RCS leakage collected in the RCDT and the water in the RCS itself. The I

I I

I I difference in density was caused by the different temperatures of the water in the RCDT and in the reactor. This defect in the leak rate test procedure produced a significant error in the test calculation and seriously compromised the accuracy of the test. Faegre & Benson Report, Vol. One, pp. 21-22; Stier

. Report, Vol. IV(A),Section III, Table III-2; id_., Appendix A, pp. A.16-18; Kirkpatrick & Wermiel Prep. St., Attachment 4, ff.

Tr. 376.

187. Average RCS temperature was approximately 581 F.

Stier Report, Vol. IV(A),Section VI, p. VI.4. The RCS leakage collected in the RCDT was cooled to approximately 85*F. The failure of the leak rate test procedure to account for this difference produced an error that was typically about 40 percent of the total RCS leakage added to the RCDT during a leak rate test. Tr. 853-54; Faegre & Benson Report, Vol. One,

p. 21. "The temperature compensation error was significant and affected virtually every leak rate test." Faegre & Benson Report, Vol. One, p. 22. The impact of the error on calculated unidentifed leakage ranged between 0.07-2.10 gpm. Stier Report, Vol. IV(A),Section III, Table III-2; Tr. 843. This was "a very significant error." Tr. 843. For example, if the actual unidentified leakage was zero but the identified leakage collected in the RCDT was 2.5 gpm, the value for unidentified leakage increased by 1 gpm due solely to this error. Tr. 844.

From mid-February 1979 to March 16, 1979, identified leakage I

, -100-I

I exceeded 2 gpm. Tr. 844-45, 848; Stier Report, Vol. IV(A),

Section IV, Figure IV-14 (RCDT collection rate and other identified leakage).

188. This defect in the TMI-2 leak rate test procedure was corrected by a Temporary Change Notice ("TCN")

that became effective on March 16, 1979, twelve days before the TMI-2 accident. Stier Report, Vol. V(C), Tab 21. The TCN required that each leak rate test be corrected by a hand calculation to compensate for the failure of the test procedure to correct the temperature of the RCS leakage to the average temperature of the RCS. Tr. 857; Faegre & Benson Report, Vol.

One, p. 22. There is no evidence in the record that explains how this defect was discovered or why it took so long to I discover it. Tr. 858.

2. Lack of Density Compensation for Additions to the MUT.

189. The leak rate test procedure failed to account for the difference in density between water in the MUT and in the reactor because of the difference between the temperature of the water added to the MUT and the average temperature of the RCS. Stier Report, Vol. IV(B), Appendix A, pp. A.16-18; id., Vol. IV(A),Section III, Table III-2; Faegre & Benson Report, Vol. One, pp. 21-22. This defect produced an error in i test.

I

-101-I

I 190. There were only 31 leak rate tests during which water was added. Stier Report, Vol. IV(A),Section VI, Table VI-l (listing 31 of 222 tests evaluated). Therefore, the failure to correct for the difference in density between water added to the MUT and water in the reactor did not affect every leak rate test. In any event, this error offset (in whole or in part, depending upon the amount of water added) the corresponding error in the leakage to the RCDT, if both occurred during a leak rate test, until the TCN was adopted on March 16, 1979. Tr. 1958 (Stier).

191. The failure to account for the difference in density between additions to the MUT and in the reactor was not discovered and corrected until August 21, 1979, after the TMI-2 accident. Tr. 878-79. No technical witness could explain why this defect was not discovered when the RCDT temperature compensation error was corrected by the March 16, 1979 TCN.

Tr. 881. It is clear that the TCN should have corrected the leak rate test procedure for both the MUT and RCDT density differences (and, to be precise, the difference in density between identified leakage other than that collected in the RCDT and the density of RCS inventory at average RCS temperature).

3. RCS Temperature Was Not Correctly Entered If Temperature Exceeded 582*F.

192. The TMI-2 leak rate test procedure failed to account for changes in the RCS temperature when it exceeded

-102-l I

I 582*F. Stier Report, Vol. IV(B), Appendix A, pp. A.18-A.19.

An RCS temperature of 582*F was used in the leak rate test procedure when the RCS temperature exceeded 582*F. This defect produced an error of 2.49 gpm in the test calculation for every degree of change in the temperature above 582*F. Tr. 885; l Stier Report, Vol. IV(B), Appendix A, p. A.18. There is no explanation in the record for the failure of the test procedure to account for changes in the RCS temperature when it exceeded 582*F. Tr. 803, 882. Mr. Kirkpatrick suggested that the test was based on the TMI-l leak rate test procedure and that 582*F was "an acceptable limit [at TMI-1] because the plant hardly ever got above 582 degrees Fahrenheit." Tr. 882. But at TMI-2, the evidence indicates that the RCS temperature exceeded 582 F in approximately 50 filed leak rate tests (Tr. 803, 883-86 (49 out of 161 tests); Tr. 885-86 (54 out of 170 tests);

Stier Report, Vol. IV(B), Appendix A, p. A.18 (54 tests)).

193. The failure of the leak rate test procedure to I account for changes in RCS temperature when it exceeded 582 F frequently produced errors in the leak rate test results of up to 1 gpm. Tr. 803, 883-86; Stier Report, Vol. IV(B), Appendix A, p. A.18; .id., Vol. IV(A),Section III, Table III-2; Kirkpatrick & Wermiel Prep. St., Attachment 4, Table 1, ff.

Tr. 376. An error of approximately 1 gpm occurred if temperature changed 0.5*F. Tr. 805, 883. However, in one test, on September 22, 1978, RCS temperature changed 1.5*F, I -103-I

4 l

I l l

which caused an error in the test result of several gpm. Tr.

I 885; Stier Report, Vol. IV(B), Appendix A, p. A.19.

4. RCS Pressure Differences Were Not Accurately Taken Into Account.

194. The TMI-2 leak rate test procedure failed to consider accurately the changes in the RCS pressure. Stier Report, Vol. IV(B), Appendix A, pp. A.7, A.ll; Faegre & Benson Report Vol. One, pp. 22-23. If the RCS was maintained in steady state (with respect to pressure), there would be no impact on the leak rate test as a result of this error.

However, RCS pressure was not a variable that was within the control of the TMI-2 operators. Tr. 760-61.

195. Changes in RCS pressure often had a "significant impact on the leak rate test" Tr. 761. A change in the RCS pressure of plus or minus 50 pounds per square inch "was typical of the test", Tr. 762, and produced an error of 1.08 gpm in the test result. Stier Report, Vol. IV(B), Appendix A,

p. A-ll.

196. Messrs. Kirkpatrick and Wermiel agreed that the failure of the TMI-2 leak rate test procedure to account for the changes in the RCS pressure "had a very significant effect on the error in the leak rate calculation." Tr. 771, 786-87.

The Faegre & Benson analysis (based on an RCS pressure change of 15 pounds per square inch) and the Stier analysis (based on an RCS pressure change of 50 pounds per square inch) are in substantial agreement. Tr. 785. There is no systematic

-104-I

I analysis of the changes in the RCS pressure at TMI-2 in 1978 and 1979. However, the technical experts indicated that the typical change ranged from 20 to 50 pounds per square inch during a leak rate test. Tr. 802-03.

197. On at least one occasion, during January 4-5, 1979, a change in the RCS pressure of 60 to 65 pounds per square inch produced an error in a leak rate test result of two I gpm. Tr. 778-80, 785.

l I 5. Instrument Errors Necessarily produced Inaccuracies in Test Results.

193. The normal instrument errors associated with the TMI-2 leak rate test procedure compromised the accuracy of the measurements used by the test in its quantification of reactor coolant system leakage. As discussed above, these measurements included the temperature in the two RCS hot legs, the l temperature in two of the four cold legs, the pressurizer level, the MUT level, and the RCDT level. Stier Report, Vol.

IV(B), Appendix A, pp. A.6-7. The impact of these normal instrument errors was quantified by the technical witnesses.

M., Vol. IV(A),Section III, pp. III.8-III.10; id., Vol.

IV(B), Appendix A, pp. A.19-27. Kirkpatrick & Wermiel prep.

St., Attachment 4, Table 2, ff. Tr. 376; Faegre & Benson I Report, Vol. One, pp. 24-26. The cumulative impact of these normal instrument errors produced an error in the typical leak rate test of between 0.7 and 1.24 gpm. Tr. 894 I

-105-I

I (1 gpm)(Kirkpatrick); Tr. 895 (1.24 gpm)(Rockwell); Tr. 896 (0.7 gpm)(Stier).

199. These normal instrument errors were, according to Mr. Kirkpatrick, "the kind of instrument error that you would expect in any kind of installation like this." Tr. 897.

The effect of these errors could have been reduced by the performance of a leak rate test for more than the one-hour test interval that was always used at TMI-2. Tr. 897. Mr. Moore testified that "we've learned a lot in these last few years and that, today, people are running longer tests. . . . Tr. 936.

200. It is very surprising to the Board that there is no NRC regulatory guidance on the accurate quantification of RCS leakage in view of these normal instrument errors. Mr.

Kirkpatrick testified, for example, that Regulatory Guide 1.45 fails to discuss normal instrument errors. Tr. 934.

201. The NRC did not discover until well after the March 28, 1979 accident at TMI-2 that a leak rate test should be performed in two-hour or four-hour intervals to reduce the random error associated with the normal instrument errors discussed above. Tr. 950-51. In June 1983, the NRC published NUREG-0986, entitled "RCSLK8: Reactor Coolant System Leak Rate Determination For PWRs." In that report, the NRC Staff recommended a test interval of four hours. Tr. 950. Prior to the publication of NUREG-0986, there was no recommendation on the part of the NRC concerning the leak rate test interval.

Tr. 951. Mr. Kirkpatrick testified that the random error would

-106-I

I I be approximately 0.1 gpm if the test was performed over an eight-hour interval, 0.2 gpm over a four-hour interval, and 0.34 gpm in a two-houc interval. Tr. 953-54. Given that it was not until June 1983 that the NRC itself recognized the advantage of performing the leak rate test over these longer intervals, we cannot fault the TMI-2 operators for failing to do so.

6. RCS Oscillations Significantly Affected the Results of the Leak Rate Test.

202. Yet another defect in the TMI-2 leak rate test procedure was the impact of RCS oscillations on the calculation of the RCS leakage by the computer during the performance of a test. Stier Report, Vol. IV(A), pp. III.13-III.15; id., Vol.

IV(B), Appendix A, pp. A.34-36; Faegre & Benson Report, Vol. I,

p. 25; i_d . , Vol. II, pp.93-108; id., Vol. III(B), Exhibits 37-42; Kirkpatrick and Wermiel prep. St., pp. 17-18, ff. Tr.

376. The oscillations occurred in the RCS temperature, the pressurizer level, and the MUT level. The impact of these oscillations produced "a significant effect" on the leak rate test results. Faegre & Benson Report, Vol. One, p. 25.

203. Mr. Russell estimated that the range of the oscillation in the RCS temperature was plus or minus 0.5*F; in the pressurizer level plus or minus 2.5 inches; and in the MUT level plus or minus 1.5 inches. Mr. Rockwell estimated that the range of the oscillations in the RCS temperature was plus or minus 0.75"F; in the pressurizer level plus or minus 4

-107-I

I inches; and in the MUT level plus or minus 2.5 inches. Tr.

1092. Mr. Stier agreed with the estimates of Mr. Rockwell.

Tr. 1093.

204. Mr. Moore testified that the oscillations were the result of a problem with the integrated control system.

Tr. 1095-96. These oscillations caused similar oscillations in the RCS, thereby affecting the leak rate test and compromising its ability to quantify RCS leakage. Tr. 1097(Russell)("the oscillations caused high variability in the computed unidentified leak rate"), 1098(Rockwell)("there was a direct correlation between the oscillation in these plant parameters and the oscillation in the leak rate test results.").

205. The impact of the oscillations on the accuracy of leak rate test was significant. Mr. Kirkpatrick testified that the typical error in a leak rate test produced by the oscillations was plus or minus 1 gpm. Tr. 1101. Mr. Stier testified that the range of the error produced by these oscillations was plus or minus 0.8-1 gpm. Tr. 1101-02.

Mr. Rockwell testified that the error produced by these oscillations was up to plus or minus 1.35 gpm. Tr. 1101-02.

In an extreme case, the error produced by the oscillations could be in the range of several gpm. Tr. 1103. The impact on I the test results "of the oscillations could cause an additional error of the same magnitude as the instrument errors." Stier Report, Vol. IV(A),Section III, p. III.15; id., Vol. IV(B),

-108-I

Appendix A, p. A.36. Such oscillations are more likely to occur during the early operation of a new nuclear power plant such as TMI-2. Tr. 1120-21.

7. TMI-2, Unlike Other Babcock & Wilcox Nuclear plants, Did Not Have An Evaporative Loss Factor in its Leak Rate Test procedure.

206. The TMI-2 leak rate test procedure failed to include an " evaporative loss factor" for losses from the RCS.

Tr. 1134. Other Babcock & Wilcox power reactors were allowed to include an evaporative loss factor in their leak rate test procedures. For example, TMI-1, Rancho Seco, and Oconee all include evaporative loss factors in their leak rate test procedures. Tr. 1135. The evaporative loss factor for TMI-1 was 0.51 gpm; for Oconee-1, 0.68 gpm; for Oconee-2, 0.73 gpm; and for Oconee-3, 0.52 gpm. Faegre & Benson Report, Vol. One,

p. 19, n.*. Assuming that evaporative losses at TMI-2 were comparable to those at TMI-1, the failure of the TMI-2 leak rate test procedure to include such a factor resulted in the overstatement of unidentified leakage in the calculation of the TMI-2 unidentified leakage of as much as 0.5 gpm. M.

Mr. Stier concluded that this was a primary problem in the performance of the leak rate test at TMI-2:

I "There were several major causes of the difficulty experienced at TMI-2. First, the TMI-l leak rate calculation permitted the I subtraction of 0.5 gpm from gross leakage to account for ' evaporative losses.' At TMI-1, this helped offset the variability of test results caused by such factors as instrument error and l -109-i I

I oscillation within the reactor coolant system.

The TMI-2 procedure contained no similar provision, thereby increasing the frequency with I which calculated unidentified leakage exceeded 1.0 gpm due to inherent test variability."

Stier Report, Vol. I, p. 21.

207. The Staff witnesses contended that the use of an evaporative loss factor at TMI-2 was inappropriate. Tr. 1138.

Staff's position, as we understand it, is that such a factor is not appropriate because, as plant operations continue,

" frequently the amount of evaporative loss would be reduced so that you didn't know what it was." Tr. 1139. That may be true, but the basis of Staff's position does not apply to TMI-2 because it had not yet become a mature plant. Tr. 1121. In any event, Staff's conclusion that evaporative losses "would be reduced" as a plant matures does not support Staff's position that the evaporative loss factor should be eliminated. See Tr. 1163 (Rockwell); Exhibit 21.

208. Obviously, it would have been easier for the TMI-2 operators to obtain leak rate test results within the 1 gpm LCO for unidentified leakage if the leak rate test procedure had included an evaporative loss factor. Tr.

1137-38.

8. A Possible " Loop Seal" Effect on the MUT Level Sensor Could Have Produced Erroneou.3 Test Results.

209. The technical witnesses agreed on the existence of a defect in the MUT level sensor that produced an error in I

I -110-I

I the leak rate test values fot unidentified leakage. This defect may have been the result of a loop in the dry reference leg of that instrument, which on occasion, it is theorized, became filled with water from the MUT. The theory is that the collection of water in the loop produced a phenomenon that is commonly referred to by the technical witnesses as the " loop seal" effect. Stier Report, Vol. IV(A),Section III, pp.

III.10-III.12; i_d., Vol. IV(B), Appendix A, pp. A.27-32, id.,

Figure A-4; Kirkpatrick & Wermiel prep. St., pp. 25-29 and Attachment 5, ff. Tr. 376; Faegre & Benson Report, Vol. Two, pp. 68-69; id., Vol. Three A, Exhibits 22-31, Vol. One, pp.

23-24; --id.

I 210. The water that may have collected in the loop would have resulted from either an MUT overflow or the condensation of water vapor in the dry reference leg of the instrument. Tr. 959-60. This defect in the loop in the dry reference leg at TMI-2 was " inherently an undesirable design I arrangement." Tr. 961, 976. The loop, it appears, was required in the instrument installation instructions provided by the manufacturer. I_d . However, these instructions also required a drain valve in the loop to avoid the problem. It is unclear whether there was such a valve in the dry reference leg of the TMI-2 MUT level detection instrument. Tr. 962-63, 982.

211. The defect in the MUT level sensor was not conclusively established in the record to be the result of a I

I -111-I

1 I

l

" loop seal" effect. Tr. 960, 963-65, 972, 995-96. A defect in the MUT level sensor existed, and the existence of the " loop I seal" effect is "a reasonable hypothesis." Tr. 965(Stier).

I However, we agree with Mr. Stier that it is not critical to l l

understand whether the defect was caused by the " loop seal" effect or by something else. d .

_I_d 212. The existence of a " loop seal" in the dry reference leg of the TMI-2 MUT level sensor could have had "a very significant effect under certain circumstances."

Tr. 968(Kirkpatrick). This impact might occur if water or hydrogen was added to the MUT, or the MUT level decreased during a leak rate test. Tr. 968-69. The existence of a " loop I seal" in the MUT dry reference leg could have resulted in an overstatement in the change in the MUT level during a leak rate test. Stier Report, Vol. IV(B), Appendix A, p. 27; Tr. 975.

Mr. Stier speculated that the " loop seal" affected leak rate tests from early February to late March 1979. During that time, the effect of the overstatement in MUT level could have been as high as 30 percent. Tr. 991(Stier); see also Tr.

992-94(Kirkpatrick).

213. A decrease in the MUT level during a leak rate test would be understated to the same extent as would an increase in MUT level. Tr. 970-71. Typically, however, there was leakage during leak rate tests that resulted in a decrease in the MUT level, but water was not frequently added to the MUT I

l -112-l lI

I during a leak rate test. Tr. 998 (Kirkpatrick)("the makeup level generally decreased when significant identified leakage existed and, as we have already pointed out, that was a good deal of the time."); Tr. 999 (Stier) ("I think that even a casual examination of the makeup level strip charts indicates

{

that decline in level was almost continuous."). Thus, the

" loop seal" effect typically increased the value of the l unidentified leakage calculated by the TMI-2 leak rate test rather than decreased that value. l 214. There were only 31 water additions to the MUT during the 222 TMI-2 leak rate tests. Stier Report, Vol.

IV(A),Section VI, Table VI-1. There are even fewer hydrogen additions to the MUT during the performance of leak rate tests.

Id., Vol. IV(A),Section VII, pp. VII.7-ll. However, the " loop seal" effect "could also cause large errors in the leak rate test results, without any operator action during the tests."

Id., Vol. IV(L), Exhibit 1, p. 35; accord, Tr. 991 l

i (Kirkpatrick).

215. The " loop seal" effect was not discovered by the NRC until 1980, during the initial investigation of the allegations raised by Mr. Hartman. Tr. 983. There is evidence in the record to suggest that Met Ed should have discovered the problem prior to 1980:

I "B&W anticipated the possibility of certain problems with a dry-reference-leg system and suggested measures to minimize those problems.

In particular, B&W suggested a reference leg with 5 -113-I

I l a condensate loop and draining capabilities. The suggestion was not incorporated in the actual l

modification as identified in the vendor's instruction manual. Incorporating the vendor's suggestion probably would have avoided the physical condition which may have allowed I hydrogen additions to the make-up tank to affect level instrumentations."

Faegre & Benson Report, Vol. Two, p. 63 n.* (footnotes omitted).

216. Met Ed should have discovered and corrected this leak rate test defect well in advance of its discovery by the NRC in 1980. Tr. 1066 (Kirkpatrick) ("there were several indications of the loop seal problem which Met Ed should have investigated but did not"); Tr. 1063 (Rockwell) ("any engineer looking at the configuration would recognize the potential for a problem"). As Mr. Kirkpatrick observed, "any competent instrument design engineer should have realized that accumulation of water in this -- in a low spot would have caused instrument problems." Tr. 1079.

I 9. The Leak Rate Test Procedure Failed to Use the Correct Reactor Coolant pressure Boundary.

217. The reactor coolant pressure boundary defined by the TMI-2 leak rate test procedure failed to correspond to the regulatory definition set forth in 10 C.F.R. S 50.2(v). Tr.

752-53; Stier Report, Vol. IV(A),Section III. p. III-3; M.,

Vol. IV(B), Appendix A, p. A.3; M., Figures A-1 and A-2. For example, the TMI-2 leak rate test procedure included sections I

i -114-I

H L

j of the reactor coolant drain system and letdown / makeup /high pressure injection system that are not parts of the TMI-2 c' reactor coolant pressure boundary. Tr. 747-52; Stier Report, H

Vol. IV(B), Appendix A, Figures A-2 and A-3. Mr. Stier testified that the TMI-2 leak rate test procedure was designed to consider sections of that system that are technically not a part of the TMI-2 reactor coolant pressure boundary. Tr. 752.

As a result, the test included more leakage than it should have. See Tr. 753-54.

10. The Leak Rate Test Procedure Failed to Account Accurately for Pressurizer Level F Changes.

218. The leak rate test procedure failed to account L accurately for changes in the level of the pressurizer. Stier Report, Vol. IV(B), Appendix A, p. A.13; id., Vol. IV(A),

Section III, Table III-2. For example, any change in mass in the pressurizer steam space was not accounted for in the test calculation. This defect produced an error of plus or minus 0.28 gpm, based on a typical change in the pressurizer level of four inches. Tr. 832-33,

11. The Leak Rate Test Procedure Failed to State p the Correct Reactor Coolant Drain Tank L Volume.

- 219. The TMI-2 leak rate test procedure failed correctly to state the level / volume relationship in the RCDT and thus incorrectly stated a parameter used in the test

-115-

I I

l calculation. Stier Report, Vol. IV(B), Appendix A, pp.

t A.14-A.16, A.32-A.34; id., Vol. IV(A),Section III, Table III-2. The error in the leak rate test results that this defect produced was dependent on the level of the RCDT during I

an actual test, but there was no " typical" RCDT level during each leak rate test. Tr. 836. During tests with high l

Identified leakage, this defect produced an error of several l tenths of 1 gpm. Stier Report, Vol. IV(B), Appendix A, l

p. A.15. Typically, the RCDT level increased during any leak I rate test; tnus this defect inevitably produced an overstatement of unidentified leakage. Id., p. A.16. The NRC l

Staff acknowledged that this defect could produce an error of i

l approximately 0.4 gpm for each 1 gpm of identified leakage; such an error was "somewhat significant" and "always present."

l Tr. 837-38.

I

12. The Numerous Defects in the Leak Rate Test l

procedure Caused Test Results Indicating Negative Unidentified Leakage.

l 220. Based on the foregoing, the Board finds that I some or the defects in ene leak rate test procedure produced I

results that overstated the RCS unidentified leakage. Other defects in the leak rate test procedure produced errors that could either overstate or understate the RCS unidentified leakage. It is apparent that if the combined errors produced a f

I negative impact on the calculated value of the unidentified leakage that exceeded the actual unidentified leakage during g

8 -11e-I

the performance of a test, the test result for unidentified leakage would have been negative.

221. Normal instrument errors will affect the accuracy of leak rate tests at all nuclear power plants that use the leak rate test to quantify unidentified leakage. "When the water inventory balance is done carefully for a reactor coolant system without leakage, the resultant leakage should be zero; repeated determinations should give results statistically distributed about zero, with some negative and positive values in the distribution." Exhibit 21, p. 3 (affidavit of Mr. Bettenhausen).

222. This statistical probability, when coupled with I the numerous defects associated with the specific TMI-2 leak rate test, produced many negative leak rate test results. Many operators and supervisors testified that because a negative value for unidentified leakage was less than 1 gpm, such leak rate tests were approved and filed. E_.3., Tr. 4570-71, 4596 (Bryan). Given that one would expect such results (as confirmed by Mr. Bettenhausen), we cannot be critical of TMI-2 operators and supervisors for accepting negative leak rate tests.

13. Conclusion.

223. The technical witnesses conducted error analyses of the TMI-2 leak rate test procedure or were familiar with such analyses. Tr. 757-58. The cummulative impact of the I

I -117-

I numerous defects in the TMI-2 leak rate test procedure is summarized in charts prepared by MPR and by the Staff witnesses. Stier Report, Vol. IV(A),Section III, Table III-2; Kirkpatrick & Wermiel prep. St., Attachment 4, ff. Tr. 376.

The cumulative impact, according to Mr. Kirkpatrick was up to 7 gpm. Tr. 890. We agree.

224. We also agree with Mr. Rockwell that the

" combined net effect [of the numerous defects] would have been to produce a widely varying and probably, to the operators, an inexplicably random set of test results." Tr. 893 (quoting Faegre & Benson Report, Vol. One, p. 25).

I E. Unidentified Leakage--Actual Amounts vs. Test Results.

225. Both Faegre & Benson and Mr. Stier sought to determine whether there were periods of time when " actual" unidentified leakage may have exceeded 1 gpm. Faegre & Benson Report, Vol. One, pp. 30-36; Stier Vol. I, pp. 103-21. No similar analysis was performed by NRR. Tr. 1893.

226. We believe that the studies performed by Faegre

& Benson and Mr. Stier are of little value for this proceeding.

To begin with, unidentified leakage is, by definition, a calculated number not susceptible to actual measurement. It is also clear that " actual" unidentified leakage may be more than 1 gpm and the leak rate test may show a result less than 1 gpm.

See Tr. 945. All of the experts recognized that the inherent I

I -118-I

variations in the leak rate test procedure and calculations were such that " actual" unidentified leakage would not likely ever be exactly what the computer calculations showed. E.g.,

Tr. 1362-63(Stier). Nevertheless, the LCO in the Technical Specifications was not violated if the calculated value for unidentified leakage did not exceed 1 gpm, regardless of whether " actual" unidentified leakage may now be thought to have been greater than 1 gpm, because the operators were instructed to rely on leak rate tests as one (but not the only) measurement of unidentified leakage. (The converse is not true; if the test result depicted unidentified leakage over 1 gpm, the LCO was not necessarily violated, because a satisfactory test result was not required by the Technical Specifications and other means of detection could be used to satisfy the LCO.)

227. Faegre & Benson conducted an analysis of unidentified leakage based upon a five-day rolling average of sump pump starts. They identified three periods when unidentified leakage may have exceeded 1 gpm: December 16-20, 1978; January 6-15, 1979; and March 27-28, 1979. Faegre &

Benson Report, Vol. One, p. 30; see Tr. 1893. From December 16 through December 20, 1978, the reactor was shut down, so the amount of unidentified leakage is of no consequence for purposes of the Technical Specifications. The second and third periods identified by Faegre & Benson correspond with two I

I ~119-I

L periods identified by Mr. Stier, discussed below. However, Faegre & Benson ultimately concluded that the rate of unidentified leakage never in fact exceeded 1 gpm at any time.

Faegre & Benson Report, Vol. One, pp. 31, 36.

I L 228. Mr. Stier focused on three periods when unidentified leakage may have exceeded 1 gpm: October 16-18, 1978; January 2-15, 1979; and March 16-28, 1979. Stier Report, Vol. I, pp. 109-21. We shall examine each of the three periods in turn.

I 229. During October 16-18, five tests produced results showing unidentified leakage exceeding 1 gpm. M., pp.

109-10. The operators reacted to those tests by attempting to locate and quantify actual leaks so that they could be subtracted from the calculated amount of unidentified leakage.

Id., pp. 110-11. Although the operators were able to locate leaks, the inclusion of additional identified leakage did not result in a reduction of unidentified leakage below 1 gpm.

This was because of a defect in the leak rate calculation, of which the operators were unaware. d Id., pp. 111-12.

( 230. Even though the leak rate tests performed on October 16-18 showed unidentified leakage in excess of 1 gpm, r

MpR's estimate of unidentified leakage exceeded 1 gpm on only one test, on October 16. M., p. 110. With respect to the remaining tests, the leakage that was identified should have

( been adequate to reduce the calculated value to less than 1

-120-

I gpm. Id., p. 111. Accordingly, it would appear that " actual" identified leakage exceeded 1 gpm for less than 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. More importantly, the evidence shows that the operators responded with a diligent effort to locate and identify leaks in order to re'uce d the calculated value. I_ d . , p. 112. When they did not succeed, the matter was treated as a reportable occurrence, leading to the filing of LER 78-62/lT. I_ d . , pp. 113-14; Vol.

V(C), Tab 29. At no time were indications of unidentified leakage in excess of 1 gpm simply ignored. Id., Vol. I,

p. 112.

231. From January 4 through January 15, 1979, Mr.

Stier estimated that unidentified leakage was consistently in I excess of 1 gpm (but never as much as 2 gpm). Id. p. 115.

However, leak rate tests showing unidentified leakage of less than 1 gpm were regularly obtained and filed by the operators during the same period. Id. The record indicates at least one satisfactory test result was filed on each day except for January 3, January 9, and January 12. I_d . Moreover, several tests were run with the stable MUT level transmitter (LT-2).

Exhibit 5-A, Attachment 5, Table 11.

232. Mr. Stier testified that a difference of less than 1 gpm between the reported leak rate value for unidentified leakage and MPR's recalculation of unidentified leakage could be accounted for by instrument error and normal oscillations and was considered not to be significant. E.g.,

I I -121-I

I Tr. 1633. Nevertheless, Mr. Stier's table comparing estimated I

unidentified leakage and unidentified leakage shown on the leak rate tests for January 2-15 contains six tests in which the difference was less than 1 gpm. Stier Report, Vol. I, p. 115.

We find that this seriously undermines the probative value of the table. Because we agree with MPR that differences of less than 1 gpm between reported and re-calculated volume for unidentified leakage could be attributed to instrument error and normal oscillations and therefore are not significant, the Board is unable to conclude that " actual" identified leakage exceeded 1 gpm betweem January 4-15, 1979. Likewise, we are unable to conclude that valid test results less than 1 gpm were not obtained and filed during that period.

233. On the other hand, there is ample evidence to support the conclusion that the operators were aware of actual leakage problems in the plant and were taking steps to address those problems. The details are contained in the Stier Report.

M., pp. 116-19. In summary, the source of the leakage was identified and a safety assessment was made to support continued operation. This appears to the Board to have been a prudent and conservative approach. Given the fact that it is not clear that unidentified leakage really exceeded 1 gpm and l that calculated unidentified leakage was within the LCO, Mr.

1 Stier's criticism that plant personnel "took no action to shut down the plant" (id., p. 117) is not warranted by the record.

I

-122-

I l l

234. Mr. Stier's analysis of the March 16-28, 1979 period is even less persuasive. Mr. Stier reports that MPR's estimate of actual unidentified leakage between March 16 and 28 showed it to be "very close to 1.0 gpm.". M., p. 120. In fact, MPR found that unidentified leakage did not exceed 1 gpm during that period. M., Vol. IV(A),Section IV, p. IV.8.

235. Mr. Stier argues that the calculated results would have exceeded 1 gpm during March 16-28, 1979, because of inaccuracy in the MUT level transmitter, except for the fact that the operators typically added water to the MUT while tests were being performed. M., Vol. I, p. 120-21. However, as we find in 1 293, 301, and 303, infra, water was frequently and I consistently added to the MUT during this period for valid operational reasons. Thus, the operators were not ignoring unidentified leakage in excess of 1 gpm. In any event, Mr.

Stier's analysis is faulty, because he began from the premise that March 1979 was a period during which actual identified leakage was in excess of 1 gpm, and shifted to the conclusion that calculated unidentified leakage would have exceeded 1 gpm but for inaccuracies in the test. We fail to see how Mr.

Stier's conclusion leads to justifiable criticism of the operators.

236. The only conclusion that can be drawn from Mr.

Stier's analysis of leak rate results for March 16-28, 1979, is that the operators were still following the "72-hour" rule and I

I -123-I

L not entering the Action Statement. But in this respects March L

1979 is no different from previous periods, so Mr. Stier's F analysis adds nothing to our overall review of the facts.

L F. Technical Analysis of Leak Rate Test Falsification.

1. Means by Which Falsification May Have Occurred.

{

237. It has been alleged that leak rate tests may have been falsified by several different means. The allegations of such falsification include unrecorded water additions to the MUT, underrecorded water additions to the MUT, hydrogen additions to the MUT, feed and bleed operations, and MUT level transmitter switching. E.g., Exhibit 5-B, p. 4.

Each of these allegations is discussed below,

a. Unrecorded Water Additions to the MUT.

238. MPR and the NRC devoted an enormous effort to determining whether there were unrecorded water additions to

{

manipulate leak rate tests. See Exhibits 5-A and 5-B; Stier Report, Vol. IV(B), Appendix H; Exhibit 1-B (Green Volume).

Appendix H to Volume IV(B) of the Stier Report sets forth the findings of Faegre & Benson, the NRC's Dr. Chung, and NRC documents provided by the Justice Department to Met Ed in 1984

{

during the criminal proceeding against Met Ed. In one or another of those three investigations discussed by MPR in Appendix H, the investigators concluded that unrecorded water

[ additions were made during leak rate tests. See Stier Report,

-124-m .

I I Vol. IV(A),Section VI, pp. VI.17-23; id., Vol. IV(B),

Appendix H, Table H-1. In only six cases analyzed by NRR does MPR conclude that there is substance to the allegations. I_d .

We first turn to a discussion of those specific Tests and draw conclusions about them, before discussing those Tests that MPR does not conclude were falsified or manipulated.

239. For convenience, we set out the six suspect tests in tabular form:

Stier CRO Test No. Date Pe r f o nned By on Log Allegations of MPR I 30 2-23-79 Hartman Booher Confirmed operator action; may have been feed and bleed operation, not water addition; CRO log shows water additions.

43 2-12-79 Kidwell Illjes Estimated addition of 150 gallons I during test; RCDT level decreased during test.

l 44 2-11-79 Mell Kidwell 300 gallon addition in CRO log; 200 gallons on chart; I pen working poorly; RCDT level decreased during test.

64 1-13-79 Hartman Booher Confirmed water addition; 117 gallons entered in CRO log.

l l

87 12-22-78 Hemmila Kidwell Confirmed water addition; probably part of feed and l

bleed operation.

1 I

I -125-I

L f

L 150 10-15-78 Germer Frederick Confirmed water

" addition; probably part of feed and bleed operation.

r L 240. First, only a few TMI-2 operators were involved 7

in the above Tests. Second, there appears to be no pattern to L

them; only one operator, Mr. Hartman, performed more than one of the suspect tests. Third, three of the Tests involved " feed

{

and bleed" operations, which NRR and MPR/Stier agree were not associated with test falsification. Tr. 1316, 1319. Fourth, two of the Tests appear to involve RCDT level decreases, indicating possibly unstable plant conditions or operater-caused changes that should have invalidated the test.

Fifth, Stier Test No. 44 involves an alleged addition during a time when the MUT level indicator pen was working poorly, making it difficult to determine what occurred; in any event, the Test was concluded at 1908 (see Table H-1 and CRO log entry at that time), whereas the CRO log shows an addition of water at 1910. Either the investigators' conclusion about the timing of the addition may have been wrong (see HT 282-86, infra, concerning the dispute about chart vs. "real" time; and see MUT strip chart showing addition outside test interval) or

{

Mr. Kidwell may have entered the water addition at the wrong time on the CRO log. Tr. 3317 (when asked about his logkeeping

( practices, Mr. Kidwell colorfully explained his sloppiness by stating, "One thing that you can smack me for is

{ recordkeeping.").

[

-126-

F L

241. In short, with the possible exception of Mr. Hartman, who has admitted occasionally adding water without entering it in the computer calculation, these six Tests do not appear to depict anything other than occasional sloppiness, by

[ a few operators, during a small number of leak rate Tests.

Because the Test performer may not have known about the possible water additions or " feed and bleed" operations during p the Tests, or because the CRO on the log (and, probably, the L

panel) may not have known that a Test was being run, we cannot

[ conclude that wilful action to manipulate leak rate test

_ results occurred during these Tests. Tr. 3305-11; see Tr. 1077 (Russell) (no requirement that test performer review MUT level strip chart before submitting leak rate test for approval).

242. As to the other 55 Tests listed in Stier Report,

( Vol. IV(B), Appendix H, Table H-1 that either Faegre & Benson's investigation or the NRC's early investigations concluded may F

L also have involved test manipulation due to unrecorded water additions, MPR has, in our judgment, conclusively disproved those allegations. During Stier Test No. 7, for example, the

( " unrecorded" water addition took place outside the Test period, and Faegre & Benson agreed (see Table H-l). A similar conclusion applies to Stier Test Nos. 15, 28, 47, 76, 79, 80, 84, 130 (also plant not in steady state), 139, 152, 156, 157, and 161. See Table H-1.

[

[

-127-

I 243. We believe that MPR's careful analysis is generally superior to those of the other three investigations, which were performed at an earlier time and, apparently, without the amount of time, detailed records, and the obvious benefit of other investigators' views that were available to MPR. The Faegre & Benson investigation, for example, was performed in 35 days. Tr. 996, 1145-46. Moreover, MPR's analysis was more objective than NRR's analysis in the sense that it relied exclusively on a technical analysis of documentary evidence. Then too, NRR did not know of, and thus did not take into consideration, such crucial documentation as evidence that leaks were repaired (Tr. 1554), thus accounting for changes in the slope of the MUT level strip chart trace.

Tr. 1550-51. Due to the significant passage of time and the consequent, almost universal, inability of the operators to recall specific tests, we regard MPR's objective approach as superior. See Tr. 2034 (Board was "most impressed" with Stier/MPR analysis, finding it "one of best expert testimony presentations" we have ever seen).

244. Because of inadequacies in the NRC or Faegre &

Benson analyses, others of the 55 Tests are not, in our judgment, instances in which leak rate tests were falsified, for the following reasons (each reason is supported in MPR's conclusions in Table H-1 and MPR's summary sheet associated with each Test in Stier Report, Vol. IV, Appendix K):

I -128-1 I

I 1

I (1) MUT instrumentation problems (Stier Test Nos. 11 (also inadvertent error in inputting data), 19, 25, and 85): neither Faegre and Benson nor Faegre & Benson nor NRR nor Mr. Stier has alleged that the operators knew of the problem (see Tr. 971-73);

(2) Use of malfunctioning MUT level transmitter (Stier Test Nos. 66, 67, 71, 72, 78, 81, 85, 88 I (also lack of steady state) 99, 100, 102, 108 and 115): NRR and Mr. Stier concluded that the operator- caused actions (or lack of action) were examples of inadvertence or negligence, not manipulation (Tr. 1272-74, 1626-29, 1756);

(3) Lack of steady state (Stier Test Nos. 52, 88, I 118, 129, 130, 139 and 152): tests need not have been run during those periods, according to Technical Specification 3.4.6.2, and neither NRR I nor Mr. Stier has alleged that that method was used to falsify tests.(Tr. 1084);

(4) Absence of indication from strip chart level indication that water was added (other than instance of water addition outside test period,)

(Stier Test Nos. 13, 14, 15 (also allegation of water addition during Test, see 1 242, supra),

22, 23, 31, 33, 34 35 (possible hydrogen addition), 37, 50, 61 (may have been " feed and

( bleed"; CRO log entry in error), 63, 70, 71 (bad MUT level transmitter), 72 (same as 71), 78 (same as 71), 79 (chart time discrepancy and NRC reliance on data from bad MUT level transmitter),

80 (same as 79), 81 (same as 71 and change in i

pressurizer level), 85 (same as 71), 96, 99 (same as 71) 102 (same as 71), 108 (same as 71), 109, 112, 1 d (lack of steady state), 129 (same as I 118), 133, 143, 144, and 145): the MPR analysis l is clearly superior, because it does not just rely on the MUT level strip chart, but also on I all related conditions in the reactor and plant to determine the cause(s) of trace deflections on the MUT level strip chart; i

(5) Possible hydrogen addition, rather than water addition (Stier Test No. 10): the allegation of a water addition is clearly incorrect based on MPR's analysis; and I

I -129-I

I (6) Confusion with Stier Test No. 45 (Stier Test No.

44): Faegre & Benson clearly confused that Test with Stier Test No. 44.

245. The conclusion is inescapable that the original NRC and Faegre & Benson allegations were, with respect to many of the Tests discussed in Volume IV(B) of the Stier Report, Appendix H, incorrect.

246. NRR subsequently has concluded that the remaining Tests listed in Table H-1 are not " questionable",

even though those Tests were earlier considered suspect. Those Tests which (according to Exhibit 5-B, Attachment 3) remain in dispute are Stier Test Nos. 13, 14, 15, 22, 23, 28, 37, 46, 47, 48, 50, 60, 61, 62, 80, 96, 108, 133, 143, 155, and 156.

247. In Exhibit 1-B (Green Volume), MPR reevaluated all of the Tests still in dispute between them and NRR.

Exhibit 1-B indicates that there is no significant disagreement between MPR and NRR on the six Tests first set out above and the Board accepts the technical analysis of those Tests as determined by MPR (with the exception of Stier Test No. 44, as to which the CRO log indicates that the water addition in question occurred two minutes after the Test, and Mr. Kidwell explained that his timekeeping practices in the CRO log were sloppy. Tr. 3317.) As to the remaining Tests, the Board accepts MPR's analysis that no falsification through water additions occurred. Stier Report, Vol. I, pp. 83-84.

l 248. The Staff's comparison of the NRR and MPR analyses in Exhibit 5-B explains why MPR disagrees with the NRR

-130-l l

I analyses of Stier Test Nos. 13, 14, 15, 22, 23, 28, 37, 46, 47, 48, 50, 60, 61, 62, 80, 96, 108, 133, 143, 155, and 156. The reasons include:

(a) normal MUT level variation. Stier Test Nos. 13, 14, 22, 23, 28, 37, 46, 48, 50, 60, 80, 108, 133, 143; (b) water addition outside test period.

I Stier Test Nos. 15, 47; (c) no water addition evident from MUT level strip chart. Stier Test Nos. 61, 80, 96; and (d) chart time uncertainty; Stier Test Nos. 62, 155, 156.

I The Board has carefully reviewed MPR's analysis and compared it with NRR's analysis, and concludes that MPR's analysis is correct. Particularly with respect to alleged " jogged" water additions, technical analysis that relies on the MUT level strip chart is incapable of adequately determining whether such an addition was made. Stier Report, Vol. IV(A),Section VI, pp. VI.13-15 (" Consequently, it appears that the ability of the makeup tank level trace to resolve so-called gradual or jogged additions of the order of 50 to 100 gallons is quite limited since the resulting deflections would be essentially the same as the normal trace fluctuations."); see Tr. 1590-91(Capra).

The MUT acted as a " surge tank" in the TMI-2 RCS system because the reactor was ordinarily full and pressurizer level was I

I -131-I

I controlled (Tr. 1540) (Harrison), thus producing level changes in the MUT due to many different plant changes or oscillations.

In any event, even if it were a close question whether MPR or NRR is correct, we have resolved the difference in favor of a conclusion that no improprieties occurred during those Tests, I because of our view that conflicting evidence whether in the circumstances improprieties occurred should be resolved in favor of the involved individuals. See T 130, supra.

b. "Underrecorded" Water Additions to the MUT.

249. MpR explained the phenomenon by which an addition of water to the MUT would result in a reduction in the I rate of unidentified leakage that otherwise would have been calculated by the computer, even though it was entered on a l gallon-for-gallon basis in the computer (so-called "underrecorded" water additions):

"As discussed in Section III in conjunction l

=

with inaccuracies in the TMI-2 RCIB test procedure, the RCIB test procedure did not convert the water additions to the same I

I temperature as the reactor coolant system. Since the density of water at the makeup tank temperature (approximately 100 F) was I approximately 62 lb/ft' and the density of water at the reactor coolant system average temperature (approximately 581*F) was about 44 lb/ft', a gallon of water added to the I makeup tank was equivalent to about 1.4 gallons at the reactor coolant system temperature. As a result of this error in the procedure, the water g added to the makeup tank was " undervalued" and 5 tended to reduce the total and unidentified leak rate as calculated in the test. This was the principal error associated with water additions I

I -132-I

• I which were properly included in the test calculations. Because of this error, the I addition of water to the makeup tank in the course of an RCIB test would always reduce the unidentified leakage calculated from the test."

{

l Stier Report, Vol. IV(A),Section VI, pp. VI.3-4. (It is important to understand that "underrecorded" does not mean that i

operators entered a lesser volume of water than they added, but rather that the volume they added and entered was less than the volume the computer "saw" as having been added. Tr. 3969-70 (Kelley, J.).) As MPR went on to explain, however, such "underrecorded" additions often made the test results more accurate:

"However, from the standpoint of the accuracy of the calculated unidentified leakage, i.e., how well the RCIB unidentified leakage represented the actual value, the addition of water may have I often made the test result more and not less accurate. This is discussed below. Until Mid-March 1979 there was a similar error in the I RCIB procedure with regard to the temperature of liquid collected in the reactor coolant drain tank (RCDT). The liquid collected in the RCDT was measured at a low temperature so that one I gallon collected was actually equivalent to about 1.4 gallons at the reactor coolant system average temperature. Prior to the correction made in i

l I March 1979, to properly account for the RCDT collection, the error in the RCDT change would lower the identified leakage and, therefore, l increase the difference between the total leakage and the identified leakage, that is to say, the calculated ' unidentified' leakage. However, when I water was added to the makeup tank in the course I of the test and included in the calculation, the temperature error for the water addition would tend to compensate for the temperature error in I the RCDT collection and, if the amount of water added equaled the amount collected in the RCDT, the net error would be essentially zero. Only if I the water added exceeded the amount collected I -133-I

I would there have been a net error to decrease the unidentified leak rate."

Stier Report, Vol. IV(A),Section VI, p. VI.5 (emphases in original).

250. Therefore, the calculated value of the rate of unidentified leakage was so inaccurate as to be essentially meaningless, due to this temperature error alone, unless the amount "added equaled the amount collected in the RCDT [in which case] the net error would be essentially zero." Id.;

Tr. 1958 (Stier). Thus, water additions to the MUT prior to March 16, 1979 (the date that the RCDT temperature problem was corrected) "would tend to compensate for the temperature error in the RCDT collection", and thus make the calculated rate of unidentified leakage more accurate.

251. MPR explained that, in nearly all cases, the so-called "underrecorded" water additions in fact made the calculated rate of unidentified leakage more accurate. MPR stated:

"There were 15 cases where water was added in the course of a test and the addition was included in the RCIB calculation (see Table VI-1) prior to March 16, 1979, when the procedure was changed to correct the error in the RCDT I collection. In 11 of these 15 tests the final makeup tank level was either below or less than an inch above the initial level. Although the I calculated unidentified leakage was always reduced by a water addition, it appears that in only four tests (011, 036, 040, and 065) were the I calculated values of unidentified leakage made unrealistically low by the addition of water."

I

-134-I -

I Stier Report, Vol. IV(A),Section VI, pp. VI.6-7. Accordingly, the Tests in which water was added, with the exception of those same four Tests (Stier Test Nos. Oll, 036, 040, and 065),

cannot be the basis for any finding of " falsification", because the water addition made the Test results more accurata. To conclude otherwise would be a contradiction in terms. See United States v. Oreida, 525 F.2d 881 (5th Cir.

I 1976)(impossibility in fact precludes conviction).

252. The essential information concerning the only four Tests run prior to March 16, 1979, in which water additions during the Test were greater than the leakage to the RCDT is set out below:

Stier CRO Test No. Date Performed By on Log Analysis by MPR 11 3-14-79 Illjes Kidwell Possible " feed and bleed"; also, opera-I tor error; additions improperly recorded and entered into computer.

36 2-16-79 Olson Coleman Additions properly recorded and entered into computer.

40 2-14-79 Faust Frederick Addition properly I entered into computer.

65 1-11-79 Coleman Olson Addition properly I recorded and entered into computer.

253. With respect to three of these four Tests (Nos.

36, 40, and 65), the Board finds that the amounts of water entered into the calculation appear to have been the precise 1

I -135-I

I amount actually added. Stier Report, Vols. IV(D) & (E), MPR summary sheets. With respect to the fourth Test, the amount I was also correct, but was inadvertently entered as a loss from the RCDT. See Stier Report, Vol. IV(C), Test 11; Tr. 3094-95.

Also, the evidence shows that no operator performed more than one such Test, and therefore no pattern of conduct can be discerned from these four Tests. As we find in TT 272-79, I infra, control room staffing at TMI-2 was sufficiently loose, and information at times was not adequately conveyed between or among operators such that we are unable to find that any one of the operators involved in these four Tests necessarily knew, during the Test, that the water addition was being made, or that, at the time of the addition, the Test was being run.

See, e.g., Tr. 3306-11; Cooper Prep. St., pp. 8-9, ff. Tr.

2835. Given the large number of tests performed at TMI-2 (Stier Report, Vol. I, pp. 54-55, 60) and given the increasing frequency of water additions necessary in January-March 1979 (Stier Report, Vol. IV(A),Section V, Table V-12), it is entirely possible that the additions during these Tests were mere coincidence. Just because the m unt of water added exceeded the amount of leakage AU ny :he Test is not conclusive evidence of falsification, particularly without evidence of knowledge by the operators of the lack of a temperature carrection. Stier Test Nos. 11, 36, and 40 were run during a period of high identified leakage id.,Section IV, I -136-I

I Fig. IV-14) and thus the water additions may have been necessary to replenish RCS inventory or for adjustment of boron concentration in the RCS. E.g., id., Vol. IV(D), Tab 40, CRO log p. 501 (note entries at 0720, 0755, and 0942 of "demin" water). (The tests performed on "D" shift are discussed at greater length in our individual findings infra.)

254. Finally, most tests between March 16-28, 1979 were run during a period in which water additions were required frequently. Stier Report, Vol. IV(A), Table VI-1. As to those tests, Mr. Stier concluded:

". . . we cannot find with certainty that from mid-March through March 28 the water additions were made with the intention to influence the I tests. There are two factors that militate against a finding of intentional conduct. First, there is no direct evidence implicating any of I the members of the two crews that performed all of the filed tests in this form of manipulation.

Knowledge of the effect of water additions on the leak rate test does not appear to have circulated I as widely as information about the effects of hydrogen. For example, Hartman stated that he was unaware that a water addition that was I accounted for in the calculation could affect a leak rate test [ citation omitted].

I "Second, the pattern of water additions between mid-March and March 28 differed from the previous period. Water was not consistently added within the last few minutes of each test as had been the case from mid-February through mid-March. In addition, the reactor coolant drain tank collection rate became so high by mid-March that water additions were required at I. short intervals to compensate for the loss to the reactor coolant system [ citation omitted}.

Operators may have added water to the makeup tank during tests out of necessity. Therefore, we cannot be certain that the water additions made between mid-March and March 28 were for the purpose of manipulating tests."

-137-I

I I_d., Vol. I, pp. 101-02. Mr. Stier also found, and we agree, that "MPR's analysis of sump collection shows that from approximately March 16 to March 28, actual unidentified leakage approached [i.e., did not exceed] 1.0 gpm." Md . , p. 101 (emphasis added). MPR concluded that actual unidentified leakage was below 1 gpm throughout the period March 16-28, 1979 until the accident. Id., Vol. IV(A),Section IV, p. IV.8.

c. MUT Hydrogen Additions.

255. In the normal operation of TMI-2, it was necessary to add hydrogen gas to the RCS by adding it to the MUT to scavenge free oxygen and to maintain net positive suction head in the makeup pumps. Faegre & Benson Report, Vol.

One, p. 42; Vol. Two, Chapter IV, p. 60; Kirkpatrick & Wermiel Prep. St., p. 25, ff. Tr. 376; Tr. 3921. These additions were made periodically to maintain hydrogen gas overpressure in the MUT of approximately 15-25 psi (the so-called " bands"). Tr.

2880-82, 2927, 3200-01, 3422-23; see Exhibit 23 (FSAR analysis of necessary hydrogen concentration in RCS)..

256. There was no prohibition against adding hydrogen to the MUT during a leak rate test. Stier Report, Vol. I, p.

88; id. Vol. V(C), Tab 19; Tr. 2859. Although Surveillance Procedure 2301-3D1 prohibited " chemical additions" during leak rate tests, the investigators agreed that the prohibition against " chemical additions" did not apply to hydrogen additions. Id., Vol. I, p. 88; Tr. 627-29, 1481. Therefore, I

I -138-I

I the addition of hydrogen to the MUT did not violate the leak rate Technical Specifications, the TMI-2 facility license, CRO I licenses, or SRO licenses.

257. Faegre & Benson identified 10 leak rate Tests in which hydrogen additions to the MUT potentially affected the results of leak rate tests. See id., Vol. IV(A),Section VII, pp. VII.7-8; Faegre & Benson Report, Vol. Two, p. 70. Early NRC investigations alleged four additional Tests in which hydrogen was added to the MUT. Stier Report, Vol. IV(A),

Section VII, p. VII 8.

258. MPR's analysis concluced that hydrogen was added to the MUT during the ten tests identified by Faegre & Benson but that the addition of hydrogen affected the outcome of the Tests in only two instances. M. at pp. VII.8-10. None of the four additional Tests identified by the NRC investigators as allegedly involving the addition of hydrogen to the MUT during the Test could be confirmed by MPR. M. at pp. VII.10-ll.

Neither the logs nor the MUT level traces demonstrate that hydrogen was added during the test, and all four were conducted during a time when there was a problem with the MUT level transmitters. I_d . The Board accepts the MPR analysis as the best evidence of record, subject to the qualifications stated below.

259. MPR concluded that there were only five instances in which a hydrogen addition had an effect on MUT I

I -139-

I level. M., Vol. IV(A),Section VII, Table VII-1. Two of those instances occurred during Stier Test Nos. 35 and 38. M.

In two other cases, the addition of hydrogen possibly affected the Test results (see M., Stier Test Nos. 6 and 95), but MpR's uncertainty causes us to find that the evidence does not l

lm l

sufficiently demonstrate falsification of those latter two l Tests.

260. With respect to Stier Test No. 38, performed on February 15, 1979, the MUT level strip chart includes the notation " pressurized MUT." Exhibit 18 (clear copy). It appears from the CRO log entry at "2100" and from the phrase

" pressurized MUT" that the notation refers to a hydrogen addition. Tr. 2862 (Cooper). The notation on the strip chart (which is very unusual -- there are almost no notations on the MUT level strip chart) strongly suggests that this was nothing more than an experiment. (We have examined the MUT level strip chart and have confirmed how unusual the notation is.) While it would perhaps have been better to have written something on the Test result to indicate that it was an experiment, we cannot find that a Test performed merely as an experiment is evidence of falsification, particularly when there are no other I tests for that same shift for which there is any clear evidence of falsification by hydrogen additions.

261. With respect to Stier Test 93. 35, the addition of hydrogen was not logged in the CRO log, but in the AO log, I

-140-I

I because this was at a time when hydrogen could not be added from the TMI-2 Control Room. Tr. 3095; Stier Report, Vol.

IV(A),Section VII, pp. VII.9-10; see MPR analysis and plant records accompanying Stier Test No. 35. The addition was logged in at AO' log at 0510, and the Test concluded at 0511, leaving substantial uncertainty whether the addition actually occurred during the Test. See Tr. 1077; Faegre & Benson Report, Vol. Two, Chapter Four, p. 70 n.**** According to I

MPR's analysis, the difference between chart time and real time was 1.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> (see summary sheet for Stier Test No. 35) and, therefore, even a slight error in the AO log, the CRO log, or the comparative analysis of chart and real time would put the addition outside the Test period.

262. With respect to Mr. Chwastyk's hydrogen experiment during Stier Test No. 34 (Tr. 3411, 3419-20, 3424-25, 3442), the Shift Foreman's log shows that the Test was crossed out (apparently indicating that someone thought the Test should have been invalidated). See documents accompanying Stier Test No. 34 in Stier Vol. IV(D). Even though the Test was filed, another Test (Stier Test No. 33) was run one-half hour after Stier Test No. 34, further suggesting that Stier Test No. 34 was not considered to be valid or to be relied on for compliance with the Technical Specification LCO for unidentified leakage. MPR concluded that there was not I

I I -"'-

I sufficient evidence that Stier Test No. 33 was falsified or invalid (see MPR Summary Sheet accompanying Stier Report, Vol.

IV(D), Tab 33) and, therefore, there is little, if any, significance to Stier Test No. 34, even if hydrogen was added to the f4UT during it.

263. The Board accepts MPR's conclusion that a hydrogen addition to the MUT could not be confirmed during Stier Test No. 33 (id.) rather than Mr. Stier's conclusion to the contrary (Stier Vol. I, pp. 95-96) because MPR, not Mr. Stier, were Mr. Stier's technical experts, and the cited basis for Mr. Stier's conclusion (footnote 160, citing MPR's documents associated with Test Nos. 33 and 34) does not support his conclusion. MPR's conclusion about the allegation was that an addition of hydrogen was "not confirmed" because the " trace deflection appears typical of others during this day" (MPR summary sheet for Stier Test No. 33), thus undercutting the sole basis for Mr. Stier's statement. The evidence is sufficiently confused that no adverse conclusion can be drawn against "B" shift for Stier Test No. 34. Mr. Chwastyk's instruction not to add hydrogen during leak rate tests (id.,

Vol. I, pp. 96-97; Tr. 3537) negates any adverse finding against him for that Test, and a review of leak rate tests after Stier Test No. 34 until March 28, 1979 shows that there is no evidence that "B" shift added hydrogen to the MUT during I

I

-142-I

I any leak rate test after that date. See Stier Report, Vol. I,

p. 97; M. Vol. III(A), Table 1.

264. Mr. Stier's " overview" of his entire Report inexplicably states that:

"Between February 11 and March 28, 1979, 45 tests were filed. Thirty-four of those tests were I influenced by either hydrogen or water additions to the makeup tank. One of those tests was influenced by both. The following chart lists those tests and indicates whether they were '

influenced by water or hydrogen additions."

M., Vol. I, pp. 90-91. Mr. Stier lists nine Tests allegedly influenced by hydrogen additions to the MUT: Stier Test Nos. 39, 38, 35, 34, 33, 32, 24, 10, and 6 (which allegedly also Was influenced by a water addition). However, the five Tests other than Nos. 38, 35, 34, and 33 just discussed were not found by MPR to have been influenced by hydrogen additions to the MUT, or MPR found that the additions had little effect.

See id., Vol. IV(A),Section VII, Table VII-1; Stier Report, Vols. IV(C) and (D) for those Tests. The MPR summary sheets, containing MPR's conclusions, are inconclusive about those Tests, as is Table VII-1.

265. We find that, although Mr. Stier may not be wrong, the evidence is not sufficiently clear to draw the conclusion that those Tests were " influenced" by those hydrogen additions (even assuming the additions occurred, as to which there is some doubt). Our conclusion that Mr. Stier overstated the weight of the evidence concerning those five Tests is I

-143-

I supported by the fact that they were performed by several different shifts and several different operators, negating any evidence of a pattern of falsification.

266. In summary, the three Tests in which a hydrogen addition to the MUT (if it in fact occurred during those Tests) may have influenced the value depicted for unidentified leakage appear to have been experiments (in the case of Test Nos. 34 and 38) or performed without knowledge of the addition (in the case of Test No. 35). It is uncertain who performed the experiment associated with Stier Test No. 38, and I Mr. Chwastyk's experiment (in Stier Test No. 34) does not.

appear to have significantly affected the Test result or to 4

(

I have been repeated by his shift. His instruction to his shift not to add hydrogen during leak rate tests and his immediate reporting of the phenomenon to the I & C department

(

(Tr. 3429-31) demonstrate his lack of intent to falsify leak rate tests. The evidence, therefore, does not support the allegation that hydrogen additions to the MUT during leak rate tests were used to falsify test results. In any event, MPR's technical analysis supports the finding that hydrogen additions, like water additions, oftentimes made leak rate test results more accurate during the period of high identified leakage during February-March 1979. Stier Report, Vol. IV(A),

Section VII; see Tr. 1958(Stier).

I I

-144-

I

d. " Feed and Bleed" Operations.

267. " Feed and bleed" operations was the term used to described evolutions to effectuate changes in boron concentration in the RCS. Stier Report, Vol. I, p. 83.

Mr. Stier explained:

"In addition [to those tests in which unaccounted-for water additions may have occurred during a leak rate test], MPR noted ten possible feed and bleed operations during tests, that is,

- water additions to and removals from the makeup tank performed to alter the boron concentration in the reactor coolant system. Although it is I not alleged by Hartman that feed and bleed operations were used to manipulate tests, it is possible that they could be used for such a purpose."

Id. Nevertheless, Mr. Stier did not conclude that leak rate test were manipulated by " feed and bleed" operations, and neither did Messrs. Russell and Capra. Tr. 1316, 1319.

268. MPR came to the same conclusion, stating:

"No firm conclusions can be drawn with respect to I whether feed and bleed operations or so-called

' jogged

• water additions were used by the operators to affect the RCIB test results. The normal fluctuations of the makeup tank level strip chart record coupled with the inherent variability of such operations do not allow their unambiguous identification."

Stier Report, Vol. IV(A),Section II, p. II-10. The Board has found no evidence from which it could conclude that " feed and bleed" operations were used to falsify test results and, accordingly, agrees with the same conclusion of all of the investigators.

I

-145-

I 269. It should be noted that NRR used the term

" questionable" to apply to all tests which Messrs. Russell and Capra concluded may have been run in violation of a requirement or which depicted unidentified leakage because of error (whether by omission or commission) on the part of the operators, even if unintentional or unknown to them.

Tr. 1472-75. The Board does not agree with this approach. See Tr. 1475, 1767-68, 4971-72. As already stated, we have found no evidence to dispute the investigators' conclusions that

" feed and bleed" operations were not used to falsify leak rate tests, and, accordingly, we make no adverse finding about such tests.

e. MUT Level Transmitter Switching.

270. The evidence shows that there were two periods of time, from October 1978 through January 1979, when one MUT level transmitter was malfunctioning. Stier Report, Vol. I.

pp. 72-82. Mr. Hartman alleged that leak rate tests could be manipulated by MUT level transmitter switching but that he was "not aware personally and don't remember if" leak rate tests were falsified by that method. Id. pp. 85-86. Mr. Stier investigated this allegation but:

"found insufficient evidence to conclude I with certainty that the faulty level transmitter was systematically switched to the computer with the intent to affect the test results."

Id., Vol. IV(A),Section II, p. II-10. In fact, MpR's detailed analysis shows that, with one apparent and two possible

-146-

I exceptions, the CROs did not switch the faulty transmitter to the computer to conduct a leak rate test. Id.,Section V, Fig.

V-5. In the vast majority of cases, the computer was already on the " bad" channel and the operator simply left it there.

Id.

271. MPR explained the dilemma faced by the TMI-2 operators:

"TMI-2 was provided with two makeup tank level instruments. The reading of one was used as input to the console mounted strip chart I recorder which was used by the operator for current information as to the tank level and its trend over the previous few hours. The other I instrument provided input to the plant computer and was, therefore, the instrument on which the calculations in the RCIB test were based. The operator had a selector switch which he could use to exchange the level instruments. This presented the operator with a problem whenever one of the instruments was malfunctioning -- if I he selected the ' good' instrument for the test, his console indication was ' bad' and vice versa.

There was a significant period of time, from early December 1978 through the early part of January 1979, during which one of the instruments had a signal which had both a higher mean value and a much higher amplitude fluctuation than the I other. The makeup tank level strip chart record shows when instrument switches were made."

Id.,Section V, pp. V.19-20. MpR further explained that this instrument problem had a significant effect not only on the validity of the test result, but also on the ability to use the MUT level strip chart trace to determine whether there were water additions, " feed and bleed" operations, or hydrogen additions during leak rate tests:

-147-

I "This instrument problem has a significant effect on the validity of the test result as well as on the ability to interpret from the makeup tank trace whether there were water additions, feed and bleed operations, or hydrogen additions during an RCIB test. In particular, if the operator switched the ' good' instrument to the I computer for the RCIB test, the RCIB test is reasonably valid; however, the makeup tank level trace is virtually worthless as a means to I confirm what happened during the test because of the very large fluctuations of the indication.

If, on the other hand, the operator failed to switch the ' good' instrument to the computer, then the RCIB test result is probably invalid.

In this case the makeup tank level trace is useful in assessing events during the test; I however, the makeup tank level values printed out in the test results would not be expected to correspond to those on the strip chart."

I_ d . , p. V.20. Nevertheless, Mr. Stier and NRR (Tr. 1272-74) concluded that the operators did not falsify leak rate tests by switching MUT level transmitters, and the Board agrees.

I 2. The Generic Defects or Inadequacies in the Analyses Prevent Adverse Conclusions from I Being Made Concerning Individual Responsibility.

a. Control Room Organization Was Such as I to Preclude a Conclusion as to the Person Responsible For a Given Test.

272. Although it would appear that we could have identified the operator responsible for a given leak rate test by simply observing who signed the test sheet as the

" performer", in actual practice it was not that simple.

Operators testified that one operator might have begun the test and a second operator might have concluded it, by answering the computer-generated questions. E.g., Tr. 1305-11, 3024, 3079, I

-148-I

3306, 3995. There even were instances where a leak rate test was begun on one shift and signed on a second shift. See, e.g., NRR Test Nos. 36, 56, 96, and 156 (Stier Test Nos. 122, 102, 62, and 1); Tr. 1623. Therefore, it would not necessarily be appropriate to assign responsibility to the CRO who signed the test even if the test result was falsified.

273. Similarly, we cannot determine from the log who may have made a water or hydrogen addition, who conducted a

" feed and bleed" operation, or who conducted any other such plant evolution during a test. Although responsibilities may have been shared differently from one shift to another, a number of operators testified that any of the CROs (and, perhaps, even an SRO) might have performed an action at the console that could have affected a leak rate test result.

E.g., Tr. 2845-46, 3025-26, 3967; see Tr. 3459-50, 274. Furthermore, the testimony was consistent that, although shift foreman or supervisors may have approved a test, they typically would not know what, if any, operator-caused actions may have occurred during a test that could have invalidated it. E.g., Tr. 3107-10, 4285-87.

275. Therefore, we cannot conclude that any given leak rate test that may have been falsified or that violated procedures was necessarily the responsibility of any particular CRO or SRO.

I

-149-I

l

b. Control Room Communications Were Not Always Such as to Permit the Conclusion that Operators Knew of Plant I Manipulations During Leak Rate Tests, or Knew that Leak Rate Tests Were Being performed During Plant Manipulations.

276. Leak rate tests were performed fairly frequently as part of a large number of periodic surveillance tests at I TMI-2. Stier Report, Vol. I, pp. 46-50; see, e.g., Cooper Prep.

St., p. 8, ff. Tr. 2835, 277. Although it was typical of some shifts for only the panel operator to perform plant evolutions, and the

" switching and tagging" operator to perform surveillance tests, i practice oftentimes deviated from what was " typical". Stier Report, Vol. I, p. 49; Tr. 3995. Because the leak rate test was routine, was performed frequently, McGovern Prep. St.,

p. 9, ff. Tr. 3148, and required little effort to commence, e.g, Cooper Prep. St., p. 3, ff. Tr. 2835; Tr. 3924, it was entirely possible that one operator would not know that another operator was performing a leak rate test. E.g., Cooper Prep.

St., pp. 8-9, ff. Tr. 2835; Tr. 3919, 3967. Therefore, one operator could perform a plant evolution (such as a water or hydrogen addition, a " feed and bleed" operation, or an RCDT pumpdown) and not know that the other operator was performing the leak rate test. E.g., Illjes Prep. St., p. 4, ff.

Tr. 3010. Conversely, an operator could perform a leak rate test, and not be aware of a plant evolution being performed during the test interval. Cooper Prep. St., pp. E-9, ff.

-150-I 1

I Tr. 2835. At that point, the operator had two choices if he became aware of the problem -- invalidate the test (as contemplated by the procedure, Stier Report, Vol. V(B), Tab 19,

p. 5), or enter the appropriate water addition or loss (from the RCDT), it either had occurred. See id. p. 4; see generally leak rate test computer sheets, calling for that information.

278. It appears that both responses took place at TMI-2. Operators testified that they threw away leak rate test sheets if, after they completed the test and answered the computer-generated questions, they became aware of plant evolutions that might have affected the test results. E.g.,

Cooper Prep. St., pp. 3-4, ff. Tr. 2835; Faust Prep. St.,

pp. 2-3, ff. Tr. 2511. Operators also submitted, and supervisors approved, leak rate tests that included water additions to the MUT. E.g., Stier Test Nos. 1-9; see, e.g.,

Hemmila Prep. St., pp. 4-5, ff. Tr. 4039; Coleman Prep. St.,

p. 3, ff. Tr. 2579.

279. The Board finds that either invalidating a test during which a plant evolution occurred, or retaining the test (if the water addition, for example, was entered into the calculation) was permissible. The applicable Surveillance Procedure 2301-3D1 permitted either invalidation or retention (if appropriate information had been entered into the computer). Therefore, the operators were correct in following either approach.

I I -151-I

I

c. TMI-2 Was a New Plant with Ongoing Plant Oscillations'that Affected Leak Rate Test Results.

280. TMI-2 became operational in March 1978 and leak rate tests were performed during Modes 1, 2, 3, and 4 until March 28, 1979, the date of the TMI-2 accident. Stier Report, Vol. I, pp.67-103. MpR explained the effect that plant oscillations would have on leak rate test results:

" Oscillations in plant conditions were observed at TMI-2 at several times as evidenced I by time-history data, particularly the relatively regular fluctuation which are often evident in the makeup tank level (See Appendix K for i examples). For example, a review of the reactimeter data from January 4-5, 1979, (see Reference 1, Faegre & Benson, Exhibit 37) shows oscillations in reactor coolant temperature, I pressurizer level, and makeup tank level, which have a period of about twelve minutes and peak-to-peak ranges of about 1*F for temperature, I 5 inches for pressurizer level, and 2.5 inches for makeup tank level. The oscillations are not in phase but appear to have a constant phase I relationship indicating they are likely all attributable to a common source.

"If the RCIB procedure were exactly correct, the oscillations in plant conditions would not affect the leak rate test results. However, the actual RCIB procedure contains approximations and l I method errors as discussed previously. The oscillations in plant conditions provide a way to change the conditions from the beginning to the l3 end of the test, and thus bringing the method g errors into play to affect the results. It appears this effect was significant as shown by the Faegre and Benson (Reference 1) analysis of I the January 4 and 5, 1979, data.

separate MpR analysis of this effect of the oscillations was not carried out, this error was Although a I included in the overall evaluation of the method errors discussed previously in Section III.C."

I I -152-I

I Id., Vol. IV(A),Section III, pp. III.13-14. These errors  ;

1 were, "at worst", "of the same magnitude as the instrument I errors", although typically they would be less. I_d . ,

l j

p. III.15. In certain periods, such as March 1979, "the method and makeup tank instrument errors had biased the test results upward by several gpm." I_d . , p. III.16.

281. The effect of plant oscillations MUT instrument errors makes it difficult to determine whether the MUT level strip chart trace or test data indicate that water or hydrogen additions occurred during leak rate tests.

I ,

d. MUT Level Strip Chart Time Error precludes a Conclusion That Operator Action Occured During Certain Tests.

282. The Board is faced with a difficult problem in

( attempting to determine whether plant evolutions, such as water additions, " feed and bleed" operations, hydrogen additions, or other actions that could have affected a test result occurred during a leak rate test. The problem is: how does one know that the " time" of an evolution, and the time of a test, coincided?

283. The investigators could not, of course, reconstruct the events, and no one could expect the operators to remember the precise time that a plant evolution occurred.

Tr. 1075-76; 1305-08. Therefore, the investigators relied on the CRC log as the time that logged events occurred, and

""~'"'""'"""~"""'"""""""'"""'"""""""'"""

I I -153-I

I whether the plant evolution occurred during the test interval.

See, e.g., Stier Report, Vol. IV(A),Section VI, pp. VI.18-24 (references to time that water additions were recorded in the CRO log); see also Tr. 1284-88 (NRR assumed clock times and log times were the same). )

284. The problem, of course, is that there is no way to know that the computer clock " time" is the same as log

" time". Operators testified, for example, that they sometimes used the wall clock, sometimes used their wristwatches, and sometimes used the computer to enter the " time" of an evolution in the logs. Tr. 3356-59; see also Tr. 3315-21, 3984. More importantly, it is clear that the time an event was logged often was recorded substantially later than the time it occured. Tr. 1288 (Stier).

285. We therefore find that the results of all of the investigations must be viewed skeptically, because the investigators relied (as they had to) on the CRO, Shift Foreman, or (in some cases) AO logs for "real" time. The logs show many entries that end in "5" or "0", suggesting that 1

actual times were approximated. The operators recalled that it might have been a substantial period of time before they would 1

have the opportunity, or would remember, to enter an operator action or plant event in the various logs. E.g.,

Tr. 3377-82(Kidwell) ("And the way we kept our records -- and we didn't do it on purpose, but the way we kept them, we've damned near made it impossible [to determine the exact time of

-154-I

I a plant evolution]."). Under the circumstances, the documentary evidence is often not reliable enough to reconstruct the sequence of events.

286. There are tests where there is little doubt that a plant evolution occurred during the test; for example, where the operators entered the water addition in the computer calculation for the test. However, in a significant number of tests, the alleged water or hydrogen additions were made so close to the beginning or end of the test, or the difference between chart and "real" time is so significant, that there is substantial doubt whether the evolution occurred during the test. See, e.g., Exhibit 1-B (Green Volume), Stier Test Nos.

15, 47, 48, 62 (NRR concludes no water or hydrogen additions likely, see summary sheet), 81, 130, 155, and 156; Tr. 1075-77, 3993-97. We find that the documentary evidence is not sufficient to conclude that the alleged operator actions occurred during those Tests because there is substantial doubt about the timing of the actions. See Tr. 1299 (26 out of 161 tests where time difference between MPR and NRR an' lyses is more than 5 minutes).

I e. MUT Level Transmitters Were At Times Inaccurate but, When Not Tagged I Out-of-Service, Could Have Been Used for Leak Rate Testing and Were Not Used to Falsify Results.

287. There were substantial periods of time when one or both MUT level transmitters were inaccurate, but the use of

-155-I

I the same transmitter for both the beginning and end of the test was consistent with Surveillance Procedure 2301-3Dl. Tr.

1243-45. The Procedure states:

I "The same sources should be used when recording initial and final RCS temperature, pressurizer level, make-up tank level and RCDT level. Differences in sources could be misinterpreted as RCS leakage when comparing successive readings."

Stier Report, Vol. V(C), Tab 19, p. 3. The instruction to the operators was only to use the "same" transmitter for the test, not the most accurate. There was no guidance to the operator concerning which level transmitter to use (Tr. 1243-44) and there was no caution in the procedure against using an unstable transmitter (Tr. 1244-45).

288. The operators were presented with a dilemma in having to choose between transmitters, because one fed data to the MUT level strip chart indicator, while the other fed the computer. Stier Report, Vol. IV(A),Section V, pp. V.19-20.

Because it was obviously important to have an accurate MUT level indicator (the evidence indicates that it may have been the most important indicator in the control room (Tr. 1279-82, 3523)), it is difficult to be too critical of the operators for using the more stable level transmitter for the strip chart, and the other for the computer (and thus the leak rate test computation). See Tr. 3933-34.

289. Mr. Hemmila succinctly explained that he was not even aware of a malfunctioning MUT level transmitter, and that, I

-156-lI - -

I I in any event, he felt justified in using either transmitter for the leak rate test unless the one used was tagged out-of-service:

"During 1978-1979, I was not aware of a period during which one of the level transmitters was malfunctioning. Unless an instrument was I tagged out of service, any operator had a right to use it and to rely on it. I was not aware of any operator manipulating a leak rate test by switching level transmitters during the test."

Hemmila Prep. St., p. 6, ff. Tr. 4039. We agree with Mr. Hemmila that, if an MUT level transmitter was not tagged out-of-service, the operator could use it for a leak rate test, assuming that he used the same transmitter at the beginning and end of the test, as provided by the Surveillance Procedure.

Stier Report, Vol. V(C), Tab 19, p. 3; Tr. 1243-45.

290. There were leak rate tests run at various times in December 1978 and Janaury 1979 with an MUT level transmitter (LT-1) that was tagged out-of-service. Tr. 2394-98; see Exhibit 5-B, Attachment 5, Table 11. Table 11 shows that no operator performed more than two such tests, and the investigators did not find that leak rate tests were falsified using the unstable level transmitter. Tr. 1756.

291. We are unaware of any evidence that an operator i switched MUT level transmitters from that used at the beginning to that used at the end of the test. Mr. Hartman stated that he was unaware of such an action (Stier Report, Vol. I, p. 86),

which is consistent with Mr. Hemmila's testimony. The

-157-I l

I investigators could only find one test during which different level transmitters were used at the beginning and end of the test, and they concluded that the use of the different level transmitters was inadvertent. Tr. 1208-09. No evidence was developed at the hearings to contradict that conclusion. Tr.

1272-74. Accordingly, we conclude that an unstable MUT level transmitter was not used to falsify leak rate tests. See Tr.

1756 (Staff witnesses came to same conclusion). We also find that the use of an inaccurate MUT level transmitter may have led various NHC investigators to conclude incorrectly that water or hydrogen additions were made during certain tests (see generally Exhibit 1-B, MPR analysis of NRR evaluation),

f. Water Additions to the RCS Were I Permitted by the Procedure and Were Often Required for Safe Operation or Maintenance of Necessary power Levels.

292. We found no evidence that operators or supervisors were trained to avoid water additions to the MUT during leak rate tests, although some had that practice. E.g.,

Illjes Prep. St., p. 3, ff. Tr. 3010, phillippe Prep. St.,

p. 3, ff. Tr. 4432. The training received was simply to answer the computer-generated questions, including the question about water additions to the MUT. E.g., Mell Prep. St., pp. 1-2, ff.

Tr. 3239; Cooper prep. St., pp. 3-4, ff. Tr. 2835, 293. As previously discussed, water additions to the RCS were not prohibited by the leak rate Surveillance

-158-I

I Procedure. The computer-generated leak rate test sheets directed the operators to enter additions of water to the MUT.

See, e.g., Stier Test No. 1. Water additions during leak rate tests were often necessary, particularly during the period February-March 1979, because of relatively high identified leakage from one of the valves at the top of the pressurizer.

Stier Report, Vol. I, p. 93; Tr. 3090-94; Faust Prep. St., p.

3, ff. Tr. 2511. The CRO log for March 24, 1979, for example, shows that 11 separate additions, totalling 2300 gallons of water, were made to the MUT during the 11-7 shift. Stier Report, Vol. IV(C), Test 4, p. 76. Several of those additions were "DW" (demineralized water) added to keep boron concentration from increasing due to the pressurizer valve leak. Tr. 3094; see Stier Report, Vol. IV(C), Test 4, CRO log

p. 76 (" pressurizer spray & heaters in manual to equalize pressurizer boron with RCS due to relief valve leak-by"); Tr.

3973-77. Given that leak rate tests were run fairly frequently, the mere fact that a water addition occurred during a leak rate test is, therefore, not evidence of test falsification or manipulation. Stier Report, Vol. I, p. 101.

294. To understand why maintaining proper boron concentration was essential to the functioning of the TMI-2 reactor, the movement and position of the fuel rods in the reactor must first be understood. At TMI-2, there were eight groups of fuel rods (groups six and seven were the " controlling

-159-

I group") that were inserted into, and withdrawn from, the reactor core so as to regulate the rate of the nuclear fission I process. Tr. 4017. The movement of the fuel rods in and out l

j of the reactor core was governed by an " integrated control system" which responded to various parameters. Tr. 4024. One such parameter consisted of the discrepancy between power demand and actual power level, which was termed the " neutron error". I_ d . The integrated control system reacted automatically to the boron concentration in the RCS. Boron in f the RCS absorbs neutrons, and thereby affects the rate of the nuclear fission process. Tr. 3225.

295. TMI-2 operating procedures set limits on the extent to which the fuel rods could be inserted into the core (this range was known as the " band"). Tr. 1314-15. For example, the controlling group was normally maintained between 90 to 100 percent out of the core. Tr. 1314. An instrument in the control room indicated the extent to which the fuel rods were withdrawn from the core (Tr. 4016), and operators were expected to check those readings periodically during the shift.

Tr. 4025. However, rod positions observed in this way were logged only at the beginning and end of each shift. Tr. 4022.

TMI-2 operating procedures required operators to take corrective action whenever the fuel rods moved outside their specified band. Adjusting the boron concentration of the RCS was one way of fulfilling that directive. Tr. 1314.

I -160-

I 296. There was a continuous circulation of water between the MUT and the RCS (Tr. 1309, 3228) that was regulated by an automatic level control valve. Tr. 3231. Thus, an operator could adjust the boron concentration in the RCS by i altering the boron concentration in the MUT.

297. Several fluids were used to alter the boron l

l concentration within the MUT. Tr. 1309. The addition of

" demineralized" water (also known as "deborated" water) reduced l

the boron concentration within the MUT. Tr. 3094, 3229. On the other hand, the addition of boric acid increased the boron concentration within the MUT. See Tr. 1309. Boron concentration could also be altered by adding the partially borated water present in the reactor coolant bleed tank. Tr.

3978.

298. One of the principal methods of altering boron concentration in the MUT was through a " feed and bleed" operation. Tr. 1312; see Tr. 3225. A " feed and bleed" operation entailed letting down (i.e., " bleeding") coolant from the MUT to either the reactor coolant bleed tank or the deborating demineralizers, and then adding (i.e., " feeding")

either demineralized water, boric acid or reactor coolant bleed tank water back into the MUT. Tr. 1309-10. The process could I also be done in the reverse order. Tr. 1310. As Messrs.

Russell, Capra, and Stier have concluded (Tr. 1316 and Tr.

1319), and the Board finds no TMI-2 operator used a " feed and

-161-

I bleed" operation to manipulate any leak rate test result. See Stier Report, Vol. IV(A),Section VI, p. VI.24. Rather, water additions to the RCS were permitted by the applicable Surveillance Procedure (see 4H 151-55, supra), and were often required for the maintenance of necessary power levels.

299. Adjustments to the boron concentration in the RCS were particularly necessary from mid-February 1979 until the date of the accident -- a period when there was substantial leakage from the valves on the pressurizer. Tr. 3090-94; 4056; Stier Report, Vol. I, pp. 82-83 (noting a high rate of RCDT collection during this period, the source of which was the leakage from the pressurizer valves). During that period, one or more pressurizer valves were leaking substantial quantities of essentially boron-free water in the form of steam, thereby creating a more highly concentrated boron solution in the pressurizer. Id.; see Tr. 3227, 3399. Because there was circulation between the pressurizer and the test of the RCS (Tr. 3093, 3399) boron concentration also increased in the RCS.

300. Numerous operators testified that they had to make frequent additions of demineralized water during the period of high leakage from the code safety valves to maintain proper boron concentration within the RCS, and thereby keep the fuel rods within their specified bands. E.g., Tr. 3090, 3399, 3816, 4056.

-162-

I 301. Mr. Hemmila's testimony before this Board illustrated how frequently additions of demineralized water had to be made during that time period. Mr. Hemmila testified that on the 11 p.m. to 7 a.m. shift on March 16-17, 1979, water additions had to be made more than once per hour "to keep the rods within their band". Tr. 4060. Thus, operators had a legitimate operational reason for adding water during that I period, even if this meant doing so during a leak rate test (a possibility that Mr. Capra acknowledged, Tr. 1314). Throughout 1978-79, the erratic and frequent responses of the integrated control system resulted in continual plant oscillations (and, consequently, control rod movements) which at times forced the operators to make adjustments to the RCS boron concentration.

See, e.g., Tr. 639 (testimony of Mr. Rockwell that the ICS

" tended to go around and around and around in circles" searching for a steady state condition); see also Tr. 4980-81, 4983-84 (Floyd).

302. In HH 238-48, supra, the Board found that MpR's analysis refuted the 55 alleged unrecorded water additions during leak rate Tests identified by Faegre & Benson or the NRC. As to the six unrecorded water additions discussed therein, the Board found insufficient evidence to conclude anything other than that those tests reflected occasional sloppiness, rather than willful manipulation. Those conclusions of the Board are bolstered by our recognition of

, I l -163-I

I I the operators' duty to maintain the necessary boron concentration in the RCS.

303. Of those six leak rate Tests, three occurred during the period of high identified leakage discussed above (i.e., Stier Tests 30, 43 and 44). There is ample testimony that operators had to add water continually during that time period to maintain necessary boron concentration. E.g., Tr.

3090-94, 4056; see also Tr. 1315 (Mr. Capra acknowledged the possibility that a1 operator might unwittingly perform a " feed and bleed" operation during a leak rate test). As we have found, control room conditions were such that one operator could add water without knowing that his shiftmate was running a leak rate test. In sum, the overwhelming weight of the evidence concerning February-March 1979 points to a legitimate rationale for water additions during that period, coupled with a credible explanation as to why such additions went unrecorded. Accordingly, the Board finds that the unrecorded water additions during February-March 1979 were made for the legitimate operational reason of maintaining proper boron concentration, and not to falsify leak rate test results.

l 304. Of course, maintaining proper boron

! m concentration was also a concern prior to mid-February, 1979.

1 Thus, the existence of this legitimate operational rationale for adding water to the MUT strengthens our finding in TV 238-42, supra that none of the alleged unrecorded water l

= -164-

I I '

additions made prior to February 1979 were made with the intent to falsify a leak rate test. In sum, evidence of leak rate test falsification through unrecorded water additions is very weak, while there are legitimate explanations for such additions.

305. For all these reasons, we cannot find that water additions to the MUT during leak rate tests were, if accounted for in the calculation, evidence of leak rate test falsification or manipulation. In any event, with the exception of the tests performed on "D" shift (which we discuss infra in the individual responsibility findings), there are no more than a few such tests for any shift, let alone any individual operator. Exhibit 5-A, Enclosure 1, p. 7.

I g. Hydrogen Additions to the MUT Were Normal Plant Evolutions, Were Not g Prohibited by the Leak Rate g Procedures, and Were Not Shown to Affect the Test Results.

306. Hydrogen additions to the MUT were normal plant evolutions which were performed by the CROs or the AOs. Stier Report, Vol. IV(A),Section VII, p. VII.l. The purpose of adding hydrogen to the MUT was to sca"enge free oxygen from the RCS and to maintain net positive suction head in the RCS pumps.

Kirkpatrick & Wermeil Prep. St., pp. 12-13, ff. Tr. 376. There is essentially no information from the plant records which can be used to establish how often a hydrogen addition resulted in i

1 -165-

I a MUT pressure greater at the end of leak rate Test than at the beginning. Stier Report, Vol. IV(A),Section VII, p. VII.3.

307. As already discussed, it has been alleged by Mr. Hartman and the investigators of TMI-2 leak rate testing that hydrogen additions to the MUT may have caused leak rate test results for unidentified leakage to be lower than would otherwise have been the case. E.g., _i_d., Vol. 86.

I I, p.

However, MpR indicates that such additions may have actually made the value for unidentified leakage depicted on the tests to be more accurate:

"It should be noted that, except where a large water addition was made in the course of a test, the makeup tank level would almost always [be]

I lower at the end of a test than at the beginning when the plant condition (s] were steady.

Consequently, the makeup tank pressure would I normally be lower at the end of the test than at the beginning and, if anything, the leakage results would be biased toward showing more leakage than the actual value."

Id., Volume IV(A),Section VII, p. VII.3. Therefore, a hydrogen addition that wholly or partially compensated for the loss of MUT pressure that would otherwise have occurred from the beginning to the end of a leak rate test would tend to offset (in whole or in part) the error produced by the effect of the loss of MUT pressure on the test results.

308. Faegre & Benson found that there may have been unlogged hydrogen additions to the MUT in addition to those that were logged. Id., pp. VII.5-6. However, MPR concluded that the documentary evidence does not permit that finding:

I

-166-I

I "Since the hydrogen effect is indistinct at best, it does not appear possible to pick out an i unlogged hydrogen addition with certainty.

Although each of those instances where hydrogen may have been a factor has been reviewed in detail, these fundamental limitations make I positive confirmation of any unlogged hydrogen additions virtually impossible."

Id., p. VII.7. Therefore, based on MPR's technical analysis, which we regard as superior to that of the other investigators, the documentary evidence alone is not sufficient to conclude that hydrogen was added to the MUT during a leak rate test where it was not logged.

309. As we have already discussed, there are only 14 leak rate tests in which it may still be contended that a logged hydrogen addition occurred in which the hydrogen addition potentially affected the test results for unidentified leakage. See id., pp. VII.7-ll. MPR concluded that hydrogen

" appeared to have an appreciable effect on the MUT level trace" on only two of those 14 tests. I_ d . , p. VII.9. In one of the two Tests (Stier Test No. 35), as we have already discussed, the hydrogen was logged only in the AO log one minute before the presumed end of the Test. I_d. Therefore, as we have I found, the hydrogen addition may not have occurred during the Test or, in any event, was not obviously directed by the operator who performed during the Test. See Faegre & Beson Report, Vol. Two, Chapter Four, p. 70 n.****. The other Test (Stier Test No. 38) appears to have been an experiment. See I

1 260, supra. In none of the remaining 12 tests can the 1 -167-I

I effect, if any, of the hydrogen addition be detected. As MPR I stated, concerning eight of the twelve tests:

"In these cases, there is no deflection of the trace which is clearly a typical hydrogen I addition and, in many cases, there is simply no effect at all. Evidently, hydrogen addition was often not effective in changing the indicated makeup tank level."

Stier Report, Vol. IV(A), p. VII.10. None of the remaining four tests "can be confirmed to have involved a hydrogen addition," because "all four of these alleged additions were in a time period when there were problems with one of the two

[MUT] level instruments." I_ d . , pp. VII.10-11.

310. The conclusion is inescapable that, with the possible exception of two Tests (Stier Test Nos. 35 and 38),

hydrogen additions to the MUT did not affect leak rate test results. Even in the case of Stier Test Nos. 35 and 38, the additions may simply have made the test results more accurate by restoring the MUT pressure that existed at the beginning of the Test.

l h. It is Difficult to Determine Accurately 3 the Amount of Water Additions or MUT Level Changes from the MUT Level Strip Chart.

311. NRR alleged that certain operator-caused change to the plant occurred during leak rate tests, using the MUT level strip chart as the best source of such information.

Exhibit 5-A, Enclosure 1, p. 4. However, MPR had serious I reservations about reliance on the MUT level strip chart:

I -168-I

I

" Examination of the makeup tank level traces shows that the idealized shape of the trace (the regular, slow drop followed by a rapid rise of an I addition) is not often present: first, various distortions occur, i.e., humps and dips are not uncommon; and second, regular fluctuations occur.

I The distortions in the trace could result from a variety of sources. Because the level in the pressurizer is actively controlled, the makeup tank is the ' surge' or variable volume in the I system. Consequently, it responds to the wide variety of events which can change the reactor I

coolant system inventory and liquid volume. For example, changes in reactor coolant temperature and pressure, changes in makeup and letdown system lineup, sampling of the system, changes in power, or changes in the steam generator I conditions can all affect the makeup tank level.

Although each individual trace distortion undoubtedly has a technical reason for its I occurrence, that reason is quite likely to be complex. The ability to assign the more complex movements of the trace to specific outside events is, therefore, quite limited."

Stier Report, Vol. IV(A),Section VI, pp. VI.12-13. These limitations in the use of the MUT level strip chart prevent the use of that strip chart as a basis for concluding that water was " jogged" into the MUT and not logged:

I " Allegations have been made (Exhibits 1 and 2 and Reference 1) that additions of water were made gradually or

• jogged' into the makeup tank.

I As long as any such additions were of the same order as the normal movements of the level trace, both distortions and fluctuations, the makeup I Id.,

level trace is simply not capable of resolving this issue."

pp. VI.13-14. After setting forth the normal ranges of fluctuation in such parameters as RCS average temperature, RCS lI pressure, and pressurizer level, MPR concluded:

" Consequently, it appears that the ability of the makeup tank level trace to resolve

-169-

I so-called gradual or jogged additions of the order of 50 to 100 gallons is quite limited since I the resulting deflections would be essentially the same as the normal trace fluctuations. Each instance must be considered on its own; however, it appears unrealistic to expect to unambiguously I identify any such instances on the basis of examination of the makeup tank level trace alone."

Id., p. VI.15. Finally, MPR explained that the period in December 1978 and early January 1979 presented an additional barrier to reliance gn the MUT level strip chart to draw firm conclusions about operator actions from the MUT level strip chart:

"An additional limitation of the makeup tank level strip charts as a means to detect water additions is evident from review of traces in the time period late in December 1978 and in early January 1979. At that time one of the two makeup tank level instruments had a mean reading which I was higher than the other and, in addition,. had a much larger fluctuation. Because of the TMI-2 instrumentation circuitry, one instrument was displayed on the strip chart recorder and the I other was used by the plant computer in the inventory balance test. As a result of this instrumentation problem, correspondence between the level values on the test printout and the level chart cannot be expected. Furthermore, if the instruments had been exchanged so that the l trace on the chart record is the one oscillating 5 a large amount, the oscillations made any interpretations of the trace virtually I

impossible. In this regard, the makeup tank level values in those tests where the instruments were M exchanged, i.e., where the malfunctioning instrument readings were used by the computer to make the leakage calculation, are likely to be seriously in error and those RCIB test esults may be invalid. See Section V.C.2.e(l) for a summary of the instrument i shifting practice of the operators."

Id., l pp. VI.15-16 (emphasis in original).

l 5

1 -170- l 1

I l l

I 312. MPR's explanation of the limited utility of the MUT level strip chart trace per se also explains why we have determined that a number of the tests deemed " questionable" (e.g., Tr. 1618, 1626-29) by NRR cannot, in fact, be the basis for any adverse findings by this Board. Because of the inadequacy of that data alone to draw a firm conclusion about any given test, we are precluded from relying on it.

I 313. The difficulty of relying on the MUT level strip chart trace is exemplified by NRR's analysis of NRR Test No.

62. The operators switched to the more stable MUT level transmitter for that Test, leaving the less stable level transmitter connected to the MUT level strip chart. See Exhibit 1-B at Stier Test No. 96 (NRR Test No. 62. As MPR explained (id.):

"During the period of the test the malfunctioning makeup tank level instrument was I recorded on the level chart. The peak-to-peak fluctuations of the malfunctioning instrument are about 10 to 12 inches and quite irregular. We do not consider that it is possible to identify a I small water or hydrogen addition in the presence of such fluctuations. In addition, the portions I of the trace before and after the test, which are the properly functioning level instrument, show no offset. An offset would be expected, if water or hydrogen had been added."

MPR noted that even NRR considered its analysis of this Test to be " questionable" (id.) and we agree. It appears to the Board that the performer of NRR Test No. 62 (it is signed by Mr. Cooper) did nothing for which he can be criticized. He used the more stable MUT level transmitter for the Test, and I

1 -171-I

I I therefore, the MUT level trace during the Test period in so irregular (because it was fed by the unstable level I transmitter) that it is impossible to conclude that anything improper occurred during that Test. Tr. 1624.

314. In general, (except for specific disagreements l 1

I discussed herein, we have relied on MPR's technical analysis of l i

tests NRR considered " questionable", because of NRR's l over-reliance on the MUT level strip chart alone as the basis for potentially adverse findings on " questionable" data. NRR Test No. 62 is the perfect example of a Test as to which it would be unfair to make an adverse finding on the basis of I NRR's analysis.

I I

I I

I I

I I -172-I

V. TRAINING OF TMI-2 OPERATORS ON LEAK RATE TESTING WAS  ;

WOEFULLY INADEQUATE; IT LARGELY CONSISTED OF MINIMAL  !

ON-THE-JOB TRAINING.

A. Classroom Training Was Virtually Non-Existen .

315. The testimony of Mr. Dennis J. Boltz, who was a training instructor assigned to the licensed operator training group at TMI-2 during 1978-79, Tr. 2218-19, made apparent the fact that classroom training regarding the performance of leak rate tests was virtually non-existent. Mr. Boltz acknowledged that, in 1978-79, the TMI-2 training department focused exclusively on " textbook type" training consisting of little more than providing a copy of the Technical Specifications to r the operators and supervisors to read, and thus it "had to rely heavily [on] the operations and on-the-job training aspects of

[the CROs'] training programs with qualified personnel on shift." Tr. 2225, 2226; see Tr. 2582, 2714, 2799, 2839.

316. Mr. Boltz searched training records from 1978-79 and found only one document that referred to leak rate testing; it merely states that a CRO must be "[c]apable of using [the]

computer for calculations of leak rate, heat balance, Reactivity Balance, etc." Exhibit 16; see Tr. 2220, 2226.

Thus, it is obvious that classroom training on leak rate testing was essentially non-existent.

317. Mr. Boltz's testimony before this Board also highlighted the crucial information concerning leak rate testing that was omitted from the TMI-2 classroom training I

-173-I I

I program. The training program did not include a discussioi. of the requirement of Administrative Procedure 1012 to log the I start and completion (or suspension) of each leak rate test performed, Tr. 2224-25, nor did it include a discussion of the Administrative Procedure 1010 requirement that leak rate tests not performed successfully or that failed to meet the acceptance criteria should be documented through the use of an exception or a deficiency. Tr. 2224. The classroom training program at TMI-2 also failed to provide any instructions regarding the performance of leak rate tests by hand.

Tr. 2229. Nor were instructions provided during the three-week course at the B&W simulator taken by all CRO trainees prior to becoming licensed. Tr. 2231.

318. The only instruction as to how to perform leak rate tests was provided as part of an operator's on-the-job training (Tr. 2225; see, e.g., Mell prep. St., pp. 1-2, ff. Tr.

3239), but the TMI-2 training department took no action to assure that the on-the-job leak rate test training was consistent among the shifts. Tr. 2224. The Board thus finds that classroom training concerning leak rate testing at TMI-2 was essentially non-existent during 1978-79, and failed to teach the TMI-2 operators and supervisors the requirements pertaining to, and proper procedures for, leak rate testing.

I I

I -174-

I B. On-The-Job Training Essentially Consisted of a Demonstration of the Manner in Which to Perform I the Leak Rate Test, with Virtually No Guidance as to the Significance of the Test or the Actions to Be Taken If Test Results Were Excessive.

319. The evidence of record is that the on-the-job instruction received by the TMI-2 operators concerning leak rate testing was confined to a basic demonstration of the steps necessary to conduct the leak rate tect.

I As Mr. McGovern testified before this Board, "[a]ll I remember about performing leak rates and learning how to do them was what I was taught by the CROs. This is how you punch it out and make the leak rate test come out. That's what I remember about leak rate training." Tr. 3207; see Mell Prep. St., pp. 1-2, ff. Tr.

3239.

320. Mr. McGovern's recollection has been echoed by the testimony of several other operators. Mr. Coleman testified that the on-the-job instruction he received concerning leak rate testing was not "very extensive."

Tr. 2583. Mr. Cooper testified that apart from being shown how to perform a leak rate test, his only instruction concerned the prescribed limits for plant leakage. Tr. 2916. Thus, there is ample testimony illustrating the limited nature of the on-the-job training received by the TMI-2 operators in 1978-79.

In essence, there was no training on the meaning or purpose of I the test. Tr. 523-24.

321. The TMI-2 operators and shift foremen were unanimous in testifying that their on-the-job training on leak I -175-

I rate testing never included any discussion of the relationship between the test and the potential safety significance of a critical crack in RCS piping. See ST 171-72, supra.

Mr. Illjes, a CRO, could not recall having received "any specific training on leak rates or the [ safety] implications" of performing them. Tr. 3082. Nor could Mr. Mehler (a shift supervisor) (Tr. 3859-60) or Mr. Adam Miller (a shift foreman)

(Tr. 3628) recall any such training. Other operators who testified to this void in their training include Mr. Cooper (Tr. 2916), Mr. Wright (Tr. 2672), Mr. Hitz (Tr. 3707),

Mr. Guthrie (Tr. 4126), Mr. Olson (Tr. 4009-11) and Mr. Conaway (Tr. 3105).

322. The on-the-job training at TMI-2 was also deficient in that it failed to instruct operators to take the prescribed actions in response to unsatisfactory leak rate test results. Numerous operators testified that they were never taught to apply Administrative procedure 1010, the Exception and Deficiency procedure, to unsatisfactory leak rate test results. Mr. Wright's testimony is typical in this regard. He stated that the requirement to apply the Exception and Deficiency Administrative procedure to unsatisfactory leak rate test results was never " impressed upon me, either in training or anything else." Tr. 2672. The testimony of Messrs. Coleman (Tr. 2636), Conaway (Tr. 3108), Hitz (Tr. 3669), Miller (Tr. 3648), Smith (Tr. 4344), and Scheimann (Tr. 2795) supports I

I I -17e-

I Mr. Wright's recollection. Mr. Mehler's testimony that failure to apply the Exception and Deficiency procedure had originated at TMI-l (Tr. 3870) illustrates that the on-the-job training program at TMI-2 never corrected the habits developed at TMI-1.

Other loose practices that were carried over from TMI-l and never rectified during the on-the-job training at TMI-2 included not logging all leak rate test results and not retaining invalid test results. Exhibit 7, NUREG-0680 Supp.

No. 5, p. 13.

323. It is also clear that the on-the-job training never conveyed the proper interpretation of the "72-hour" rule and the " Action Statement" requirement. The TMI-2 operators believed that they were obligated to enter the Action Statement only upon failing to obtain one satisfactory leak rate test result within a 72-hour period. E.g., Illjes prep. St., pp.

2-3, ff. Tr. 3010; Scheimann prep. St., p. 2, ff. Tr. 2831; Cooper prep. St., p. 5, ff. Tr. 2835. Only a few TMI-2 operators and supervisors seemed to have correctly understood that the requirement to invoke the "4-hour clock" in the

" Action Statement" applied when a valid leak rate test depicted unidentified leakage in excess of 1 gpm. Frederick prep. St.,

p. 2, ff. Tr. 2447; Zewe prep. St., p. 2, ff. Tr. 2946. As discussed in 11 340-389, infra, management's and l Mr. Haverkamp's efforts to correct this pervasive misconception were ineffective.

l I -177-

1 4

I l I

324. One consequence of this void in on-the-job training (and of their day-to-day experience with widely l .

varying test results) was that the TMI-2 operators viewed the leak rate test for unidentified leakage as an unrealistic and perfunctory administrative burden, not necessary for the safe operation of the plant. Stier Report, Vol. I, p. 109. As such, they did not view the test as establishing whether the plant complied with the 1 gpm LCO for unidentified leakage.

I_d . !Ievertheless, TMI-2 operators were very concerned about, and 2 acted promptly to, any increases in leakage that they perceived from other plant parameters. Id.

325. Actual unidentified leakage exceeded 1 gpm only briefly during October 16-18, 1978. See 9 230, supra. During that period, TMI-2 operators and supervisors promptly reacted to test results showing unidentified leakage exceeding 1 gpm by doing what the " Action Statement" required of them -- they ran additional tests, identified and repaired leaks, and evaluated the safety significance of the leakage. H 229, supra; Stier Report, Vol. I, pp. 109-12.

326. The Board finds that on-the-job training at TMI-2 essentially consisted of a demonstration of the manner in which to perform the leak rate test, with virtually no guidance as to the significance of the test or the actions to be taken if test results were unsatisfactory. See Tr. 4967-71 (Board summary of testimony). The conclusion is inescapable that the I

I -178-

lack of meaningful training was a major -- perhaps the major --

" cause of the problems with leak rate testing at TMI-2.

e L VI. THE " INACCURATE AND MEANINGLESS" LEAK RATE PROCEDURE PRECIPITATED OPERATOR INDIFFERENCE TO TEST RESULTS.

{ 327. An overwhelming majority of control room shift personnel at TMI-2 testified that they did not regard the leak rate test as a reliable means of measuring unidentified leakage. One operator, for example, described the leak rate l

test as "almost meaningless." Cooper Prep. St., p. 6, ff. Tr.

l 2835.

328. What caused the TMI-2 operators to lose faith in l the leak rate test? One crucial shortcoming observed by TMI-2 operators was that leak rate test results were often l

inconsistent with other indicators of plant leakage. Conaway l

Prep. St., p. 3, ff. Tr. 3097; accord, Exhibit 5-A, Enclosure 1, p. 11 (concluding that the perception of the operators was l correct). Consequently, most operators became skeptical of leak rate test results, and placed greater emphasis on other i

indicators of plant leakage. Tr. 2714-15, 3298-99; see also Phillippe Prep. St., p. 2, ff. Tr. 4432; Guthrie Prep. St.,

p. 2, ff. Tr. 4113. Such other indicators included the containment sump inventory and containment radiation monitor I

(Conaway Prep. St., p. 3, ff. Tr. 3097), MUT valve position and RCDT level (Hitz Prep. St., p. 3, ff. Tr. 3664; Tr. 4246), and the levels of the MUT and the pressurizer (Tr. 4242, 4244).

r>

-179-

I The operators and their immediate supervisors felt no obligation to rely exclusively on the leak rate test to measure unidentified leakage when they believed justifiably that they could monitor that aspect of plant safety through a host of other methods. As one supervisor testified, "[t]he leak rate test was only one of many, many items we looked at to determine leakage, and it just was not something that we dwelled on or focused on, which is being done now." Zewe Prep. St., p. 7, ff. Tr. 2946; see a1so Tr. 3015.

329. In addition to being inconsistent with other plant parameters, the leak rate test procedure was fraught with defects. 1H 184-224, supra. Numerous operators and I supervisors testified to the deficiencies that they believed existed in the computer program used to run leak rate tests.

E.g., Tr. 4239, 4348, 4463. CROs were also aware that the MUT level transmitters provided inconsistent information. h, Tr. 3255, 3934; see McGovern Prep. St., p. 7, ff. Tr. 3148.

330. The fact that, as a new plant, TMI-2 underwent frequent oscillations also contributed to the erratic leak rate test results observed by many operators. E.g., Hitz Prep. St.,

p. 4, ff. Tr. 3664; Adams Prep. St., p. 3, ff. Tr. 3776; see also Tr. 4805. Finally, some operators questioned the ability of the leak rate test to detect a mere 1 gallon per minute of unidentified leakage from an RCS containing approximately 80,000 gallons of water. As one CRO testified, " searching for I

I -180-I

I one gallon of leakage in a system with many thousands of gallons of water was similar to searching for a needle in a haystack." Olson Prep. St., p. 3, ff. Tr. 3911.

331. Most of the TMI-2 operators and supervisors believed, in 1978-79, that their leak rate test practices were appropriate. One reason for this belief evidently was that all TMI-2 leak rate test practices originated at TMI-1. E.g.,

I Floyd Prep. St., pp. 2-3, ff. Tr. 4894; Tr. 4965; see Exhibit 7.

332. Several other factors contributed to the operators' loss of respect for the leak rate test as a reliable indicator of leakage. As discussed supra in HH 131-61, supra, TMI-2 operators were confused by the lack of clarity in the Technical Specifications and Surveillance Procedure. TMI-2 operators were never taught the safety significance of the leak rate test in the course of their on-the-job training.

HH 315-26, supra.

333. The adoption of the TCN in March 1979 vindicated the longstanding belief of many TMI-2 shift personnel that inaccuracies in the leak rate test procedure caused test results to be inconsistent with actual unidentified leakage.

The Board found in H 192, supra, that the TMI-2 leak rate test procedure failed to correct for the difference in density between RCS coolant at average RCS temperature and RCS coolant in the RCDT, seriously compromising the accuracy of the test.

I I -181-I

I TMI-2 operations personnel lived with this significant error until March 16, 1979, when the TCN was implemented to correct that problem (although as the Board found in 191, supra, the TCN provided only a partial solution). Mr. Chwastyk testified that, upon the implementation of the TCN, leak rate test results showing excessive unidentified leakage ceased. Tr.

3587. Mr. Chwastyk believed that the TCN had solved the problem with the leak rate test procedure that had existed until it was adopted. Tr. 3587-88. The Board finds that the reduction in unidentified leakage depicted on leak rate test results because of the TCN supports the reasonableness of the belief of many TMI-2 shift personnel that the leak rate test had been inaccurate. It is not reasonable to have expected TMI-2 operators and supervisors to have realized that a density correction for additions to the MUT was also necessary.

334. It should be noted that Mr. Floyd, the Supervisor of Operations at TMI-2 in 1978-79, did not share the operators' and supervisors' attitude about the leak rate test procedure at the time. Floyd Prep. St., p. 5, ff. Tr. 4894.

While he did not believe that it could accurately measure leakage to the degree of precision implied by the test's use of l

ll 5

four decimal places for leakage values, he thought that it was a " valid indicator of leakage." Tr. 4911-12. It was not until after the Hartman allegations became public in 1980 that Mr.

Floyd analyzed the leak rate test procedure at TMI-2; at that

-182-J

L time, he quickly concluded that the procedure was not capable L. of producing accurate results. Tr. 4984-90.

r 335. The technical witnesses heard by this Board were l

unanimous in support of the operators' perception of the leak l L rate test as a meaningless indicator of actual plant leakage.

Mr. Rockwell testified that the leak rate test was I

L " dysfunctional" and that it " deadened [the operators]

sensitivity" by producing "obviously inappropriate results."

Tr. 450-51. Mr. Kirkpatrick conc.uded that the leak rate test I was "a bad test and that you couldn't trust the results and that the operators should have been suspicious of the results."

Tr. 1230-31. Mr. Russell observed that the substantial I variation in leak rate test results ir.fluenced TMI-2 operators l

to perceive the test "as essentially a meaningless test that l was giving them random results." Tr. 1237; accord, Tr. 1232-33 (Mr. Stier concurring in that view).

l 336. Continual plant oscillations were partly responsible for the inconsistency of leak rate test results l

observed by the operators, according to the technical

( witnesses. Mr. Russell concluded that plant oscillations at TMI-2 essentially "made the test a random number generator."

Tr. 641.

337. The technical witnesses agreed that the operators' confidence in the leak rate test was undermined by the fact that leak rate test results did not comport with other

-183-

I indicators of plant leakage, such as sump collection and the trend on the MUT level strip chart. Tr. 1232-33 (Mr. Stier).

Finally, numerous defects in the leak rate test procedure (see 1(1f 188-228, supra) diminished the reliability of the test.

Tr. 1236 (Mr. Stier), 893 (Mr. Rockwell).

338. The technical witnesses also testified that the format of the Technical Specifications confused many TMI-2 operators. Mr. Stier observed that a CRO could note the

" variety of sources of calculating leakage that are indicated there [in Technical Specification 3.4.6.1], and not come away with as clear an understanding as we now have in retrospect, that the inventory balance test was as important as it was at I TMI-2." Tr. 517. Mr. Russell agreed with Mr. Stier. Tr. 526.

Consistent with that testimony, the Board has found that there was no guidance in the Technical Specifications to the TMI-2 operators as to their obligation if the results of the various means of detecting RCS leakage were in conflict. See 1f 150, I supra.

339. The Board finds that TMI-2 control room shift l

personnel generally had little faith in leak rate test results as an accurate indicator of plant leakage, and that they were justified in that belief, based on the undisputed evidence that l

5 the test results were, in fact, inaccurate and meaningless.

See Tr. 4406 (Kelley, J.) (". . . cynicism about leak rates at i TMI-2 . . . well-founded. . . ."). The NRC inspector, I

I -184-I

I l

Mr. Haverkamp, and Mr. Seelinger, tried to convey the message l

that the test results should not have been discarded and should have triggered the " Action Statement" portion of the Technical Specifications when results exceeded the LCO for unidentified l leakage, but the next two Sections of this Recommended Decision demonstrate that they did not accomplish their objective. l l

VII. THE TMI-2 OPERATORS NEVER ADOPTED THE i HAVERKAMP/SEELINGER INTERPRETATION OF l

THE LEAK RATE SURVEILLANCE REQUIREMENT.

A. Leak Rate Tests Between October 15-19, 1978.

I 340. NRR Test Nos. 12, 12A-E, 13, 14 and 15 (Stier l

Test Nos. 142-150, Stier Report, Vol. IV(H)) were performed during October 15-19, 1978, and are particularly important to the issues in this proceeding for the reasons ' hat are set forth in the remainder of this Section. The Tests depict the following information:

Identified Leakage (Entered Unidentified Test Number Date and Time By Operators)* Leakage NRR Test 12 October 15, 1978 1.21 gpm -0.3504 gpm (Stier Test 150) 1928-2028 NRR Test 12A October 16, 1978 -1.21 gpm 2.5645 gpm (Stiet Test 149) 1935-2035 NRR Test 12B October 17, 1978 -2.16 gpm 2.0738 gpm (Stier Test 148) 1328-1428 NRR Test 12C October 18, 1978 -2.16 gpm 1.7754 gpm (Stier Test 147) 0514-0614 NRR Test 12D October 18, 1978 -2.16 gpm 1.2939 gpm (Stier Test 146) 0736-0836

-185-I

I NRR Test 12E October 18, 1978 -2.16 gpm 1.3219 gpm (Stier Test 145) 0859-0959 NRR Test 13 October 18, 1978 -2.75 gpm 1.0246 gpm (Stier Test 144) 1018-1118 (handwritten (handwritten

-2.776 gpm) round-off to 1 9pm)

NRR Test 14 October 18, 1978 0.196 gpm 0.1001 gpm (Stier Test 143) (handwritten (handwritten I 1221-1321 change to

-0.196 gpm) correction to

-0.283 gpm to conform to sign change for identi-fied leakage)

NRR Test 15 October 19, 1978 0 0 (by round-(Stier Test 142) 0746-0846 off, 0.2429 I if identi-fied leakage is subtracted from gross leakage)

• The value for identified leakage is different from the value I for identified leakage calculated by leak rate procedure because that amount (which is subtracted from gross leakage) includes leakage to the RCDT, whereas the amount entered by the operators corresponds to measured leakage observed by plant I personnel. Stier Report, Vol. IV(A),Section IV, p. IV-27.

341. Several conclusions follow from the above table, which will be discussed throughout this Section. First, on two Tests (NRR Test Nos. 12 and 14), the operator entered the wrong sign for identified leakage, thus making the indicated value for unidentified leakage erroneously high. (NRR Test No. 14 was corrected in handwriting, and the Board finds that the correction were appropriate because the change in sign for l identified leakage corresponded to the leak rate test program l

(see instruction at top of leak rate test computer printout).

lI l

-186-

L, u

Tr. 4516.) Second, leakage was identified between October 16, r~

L 1978 at 2035 and October 17, 1978 at 1428 (as demonstrated by comparing the indicated values for identified leakage entered by the operator on NRR Tests 12A and 12B). Third, additional leakage was identified sometime on October 18, 1978, as

{

demonstrated by the increased value for identified leakage entered that date. See also Tr. 3891, 4692-93. Fourth, MPR's analysis demonstrates that leakage was repaired at approximately 1130 on that date. Tr. 1549-51 (Stier discussion

{ of leak repair). Fifth, the leakage was repaired by October 19, 1978 at 0846, because the value for identified leakage entered on NRR Test No. 15 (Stier Test No. 142) is 0, indicating that the leakage was identified before that time had

[ been eliminated. Therefore, the Board finds that the evidence supports a finding that RCS leakage was identified during October 16-18, 1978 and repaired on October 18 (and, possibly,

[ on October 19), 1978.

342. These Tests also depict several other things

[ that are important to an understanding of the events of mid-October 1978. The rounding-off of values for unidentified

{

leakage first occurs in this period, with rounding off performed in handwriting by Mr. Floyd (Tr. 4914) and Mr. Mehler (Tr. 3788, 3891), and by the computer on NRR Test No. 15 and thereafter, until October 27, 1978. Also, NRR Test No. 12C contains the only written entry into the " Action Statement"

[

-187-

I under the leak rate test Surveillance Procedure (albeit after-the-fact, note time of "1200", Tr. 4914-15), indicating Mr. Floyd's understanding of the " Action Statement" requirement as of that date. The entry is recorded by Mr. Floyd's handwriting on the leak rate test computer sheet itself.

Tr. 4914. Mr. Mehler testified that he believed that TMI-2 was in the " Action Statement" of the Technical Specifications on October 18, 1978, separate and apart from Mr. Floyd's notation, because of the actions he can recall on that date.

Tr. 3845-49. That is the only documented entry into the

" Action Statement" at TMI-2. Tr. 4962, 5011.

i 343. The Tests just described also support the finding that when TMI-2 personnel believed that they might actually be experiencing relatively high RCS leakage, they did what the " Action Statement" portion of the leak rate procedure required of them -- they ran additional tests and they identified (and repaired) leaks. Stier Report, Vol. I, p. 112.

The Tests demonstrate the involvement of senior personnel, presumably because of one or more of the following three things: the imminent expiration of the 72-hour period for obtaining a " good" leak rate test (Tr. 3845-49), the involvement of Mr. Haverkamp in the problem (Exhibit 20, hereinafter cited as "Haverkamp Prep. St."), or the actual I presence of leaks (see NRR Tests Nos. 12-14 and Tr. 4688).

precise cause(s) of that involvement are not important to the The i

l I  !

1

-188-

E issues herein, but that involvement, and the findings set forth L

below, do suggest the unusual nature of October 18, 1978, insofar as leak rate testing was concerned.

B. Discovery of the problem on October 18, 1978.

344. On October 18, 1978, while conducting an inspection of TMI-2, Mr. Haverkamp overheard certain TMI-2

{

licensed operators or supervisors (or both) discussing leak rate tests that depicted unidentified leakage of greater than 1 gpm. Haverkamp prep. St., pp. 2-3; Tr. 2054-61, 2104-06.

Mr. Haverkamp joined the conversation and determined that the e

situation was being taken care of properly. Tr. 2054, 2103-04, 5242-43. For this reason, he decided that a notice of violation was not warranted. Tr. 5242-45. Nevertheless, the problem was important at that time because the last " good" leak

[ rate (i.e., one depicting unidentified leakage under 1 gpm) was performed between 1928-2028 on October 15, 1978, and thus,

{

under "the 72-hour clock" (rather than the "4-hour clock"),

because of the prevailing interpretation of the operators and supervisors, Floyd prep. St., pp. 3-4, ff. Tr. 4894, TMI-2

[

shutdown actions would have had to have begun later that day.

Tr. 3846-49 (Mehler).

C. NRC Inspector Haverkamp's Response.

345. After learning of the problem, Mr. Haverkamp

[ went to Mr. Seelinger's office to discuss the leak rate test

{

-189-

results of which he had just become aware. Tr. 2050-51:

Haverkamp Prep. St., p. 4. At that time, Mr. Seelinger was 1 Chairman of the TMI-2 Plant Operations Review Committee

("PORC") and was TMI-2 Superintendent, Technical Support.

Tr. 4628. He was responsible for interpretation of TMI-2 Technical Specifications, and was Mr. Haverkamp's principal contact at TMI-2. Tr. 4631-32.

346. In the conversation with Mr. Seelinger, Mr. Haverkamp explained that the failure of the operators and supervisors to invoke the " Action Statement" of Technical Specification 3.4.6.2 upon receipt of leak rate test results depicting unidentified leakage over 1 gpm violated the leak rate Technical Specification. Tr. 2059-61. Mr. Seelinger agreed to implement this interpretation, because it was the NRC's interpretation of the Technical Specifications and because he had been inclined to adopt that interpretation in any event when he first became aware of the problem approximately two weeks earlier. Tr. 4745.

347. During the same conversation, Mr. Haverkamp suggested to Mr. Seelinger that the TMI-2 operators were not taking full advantage of the leak rate Technical Specification because it referred to "1 gpm" as the LCO for unidentified leakage, not "1.0 gpm", meaning that the value for unidentified leakage could be rounded off. Seelinger Prep. St., p. 7, ff.

Tr. 4623; Tr. 4771; see Tr. 4698-701. Mr. Haverkamp finally 1

-190-I

l acknowledged at the hearings in this proceeding that he initiated the discussion of rounding off the unidentified l i

1 leakage test results and that Mr. Seelinger could have I l

interpreted the conversation as a suggestion to do so.

Tr. 2111-12, 2127-28, 2141-42, 5240-41, 5244-58. Mr. Seelinger testified that he would not have decided to round off the values for unidentified leakage had it not been the NRC Inspector who initiated the idea. Seelinger prep. St., p. 7, ff. Tr. 4623; Tr. 4770-73.

348. The Board finds that Mr. Seelinger's explanation for the rounding off practice is entirely plausible and credible, because unidentified leakage values were not rounded off before October 18 (e.g., NRR Test No. 12A) and were rounded off after October 18 (e.g., NRR Test No. 15). Furthermore, when Mr. Haverkamp advised Mr. Seelinger on October 27, 1978 that.the rounding-off practice was no longer acceptable, the practice was immediately discontinued. Tt. 4773-74; see Exhibit 4; NRR Test No. 26; Stier Report, Vol. I, p. 72. The pattern of events, and Mr. Haverkamp's testimony, confirms that the practice of rounding off would not have been adopted had it not first been suggested by the NRC. Although the Board believes that rounding off may not have been appropriate, it does not appear to have led to a violation of Technical Specifications, because a review of NRR Tests Nos. 15-26 (Stier Tests Nos. 143-132) discloses that in every instance the

-191-

~

1

I unidentified leakage would have been below 1.0 gpm even without I rounding off. Stier Report, Vol. I, p. 72. In any event, h= neither Mr. Seelinger nor Met Ed can be faulted for adopting a 1

practice that was suggested to them b1 the NRC and which has support in the Technical Specification's use of "1 gpm" rather than "1.0 gpm."

349. On the other hand, Mr. Haverkamp's instructions to Mr. Seelinger on the Action Statement requirement led to a variety of actions, including identification of leaks I. (Tr. 3891, 4692-93), performance of additional leak rate tests (NRR Tests Nos. 12C-14), performance of a satisfactory leak rate test (NRR Test No. 14), and repair of leaks (see NRR Test No. 15; identified leakage reduced to O). However, virtually none of the TMI-2 personnel ever understood the change in interpretation of the " Action Statement" requirement, although Mr. Seelinger did. Tr. 4708-20.

h D. Mr. Seelinger's Response to the Conversation with l 3 Mr. Haverkamp.

l 350. The pORC met in the afternoon of October 18, 1978 to discuss, inter alia, whether to report the failure to enter the " Action Statement" upon receipt of leak rate tests depicting excessive leakage. Haverkamp prep. St., Exhibit G, pp. 1-3; Tr. 4780. Following what Mr. Seelinger believed to have been Mr. Floyd's concurrence at "1000 hours0.0116 days <br />0.278 hours <br />0.00165 weeks <br />3.805e-4 months <br /> on October 19, 1

I 1978", Mr. Seelinger properly sent a " reportable occurrence" 1

-192-

I letter to the Commission (Region I), informing the Commission that the " Action Statement" had not been invoked when l unidentified leakage exceeded the 1 gpm LCO. Stier Report, Vol. V(C), Tab 31; Tr. 4780. Mr. Seelinger's letter indicated that "[ulnidentified leakage was reduced to [ Technical l Specification] limits on October 18, 1978" (id.), which was true (Tr. 4661) because, on the date of Mr. Seelinger's letter, I the rounding-off practice suggested by Mr. Haverkamp (Seelinger Prep. St., p. 7, ff. Tr. 4623; Tr. 3845-47, 4775-76) was in place. Unidentified leakage had also been " reduced" by the additional identified leakage discovered during October 16-17, 1978. Compare NRR Test No. 12A with NRR Test No. 12B.

351. Mr. Seelinger also did several other things to attempt to correct the problem. Tr. 4670-74. He discussed the problem with Mr. Floyd, which led to Mr. Floyd's " Operations Memorandum" No. 2-78-19 (about which more will be said in the next subsection). See Stier Report, Vol. V(C), Tab 37 I (enclosure to letter); Seelinger Prep. St., p. 8, ff. Tr. 4623.

He discussed the problem on two or three occasions with TMI-2 CROs and supervisors, informing them of the correct interpretation of the Technical Specification. Seelinger Prep.

St., p. 8; see also Floyd Prep. St., p. 4, ff. Tr. 4894. He satisfied himself that Mr. Fels was working on changes to the inputs to the computer and the computer program to make it more lEg f clear to the operators. Seelinger Prep. St., p. 8, ff. Tr. 4623.

I i

W -193-

I He called Mr. O'Hanlon, the then-Superintendent of TMI-1, to ensure that Mr. O'Hanlon was aware of the correct interpretation of the Technical Specification. I_d. Mr. Seelinger also raised the question again when he became TMI-1 Superintendent, to satisfy himself that the proper interpretation was being followed at TMI-1. Id. While it is obvious that the message did not get through to the TMI-2 operators and supervisors, the Board finds that Mr. Seelinger did attempt to rectify the problem. As he put it, "I never tried harder on any other LER. . . ."

Tr. 4680. Obviously, however, the TMI-2 operators and supervisors failed to understand about the " Action Statement" requirement, which failure may have been related to the language of the LER Narrative (see HH 358-64, infra) or the cancellation of Mr. Floyd's Memorandum. Tr. 4680.

352. Following his discussion with Mr. Haverkamp on the morning of October 18, and a subsequent discussion with Mr. Floyd (Tr. 4766, 4778), Mr. Seelinger reported the problem to the Unit Superintendent, Mr. Gary Miller. Tr. 4722-23.

Messrs. Miller and Seelinger discussed the matter with a senior person in Met Ed's Reading office, the identity of whom Mr. Seelinger can no longer recall. Tr. 4722-24. The decision was made at that meeting to treat the event for reportability purposes just like any other event, even though it had been the NRC that brought the problem to Mr. Seelinger's attention. I_d .

Mr. Seelinger did promptly report the event to the NRC once the I

ll -1e4-ii

E h

PORC and Mr. Floyd concurred that it should be reported. Stier

+ Report, Vol. V(C), Tab 31; Tr. 4780-81; 1 350, supra.

)

L E. Mr. Floyd's Response.

353. Mr. Floyd, the Supervisor of Operations of TMI-2 l

during the relevant period (March 1978-March 1979) (Floyd Prep.

St., p. 1, ff. Tr. 4894), was informed by Mr. Seelinger of Mr. Haverkamp's views on the leak rate issue. See Tr. 4673-74, 4778 (Seelinger). Mr. Floyd responded by, inter alia, issuance of his Operations Memorandum No. 2-78-19 to inform the TMI-2 operators and supervisors of the correct interpretation of the Technical Specifications. Stier Report, Vol. V(C), Tab 37.

Mr. Floyd's Memorandum correctly describes the appropriate b interpretation of the Technical Specifications (i.e., enter the

" Action Statement" when leakage exceeds the applicable limits, applying the "4-hour clock") and advises the shift supervisors to keep a list of identified leaks. Id., Vol. 1, pp. 113-14.

The identification and evaluation of leaks is, of course, the purpose of the Surveillance Requirement.

354. Mr. Floyd's Memorandum also incorporated rounding-off, which was in effect on the date (October 20, 1978) when the Memorandum was issued. Therefore, the amounts

, of unidentified leakage would be stated as a whole number: 0, 1, 2, etc. Because a value of "1 gpm" was (under the rounding-off practice) in compliance with the LCO, the lowest value not in compliance with the LCO would have been 2 gpm.

-195-

Therefore, the Memorandum instructed operators and supervisors that the "4-hour clock" would commence with a value from the computer for unidentified leakage of 2 gpm or more. See Tr. 4935-39.

{ 355. As already discussed, the practice of rounding off the value for unidentified leakage was stopped on October 27, 1978. Stier Report, Vol. I, p. 72; Exhibit 4. It appears that the operators and supervisors may have been confused about the appropriate interpretation of the Technical Specification after that date, because Mr. Floyd's Memorandum was cancelled thereafter. See Tr. 4680-81; Stier Report, Vol. V(C), Tab 34

("X" through Memorandum); see also Tr. 5004. Therefore, when

, the practice of rounding off stopped, the operators and supervisors may have believed that the old interpretation of the Technical Specification (i.e., the "72-hour" rule) was back l ,

in effect. Tr. 4680-81; Stier Report, Vol. I, p. 114. In any I event, most of the TMI-2 operators and supervisors did not change their interpretation of the Technical Specification, and l

continued to apply the "72-hour" rule (with the apparent exception of Messrs. Frederick, see Frederick prep. St., p. 2, l

ff. Tr. 2447, and Zewe, see Zewe Prep. St., p. 2, ff. Tr.

2946). The Board finds that Mr. Floyd's Memorandum was an

?

/ appropriate effort to correct the interpretation of the Technical Specification, but that the circumstances surrounding L

its issuance and cancellation may have prevented it from being effective.

f

-196-l

r F. Mr. Fels' Actions.

356. In response to Mr. Seelinger's decision to round off values for unidentified leakage, Mr. Fels reprogrammed the computer so that the value seen by the operator would be a whole number. Floyd Prep. St., p. ff. Tr. 4894; Tr. 4935;

( 4, see, e.g., NRR Test No. 15. At a later time, Mr. Fels also caused a reminder to the operators to be printed out at the top of the leak rate test computer sheet, stating ". . . You must

(

enter Dec[imal] P[oin]t with leakage values . . .

to clarify

{ for the operators the correct way in which to enter data to the l computer. Compare NRR Test No. 40 with NRR Test No. 41. He also caused the RCDT level to be depicted in inches, rather than require the operator to enter that level in volts in the computer as had been the case through November 9, 1978 (NRR Test No. 40). See NRR Test No. 41. These changes were the bases (or some of them) for the statements in the November 1,

( 1978 Licensee Event Report that "[ilnput data for the computer program which calculates unidentified leakage has also been clarified." Stier Report, Vol. V(C), Tab 29; Tr. 4704-08.

G. The Plant Operations Review Committee's Actions.

1. The October 19, 1978 Reportable Occurrence Letter.

357. Mr. Seelinger's October 19, 1978 " reportable occurrence" letter has already been discussed. 1 350, supra.

It was sent after Mr. Miller's instructions to Mr. Seelinger to

-197-

I treat the event as he would have if the NRC had not first brought the problem to Met Ed's attention (Tr. 4727), and, following PORC review, the letter demonstrates that the matter was so reported. Haverkamp Prep. St., p. 7.

I 2.

358.

The November The November 1,

1, 1978 Licensee Event Report.

1978 Licensee Event Report has generated considerable controversy. See Stier Report, Vol. V(A), Tab 2, pp. 15-16. The LER was "the followup report" referred to in Mr. Seelinger's " reportable occurrence" letter to the NRC and was Met Ed's explanation of the violation of Technical Specifications and indication of corrective action to be taken. Stier Report, Vol. V(C), Tab 31. There is no dispute that it contains errors (Seelinger Prep. St., p. 7, ff.

Tr. 4623; see also Tr. 4617-18, 4663-64), that differing versions were sent to the TMI-2 operators and supervisors, on the one hand, and the NRC on the other (compare Stier Report, Vol. V(C), Tab 29 with Tab 30), and that the LER, together with Mr. Floyd's Operations Memorandum, were the only written instructions to the operators and supervisors about the change in interpretation of the Action Statement. Tr. 4710-11. Only two TMI-2 operators (Messrs. Chwastyk (Prep. St., p. 4, ff.

Tr. 3407) and Phillippe (Prep. St., p. 4, ff. Tr. 4432)) have any contemporaneous recollection of the LER.

359. The Narrative to the LER contains at least two errors. They are: (1) the time and date that begin the I -198-

L Narrative ("[ alt 1000 hours0.0116 days <br />0.278 hours <br />0.00165 weeks <br />3.805e-4 months <br /> on October 19, 1978. . . "), which I

Mr. Seelinger evidently questioned but someone else did not correct (Tr. 4617-18), and (2) the reference in the Narrative's L

second and third line to "the last recorded acceptable surveillance performed at 1935 on 10-16-78" (a reference to NRR

[

Test No. 12A) should have been a reference to "1928 on 10-15-78" (corresponding to the actual "last recorded l acceptable surveillance", NRR Test No. 12). That time and date was also filled in by someone other than Mr. Seelinger l (Tr. 4618), evidently Mr. Stair. Tr. 4664.

360. The Narrative in the version of the LER sent to the NRC also contains this sentence:

"However, action was being taken to reduce the j Unidentified leakage to within allowable limits I

l and was accomplished at 0735 on 10-18-78, by determining a portion of this leakage to be identified leakage from the Reactor Coolant System and to be well within the limits of I

l

[ Technical Specification] 3.4.6.2.c."

This sentence was altered after the pORC draft of the Narrative was sent to Met Ed's licensing personnel in Reading. Tr. 4658; l

see attachment to Seelinger prep. St., p. 4, ff. Tr. 4623; Stier Report, Vol. V(C), Tab 29; compare i_d., Tab 30. The sentence is l

true without the words that begin with "by determining. . . ";

f those words were added after the Narrative was sent to Reading.

i Tr. 4693.

361. Furthermore, as NRR Test No. 14 demonstrates, the sentence would have been substantially correct (except for

-199-

l the Technical Spe'cification reference, which should have been a reference to T.S. 3.4.6.2.B, not 3.4.6.2.C (Tr. 4653)) had the time reference been to "1221" instead of "0735" on October 18, 1978; by that time, there had been additional leakage identified (see NRR Test No. 12E) and repaired (note significantly lower number for identified leakage on NRR Test No. 13).

i Mr. Seelinger evidently questioned the accuracy of the reference to "0735 on October 18, 1978" in the Narrative.

Tr. 4617-18. The Board finds that the use of the erroneous time of "0735" instead of "1221" must simply have been a mistake by someone other than Mr. Seelinger. Tr. 4664 (citing testimony of Mr. Stair). In any event, there was no motive to insert an incorrect time in the Narrative, because a reference to the identification of leakage by 1221 would have been equally useful to Met Ed's Licensee Event Report to the NRC.

362. The reference in the sentence to " allowable limits" is correct based on rounding off. Tr. 4701. In any event, it would have been correct even without rounding off had the time reference been "1221" instead of "0735", because unidentified leakage in the Test begun at 1221 was, in fact, below 1 gpm. NRR Test No. 14.

363. The Narrative also contains the following "The appropriate personnel will be instructed on i the requirements of the applicable sections of the [T]echnical [S]pecifications and the i

I -200-I

I requirements to . . . immediately invoke applicable action statements when the provisions of [the LCO's] are not met."

Tr. 4709. The Board finds that this statement, while in conformance with the Technical Specifications, simply did not

, convey the correct interpretation to the TMI-2 operators and supervisors, because none of them ever changed their interpretation of the "72-hour" rule. Tr. 4942 (Kelley, J.).

The sentence does indicate, however, as do Mr. Seelinger's efforts at implementing corrective actions (Seelinger Prep.

St., pp. 7-8, ff. Tr. 4623), that the PORC did understand the problem and attempted to inform the operators and supervisors of the correct interpretation. Stier Report, Vol. I, p. 114.

364. In the Board's view, the LER Narrative was a sincere effort at informing the operators and supervisors of the correct interpretation, but was obviously inadequate, as the next portion of this Section shows.

I 3. The PORC " Action Item" Instruction to the TMI-2 Operators and Supervisors.

365. By means of a so-called " Action Item", the PORC instructed Mr. Marshall, an Operations Department Engineer, to

"[i]nsure the following is documented by Ops review of LER 78-62/IP(1T) [t.here follows the penultimate sentence of the Narrative to the LER]". Id., Vol. V(C), Tab 30. With the possible exceptions of Messrs. Chwastyk and Phillippe, however, no operator or supervisor could even remember the LER or the i

I -201-

I l l

I pORC " Action Item", although each of them (except Mr. Bryan, I

)

Tr. 4608-09) agreed that he must have reviewed it because his I initials are on the " Action Item". Stier Report, Vol. V(C),

Tab 30. Even though the supervisors and operators did review the " Action Item" and did receive, or were required to review, I Mr. Floyd's Operations Memorandum No. 2-78-19 (Tr. 4943-44),

the import of those two documents simply did not get through to the operators and supervisors. E.g., Tr. 4675, 4942. The Board agrees with Mr. Seelinger that the confusion may have been created by the rounding off issue, which apparently led to cancellation of Mr. Floyd's Memorandum (Tr. 4680), or by the lack of clarity in the Narrative and the Memorandum themselves.

Tr. 4935-42.

366. Mr. Marshall, by placing the LER in the

" Required Reading" book in TMI-2 control room did what was required of him (Tr. 4397-99, 4664-65), and therefore the Board cannot fault him for the absence of a change in the operators' interpretation of the Technical Specification.

367. Mr. Haverkamp's Inspection Reports sent to Met Ed subsequent to the events of mid-October 1978 concluded that the leak rate test practices then in place at TMI-2 were appropriate and in compliance with regulations. As. Mr. Stier concluded:

"Three NRC inspection reports were submitted I to management between October 1978 and March 1979. None of them suggested that leak rate testing was being performed in violation of I

i

-202-I

I I procedures after the incident of October 18, 1978, which had been the subject of the LER.

Indeed, the Inspection Report dated January 26, 1979, indicated that the LER had been selected I for on-s'"e follow-up and stated:

I [T]ht inspector verified that the reporting requirements of Technical Specifications

. . . had been met, that appropriate I corrective action had been taken, that the event was reviewed by the licensee as required by Technical Specifications, and I that continued operation of the facility was conducted in conformance with Technical Specification 1imits.

"The report concluded that the LER had been I satisfactorily resolved."

Stier Report, Vol. I, pp. 137-38 (footnotes omitted).

368. In any event, the Board agrees with the conclusion of Mr. Haverkamp's testimony that:

" Subsequently, in April or May 1980, in support I of an NRC investigation of alleged RCS leakage measurement falsification by TMI-2 operators, I reviewed actions taken by the PORC and facility I management concerning previous excessive RCS leakage (See Exhibit G with enclosures). The PORC review and licensee management reporting of the excessive RCS leakage was performed as I required by the (Technical Specifications]."

Haverkamp Prep. St., p. 10; see also Exhibit G thereto. The Board finds that PORC did what was required of it under the Technical Specifications, but that, through failures of communication, the point never got through to most of the TMI-2 operators and supervisors. E.g., Tr. 4942 (Kelley, J.); Stier

.. Report, Vol. I, pp. 134, 139. Most operators and supervisors interpreted the Surveillance Requirement as permitting the I invalidation of leak rate tests due to the perception that the I -203-I

l reactor was not at " steady state" (see, e.g., Floyd Prep. St.,

l pp. 3-4, ff. Tr. 4894), and this practice never changed despite the LER. We found no evidence to demonstrate that, through the time of the TMI-2 accident, the PORC ever became aware that the

-' directive of the LER was not being followed.

L VIII. THE NRC EXACERBATED THE PROBLEM IN TMI-2 LEAK RATE l TESTING AND FAILED TO MAKE CLEAR TO TMI-2 OPERATORS m AND SUPERVISORS THE ACTIONS REQUIRED OF THEM.

~"

A. The TMI-2 Leak Rate Technical Specifications and L_ .

Surveillance Requirement Were the Product of NRC Licensing Actions.

369. As discussed in 1H 131-70, the Commission's regulations (10 C.F.R. S 50.36 (1986)) and the TMI-2 Facility License required Met Ed to have Technical Specifications that constituted legal requirements on Met Ed and the TMI-2 operators. The TMI-2 Technical Specifications were the NRC's standard technical specifications for B & W reactors, rather than being separately developed for the TMI-2 reactor. See Stier Report, Vol. VI(I), 2/15/85 Deposition of G. Miller, pp. 40-41. Because TMI-l had different Technical Specifications than did TMI-2 for leak rate testing, including different time clocks for obtaining a satisfactory leak rate tests, id., pp. 19-20; id., 2/7/84 OI Interview of G. Miller, pp. 47-48, the Commission, through its licensing process, created two different sets of licensing requirements with which the TMI management had to comply.

t

-204-

1 l

B l

370. Met Ed established a policy at TMI-l of running a leak rate test every shift, rather than once every 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, the required test interval. Floyd prep. St., p. 2, ff.

l Tr. 4894. The two surveillance periods, and the policy of performing one leak rate test per shift, created the potential for confusion for the operators and supervisors as to which set L oc data to believe te more tests were run than the number required. A:1 Mr. Gary Miller, the Station Superintendent, testified-

. . . You know, two tech specs. You look at them sitting on the Island with two different tech specs, although they are not totally different if you read them, they are hard to L read. They are hard to read the same. They are totally different in the way they are laid out.

l

[ "We started out in Unit 1. I was aware we had to do a leak rate every twenty-four hours, I believe. Jack Herbein set a policy--I don't think it was written, I've been asked this before

-- doing it every shift. And the purpose of that was to be sure we did one within the law, every twenty-four hours, not (to] be sure you do enough h of them, okay.

"I realize there is another side to that coin. Unit 2 we just couldn't continue that.

Remember I had the same Shift Supervisor. I had enough burden on him remembering two different specs much less telling him to remember to do one here and do one there, because there were different frequencies in the law. One every shift was the rule. And I thought we were adhering to that.

"I may have learned subsequently we weren't adhering to that well. In a lot of areas, we've all learned a lot. But I thought we were doing one every shift basically knowing they would get a shift when they couldn't do one. And that's my memory of that."

-205-

Stier Report, Vol. VI(I), 2/7/84 OI Interview of G. Miller, pp. 47-48. Thus, the TMI-l practice of running a leak rate test every shift unless one could not be done led to the fact that there would be several sets of leak rate data during the 72-hour period for performing such tests at TMI-2, and thus more potential for confusion than at TMI-1, because of its 24-hour interval for such tests.

l 371. The TMI-l Technical Specifications also included . _ - _ _

a provision for evaporative losses, the so-called " evaporative loss factor". Exhibit 21 (Bettenhausen Affidavit). As Mr. Bettenhausen explained, the inclusion of an evaporative loss factor had no safety significance. Id. Accordingly, the LCO for unidentified leakage at TMI-l could appropriately have been as high as 1.51 gpm (1 gpm + 0.51 gpm evaporative loss factor, see H 206, supra). Mr. Rockwell agreed, stating: "I might add we observed in our report that there was not an evaporative loss factor and that we felt that one could have been provided for TMI-2." Tr. 1163. Clearly, therefore, had the same factor been used at TMI-2, acceptable leak rate tests

• results would have been easier to obtain.

372. The operators' and supervisors' confusion developed from the differing Technical Specifications for the two TMI Units.

-206-f - ---

B. Lack of Clarity in the Technical Specifications and Leak Rate Surveillance Requirements Confused the Operators.

373. We have discussed at length the confusion in the TMI-2 Technical Specifications and Surveillance Requirements.

See 1H 150-61, 369-72, 480 supra. It is obvious that this lack of clarity misled the operators and supervisors.

374. It became apparent to the Board that the operators thought that they could invalidate a leak rate test if the plant was not at " steady state" (an undefined term).

E.g., Tr. 4675; see Floyd Prep. St., pp. 3-4, ff. Tr. 4894.

The Procedure does provide that tests can be invalidated (see Stier Report, Vol. V(C), Tab 19) and it does require that tests be run at " steady state". Mr. Seelinger understood, after his conversation with Mr. Haverkamp on October 18, that the operators were required to live with the results of a test once it was begun (Tr. 4674-76), yet that interpretation never got through to the operators. See, e .g . , Hitz Prep. St., pp. 2, 6, ff. Tr. 3664; Tr. 4675-77 (Kelley, J.). Clearly, the interpretation provided to Mr. Seelinger was " conservative" (Tr. 4677), in light of the provision in the Surveillance Procedure for invalidating a test if unaccounted-for operator action occurred during the test (Tr. 2206-09) (Capra). Even that interpretation did not get imparted to the operators by Met Ed or the NRC, because the leak rate testing practices did not change after the November 1, 1978 LER.

5

-207-I

375. The TMI-2 operators and supervisors understood Surveillance Procedure 2301-3D1 to permit water additions to the RCS during leak rate tests. The NRC Staff's post hoc interpretation that such additions were not appropriate unless unavoidable was never conveyed to the operators by the NRC in 1978-79.

376. The Technical Specifications are unclear in yet another respect; by providing four methods of leakage detection not providing any guidance to the operators as to what to do if I the various methods of leakage detection conflicted, the NRC created the confusion that existed among the TMI-2 operators and supervisors on this subject. See T 150, supra.

377. The Surveillance Requirement's " steady state" terminology also suggests that judgment, not mechanical application of leak rate test results to the 1 gpm LCO for unidentified leakage, is appropriate. Personnel as senior as Mr. Floyd, the Supervisor of Operations, understood that the application of judgment to leak rate tests depicting unidentified leakage over 1 gpm was permissible. Floyd Prep.

St., pp. 3-4, ff. Tr. 4894. If the proper interpretation was unclear to Mr. Floyd, the Supervisor of Operations, it is understandable that it was unclear to those who answered to him.

378. We believe that the Commission should, if it intends that operators and supervisors to be bound by the I

-208-I

provisions of its standardized Technical Specifications, see to it that they are written in plain English so that this type of confusion cannot occur again.

C. The NRC Caused the 1 GPM LCO on Unidentified Leakage At TMI-2 to Be, in Reality, 0.72 Gpm.

379. MPR's technical analysis of the TMI-2 leak rate tests for Mr. Stier explained that the test procedure had a major error which made the LCO for unidentified leakage more stringent than intended by the Commission. MPR explained:

"The RCIB test, as run at TMI-2, included a number of technical errors and inconsistencies.

I The majority of these errors had'only a small effect on the results in most actual conditionr..

Furthermore, the TMI-2 RCIB procedure was not I significantly different from the procedures being used by plants contemporary with TMI-2 except for one major error. This major error was that water additions to the makeup tank and the water I collected in the reactor coolant (RC) drain tank and the change in inventory of the reactor coolant system were not corrected to a common temperature."

Stier Report, Vol. IV(A), p. II.L. MPR also explained:

. . . the unidentified leakage limits at TMI-2 were more stringent than required and more stringent than the current (19PS) practice in the I nuclear industry. (This occurred primarily because the unidentified leakage at TMI-2 was limited to a volume rate (gallons per minute, I gpm) of one gpm which was evaluated at reactor coolant system (RCS) average temperature (about 581* F) rather than at room temperature. At room I temperature the coolant is substantially more dense than at reactor coolant system average temperature. Therefore, the one gpm unidentified leakage limit at TMI-2 was, in effect, 0.72 gpm i

I when evaluated at room temperature as would be done if current industry practice were followed."

IJ 1. , pp. III.1-2.

l -209-l L

380. The failure to use the appropriate temperature in the TMI-2 leak rate procedure was attributable to the NRC; it established the 1 gpm LCO (see 11 369-72, supra; see also Faegre & Benson Report, Vol. Two, Chapter One, pp. 15-16 (discussing NRC involvement in revisions to leak rate Technical Specifications)) and participated in the selection of room temperature, rather than RCS temperature. See Stier Report, Vol. IV(B), Appendix B. As MpR stated:

" Based on a review of the applicable NRC I Regulations and guidelines, room temperature, i.e., about 70*F, is the temperature which is applicable to the one gpm unidentified leakage I limit. This result is consistent with the definition of leakage rate in ISA Standard ISA -

S67.03 (Reference 5) which states that 20*C (68*F) is the applicable temperature for I determining leakage rate. The NRC participated in the preparation and acceptance for issue of ISA-S67.03. Furthermore, the results of NRC I evaluations show that nuclear power plants other than TMI-2 were operating in 1978-79 using room temperature as the applicable temperature for the one gpm unidentified leakage limit, and the NRC I indicated that these plants were using the

' appropriate' method. Finally, there is no safety concern at TMI-2 which ties the one gpm I unidentified leakage limit uniquely to a specific temperature. Although the crack size analyses in the TMI-2 FSAR were performed assuming 557*F as I the applicable temperature for the one gpm unidentified leakags limit, the conclusion of these analyses remains unchanged when room temperature is taken to be the applicable I temperature, i.e., the crack size is still a small fraction of the critical crack size.

"When unidentified leakage rate was determined at TMI-2 using the reactor coolant inventory balance procedure, the value was I determined at the average reactor coolant system temperature and not corrected to room temperature. Since the calculated leak rate was I

-210-lI t

I compared to the one gpm limit, this approach meant that a more stringent limit was being applied at TMI-2 than warranted by applicable I regulations, i.e., the limit on unidentified leakage at TMI-2 was, in effect, only 0.72 gpm instead of one gpm."

I_dl c . , pp. B.16-17; Tr. 543; see also Faegre & Benson Report, Vol. Two, Chapter One, p. 14 (discussing failure to correct for I density differences at different temperatures). Thus, the Commission created a problem for the TMI-2 operators and supervisors by imposing on them (in effect) a more stringent LCO (0.72 gpm, instead of 1 gpm) for unidentified leakage than was intended, and more stringent than the LCO applicable to I

TMI-l or other pWRs licensed by the Commission.

I D. Frequent NRC Inspections Not Only Failed to Correct Improper Leak Rate Test procedures and practices, They perpetuated and Compounded Them.

381. Mr. Haverkamp was the principal NRC Inspector for TMI-2 in 1978 and 1979. Haverkamp Prep. St. He testified before the Board about his role in the leak rate testing problems on two separate occasions. Tr. 2042-67, 2090-2210, 5237-58.

382. As the principal TMI-2 inspector in 1978-79, Mr. Haverkamp's responsibilities were, in his words, "to assure compliance with the technical specifications, that is, that the licensee demonstrated compliance with the technical specifications and to ensure that they had what was in my view, an appropriate perspective of safety." Tr. 2158. In carrying out his responsibilities, Mr. Haverkamp inspected TMI-2 on a

-211-I

I monthly basis; in all, he inspected TMI-2 fifteen times between February 1978 and March 1979. Tr. 2158-59. Six inspections during that period focused specifically on leak rate testing.

Tr. 2173-75.

383. As part of his inspections, Mr. Haverkamp stated he would spend several hours reviewing important plant records and logs, observing control room activities and plant operations, and conferring with plant personnel. However, in his review of the records and logs, he paid no attention to I details; for example, he never noticed that leak rate test start and stop times were not being recorded, and he never looked at the MUT level strip charts. Tr. 2048, 2056, 2190-91.

He never questioned the validity of the tests nor did he do hand calculations to verify test results. Tr. 2055-56, 2097-99. Even when he reviewed records of leak rate tests "in conjunction with a broad view of surveillance tests and the performance of surveillance tests", Tr. 2102, he noticed nothing awry. In sum, his examination of the logs and r eco r d.s never alerted him to ask any questions about TMI-2 leak rate testing after October 18, 1978. E.g., Tr. 2160-61; see Tr. 2130.

384. Mr. Haverkamp's frequent visits, and his observations of control room personnel, also failed to indicate to him that thete were any problems with leak rate tests.

perhaps this was because he only talked with the CR0s infrequently, spending most of the time with supervisors.

-212-I

! Tr. 2093-94, 2136. Or perhaps it was that he paid no attention to leak rate tests when they were performed because "it's not very exciting to observe . . . . Tr. 2160; see Tr. 2183. He acknowledged that there was no apparent effort to hide invalid test results from him. Tr. 2105-06.

{

385. His discussions with plant personnel were also uninformative on leak rate testing, apparently because he never brought it up. In his view, leak rate testing was an " operator interpretation problem", and his conversations were with management and supervision about hardware-related problems.

l Tr. 2093; see Tr. 2106-07.

386. Not surprisingly, then, none of the NRC's inspections, including the six which were leak rate test specific, yielded the slightest hint that anything was chronically amiss with leak rate tests or the test procedure.

Even Mr. Haverkamp's involvement with the October 18, 1978 leak rate test episode during an inspection was a mere coincidence.

See 1 344, supra. He became aware that some operators misunderstood the circumstances in which they should enter the Action Statement when he overheard personnel discussing the issue. Tr. 2059-61, 2105, 5242. Once he satisfied himself that plant supervision understood and would correct the problem, he took no further action, either to see that the misinterpretation was dispelled, or to see if there might be any other problems with leak rate tests or the test procedure.

Tr. 2113-15, 2143-44, 2177, 2062-64, 2190.

. -213-r

I 387. Instead of leading to a resolution, Mr.

Haverkamp's role in the October 18 incident was the source of the short-lived policy of rounding off test results. The subsequent rescission of that policy may have caused the cancellation of the only arguably clear explanation to the operators and supervisors were supposed to enter into the Action Statement. See 1f 355, supra. It was during his conversation with Mr. Seelinger about the operators' misunderstanding concerning the Action Statement requirement that Mr. Haverkamp initiated a discussion of the technical basis for the 1 gpm limit and whether test results could be rounded off. E.g. Tr. 5249-50. It is clear that Mr. j Haverkamp's comments led to the adoption of the rounding off policy. See 1[9[ 347-48, 354-55, supra. This, in turn, led to Mr. Floyd's Operations Memorandum No. 2-78-19 which incorporated the rounding off practice and, simultaneously, attempted to correct the operators' misunderstanding of when the Action Statement should be implemented. See 1[1f 3 53-55,

, supra. When TMI-2 supervisory personnel took steps several l days later to stop the rounding off practice, that Operations Memorandum was cancelled in its entirety, leaving a void as to I

the correct interpretation of the Action Statement. Tr. 4680, 5004.

388. Mr. Haverkamp's frequent inspections helped W perpetuate the leak rate test problems in another way, i

I

-214-

.I t

Although he would personally review his findings with plant management at the conclusion of each inspection, Tr. 2118, he obviously never raised questions about leak rate testing because he was not aware of any problems. For example, in following up the events of October 18 three months later, he reported that plant practices, including leak rate test practices and the response to LER 78-62/lT, were acceptable:

"For those LERs selected for onsite follow-up (denoted in paragraph 5), the inspector verified that the reporting requirements of technical I specifications and Gp 4703 (Original) had been met, that appropriate corrective action had been I taken, that the event was reviewed by the licensee as required by Technical Specifications, and that continued operatior, of the facility was conducted in conformance with l Technical Specification limits."

Haverkamp prep. St., Ex. E, Encl., p. 11; see Tr. 2117-19. As Mr. Stier succinctly observed of this report, it indicated "everything was all right. In fact, everything was not all right." Tr. 1924. The absence of any notification from the NRC inspector about TMI-2 leak rate testing practices reasonably could have been interpreted by Operations Department personnel that there were no problems with the manner in which they ran leak rate tests. Had Mr. Haverkamp found and pointed I out deficiencies in leak rate testing, the Operations Department could have corrected them. See Stiet Report, Vol.

V(D), Tab 55 (Mr. Floyd's January 25, 1979 instruction to personnel to correct deficiencies that the NRC had pointed out to Met Ed).

I -215-I _

1 5

389. Based on the foregoing, the Board finds that it must agree with Mr. Stier's assessment "that there was a failure on the part of an inspection mechanism and a reporting mechanism, to bring to light the underlying problem that existed in this test and to resolve that problem." Tr. 1924.

The Board also agrees with Mr. Haverkamp that the NRC inspection program has changed for the better since the TMI-2 accident and had today's program been in place then, it likely would have brought the problem to light. As Mr. Haverkamp testified, resident inspectors now

" spend a lot more time and devote a lot more attention to operating practices, observation of surveillance, rather than relying on record I review [,] observation of maintenance practices, again, instead of relying on records. So we are in the plant quite a bit more, I think we are I much more familiar with the day-to-day practices, and I think today's program is much more likely to uncover this type of problem."

Tr. 2141; see also Exhibit 26.

I I

I I

I I

-216-I . --.

I Respectfully submitted, I Of Counsel:

HARRY H. VOIGT LeBOEUF, LAMB, LEIBY & MacRAE By h & l} &

Partner I ROBERT ST. JOHN ROPER MOLLY S. BOAST JAMES W. MOELLER 1333 New Hampsbire Avenue, N.v .

Suite 1100 I MARLENE L. STEIN C. CHRISTOPHER SPRAGUE Washington, DC 20036 (202) 457-7500 SMITH B. GEPHART KILLIAN & GEPHART JANE G. PENNY 216-18 Pine Etreet TERRENCE G. McGOWAN Harrisburg, PA 17108 (717) 232-1851 Attorneys for the Numerous Employees I January 23, 1987 I

I I

, I I

I I

I I

I ,

I APPENDIX A List of Exhibits:

TMI-2 Leak Rate Proceeding Exh. Identified Received No. Desc ription at Tr. Page at Tr. Page 1-A TMI-2 Reactor Coolant Inventory Balance 388, 391/as 388/as Testing, Prepared for GPU Nuclear Corp. modified modified by Edwin H. Stier, September 5, 1985, at 569 at 570

( all volumes.

1-R Review of NRR and OI Reports on TMI-2 388, 391/as 388/as I Reactor Coolant Inventory Balance modified modified

! Testing, Prepared for GPU Nuclear Corp. at 569 at 570 by Edwin H. Stier, August 28, 1986 1-C Letter from Edwin H. Stier to Philip R. 388, 391/as 388/as Clark, dated September 2, 1986 modified modified at 569 at 570 2 Results of Faegre & Benson Investigation 388-89 389 of Allegations of Harold W. Hartman, Jr.

I Concerning Three Mile Island Unit 2, September 17, 1980, all volumes, but excluding Vol. 1, Sec. IV and Vol. 2, Ch. 9.

3 Portion of Page 12 of Report of GPU 389 389 Assessment Panel for Individuals Involved in TMI-2 Leak Rate Testing I (in 1978-1979), January 6, 1986.

(re Mr. Herbein) 4 TMI-2 Computer Log (October 27, 1978). 389 389 5-A Results of NRR's Investigation and I

389/as 389/as Evaluation of Ten Licensed Operators modified modified involved in TMI-2 Preaccident Leak at 569-71 at 571 Rate Testing Irregularities, including I attachments and supporting documents.

(Per the Board's 6/24/86 Order, p. 7, undifferentiated references to

" management" are to be disregarded.)

5-B Memorandum from William Russell to 389/as 389/as Harold Denton, dated September 20, 1985, modified modified with enclosed report. at 569-71 at 571 I

8

I Exh. Identified Received No. Description at Tr. Page at Tr. Page I 6 Office of Investigations Report entitled: Three Mile Island-2:

389-90 390 Investigation of Individual Operator Actions Concerning the Falsification of Leak Rate Test Data, including all attachments.

7 U.S. Nuclear Regulatory Commission, 390 390 Office of Nuclear Reactor Regulation, TMI-l Restart: An Evaluation of the Licensee's Management Integrity as It Affects Restart of Three Mile Island Nuclear Station Unit 1, Docket 50-289 (July 1984) (NUREG-0680, Supp. No. 5, Sec. 4.0.)

8 Memorandum of October 27, 1978 390 Withdrawn 5103 I 9-A Photographs of TMI-2 Control Room thru 9-F 377-381/as renumbered at 388 381/as renumbered at 391 9-G Photograph of the TMI-2 makeup 1182 1183 storage tank level indicator, I pressure indicator and temperature indicator.

10-A Sample SRO license 1025-26 1027 10-0 Sample RO license 1026 1027 I 11-A and 11-B J. Moore Charts, Effect of Oscillations 1025 5104 12-A Administrative Procedure (AP) 1036, 1354-55 5104 Instrument Out-of-Service Control, Rev. O, March 1978 12-B Administrative Procedure (AP) 1036, 1354-55 5104 Instrument Out-of-Service Control, Rev. 1, August 1978 13 Drawing of peak-to-peak v. slope 1845/as 5104 modified at I offset calculation (included in Exhibit 24.) 5103-04 14 Stier cover letter to Board 1862/as 1862 I with two strip chrats showing hydrogen additions (included in Exhibit 24.)

modified at 5103-04 I

Exh. Identified Received No. Description at Tr. Page at Tr. Page I 15-A and Curricula Vitae of Dr. Harrison and Mr. Cole 2008 2008 15-B 16 Training Department Administrative 2219-20 2234 I Memorandum Number 5, Change 2, dated October 8, 1976 (Subject-Category IV CRO Training Program.)

17 Excerpts from Instrument Out of 2400 2400 Service Log I 18 RCIB Test of 2/15/79, portion of CR0 log and MUT Strip Chart 3470 3470 19 TMI-2 Control Room Layout Diagram 3500-01 3501 20 D. Haverkamp prefiled testimony and 4336 4337 attachments 21 Bettenhausen Affidavit 4336 4337 I 22 TMI-2 FSAR Pages 9.3-26 thru 9.3-30 and 5.2-16 thru 5.2-27 4431 4431 23 TMI-2 FSAR Section 5.2.3.4, and 4431 4431 Table 5.2-12 24 Stier letter to Judge Kelley, dated 5104 5221 September 30, 1986, with at tachments 25 MPR letter to Edwin Stier, dated 5104 5221 September 30, 1986 26 November 3, 1986 NRC (Region I) 5266-67 Rejected; Inspection Report on Rochester Gas I

5267; Offer and Electric Corp.'s Ginna Plant of Proof 27 Letter from James Moore to Winthrop -- --

I Rockwell, Octobot 1986 (Admitted subsequent to close of record by Board order dated November I 19, 1986)

I I

APPENDIX B CORRESPONDING NRR AND STIER TEST NUMBERS FOR TMI-2 LEAK RATE SURVEILLANCE TESTS PERFORMED BETWEEN SEPTEMBER 30, 1978 AND MARCH 28, 1979*

STIER TEST NO. NRR TEST NO.

1 156 2 ---

3 155 I 4 5

6 154 153 152 I 7 8

9 151 150 149 10 148 11 147 12 146 13 145 I 14 15 16 144 143 142 I 17 18 19 141 140 139 20 I 21 22 138 137 136 23 135 24 134 1 25 133 26 132 I 27 28 29 131 130 129 30 128 I 31 32 127 126 33 125 I 34 35 36 124 123 122 37 121 38 120

• Tests run prior to September 30, 1978 are not listed because NRR did not analyze them, and thus did not assign them a number. Exhibit 5-A, Enclosure 1, p. 1.

I .- - ..

I ll s11ee res,so. see ress so.

39 119 40 118 41 117 42 116 I 43 44 45 115 114 113 46 112 47 111 1 48 110 49 109 I 50 51 52 108 107 106 53 105 1 54 55 104 103 56 102 57 101 5 58 100 59 99 I 60 61 62 98 97 96 63 95 I 64 65 94 93 66 92 I 67 68 69 91 90 89 I 70 71 72 88 87 86 73 85 I 74 75 84 83 76 82 I 77 78 79 81 80 79 80 78 I, 81 77 82 76 83 75 I 84 74 I

ll '

l .- -- -- - -

l

II l STIER TEST NO. NRR TEST NO.

85 73 I 86 87 88 72 71 70 g

89 69 3 90 68 91 67 92 66 93 65 8 94 64 95 63 I 96 97 98 62 61 60 99 59 1 100 101 58 57 102 E6 I 103 104 105 55 54 53 I 106 107 108 52 51 50 I 109 110 111 49 48 47 112 46 I 113 114 115 45 44 43 I 116 117 118 42 41 40 119 39 I. 120 121 38 37 122 36 I 123 124 125 35 34 33 l

g 126 32 3 127 31 128 30 129 29 130 28 5

I II I - - . -

I l s11em sess me. mee esse me.

131 27 132 26 25 133 24 I 134 135 136 23 22 21 g 137 20 E 138 19 139 18 140 17 I 14 1 142 143 16 15 14 144 13 1 145 12E 146 12D g 147 12C g 148 12B 149 12A 150 12 151 11 1 152 10

'53

. 9 154 8 I 155 7 156 6 157 5 I 158 159 4

3 160 2 161 1 I

1 l1

.I I

!I . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ . . . _ . . _ _ _ _ . . _ _ _ . _ _ _ . _ _ . _ . . _ _ _ _ - _

l E APPENDIX C

\

i I

UNITED STATES DISTRICT COURT FOR THE MIDDLE DISTRICT OF PENNSYLVANIA UNITED STATES OF AMERICA )

m , .

k

v. ) MIT4 TNAL NO.

METROPOLITAN EDISON COMPANY

)

)

FILED wanassauna on I NDICTMENT DONALD EERY, CLERK THE GRAND .7(JRY CHARGES : m /

CE*Jfv 4(an' At all times relevant to this Indictment:

1. The Nuclear Regulatory Commission (NRC) was an agency of the United States of America and was responsible for the I regulation of nuclear power plants in the United States.

2. The NRC administered and enforced the Atomic Energy Act, Title 42, United States Code, SS2G11 et seq., and regulations established thereunder, contained in Title 10, Code of Federal Regulations, Chapter I.

3. The Atomic Energy Act authorized the NRC to issue licenses to operate nuclear power plants and to prohibit the operation of any such plant except under and in accordance with a valid license. .

4. The Atomic Energy Act, Title 42, United States Code, Section 2201(o), authorized the NRC.to require such re p.o r t s , the keeping of such records with respect to, and to provide for such inspections of, activities under licenses issued pursuant to I Section 2133of}itle42, United States Code.

• I '

i .

l I l 1 5. The Atomic Energy Act, Title 42, United States Code, Section 2201(i), authorized the NRC to establish regulations to govern the operation of nuclear power plants, in order to protect health and to minimize danger to life or property.

6. The defendant, Metropolitan Edison Company, was a wholly owned subsidiary of General Public Utilities Corporation. The defendant company was incorporated under the laws of Pennsylvania on July 24, 1922 and became a subsidiary of General Public Utilities Corporation in 1946. Metropolitan Edison Company maintained its headquarters at 2800 Pottsville Pike, Muhlenburg Township, Berks County, Pennsylvania 19650.

7. The defendant, Metropolitan Edison Company, along with Jersey Central Power and Light Company (named herein but not indicted) and Pennsylvania Electric Company (named herein but not indicted), jointly owned the Three Mile Island Nuclear Generating Station consisting of Units 1 and 2, located on Three Mile Island in the Susquehanna River, Londonderry Township, Dauphin County, Pennsylvania, approximately ten miles southeast of Harrisburg, in the Middle District of Pennsylvania. The defendant, Metropolitan Edison Company, owned 50 percent of Units 1 and 2. Jersey Central W Power and Light. Company and Pennsylvania Electric Company each owned 25 percent of Units 1 and 2.

8. On February 8, 1978, the NRC, pursuant to Title 42, United States Code, Section 2133, issued Facility Operating License No. DPR-73 to the defendant, Metropolitan Edison Company, to possess, use and operate the Three Mile Island Nuclear Station, t 2 1

I I

Unit 2, a pressurized water nuclear reactor with associated equipment.

9. In order to obtain an operating license f rom the NRC ,  :

l E the defendant company was required to develop a set of rules by i 1

which it agreed to operate the Three Mile Island Nuclear Station, Unit 2. Such rules, known as technical specifi. cations, established standards for operation. Violation of such standards could result in the shutdown of the facility.

10. As a licensee, the defendant company was subject to the conditions, limitations and procedures contained in the rules and regulations of Title 10, Code of Federal Regulations, Chapter I, and the conditions of the operating license, including the technical specifications incorporated therein.

11. The operation of the Three Mile Island Nuclear Station, Unit 2, required reactor coolant, which was used to: (a) keep the facility at a safe temperature; and (b) transfer heat generated by the nuclear reaction.

12. To avoid the loss of reactor coolant, the technical specifications applicable tc Three Mile Island Nuclear Station, Unit 2, limited the amount of reactor coolant leakage duriag operation to one gallon per minute of leakage from unidentified sources. If unidentified leakage exceeded the established limit, the technical specifications required the defendant company to take certain actions set forth in a procedure called an Action Statement. The Action Statement required the company to reduce the leakage to within allowable limits within four hours of 3

1

I I discovery. If that could not be accomplished, the Action Statement required the defendant company to commence a power reduction so that the facility would be shut down within the following 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br />.

13. Pursuant to Title 10, Code of Federal Regulations, Chapter I, and the provisions of its license, the defendant company was required to establish, implement and maintain appropriate procedures for surveillance tests. Among such tests was the reactor coolant system water inventory balance, also known as a leak rate test, by which the defendant company was required to demonstrate that reactor coolant leakage was within allowable limits. The leak rate test was to be conducted by use of a computer available for such purpose or by manual calculation.

14. If the leak rate test indicated that allowable limits of leakage were exceeded, the defendant company was required to: (a) enter the Action Statement, as described in paragraph 12; or (b) disqualify such test by filing a written Exception or Deficiency, with an accompanying explanation why such test was being disqualified, as required by the defendant company's Administrative Procedure #1010, Technical Specification Surveillance Pro. gram.

15. Absent the filing of an Exception or Deficiency to disqualify the test, each time the 1.eak rate test indicated leakage above allowable limits the defendanc company was required to enter the Action Statement, as described in paragraph 12.

I 4 I

B

16. Pursuant to Title 10, Code of Federal Regulations, Chapter I, and the provisions of its license, the defendant company was required to retain for at least five years all records of surveillance activities required to be conducted, including the leak rate test. In addition, the defendant company was required to maintain all other records not specifically mentioned elsewhere 1

in the regulations or its license until the NRC determined the appropriate dispositi n of those records.

17. Pursuant to the provisions of its license, the defendant l

I company was required to notify the NRC if it failed to enter the Action Statement after receiving information indicating that the reactor coolant system was being operated with leakage above l allowable limits.

l

18. Pursuant to Title 10, Code of Federal Regulations, l Chapter I, and the provisions of its license, the defendant company was required to identify problems encountered with its 1

surveillance testing, including the leak rate test, by making entries relating to such problems on an Exception and Deficiency List, as required by the defendant company's Administrative Procedure $1010, Technical Specification Surveillance Program.

19. Pursuant to Title 10, Code of Federal Regulations, Chapter I, and the provisions of its license, the defendant company was required to make entries in its operating logs relating to the conduct of surveillance testing, such as the leak rate test, including the starting and completion times of such I

I e I

I I testing and the results thereof, as required by the def endant company's Administrative Procedure #1012, Shift Relief and Log Entries.

20. Beginning sometime in 1978, af ter the licensing of Three Mile Island Nuclear Station, Unit 2, the defendant company was on notice that the leak rate test it implemented did not accurately and meaningfully measure the amount of reactor coolant leakage.

21. Notwithstanding its duty to maintain an accurate and meaningful leak rate test, the defendant company continued to use E

a test for measuring reactor coolant leakage that was inaccurate and meaningless.

22. The defendant company engaged in a pattern of criminal conduct designed to permit continued operation of the Three Mile Island Nuclear Station, Unit 2 by:

(a) Taking various actions designed to conceal f rom the NRC serious deficiencies in the leak rate test; (b) Failing to take actions which might have revealed to the NRC serious deficiencies in the leak rate test.

23. The means and methods used by the defendant company to accomplish the objectives outlined in the foregoing paragraphs, are the criminal acts more fully described in the following counts:

I I

I e I

l

.l l

1 COUNT 1 THE GRAND JURY FURTHER CHARGES:

1. Paragraphs 1 through 23 of this Indictment, set forth in 1 pages 1 through 6, are hereby realleged and incorporated as though l restated in full herein.

2. The defendant company's license to possess, use and operate the Three Mile Island Nuclear Station, Unit 2, required it to establish, implement and maintain an accurate and meaningful reactor coolant system water inventory balance procedure to demonstrate that unidentified leakage was within allowable limits.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendar.: company willfully possessed, used and operated a utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of the conditions of the license issued by the NRC; that is: the defendant company was on notice that its procedure for the reactor coolant system water inventory balance did not accurately and meaningfully measure the amount of unidentified reactor coolant leakage. Despite such notice, the defendant company continued to use such inaccurate and meaningless procedure in an effort t;o generate results which appeared to establish that reactor coolant leakage was within allowable limits.

All in violation of Title 42, United States Code, Sections 2272 and 2131.

I 7 I

I COUNT 2 THE GRAND JURY FURTHER CHARGES:

1. Paragraphs 1 through 23 of this Indictment , set forth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. Title 10, Code of Federal Regulations, Chapter I, required the defendant company to establish, implement and maintain an accurate and meaningful reactor coolant system water inventory balance procedure to demonstrate that unidentified leakage was within allowable limits.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of regulations promulgated by the NRC; that is: the defendant company was on notice that its procedure for the reactor coolant system water inventory balance did not accurately and mean:agfully measure the amount of unidentified reactor coolant leakage. Despite such notice, the defendant company continued to use such inaccurate and meaningless procedure in an effort to generate rosults which appeared to establish that reactor coolant leakage was within allowable limits.

All in violation of Title 42, United States Code, Section I-2273 and Title 10, Code of Federal Regulations, Part 50, Appendix B.

8 I

I COUNT 3 THE GRAND JURY FURTHER CHARGES :

1. Paragraphs 1 through 23 of this Indictment, set forth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. The defendant company's license to possess, use and operate the Three Mile Island Nuclear Station, Unit 2, required it to demonstrate that unidentified reactor coolant leakage did not exceed allowable limits by performance of a reactor coolant system water inventory balance procedure.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facilitv. to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of the conditions of the license issued by the NRC; that is: the defendant company intentionally manipulated the reactor coolant leak rdte tests by the addition of water and hydrogen to the reactor coolant system during the course of such tests and in this manner generated test results which appeared to, but did not in fact, fulfill the defendant co mpany 's license requirements.

I All in violation of Title 42, United States Code, Sections

'I 2272 and 2131.

I 9 I -

I COUNT 4 THE GRAND JURY FURTHER CHARGES:

1. Paragraphs 1 through 23 of this Indictment, set forth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. Title 10, Code of Federal Regulations, Chapter I, required the defendant company to demonstrate that unidentified reactor coolant leakage did not exceed allowable limits by performance of a reactor coolant system water inventory balance procedure.

3. Beginning sometime prior to October 18, 1978, the exact l

date being unknown to the Grand Jury, and continuing up to and

)

I including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a i

utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of regulations promulgated by the NRC; that is: the defendant company intentionally manipulated the reactor coolant leak rate tests by the addition of water and hydrogen to the reactor coolant system during the ccurse of such t

l tests and in this manner generated test results which appeared to, but did not in fact, fulfill the defendant company's license requirements.

I All in violation of Title 42, United States Code, Section 2273 and Title 10, Code of Federal Regulations, Part 50, Appendix B.

5 10 5

L F

1 COUNT 5 THE GRAND JURY FURTHER CHARGES:

! 1. Paragraphs 1 through 23 of this Indictment, set forth in pages 1 through 6, are hereby realleged and incorporated as though I

restated in full herein.

2. The defendant company's license to possess, use and operate the Three Mile Island Nuclear Station, Unit 2, required it to enter the Action Statement when the reactor coolant system )

water inventory balance procedure indicated that unidentified reactor coolant leakage exceeded one gallon per minute.

3. Beginning sometime prior to October 18, 1978, the exact e=

date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of the conditions of. the license issued by the NRC; that is: the defendant company failed to enter the Action Statement when the reactor coolant system water inventory balance procedure indicated that reactor coolant leakage from unidentified sources exceeded one gallon per minute.

All in violation of Title 42, United States Code, Sections 2272 and 2131.

11

] COUNT 6 THE GRAND JURY FURTHER CHARGES:

1. Paragraph 1 through 23 of this Indictment, set forth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. The defendant company's license to possess, use and operate the Three Mile Island Nuclear Station, Unit 2, required it to retain records of all surveillance activities required by the Technical Specifications for a period of at least five years. The 1

reactor coolant system water inventory balance procedure was a surveillance activity required by the Technical Specifications.

H

3. Beginning sometime prior to October 18, 1978, the exact V

L date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of the conditions of the license issued by the NRC; that is: the defendant company destroyed, discarded, and failed to retain records relating to the reactor coolant system water inventory balance procedure.

All in violation of Title 42, United States Code, Sections 2272 and 2131.

12

I I

COUNT 7 THE GRAND JURY FURTHER CHARGES:

1. Paragraph 1 through 23 of this Indictment, set forth in I pages 1 through 6, are hereby realleged and incorporated as though l

restated in full herein.

2. Title 10, Code of Federal Regulations, Chapter I, required the defendant company to retain records of all surveillance activities required by the Technical Specifications I for a period of at least five years. The reactor coolant system water inventory balance procedure was a surveillance activity required by the Technical Specifications.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of regulations promulgated by the NRC; that is: the defendant company destroyed, discarded, and failed to retain records relating to the reactor coolant system water inventory balance procedure.

I I All in violation of Title 42, United States Code, Section 2273 and Title 10, Code of Federal Regulations, Section 50.71 and Part 50, Appendix B.

ii I 13 I

I COUNT 8 THE GRAND JURY FURTHER CHARGES:

1. Paragraphs 1 through 23 of this Indictment, set forth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. The defendant compar.j 's license to possess, use and operate the Three Mile Island Nuclear Station, Unit 2, required it to notify the NRC if it f ailed to ent( r the Action Statement after receiving information indicating that the reactor coolant system was being operated with leakage above allowable limits.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facility, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of the conditions of the license issued by the NRC; that is: during the operation of Three Mile Island Nuclear Station, Unit 2, the defendant company failed to notify the NRC that it had not entered the Action Statement after receiving information indicating that reactor coolant system leakage was not,within allowable limits.

I .

All in violation of Title 42, United States Code, Sections 2272 and 2131.

I I "

I

'I COUNT 9 THE GRAND JURY FURTHER CHARGES:

l l 1. Paragraphs 1 through 23 of this Indictment, set forth in l

pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. Title 10, Code of Federal Regulations, Chapter I required the defendant company to make and maintain adequate operating logs relating to the conduct of surveillance testing, such as the reactor coolant system water inventory balance procedure, including the starting and completion times of such testing and the results thereof, as described in defendant company's Administrative Procedure #1012, Shift Relief and Log Entries.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facilitiy, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of regulations promulgated by the NRC; that is: the defendant company failed to make and maintain adequate operating logs relating to the conduct of the reactor coolant system water inventory balance procedure, including the starting and completion times of such testing and the results thereof, as described in defendant company's Administrative Procedure #1012, Shift Relief and Log intries.

All in violation of Title 42, United States Code,Section I 2273 and Title 10, Code of Federal Regulations, Part 50, Appendix 8.

I

L /

? .

l 5

~

l COUNT 10 THE GRAND JURY FURTHER CHARGES:

L i. earagrapns t encougn 23 of this indictment, set earth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. Title 10, Code of Federal Regulations, Chapter I, required the defendant company to identify deficiencies and non-conformances encountered with its surveillance testing activities, including the reactor coolant system water inventory balance procedure, by making entries relating to such problems on an Exception and Deficiency list, as described in defendant company's Administrative Procedure #1010, Technical Specification Surveillance Program.

3. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company willfully possessed, used and operated a utilization facilitiy, to wit, the Three Mile Island Nuclear Station, Unit 2, in violation of regulations promulgated by the NRC; that is: the defendant company f ailed to make entries identifying deficiencies and non-conformances encountered with its reactor coolant system water inventory balance procedure on an Exception and Deficiency list, as described in defendant company's Administrative Procedure #1010, Technical Specification Surveillance Program.

All in violation of Title 42, United States Code, Section 2273 and Title 10, Code of Federal Regulations, Part 50, I Appendix B.

16

COUNT 11 THE GRAND JURY FURTHER CHARGES:

1. Paragraphs 1 through 23 of this Indictment, set forth in pages 1 through 6, are hereby realleged and incorporated as though restated in full herein.

2. Beginning sometime prior to October 18, 1978, the exact date being unknown to the Grand Jury, and continuing up to and including March 28, 1979, in the Middle District of Pennsylvania, the defendant company knowingly and willfully falsified, concealed, and covered up by trick, scheme, and device, from the Nuclear Regulatory Commission, an agency of the United States, a material fact, to wit, that the reactor coolant system water inventory balance procedure did not accurately measure the amount of reactor coolant leakage from unidentified sources at Three Mile Island Nuclear Station, Unit 2.

i I In violation of Title 18, United States Code, Section 1001.

a.u. L. d .uu a Ma LO FOREPERSON November 7 , 1983 I '

.n_-,_n) L'l DAVID DART QUEEN g I United States Attorney i ,,

I

APPENDIX D

'I i 1

2 IN THE UNITED STATES DISTRICT COURT FOR THE MIDDLE DISTRICT OF PENNSYLVANIA I 3 4

UNITED STATES OF AMERICA  :

5 v. Criminal No. 83-00188 ,

METROPOLITAN EDISON COMPANY,  :

'I 6 7

Defendant  :

8 l

9 TRANSCRIPT OF PROCEEDINGS CHANG E OF PLEA AND SENTENCING 10 i BEFORE: HON. SYLVIA H. RAMBO, J.

!E W 11 DATE  : February 28 and 29, 1984 12 PLACE : Courtroom Number Two 13 Federal Building Harrisburg, Pennsylvania 14 15 COUNSEL PRESENT:

16 DAVID DART QUEEN, United States Attorney JAMES WEST, Assistant United States Attorney 17 SALLY A. LIED, Assistant United States Attorney I 18 STANLEY ROTHSTEIN, Assistant United States Attorney For - United States of America 19 PAUL J. CURRAN, Esquire I DAVID KLINGSBEPG , Esquire 20 PAUL H. RHOADS, Esquire For - Defendant Metropolitan Edison Company I 21 22 5 23 Vicki L. Fox, RPR Official Repoc e:

24 25 I

v:CKl L. FOX. QFFTC!AL COURT REPORTER. 801 383. FEDERAL su L3:No. saggi$gygg,pa ,7tes I

o -

65 i l

I I

]

1 There was observation -- and I think if I 2 understood Mr. Curran correctly he mace the assertion that 3 the one million dollar payment uas che enthusiastic anc I 4 heartfelt offering of the company and hadn't a thing ec co 5 with these proceedings. That was as a result of a negotiated 6 settlement when the United states Attorney's office.-- I am 7 the United States Attorney -- when I told him that under no 8 circumstances would we accept noic pleas across the board; 9 there had to be a guilty plea to a significant count. That 10 is what we demanded. Count 2 is such a count.

11 If one looks at the language contained in it, it 12 goes to the very heart of the leak rate test, the primary j 13 method of measuring leakage frem unidentified sources. And 14 what it says in there and what I believe through all of the 15 remarks made through Mr. Curran still nevertheless remains 16 undenied is that that leak rate test didn't work; it was the 17 primary method of measuring leakage from unidentified 18 sources; supervisory and louer level employes at Unit 2 knew '

19 it, and in the face of that they continued to do really in 20 essence all else that follows in the indictment.

21 I believe that a plea of guilty to Ccunt 2 --

22 which is the really pivotal issue, far more so than the nolo 23 contenderes -- really sets the stage for all of the conduct 24 that is contained in the counts that follou.

25 Specifically, whether one is jogging water, adding 7

.g t -

E vlCKl L. Fox. OFFICIAL COURT REPCRTER. 80X 983. FEDERAL SulLDING. HARRISBURG. PA t7t08 I

- -

• 66

/-

~

1 hydrogen, destroying leak rate tests, all of that becomes

]

2 relevant and necessary for the Defendant to perform only if 3 there is a leak rate test that doesn't uork. Eecause if that 4 leak rate test did work, then all of the chenanigans 5 contained in the balance of the indictment would have been 6 utterly unnecessary. To that extent, I think accepting the 7 plea of guilty to Count 2 is accepting a plea to a meaningful 8 count -- an e::tremely meaningful count.

9 I want to address one additional observation.

l 10 11 There were a numoer of self-serving observations made in che Defendant's response. One is basically a fairly granc 12 promise to do in the future uhat in my judgement the company l I

should have been doing the whole time it was operational in j I

}

.x 13 14 197 0 and 197 9; that is to follou the technical l

15 specifications, follou the regulations and operate the plant i

I 16 the way Mr. Curran now says in the light of all that has been:

I l

17 learned they promise to operate it in the future.

la Ironically what Mr. Curran is really saying I .

l 19 guess for public consumption is no more than what those 20 Unit-2 employes said to the URC inspector on the 18th of 1

21 October -- ue promico to be doing cur job properly. .ind for 22 all the weeks and months that followed, they didn't. That is 23 an undeniable and undi puted fact; that they persisted in the 24 dectruction of leak rate tests only if thcce tests showed 25 excess leakage.

l1 h

I i

I VICKI L. FOX. OFFICIAL COURT REPORTER. 801 983. FEDERAL SutLD!NG. H ARRl58URG. P4 17108 I

o.,

I

.- 1 I 'i' ~ 1 MTER RECESS 2 7:IO COUNT: lIr. Cucen anc Ur. Curran, I regrac the 3 Jact tha: the cocicion I make may croata sc a difficulticc 4 ror councel. Thic Court did not receive any of the cocuments 5 in this case until after five o' clock last night to I was not 6 aware of what the contents of this hearing was going to i:e.

7 Ue have just now had the oral argumentc, anti I do 8 feel it vould not be in the boat interests of ju.:tico to take 9 a decicion pecmptly at thic time. I would like coveral more 10 hours, and I therefora vill hopefully decernine the outconc 11 on this tocorreu at nino-thirty.

12 U R . CU E Cli: If you don' t mind seeing =c in the 13 same suit, I will bo here.

I 14  !*R. CURRAU: Here, Your Honor?

15 THE CCURT: This courtrocs. Ccurt is aC 3 curned.

16 (Uhereupon, the prococdingc were concluded.)

17 l

10 19 20 21 22 23 24 25 5 -l sJ vlCXI L. FOX. OFFICIAL COURT REPORTER. SCX 963. FEDERAL SUILOING. H ARRl58URG. P A 17108 I

~

75 I .

- 1 FCSEUARY 29, 1984 I

THE CCUnT: Good corning, everyone.

2 3  !!R. GUEC::: Good corning.

4 i:R. CUnr.::: Good corning.

5 THE COURT: I have come questions that nceu 6 cicrification. Uith ragsrc to paragraph ton of the propocou I 7 agreccone, it appear: that the cmcrgency preparecnc s conc l

8 cpecified by ;nc whien 20:1 uculd so conitoring really 9 depence upon thu continucc cnictcnce and operation et the T".I 10 facility; is that correct?

11 F.y inquiry i for s.urscces of argument sc auu:.c.y I 12 that the plant ncycr cponed fer c':cration or did net continuei i

13 to axict, what uculd happen :c the aillion dollars? ,

I.i

\-

Ycur Coner, it is ny understancing 14  ::R. CU3Ct!:

15 that there wculd continue ac a reuuic of the unccupleted 1G clean-up activity of the damaged reactor Unit-2 to be a 17 viabic need for emergency preparcano:s anc cvacuaticn within l 10 the twenty-nile zone. If Unit-1 10 net licenced for rectart l 19 by the tiucicar neguistory Cc==icsion, it beco=ca secewhat i

E 20 probiccatic as to whether or not there iu a c;illion dollars 21 uor h of evacuation pisaning curroundin<j Qe clean-up 22 cituation.

23 It is very difficult to precict hcu tuch it uault 24 cost, cut I thi..k the Court coula safely conclude that

_ 25 uncther or not Unit-1 ic restarted, there aces con inue to ce!

i viCKt L. FOX. OFFICIAL COURT REPORTER, 80X 983. FEDERAL BUILDING. H ARR158URG. P A 17108 I -

-. - 79 w .

El ~

which I had and which I used yesterday. Tne facts or the

[ 1 2 statements chat it contains were made a =atter of recorc by

{ 3 me yesterday.

4 I have absolutely no objection to the document 5 itself beine a matter of record. I do not have a copy of it.

( 6 MR. QUEEti: Your Honor, I realice this doesn't 7 directly address your point, but I do think the record should 3 reflect that until you just mentioned thi: document 9 circulating in the hallway, I was completely unaware of it 10 both in terms of the nature of the contents or the fact it

, 11 was going to be circulated.

12 Obviously from the standpoint of the government, j 13 we did not yesterday and do not now embrace the accuracy of

[ 14 everything said in whatever might be circulating through the 15 halluays of this Courthouse.

16 THE COURT: I just want it clear that this Court 17 has not had a copy if it, and that whatever it contains, I 18 don't know. I only read the first part -- the first sheet.

[ 19 It was obtained by one of my staff from one of the guards 20 downstairs.

21 For the record, I would like to state the 22 following: In reaching a decision on this matter, this Court -

{

23 has been recuired to determine what better result in the

( 24 overall administration of justice could be achieved by an 25 adjudication of guilt on all eleven counts af ter a trial as

[-

VICKI L. FOX. OFFICIAL COURT REPORTER. BOX 983. FEDERAL SUILDING. HARRf58URG. P4 17108 s

E - - - - - - - - -

00 In 1 cpposed to accepting the pleas as tendered.

1 2 The public is under the m.tsconception that a 3 finding of guilt on these eleven charges would ampouer this 4 Court to determine issues affecting management and the 5 continued operation of the TMI facility. It must be made 6 clear that the province of this Court is to ensure the properi 7 conduct of a trial, and it is the function of the jury to  !

determine if there have been violations of the United States !

l 8

I 9 Criminal Code and nuclear license requirements.

i r

10 It is not the function of this Court or any jury ,

11 to draw conclusions as to the integrity and competence of 12 management. Even after a trial and conviction on the issues .

13 in the indictment, this Court could only assess penalties I] 14 provided by lau, that is the ma::imum $85,000.00 penalty.

15 Issues affecting management and/or the continued operatica of 16 TUI remain with the regulating agency, I 17 There are many who believe that a full trial in i

18 this case will establish once and for all what the facts were 19 surrounding the incident of March 23, 19.'9. In just this 20 case alone, this Court has cefore it numerous motions and 21 briefs consisting of more than one thousand pages. With the 22 knowledge gleaned from these, the Court is of the opinion 23 that the facts surrounding the issues in this indictment uill 24 in all likelihood continue to be clouded by the contradictory 25 assertions of the scientific e:cperts. Even after a trial in i i

l J i I(. VICXI L. For. 0FrtCIAL COURT REPCMER. 80X 983. FEDERAL SUILDING. HARRIS 8URG. P4 17108 l

I

L, . 81 e

L

/\

[

~

1 this matter, it is highly doubtful that che issues will be 2 much c'learer or that convincing conclusions will be reacned.

e l

H 3 I believe that the count on unich the Defendan:

1 p 4 has tendered a plea of guilty -- Count 2 of the indictment -- l 5 is both a meaningful and significant admission. In fact, I 6 agree with the government that Count 2 is one of the pivotal 7 counts in the indictment.

8 Furthermore, the $1,045,000.00 non-deducti.ble 9 penalty payment agreed to by the Defendant should act as a 10 substantial deterrent.  !! ore importantly, one million of

. 11 those dollars will benefit the public, especially those 12 persons residing in the impacted area. By way of contrast,

'] 13 the maximum penalty which could.be imposed on the-Defendant 14 after an adjudication of guilt on all eleven counts following 15 a trial would be $85,000.00 -- none of which would go to the 16 direct benefit of the citizens negatively impacted by the 17 Defendant's conduct.

18 Still another consideration in my determination of 19 whether or not to accept this plea is the cost in time and 20 money that this highly complex and sure to be protracted 21 trial would exact. It is my opinion that these costs of t i.- .e 22 and money are not likely to be counterbalanced by the 23 penalties which would follow conviction after trial.

24 Variou other arguments and assercions have been 25 made by counsel in support of their request that I accept In -

VICKI L. FCI. OFFICIAL COURT REPORTER. 80X 983. FEDERAL BUILDING. HARRISBURG. PA I7108

o - a u

~ _ ..

s l's 1 this plaa. I reject many of those argument.3.  :!owever , I i l

2 believe that the three reasons cited by the Court, taken 3 togacher, are compelling enouga for this Court to accepn the 4 plea agreement.

5 Accordingly, the Court will enter the f ollowing-6 order: AND UCW, this 29th day of February, 1934, I find that 7 the Defendant Corporation is acting voluntarily and not as a 3 result of force or threats or promises apart from the , lea 9 agreement. The Court further finds that the Defendant I i i

understands its rights and the coacequences of its plea and 10 g

l 11 voluntarily waives its right to trial.

i 12 The Court is satisfied that the plea of guilty to l

l 13 Count 2 has a basis in fact and contains all of the elements 14 of the crime charged. The Court, therefore, permits the l I

15 revocation of the not guilty pleas and accepts a plea of

} 16 guilty to Count 2 and pleas of nolo contendere to Counts 4,  ;

17 5, 7, 8, 9 and 10. l i

18 I do hereby direct the entry of a judgment of I

l 1

19 quilty on the plea on Count 2 and j udgment of conviction on i

20 Counts 4, 5, 7, 8, 9 and 10. j I 21 It is customary on individual Defendant cases for 22 the Court to order a presentence investigation. However, the 23 same may be waived by the Defendant.

24 Uhat is the intention of the Defendant?

I, 25 MR. CURPJdI: It is waived, Your Honor.'

Ne intend l j(l

- i I

VICKI L. FOX. CFFICIAL COURT REPORTER. 80x 983. FEDERAL BUILOING. H ARRISBURG. P A 17108