NUREG-0680, Applicant Exhibit A-7,consisting of NUREG-0680,Section 4.0, TMI-1 Leak Rate Falsification
| ML20237H085 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 09/08/1986 |
| From: | NRC |
| To: | |
| References | |
| RTR-NUREG-0680, RTR-NUREG-680 LRP-A-007, LRP-A-7, NUDOCS 8708170048 | |
| Download: ML20237H085 (20) | |
Text
NUCLEAR MGULATO COMM!$$t0N Dockot No. ---
Offi No.
in the matter of NM il Staff IDthfinED V
W Appficant RECENED
'87 E 12 AM :10
' * = =
"D Cont's Offr r_
Contractor DATI.
4.0 THI-I LEAX RATE FALSIFICATION othn mim: Wu#
4.1 Background
88 Pat" In July 1983, Re coolant system (gion I commenced <a special, unannounced inspection of reactor RCS) leak rate test procedures, equipment, and records at TMI-1. The basic objective of the inspection was to determine if the records of RCS leak rate surveillance testing at TMI-1, for the period April 1,1978, through March 31, 1979, showed any indications of practices similar to alleged irregularities at TMI-2. The allegations related to THI-2 RCS leak rate test-ing involved the following:
(1) Tests were often repeated until the resul.ts met the acceptance criteria of the Technical Specifications (TS); the unfavorable results were discarded.
(2) RCS water inventory was adjusted during the leak test period. Contrary to procedural requirements, these water additions were neither recorded in the Control Room Operator's Log or the Shift Foreman's Log nor were they factored into the RCS leak rate test calculations.
(3) Hydrogen gas was added to the makeup tank (MUT) in such a manner as to influence leak rate test calculations in a favorable way.
(4) Computer data entries for the leak rate calculations were altered to make the leak. rate test calculations appear acceptable.
01 LR-1 Ex. I at 5.
The results of the special inspection are documented in Inspection Report 50-289/83-20 dated September 21, 1983, and its subsequent revision dated February 24, 1984. From the 645 test records examined the revised inspection report identified 13 instances of water additions and 11 instances of hydrogen additions to the RCS MUT during the performance of PCS leak rate surveillance tests without these additions being properly accounted for in the test calcula-tions. Additionally,13 feed-and-bleed operations were identified as occurring during the performance of leak rate tests for which the leak rate calculations were not properly corrected for these evolutions. One instance of a combined hydrogen addition, water addition, and feed-and-bleed operation was found.
These 38 instances amount to 5.9% of the surveillance tests examined. After these evolutions were considered, recalculation. by the inspection team showed, with the exception of three instances, that the TS acceptance criteria would have been met even if these evolutions had not occurred.* OI LR-1 Ex.17 at 7.
- The revised inspection report concluded four instances would not have met TS limits; however, as discussed in Section 4.2.1.3, the test beginning at 0047 on May 12, 1978, would not have. exceeded 1 gpm and, therefore, was acceptable.
4 NUREG-0680 4-1
' PDR. ADDCK 05000320 0
PDR h_
_._.,.m-_-.
s On the basis of the special inspection findings," T. E. Murity, Regional Admin-istrator, Region I, requested on September 21, 1983, that the Office of Inves-tigations (01) conduct an investigation into suspected falsification of RCS leak rate test data at TMI-1. Murley requested that the investigation determine if there was a systematic pattern of falsification of RCS leak rate surveillance test data by control room operators at TMI-1 and, if so, to what extent licens-ee's management was cognizant of and/or involved in such activity. 01 LR-1 at 1, 4.
The ! investigation was completed by the 01 Field Office, Region I, and two reports dated April 16, 1984 (Investigation Report 1-83-028), were submitted.
The first report was submitted as pending; the second report (supplemental) completed all investigative leads [ Memorandum from B. B. Hayes (01) to T. E. Murley (Region I) dated April 16,1984]. A discussion of the results j
of the investigation is presented below.
4.2 Investigation Results The principal conclusions of the investigation may be summarized as follows:
(1) The percentage of questionable leak rate test results was small in com-parison with the total number of leak rate tests examined (38 out of 645 or5.9%). NRC recalculation of these tests concluded that the TS accept-ance criteria would have been met, except for three tests, even if the unaccounted-for evolutions had not been made. OI LR-1 at 1.
(2) The investigation did not. disclose conclusive evidence to indicate that any TMf-1 licensed operator deliberately falsified RCS leak rate surveil-lance test results (0! LR-1 at 2; OI LR-1 Supp. at 1-2).
(3) The investigation did not identify a systematic pattern of falsification of THI-1 RCS leak rate surveillance tests, although the 11 instances of hydrogen cdditions during leak rate tests occurred only on 2 of the 6 operator shifts and was limited to 5 control room operators (OI LR-1 at2).
(4) It was common practice in the preaccident period for control room per-sonnel to discard test results that were deemed invalid (OI LR-1 at 2).
(5) No apparent motive existed for the operator's to attempt to alter leak rate test results by making unaccounted-for water or hydrogen additions (OI LR-1 at 1).
(6) No evidence was developed to indicate that any licensed operator was either directed or pressured by supervisory or management personnel to manipulate RCS leak rate surveillance tests (0I LR-1 at 2; OI LR-1 Supp, at1-2).
(7) A significant amount of infonnation was available to plant management at THI-1 regarding the effect of hydrogen additions on RCS MUT Tevel indi-cation (specifically, preaccident TMI-1 plant maintenance work requests and the September'1980 Faegre & Benson Investigation Report on TMI-2).
Despite this information, the root cause of this effect (the existence NUREG-0680 4-2
of'a loop seal in the MUT level instrument reference leg), was not dis-covered until the Region I special inspection in the summer of 1983.
Memorandum from 8. 8. Hayes to T. E. Murley, April 16, 1984.
A detailed discussion of each of these conclusions and their supporting bases is presented in the following sections.
Section 4.2.1 addresses conclusions 1, 2, and 3 and Sections 4.2.2 thrown 4.2.5 cover conclusions 4 through 7 respectively.
4.2.1 Whether Leak Rate Tests Were Deliberately Falsified On the basis of the relatively small percentage of questionable test results identified by the Region I special inspection, the investigation attempted to determine whether or not there was a consistent pattern to the questionable test results and whether a motive existed for the operators and their super-visors to attempt to manipulate leak rate test results (memorandum from B. B. Hayes to T. E. Murley, April 16,1984).
As part of the investigation, all preaccident and current THI-1 control room operators (CR0s), shift foremen, and shift supervisors, who actually conducted leak rate surveillance tests during the period under investigation, were inter-viewed under oath.
In addition, a large number of preaccident and current staff personnel (operational, engineering,' instrumentation and control (I&C),
and maintenance) and site and corporate management officials were also inter-l viewed (0I LR-1 Ex.33-112; 0! LR-1 Supp. Ex. 1-12).*
Every individual interviewed testified that they were not aware of any willful attempt on the part of any operator to alter RCS leak rate surveillance tests at TMI-1 by making unaccounted-for hydrogen or water additions to the MUT or by any other method. Every individual interviewed also testified, that they were never involved in, condoned, or were aware of any systematic pattern of falsification of leak rate tests. O! LR-1 at 17-18; 01 LR-1 Supp. at 1.
As discua. sed in Section 4.2.3 under motive, there were a number of differences that existed between TMI-1 and THI-2 in the RCS leak rate test area that re-sulted in significantly less motivation for test falsification on Unit 1.
These included a more accurate measurement capability, less restrictive TS leakage requirements, and lower leak rates at Unit 1 (OI LR-1 Ex. 21 at 11).
The Unit 1 Technical Specifications required, in part, that if unidentified RCS leakage (excluding nonnal evaporative losses) exceeded I gpm, the reactor should be placed in hot shutdown within 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of detection (OI LR-1 Ex. 2 at3-12). The ] gpm value is a typical specification for pressurized water reactors (PWRs). The Technical Specifications also required that RCS leakage should be evaluated daily when RCS temperature was greater than 525'F (OI LR 1 Ex. 3 at 4-8).
The procedure for conducting RCS leak rate tests was contained in Surveillance Procedure (SP) 1303-1.1 (OI LR1 Ex. 4). The plant computer was normally used to compute RCS leak rates. The computer would calculate leak rates on the basis of a 1-to-8-hour time interval. The majority of operator testimony indicated that a leak rate test interval of I hour was nonna11y
- Exhibits. to the OI Supplemental Report are so indicated by dSupp. Ex."
NUREG-0680 4-3
l selected (e.g., 01 LR-1 Ex. 36 at 5; Ex. 38 at 5). All but two of the ques-tionable leak rate tests evaluated in the Region I inspection report were con-ducted with a 1-hour test duration. Provided the plant is s intained in a steady-state condition for the entire test period, the error inherent in the leak rate calculation is reduced as the test duration is extended. 01 LR-1 Ex. 17 at 26. During the time interval that the test was being run, the pro-cedure cautioned the operators to avoid evolutions such as chemical additions
(
and inventory changes to the RCS. However, if changes were made to the RCS inventory during the test, the known values of the changes were to be provided i
as additional input to the computer at the end of the test, before the final leak rate calculations were performed by the computer (0I LR-1 Ex. 4 at 3.0).
There were nonnally two licensed control room operators (CR0s) and one shift foreman [ senior reactor operator (SRO)] on watch at each unit.
In addition, a shift supervisor (SR0 licensed on both units) was also assigned to each shift and split his time between TMI-1 and TMI-2. One of the CR0s was desig-nated the " panel" or " console" operator and the other was assigned duties as the " switching and tagging" operator. The " switching and tagging" operator would normally be the operator who would interface with the plant computer to conduct the leak rate test. The operator at the panel was the individual who would control all operational evolutions, including water or hydrogen additions to the RCS MUT and feed-and-bleed operatiens.
E.g., OI LR-1 Ex. 35 at 6.
l Because the operator running the leak rate test was not the same operator at j
the console, it was necessary for the operator running the ' test to inform the i
console operator that the test was in progress and for him to avoid making any unnecessary changes to critical plant parameters (e.g., 01 LR-1 Ex. 38 at 5).
From testimony provided by operators, it is apparent that at times the oper-ator conducting the test either forgot to tell the console operator that the j
test was in progress or that the console operator forgot the test was in pro-gress and altered the steady-state condition of the plant in such a manner that the results of the leak rate test were affected without the knowledge of the operator conducting the test (e.g., Of LR-1 Ex. 36 at 8). In most cases, when this was detected, the operators would " invalidate" the test and start the leak rate test over again.
(The discarding of invalid leak rate tests is discussed in Section 4.2.2.)
The preponderance of sworn testimony by the oper-ators indicated that when the plant was maintained in a steady-state condition, with no operator-induced changes, there was little problem getting valid leak rate test results that met the acceptance criteria of the Technical Specifica-tions (e.g., O! LR-1 Ex. 39 at 10; Ex. 42 at 6; Ex. 44 at 7).
While the Technical Specifications required that RCS leak rate be determined every 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, actual leak rate tests were conservatively run once per shift (every 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />) as a matter of routine. Therefore, even if the operators ex-perienced some difficulty in obtaining a valid leak rate test on their shift because of plant conditions, they did not feel pressured by management or supervisory personnel to obtain a valid leak rate test because it could be performed on the next shift (e.g., 01 LR-1 Ex. 41 at 18).
~
During the course of the interviews, operators were questioned at length regarding the three principal methods by which leak rate test results were
{
NUREG-0680 4-4
i called into question (that is, unaccounted-for additions of water and hydro-gen to the RCS MUT, and feed-and-bleed operations). Each of these is dis-cussed below.
~4.2.1.1 Hydrogen Additions The addition of hydrogen to the RCS MUT was required periodically to limit
{
the oxygen content in th; RCS and to provide an increased net positive suc-tion head for the RCS makeup pumps. The addition of hydrogen, theoretically, should not have' affected MUT indicated level; however, because of the config-uration and environment of the MUT level detection instrumentation system at
~
THI-1, water condensed and collected in the low-pressure reference, leg. Under this condition, hydrogen additions to the MUT could cause a temporary increase in the indicated MUT level without actually adding water to the tank. There-fore, the addition of hydrogen at the appropriate time (after the initial data collection and before the final data readings) could affect the leak rate results in a nonconservative manner (that is, the calculated leak rate would be less than the actual leak rate) (0I LR-1 Ex. I at 27-28).
The RCS leak rate test procedure, Sp 1303-1.1, prohibits the addition of chemicals during the test, although the procedure did not specifically iden-tify hydrogen as a chemical. Some of the operators interviewed stated that they did not consider the addition of hydrogen to the MUT as a chemical addi-tion prohibited by the procedure (e.g., OI LR-1 Ex. 35 at 39).
A review of the 12 hydrogen additions (including the combined hydrogen addi-tion, water addition, and feed-and-bleed operation on January 6,1979) dis-closed that these additions were made only during two shifts, "A" and "D."
Five CR0s were identified as being involved in these additions and five shift foremen provided approval signatures on the associated RCS leak rate surveil-I lance test calculation documents. OI LR-1 at 18. Four of the five CR0s test-ified they were not aware of the effect of hydrogen additions on leak rate tests at the time. One CR0 was aware of the effect (0I LR-1 at 18). Three j
of the five shift foremen testified that they were not aware of the effect of i
hydrogen additions on leak rate tests at the time. Two shift foremen were aware of the effect (0I LR-1 at 18-19; OI LR-1 Supp. Ex.12 at 30).
A technical evaluation of the THI-1 hydrogen additions was performed by IE Headquarters (OI LR-1 Ex. 21 at 2). Any hydrogen addition while the leak rate test was in progress, had potential to affect the leak rate calculations.
However, this evaluation showed that to be most effective in reducing the indicated leak rate, the hydrogen additions would need to be made during the latter half of the leak rate test. The majority of the additions evaluated were not made near the end of the test. Those that were done near the end of the test were accomplished during a period when the actual leak rates were quite low; therefore, there would have been no reason to attempt to reduce the test results.
In one case, the hydrogen was added right before the test began and any error caused by adding hydrogen at that' time would tend to result in an increase in calculated leak rate. OI LR-1 Ex. 21 at 2-3.
The possibility that the hydrogen additions were made at random (as a result of normal plant operations) was considered by IE.
In reviewing copies of the strip charts for MUT level indication provided in the Region I inspection l
NUREG-0680 4-5 l
l
e I
i-l i
I report, IE identified three other hydrogen additions of similar short dura-tion during periods where no RCS leak rate tests were being performed.* These l
additions would tend to support the claim of operators during OI interviews that similar hydrogen additions were made for legitimate operational reasons during normal plant operations (0I LR-1 Ex. 21 at 4).
It is not clear from operator testimony that the addition of hydrogen could always be accomplished from the control room alone during the period under investigation. Some testimony indicated that for safety reasons the hydrogen bottles were left isolated (valved-out) locally.
In these cases an auxiliary operator (AO) would be sent to open the local valve first when the operator in the control. room wished to add hydrogen. Because of this uncertainty l
regarding whether hydrogen was available to be added from the control room alone, some operators testified that they would check the valve lineup by opening the hydrogen addition valve from the control room and immediately shutting it again to see the effect on MUT pressure (e.g., OI LR-1 Ex. 35 at36). Thus, this " testing of the valve lineup" provioes a partial explana-tion of why only small amounts of hydrogen were added to the MUT during some of the leak rate tests.
Log entries were not normally made in the CR0's Log for hydrogen additions (e.g., 01 LR-1 Ex. 76 at 54). Therefore, if the console operator inadvertently added hydrogen to the MUT during the performance of a leak rate test and for-got to inform the operator performing the test, there would be no way for either the operator performing the test or the shift foremen reviewing the test results to know that such an error had been made.
In summary, there were legitimate operational reasons why hydrogen ~was added to the RCS MUT periodically. The frequency of addition varied with plant l
. conditions.
i Hydrogen additions were not routinely logged in the CR0's Log.
Miscommunication between the operator conducting the leak rate test and the operator at the console could lead to inadvertent hydrogen additions being made during the performance of leak rate tests.
1 A detailed review, by Region I of the MUT strip charts of 645 surveillance tests conducted during l
?
the period under investigation, identified only 12 occasions where hydrogen was added to the MUT during the perfonnance of leak rate tests.
While only 5 CR0s and 5 shift foremen were involved in the 12 hydrogen additions, 4 of the CR0s and 3 of the shift foremen testified at the time that they were not i
aware that hydrogen additions could impact leak rate test results. All 10 of the operators testified that they never added hydrogen to intentionally alter the results of leak rate tests.
In addition, the IE evaluation of the 12 hydrogen additions concluded that none of the additions would have affected leak rates in such a way that if the additions were not made, the limits for RCS leakage would have been exceeded.
On the basis of the infonnation pre-sented in the investigation, the staff finds that the evidence does not sup-port a conclusion that the operators at THI-1 intentionally added hydrogen to affect the results of RCS leak rate surveillance tests.
- A telephone conversation between NRR staff and Region I inspectors on June 6, 1984, confirmed that the MUT level indication strip charts for the l
1-year period under investigation show numerous hydrogen additions of short 1
duration during periods when no leak rate tests were in progress.
l NUREG-0680 4-6
)
1 i
4.2.1.2 Water Additions When water is'added to the MUT during a RCS leak rate test (after the initial data set is taken, but before the final readings are taken), there will ~always f
be a reduction in the calculated ' leak rate unless the addition is properly 1
acnounted for in the calculations. The leak rate surveillance procedure, j
SP 1303-1.1, cautioned the operators to avoid the addition and removal of.
i water from the RCS during the test; however, the procedure provided a data sheet for accounting for operator-caused changes to RCS inventory (0! LR-1 3
Ex. 4 at 1.0 and 18.0). However, because of an error in the calc 91stional procedure, the program'did not account for water expansio'n as tha cold water l
from the MUT (125'F) heated up in the RCS (579'F); consequently, even a cor-j rectly entered water addition would result in an error (OI LR-1 Ex.1 at 37).
Administrative Procedure 1012 " Shift Relief and Log Entries," Section 3.3.11, required that log entries be made for the addition of boron or dilution of RCS
(
i boron concentration (0I LR-1 Ex. 7 at 7,0). Therefore,.during the time period
/
i under investigation, water additions to the MUT should have been recorded in l
the CR0's Log.
By reviewing the MUT strip chart recorder traces of 645 surveillance tests performed during the period under investigation, the revised Region I inspec-
)
i l
tion report identified 14 water additions to the MUT (this includes the com-t bined hydrogen addition, water addition, and feed-and-bleed operation on l
January 6,1979) that were made during periods in which RCS leak rate surveil-J 1ance tests were in progress. Ten of the additions were not logged in either the CR0's or the Shift Foreman's Logs. Thirteen of the 14 additions were not t
factored into the leak rate calculations. The one addition that was included l'
in the calculation did not account for.the correct amount of water added. OI i
LR-1 Ex. 17 at 39. Both the revised Region I inspection report and the IE review of the Region I inspection report identified some additions where the water appeared to be added gradually (jogged) in several small steps, a proce-4 dure that could. indicate an attempt to hide the addition (OI LR-1 Ex. 17'at 38; Ex. 21 at 4).
However, a review of the 14 water additions did not disclose any visible pattern regarding either a particular operator shift, CR0s, or shift foremen (OILR-1at19). A total of 12 different CR0s and 7 different shift foremen I
wereinvolvedintheadditions(OILR-1Ex.19).
i Recalculation of the leak rate. data, taking into account these water addi-tions, identified only three instances where the leak rate test results were 1
affected in such a manner that, had these additions not been made, the cal-i culated leak rate would have exceeded the TS acceptance criteria for unidenti-fied leakage (OI LR-1 Ex.17 at 38)..These three additions were made by three different CR0s approximately 1 month apart (April 30, 1978; May 26, 1978; and-June'21,1978). The leak rate surveillance tests that were performed during these additions were reviewed and approved by three different shift foremen, The 19 operators involved, either as CR0s or shift foremen, in the 14 water additions, along with other operational personnel who were interviewed, denied any knowledge of, or involvement in, any deliberate attempt to alter RCS leak rate surveillance test results by unrecorded and unaccounted-for water addi-tions to the MUT. The interviewees did admit that water additions could have NUREG-0680 4-7
R c
I l
been inadvertently made during the performance of RCS leak rate tests by the console operator failing"to notify the operator conducting the surveillance test that water had been added. Additionally, the interviewees denied grad-ually adding water during leak rate tests. None of the operational personnel interviewed were aware of any incident in which unaccounted-for water addi-tions were knowingly made and the leak rate surveillance tests were accepted as valid (0! LR-1 at 20).
In summary, out of the 645 leak rate surveillance tests reviewed,14 showed evidence of water additions being made to the MUT during periods when leak rate tests were being conducted. Ten of these additions were not logged in either the CR0's or Shift Foreman's Logs. Operator testimony classified the failure to make log entries as an operational mistake. The leak rate surveil-
' lance procedure, SP 1303-1.1, required operator-induced inventory changes (for example, water additions) to be taken into consideration in the leak fate test calculations. Operator testimony characterized this as miscomunication between the console operator making the addition and the operator performing the surveillance test. While these actions appear to indicate an atmosphere of operational laxity and a failure by operators to comply with approved plant procedures, other facts identified in the investigation supported the argument that these additions were not intentionally made for the purpose of altering leak rate test results at TMI-1. Only 2% of the surveillance tests examined exhibited evidence of unaccounted-for water additions. Of these, only three cases were identified where the TS acceptance criteria would not have been satisfied had the additions not been made. There was no visible pattern of involvement by individual CR0s or shift foremen in making the additions.
The 14 additions involved 12 different CR0s and 7 different shift foremen with no one individual being involved in more than 2 unaccounted-for additions. The sworn testimony of all operators was unanimous in stating that they never par-ticipated in, or were aware of, any unaccounted-for water additions knowingly i
being made for the purpose of altering RCS leak rate surveillance test results.
On the basis of the information presented in the investigation, the staff con-cludes that the evidence does not support a finding that there was any willful or systematic pattern of falsification of leak rate surveillance tests at TMI-1 by unaccounted-for water additions during the period in question.
4.2.1.3 Feed-and-Bleed Operations Feed-and-bleed operations are usually done to change the RCS baron concentra-tion and the amount of coolant removed is normally equal to the amount added.
Performing this operation during a RCS surveillance test would tend to reduce the accuracy of the leak rate test results; however, it would not be an effec-i tive means of reducing the meaured leak rate unless the amount of coolant added exceeded the amount removed (OI LR-1 Ex. 21 at 6). The leak rate sur-l veillance procedure, SP 1303-1,1, cautioned the operators to avoid boration or deboration while performing leak rate tests; however, the procedure provided a data sheet for accounting for operator-caused changes to RCS inventory (0I LR-1 Ex. 4 at 1.0, 18.0).
Administrative Procedure 1012 " Shift Relief and Log Entries," Section 3.3.11,_
required that log entries be made for the addition of boron or dilution of RCS boron concentration (0I LR-1 Ex. 7 at 7.0). Therefore, during the time period under investigation, feed-and-bleed operations should have been recorded in the CR0's Log.
NUREG-0680 4-8 i
The revised Region I inspection report identified 14 feed-and-bleed operations (including the combined hydro operation on January 6,1979) gen addition, water addition, and feed-and-bleed that were conducted during periods in which RCS leak rate surveillance tests were in progress. Ten of the 14 operations were n'ot logged in either the CRO's or Shift Foreman's logs. Only 1 of the 14 operations was taken into account in the leak rate calculations. OI LR-1 Ex. 17 at 15, 36.
The revised Region I inspection report did not provide the same detailed evaluation for the identified feed-and-bleed operations that it did for the hydrogen and water additions to the MUT. The report listed 14 uncompensated feed-and-bleed operations and, where possible, identified the corrected leak rate values. Nine of the tests were determined to be invalid and corrected leak rate values were not provided. The corrected leak rates for the five remaining feed-and-bleed operations showed that the TS acceptance criteria would not have been exceeded had the feed-and Oleed operations not been con-l ducted (OI LR-1 Ex. 17 at 15 and 36).
The OI investigation report identified the feed-and-bleed operation that was
{
conducted during the leak rate surveillance beginning at 0047 on May 12, 1978, as having a corrected leak rate value that exceeded the TS limit for unidentified leakage (OI LR-1 at 15).
g The narrative section of the Re inspection report also supported this conclusion (OI LR-1 Ex.17 at 7)gion I However, the corrected leak rate for this evolution was stated as being 8
1.000 gpm (OI LR-1 Ex. 17 at 15). Section 3.1.6.2 of the TMI-1 TS for leak-1 age states that "[)i]f unidentified reactor coolant leakage (excluding normal J i
evaporative losses exceeds one gpm or if any reactor coolant leakage is evaluated as unsafe, the reactor shall be placed in hot shutdown within 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of detection" (0! LR-1 Ex. 2 at 3-12). Thus, the recalculated leak e
rate was at the TS limit, but was not in violation of the TS acceptance cri-teria of greater than 1 gpm.
IE Headquarters conducted a technical review of i
13 of the 14 feed-and-bleed operations identified in the Region I inspection j
(The combined hydrogen addition, water addition, and feed-and-bleed report.
i operation evaluation was not included.) 01 LR-1 Ex. 21 at 6-8.
The review i
concluded that in only four cases could it be shown that the feed-and-bleed operations were effective in reducing the measured leak rates for the associ-l ated tests and that the corrected leak rates for these four were within the allowable limits.
l Because no detailed technical evaluation of the feed-and-bleed operations was contained in the Region I inspection report, the OI interviews of the CR0s and shift foremen involved in the feed and-bleed operations concentrated on t
possible falsification of leak rates by the addition of water or hydrogen to the MUT. The operators were asked, however, if they were aware of any other i
method besides hydrogen and water additions that could be used to manipulate leak rate test results (e.g., OI LR-1 Ex. 33 at 67; Ex. 35 at 60; Ex. 74 at 42).
None of the operaters interviewed identified feed-and-bleed operations i
as a possible method for obtaining altered leak rate test results.
When the technical evaluation of the feed-and-bleed events was provided by IE, OI used i
this information in later interviews with five of the six shift supervisors (OI LR-1 Supp. Ex. I at 47; Ex. 2 at 39; Ex. 3.at 69; Ex. 4 at 75; and Ex. 5 1
at47).
Miscommunication between the operator performing the test and the operator conducting the feed-and bleed operation was the only explanation NUREG-0680 4-9
p j;
l '.
A offe' red by the shift supervisors as to the reason why a: feed-and-bleed opera-tion would be conducted during' the performance of a leak rate test: (OI 'LR-li
'l Supp'. Ex. 1 at 48; Ex. 2 at 40;.Ex. 5 at 47).
A review of the 14 ' questionable feed-and-bleed operations did not disclose q
any visible pattern regarding particular operator shifts,'CR0s, or shift I
foremen. - A total of 10 different.CR0s and 8 shift foremen were involved in.
these evolutions.No individual CR0 was involved in more than two of the' l
questionable. tests and no shift foreman was involved in the approval of more:
than three of the.affected tests.. OI LR-1 Ex. 20.
In summary, out of. the 645 leak rate. surveillance. tests evaluated,14 showed evidence of possible feed-and-bleed operations during the periods when the leak rate tests were being conducted.~ Contrary to Administrative Procedure.
1012, ~ only-four.of.these feed-and-bleed operations were. logged in either the -
CR0's. Log or the Shift Foreman's Log. The OI: investigation did not determine
'the reason why these evolutions were not logged.~ Limited testimony, provided by the shift supervisors during this period, offered misconnunication.between.
the console operator performing the; feed-and-bleed operation _and the operator-conducting the leak rate test as a possible reason why such a situation could -
The actions specified 'above appear to be additional examples of. opera-occur.
tional informality and failure to comply with approved plant procedures.' How-ever, other facts identified in the ' investigation tend to support the argument '
that these feed-and-bleed operations were not performed for. the purpose of altering leak rate test results at TMI-1. Only 2% of the surveillance tests examined exhibited evidence of possible feed-and-bleed operations being done lI in parallel with leak rate testing. Of these, no cases were identified where the TS acceptance criteria would have been exceeded had these operations not been done. No visible pattern of involvement by individual CR0s or shift foremen was identified. The sworn testimony of these CR0s.and shift foremen
'did not specifically address feed-and-bleed operations as a possible method for manipulating. leak rate data. However, each was interviewed extensively on hydrogen additions and water additions to the MUT'as other possible methods.
Each of the operators denied being involved in any' type of leak ~ rate manipula-tion. On the basis of the information' presented in the investigation, the staff concludes that the evidence does not support a finding that there was.
any willful or systematic pattern of falsification of leak. rate surveillance tests at TMI-1 by operators manipulating test results through the conduct 'of RCS feed-and-bleed operations.
4.2.2 Discarding of Invalid Leak Rate Surveillance Tests TMI-1 Technical Specification Table 4.1.2,." Minimum Equipment Test Frequency,"
required that RCS ~ leakage be determined daily when the RCS temperature was greater than 525"F-(OI LR-1 Ex. 3). Surveillance Procedure 1303-1.1, "RC System Leak Rate," was the approved procedure that governed how the leak rate-tests were.to be conducted (OI-LR-1 Ex. 4). Testimony provided by operators-identified that a more conservative test. frequency was established than re-quired by the Technical Specifications. The leak rate surveillance tests were routinely perfomed every shift (every 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />).
E.g.~, OI LR-1 Ex. 72 at 6.
The surveillance procedure could be run either by using the plant computer.or by performing. hand calculations.
In almost all cases, the computer was used' NUREG-0680 4-10 m
- f z -,,
to perform the test during the period under investigation.
0' I LR-1 Ex.17 af
~
16.
I The test procedure included a specified test duration of 1.to 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />.:
\\
.During this period, the procedure included a precaution to avoid the addition 2and removal of water from the reactor coolant and makeup systems. '. Makeup or chem.
~1 ical addition to the makeup system and boration or deboration were. listed as two of six evolutions that should not be conducted during,the test. OI LR-1 y
.Ex. 4 at 1.0.
On initiation of.the computer calculation, data was taken auto-matically from hard-wired, predesignated computer points. Manual actions; required by the computer method were input specifications, such as test inter :
val (nonnally I hour) and.any identified. leakage and any operator actions a
that would affect leak rate results,- such as makeup water additions and reactor coolant drain tank (RCDT) pumping. OI LR-1 Ex. 17 at 16.-
l
.If, following completion of the leak rate test, the RCS leakage was in excess of the acceptance criteria,. the procedure stated that the operator was to proceed as follows:
6.4.1 Perform another determination'n of RCS leak rate.
l 6.4.2 Insure-that no unaccounted for operator action has occurred that would cha~nge the RCS inventory....If such an action has occurred, it invalidates the measurement. Enter this in the Remarks section of the data sheet, clearly. describing the action that invalidated the measurement.
l OI LR-1 Ex. 4 at 3.0.
TS 6.5, " Station Operating Records," required that records.of periodic checks, tests, and calibrations-be prepared and retained for a period of 5 years (OI.
-l LR-1 Ex. 5 at 6-5). The licensee's Administrative Procedure (AP),1010. " Tech-nical Specification Surveillance Program," required that when the surveillance-tests were completed, the results were to be compared to the acceptance cri-teria.
If any part of the result was unsatisfactory or if problems were encountered while performing the test, it was to be. recorded on an " Exception and Deficiency" list, (OI LA-1 Ex. 6 at 9.0 and 10.0). AP 1012. " Shift Relief and Log Entries," required that the start and completion times be. recorded
(. OI LR-1 Ex. 7 at 7.0).
..1 The RCS leak rate test procedure utilized at TMI-1 during the year prior to the accident at THI-2 contained several inadequacies. These-would, in most instances, have calculated leak rate values that were less than the actual.
leak rates (OI LR-1 Ex. 1 at 6). Because of these errors in the test calcula-1 tions, the surveillance test records frequently showed negative leak rate results. -For example, leak rate test results for the months of May and June 1978 showed that almost 40%-(48 out of 121) of the calculated and recorded
'L' unidentified leak rates had negative values (OI LR-1 Ex.17,12 14). The leak rate procedure did not prohibit acceptance of test results showing nega-
)
tive leakage. Nevertheless, many of the o>erators testified that they nor-I mally would consider the test invalid if tie calculated leek rates were more-i l
NUREG-0680 4-11 j
i
~
L a
f.
v.
f g-negative than approximately -0.5 to '-1.0 gpm (e.g., 01 LR-1 Ex. 38 at 9; Ex. 48 at 19; Ex. 72 at 9).
The investigation determined,Lthrough sworn testimony by the preaccident operators, that it was common. practice at TMI-1 to discard leak rate test results that were determined to be." invalid." Fourteen of the 18 CR0s, 6 of the 8 shift foremen 4 of the 6 shift supervisors, and the Supervisor of..
l Operations all testified that they were aware of this practice (_e_... OI LR )
g.
Ex. 36 at 7; Ex. 75'at 8; Ex. 107 at 13; Supp. Ex. 5 at 10). Except for a j
former -TMI-1 Plant Superintendent (J. O'Hanlon), none of thefline managers responsible for TMI-1 stated that they were aware of the practice of. dis-carding invalid leak rate tests [0I' LR-1 (O'Hanlon) Ex. 58 at 21; (Seelinger)
.Supp.Ex.'9at13;-(Miller)Ex.106'at22;(Lawyer)Ex.104at38;-(Herbein)-
Ex. 102 at 16].
~
According to testimony of the operat' rs, there was no formal guidance pro-o vided by management for. determining whether a leak rate test was classified i
as valid or invalid. However, most operators agreed that either large, posi-tive leak rates that were not representative of what other plant instruments-i tion indic'ated or large, negative leak rates were routinely considered invalid (e a, OI LR-1 Ex. 36 at 8; Ex. 44'at 7; Ex. 71Lat 6). When an invalid leak rate test result was obtained, the operators would ' start a new leak rate cal-culation. When a, valid leak rate result was obtained, the invalid test was'
(
thrown away.
Invalid tests were not recorded in the log (as required by:
AP 1012); the reasons why the tests were considered invalid were not written in the remarks section of the data sheet (as required by SP 1303-1.1); prob-1 ems encountered while performing the tests were not recorded on " Exception and Deficiency" reports (as required by AP 1010); and test results were not -
maintained on file for a period of 5 years (as required by TS 6.5.4).
The majority of operators testified that invalid leak rate tests were caused i
by several factors such.as plant oscillations or transients during the test; operator actions, such as water additions or pumping of the reactor coolant drain tank; computer input' error by the operator perfonning the test;-instru-ment errors; and possible problems with the ' computer program itself.(e.1..
OI LR-1 Ex. 70 at 6; Ex. 75 at 7; Ex. 45 at-11; Ex. 33 at 16).. The opera.-
tors indicated that invalid leak rate tests were not indicative of actual plant conditions and, because the infonnation obtained from toe computer was in error, they did not believe it was required to be kept.
For e' ample, x
D. C. Janes, a shift foreman during the period under investigation,' stated:
"If there were obviously an error, our policy at the time was to retain the test that was bad and rerun it, and when we had a valid test, then we would discard the invalid one and submit the valid one....It seems to the best of ry recollection that the operators and foremen at the time thought that the intent was to get an accurate leak rate. And if it was obviously way off, f
we felt that it was in bad faith to turn it in."
0I LR-1 Ex. 75 at 8. 9.
None of the testimony indicated that the discarding of the leak rate tests that were deemed invalid, was done with an ulterior motive to conceal deroga-tory information about plant conditions from regulatory officials (OI LR-1 i
at 25).
Interv'iewees indicated that there was minimal difficulty in obtaining good leak rates. By and large, they considered the leak rate surveillance proce--
dure to be a valid procedure. Except for one former CR0 (J. C. Banks) (Ex.:33 1
NUREG-0680 4-12
- ~.
q.
a
~
l at16),theopiratorsbelievedthat'thetestswerereliableandconsistent with the ' status of the plant and that they routinely experienced few problems.
in obtaining satisfactory leak rate tests. OI LR-1. at 25.. However, it. is,
apparent from interviews conducted with the CR0s that the performance of RCS.
leak rate tests at TMI-1 was considered mundane and' repetitive and that the actual performance of the -tests was approached in a very perfunctory manner.
[ memorandum from B. B. Hayes (OI) to T. E. Murley (Region I). dated Aprf1l16,.
l 1984 at 2].
'In sunnary, it appears, from the testimony < taken during the investigation.,
that the practice of. discarding;what operators deemed invalid leak rates was-connon practice at. TMI-1 as far back as any. of. the interviewees could recall.
There was no formal written policy from supervisory or management personnel l
that either provided criteria for-determining the validity of leak rate
{
results or directed that invalid leak rate tests be discarded.. Not logging i
all tests, not providing proper justification and documentation for invalid 4
tests, and not retaining the invalid test results are violations of plant
~
Technical Specifications and approved plant procedures..However, the evidence does not support a finding that operators were either performing these actions. q ~
as.a deliberate attempt to conceal: actual leakage that was in violation of TS acceptance criteria or attempting to concea'1;this information from the NRC.
4.2.3 Whether There Was a Motive for Leak Rate Falsification I
An important element in trying'to determine whether there was leak rate test falsification at TMI-1 similar to that which was alleged to occur at TMI-2 t
4 was to determine whether there was a motive or need to manipulate' leak rate results at YMI-1.
The IE Headquarters review of the Region I inspection report ' discusses several
. key differences between TMI-1 and TMI-2 (OI LR-1 Ex. 21 at 8-10). These dif-ferences may be summarized as follows:-
(1) At Unit 2 the wrong water density was used in the computer calculation of the identified leakage collected in the reactor coolant drain tank (RCDT). As a result, any time that the identified leakage exceeded approximately 2.5 gpm, the computer should have, theoretically, calcu-lated an unidentified leakage rate in excess of the allowable 1 gpm limit, even if the actual unidentified leakage was zero. For the m4ority of time, during the last four months of operation of TMI-2 the identified leakage did exceed 2.5 gpm. OI LR-1 Ex. 21-at 8.
(2) Both TMI-1 and THI-2 used a temperature versus RCS density conversion table in their computer programs that only went up to 582*F. - This was.
not a problem for Unit I because RCS temperature rarely exceeded 582'F.,
Unit 2 operated at a higher temperature than this for about a third of 3
j the tests reviewed by IE. This deficiency introduced an error of-i approximately I gpm into the results of those tests which exceeded 582'F' (0I LR-1 Ex. 21 at 9).
(3) At Unit 1, RCS losses to certain connecting systems such as the RCDT, were not considered leakage and, thus, were subject to the limit for ~
1 I
NUREG-0680 4-13
{
\\
~
leakage plus losses of 30 gpm. For Unit 2, these losses were considered to be identified leakage and, therefore, subject to a limit of 10 gpm.
01 LR-1 Ex. 21 at 9.
(4) The Unit I limit for unidentified leakage was 1 gpm after subtracting an evaporative loss factor of 0.51 gpm. The Unit 2 limit did not include an evaporative loss factor. OI.LR-1 Ex. 21 at 9.
(5) The final significant difference between Unit 1 and Unit 2, pointed out by IE, was that the actual leak rates at TMI-1 were lower. Therefore, Unit 1 could tolerate a larger measurement error before calculating an unacceptably high value. 01 LR-1 Ex. 21 at 9, 10.
The Region I inspection report identified several procedural inadequacies associated with the THI-1 leak rate procedure. However, these inadequacies would, in most cases, have lead to a less conservative leak rate calculation (that is, the calculated leakage would have been less than the actual leak-age) OI LR-1 Ex. I at 6, 22-24.
The consensus of testimony provided by operators and supervisory personnel classified TMI-1 as a " tight" plant in terms of leakage. Operators experi-enced minimal difficulty obtaining acceptable leak rate test resultL (le..,
0I LR-1 Ex. 41 at 15; Ex. 44 at 13; Ex. 76 at 76; Ex. 77 at 18; Sup7. Ex. I at 8; Supp. Ex. 2 at 8-11). As discussed in Section 4.2.2, when inv'hlid leak rates were obtained, operators identified the problem in most cases"as being caused by plant oscillations or operator-induced errors. None of the opera-tors, supervisory personnel, or management officials interviewed were aware of any actual leak rate problems at THI-1 during the period under investigation.
1 In summary, the investigation does not support a finding that there was a motive or need to cheat on leak rate tests at Unit 1 as appears to have been l
cne case at Unit 2.
The recalculation of leak rates by Region I and IE Head-quarters did not identify actual leak rate problems for Unit I during the l
1-year period that was evaluated. On the basis of the technical analysis per-I fonned by IE and Region I, combined with the sworn testimony of the operators i
and management-personnel, the staff concludes that there was no apparent motive or need to falsify leak rates at TMI-1.
4.2.4 Whether There Was Management Involvement in Any Leak Rate Falsification As stated in Section 4.1, the purpose of this investigation was to determine if there was a systematic pattern of falsification of RCS leak rate surveil-lance test data by operators at THI-1 and, if so, to what extent licensee's management was cognizant of and/or involved in such activities. This section addresses the latter half of this question.
j On the basis of the infonnation presented in Sections 4.2.1 through 4.2.3, the staff has concluded that the evidence does not support a finding that I
there was any willful or systematic pattern of manipulation or falsification i
of leak rates at TNI-1.
In addition, there was no apparent motive for oper-I ators to manipulate leak rate tests. The operators experienced only minimal I
difficulty performing leak rate tests and obtaining results within the allow-able TS limits. The investigation did identify apparent violations of the i
NUREG-0680 4-14 l
+
l y
7 4 -
9..
a Technical Specifications-and approved plan't procedures by operators not '..
logging all tests, not providing proper justification and documentation for
. invalid tests, Land not retainingLtest results. deemed to be invalid. This latter practice was' common and known to be occurring at the time. by: the majority.of CR0s, shift foremen,l shift supervisors,-and the Supervisor. of
. Operations.. However, the evidence did. not, identify that these actions were being perfonned either as'a deliberate attempt to conceal. actual leakage that; was' in violation of the Technical Specifications or. to conceal this informa -
tion from the.NRC.
During each. interview, licensed and unlicensed CR0s were specifically ques '
tioned regarding real or. perceived pressure from supervisory or management.
i personnel to. obtain valid leak rate tests.. They;were also asked if. they'were j
ever directed to perform any actions that would. result. in the manipulation.
1
- or falsification of leak = rate data by either supervisory or management per '
d sonnel. Other than the nonnal pressure placed on the operators to do'all.
things right, testimony by CR0s was unanimous that"they were_ never directed -
by any supervisor. or management official to manipulate test' data -in any L
. matter. None of the_ operators testified that they ' felt undue pressure by supervisory or management personnel to obtain leak rates that were within Ex. 38 at 30; Ex_E.1._ OI LR-l' Ex. 33 at 23; Ex. 35 at 47; Ex.' 36 at 17; the TS limits.
. 39 at 15; Ex.' 40 at 16; Ex. 41 at 18; Ex. 43 at-35.-
Similar testimony was obtained from the SR0s.who 'were assigned as shift foremen and shift supervisors during. the period' under investigation. These supervisory personnel stated that they were never pressured to manipulate or obtain good leak rates at TMI-1 by. any management official.-(_e.g_., OI LR-1 Ex. 71 at 34; Ex. 72 at 31; Ex. 74-at 39; Supp. Ex. 1 at 18; Supp. Ex. 4 at 32; Supp. Ex. 6 at 22).
In like manner, TMI station management and corporate management officials also were interviewed under oath.. Each of these inter-viewees (e.g., Arnold, Herbein, Mil.ler, Ross,.and Shipman) stated that they neither directed nor were aware of any undue influence placed upon TMI-1 operators or' supervisory personnel to manipulate leak rate tests for the purpose of obtaining good leak rate results at TMI-1-(_e.g_. 01 LR-1 Ex. 97 at 23; Ex. 102 at 27; Ex. 106 at 51; Ex. 107 at 73; Ex. 109 at 27).
In sununary, there is no basis identified in the investigation that would support the conclusion that management or supervisory personnel placed pressure on the operators at TMI-1 to manipulate or falsify leak rate test results.
3 4.2.5 Followup on Makeup Tank Level Indication Problems On April 16, 1980, the law firm of Faegre & Benson was contracted by Met-Ed j
to direct an investigation of allegations-made by H. W. Hartman, Jr. ' after-WOR-TV, New-York, broadcast an interview with him on March 24,'1980. Hartman-alleged that leak rate test results were intentional {MI-2. manipulated for the-purpose of obtaining results within the TS limits-at.
He stated that one of the techniques utilized by operators at TMI-2 to' alter leak rate.
results was to add hydrogen gas-to the makeup tank (MUT) during the. perfonn-ance of test. OI LR-1 Ex. 113 at 1.-
4 NUREG-0680 4-15 I
p
+
3 4
According.to the Faegre &'Benson Report dated September 17 1980:
Under operating conditions, there.is no theoretical reason why hydrogen added to the make-up tank even during the per-fonnance of a leak rate test would improve the results.
Because Hartman had alleged that hydrogen additions were used to improve the results,' we reviewed the make-up tank system to see whether any abnormalities might permit this 4
effect to occur.
i We discovered that a loop existed in the 'ifry-reference leg '
tubing which runs ~from the top of the make-up tank to the.
make-up. tank level sensing instrument. The configuration of j
the loop is such that any condensation which might gather at'
]
the bottom'of the loop could form a water plug..Such a. water j
plug, because its density-is greater than the density of.
j hydrogen, would distort the effect..of pressure changes in the
- j make-up tank which occur when hydrogen.is added.. Assuming the existence of a water plug, increases in hydrogen overpressure
-1 l
would increase indicated water level in the make-up tank even H
though the make-up tank water level remained the same. Thus, I
increases in hydrogen overpressure during a leak rate test could decrease the calculated unidentified leakage in some Cases.
OI LR-1 Ex. 113 at 42, 43.
'i On July 11, 1983, as part of the Region I inspection of reactor coolant' leak j
rate tests at TMI-1, the inspector discovered the' existence of a similar loop
,j seal in the dry reference leg of the MUT level instrumentation at TMI-1.
Subsequently, the insp(ector, accompanied by a licensee representative, veri-fied that the U loop loop seal) in the dry leg was more than 4 feet in height (OILR-1Ex.at30). Followup tests conducted by the licensee and Region I j
inspectors on July 28, 1983, confirmed that the addition of small' amounts of hydrogen to the MUT could, under the proper conditions, cause' indicated level in the MUT to increase (0I LR-1 Ex.1. at 31, 32). As discussed in Section f~>
4.2.1.1, the addition of hydrogen at the appropriate: time during leak rate testing can affect the leak rate results in a nonconservative manner.
l The OI investigation developed testimony regarding two subtopics related to I
the loop seal. First, during the preaccident period, was the licensee aware 11 of the existence of the loop seal on TMI-1.7 -Second, if the licensee was:not i!
aware of the existence of the loop seal on TMI-1 in the preaccident period, why had the existence of the loop seal not been identified by the. licensee L'
following the publication of the Faegre'& Benson Report?
Evidence existed, during the period under' investigation, as documented in'
-i, r
' maintenance work requests, that there.were periodic problems;with MUT level indication (0! LR-1 Ex. 11-15). Entries on these work requests indicated that some members of the plant staff were aware of the effect of water con-densation in the low pressure leg of the level instrument. 'A conclusion can be drawn from the testimony of the operational personnel, instrumentation and control (180) technicians, and the plant engineering staff that site personnel L
failed in the preaccident period to recognize that the collection of water in 4l 4
NUREG-0680 4-16
[
t 6
y t.
the dry reference leg.and the resultant problem with the MllT level recorder resulted from the existence of the loop seal in'the sensing'line. Except for-periodic draining of the dry reference leg, when a work request was submitted, 1
' the licensee did not formally analyze the problem to determine its root cause.
01'LR-1 at 22.
During the' postaccident period, the licensee's own' internal investigation into the Hartman allegations enumerated leak rate. test discrepancies and-abnormalities at TMI-2, including the identification of a loop. seal on the dry reference leg of the MUT level instrument; system. Several management personnel were questioned to determine whether the licensee' tried to ascer-i tain if the TMI-2. problems pointed out in the Faegre & Benson Report were equally applicable to TMI-1. The majority.of.these. individuals. testified that TMI-1 leak rate procedures and hardware were examined as a result'of
' the. report; however, the existence of the loop seal was not identified until discovered by the Region I inspection on July 11,~1983. OI LR-1 at 24.
4 On the basis of the'information identified during the investigation OI has concluded that a'significant amount of. infomation was available to plant management regarding the hydrogen effect as a result ~of.the loop seal in.the RCS MUT at TMI-2.and its apparent cause and effect on leak rate surveillance tests.- Despite this information, no affirmative actions were initiated to
. determine if the potential existed for the same problem at TMI-1 until the J
NRC Region I inspection.
'I The staff concludes that, in hindsight, there was infomation available to the licensee that could have led to the discovery of the loop seal before July 1
1983.
However, because approximately half of. the operators testified that they were aware of the effect of hydrogen additions on.MUT level, the exact-cause or mechanism that allowed this-phenomenon to occur is not considered l
significant in determining whether leak rate tests were falsified at TMI-1.
4.3 Staff Findings
?
On the basis of an independent review of the information contained in Investigation Report 1-83-028, "Three Mile Island Nuclear Generating Station (NGS) Unit 1 - Possible Falsification of Reactor Coolant System Inventory Leak Rate Tests," and'its exhibits, the staff concludes that the evidence does not support a finding that leak. rate surveillance tests at THI-1 were l
intentionally or systematically falsified during the period investigated.
}
. The staff notes, however, that it is impossible to exclude the possibility i
that individual operators may have attempted to manipulate test results for-i some unknown reason. The specific staff findings discussed earlier in tgis l
section are sunnarized below.
6 (1) Only a.small percentage (5.9%) of'the 645 leak rate surveillance tests that were conducted at TMI-1. between the period April 1,1978 March 31, 1979, and that were examined by Region I, were accom..and plished during periods when operator-induced evolutions occurred that would call =
into question the validity of these tests. These include 38 unaccounted-i for hydrogen and water additions to the MUT and feed-and-bleed operations.
See Section 4.1.
4 NUREG-0680 4-17
'l.
L x.W:
3
7 J
g:
j.;
J a
(2) Of the 38 questionable tests,. technical analyses by Region'I and.IE Headquarters, show that, except-in three instances,,the TS acceptance-criteria for unidentified leakage would have been satisfied had the.
operator-induced evolutions not occurred. This represents only 0.5% of:
the 645 test records examined. See Section 4.1.
(3) There is no conclusive evidence'to indicate that any TMI-1 licensed or unlicensed operator intentionally performed plant evolutions during leek rate testing with.the intended purpose of menfoulating or falsifying leak rate results. There is also no pattern of specific operator involvement' in the questionable tests.
lhe 38 questionable tests involved 12 of.]
licensed and unlicensed CR0s and all 10 shift foremeri.'
I The example that would come the closest to showing a pattern would be.
{
shift D in which two CR0s (P. Chalecki and D. Wooddell) and their Shift
.j Foreman (D. Janes) were involved in 9.of the=38 tests. These_ tests in.
3
.volved four hydrogen additions, two water' additions, two feed-and-bleed-
]
operations, and one combined hydrogen addition water addition, and..
feed-and-bleed operation during periods when leak rate tests were being
'q conducted.
Both Janes and Wooddell-have testified that they were unaware..
i of the effect of hydrogen additions on leak rate test results...while Chalecki stated he was aware of the effect and was told not to make hydro '
l gen additions during leak rate testing. All three operators ' admitted the l
possibility that accidental additions could have been made during these
]
tests; however, each denied, under oath, that they ever, performed any type of evolution to intentionally affect the outcome of leak rate tests.
H None. of the nine tests involving.these operators had recalculated leak rates in excess of the TS acceptance criteria.
During the period under investigation,. there were six rotating shifts 1
assigned at TMI-I. Therefore, it is reasonable to assume that these operators performed approximately 1/6 of the 645 tests examined or 108 tests. Thus, the nine questionable tests involving these operators represents only ap these individuals. proximately 8% of the leak rate tests performed by See Section 4.2.1.
(4) There was no apparent motive or need to manipulate leak rate tests at THI-1. Operators experienced only minimal difficulty in conducting leak rate tests.
In most cases, operators attributed any difficulty in per-forming the tests to either miscommunication between the panel operator and the operator performing the leak rate test or to failure to maintain steady-state conditions during the period the. leak rate data was being obtained.
As a result of the calculation inadequacies in the leak rate program, negative leak rate :results were often obtained. Technical.
analyses by Region I and IE Headquarters supported the fact that there was no actual leak rate problem at TMI-1 during the period investigated.
a See Section 4.2.3.
(5) The investigation did not identify any evidence that would indicate supervisory or management personnel placed pressure on the operators at TMI-1 to manipulate or falsify leak rate test results. See Section 4.2.4.
p 1
h p
NUREG-0680-4-18 k
I i
j (6)
It was common practice, during the period under investigation, for I
operators to discard what were deemed " invalid" leak rate test results.
1 There was no formal written policy from supervisory or management per-sonnel that either provided criteria for determining the validity of leak rate results or directed that invalid leak rate tests be discarded.
This practice was apparently known and condoned by plant supervisory and management personnel up to the level of Manager of Plant Operations for TMI-1 (M. Ross). Testimony of the former THI-1 Su (J. Seelinger), Station Superintendent (G. Miller)perintendent
, and Metropolitan Edison Vice-President of Generation (J. Herbein) indicated that they were not aware, at the time, that operators were discarding invalid leak rate tests at THI-1. While the practices of not logging ail surveillance tests not providing proper justification and documentation for invalid tests, and not retaining the invalid tests are violations of plant Technical Specifications and approved plant procedures, the evidence developed during the investigation does not support a finding that the operators were either performing these actions as a deliberate attempt to conceal actual leakage that was in violation of the TS acceptance criteria or to conceal this information from the NRC. See Section 4.2.2.
(7) On June 26, 1984, the staff received the ifcensee's report entitled "TMI-1 Reactor Coolant Inventory Balance Testing" (Stier Report). The staff perfonned a review of the report (including appendices) and finds that the conclusions of the Stier Report are consistent with the staff's eval-uation of the TMI-1 leak rate issue presented in Section 4.2 with the exception of two points: First, with respect to the design configuration of the MUT level indication, the Stier Report claims the installation of a drain valve on the low point of the " loop seal" at THI-1 is a major dif-ference between THI-1 and THI-2. This valve allows draining of condensa-tion from the loop seal, rendering the effect of hydrogen additions to the MUT at THI-1 negligible. The staff disagrees. Approximately half of the operators at THI-1 testified that they were aware of the effect of hydrogen additions on MUT level. Thus, while the drain valve was installed, the frequency of its use is not cicar. During an OI invest-igation, I&C technicians were interviewed regarding maintenance on the MUT level instrumentation.
It was established that work requests were initiated by operations personnel because of erratic behavior of the MUT level indication. However, the corrective action did not include using the drain valve to remove condensate from the reference leg (Stier Reportat32). Also, no regularly scheduled preventive maintenance i
program was established to ensure that condensation was drained from the loop seal at a frequency sufficient to preclude hydrogen additions from having an effect on MUT level indication. The Stier Report argu-ment that the THI-1 level instrument is not defective because of the existence of a drain valve is moot because condensation and/or valve
{
1eakage collected in the reference leg causing erratic MUT level indi-cation. Second, the Stier Report claims it was unable to confirm any of J
the hydrogen additions and could confirm only 3 of the 14 water additions and 5 of the 14 feed-and-bleed operations. The staff agrees with the aler Report in that the method of identifying water additions and hydrogen additions to the makeup tank is necessarily subjective (Stier l
i f
at 53), and, therefore, disagreement in interpretation can be expected.
The staff considers its evaluation of the IE inspection and the 0! invest-igation sufficient to conclude that the preponderance of evidence does l
)
NUREG-0680 4-19 i
I
. - -. - - - = - =.
_i not support a' finding that leak rate test results were intentionally manipulated at TMI-1.. In sum, the staff agrees with the statement in the Stier Report that "even if all the suspected additions had,'in fact, taken place, they would have involved such an insignificant number of tests that no inference of test manipulation could. reasonably be drawn" (Stier at 60).
l.
i b
f I
a NUREG-0680 4-20
-