ML20082B489

Corrected 831107 Rebuttal Testimony Re Shutdown Mechanism

ML20082B489
Person / Time
Site:	05000142
Issue date:	11/17/1983
From:	CALIFORNIA, UNIV. OF, LOS ANGELES, CA
To:
Shared Package
ML20082B483	List: ML20082B482 ML20082B489
References
NUDOCS 8311210209
Download: ML20082B489 (37)

Text

.o eP l

CORRECTIONS )

1. Rebuttal to CBG's Panel I Rebuttal

p. 6, 4th line from bottom, the sentence should read: "The reactor's concrete shield plugs must be unstacked to replace the start-up source."

p.10, A.4, 9th line, should read: . . . excites . . ."

p. 14, A.7, 3rd line, should read: . . . exits . . ."

2. Rebuttal to CBG's Wigner Energy Testimony

p. 4,15th line, should read: ".. 100 kw. . . . "

p.10, A.3, 3rd line, should read: "

. . 3 x 10 12 , ,_,.

3. Calculation of Wigner Release for the UCLA Reactor by Dr. H. Pearlman with Supporting Calculation by N.' Ostrander

p. 1, A.1, 12th line, delete: "p"

p. 2, equation at bottom of page: insert division symbol ("/")

between first "dE" and second integral symbol.

p. 3, 2nd line from bottom, should read: "

. . . conservatism . . ."

p.17, 8th line from bottom, should read: I' . . . occurs . . .

Insert the attached p.19 (" Figure 1") which was inadvertently omitted.

4. Calibration of the UCLA Nuclear Reactor, Emil K. Kalil, Ph.D.

p. 2, throughout the page, the decay constant should be: " Lambda" not "Landa."

p. 2, the last sentence on the page should read: " Substituting the 22 to 25 second half life in the above equations, gives a neutron yield

of approximately .0047 to .0060."

5. Rebuttal on Credibility of Graphite Fire at the UCLA Reactor l p. 3, item "D" under "Windscale" should read: " Area /vol . = 0.16"

p. 9, 1st line, should read: " summa rizes"

p. 10, 4th line: replace the colon with a comma, p.10, A.9, lith line, should read: . . . photographs . . . "

p. 12, A.11, 2nd line, should read: .. . . concrete . . ."

F3)1210209 831117 PDR ADOCK 05000142 -

T PDR. ;

3 6. Rebuttal on Credibility of CBG's Fission Produce Release Model p.11 -12th line, should read: "

. . . (1978). . . ."

7. Rebuttal'to CBG's' Testimony on Dispersion

p. 2,. 7th line from bottom, should read:

. . . conditions . . ."

8. University's Exhibit List

p. 2, IDENT. N0. 21., 3rd line, should read: "

. . 100-16528 . . ."

4 i

l- -

[ ,.

19 GR APHITE , FUEL +H2 0; ,

GR AP HITE ,

FUELt H 2 0,, G R A P HIT E i

i l

i I

I E

N s P

, W I

e I

I J\ J\ jy ,

y V V V EXTERNAL NORTH CENTElf ISL AND SOUTH EXTERNAL REFLECTOR CORE COHE REFLECTOR I

z=0 21.6 c m 6.35~ cm PLAN VIEW OF ARGONAUT REACTOR Figure 1

.?

'g . REBUTTAL CONCERNING THE SHUTDOWN MECHANISM IN THE UCLA ARGONAUT 4

Q.1. Please describe the shutdown mechanism in the UCLA Argonaut reactor.

A.1. The UCLA reactor shuts down in response to rapid and 1,arge reactivity insertions by the same fundamental shutdown mechanism 1! hat operates in the class of plate-type water-moderated reactors represented by SPERT and Borax: temperature rise in the fuel causing density changes in the fuel and water and followed by the voiding of water as the result of sub-cooled n;:cleate boiling. In the models that have been developed to describe the mechanism the temperature rise in the fuel plates is a function of the reactor period, which is jointly determined by the reduced prompt neutron lifetime and the reactivity change expressed in dollars, and the shutdown (or reactivity) coefficient.

The total energy release is a function of the temperature rise or energy density of the fuel plates and the active mass of the core. In the region of a 14 msec period, which in the UCLA reactor would result from the instantaneous insertion of the proposed licensed limit of

$3.00 excess reactivity, SPERT I-D and Borax-I data indicate that the ,

peak temperature calculated adiabatically would be about 400 C, which would be considerably above the peak temperatures that would be expected on the basis of the temperature rises actually observed during SPERT and Borax tests (see " Figure 3" reproduced here from my earlier testimony). It should be noted that the Battelle generic study

• l analyzed the case of an instantaneous insertion of $4.00 excess reactivity which put the reactor on a 7.2 msec period. This represents considerable conservatism with respect to the maximum credible reactivity accident at the UCLA reactor.

t

,-. e

l l l l l 5 l l l l l BOR AX DESTRUC T TE ST -

- ~

BORAX-l SPERT_l_

_ ADIABATIC RISE o .

OBSERVED RISE a o ou 1000 -

• 1 m 800 -

~

.. g . OBSERVED TEMPERATURE RISE E

m 600 =

JJ UPPER BOUND a

p 500 -

< o a

r. o

@ 400 -

o Y, -

a o O

• E ,

3 o o Ct9) b * " N g 300 - ,og

~

g' " -

3 250 -

g -

!< 200 -

,o -

E o a g 150 -

c -

D&o 0 0 100 " ' ' ' ' ' ' ' ' ' ' '

20 30 40 50 60 7080 100 200 300 50 400 RECIPROCAL PERIOD ~ Sec~l 25 20 10 5 2.5 PERIOD ~ Millisec ADI ABATIC ANb OBSERVED ~ TEMPERATURE RISE ,

VERSUS REACTOR INVERSE PERIOD (seconds-1) ..

FROM N. OSTRA NDER's TESTIMONY page 17 APPLICANf'S FYHIRIi

Q.2. What about the fact that SPERT and Borax were " swimming-pool" type reactors and the UCLA reactor is not.

A.2. The fact that the UCLA Argonaut reactor is not a swimm,ing-pool type rea.ctor does not change the basic conclusion, although it is necessary to examine the shutdown process in more detail to explain how much voiding is required and where it is accommodated. The SPERT-I-A-17/28 swimming-pool core was a closed system in which recriticality was restored after the first power burst for transients of initial periods i

down to 7 milliseconds. The UCLA Argonaut reactor will behave timilarly for an initial period of 14 milliseconds. The system pressure following the first burst will rise to about 47 psig and expel sufficient water l in about 100 milliseconds to terminate the event. That expulsion time is short compared to observed oscillatory and chugging periods and none are to be expected for the UCLA Argonaut reactor.

Q.3. How does the analysis of this event proceed?

A.3. There are four steps. The first step is to find the common denominator i

for a spectrum of reactors characterized by various values of the

" shutdown parameter" which is the ratio of the void coefficient to the reduced prompt neutron life time. It is found that the peak power density is inversely proportional to the square root of the shut down .

parameter. Based on empirical data the constant of proportionality is 73 + 16% for reactor periods of 14 milliseconds in the A, B, and D cores of the SPERT-I series. The numerical value yields peak power density in units of kw per cm3 ,

. . .--, ,-. , , _ . _ - - . , _ . .- .- , ---__,-4 ,- -, ,~,..m.-

1 The second step is to insert the value of the shutdown parameter in the generalized relation to calculate the expected peak power in .

the UCLA reactor for a transient of the same period. The total energy ,

density generated in the first burst is estimated as if the peak power persisted for two periods, a rectangular power pulse of 28 milliseconds dur'a tion. The temperature rise is then calculated.

The third step is to estimate the voiding rate and the disposition of the displaced water. It is found that the air. space above the f

normal core water level is sufficient to accommodate a volume equivalent to about -6.40 dollars of void. The reactor is well on the way to shutdown at this moment.

i In the fourth step, the subsequent expulsion of water is treated by an energy imperative, the system pressure will rise to whatever '

pressure is necessary to expel the water at the same rate that steam is being generated'.

Q.4 Please describe the shutdown coefficient discusted by Forbes et. al.

(I00-16528, 1959).

Q.4. That shutdown coefficient is a " constant" which relates the instantaneous reactivity, measured in . inverse seconds, to some function of the cumulative energy generated. The mathematical form chosen by l

Forbes el a_1,is:

fh=a - gb[E(t- T)]1/2 ,

l 9

e

-r= + + -

The initial reactivity is a, invers~e seconds, T represents a delay time, b is the shut down coefficient, and E(t -T) is the cumulative.

energy generated T seconds prior to time t. The time t is an ,

independent variable which has no characteristic value. $ = power.

Q.5. Wha't do you mean by the phrase reactivity measured in inverse seconds?

A.5. In my previous testimony of June 1983, p. 22, I described a simplified version of the in hour equation as

$=w(h)+1 .

l The symbol w repre:ents the inverse period which by algebraic re-arrangement is

.,$-1 t/6

• g During the time interval between an abrupt positive reactivity insertion and the onset of shutdown, w=a g , the initial inverse period.

l Q.6.

What factors influence the numerical value of the shutdown coefficient b?

A.6. The shutdown coefficient is not dimensionless, and certainly the .

numerical value will depend upon the units that are chosen. In particular, if power density is used instead of total power, and if the void coefficient Cy is expressed in dollars per percent void then for 1

1 1

j w - . , . , . --,,-e.- - . . - , , -

^

a 14 millisecond period the four aluminum plate reactors of Forbes and the SPERT-1-D reactor described by Miller, Sola, and McCardell (ID0-16883,1962), can be represented by ,

3 fCY Bi -1/2 ,

Power density (kw/cm ) = k -

g }

with k = (73 + 16%)

In other words, for this class of reactor with the proportionality constant k = 73, one can predict the peak power density from the known values of the shutdown parameter, Cy .8/1 Q.7. Please explain why you use power density instead of total power.

A.7. The power density represented by that equation is the maximum power density of the excursion, maximum in time, but average in space.

If Forbes had provided total energy, I would have reduced that to energy density. They did not do so, and it is necessary to estimate the maximum energy' density.

Q.8. Why do you want to calculate energy density?

A.B. The need for power and energy densities was remarked by Dietrich in discussing the Borax I and Borax II tests. For a fixed energy density, a large reactor will generate more total energy than a small reactor, but the temperatures will be similar. The temperature is important in two ways. First, it is the driving potential for heat transport, steam generation, e

v -_. . .y

and shutdown. This is implicit in the use of temperature coefficients.

Second, there is the practical matter of the maximum temperature that a fuel plate reaches in an excursion. Energy density is another name ,

for temperature and it is the temperature that is important, not the total energy. '

Q.9. What examples can you cite to illustrate this point?

A.9. Figure 11, p. 21 of Forbes shows the maximum temperature versus reciprocal period for four aluminum plate type reactors. At a reciprocal period of 70 per second the sequence of temperature rises versus core type is the same as that for power density in my Table I which follows.

Q.10. What does the uncertainty of k represent? '

A.10. Values of power density for a reactor period of 14 milliseconds were calculated for the five reactors from the graphical total power data of Forbes, et. al, and from the data of Miller, Sola and McCardell for tne SPERT-I-D core. The total power was divided by the active fuel plate volume. These power densities were then multiplied by the square root of Cy 8/1. The value 73 is the arithmetic mean of the extreme values, and the 16% uncertainty encompasses all of the k values.

l

Table I. 14 Millisecond Transients in Various Reactors

~

CORE B-12/64 B-16/_40, A-I7/28 B-24/32 D-12/25 Argonaut Number of olates 768 640 4 76 76 8 270 264 Number of channels 704 600 448 736 245 240

. Water channel cross section, cm2 2071 1185 812 741 e Active height, cm 60 60 co 60 60 II of water volume 1243 711 487 445 C, cents /mi -0.0093 -0.029 -0.046 -0.073 c, A/% void -0.116 -0.206 -0.224 '

-0.325 -0.36 -0.277 t/s,alllisec 11 10 7 7 8.16 29.2 c, + s/t. 4/1-sec 10.55 20.6 32 46.4 44.1 9.5 Max. power. Mw 950 550 400 550

' 194 Active volume, cm 3 43500 36240 26960 43500 15300 Max. Power Density (P) kw/cm3 for 3- 70 m'I 21.8 15.2 10.8 12.6 12.7 k=P-(c,f)1/2 M.8 69.0 61.1 ' 85.8 84.3.

e

.g-Q.11. What is the source of the uncertainty?

A.ll. The reduced prompt neutron life is not recorded to more than -

about one part in 10 and could introduce an error of approximately 5%. The graphical data for the SPERT A and B cores is ' difficult to read to within + 5%. There appears to be a residual dependence upon C y B/t with the smaller values of k associated with the smaller values of Cy B/t. The various values are shown in Table I.

Q.12. How would you use this data to predict the temperature of a fuel plate in an Argonaut reactor under a 14 millisecond excursion.

A.12 I would take the mean k value, k = 73, Cy = -0.277 dollars /% void and 1/8 = 0.0292 sec. This yields a power density of 23.7 kw/cm3 ,

Forbes, et. al, do not provide total energy data, but Dietrich (Borax-1) and Miller, Sola, and McCardell (SPERT-I-D) both indicate that the relation E(total) = 2T 4(max) is applicable in the region of T = 0.014 seconds. With this I calculate 664 joules per em3, For a specific heat of 2.83 joules /cm3,o C, the average temperature rise would be 2350 C. For a peak-to-average power ratio of 1.63, the maximum temperature rise would be 382 'C to a final ter.perature of 402 C.

Q.13. Why did you use k = 73?

A.13.

. It is the mean value for the data that I reviewed. It is conservative.

, In terms of CyB/t, the Argonaut reactor resembles the B-12/64 core with Cy 8/1 = 71. The calculated adiabatic temperature rise will -

always exceed the true temperature rise. Forbes in Figure 11 (which fol. lows) indicates a total temperature of about 3000 C for the B-12/64 core at a reciprocal period of 70 per second.

i Q.14. You take E(tot) = 2T 4(max) as some kind of fundamental relation.

Please explain that.

A.14. It is really an experimental observation derived from the Borax-I and SPERT-I-D reports. Those two reactors differed markedly in size and in other respects. The rule is independent of whether the Borax-I reactor starts from a subcooled state or near saturation temperature.

1 It means that the total energy, E(tot) can be represented by the maximum power,4(max), acting over a time interval 2T.

This indicates to me that there is an energy density imperative, the reactor will shut itself down even if it must generate large maximum power and total energy. The energy, power, temperature, and 3

pressure histories may, and did differ appreciably-between Borax-I and SPERT-I-D but the 2r rule was valid for both in the region'of' T = 14

• milliseconds, i

l i

I I

k w - r --r * ,- - , e , -+,

FtEL PLATE SLA4 FACE TEhPERATL#tE .

vs RECIPROCAL PERIOD '

1. i-20'C Hood = 24 inches -

id I ~~ ~~ ~ ~ %TM TEdRATUK OF STAINLb5 STEEL g _______________.________

L _ Z-MELTING TEMPERATURE OF ALUMINUM , ,g a-(12/s4) ni P-te i id - - - - p/19)g

,,... .(g7fg) rge-(m/40)

L - BM /32) sATunAmN 7- -

g N TURE uaxlMUM TEMPERATURE /

10 -

g -B -(2/40)

L #

W, -SATURATION TEMPERATURE ,' , .. . /28) 5:10 -

t

<- k... P-(S/19) -~~

~

g B-(12/64) _ , ,

<r - -

B-(24/32) a-10 _

TEMPERATURE RISE AT MAXIMUM POWER unim f i i f i i 1 , , f . ,

16' 10' I 10 10' 10' RECIPROCAL PERIOD (seE') smie Fig. 11 -

Reproduced from Forbes et al.,1959 [2]. p.21 '

p e

g $

i r 3

l I i Y 3-

-C i A

'f l .

.  ?

l l

. _ . _ _ . . _ , . . . _ . _ _ . _ . . , _ . . . . . . _ . - - ..q__.. , . _ . . . _ . . . . _ . . '. . _ . . _

s Q.15. How did you obtain the void coefficients in units of dollars per percent void? , ,

l A.15. The values,used are for the uniform void coefficient. . A value of

$0.36/% void is given by Miller, Sola, andEcCardell for the "D" core. The values for the A and B cores were taken from Forbes in units of dellars per cm , multiplied by the estimated water volume for the ccre in question and divided by 100 to obtain dollars per i

percent.

Q.16. What uncertainkies are there in choosing void coefficients?

A.16. The principal uncertainty is in kno' wing what an investigator means when reporting a void coefficient. There are central void coefficients, peripheral void coefficients, and uniform void coefficients. Although these regions are easily identified qualitatively, the precise domain is seldom identified in terms of a description of the volume over which that coefficie it is applicable.

Q.17. How was the void coefficient determined for the UCLA reactor?-

A.17 The value is derived from measurements described and documented in the startup, report for the UCLA reactor. It is the only documented value1

,for the UCLA reactor. The startup was supervised and reported by

,t General Nuclear Engineering. The reactivity effect of local. simulant u, -

d

~

s

('

_-- .- ,c i

)

.A e,

. .. . . - - _ _ . - _ _ _ _ _ . _ - _ _ _ _ _ . _ - _ _. _. _ =_

i .

I j voids of 110 ml (1 ml = lem 3) was sampled in a number of locations within the core. An average or uniform coefficient for the core was reported to be 23 inhours per void (110 ml), 5.3 x 10'4%k/m1, and ,

0.25k/% void.

'l 1

Q.18. How do you convert those numbers to dollars per percent void?

A.18. One inhour is approximately 0.00360 dollars, a conversion that is independent of the delayed neutron fraction. I define 100 percent void as all of the water between the fuel plates, 2.845" x 24" x 0.137" x 240 spaces. For the UCLA reactor, this is approximately 36800 ml and 1% of that volume is 368 ml per % void. The result is 1

Cy = -(23/110) x 0.00360 x 368 i Cy = -0.277 dollars /% void '

Q.19. You have cited oth'er numerical values for the void coefficient in previous testimony. Please explain the apperent discrepancy.

A.19. A lower value was predicted prior to the startup measurement. A still smaller value was reportedly measured by students around 1963. However, there is no existing documentation of such a measurement; nor is it l

known what units were used to report tha coefficient. A member of- .,

UCLA's nuclear enginee 'ng faculty who participated as a student in the measurement was unable to confirm the measurement.

Not much cel be done with that kind of information.

i -

I went back to the documented measurements and made a conversion

'. to dollars which does not require speculation concerning the delayed neutron fraction. I related it to a well defined volume of water. -

I don't know how to find a better value than the value. C, = -0.277 dollars /% void. For comparison, if the smallest reported (undocumented) value is used, the predicted temperature rise is increased about 20" C, which is a relatively small effect indicating that the calculation is not particularly sensitive to any uncertainty in the void coefficient measurement.

Q.20. Why do you choose to use a void coefficient in units of dollars per %

void rather than in dollars per milliliter?

A.20 If one is analyzing a single reactor, it makes no difference which 1

units are chosen. However, if one is studying a spectrum of cases which range from small, undermoderated reactors to large overmoderated reactors, it is evident that a milliliter of void will have a much larger effect upon a small reactor than a larger reactor. The spectrum of void coefficients is appreciably narrower when expressed .

in the % unit than when expressed in the milliliter unit.

Q.21. Why did you not use the Borax-I data?

A.21. The Borax-I reactor was superficially very similar to the SPERT-I A core in number m' plates, plate dimensions, water channel width and I

i-a

+c- -, , s , e . , , . ,,y .,. ,

= .- . -. . .- _ .

. 15-I fuel plate loading. Its performance was qui +.e different. The fuel plates were curved, their configuration in the fuel channels is not clear in the single drawing I have seen (Dietrich, Figure 7) and the static head above the core was 3 to 4.5 feet versus 2 feet above the fuel plates in the SPERT A and B cores. I do not know whether there are other differences which might explain the energy density pulses and the pressure pulses which were significantly higher in Borax-I than in the apparently similar SPERT-I A core.

Q.22. Why did you not consider the Borax-II reactor?

A.22. I did consider that reactor. It was a unique reactor in that it contained perimeter fuel loadings that contained plates of much higher uranium -235 content than the central plates and the plates of Borax-I.

The effect was to markedly flatten the flux profile relative to Borax-I.

Borax-II was a remarkably responsive reactor. With an absolutely flat peak-to-average power or flux ratio, all parts of the reactor reach the point of incipient boiling at the same instant and the time scale for shutdown is compressed relative to that for a reactor with a high peak-to-average power. In the latter, steam generation spreads more smoothly throughout the core as successive portions of ~

the core are driven to the point of incipient boiling. This induces a delay, but the pressure pulses will be smaller. -

e S

i 54 v -

1 It was this consideration that led to my introduction of the

', peak-to-average power ratic in my previous testimony.

i I

Q.23. Why did you not use the SPERT-II heavy water test data? l A.23.

With the reported value of 1/8 = 0.1 seconds for the SPERT-II reactor, it would take a reactivity insertion of over 7 dollars to excite a 14 millisecond period. Further, with a void coefficient of 0.07 dollars per % void (about 1/4 the UCLA void coefficient value).

Cy B/t = 0.7 and even with k as small as 60, the resultant energy density of 2000 joules per em3 would be well into the melting region. I The UCLA Argonaut core has a shutdown parameter value of above 9.5 which is similar to the SPERT B-12/64 core value of 10.6 and of the same magnitude as the other SPERT I cores. However, the UCLA Argonaut shutdown parameter value is about 14 times that of the SPERT-II reactor.' Obviously, the parameters affecting shutdown in the two reactors are sufficiently dissimilar to make useful comparisons unlikely.

Q.24 Why do you think your correlation with k=73 should apply to the SPERT-II heavy water?

A.24.

I did not mean to imply that the correlation should be applicable to an arbitrarily large class of reactors. I only indicated that the correlation was in qualitati've ' agreement with the observations that had been made

• 4 u----_____- - _ _ _ -

i l

about SPERT II and would remain so even if a k value substantially less than 60 might correlate the non-existent data.

Q.25. Please explain the significance of the Robles data regarding the graphite temperature coeffi~cient.

A.25. The value for the graphite temperature coefficient measured by Robles and reported in his thesis (Robles Primitivo, "A Theoretical and Experimental Dynamic Analysis of the UCLA 100 kW(t) Nuclear Reactor", UCLA Masters Degree Thesis,1972) is in excellent agreement with the other reported measurements of the coefficient. Perhaps i

confusion has resulted from the different choices of temperature that may be used to define the reactivity coefficient. Robles presented a graph which plotted % k versus temperature. The graph had two 1

essentially straight lines, one for the reactivity change versus a temperature observed near the core center, the other related the reactivity chang'e to an average and lower graphite temperature.

Both lines are correct, but one must be consistent in using the temperature for which the coefficient was defined. In particular, it is improper to " average" coefficients which were differently defined.

..r for the Robles coefficients based upon the average graphite tenperature, the other measured values for Argonaut reactors are:

4 s

?

6 0

% k/ F U. of W. 0.0014 UCLA (Robles) 0.0014 '

UCLA (Ostrander) 0.0013 .

These values consistently relate to the temperature at the center of the graphite center island. The agreement is excellent.

Q.26. What is the significance of this postive coefficient in influencing a transient?

A.26. None. Most (85%) of the fission neutrons deposit their energy in the water. The graphite is heated primarily by gamma rays. The amount of energy per fission that is promptly released and potentially available to heat the graphite is about 3% of the total event energy.-

If half of this energy is deposited in the graphite center island

~

(roughly 3 cubic feet of graphite or 145 kilograms of specific heat 0.745 joules per gram-deg C at 25 C) then a 30 mega-joule transient (SPERT-I-D destruct) would yield an energy deposition in the graphite of 6

30 x 10 x (0.03/2)/145 x 103 = 3.1 joules /gm.

The resultant adiabatic temperature rise of 3.1/0.745 = 4.2 deg C to '

contribute about 1.5c of incremental reactivity to the event.

e 7

Q.27. Why do you use only one-half of 3%?

A.27. I assume the other half went into the outside reflector.

• l l

Q.28. In'.your earlier testimony you deduced a reactivity change of 1.3 cents for the same event. Why are you changing your testimony?

A.28. This is not a change. The earlier testimony on this point was based upon experimental data regarding the temperature rise in the graphite within 5 minutes of 100 kw operation from a cold start (30 megawatts).

At that time I did not know that only 15% of the prompt neutron energy was deposited in the graphite. I am now showing that you can come to nearly the same answer,1.3 versus 1.5 cents, in a different and

~

independent way.

A.29. Why are you so sure that 85 percent of the neutrons are moderated by the water?

A.29. The question of the relative importance of the graphite as a moderator versus its reflector function has been around for a long time. An answer to that question would clarify the related questions of graphite j heati_ng and void coefficient.

l The question became fundamental to the discussion of energy deposition in graphite for the Wigner stored energy calculation. It is.also relevant to any discussion of graphite radiation damage effects in the e

3 - - - = , e

UCLA reactor. I made two calculations, one somewhat intuitive and

a second of a confirmatory nature.

, They agreed very well that approximately 85% of the neutrons are moderated by water. As this .,

was done in the context of the Wigner energy question , I provided my description of the methodology and results to Dr. H. Pearlman.

The details are described in my attachment to the testimony on Wigner energy.

Q.30. Why do you use 1/B = 29.2 milliseconds to characterize the UCLA Argonaut transients?

A.30. A number of pile oscillator experiments were performed in the UCLA reactor in the 1960's to obtain measurements for the quantity R/t.

The value of B/1 chosen from measurements in the range 20 to 50, was 34.2 which is consistent with the lowest conceivable 8 of 0.0065 and' a reported prompt neutron lifetime of 1.9 x 10-4 seconds. A recent KENO code calculat' ion of the UCLA prompt neutron lifetime done for the NRC produced a value of 1.88 x 10-4 seconds, which is very close agreement. .The reciprocal of 34.2 is about 0.0292 seconds. The basis of that number is experimental observations which were related ,

to plausible values of L and B.

Q.31. Is the Argonaut reactor an open or closed system? .

A.31. For low flow rates, the Argonaut reactor will not sustain a pressure significantly above atmospheric pressure. Also, it can remove any  !

I i

amount of water if the pressure is high enough. .

i l

1 i

Q.32. Were the Borax and SPERT reactors open systems?

A.32. The Borax-I and SpERT A and B systems were closed in a cerzain sense. .

The vessel walls were sufficiently high that for 14 and 16 millisecond pulses, they could not eject enough water to prevent recriticality.

After the first pulse they settled to a more or less stable boiling mode (Dietrich, Figure 11 and Schroeder et. al, Figure 8).

Q.33. How would you portray the UCLA Argonaut reactor in this regard?

A.33. In regard to inerdal forces, the UCLA Argonaut reactor with a 10 inch water overburden, is less " closed" than either the SPERT-I or Borax-I reactors with overburdens of 2 to 4.5 feet. On the other hand, the larger void requirement demands a greater voiding rate if water eviction is to be achieved in a timely fashion. '

Q.34 What is your estima'te of the shutdown void and void rate for the UCLA Argonaut reactor?

A.34. It has been commonly recognized that to effect shutdown it is necessary to cancel or compensate only the supctprompt critical reactivity. In the UCLA Argonaut reactor experiencing an instantaneous reactivity change of $3.00 this would mean -2.00 dollars reactivity -

compensation would be required. However, I chose to consider the introduction of -4.00 dollars of reactivity in order not only to turn l the power rise around but also to completely shut down the reactor.

S 9

_. ~ . - ~

EXPERIMENTS ON WATER-MODERATED REACTORS sees b --

=.--- - - - - .-.

e ne: - I

-neim e e.e.e . j l I e *

• I ~

- c ..es l l N"':

f:;- %V -fi v _

gA{M p _.

~

N ..e g s

' h L

~

j i.

e e s e a . . , , , ,, ,,

flaat AFTER Antif aAaY 2E00, ese Iself tAL WATER TEenPERATutt 30 F ne m ii. no , m, i ; u to .a . v.a i i.nl .ec.r.i .t i4 m.o .d p.,w BORAX - I Reproduced from Dietrich,1955 [1]. p.95 l

, 10'

/

ii l

TRANSIENT No 1469 - i0 T=66 wee , ,

}

b ,_ h_ _ _ _ _ - _ 0 -

l TRANS40if MIESSURE 1v

" -S E

~ (44 temer end bos) -

i0 AATE RMRAW -

250

' SS-172 + 0 3 Crus -

h  ! -

200P f, _

L REACTOR POWER

- ISO

____m iOO f LTK OF SATIAIATION TDIPERATURE ~97* C - 50 g* ' g1 PEAK power f f f f I f W

0 i f 0 01 02 03 04 0.S 06 07 08 09 10 TIME AFTER ARSITRARY ZERO (sec)

Ftc. 8. Itepre-entative behavior following I.fr~e renetivity addition (e = 62 sec-8).

TadE 30 g ,. TRANSENT No.1530 -

i g_ T*7.0 mesc -; to L 4 '\_T=A aT =S uRE %r -

300

. (44 laser ens ten)

SATURATION TDrDtATIAEa-97*C- ~

Pun TneDtATURE 200 l0 g

l 4S-02 + 0 scRm g

15 0

, p- --- -

iOO IEACTOR PUNDI -

50 g' d , , , , , ,

0 0 0.2 04 06 Os 1.0 12 TaE AFTER AlWITRARY 2010 (esc)

Fm. 9. Itepre-entative lehavior following 1.4c'c reactivity addition (a = 143 nec-8).

Reproduced from Schroeder.1959 [4]. p.102,103 l i

t e

The introduction of -4.00 dollars of reactivity with a void coefficient of -0.277 dollar per percent requires a void of 14.4%.

The voidable water per fuel box is about 6140 ml of which 886 al -

must be displaced by void to create 4.00 dollars of negative reactivity.

The time scale for voiding is approximately 2T where T is the i initial reactor period. This is based upon the observation that E(tot) = 2 T - P(max), and is also the approximate duration of the pressure pulse in a 9 millisecond transient of SPERT-I-D, run number 24 (Miller, Sola, McCardell 1D0-16883, p.92). The average flow is about 886/2 x 0.014 = 31700 ml/sec. for each fuel box.

Q.35. Explain whether the rupture disk will provide a release avenue.

A.35. The rupture disk is not a reasonable avenue for a time scale of 28 milliseconds. The main impediment is the inertia of the water

~

resident in the 30 foot length of 3 inch line that runs from the bottom of the fuel boxes to the rupture disk.

Q.36. What about the overflow lines from the fuel boxes to the dump tank?

A.36. Those lines are one inch in diameter and provide a cross section of 2

5.07 cm . The velocity in those lines would have to be about 62

• meters per second. The principal pressure drops would be entrance and exit losses which i~f taken as one velocity head (1/2pV2)-each, will sum to about 570 psi. -

l -

l l . .

l . .

l l

1 Q.37. What about the deflector plate apertere?

9 A.37. I misspoke earlier when I said that they faced a graphite wall. They -

actually face a wall of lead bricks, but the situation is the same.

The shield plugs weigh about 100 pounds each and are manually removed and reinserted. To accomplish this, a clearance is required which I and others estimate to Be about 1/8 inch. The aperture is 2 inches by 6.5 inches, and the 17 inch perimeter by 1/8 inch clearance provides a flow area of about 13.7 cm2 . If the water had to escape by that '

avenue, the velocity would be about 23 meters per second. (Diagrams of the cooling water system and the fuel box and plug assembly follow.)

Q.38. What pressure do you associate with that veloci'ty?

A.38.

There are several pressure drops associated with that avenue, but the entrance and exit losses at one velocity head each '4111 sum to about 80 psi.

Q.39. What if you consider these avenues jointly?

A.39. 2 The cross sectional flow area becomes 18.8 cm . Both exits will have.the same pressure and hence the same velocity of about 17 meters j

- per second. The required pressure is about 42 psi. ~

l l .

~

\

W 4

4

\

e/p

• w

~ ,3v

/ e n

s' ' Y

?

C$t h

\2.6 ,

' pp \

s s

\\

pWP

1. 1. Ns

\

h es CflN

/ N

\\

r l$ .

\

=

\

PR Ping gy '

uyp r.

)

• 91 W t

e

1"

,, -l: :l

[

I I

I I

i . l l

l 1

l 3-3- 14 g -

16 g I l

I ,,

I j ... g '

2-

\

fe i' FUEL BOX TOP i f! elevation 12" i- "h -- A OVERFLOW LINE l elevation = 10" ,

l l

I

,,g [ FUEL PLATE TOP elevation = 0"

~

/

5" 6" Fuel Box and Shield Plug Geometry t

0

,, - , - , , - - - , r -

Q.40. What other avenues are there for expulsion?

A.40. The shield plugs weigh about 100 pounds each and have a cross .

tectional area of about 29 square inches. The shield plugs rest on top of the fuel boxes and are not secured in any way. A gap exists between the top of the plugs and the shield blocks on top.

The plugs will lift at a pressure of 3.45 pounds per square inch (3.45 psi).

Q.41. Will the air above dthe normal water level be compressed?

A.41.

The peak air displacement rate of 31700 ml/sec could exhaust through the deflector exit aperture gap of 13.7 cm2 at a speed of 23 meters per second. For air of density 0.0012 grams per m1, the adiabatic expansion requires a pressure of about 0.05 psig to achieve that velocity. The pressure requirement, if doubled is approximately two velocity heads (0.09 psi). The air flow is essentially unrestricted.

Q.42. You are assuming that the air pressure will break the membrane at the bottom of the shield plug. Why do you assume that?

A.42. The membrane is an aluminum rectangle 5 inches by 6 inches and a thickness of 0.020 inches. .

+

4

O l

UsinD " Marks" Standard Handbook for Mechanical Engineers, seventh edition (Baumuster, ed. McGraw-Hill), section 5, pages 68, 69, and 70, case 16 leads to a uniform loading of about 1.5 psi to '

rupture aluminum of 24000 psi ultimate tensile strength. That tensile strength is the upper limit cited in the same reference, same section, page 5.

Q.43. How does this back pressure influence the excursion?

A.43. The effect is negligible. The air pressure will simply rise to about 1.5 psi and rupture the membrane. The water will continue upward to fill the deflector region of approximately 5 inches by 6 inches in horizontal cross section with an average height of one inch. The available air volume above the original water level is about 1430 mi per fuel box.

This volume'will be occupied at about 45 milliseconds from the time of peak power.

The steam void that has displaced the water is about 23% of the water volume and has a worth of about -6.40 dollars.

Q.44. What happens to the water. '

i 4

S

A.44. It is helpful to look at Figures 8 and 9 Schroeder et. al. At approximately 50 ms after peak power the reactor is approaching a minimum in power after which criticality is restored and the '

power again rises. This is true for both the 16 ms and 7 ms transients . They did not expel enough water to prevent recriticality, and the same can be expected of the UCLA Argonaut.

To answer the question more directly, the power will rise and generate steam at a pressure that will balance the steam generation rate with the water expulsion rate. That pressure will be about 47 l

psig and the event will be over in 0.1 to 0.2 seconds after the first minimum.

Q.45. How do you define the event termination?

A.45. The event will be over when enough water is expelled to prevent recriticality. I define that quantity of water as all of the water overburden, 4.8 liters, and 15 percent of the core water, an additional liter. The total quantity to be expelled for termination is 5.8 liters.

Q.46. How do you know the event will be over in 0.1 to 0.2 seconds after the first minimum? "

9 e

4 e

l

l A.46. The prolonged first minimum is because the water cannot get back into the core, there is too much steam from the hot fuel plates.

3 When they cool, water will reenter and restore criticality. The

( .

Power will rise and generate more steam. Assuming that the Argonaut j

will rise to an average surface heat flux of 12 watts per cm2 , the I steam generation rate will be about 145 grams /sec per fuel box. At 47 I i

psig the steam volume rate will be about 64 liters per second per l l

fuel box. Also, the expulsion rate of 64 liters per second can be I i

j i

satisfied at a pressure of 47 psig. The cross section area for ~

l 2

expulsion consists of the overflow line (5.1 cm ), the deflector 2

gaps (13.7 cm ) and the shield plug gap (16.5 cm2 ). The water l

l velocity through those areas will be about 18 meters per second, l and the loss of two velocity heads is about 47 psig.

! l l l

The calculated expulsion time is 5.8 liters divided by 64 -

l liters per second or 0.09 seconds.

l o

Q.47. What will the fuel plate temperatures be during this phase?

A.47.

According to Figure 8 of Schroeder, et. 13 ., the temperatures are still l

declining. However, a heat flux of 12 watts /cm2 requires a temperature

(

excess of 10 deg C above the saturation temperature. The saturation i

temperature at 47 psig is about 146 deg C, and the temperature cannot . l i

l fall below about 156 deg C until the event is over and the system drops l to atmospheric pressure.

. ~-

S

,- m , ,---m-

w. -

Q.48. Will there be oscillations or chugging?

A.48. No, the event is over too soon. A second small pulse can be .

expected, but it can't get very far because the systen reactivity

, l is disappearing rapidly. '

Q.49. Where did you obtain the heat flux of 12 watts /cm27 A.49. Figure 8 of Schroeder indicates a peak power at the second pulse of about 4 megawatts. The heat transfer area of that reactor was 2

about 4 x 10 cm . The fuel plate temperatures were not far above the saturation temperature, a condition indicative of normal boiling 2 2 with a heat flux of about 10 watts /cm . The choice of 12 watts /cm for the Argonaut reactor is in recognition of the smaller surface to volume ratio of the Argonaut reactor relative to the SPERT I A '

reactor.

4 1

Q.50. How long does it take to raise the shield block plug and open that 0.3 inch gap?

A.50. The momentum of the moving water can open the gap. Whether it stays open depends upon the subsequent pressure. With a pressure of 47 psi l acting on the 30 square inch projected area of_ the 100 pound plug -

will accelerate it at 14.0 g's. Deducting the gravitational accelera-l tion the net upward acceleration will be about 12700 cm/sec2. The gap g will then open in about 11 milliseconds. The rise time is small compared l to the water expulsion time. .

t

i Q.51. Please summarize your conclusions regarding the self-limiting '

shutdown of the UCLA Argonaut Reactor?

A.51. There is more than an adequate margin of safety to preclude fuel melting in the first burst initiated by a 14 millisecond transient. Following the first burst, the rate of water return is limited by evolving steam, and can represent only a ramp insertion of reactivity. The subsequent rise of power will be similar to that of a boiling water reactor with water leaks which remove moderator and in which the energy generation is dissipated by producing steam. Although I have ignored the rupture disk line and the analysis, it could become a useful release route if the boiling stage is at lower pressure and hence more prolonged than I have estimated. I find no phenomena here that can induce melting and fission product ,

release, and hence there is no potential for endangering the public health and safety in such an event.

l 1

l l

t 6 m e

y, - - "

-gy. ^

t l REFERENCES l .

( 1. - Dietrich, J. R., " Experimental Detenninations of the Self-Regulation and Safety of Operating Water-Moderated Reactors," Peaceful Uses of Atomic Energy, Geneva Convention, V.13, p.88-101,1555., '

2. Forbes, S. G. , Bentzen, F. L., French, P. , Grund, J. E. , Haire J. C. ,

Nyer, W. E., and Walker R. F., " Analysis of Self-Shutdown Behavior 1 in the SPERT-I Reactor," AEC Research and Development Report. l 100-16528, 1959.

3. Miller, R. W.,

Sola, Alain, and McCardell, R. K. , " Report on the Spert I Destructive Test Program on an Aluminum, Plate-Type,

' Water-Moderated Reactor," AEC Research and Development Report, 100-16883, 1964.

4.

Schroeder, F. , Forbes, S. G. , Nyer, W. E. , Bentzen, F. L. , and Bright, G. O., "Expirimental Study of Transient Behavior in a l

Subcooled, V.2, Water-Moderated p.96-ll5, 1957, Reactor," Nuclear Science and Engineering, i

1 e

go I

s k

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

"