ML20107F334

Testimony of El Zebroski Re TMI Incident.Related Correspondence

ML20107F334
Person / Time
Site:	Three Mile Island
Issue date:	11/01/1984
From:	Zebroski E ELECTRIC POWER RESEARCH INSTITUTE, GENERAL PUBLIC UTILITIES CORP.
To:
Shared Package
ML20107F324	List: ML20107F330 ML20107F334 ML20107F338
References
SP, NUDOCS 8411050437
Download: ML20107F334 (12)

Text

_ _ _ _ _ _ _ - _ . . _ _ _ _ _ _ _ _ _ - - _ _ _ _ _ _ - _ . ___ _ _ . _ _ _ _ _ _ _ - _ _ - _ . - _ _ _

t Y D Q Tzyg ; ,

j>';; W Zit m 00CMETEf UShec 1984 November 1,'84 IOV -2 R2:10 UNITED STATES OF AMERICA

- ' NUCLEAR REGULATORY COMMISSION BEFORE THE ATOMIC SAFETY AND LICENSING BOARD In the Matter of )

-)

METROPOLITAN EDISON COMPANY ) Docket No. 50-289 SP

) (Restart-Management Remand)~

(Three Mile Island Nuclear )

Station, Unit No. 1) )

TESTIMOMY OF E. L. ZEBROSKI My name is Edwin L. Zebroski. My current position is

' Chief Nuclear Scientist at the Energy Study Center, a part of the Electric Power Research Institute (EPRI) in Palo Alto, California. EPRI is the research arm-of the electric utility industry. Prior to joining EPRI, I held various design and de-velopment positions in. Stanford Research Institute, (Physics Department),.and in the General Electric Company, Research Lab-oratory,..and in the Nuclear Energy Division. My training in-cludes degrees in Science from the University of Chicago and the University of California. I am a registered Professional

. Engineer, and a member of the National Academy of Engineering.

I have authored or co-authored over 120 technical publications g 'and patents relating to the basic and applied science of nucle-ar energy. A major area of my specialization during the period 1965-1976 was the behavior of nuclear fuel under various operating conditions, including transients and accidents.

8411050437 841101

- PDR ADOCK 05000289 T PDR u-.

' ~

fb _ .f. .

T: ;g 1

shi--

.The purpose.of my testimony is to cover three-main points,

a. -

- ; based-on my.personalfobservations and involvement as a member.

of-the. Industry-Advisory Group,' convened at Three Mile Island in the_early days following~the accident:

1. The extent to which there was a rapid learning curve evident-in.the days'immediately after.the acci-dent', in respect to organizing, and interpreting, the large volume of plant data, and in sorting out dif-ferent views.and speculation'as to the extent and na-ture of-the damage to the reactor, by focusing on generation of hydrogen;as illustrative of this learn-ing curve.

2. The extent.to which related uncertainties re-s mained for months after the accident, reflecting the limited ~ general state of knowledge of severe core accidents at that time.

3. The extent and nature of the involvement of Mr.

Herman Dieckamp-in the activities and technical dis-cussions of the Industry Advisory Group during-the period of my observation.

At the time of the TMI-2 accident, I was Director of the Nuclear Systems and Materials Department at EPRI, which con-

' ducts research and development programs-aimed at improved life-

-time, reliability, and cost-effectiveness of components, fuels, and systems of Nuclear Power Plants. (Mr. Dieckamp was gener-

, ally. aware of these programs through his participation in prior years in two of the advisory committees which serve EPRI.) At a Research Advisory Committee meeting in Scottsdale, Arizona, Dr. S. Bartnoff of GPU reported to the Committee on March 29

.and again on the. morning of March 30, that an incident had oc-curred at TMI-2. Later in the morning of March 30, Mr.

Culler, the President of EPRI, reported to the same meeting on a phone call from Mr. Dieckamp which indicated that the situa-

-tion had deteriorated relative to the perceptions on the previ-ous day and that technical support help from EPRI was needed.

Mr. Culler agreed to send technical assistance to TMI, ini-tially consisting of Mr. Milton Levenson, then Director of the Nuclear Division at EPRI, and myself. Mr. Dieckamp outlined four basic tasks which needed technical support; I was asked to undertake the first task which was Core Damage Assessment.

After a conference call on March 30 with Mr. Robert Keaten of GPU, I traveled to TMI, arriving on the morning of March 31.

Office and conference space was made available at the National Guard Armory adjacent to the Harrisburg airport. An initial meeting to review the situation was organized and the technical review discussion was led by Mr. Dieckamp on the afternoon of

-March 31. Sometime during March 31, I became aware of the pressure spike which occurred shortly before 2:00 p.m. on March 28, 1979. I remained at TMI intermittently for the next four weeks, serving as co-leader of the Industry Advisory Group I

which was assembled to provide calculations, and evaluation of options for maintaining control and safety of the reactor sys-tem.

My investigative efforts on core damage at TMI during the initial days following the accident centered on several ques-tions: namely, (1) the postulated hazard from the gas bubble in the reactor, (2) the possible extent of core damage, and (3) possible means for removing the gas bubble.

The gas bubble evident in the reactor was postulated to be potentially subject to ignition and explosion creating a sense of immediate potential for catastrophe. This potential appar-ently was first postulated about March 30th, and was reported in the national media with banner headlines. Various people from national laboratories discussed the explosive potential.

The President's Science Advisor was reported to have commented that New York City and Philadelphia might be exposed to severe radiation if the bubble were to explode. A helicopter re-portedly was dispatched to bring sacks of oxygen-absorbing chemicals (like sodium hyposulfite, a chemical used in photog-raphy).

In the telephone call with Mr. Robert Keaten of GPU on March 30 (mentioned earlier) he noted that he had become aware of a gas bubble in the reactor vessel but did not know its source or its full composition. He hypothesized that it might contain some air, from air dissolved in the borated water used to assure safe nuclear shutdown of the reactor.

pm -

a-

~

n '; :n .

g ,

4

, < I: stated to Mr. Keaten my belief that oxygen could not be L fpresent in:the reactor vessel and that no explosion was possi-m ble; . I repeated this position later in meetings with NRC peo-spie on= site-(Stello,.Vollmer,nand Mattson) during the period g ,

JMarch 31.toTApril--2, quoting the extensive literature on this -

~

t l subject-dating to the 1950's and 1960's. The basic scientific information-was that the presence of.even small amounts of hy-drogen' suppressed the effect of radiation on water. (In-the absencefof excess hydrogen, radiation acting on water can pro-1 duce hydrogen and oxygen in a volume ratio of 2 to 1,2which is an explosive mixture.)

-Apparently_none'of the staffs or the officials of the var-

~

ious1 government agencies involved were awarc that since the mid-1950's, hydrogen was. routinely used in all pressurized

water reactors - :both Navy and civilian power -- to prevent the formation of oxygen-hydrogen mixtures by radiolysis. . I a f urged the NRC' representatives to'make. telephone calls to.the lc national laboratories (Brookhaven, Argonne, and Oak Ridge)

.where'the scientific and test work had been done to check out

.this information. This work'was widely published'in technical a l papers, and covered in textbooks on nuclear enqineering.

By about April 2nd or 3rd, the NRC decided that the evi-

-dence against the possibility of a hydrogen-oxygen explosion was indeed unquestionable, and the bubble was disappearing.

This was announced publicly, with the comment that previous 1

L.

(

I I

concerns of possible explosion were due to overly conservative calculations. (It was later evident that the earlier inquiries to the national laboratories did not indicate that hydrogen was present in the reactor vessel.)

During the same days I was also continuing my efforts to estimate the extent of core damage. The prevailing state of knowledge on possible reactor core damage as of 1979 was the analysis in the report Wash-1400. This report, and the related NRC calculations used in licensing, postulated that if cooling water was lost, the fuel would fail (distort and leak) due to high temperature, and that the reactor core would then proceed to melt down with extensive spread of the bulk of the radioac-tive elements in the fuel (up to 70% of the total). The infor-mation available to me March 30 through April 4 did not corre-spond to such a degree of severity. The observat'ons available March 30 and 31, (including the pressure spike and the indica-tions of high levels of gaseous radioactive elements, but only small amounts of iodira and cesiu.a) wat that a significant fraction of the fuel was certainly perforated, releasing most of the rare gases. The apparent evidence that only a small fraction of the iodine and cesium were released was consistent with perforation of fuel cladding, but not necessarily gross disruption or melting of fuel. If major core damage were present, a large fraction (up to 70%) of tnc iodine and cesium would be expected to be volatilized -- according to the pre-valling calculations accepted by the !!RC.

tr - '

X;V s

W s,._ .

ig The pressure spike was-evidence offthe probable presence c of enough hydrogen to burn, but of itself was not evidence of h'ow much had been' produced. Small amounts of hydrogen -- as

- low as 4% in air, are known to be capable of ignition -- which

~

would result in a pressure pulse, even if there was simply burning rather than-explosion. I was aware that hydrogen gas from gas cylinders is routinely used.to provide a small amount of hydrogen dissolved in the reactor. coolant. As noted earli-er, the dissolved hydrogen'is used to prevent the decomposition of water lur radiation'(radiolysis), which would otherwise form

- oxygen and hydrogen. There was an evident need to determine

- whether some hydrogen cylinders or piping might have leaked hy-drogen into the containment, which then could be ignited when a relay.or motor was actuated or started.

Another possible source of hydrogen was recognized to be

'from the reaction of zirconium with steam at high temperatures.

This'was also plausible but did not of itself necessarily imply more extensive fuel damage than just perforation from localized overheating. Localized overheating alone could cause clad bal-

- looning and rupture, with or without the added effect of oxida-tion of zirconium.

One of the major technical surprises of the subsequent in-vestigations of the TMI accident has been the low extent of mo-bility of iodine and cesium, despite what we now know to be 4

major core damage, with oxidation of a large part of the

F-a s

cladding. On the basis of the licensing analyses prevalent at the time of TMI, major core damage should have been accompanied by the release of large fractions of iodine and cesium (up to 70% of total inventory) to the primary system, and from there to the containment building air, and to any leakage paths to the auxiliary building. It is now known scientifically (although not yet fully accepted for regulatory purposes), but was not known or accepted then, that iodine and cesium, under conditions prevailing in a PWR loss-of-coolant accident, have a very strong affinity for water. The relatively large amounts (over 5%) which are now believed to have escaped from the fuel at TMI-2, have remained almost entirely in the water.

The small amount of iodine that did escape to the air (a small fraction of 1%) was readily detectible in the containment building and the auxiliary building. Had the postulated amounts of iodine been released, much larger emissions of io-dine to the containment, and via leakage paths to the auxiliary building, would have been expected. In the absence of such ob-servations, the expectation that core damage was limited to leakage or perforation of some fuel was plausible. If the fuel were only perforated, then it would still be possible to remove it and replace it using conventional underwater mechanical han-dling equipment. A small degree of fuel perforation

(" leakers") is often present in the normal periodic refueling operations.

r H+

t m4 ;g . - - .

y --,4.5' x

is cThe question'of how much hydrogen was evolved was the key 1

$ .to determining whether-the core damage was limited to'perfora-v_

x '

tion.or whether'there'had.been more extensive or'even severe. -

" disruption of the core structure.

The first-solid evidence of Ethe amount'of hydrogen produced came from the analysis of. gas -

samples taken from~the containment building on March 31 at-r 0600. :These showed significant oxygen depletion (4.4% to 5.270 belowethe normal value in air, respectively). .This corresponds

( Eto extensive reaction of zirconium-(later calculated to be 45 s -

'to 52%,of the core. inventory).. However, at the time, these re-g sults were questioned. Eight.more gas samples were takennon

~

April:1?and April 2. - These showed'substantially smaller oxygen i '

'depletionj(average value of 2.3%, but with.a wide scatter, some-samples showing normal oxygen levels or. higher). Later samples have confirmed that the initial values from :the , samples of March 31'are most likely_to be' valid. L(There is an apparent possibility that in-leakage of air to the gas samples caused the. error and scatteriin.ths April 1-2 _ samples.~)

Even with 50% cladding' oxidation, the preservation of much of.-the' core structure was judged to be possible. This assumed.

p .that the' oxidation of the zirconium _ cladding, produced a layer t

4 of oxide, but-leaving a metal tube intact under the oxide. "

'(Somewhat analogous to rusted iron pipe with an average of half

.offthe iron _still intact.) The near-normal readings on

_thermocouples-in_the core region also seemed to indicate that-the core' structure was mostly intact.

E

_g.

e--,- .  : .

'i f

-i, '

_:(

L 19

z -After.the situation at TMI had been stabilized, late in-

. April, 1979, the EPRI. Board of Directors authorized EPRI to set

- .up an investigation team to assess.the facts, causes, and les-

-sons-learned 1from the accident.

This led to the organizing of the Nuclear Safety Analysis Center-(NSAC) at'EPRI in May, 1979, for which I was named the.

~

' Director. In the next few months, a total.of 80 technical peo-ple were enlisted in the investigation for a total of 12 h

- man-years of. effort. This effort produced a report'(NSAC-1) issued July, 1979 on the sequence of events, with supplements fin-succeeding months. A' final report including the supplements was' distributed in March 1980. Many other investigations were proceeding which involvyd exhaustive interviews with plant per-sonnel. Thesefinterviews apparently were finding a consider-able range of conflicting recollections and perceptions. It was-decided that the NSAC study should rely on the' detailed

. analysis of instrument records and to avoid reliance on recol-lections or interpretations-by plant personnel. .

I Accordingly, we did not interview any of the plant person-nel. (We did have full support and help from GPU:and plant personnel in finding and copying any instrument records and q logs. This eventually amounted to over 50,000 pages of

. records).

Some months later, in Palo Alto, analysis of the instru-ment records brought out awareness (in NSAC) of an apparent c , >

',' v

.;g

-thermalishock to the reactor _ core, possibly from a rise of water level in the core at about 7:47 a.m. on March 28. The nuclear!' instruments-also showed a change in readings which

'could becinterpreted as relocation of fuel by slumping or col-lapse =of fue1Lrods previously'embrittled by oxidation of the

-, cladding. (I was aware of the results of two incidents in which experimental fuel was operated without adequate cooling

.and which resultedtin fragmentation of the fuel rods.)

The NSAC analysis reported in NSAC-1 suggested that roughly the upper two-thirds of the core had been uncovered and subsequently. overheated. Given that about 50% of the total

-zirconium was converted to oxide, the. local oxidation in the

upper part of the core would have to be near 100%. .The fuel cladding in this region would be almost completely converted to a ceramic oxide. The sudden cooling of a hot brittle ceramic can result'in fragmentation. The likelihood that core struc-ture was preserved in this region was then recognized to be small. From this emerged the hypothesis published in NSAC-1 report, that a region of the core shaped like an inverted bell, reaching to within about 3 to 5 feet of the bottom of the core,

was.most likely fragmented into a rubble bed.

(This analysis was confirmed conclusively only in July-August 1982, when a TV zcamera was lowered into the reactor core region.)

During the period that I was at TMI, Mr. Dieckamp contin-ued to keep'in touch with me and Mr. Levenson and to

~

C, J h~

y participate lin the. technical di'scussions after the initial 1 meetings of the: Industry' Advisory Group (IAG), to.which I re-

'ferred earlier. There was also an operating support group at TMI: led by Mr. William S. Lee (of Duke Power) for a time and-then later by Mr.: Byron Lee (of Commonwealth Edison.Co.). Mr.

Levenson and I met with .this group daily tx> review our findings and. recommendations. Mr. Dieckamp participated actively in these discussions. There were also daily meetings with the

. principal'NRC representatives (led by Mr. Victor Stello) to 1 discuss.our findings and recommendations, also with active par-ticipation by_Mr. Dieckamp.

4 11.

-.