ML20080K578

Forwards Rebuttal Outline & Corrections to Direct Testimony

ML20080K578
Person / Time
Site:	05000142
Issue date:	09/26/1983
From:	Hirsch D COMMITTEE TO BRIDGE THE GAP
To:	Bright G, Frye J, Luebke E Atomic Safety and Licensing Board Panel
References
NUDOCS 8309290248
Download: ML20080K578 (25)

Text

m COMMITTEE TO BRIDGE THE GAP 1637 BUTLER AVENUE =203 00CKETED LOS ANGELES, CALIFORNIA 90025 USNRC (213) 478 0629 Septente 26 lh John H. Frye. III. Chairman Administrative Judge Glenn O. 3right

[0C SEC G A gTp :

Administrative Judge T

Atomic Safety and Licensing Scard Atcmic Safety and Licensing ho=.rd BRANCH U.S. Nuclear Regulatory Commission U.S. Nuclear Regulatory Cercmission Washington, D.C. 20555 Washingten D.C. 20555 Dr. Emmeth A Luebke Administrative Julge Atomic Safety and Licensing Zoard U.S. Nuclear Regulatory Commission Washington, D.C. 20555 In the Hatter of The Regents of the University of California (UCLAResearchReactor)

Docket No. 50-142 (Proposed Renewal of Facility License)

Dear Administrative Juiges:

Enclosed please find the requested rebuttal outline.

Also enclosed please find corrections to C2G's direct testimony.

Not all of CEC's uitnesses have yet reported in with corrections, but in order to expedite the October hering these corrections are being provided prior to hearing.

For your convenience, replacement pages are provided.

A Sincerely, Daniel Hirsch cci service list (w/ enclosures) by Express mail to Board, Staff, and UCLA i

4 REEUTTAL CUTLINE L

i The panels are contemplating responses to the following matters raised at i

theJuly/Augusthearing:

i Fanel i

• Basic concepts of safety analysis: what is a generic analysis what is a conservative analysis what are adequate safety m rgins importance of taking into account error lars Fanel I

• Appropriate values for void and temperature coefficients,l eb, lambia, insertion rate, peak-to-average flux, etc. reliabilitys sensitivity of power excursion analyses to different input variables Eaximum reactivity worth of core and of inserted materials: rapid reactivity l

I insertion / withdrawal mechanisms

• BORAX /SPERT/SL-1/ Argonaut similarities and differences and relevance to power excursion analyses Farel II 1

r

• Combustibility of reactor materials and conditions required therefor

• Graphite thermal conductivity effects

• extensiveness of potential metal-water chemical reactions and relevance to

~

safety analyses

• conservatisms/non 6n characteristics and effects on fire and Wigner analyses

• site and core desi conservatisms in Battelle Wigner energy models correct input 4

I numbers for Wigner energy estimates i

Panel III 4

• assertions as to conservatisms in seismic LOCA analysis

• undermoderation and potential for reactivity sur6es via core disruption effects

• conservatisms/nonconservatisms in Battelle smashed fuel estimates Fanel IV i

• intermittent running, ahutdown, release point, V/Q's, dose conversion factors i

and standards, site characteristics-appropriateness of modelling assumptions and effects on consequence concitsions.

4 Note: some atters may cross over to more than one panel a

i l

l 1

i l

e,

,,w e

.,_,-,,-9

,e-

+,, - - - -,,,~

+y e--w-.c-

.,,-re*

.--y c,

r+*

c----

Errata Sheet Panel Page Correction I

9 add comna af ter 0.65% on line 4 of paragraph 29 10 "c" had been lef t out of " approaches" on line 4 of paragraph 35 18 semi-colon changed to comma, end of line 3 of paragraph 67 20 last two sentences of footnote removed 22 "a'_' had been lef t out of " assumption," line 1 of paragraph E0 24 parenthesis cark added to last line of paragrab 87 27 underlining of "would reduce" in quote in paragraph 99 recoved 28

" correcting" changed to "a correction", "at least at first" added, to last two lines of paragraph 103 31 "necessarily" added to 8th line, "would" changed to "could" I 1171 3rd sentence in paragraph 118 shortened 35 change "descrived" to " described" in I 135 37 add "being exceeded" to last sentence I 144 II 7

change "sithin" to "within", 1 29 11 change "125 and 210" to "115 and 190", line 3, I 41: remove "to 10000"' from 3rd sentence, I 41, and cut rest of sentence; 0

remove "--of roughly 120 C- " from next sentence; I 42, change "3. 3"' to "3": footnote reworded as indicated.

13 In 146 and 48, change "3 3" to "3"; 2 48, change "125 to 208" to "113 to 189": include sentence beginning " Integrating...";

substitute in footnote "**" reference to George B. Taylor, instead of George B. Bradshaw, thesis.

19 change reference for Exhibit C-II-3 to George B. Taylor thesis C-II-3 substitute Taylor thesis excerpt for Eradshaw III 1

change in P 2 to"available" from "availalba" l

Ex. sheet add C-III-9, " Photos from within NEL" IV 10 remove "k" from "km" at bottom 11 in P 23, make " panel" int o " panels" t

l 9_

3RIEF EXFLANATION CF 3A3IC NUCLEAR Ph73ICS CLNCEFT3 RELV~T) TO FCVER EXCURSICM3

26. This section constitutes a brief explanation of some of the basic concepts of nuclear physics relatec to power excursion phenomena.

Brevity will necessitate considerable simplification.

Criticality and relayed Neutrons

27. A reactor such as the UCLA Argonaut-type reactor operates by fissioning atoms of uranium-235 which, when hit by neutrens of particular energies, will split apart. In the process of fissioning, uranium-235 atoms release additional neutrons which can then cause other uranium-235 atoms to fission, creating a chain reaction, which can te self-sustaining.

29. A reacetr is said to be critical when the number of neutrens in one generation is equal'to the number in the next. In such a situation, k, the criticality or effective nultiplication factor, is said to be equal to unity. " hen the chain reaction is increasin6 with time, so that the number of neutrons in one generation is larger than the number in the previous, the reactor is said to be supercritical. A supercritical

actor is thus defined by k >l.

k(1 means the reactor is subcritical.

29 A nuclear chain reaction runs on tvo different kinds of neutrons prompt and delayed. 3y far the majority, approximately 99 355 for UCLA, are prompt, being produced virtually instantaneously at the moment of fission. A small fraction, approximately 0.65%, are delayed neutrons, i.e.

l neutrons produced by decay of the fission fragments created by the nuclear fission process.

These delayed neutrens are produced in periods of time ranging from microseconds to hours. If it were not for the delayed neutrons, i

nechanical control of a reactor would te impossible.

30. Neutron generation times are measured in milliseconds; if the only neutrons upon which the chain reaction is based were prempt neutrons, there would simply not te enough time for either human or mechanical interventien to prevent a runaway condition. Crowth in reactor power (a function of growth in the neutron population) is essentially exponential.

Even a small rate of growth from one 6eneration of prompt neutrens to the next could cause reactor power to increase to such a level that fuel melting could cccur long tefore human or mechanical intervention (e.6.

insertion of control rods) could be completed in order to prevent such a runaway condition.

Surercriticality and Premot Neutrons

31. ' Such a runaway condition, where the neutron population grows uncontrol-ably, is called a prompt supercritical power excursion.

If the power excursion is severe enough (i.e., if 6

h before a shutdcun mechanism can be activated) power gets hi h enou6

~

, fuel melting can possibly cccur as well as a steam explosion or explosive metal-water reaction. It is very. important, 4

. therefore, that a reactor not be able to run away at a rate faster than its shutdown mechanisms can respond; i.e., that it not teceme

" prompt supercritical," or supercritical on prompt neutrens alone.

32. 'dhen a reactor is running on delayed neutrons-- i.e., when the reaction needs both the 99 35% of neutrons that are prompt and the 0.655 or so that are delayed in origin-.the delayed ncutrons provide a margin of safety that peratits interventien of control rods or other shutdown features. in time to prevent an increase in power that is so rapid that melting can occur. The delay required for Eeneration of these neutrons provides time for electronic indicators to report an abnormal growth in neutron population, inform a control operator who can take appropriate actions or activate an autocentrol which can mechanically de likewise.

33 However, when a reactor is running on prcmpt neutrons alone, it has lost that protection. An increase in neutron population can occur so suddenly, and continue to increase exponentially so rapidly, that inter-vention by human response or en6 neered safety feature is not possible.

1 Thus, this situation is stron6 y to be avoided.

1

,Ixponential Increase in power in Milliseconds 34

'4 hen a reactor is supercritical, i.e. k')l, meanin6 that each generation of neutrons is larger than the arevious, the power rin is exponential. The exponential period (T) is that amount of ti te it takes the power to increase by a factor e, or approximately 2.718.

Thus, in five exponential periods, the power would rise by e5 or about 150 times for example. The ability of reactor power to rise astronemically on a very short period is thus evident, and explains why supercriticality on prompt neutrons can te so dan 6erous.

35 The effect of the delayed neutrons is to elongate the exponential period T quite substantially, givin 6 time for human or en61neered features to come into play before the exponential imperative brings a dan 6erous power level. But as a reactor approache, prompt supercriticality, the l

expenential period becomes exceedingly short, cakin6 possible massive power rises in very small fractions of a second, Elven by the following general equations t

[

= e o

where F@ is the initial power and F is the power af ter the lapse of time t.

For ver short reactor periods T, then, very large power rises can eccur in very short time intervals t.

And when a reactor is supercritical on prompt neutrons, the period T becomes exceedin6 y short.

1

36. Thus, it would be quite incorrect to assert that prompt critical is just another point on the curve. Near prompt critical, the exponential period jumps from a manageable ran6e measured in seconds or hours to l

periods measured in milliseconds, making engineered safety features such I

as control blades and dump valves potentially useless should the excursion g'o unchecked.

(Any remaining intrinsic safety features will te discussed shortly.)

t

o. has been trought into serieus question by lack of accurate calibration.

Third, the amount of centained fission products is no longer small relative to twenty years ago since the power limit has increased tenfold.

And fourth, there are a nutter of credible ways in uhich fission products can te rade to escape, including power excursions made possible by the increase in excess reactivity available and other factors.

67. In addition to the quadrupling of excess reactivity, far teyerd the prempt critical limit prescribed cy the Ea:ards Analysis ard up to tne l

precise level which its calculations indicate could cause fuel melting, l

and in additien to the tenfcid increase in reacter power, a nuater ef l

l other developments over the years at the UCIA reacter have censiderably

(

reduced the safety margins presumed initially.

Sese include discovery l

of smaller-than-expected negative reactivity coefficients (in additien to the unexpected discovery of several positive reactivity effects);

apparent lack of,deflecters, as designed, to prevent repeated criticality; enlargenent of irradiation pcrts, making possitie insertien of larger samples; and the addition of a pneumatic tube "rathit" system, which makes pessible new mechanisms for rapid insertien ard removal of reactivity.

(Other eversights in the original design review, such as errors about centusti-bility nf the reacter constituent materials and Wigner energy sterage, are discussed elsewhere; they too can impact upon effects of reactivity accidents at this reactor.)

t 68.

The nriginal design for the UCLA reacter called for substantial i

inherent safety features as well as large :argins of safety: 10 ka:n pnuer limitatien, large prompt negatim temperature and void coefficients, and excess reactivity below that necessary for prompt critical. As fer the restriction en excess reactivity to telow 0.65d k/k, the Eazards Analysis said:

"it is possible to operate the reactor with an amount of excess reactivity which is well below that required for prompt criticality.

Under these corditien, the reactor meets the safety requirements of a training reacter ard can tolerate considerable operational error without damage. (p. 19)

If the reactor ever met those safety requirements, it no longer does.

IEE DAUGER OF EXTRAFLIATING, VITEUT VERY IARGE IRRLR 3ARS vR SAFETY MARGIys, FRCM 3FERT AND 3CRAX TESTS TO THE UCIA ARCONAUT

69. UCIA argues that nene of the alterations or problems that may have ecciarred during the reactor's operating history to date are of consequence because the reactor is protected by inherent design against significant fission product release. In particular, UCIA argues in its license renewal request that its reactor can safely tolerate a far larger excess reactivity insertion than the reactor's original design limit. UCIA appears to rely heavily on an assertion that the ECRA7 I ard 5FERT I tests corducted at the NRTS in Idaho in the 1950s ard 19c0s " proved" that the requested level b

of reactor fuel (e.gt, flat plate) could survive a $3 50 insertion in any imaginable reactor.

74 The important question, then, is not what reactivity insertion destroyed 3FERT or 3CRAX or SL-1, or even uhat insertion could te expected to te the minimum necessary to induce melting in those reactors, but rather, what level is a safe level for the UCR Argenaut, with sufficient margins of safety consonant with student operation in a densely populated location. Af ter all, SPERT, 3CRAX ard SL-1 were all destrcyed in the_

Idaho desert far from any populated center. And the CCM Argonaut-type reactor is a substantially different reactor than the three Idaho reactors mentioned above.

75 The differences are significant. Plate and meat thicknesses are different, as are coolant channel widths. The STERT tests us&d essentially fission-product free cores, with fresh cladding.

UCI.A's fuel has teen irradiated for tuo decades, can te irradiated for another two decades if relicensed, and has been sitting intermittently in water, allowing for some degree of corrosion, for many years. Each of those factors might affect the heat transfer time to the water *, potentially elongating the transient and increasing the energy release, factors not analyzed in the existing reports.

76. Furthermore, SIERT and 3CRAX were entirely water-nedorated ard

-reflected, as was SL-1.

UCIA's reactor is moderated by toth water and graphite, and reflected by graphite. This lengthens the neutron lifetime, producing a longer period for any given reactivity insertion, but it also significantly reduces the value of the shutdown feature caused by expulsion of the water portion of the moderator.

In the UCR case, part of the moderator ard reflector, i.e. the graphite, cannot te expelled from the core during the normal course of an excursion, thus reducing the effectiveness of moderator voids in limiting the peak power reached.

And further, the reported void coefficient is smaller for UCR than 5FERT or 3CRAX, as is the temperature coefficient for the water portion of the mederator.

"he positive coefficient for the graphite further weakens the si::e of the shutdown mechanism for UCLA, and the positive reactivity effects noted when water level initially drops in the core and when fuel plate spacing (and/or bundle spacing) is altered, as by oscillation, are other important differences.

77. These differences can te very significant in determining the energy release from any particular excursion and uhether fuel melting will result.

Even different reactors of the same general type produced widely different energy releases for the same period, as is shoun in the plot of energy versus reactor period on the next page, taken from Thompson and 3eckerley's A potentially significant factor not considered in the analyses to date is the reduction of thermal conductivity in the fuel due to irradiation.

I t

Technolerv of Nuclear Feactor Safety, p. 675 As is shewn there, 3CRAX produced substantially more energy than SPERT, ard 3L-1 acre than either, given the sane initial reactor period.

(This is an important reason why esticating the energy release for an excursion of a particular period at UCLA directly from the release for 5FERT at the same peried is se non-conservative-- the same period produced far higher excursions in otherreactors.) Seemingly minute differences in metal-water ratics, temperatures and void coefficients, etc., had marked effects on total ener6y released.

78. This is understandable when one realizes that the process of a power excursion is essentially exponential.

The nature of the ex;cnen-tial rise is that very minor decreases in exponential period (the "e-foldir.s time") or increases in total time of the excursion (by delay in the shutdeun mechanism) can cause the power to increase by large a: cunts.

Thus a delay of a few millisecords in the transfer of heat from the fuel meat to the clad ard then to the ecolant (caused, for example, by thicker fuel plate er lowered thermal conductivity because of cerrosien or irradiation) can mean the difference between an excursion terminated safely ard one resulting in melted fuel and substantial fission product release.

Thus, minor errors in calculation or extrapolation can have potentially disastrous results. -

79 In the absence of actual 37ERT-type excursion tests with an Argenaut-type reactor, it is understandable perhaps that hazards analysts would attempt to extrapolate from the excursion tests that have been performed, albeit en reactors of different type.

Thus UCLA's cun 1960 Hazards Analysis, the Hawley e_t, al review, and the Necgy memorandum all rely en the power excursion tests performed at the NRT3 in Idaho. UCIA relies largely on the 3CRAX tests in its original analysis; Eauley et al on the SFERT ID series of tests; and Neegy on the 5?IRT IA series'.--

(Surprisingly, none even touches on the SL-1 accident.) All are based on the fundamental assumption that one can extrapolate with extrenely high precision frem the 3?ERT or 3CRAX tests to the UCLA Argonaut.

80. We take substantial issue with such an assumption. First of all, I

the 3FERT tests were not intended to be used in such a fashion.

5?IRT was an attempt to unde. stand the mechanisms of shutdown in peuer excursions, not to produce an absolute number that could be plugge:i into reactor analyses for significantly different kinds of reactors.

In particular, it was never intended that a hazards analyst would simply look at the exponential peried at which some melting uas expected to tegin at 5?IRT and say that therefore substantially different reactors could safely handle precisely the same period.

The 3FERT tests simply do not permit such extrapolation to :iifferent reactors without an extremely detailed accounting for differences between the reactors, which is very difficult to do, and very significant error tars to take into account the si nificant 6

uncertainties in such extrapolation.

81. If the SPIRT core was destroyed with a $3 50 insertion, it would have been of considerable cencern if a reactor operator used that fact as tasis for a $3.40, or $3.00 linitation for another reactor, particularly of a different type and in an urban environment.

The SPERT tests were never interded to be so used-- the uncertainties are just too large.

To say, as the Hawley et al review essentially does, that the 3FERT ID core indicating melting beginning around a 7 msee period meant that the UCLA Argonaut coul:1 tolerate a 7.2 msec period excursion without any melting

. zeltin6 point...

"The first step in the procedure is the esti=atien of the ex;cnential peried corresponding to the excess reactivity which uculd have characterized a power excursion of similar effect in 3CRAX I."

(p. III/A-3, emphasis added)

67..The Analysis then attempted to correct for the different void

• coefficients, coolant channel width, fi6ure of merit for fuel perfornance, and peak to average power ratio, cencluding that the limiting excursion for UCLA is 9.1 milliseconds.

Correctin6 for the different prompt neutron lifetimes, it was stated that that period corresponds to an insertion of 2.3% Ak/k.

(It is interesting to note that the Hazards Analysis estimated that the UCLA reactor could tolerate a considerably realler pouer excursion in terns of energy release than could 3CRAX, because of the different characteristics of the reactor-- 41 Ew-sec, plus the energy to bring the water to saturation, as the limit for 3CRAX, and 26 EW-sec for UCLA. Conversely, 3CRAX was stated to reach its limit with a 6.7 msce period, UCLA with a 9.1.

This shows the problems with assumin6 that if 57FRT, for example, could tolerate a 7 msec period, so tco would UCLA.)

l As the ori 1nal Hazards Analysis calculations make clear, 2 3%d k/k 88.

6 would be sufficient' to cause fuel melting at UCLA, if the assumptions employed are correct.

We have made clear above our objections to such There is some confusing language in the text of the Analygis on this point. The calculation.s make perfectly clear that, if the Analysis is correct, a 2 3% reactivity insertion will brin 6 the hottest part of the fuel meat to the celting point of aluminum. Yet it is stated at one point that the reactor will tolerate a power excursion of at least that ma6nitude without melting occurring at the hottest part of the fuel.

This is primarily a semantic difference, asserting that a certain estimated point is the end of the safety zone instead of sayin6 it is the be61nning of the danger zene.

Some of the confusion can be traced to the fact that UCLA cepied its raised to 2.3%) ysis Report (by which time the reactivity limit had been 1980 Safety Anal from its 1960 Hazards Analysis (at which time the limit was C.65), which in turn was copied from a 1959 AMF analysis, which in turn was copied frem a 1958 analysis for the University of Florida reactor.

(seeattachments). A comparison of the analyses indicates that while the lan6uage was copied virtually ver':atim, there was a significant difference between the University of Florida reactor, upon which the original analysis was based, and the UCLA reactor.

The fuel at the former was 20% enriched, 90% enriched for UCLA:

the U of F fuel was 46 w/o U-A1, whereas UCLA's is right at the eutectic point, 13 4 w/o.

(See pa6e 1 of U of F and UCIA's " Estimation of Effects of Assumed large Reactivity Additions.")

The uranium-aluminum alloy in the U of F fu:1 meat melts considerably above the melting point of aluminum, unlike the alley in UCIA's, which melts below the critical temperature of aluminum.

1 Furthermore, the U.of F fuel had a Doppler c?ntribution to shutdown, since it was LEU, whereas UCIA practically does not. The 1960 UCLA 1

i p

.. as initial moderator temperature, to name just a few.

Al+ hough the Analysis conservatively assumed the 2 3% 6k/k insertion to occur in a subcooled reactor, the Fawley review at p. 15 rightly points out that excess reactivity is normally measured at normal operatin6 tempera ures of the reacter and that negative temperature coefficients for the water would make, for example, 2 3% at operating temperature actually much more at lower-than-normal temperature. Conversely, if 2 3% is dam 6erous en a cold day, far less than that amount must te installed if measurement is under warm moderator corditions.

96 Ard, as discussed in more detail later the positive coefficient.for the graphite can likewise mean that 2 3%dk/k measured when the graphite is cool can result in more than 2.3% Ak/k teing available after its temperature has risen. That factor, plus positive feedtack effects in an excursion (such as the positive coefficient for the graphite, the positive void ccefficient in a portion of the water mcderator, ard the positive effects from chan6es in plate ard burdle spacing that might accompany the initial stages of the excursion) further damatically reduce the " safe" level.

Proper inclusion of adequate error tars fer the various steps in the calculation, pushes the level even further down.

97. Thus, given the tasic assumptions employed in the Hazards Analysis, and the numerical values utilized, the Analysis' calculations predict fuel melting with insertions in the ran6e of 2 35

~4 hen a few of the numerical values ar : hanged to reflect more appropriate values (e.g.,

/2, veid ccefficient, and eutectic melting point), substantially less than 2 35dk/k uculd appear to be sufficient to irduce melting-- if the method-clo61 cal assumptions employed are correct. If other facters are included, even smaller levels are tolerable.

98. There are problems, as indicated at the outset, with extrapolating from one resctor to a different one-- to three significant figures--

witheut error tars. ~his assumes that there exists a complete knowledge of all the differences letween the reacters ard how those differences precisely affect behavior. As has been shown, a number of differences were not censidered, and to assume that uhat differences are considered can te corrected for using simple linear relationships is inappropriate for the level of precision assumed.

For example, the Hazards Analysis assumes a linear relationship between void coefficients and total energy release, which is unlikely to be correct, 61ven the exponential nature of energy release in a power excursion.

99. The Ea a:ds Analysis merely declares that the 0.6% k/k limit has a reasonable safety margin to compensate for the potential errers in extrapolating from the 3CRAX data. It is filled with terms describing the calculations clearly as estimates ard extrapolations, based en unverified assumptions:

On the assumptien that this minimum value is the true value, a rise of water temperature from near 00C to c0cc would reduce reactivity by 0.6% keff.

III/A-2emphasisadded

. _ 103 The section of the Hawley, el al, report dealing with excess reactivity

,[

issues appears to consist almost exclusively of a brief literature review and sone extrapolations from the SE RT I tests. 'ihereas the 1960 Eazar:is

+

Analysis took into account a number of differences between the UCI.A Argonaut and the 3CRAX reacter, from which it was extrapolating its data, the Hawley f

review does not account for several of the UCI.A-SPERT differences, partic-ularly UCI.A's smaller void coefficient, which would tend, if not otherwise com-pensated, to suggest that an excursion of the same period in 5FIRT and the Argonaut would produce greater energy release at UCLA. The Rawley report's primary consideration of differences between the tuo reacters consists

)

of a correction for the longer neutron lifetime at UCLA, a factor which is, at least at first, helpful to UCLA.

a 104 The Hawley approach was extremely simple-- calculate the period l

produced by an insertion of available excess reactivity, estimate the i

energy release an excursion of similar period would have preduced at SEIRT ID, and then scale temperature linearly to the peak temperature estimate during the 5FIRT ID destruct test.

105 And yet, even without taking into account factors such as void coefficient differences, which would tend to produce higher temperatures, 0

{

the analysis estimates peak fuel temperatures only about 50 telow the melting i

temperature. No error bars whatscever are provided for the extrapolation steps nor for the final conclusion.

(There appears to be a subtraction error in that Eawley et al assert on page 19 of their report that a hot j

spot of 5860C would be 7DC below the melting point of the fuel teat, j

which they cite on the previous page as being 6400C. )

0 106, 50 is not an adequate margin of safety, particularly when so many cf the differences between SFERT and the UCIA Argonaut were not taken into l

account. Furthermore, significant effects may appear just telcw the melting point, such as volumetric expansion of the fuel resulting h j

cladding failure, or censiderably increased diffusion of fission products i

through the hot metal. It was noted at SPIRT, for example, that seme.. the fuel plates were very substantially softened and warped, even though not i

truly melted, and that they would stay-in that softened form for several j

days thereafter, tehaving something like a wet noodle. This was prior to j

the final destruct test.

107. So even if Hawley d al were correct in their estimate of peak temperatures 50 er so degrees below the melting temperature, there would l

still te concerns. However, questionable assumptions used by Hawley ej al suggest far greater temperatures could te achieved in the UCIA Argenaut than those estimated.

Questionable Assumptions 108. Ferhaps the most questionable assumption is that a 7 2 maec period l

would produce a 12 E4-sec energy release in the UCI.A Argonaut. Given the linear scaling assumgtion of temperature to energy release employed by Eawley (p. 19: 1500 C per 30.7 E4-sec. er about 49 C/rt-sec), a 13 ri-sec energy release would cause melting, if "awley's assumptions are accepted.

3at is not much of a margin of safety if his 12 E4-sec estimate is correct.

-n

---

e-r~-

w

--~--r,

. 117. Taking the example given above, and assuming a very modest difference of 10% in speed of shutdown, representing a feu milliseconds, one additional e-folding period would occur at UCIA before shutdown than at SPERT, from which Hawley obtained his 12 ?X-sec estimate.

This could mean, thus, peak power 2 7 times hi6her, just because of a delay of a few thousandths of a second in transferring heat to the coolant, voidin6 the coolant, or the reactivity worth of voiding the. coolant.

In other words, a few percent less prompt or less effective shutdown mechanism does not necessarily mean a few percent higher peak power, but because'of the exponential nature of the rise, could mean several times higher peak power.

118. All indications are that the shutdown mechanisms for UCIA could be substantially slower and smaller in effect than those of the SFERT or BORAX reactors with which they are being compared.

The 1960 Hazards Analysis made clear that just correcting for a few of the differences between UCIA and ECRAX, the minimum period UCLA was expected to be able to tolerate was considerably longer than that estimated for ECRAX. The void coefficient is smaller, which is quite important, and potential effects like reduction in thermal conductivity in the fuel caused by corrosion or irradiation could elongate the time interval for the heat generated in the excursion to be transferred to the moderator for eventual shutdown. Given the exponential nature of the rise, and the exponential period measured in milliseconds, delays of a millisecond or two in transferring the heat, and differences of a few percent in the effectiveness of the voids formed in the coolant, mean melting could occur substantially below the reactivity insertions assumed by Hawley or the original Hazards Analysis.

Based on the analyses done to date, insertion of either $3.00 or $3 54 must be considered a credible cause of fuel melting.

119.

It should be noted once again, however, that the methodology of very simplified extrapolation from SPERT or EORAX data to the UCLA Argonaut case, as done in the Hawley report, seems most inappropriate given the differences in the reactors and the difficulties in correcting for those differences.

The SL-1 accident, which took the lives of the only people nearby at the time, was "non-credible" in Hawley's terms, yet it happened.

It released several times more energy than Hawley's extrapolations from 5FERT ID would predict, even though it was much more similar to SPERT than is the UCLA Argonaut.

The Haalcy extrapolations cannot be relied upon to prevent an SL-1 type accident at UCIA, one that would occur not in a remote federal testing station but in the midst of tens of thousands of people.

IT. Ostrander, in his September 1,1982, declaration, at page 10, asserts that the reason why 20RAX data suggest a so much larger power excursion for the same period than does SFERT (and why he believes it appropriate to ignore the more conservative ECRAX data) is because of different active core height producing hydrostatic pressure and inertia forces which impede boiling more in the 20RAX case. This is an interesting hypothesis un-fortunately, its validity has not been demonstrated.

However, assuming for the monent that it is correct, such an effect may well be very unfavorable for UCIA' because among the many differences between SPERT, BCRAX, and CCLA, a clear one isthat the former were open tank reactors at atmospheric pressure. There was nothing to impede the expulsion of the moderator out of the core.

In the UC LA case, the moderator is in a closed system in order for the coolant to be expelled, a pressura pulse must be generated in the core region, transmitted through the coolant

1 l

~

could result in positive reactivity being added in the midst of an excursion which might not, of itself, be sufficient to cause melting. Similarly,

)

expansion or bewing effects that increased the plate spacing could push l

an excursion "over the top," as could the initially positive effect noted j

upon dropping the water.

135. There are numerous other possibilities as well. cne entails a power excursion not sufficient to cause melting by itself but which does involve expulsion of the water moderator. It was noted with the l

SPERT reactors that such expulsion would on occasion lead to repeated criticality as the expelled water condensed and dropped lack into the core. An excursion limited by moderator expulsion, as at SPERT or BCRAX, can send a plume of water and steam high in the air. When that water returns, it does so at a significant velocity, which amounts to a very rapid insertion of substantial excess reactivity.

Such behavior is l

called " chugging", and on several occasions incidents occurred in which the initial reactivity insertion was not sufficient to cause damage, but the repeated excursions caused by repeated reintroduction of the moderator after expulsion caused increasingly larger excursions which, had the event not been terminated through scramming the reactor, might have essentially torn the reactor apart.

(A history of sticking control blades which could make final termination of such a series of excursions impossible would thus have safety significance. Similarly, the lack of deflector

• plates described in h original Hazards Analysis as designed to prevent such repeated excursions by preventing expelled I

water from returning to the core, means that an important safety feature I

is missing.)

136. The positive temperature coefficient for the graphite is troubling as well. A research reactor used by students needs to be inherently safe. Inherent safety necessitates large negative temperature and void coefficients. Any positive coefficients (which are thereby autecatalytic) are to be strongly avoided. This is especially true when W value at-tributed to the positive temperature coefficient for the graphite

(+0.006%ok/k/0F) is larger than the negative temperature coefficient cited for the water (-0.0048% Ak/k/'F).

137. During a power excursion the positive temperature coefficient of the graphite could provide a feature which makes the excursion more destructive than would otherwise be the case. A portion of the energy lib-ersted in a power excursion is given off as prompt neutron and gamma radiation, resulting in a prompt temperature rise in the graphite and oth.

surrounding materials bombarded by that radiation. Even a few degree i

rise in the graphite temperature would mean the addition of positive f

reactivity at a time when n*gative reactivity is needed to limit the power excursion. The addition of even relatively

===11 amounts of positive reactivity can produce a slight delay in the shutdown mechanism taking hold because of the exponential nature of the excursion, even a milli-second additional delay can be significant. Given the extremely = mall margins 0

of safety, e.g., Hawley's 40-50, even assuming all the sssumptions made are correct and the absence of other uncertainties, a slight addition of positive reactivity during the excursion can cause a small margin of safety to become far an=11er.

l 138. Hawley (p. 15) has pointed out that excess reactivity in Argonaut-type reactors is usually measured under normal operating conditions and that the negative temperature coefficient of the water thus makes it possible that a reactor with a measured level of excess reactivity of, say, i

~37-343 Thus, factors such as those discussed above could mean that the reactor, at the time when reactivity was measured, was below the licensed limit, but at other times, due to positive reactivity effects, was above.

Furthermore, the existence of an excess reactivity limit in the license does not mean that that limit will not otherwise be exceeded, unless the reactor's inherent design does not permit any more reactivity than that level, which is not the case with the UCLA Argonaut. Reactivity is controlled by the amount of the fuel and the effectiveness of the moderator, both of which are easily modified in the Argonaut.

(Note the large quantity of heavy water stored next to the UCIA reactor, for example. )

UCIA has a substantial quantity of spare fuel on sites reactivity is readily added by removin6 dummy fuel plates from the core and replacing with actual ones. This is how refueling is done to compensate for burnup and other factors.

144 Therefore, the fact that the Technical Specifications may contain a limitation of $3 54, or $3 00, on excess reactivity does not mean that that limit will not be overshot from time to time, given errors in measurement or violations of Technical Specifications. A history cf measurement errors or Tech Spec violations at such a facility would i

substantially increase the probability of excess reactivity limitations being I

exceeded. I 145 In'teractions may potentially occur between power excursion accidents and accidents of other types. For example, various core disruption events could cause or contribute to positive reactivity insertions. Flooding, be it by pipe break or other event, could add moderation (because of the dangerously undermoderated nature of the reactor) and thus cause a positive excursion. Core-crushing could move the core to more of an l

optimal arrangement for moderation. Seismic jolt could cause a negative sample, or a control blade, to move out of the core rs5 on. An event which 1

caused the fuel bundle spacing or fuel plate spacing to alter could like-wise contribute positive reactivity. A saml1 seismically-induced excursion, not sufficient in itself to cause melting, could increase the maximum more detail in the panel on core disruption.)(These will be discussed in fuel temperature reached in a crushed core.

i 146. Fire could likewise cause some positive reactivity effects. Were i

the low-melting cadmium control blades to melt out of the core region.

a positive reactivity effect could be observed. Were the graphite to heat up substantially, the positive coefficient could add reactivity to the core. Were firefighters to use water (or perhaps other moderating substances) to fight the fire, a6 min a positive insertion might result.

A power excursion could provide the initial heat necessary to start such a fire. These matters will be discussed in more detail in the chemical reaction panel, as will be the steam explosions and explosive metal-water reactions which have accompanied several destructive power excursions such as SL-1, 3CRAI and SFERT. And a power excursion of insufficient mag-i nitude to melt the fuel by itself may be sufficient to trigger Vigner energy release, which could add sufficient energy to either melt the fuel or ignite parts of the core.

CCNCLUSION 147. The UCIA. Argonaus, in its current configuration, is not inherently safe. Because of the large amount of excess reactivity, and features by

_ _ _ _ _ _ - Fire Scenarios

29. De Hawley report presents a nuator of potential fire scenarios.

Among them welding torch accidentally igniting outer graphite power excursion sufficient to ignite a flammable solvent ( a conson mode scenario for this event would be a power excursion caused by breakage of W sample container in which a large sample dissolved in solvent is being irradiated removal of the neutron-absorbing material from the core could initiate the power excursion which, even though perhaps insufficient to melt the fuel itself or 1 nite the graphite, could ignite the solvent with its lower 6

flash point): nuclear heating of inserted natarials "to a temperature high enough to ' ignite various flansable substances seems well within the realm l

of Possibility": building fires and so on.

One can suggest numerous others i

as well, but it is sufficient to indicate that the reactor is not inherently j

protected against fire.* Re Hawley report indicates that a number of these scenarios could put the fuel at risk if proper and prompt response were not ande to suppress the fire. De report also indicates that because graphite produces little smoke when it burns, the fire might go unnoticed for sub-stantial periods of time. Jhere is no procedure in the energency plan for actually fighting a reactor fire.' Given these factors, a reactor fire can occur and can put the fuel at risk. Fires are conson events.

30 De NRC Staff has asserted that a graphite fire in the UCIA reactor would occur only if an experiment failed and a general building fire occurred and the reactor's graphite blocks were exposed to a free flow of air.

The Staff cites pp. 41-43 of the Hawley report. We anst not be randinf the same report. Page 41 refers to a credible scenario in which a Wilding fire occurred while the shield blocks were removed: there is no mention of r.he necessity of a failed experiment as well.

Credible common-sede causation is suggested by the authors. Page 42 of Hawley describes a credible accident scenario caused by a failed experiment alone.

De bottom of p. 42 con-tinuing onto p.43 describes another credible scenario, a simple building fire while h shield blocks were removed.

De Staff appears to have misread its consultants' report.

Sufficient Airflow fer a Fire 31 De current Safety Analysis Report of UCIA (1982) no longer makes the mistake of its crisinal Kazards Analysis in denying the combustibility In addition to the reactor itself catching fire, significant hasa.eds could

+

occur through radioactive release due to other kinds of fire at the facility.

For exmaple, there could be considerable danger if the plutonium source at the facility uns involved in fire.

Other radioactive substances (for example,

" hot" samples that had been irradiated in the core) could ignite, either in-core or outside. A fire in the " rabbit room", where the samples return after being irradiated in the core, could be guite serious.

(See photos of the rabbit room, showing plastic bags containing hundreds of plastic vials coa +=4= ins radioactive samples, being stored.) A fuel h adlins accident could likewise involve fire, were, for essaple, the fuel placed in a vat et solvent to clean off surfaces for inspection and were the solvent to ignite.

. 40 1he Hauley, Kathren, and P.obkin review treats the Wigner matter in two brief paragraphs on page 37 of their report. They conclude that the amount of stored energy that may have accumulated in an Argonaut-type reactor like UCIA's is approximately 5 cal /g, which they indicate is insufficient, to heat the graphite by more than a trivial amount.

41.

The Hawley. et al. estimate, however, is low by a factor of at least approximately l 25-40 The true level of Wigner energy that may be stored in tne graphite of an Argonaut-type reactor such as that at UCIA is between 115 and 190 l

cal /g, given the calculational assumptions employed in the Hawley report and substituting numerical values that are more correct for the UCLA case than those used by Hawley. Such a level of stored energy is sufficient, if released, to raise the graphite temperature 6000C.

In sum, an_ incident involving a relativelv modest initial temnerature rise in the graphite could te sufficient to trirrer release of sufficient Wigner energy to icnite the craphite or otherwise put the reactor fuel at risk of igniting and/or melting.

• 42.

The lhuley report underestimation is caused by a series of. cumulative 0

errors. First of all, the value chosen for the rate of energy storage at 30 C is low by a factor of between 1.2 and 2.

Next, the ratio of energy storage at 50 C to that at 30 C is low by about 40%. In addition, Hawley uses 0

a thermal flux that is low by a factor of 3, insed on empirical measurements I

at UCIA. And he estinate a total operating history of 12 FN-days, thereas the UCIA reactor has already run 19 FV-days in its first 20 years and, if relicensed, can run an additional 37 FV-days through the licensed period, given the operating restrictions at the facility. This is a further error of 4.7.

The cumulative effect of these errors (1.2 x 1.4 x 3 x 4.7 = 24 to 2 x 1.4 x 3 x 4.7 = 40), a factor of 24 to 40, de ending on which initial valueischosenfortherateofenergystorageat30gC,isquitesubstantial.

The errors are discussed in more detail below.

43 The Nawley report takes the value of.5 cal /g per FV-day /At as the best value for the rate of energy storage in graphite irradiated at 30 C, citirs Nightingale's Nuclear Graphite, p. 328.

However, on page 345 of the same text (attached), Nightincale states that "more accurate" values at low exposures range from.6 to 1.0 cal /g per FV-day /At.

44 In order to correct these rates for the somewhat higher temperature he says is fourd in the Argonaut's graphite, cited to be approximately 500C, Hawley uses a correction factor of 3 5ths. Data given by Nightingale (p. 330) for the change in the rate of energy storage with temperature, however, when graphed (see next page) produce an actual ratio of 5 6ths (inverse 1.2).

This yields storage rates of.5 to.83 cal /g per FV-day /At at 500C, as opposed to the.3 assumed in the Hawley report at this stage of the calculation.

1.o., assume graphite is at its normal temperature and some incident raises 0

that temperature, not 600 K to the ignition point, but rather a mere 1200 to above the temperature at which Wigner energy is rapidly released.

Ascuming adiabatic conditions, the released stored energy could b sufficient to raise the graphite to 10000K, above the ignition temperature of the graphite or the ignition / melting temperature of the fuel. A higher initiating temperature will, of course, result in an even higher final temperature.

1

. - _ 45 Using the equation given by Nightingale relating thermal flux and )Xd/At (p. 328 of Nightingale), Hawley then obtained a rate of energy storage in the UCLA reacter.

De Nightingale approximation

• is:

Therral nvl (EEPO equivalent) 6.4 x 10 7

=

FX-day /At in the UCLA reactor of 7.8 to 13 x 10- 9 cal-cm /g y storage for graphite Inserting the correct values yields a ate of energ 2

n, compared to Hawley's value at this stage of 4.7 x 10-19 46.

Hawley t en attempted t.o estirate integrated thermal neutron flux (nyt, in n/cm ) in order to convert, through the approximation provided above, into cal /g12 To estimate integrated flux, Hawley assumed a flux 2

n/cn-sec." This order of magni de estimate was rate of "about 10 quite crude, as Hawley assumed the flux to be 1_ x 10p' whereas actual teasurements made at UCLA indicate thermal neutron flux as high as 3 x 1012. **

47.

Hawley then assumed that the reactor had locged 120 full power days, in order to estimate integrated flux (i.e., flux in n/cm2 per second as determined in 46 above, times number of seconds, to produce n/cm2 integrated dose.) However, UCIA reports (Amended Application, p. III/8-7) that it had logged 19.4 FM-days (or 194 full power days) in its first 20 years.

In addition Hawley failed to consider the next 20 years for which UCIA has requested the license. At a 5% operating limitation, as in the Technical i

i Specifications, that would be approximately an additional 37 PN-days, for a

{

l total of about 560 full power days to the end of the licensed period, in

{

}

contrast to the 120 assumed in the Hawley report. ***

I I

48.

Inserting the more correct integrated thermal neutron flux into the relationship obtained from Nightingale in 45 above one gets a potential stored energy of 12/cm2-sx7.8to13x10-19 cal-cm2 560 full power days x 86,400 sec/ day x 3 x 10 n

~

yielding a potential stored energ/y of 113 to 169 cal /g of graphite.This is in sharp contrast to the 5 cal g estimated in the Hawley report. Integrating over teperature the specific heat of graphite, one determines that such energy corresponds.to a temperature rise of approximately 6000C.

• Hawley does not demonstrate that this approximation from Nightincale is universally applicable.

It is used here only in following the Hawley methodology in order to demonstrate that given the methodological assumptions employed, but using more correct numerical values, a substantially different result is obtained.

• • "The Neutron Flux Distribution Within Three Energy Groups of the UCLA Educator Reactor" by George E. Taylor, Fasters Thesis,1962, p. 95 The study measured neutron flux at a series of locations in the graphite.

Furthermore, there appears to be some uncertainty as to the past irradiation history of the UCIA reactor's graphite--whether, for example, it might have been previously used in another reactor p.rior to the con-struction of the UCIA reactor. Thus, the true maximum exposure may be greater than,the 56 FN-days assumed here.

, CHEMICAL REACTIONS Exhibit List Exhibit Number Description C-II-l "Ihe Windscale Incident", by C. Ro6ers McCullou6h C-II-2 Nuclear Graphite by Nightir16 ale (excerpts)

C-II-3 "The Neutron Flux Distribution Within Three Energy Groups of the UCLA Educator Reactor", FS thesis, by G.3. Taylor, 1962, flux plot C-II-4 L.A. Fire Dept. Emergency Response Plan for fire at NEL C-II-5

" Aluminum Tank Explosion" (RCE 70-3)

C-II-6 Photos taken within the Nuclear Energyilab l

l I

l l

l

EXHI3IT C-II-3

p. 1 of 3 UNIVERSITY OF CALIFORNIA, LOS ANGELES IMC NEOTRON FLUX 01STRIBUTIGN #1 THIN THREE ENERGY 3ROUPG OF THE.'CLA EDUCATCR REACTOR J

1 A Thesis susulvTED 1:4 PARTIAL SATIOFACTION OF T r4 E RERVIREMENTS FOR THE DEG9EE MASTER OF CCIENCE IN ENGlNEERlNG BY I

3EORGE 3RUcE TAYLOR

\\

AUGUST, 1C6g I

t r

e I

(

1 L

I o

.rwww.w.m.mn..

EXHILIT C-II-3

p. 2 of 3 l

\\

l (INE TMCGIC Or '3CORGE 99UCE TAYLOR IS APPROVCO:

I L

4 f

s 0 m r ~ x)

- 6 g

A

?'I j

~

1

) fe

'\\ m s,n +1LM t

v w C '.:*.1 I T T E C J1 A I R t.t A ta l

l t'

t.

U 'l l V C R u l T / OF CAllrCANI4, l.C 3 A '4 3 C L C C A U '3U C T, 1#$2 i

4

?

l e

x s-.-m sm.m

EXHIBIT C-II-3

'P. 3 of 3 J

~

m:.=_. g..=...#.. a..p. m 2.+. v. =. - t'. =..+=- i -,. a.. - :-.=.=- w--90. c:4. 9: w-+=-

• E.-

r-a r-

=*.- E. h. 4 = ?=.t_:: =L.

• ~.' 3.=';* W.... -[; Y f."L-

"~

~ :.. ; ^I.'V. ',.R: :- L. n._ '.f P- - _.. - a. ~.:.W.*=*1:- f... ? *D :..

• . T

• - : :.--. z-.+- - t -:-

K*'.*-.%.L.L f.* * '

-M.14..

g t

L.2 -

w^

. a..;

• 4.=;^=

.C

.^

',.k f

. - -:~.

-*^t-----

• V ' M '

c _t

r. r -

-~

- *

• M '.~:ba.n :.: tur-+b':.-.7.0 c t:ricr L.J-: t.2 =

7

' ss

• - * ~ - -

,J. -.-"4* ~* *:-' "!.

• L'*.':~ *-*'J :..

a---.a-*^

j..-$

4 Mi - C.N' : ~~ X': _... - 32f-~ ^ * ---

C

.'_.-6.---

D.

-r

.. r u. _. - -. - :: _-+g, W._ _

z.<-.=.:.: :.. t' 3.._. - $ m.r--'

- ~.-.. r.. : --.= = ~. = f- =-

u.:=

.p

-- - m-

. _3 - -~

r --

3

.Qet 1,.d

~1.7--n~*--*'..----

. -.-~

i. E

. 7..~~

N

.(

- *-

• I* -

• l

.i s..

. T uA,.4, g s.L..

.=6.muns.wA.c.L a

--._45--...

r r...

.*.o H4

.-u

..*.6-~-

- - - ~ ~

-._p..-

,. 5. L s

- - ~

r-n

..~

. 4

_...a

..:. T C. r

-~ ::....-

{..:

.r.c. : :n_ =. - %r........",- :.;.: : ;. 7_..

r.,i _- -... - }. :.. ?. n

(*.

.:. 4_-..... :-._. _.7..

a

..*.... r_

s I

n.... f..

L.

3 1 !.:. -

.2._.- :* m...

..l.

N.....-~l- *~ 4.&. - ---..-

.t w

>w

.. - *. 3g.-....

g w

(g

--M

.%' r.-

.-\\..

c...

I..

\\i.

..l..

4-....

6 a s

A.

V

-s-.

l ;.

e Ao

,e

.--e..'

M

}.

=)

. a.

- = =.. ~

/ %......

... } p.~.... _

sh.,-

.)

. Jp l.[.. _....... _ _

-... p+...

~

.u.

• Q%

.1 y

. g.:

y

,3

..}.

p.- (

--. \\.. -- - -

m.

s... _

- -. ~

..\\.

.... -. ~.

4__

s f.\\g..

j i

4-

.i

,i 4

iei

../

1,g

,. - [l..

.g=

t i

S W'-

~..

e i-

.u :.

.t.

l. f.... f a a_r.

. + = ' -.W

.b u "ib y W

== "

- [

g j

j f -

_y

.[

~

5 E-it:-t-plE-s l

h;

.._ i=% 2i-WM d==

L-t?

i==W.M. tM - ~.+ #W;&T.~.=

i t

i.-

I

... '.t.

t'..-

s s

..L r n $.... :

g

.........-.r.

.. --. M..t2 f.r. ;;. :. r

\\

2%

6

..p....

-s..

31, - (

!,. h 33

...._l...._..,

v

-. +..

.......)

i,.

s

...f.,.-.--

E.

3

. ~ -...

h.,

. -. +. -

-.s....-

o, r,sf.

2

.w

-l

.~.....i....

i.

i, i

D / C$n =..

6..

1.. v l_....

.. j e.

l 4-----

..l

...-..-.... - {

.}...-..,.*.-...

~ g

)

m.-

r I;

~

i

- '.:. [ f _ i :f

- [..........,

2...-

4,.-.{_n _,_~.

g

.. g

, t.fu..

gg 3__

j

.. j/

. [.

[.;...

.. t.

y g

.. :t.g. g_,,

.qg,

.[-...

.Y v,.,y,

.. ~...

.p s

.. _... r.

g.,.

t,.

..o

..s

.g i

e.,

e n

y

[

...-.W_.

J.

s.,.,

p..

... ?

_.1

...l....

-.1_

._4 l.

.J..._....,l_..

~.L. v i W.

...g-

yl

..r-..--

'.. /

1 p.i.

k r.,n,,.

i

i.. ; f....

. V.. ;l....

o,

. L

.i, _ s....

g g

i

,,.7. ' *'%-".. -

p.

y

...__h....

4_.

g p

}gL

! j.

.R h-

.+

.M.N

~ O-i

r. -p

...y 3

f i.

r I' l f

i

?. %.W i

1 I

.1 j a : :. _. :_ :... j..

.. ;_-- - l

+n-+. _-.v ;;,

y-7 f-

. O pr,,.,- -.. :-;;.:._. :.' 5

-......f. l._

.t

.w.

t.

... ~. ~

... _._ : t, :. =_~._..w_ __

W,T.

,,,,.5.,.

, w. c.,

.t_-

.w.,,e.w s l

p i

i.

.W P*.

[

..,.q.

j..

.._1,_,._._...

.n.

6

_5._....

...%+

,,, y

i. i._.._.

l l..,.

.....a.. j.

,c v.,p a

p

[..

v. >y#

\\

s.

C m

- * ~ ~

~

roa.,th._.].2

..:: i:

.. l-

~.

.e. - _.

_ =:h.L-.. :.

W L> r.:y'..

f.

t

< =.. -

u.~. =...:.-

_...:t-5. :

. ~ * :; \\

\\

+

p.. p n\\

._.s.._.

m g m,u. e.1Fegg

..... t..

w _.

sc..

. m...... rc, ___ _.o.._.._..s o..

.._.....J. I'.. y,... _.

.. :r__

ve.

. w.l 4

..._.-1---- n %..

...44 -.

. _.. _c% %

-. _t,__- -. - - - -

E.,w<.

Inc es l.._._.._ _

...,..a

.L

.,c.

• , s..,o t.c,

_ _ m..

4

- _l I

. -. I....

. ~.. l

_......... }.

J

@. y. '...,

....,.2

.._. J. -..

,.._....._ p

...i.

./. i

.u p <

i

. y.... p 9._... _

... g __ ;.

4

.sd. l

...[.....--

l

.. e.u.. \\

.l.....

, e

__w

.a_. _

.. y.

q 3_

N @,.*gs ;

q_

. :.J.

p._L

'M e l

._.. q...

_.. g., g.

. ~.. - -..._u_L.;

,gg,., i Ii,

,i i

1 g--

ar.

... i ii) i

. _. _. l r.

i,

4 t

i

.,il

.,wQ.

w, sh -

n W. >.8 t

fU

't 8,HP1pGg.n m.w-A.

u

-~

e-x,,,

l PA N E L g CCRE DISRUPTICN AND REIATED ACCIDENTS Seismic Dep_ age, 1.

Re Hawley study states that the consequences of a core-crushing accident "would be some multiple of the consequences of the fuel-handling accident" analyzed in the study. (p.26).

Me damage to a single bundle l

in a severe core-crushing accident induced by collapse of the building above onto the core in a major earthquake would be substantially greater than the damage induced in a fuel-handling accident to a single bundle.

Furthermore, one must presume most, if not all of the fuel bundles in a core-crushing accident would be similarly affected.

At minimum, then, the consequences would be twenty times as great for a core-crushing i

incident as for Hawley's assumed fuel-handling incident.

• I 2

likrthquake-induced structural damage is often acecapanied by fire.

In this case, the structural damage could expose the core interior to more air than might be available were the core intact, mak%g propagation l

of fire even easier.

3.

Me history of tie-bolt failures for the fuel and the unreported finding from the vibration tests of reactivity oscillations due to the severely undermoderated current configuration, particularly with regards coolant channels h t are half the optimum width, creates potential for reactivity surges due to bowing or other plate and bundle spacing changes induced by the seismic shock. An earthqueke could also readily cause a i

large negative worth sample to be removed from the core region rapidly, without time for intervention of the control blades to aospensats, resulting in a power excursion.

(one particularly worrisomescenario would be a hrge negative worth sample in an irradiation port, the sample being in liquid form in a container which is squeezed or shattered by the compressive forces in the earthquake, rapidly expelling the contents from the core region.

In addition to effecting a positive reactivity insertion, if the liquid were a solvent, the reactivity-induced temperature rise could ignite the material.) A seismic jolt could jerk a control blade out of the core, er cause a large object to impact the drive mechanism outside the core, perhaps initiating an excursion.

4 Se Argonaut reactor is severely undermoderated.

Se University of Washington's Argonaut reactor reports that just a===11 change in the gap between fuel bundles can cause a significant change in reactivity.

(see Exhibit C-III-1). At UCLA, it has been determined in addition that the narrow plate spacing within the bundles creates an " extremely undermoderated" situstion so ht any incident which caused an increase in that plate The Staff attempts to make some comparison to guillotine-type breaks in the fuel. First of all, the fuel is unlikely to shatter in clean, guillotine-type cuts. S e jagged exposed surfaces will have substantially more surface area exposed, and therefore result in greater fission product release, than h theorized clean cuts.

Se fission product release rate from jagged surfaces will not be substantially slower. But more importantly, the assumption of a more three clean cuta per plate in a severe seismically-induced cc w aushing incident is questionable. It is unrealistic to assert that the damage that a severe core-crushing accident, as from a major earth-quake which collapsed the building above the reactor onto the reactor core, would produce the same or even less damage than that which could result from a fuel-handling accident to a single bundle.

\\-

CORE DISRUPTION AND RELATED ACCIDENTS Exhibit List Exhibit Nusber Description C-III-1 letter, 11/2/81 Univ. of Washington to NRC C-III-2 abstract of thesis by J.A.Vitti C-III-3 excerpts frca thesis by R.L. Rudman on Earthquake-Induced Vibrations C-III-4 UCLA Daily Bruin article on water leak, 11/21/79 C-III-5 1963 operating Los excerpts C-III-6 1964 Operating Log excerpts C-III-7 1965 Operating Log excerpts C-III-8 Testimony of D. Hirsch before California Highway Patrol C-III-9 Photos from within NEL l

l v--

y-m

u_

....~

m.s

._m s.m

--._m_

m,

n e

• 4

• . 2 :_.

%*.l=. ="_ __ _ _

• -.--*:*4?---

...:qr_ __ _ _

J,.-_..

y,, _- _

~

_..-~-__ _..-

3,,e, _. _ _ - -.

....__.:'_.".7*-""-'1A 2

e

~~

..'_;-~"...

- - ' _ ~ ~ - - ___.

-: -. - = -. _ MB 1

.'=_.

' ^ C' a G..

=.*.,...-

_.-=?_

M :=.

• ~. _. =:.-.-

~~

..~

.= _. -

~~

b_

_.m e.-.

~ '. - ' ~. _

.+e g

..g. _ _. _ - _ _-.e _w..-._--e

_.. _.e,_.._

.+.

m.-

.+*.g g. g M.

u

...m*.*.._

.e..

• m

_e.

N 6

-N m _.. _..

6 p..

. __. =.. _ =

S

-. g e

_e...

.3.-

ea h_.e.

em.

6

_.M

.M.

l

.*.a 6

.T N -

6

+_*9*

10;- _\\

w _.._.__

s...

_ =

T.=_._:.._._I_J'_.

- =- E.__9_..... _.. _ E_ -2..i:.~-- -

G.

. _ _ _.__ _._._._o..=_

- = = = - = ::. _ _ - _< w. _c__ 2 m =_

=

o y

. _ :. :__.. m=. c.. _=. _ = _

_ _. _, =. -. -.

og ;

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

.=.

._.v_

2

_. _ _ s.

__N Y.

Y w w -

1 " -

-\\

m=

pm.

pay;: gy q ggm..

m.,

s x

s w

x x

ea N

A x

l -

X

. m_3.

.. w_. _.

-._.._2.

e

..- >.: _:*: = w""_~-_

• ~.=2-*

~ _ " _

- :_ :- - _; -.=- -.

^

w...

A 1

__--_-.=%

• I "** _ _

V.

.= _._ _.._ _

....o

_..=.

4

.)

r lo Go 30.

Se 60 W

PISTAN CE m

'I f

h e 25% Radiciodine Release 23 The industry standard for research reactor site evaluation (ANSI /ANS-15 7-1977) irdicates 25% of the radiciodines and 100% of the noble gases should be presumed released.

This is the fraction of release assumed by the University of Florida in its 1981 Safety Aralysis Report for its Argonaut reactor. And, as indicated in the testimony of the C3G panels on accidents I

and release fractions, a 25% radiodiodine release is a realistic estimate for several different accident scenarios at UCLA.

Indeed, there are accidents, particularly those involving fire, which could release a substantially larger fraction than 25%.

(25% is approximately 600 curies of I-131, based on Hawley's calculation that.189% is equal to 4.4 curies).

24 Using Reg. Guide 1.145, as with the 4.4 curie release, for estimating dispersion greater than 100 meters from the reactor room uall produces the results in Table 1 and Figure 1.

Doses are 5200 rem at 100 meters; an EFZ would be necessary out to 23 kilometers as indicated by'the 5 rem dose at that distance; and an urtan boundary should not occur until 75 kilometers from the reactor, as indicated by doses in excess of the ANSI /ANS site criteria out to that distance. As indicated, doses exceeding 10 CFR 20 criteria would extend several scores of kilometers. There are obviously millions of people within both the EFZ and the ANSI /10 CFR 20 zones because of the placement of this particular reactor in the midst of one of the largest cities in the world.

25 Dese estimates for the elcae-in areas of the unrestricted zone near the reactor were made for the 600 curie release in the same fashion as for the 4.4 curie release.

We results are recorded on Table 2 and in Figure 3 As is seen, doses of about 1.2 million rem to the thyroid are found about three feet from the reactor room wall, i.e. in the unrestricted public area outside the reactor facility.

26. Within the reactor room, 600 curies of iodine-131 (and the stardard assortment gf the other iodine isotopes) would produce a concentration of 0.4 Ci/m> of I-131, or approximately 33,000 times the concentration s

assumed by Hawley for the downwind observer to the fuel handling a'ecident,'

said to receive 43 3 rems to the thyroid. The dose at the reactor roem wall would thus te about 1.4 million rem.

27.

The magnitude of these doses near the reactor boundary is confirmed' by the 1980 UCLA Safety Analysis Report (ard 1960 Eazards Analysis). -The 1800 rem estimate at 15 meters was based on the assumption of a 3 curie release to the environment, based on several non-conservative assumptions such as only 10% relasse ard 10 kw instead of 100 kw operation, as now licensed.

Correcting for a 600 curie release, doses 200 times higher at 15 meters, or about 360,000 rem to the thyroid'are fourd.

Thus, the different tl calculational methods result in thyroid doses of 1.4 million rem at the reactor room boundary,1.2 million rem about three feet away, and about 360,000 rem about 50 feet downwind.

4

., ~

u j