ML19289C374

Forwards Rept to NRC on the PSC Document Entitled Core Fluctuation Investigation Status & Safety Evaluation Rept. Concludes It Is Unlikely That Fluctuation Problem Is Causing Serious Structural Damage as Result of Core Motion

ML19289C374
Person / Time
Site:	Fort Saint Vrain
Issue date:	12/13/1978
From:	Bennett J LOS ALAMOS NATIONAL LABORATORY
To:	Tokar M Office of Nuclear Reactor Regulation
Shared Package
ML19289C364	List: ML19289C374 ML19289C376
References
NUDOCS 7901120104
Download: ML19289C374 (17)

Text

.

REPORT TO NRC ON THE PSC DOCUMENT ENTITLED

" CORE FLUCTUATION INVESTIGATION STATUS AND SAFETY EVALUATION REPORT" by J. G. Bennett, 0-13 C. A. Anderson, Q-13 Los Alamos Scientific Laboratory L8hSL

5==-

7901126i0'

REPORT TO NRC ON THE PSC DOCUMENT ENTITLED

" CORE FLUCTUATION INVESTIGATION STATUS AND SAFETY EVALUATION REPORT" INTRODUCTION The Reactor and Advanced Heat Transfer Technology Group (Q-13) of the Los Alamos Scientific Laboratory (LASL) acting as a consultant to the Nwlear Regulatory Comission (NRC) has reviewed a document submitted to NRC by the Public Service-Company of Colorado (PSC) entitled " Core Fluctuation Investigation Status and Safety Evaluation Report". The document was submitted to NRC on ,

August 11,1978 in support of an application for further testing of the Fort St. Vrain (FSV) reactor, a High Temperature Gas Cooled Reactor (HTGR), in a temperature fluctuating mode at 70% power, and to operate beyond 70% power.

Group Q-13 of LASL has had an on-going effort in HTGR structural evaluation and safety for a nurber of years and in this regard we are familiar with the geometry, -

materials, and components of the FSV reactor. Thus, it is believed that we are uniquely qualified to c..m.ent on the structural aspects treated in the report and the references cited. Table I lists the references given in the report and indicates those that were reviewed.

r CORE MOTION EVIDENCE There is no doubt that some type of core motion is occurring. Figures 1 and 2 (Figs. 4-5 and 4-6 of the report), reproduced here from the report, show the time histories of data recorded during an oscillation occurrence. The coinci-dence of enhanced Prestressed Ccqcrete Reactor Vessel (PCRV) motion indicated by the displacement probe DP-2, and the increased nuclear activity indicated by the nuclear channel detectors are firm evidence of such motion.

THERMALLY INDUCED MOTION MECHANIS?t The core motion of concern, indicated in Figs.1 and 2 by the sudden in-crease in amplitude on the DP-2 record is probably not flow-induced flutter .-

in the sense of conventional structural vibrations, and, as such, is probably not relatable to any lateral pressure gradients. (This statement does not mean that we do not think that flow induced motion may be occurring. Indeed, the small amplitude background motion of the PCRV may well be caused by a flow-induced low amplitude motion of the internals.) Because of the long period (5-20 minutes) of the fluctuation, and because the data shcwn in Figs.1 and 2 indicate an I

, . M C'Jr.E t-3 i . i .. .

j* . .

I ..1 -

• f 4 erit

i, io7a -

i pas

, 12:t 1:36 . . -

l -

12:10:14

-: -<- 10 SECONDS - + ,

-w.~ - ^ ^

- ~- - ~ -

__ , _- 4 i, . . NUCLEAR dlANNEL S

. . , .--_ . ~ ~ ~ - ~ ~ ~ ~ #'

- NUCLEAR CllANNEL 7

. .?

. _. - _- r n__ _._

-' ~ - '

I NUCLEAR 'CilANNEL 3 .

s..

< ~

~ - - ' -

= ~~

7 .

'l

- - NUCLEAR CllANNEL 4

.~

s- .

__- w. -- .l . . *

. . NUCLEAR CitANNql S

_ ;_ _=_ -_=1.-

p- -

NUCLEAR Cl(ANNEL 6

[ , ,.

i;

'% fV' ,

%'%A%\fvb qO C0llE PRESSURE DROP j .... .

MM

- ~ -

1 N@WMh , e'""2"'"*"*'"" ""

3.

Fig. 1.  %

Time history of data record during temperature fluctuation.

as (Fig. 4-5 of the report)

Al' AIL 2,r, .l S *.'.C . --

13:01:05 ~~, 5 ~ Sil:t :5T. ' .

.10 SECG10S - .

3 .M -

~

-- - -- -- - .flHE .

.- - - - - - = - NUCLEAR CHANNEL 8

=-

, NUCLEAR CilANNEL 7

~

6 .

, _[ . _

1- - -

NUCLEAR CitAIDfEL 3 r _

/

__ l- NUCLEAR CllANNEL,4

~. .

NUCLEAR CilANNEL.5 .

a

--- m - HUCLEAR CilANNEL 6 NM MW g .

M

,. .. CORE PRESSURE DROP h f& 2 i  ! .

fig. 2. Continuation of the recorded data from Fig 1. (Fig. 4-6 of the report)

abrupt change in PCRV motion, we think it likely . at . e motion results from a sudden shif ting of the core to a new equilibrium :3siti:n. We do not dismiss the possibility of a lateral pressure gradient initiati::g the shift, but the strain energy buildup and release resulting in the core motion is likely to be thermal in nature since themal changes have such :haracteristically long periods.

The shifting of the core will produce a new set of flow paths and may create conditions for a later shifting and return of the core to its original configura-tion. This cyclic, themally induced shifting motion could go on indefinitely if not arrested.

The analytical model given on page 21 Section 5.3.1.5.2 of the subject report, " Lateral Pressure Gradient in the Core Pedel," does not appear to have any thermal strain energy buildup or release mechanisms available. In addition, the model assumes a priori tha't the core motion is flow-induced motion because of the lateral pressure gradients. Such a model leads to . continuous PCRV motion that might be characteristic of the backgrcund ration indicated in Figs. i and 2 by the displacement probe. .The displacement amplitudes of the PCRV obtaine:

from the model are highly dependent upon the assumptions made in the modeling of the flow and the gap pressure differences. Thus, it is possible to approximate the PCRV displacement and gage maximum amplitude with the computer code output. f The modeling as given suffers from other deficiencies; namely, the result does not indicate the same form (i.e., the characteristic of a lowest mode transient response and decay of the PCR'l to some type of impulsive loading or core shift) that is shown by the data in Figs. I and 2. In addition, we have some reservations about the modeling itself, particularly with regard to some of the stiffnesses chosen., the masses represented, and the method for input of the forcing function from tne flow data.

Therefore, the 3 in./s velocity from this model given on page 22 and later -

used in section 5.3.1.5, " Fuel Element Impact Loadings," to calculate loads on individual elements seems questionable.

ENERGY IN THE PCRV MOTION AND IMPACT VELOCITIES We agree that the amount of energy involved in the motion of the PCRV is small'as shown by our calculations, which are sura:arized in Appendix A. However ,

we must point out that if the displacement transd;cer is not aligned with the direction rf motion of the PCRV, then the actual energy involved can be higher than the energy calculated by I

calcu1&ted Eactual = cosa e where e is the angle between th7 actual displacement vector and the measured one. In this regard,1f we assume the maximum PCRV displacement is being measured, our lowest mode calculations based on Figs. I and 2 indicate a change in kinetic energy of 2.0 in.:1b as compared with the 1.5 in.-lb value given on page 21 of the report. If the transducer is off in placement as much as, say, 80 *, this value will be 66.3 in.-lb.

Unfortunately, the calculation of the impact velocities based on momentum conservation depends upon the mass assumed to be impacting the PCRV. We feel that the PSC/GA assumption that an entire fuel region is involved is not cormletely justifiable, though not necessarily unconservative. It is possible -

that shift of the entire core mass could be involved, in which case very low impact velocities will be obtained. It is also possible that only a portion of a

, nel region. is involved in the impact in which case very high velocities can be tained.

It need also be pointed out tha't a conservatien of comentum analysis, that also assumes conservation of energy during the impact, does not necessarily give the upper value of impacting velocity. Again, the calculated initial impacting velocity depends on the initial conditions and masses assumed to be invo,1ved in the impact. For example, if conservation of energy is not assumed and the single impact model with a coefficient of restitution of 0.3 is used, a single fuel region could be moving at about 13 in./s and give the data quoted on page 21.

Appendix B has the details of this calculation. Therefore, we do not agree that "the velocity of 5 in./s" necessarily " represents an upper -limit value" (pg. 21).

DOWEL PIN / SLOT SAFETY Regarding dowel pin failure, the configuration and kinematics involved in any shifting motion of the core. do not present a reasonable opportunity for a single dowel / slot inact to. occur. Such an impact would require relative motion between individual fuel elements. Furthennore, for a single dowel to take the ireact loading would certainly require some nechanism for a vertical separation betwee.n blocks to occur or else would require an extraordinary set of clearances between dowels and mati,ng slots to exist.

I

In reviewing Ref. 5-3 of the rano-t (Table I), which is cited as the reference for dowel and socket strength in fatigue, we find that it is for 4 and 5 dowel /s lot configurations. This reference also is the source of dowel / socket stiffness measurements. The dowel / socket stiffness value is critical to demonstrating adequate reserve strength. If such measurements exist for FSV fuel blocks they should be documented more fully.

Using a spring stiffness value of 30,000 lb/in. (Ref. 5-3) and a static failure load of 950 lb (pg. 25 of the report), the energy absorbing capability of 2 (upper and lower) dowel / slots is approximately 30 in.-lb. This value means that an extraordinary set of clearances coupled with a relative impact velocity of 9 in./s would be required to cause pin fa: lure. For these calcula-tions, see Appendix C. This calculation differs from our earlier calculation where we assumed a dowel stiffness of 610 lb/in. and thus we could not categori-cally rule out dewel pin failures. We now feel that it is most unlikely that .

any dewel pin hilures have occurred.

A General Atomic (GA) Quarterly Progress Report (GA-7314) for the period ending June 30, 1966 gives the results of 4 dowel socket impact tests for the FSV fuel block. From these tes ts, it appears that the energy absorbing capability of a single dowel / socket ranged from 66 to 90 in-lb. This data reflects the fact f that many materials have higher impact failure strengths than static failure strengths, providing no repeated impacts ara involved (see Appendix C).

FUEL ELEMENT SAFETY It is most likely that all motion from shifting of the core results in flat faced or edge and corner impact loadings. In reviewing the references cited (Ref. 5.1, 5.2) we find that the most significant sentence is in the conclusion of Ref. 5.2,."The only method of being sure of the corner and edge impact effect on the seismic design is to perfonn failure impact tests for angular impacts."

Again, such data, if available for FSV fuel elements,should be presented to demonstrate the adequate safety of the elements.

For the 13 in./s impact velocity quoted above and from data given in -

Ref. 5.2, the flat face impact energy absorbing capability of an element can be s, hewn to be adequate by a factor of 6 (Appendix C). We feel that it is unlikely that any damage to individual elements is occurring or will occur. -

DATA NEEDED We see no' likely fix" without having more data. We think that in-core mechanical instrumentation i. eeded to obtain the data. We also note that to I

support any of the structural response related arguments put forth in this document, more PCRV moticn transducers are needed. It would seem wise to have at least four around the periphery and two vertical ones. The vertical motion transducers will indicate any rocking motion of the PCRV.

REC 0te:ENDATION It is probably unwise to operate in the fluctuating made for any extended period of time without having a more definite indication of the magnitude of the intamal motio'n of the care (i.e., until more data acquisition capability is in place). An analogy that might be given is that the problem is akin to being able to measure the magnitude of beating of an interior wall with measurements taken on the exterior of the building. ibout the only thing you are likely to detect is some type of lowest mode response of the building. We recommend that any time fluctuations are noted that steps imediately be taken to arrest them.

SUMMARY

In sumary, we %el that it is unlikely that the fluctuation problem is causing any serious structural damage as a result of core motion, but without in-core mechanical instrumentation, the case for precluding potential damage is weak. We think that the core motion may be due to a thermal strain energy release' and re:ultant core shift, and that any analytical model should allow for this mechanism. We recomend that any time fluctuations are encountered that imediate steps be taken to arrest them. We recomand that in-core mechanical instrumentation be planned and placed as soon as is practical.

None of our coments should be construed to mean U.at we.do not believe that the FSV reactor should be operated beyond 70% power since the repor- indicates that is is possible to operate beyond 70% power without fluctuations occurring.

The report also adequately demonstrates that the fluctuations can be arrested by

~

reduction of power.

e

4 The following references were reviewed in conjunction with this report:

TABLE I ,

REFERENCES GIVEN IN THE REPORT 5-1. Sevier, L., "HTGR Graphite Fuel Element Seismic Strength," General Atomic Report GA-A13920, April 30,1976.

5-2. Shatoff, H.D. " Approximation of Corner and Edge Loads from HTGR Core Seismic Analysis Codes," General Atomic Report GA-A1427, April 1977.

5-3. Chiang. D. D. , " Fatigue Tests of Dewel-Socket Systems," General .

Atomic Report GA-A13861, June 15,1976.

• 5-4. Price, R. J., " Cyclic Fatigue of Near-Isotmpic Graphite:

Influence of Stress Cycle and Neutron Irradiation," General Atomic Report GA-A14588, Novecter 1977.

t

• This reference was not reviewed.

1 9

I

APPENDIX A CHANGE IN KINETIC ENERGY STORED IN THE PCRV

. Assuming the PCRV is oscillating in a low mode configuration, the total kinetic energy change is given by AXE = h PCRV M ~

where ,

4 I MPCRV = mass of the PCRV = 8.02 x 10 Vf = final velocity during change in KE Vj = initial velocity during change in KE Since the motion is sinusoidal, t

V = X ta where .

W. = the natural frequency of the motion X = the amplitude of the motion From Page 21 of the report.

X j = 200 x 10-6 in.

-6 in.

Xf = 600.x 10 From Figs.1 and 2, (Figs. 4-5 and 4-6 of the report) ,

f = 2 Hz I

u = 2nf = 4n rad /s .

Therefore ,

"2 AKE = f [8.02 x 104I~'2) n x 10-3 x 16:2 (36-4) 2 AKE = 2.02 in-lb.

APPENDIX B . '

CALCULATI?N OF VELOCITIES FROM A SINGLE IMPACT MODEL Consider the central impact of two bodies as shcwn belcw:

. V fB '

B - -

.V -

Yk - -

A M 0 - N B A A w ,- 3 x ,

. . . . BEFORE IMPACT. . . . . AFTER. IMPACT . ..

Conservation of momentum requires, MYAA+NY B B " N AYd + N BYb .

(I) where Vj = velocity of body i (positive to the right)

Mg = mass of the body i and primes denote the velocities after impact. .

A enmmon method of accounting for energy loss during impact is to relate the relative velocities before and after impact by a coefficient of restitution.

I

For the situation in the sketch above with velocities as shown, the coefficient of restitution e is given by YB~Y

• • VA-VB or, e(VA~Y)=Vj-Vj.

B - (2)

We note that in this case, if "B" is the PCRV, we know V and V' fr m the dis-B B placement probe data.

Equation 1 can be rearranged as M

VA-Y (Y5 + YB ) (3)

Equation 2. gives, e eVA+Yd*Yb~'YB (4) adding Eqs. (3) and (4) and solving for V A, we have V (Y" ~ YB ) + Y'B + eV B -

(5)

A " fi + e B Consider the following examples implied by the data in the report.

EXAMPLE 1 - Bodies having velocities with like signs.

Let, M = the mass of a single fuel region = 17,500 lb/g MB = the mass of the PCRV

= 29.3 x 106 gjg where g is the acceleration of gravity.

I

Frequency data fmm -6 I

VB = (200 x 10 in) (4r rad /s) = Bn x 154 in/s Figs. I and 2. Amplitude I data from page 21 of the iY'=(600x10-6 g in) (4x rad /s) = 24n X 10-4 in/s report. i Let us assume that both the fuel region's velocity and PCRV's velocity are to the right. Furthermore, assume the impact.to be perfectly elastic and that total energy is conservad. This case is given by e = 1.

Substituting into Eq. (5), we obtain e '

4 29.3'x 10 V

A" 1 l7. x 10 .

(24-8)wx10~4

+ (24+8)w x 10-4 > in/s s negligible ;

VA = 4.2 in/s .

This VA is the velocity of the fuel region under the assumptiens made.

EXAMPLE 2 - Bodies approaching one another.

Let all data remain the same as that given in Example 1 except that we will assume that the initial velocity of the PCRV in the sketch is to the left.f.e., '

V B = -8n x 10-4 in/s.

Also,it is reasonable to assume some energy is lost during impact. For example GA data shows that impact graphite fuel blocks have a coefficient of restitution of about 0.3. Let e = 0.3.

Then, substituting into Eq. (5),

- 3

) 29.3 x 10 6 -

VA =( I1.3

- <1 j 17.5 x 10 3 -

(24 + 8)w x 10~4

+ negl_igible term >

• s VA = 12.9 in/s . .

This V is the velocity of the' fuel region before impact under the assumptions A

given.

E

These examples are giYen to illustrate the strong dependence of the velocity obtained on the assumptions made. Note that even higher velocities will be obtained if a smaller MA is used.

APPENDIX C FUEL BLOCK DA!MGE POTENTIAL A method for estimating the potential damage to a body undergoing an impact loading is to assume that the kinetic energy stored in the body is absorbid elastically during the impact. The maximum energy that can be absorbed by the body can be estimated from the static failure load and the spring stiffness of the member. This method can be expected to be accurate for brittle (graphite) materials, but appears to be conservative since normally measured dynamic failure ,

loads are typically two or more times as large as static failure loads from a single test. On the other hand, repeated impact can be expected to.. decrease failure loads by a factor of two or more. As an example, the large H327 graphite HTGR fuel block in flat faced single impact tests exhibits an initial failure load of about 80,000 lb. However, after about 1000 impacts at 40,000 lb, failure f will nomally occur .;,(Ref. 4.) Table _I). _._

Tiius, using the static failure load may at first appear excessively con-servative, but since the pntvious impact history is unknown, the degree of con-servatism vanishes. The following calculations demonstrate the methods we have used in evaluating.the report:

. EXAMPLE 1 - Energy absorbing capability of single pin-slot menters.

The energy absorbed by an elastic spring.under' load is given by 2

F EA"N*

' here W F is the applied load and K is the spring stiffness of the body.

Using the static failure load for the Fort St. Vrain pin / slots of 950 lb (given on page 25 in the report) and the equivalent spring stiffness of K=30,000h as given in Ref. '5-3, we have for a single pin slot, E

2 E = (950)2 lb 15 in-lb.

2(30,000)h

, Because of kinematical considerations and a required clearance mismatch between all six pins and slots for a given fuel block it is not likely that a single pin impact can occur.

~

At least two pins, one upper and one lower (and prebably more), wiil share the impact. Thus we estimate the maximum enargy absorbtion capability of 2 pin / slots to be about 30-in-lb. Assume that the impact is shared by 1 upper and 1 lower pin / slot. .The relative velocity between blocks in a column required to cause damage is then given by f MV2 = 30 in-lb.

For the 300 lb block , .

V = 8.8 in/s.

r EXAMPLE 2 - Flat faced impact for an impact velocity of 13 in/s. The hnetic energy in the block Eg , is Eg= =f in s' 13)2 n/M Eg = 65 in-lh .

For flat face impact, the spring stiffness K is estimated in Ref. 5-1, Table I, as K = .1.93 x 106g and the failure load from Table I, Ref. 5-2 (after 1000 impacts) as F = 40,000 lb .

The energy that can be absorbed is,':then, lb 2 E

A =2(1.93 B 0000)2 6 415 in-lb x 10 )

Safety Factor = h = 6.

e

/

9 0

I