ML20086P892

Comments Re Igniters in Upper Plenum of Ice Condenser of Sequoyah & McGuire Nuclar Power Plants, Prepared for ACRS Subcommittee on Class 9 Accidents

ML20086P892
Person / Time
Site:	Mcguire, Sequoyah, McGuire, 05000000
Issue date:	08/17/1981
From:	Sichel M MICHIGAN, UNIV. OF, ANN ARBOR, MI
To:
Shared Package
ML20083L323	List: ML20086P892
References
FOIA-84-7 NUDOCS 8402270374
Download: ML20086P892 (20)

Text

c..

(

U Comments Regarding m.

siters in the Upper Plenum of the Ice Condenser of the Sequoyah and Mc Guire Nuclear Power Plants Prepared for the Subcommittee on Class 9 Accidents of The Advisory Committee on Reactor Safeguards of the Nuclear Regulatory Commission by Professor Martin Sichel Department of Aerospace Engineering The University of Michigan Ann Arbor, Michigan 17 August 1981 8402270374 840130 PDR FOIA HIATT84-7 PDR

(

t.,

I.

Introduction and Summary A system for deliberate ignition of the hydrogen generated in any

\\

loss of coolant accident (LOCA) has been incorporated into the Sequoyah J

and Mc Guire power plants. This interim deliberate ignition system (IDIS) consists of glow plugs placed in the lower and upper containments and at the entrance and exit of the ice condensers. Sandia Corporation has recommended that the igniters at the top of the ice condensers (IC) be 7:emoved [l ],[ 2 ].*

This recommendation is based on the possibility of steam inerting of the hydrogen air mixtures generated in the lower containment volume during some phases of a LOCA. Then, because of condensation, a detonable H -air mixture may issue from the ice condenser discharge.

2 Sandia suggests that ignitersin this location may cause this H -* #

2 mixture to detonate resulting in containment failure.

Because of the many uncertainities involved, it is also possible to postulate a sequence of events which indicate that removal of the igniters at the condenser discharge will increase the danger of cr ntainment failure.

Consequently, the author of the present comments recommends against removal of the igniters at the IC exit. These ideas are developed in detail below.

Numbers refer to references at the end of these comments.

s

]

l i

1

?

1

(.

p II. The Sandia Argument for Igniter Removal The Sandia argument is based on the possibility of steam inerting of the hydrogen air mixture in the lower compartment so that the glow plugs in this region fail to ignite and dispose of the hydrogen. As this mixture flows upward through the ice condenser. much of the steam condenses so that the mixture of hydrogen. air at the condenser exit may t

be detonable. For example, Berman et. al. [ 2 ] show that an inerted mixture consisting of 9% H and 60% steam will after steam condensation 2

have been transformed to a hydrogen air mixture with 22.5%'H, which is 2

clearly at the detonable level. The situation is shown schematically in Fig. 1.

There could thus be a partial 300 torus of detonable gas just above the ice condensers. The main danger then arises from the detonation of this toroidal region. The hydrocode CSO, which is available at Sandia, was used to assess the damage which could be caused by the detonation of this region [1],[ 3 ]. CSQ is a two dimensional code which can only be used to deal with axisymmetric or plane problems. The first calculation made considered the instantaneous detonation of a 360 toroidal region of H air above the ice condenser. While such symmetrical ignition is unlikely.

2 CSO, since it is a two dimensional code, can only handle such cases in the cylindrical geometry. Because of shock interactions and focusing. the resultant impulsive loadings on the containment were extremely high.

The unlikelihood of the symmetrical detonation described above l

was recognized by Sandia, and so they also attempted to estimate the effect 2

/

)

6

.,,_y,-

n-~,

'v),,.c,,,f,;n,,[l,,f

(.F p

4/ome Combv.sNb/e Nu-4;,-

fflMixfam

+s

• 'b

\\

L.

\\

p

.Il E//Y 4f g# ff

); {

A L

Plesso,,4 u

i

/ /

Ce & denser ~s i

j k

/

XX u

I a

I l

"R Inecled' f/ '"

81h d 4 - M g kit' $/X/JM Lowe, Gfas,meir/ Yo/ume l

Figure 1.

Schematic cross sectional diagram of Reactor l

Containment (Based on Fig.

~v'-l of Ref. 6).

3

-[

---

,---.-...,.---,.---.-,-------r--

.?.

(

i r

of asymmetric detonations using CSO. Since this code is two dimensional, these studies were restricted to a plane geometry. Again high impulsive loads were computed.

Another important element of the Sandia argument is the mechanism by which the glow plug igniters can initiate detonation in the hydrogen rich region above the ice condensers. One explanation is based on recent large

'4 scale experiments by J. H. Lee and co-workers [4],[5 ] showing the manner in which turbulence inducing obstacles or baffles can accelerate transition to detonation. Lee has described two types of tests. In one sequence of tests significant overpressures were developed when a flame passed through an open ended tube with L/ D (length to diameter ratio) of 4 filled with a stoidiometric methane air mixture provided obstacles to the flow were present in the tube. Detonations were also induced in a large tube containing a stoichiometric propane air mixture by using a smaller ignition tube containing the same mixture and exhausting into the l' rger a

tube. The critical element was a -set of turbulence producing baffles placed at the end of the smaller tube. Without these detonation did not occur.

.In applying the Lee results Sandia has suggested that the upper plenum of the ice condenser may act like one of the tubes described above with various obstacles and condenser doors near the condenser exit acting as turbulence generating elements. Two rather related sequences leading to detonation have been suggested by Sandia. Quoting from the Butler Memo [ 5 ] :

(.

l l

i

(

U,

~

"..., it has been hypothesized by Dr. Berman that under degraded core accident conditions, the ignition of a hydrogen mixture in the upper plenum results in propagation of the flame b 'ck into the ice condenser where obstructions pro.

mote turbulence and acceleration of the flame front. Tran.

sitions to detonation are then possible in the confined spaces of the ice bed and the highly accelerated flame front, fed by ~

high concentrations'of hydrogen (as a result of inerting in the lower compartment) emerges into the upper plenum as a strong ignition source. It is further postulated that the geometry at the interface between the ice bed and upper plenum, and the chemistry of the mixture in the upper plenum are conducive to transmitting the detonation into the upper plenum. "

A second initiation sequence was described by Dr. Berman from Sandia during The Mc Guire Operating License Hearing. When hydrogen comes through the ice condenser at a low rate, it may burn at the condenser exit like an array of Bunsen burners. The exit of the ice condenser thus acts as a flame holder. Atlow H e neentrations downward propagation 2

or flashback would be unlikely. Now however quoting the Berman testimony

[1]:

" However, if this occurs, if you had these flames going on, and subsequent to that a very large pulse of hydrogen air came through at the same time the burning was going one. I do not 5

__iii

(

i believe that we can guarantee that this kind of quiescent burning would continue. "

The Berman [ 1] suggested that the combustion would become turbulent and that transition to detonation at the ice condenser exit might occur by a mechanism similar to that studied by Lee and his co-workers [4 ],[5].

On the basis of these considerations, Sandia urged removal of igniters at the ice condenser discharge in order to eliminate the dadger of detonation in this region. Instead Sandia suggested that igniters near the dome over the upper compartment of the confinement would provide for slow safe elimination of the hydrogen. Sandia anticipates sufficiently low concentration of H in this region that the hydrogen burns would occur in 2

a slow and lazy fashion. They argue that the H e neentration there, will 2

be too low for downward propagation so that flashback to the ice condensers is unlikely.

III. Discussion of the Sandia Arguments The discussions of the sequence of events which might accompany a LOCA involve many uncertainities and often are speculative in nature.

f i

The uncertain nature of the arguments for removing the upper ice condenser (IC) igniters are recognized by Sandia [1],[2], but nevertheless the sequence of possible events postulated by them led them to their recommendation. The author claims no greater certainity for the discussion of the Sandia arguments which follows. However, as shown below, it is also 6

, ~... -.,..

.,-....,?--~-.

. -,,, J r.

possible to envisage a sequsnco cf evsnto which provida ctreng crgumsnts i

for retaining the igniters at the IC exit.

Gteam inerting of the lower compartment (LC) of the containment is j

the key element of the Sandia arguments. There is disagreement as to whether such inerting is likely to occur in a small break LOCA sequence.

Both the NRC staff report which was included as exhibit K in the Mc Guire hearings [1) and the TVA report on IDIS [6] suggest that the maximum steam generation rate and the bulk of the hydrogen generation will occur

~

at different periods within the failure sequence making steam inerting in the LC highly unlikely. The question of whether or not inerting occurs is beyond the competence of the present author. For the purpose of discussion,-

it therefore will be assumed that inerting does indeed occur in the LC of the containment.

The results of the CSO calculation for the completely symmetric detonation of a torus of detonable' H -air above the IC are alarming. However, 2

there is an extremely smalllikelihood that anything approaching this situ.

ation could ever

• rise. Even if it were desirable to set off a detonation wi'sh such symmetry in order to provide very high pressures, it would be

)

a technologically difficult task. Even if a torus of detonable gas were established above the IC, its composition would be highly variable, and f

initiation of detonation, if it did occur, would problably only occur for a few sections of the toroidal regions and certainly not simultaneously. The CSQ calculations, which are no doubt correct, are not relevant to the.

situation which might arise in a reactor containment.

7 e

1. esame e e en meegam eeense*4 em

,e.

n

.-~,.,.n,_

-n,...._--.-a..

-c

--n--

Recognizing this limitation, Sandia also made some asymmetrical CSQ runs, but now limited to a plane geometry [ 1]. The absence of the side relief present in three dimensions can here also lead to erroneous results. A point explosion in the plane actually represents the explosion of a line of explosive perpendicular to the plane and of infinite length.

1 Thus, it appears to the author that the CSQ calculations carried out by

~

Sandia greatly exaggerate the containment loads generated by the detonation of segments if any H -air mixture present in the upper chamber of the 2

containment. The estimates of pressures due to point detonations presented in the December 1980 Sandia report [ 1] or the impulse estimates developed by the present author [ 7] may be more realistic. The effect of detonations upon the containment requires further study; however, the CSO calculations are not reh11y releva.ut.

1 The experimental studies of Lee and co-workers were used as a basis for showing how detonations might be initiated in the region above the IC exit. The conditions in the Lee experiments and at the IC exit are however, somewhat different, a fact also recognized by Lee. Thus, quoting from the Butler memo [5] :

"Dr. Lee made it clear from the outset that his remarks on transition to detonation were to be taken in the light of characterizing a phenomenun that has only recently been subjected to experimental scrutiny and should not be used to prove or disprove the likelihood of its occurrence in an l

t 8

e

(

t, ice condenser under postulated degraded core accident conditions. He stated that the likelihood of events occurring in an ice condenser containment that lead to the subject phenomenon are beyond the scope of his experience. "

Lee et a1 [ 4) ignited stoichiometric methane air and propane air mixtures which were maintained in a quiescent uniform state inside a pipe prior to ignition. Conditions in the ice condenser plenum will be different especially if the four igniters indicated in Ref. 6 are present. First, the mixture will not be stoichiometric decreasing the likelihood of deflagnation detonation transition. Since the igniters will continually result in hydrogen burning it is unlikely that the mixture in the IC plenum will be uniformly detonable throughout.

The hydrogen air mixture in the IC exit plenum, will not be quiescent in general. The complex sequence of events in a LOCA may result in large surges of flow. The CLASIX code calcuations described in Ref. 6 have dealt with this problem; however, a detailed consideration of the flow history in the IC is beyond the scope of this report. It is, however, of interest to establish the IC exit plenum flow velocities y lume reactions and flow rates. To estimate corresponding to various H2 these velocities it was assumed that 18% and 22% H -air mixtures 2

1 corresponding to hydrogen flow rates of 20 lb/ min and 70 lb/ min (esti-

-1 i

mated range of generation rate, Ref. [1 ] ). leave the condenser exit at a pressure of one atmosphere and a temperature of 300 K (it has been 9

.-n

,_4.

~

~

~

~'

{y assumed here that the LC of the containment is inerted). The detailed calculations are described in Appendix A.

The maximum velocity at the intermediate deck door is 40 cm/ sec, and within the plenum 12. 5 cm/ sec.

The laminar flame speeds for 18% and 22 % H -air mixtures are 60cm/ sec, 2

and 110 cm/ sec respectively [ 9 ]. A similar calculation can be made considering only the 80,000 ft / see of moved from the upper to the lower compartment by the recirculation fans. If this volumetric'ilow rate passes through the ice condenser, the velocity at the intermediate deck door and the plenum will be 41 cm/ see and 13 cm/ sec respectively.

These velocities, which are all below the laminar flame speed are quite low, but of course ignore the significantly increased flow rates which could be induced by pressure differences between the lower and upper containment.

Conditions according to these calculations are almost stagnant in the upper IC plenum and represent a worst case in that a flame could propagate backward from the IC upper plenum into the ice condenser However, if the flame did propagate into the IC passages it could passages.

only go as far as the region where the steam concentrations reaches the inerting level at which point one would expect the flame to extinguish.

The flame could not propagate back to the plenum since insufficient hydrogen would be left for combustion. If flow velocities at the IC exit are two or three times the laminar flame speed, flashback would be highly unlikely.

10

c

(

Exactly where and how the combustion induced by the upper IC plenum igniters will occur is not clear. It could be restricted to the vicinity of the igniters or might occur at the open intermediate deck doors which might act as flame holders, or combustion might occur throughout e n entration exceeded a minimum the IC exit plenum whenever the H2 I

value. What is clear is that as H starts to appear at the IC exit, 2

combustion will occur so that the mixture in the IC exit picnum will tend to have a lower percentage of 1( than the mixture issuing from the IC.

Thus, unless the igniters are overwhelmed, there is some question of whether the mixture in the IC exit plenum will reach a detonable level.

The overwhelming of the igniters by a large pulse of H -air is, of course, the second initiation mechanism suggested by Berman. The author cannot e

refute this possibility absolutely-it could happen. The amount of hydrogen involved in such an event would probably be that contained in tne IC exit should not lead to an plenum, and in the absence of detonation focusing excessive pressure rise.

If the igniters are removed from the upper IC, plenum there is however, another possible sequence of events which the author considers far more dangerous than those posed by Sandia.

l It has been assumed here and by Sandia that the hydrogen. air mixture in the lower compartment (LC) of the containment is inerted by It appears quite likely to the author that at some point in the course steam.

I of a LOCA the steam concentrations will drop below the inerting lev'el so l

that the LC igniters will initiate combustion. Now, however, both the ice condenser, and in the absence of upper IC plenum igniters, the upper 11

_ =

(

IC plenum and the region above it. will contain a combustibh or even detonable H -air mixture. Then, as indicated by Strehicm[IO], a "

2 flame propagating upward through the ice condenser could accelerate and lead to high pressure waves or even to a detonation whi:h then could ignite the H -air mixture above the IC exit. With sufficierchydrogen.

2 air accumulation, the resultant rapid burn would cause an appreciable rise in pressure. It appears apprcpriate here to quote the pertiment section of Strehlow's report [10 ] :

"Unfortunately in the Sequoyah configuration the opper and ' -

i lower compartrnents are not independent but are ccnnected by the ice condenser. In' my opionion, this is avery dangerous configuration because it would generate pressure waves which could possibly lead to local over pressures that corld breach the containment. This is because the ice condenser contains hundreds of tubes (the spaces between the basket:s) which have a very large L/ D and which could cause si "i*' ant E

flame acceleration and possibly even transition t:o chtonation.

l Sandia has suggested tha: safe H rem va Wu e Prodded by.

2 the igniters near the dome of the upper compartment (UClif the confine-ment, since the hydrogen concentration here would be too hw for down.

l ward flame propagations. The detailed H sr u n

de n the 2

absence of the igniters at the IC exit is, however, not k:nove and is a major source of uncertainty in the Sandia argument. Thus. it is also 12

, ~..

,y

,..c y

1

(

(i

• • * ** "8 '#

• conceivable that the light H2*#"

fraction upward along the UC walls forming a region in which the H2 increases from the dome to the IC exit. Then it is not inconceivable that a flame will propagate from the dome igniters downward into the region of increasing H Y lume ra ti n re8ulting in the rapid burn of a significant 2

quantity of hydrogen. In any case, it appears to the author that there le no guarantee that in the absence of IC exit igniters, H2* * ""

occur " lazily" near the dome, IV. Major Sources of Uncertainity The major sources of uncertainity which arose in connections with both the Sandia argurnents and those presented here, as they appeared to the present, author are listed below:

1.

How likely is the occurance of steam inerting in the lower com-partment of the containment?

2.

How does combustion of H ignited by glow plugs in the upper IC 2

plenum occur? That is, wilt the combustion be localized near the igniters, will the igniters act as pilot flames which ignite flames attached to the IC exit doors or the upper deck covers, etc.,

Schott [11 } has, on the basis of possibis high flow rates in this region, recommended that the igniters be placed just above the IC exit plenum rather than inside as at present.

13 4-w

(

(>

i 3.

What is the nature of the combustion initiated by igniters in the dome of the upper compartment, espe cially if the igniters at the IC exit are absent.

J 4.

What will happen if a flame propagates through the ice condenser passages?

5.

What containment 1oadings will be produced by the detonation of limited quantities of H 2-air within various portions of the containment?

V.

Conclusions The igniters at the IC exit should be retained. In the absence of these igniters an appreciable quantity of H e u d accwnlate;.. -

2 regions above the IC. Deinerting of the lower containment could then lead to a sequence of events which would cause rapid combustion or even detonation i of this hydrogen accumulation. With the many unknown factors involved in both the Sandia and the present arguments, it appears desirable to avoid the accumulation of significant quantities of H2 "" *

• all circumstances, and removal of the IC exit igniters is counter to this objective.

14 l

9 w

aw-y

c APPENDIX:

l i

Estimates of Gow velocity in the ice condenser plenum.

According to the sketches in Ref. 6, the inner and outer radius of the ice condenser exit plenum is 41. 5 ft. and 54 ft respectively and the ice condenser extends around 300 of the containment. Hence, 300 2

2 l

(

= A

=

plenum exit area

~

p 360 290.3 m

= 3125 ft

=

If V is the average gas velocity in the plenum and O is the volumetric flow rate, then O=A VP P or II)

V

= --

p A

P The value of V thus depends on the estimate developed for the volumetric p

flow rate O.

At first, the effect of the recirculation fans will be considered.

3 3

37. 73 m / sec.

These have a total capacity of O = O = 80,000 ft / min

=

f Using this value for O in Equation (1),

0.130 m/ sec I

13.0 cm/ see (V )

=

=

P

~

fan Next estimates for limiting values of O during a LOCA will be developed. For this purpose EnH2 i

15 l

l

(

(f

~

no tho volumo frcetien cf H in tho H ~U #

~

hydrogen releans and y 2

2 mixture leaving the ice condenser exit. T and p are takea as the is the total volumetric temperature and pressure at the IC exit and Og flow rate. It will here be assumed that the mass flow rate at the IC exit equals the H release rate, which, of course, is a gross simplifi-2 cation.

Now if p is the density of the H -air mixture at the IC exit, 2

(2) pQ total mass flow

=

rn

=

T while

= '

9M

+(~"}

(

H A

l P

2 em em ar weights of H and air. If Y with %

2 H

H A

2 2

is the mass fraction of H in the H air mixture, then 2

2 m

H H2 and since 9

H (5)

Y"2 y 7Q g +(1-n)Q

=

l 2

1 it follows that MH p

yO (6) rh

=

H L

T Combining Eqs. (1) and (6) the plenum velocity V can be expressed p

in terms of n and rhH2 16 q

(

i "H

kT 2

(7)

V

=

P nA P

p According to estimates discussed in Ref. (1), the maximum H2 N * **

• during a LOCA, ranges between 20 3b/ min and 70 lb/ min. The correspond.

ing values of V, computed using Eq. (7) are tabulated below. for n = 0.18 l

p corresponding to the lower limit of detonability given by Ref. 6, among others, and for n = 0.22, which is a possible value suggested by Sandia.

For the purpose of this estimate it has been assumed that T r 300K and p = 1 atm at the IC exit. The laminar burning velocity So, taken from Ref. 9 is also indicated.

The H -air mixture must pass through the intermediate deck door 2

before entering the upper IC plenum. These doors have a maximum open area of 982 ft (Ref. 6) and the gas velocity V at these doors will be correspondingly greater. V is also indicated in Table A 1.

It should be emphasized that flow surges due to pressure differences between the upper and lower compartments of the c.ontainment are completely ignored here.

17

(-

t. }

Table A-1.

Estimate IC Exit Flenum Velocity (Effect of Uc-LC pressure difference neglected) n = 0.18 n = 0. 22 Su = 60 cm/ se,c Su = 110 cm/ see In lb/ min 20 70 20 70 H2 Kg/ sec 0.151 0.530 0.151

'O.530 f

rng 2

O m / see 10.33 36.2 8.45 29.6 g

O m / sec 37.7 I

f V

cm/ sec 3.56 12.46 2.91 10.19 P

V cm/ see 11.33 39.7 9.26 32.43 V

13.0 F

pf cm/ see volumetric flow rate of H "" #

• O

=

g 2

average velocity of II -air mixture at IC exit plenum V

=

2 p

average veloch of H ~" #

• V

=

2 pd doors volumetric flow rate of recirculation fans O

=

f average velocity at IC exit plenum due to recirculation V

=

Pi fans only.

I r

= H generation rate during a LOCA H

2 2

l l

l l

i l

y, *:'

(

.{-t References 1.

Transcripts of Operating License Hearing, Duke Power Company,-

Mc Guire Units 1 and 2, 10 11. March 1981. Docket Nos. 50-369-OL, 50.370 OL, pp. 3998 4372.

2.

M. Berman, M. P. Sherman, J. C. Cummings, M. R. Baer, and S.K. Griffiths, " Analysis of Hydrogen Mitigation for Degraded Core Accidents in the Sequoyah Nuclear Power Plant", Prepared by Sandia Corp. for Div. of Reactor Safety Research, NRC, Dec. 1,1980.

t Detonations in Zion and f

3.

Byers, R.K., "CSQ Calculations of H2 i

Sequoyah," paper presented at the Workshop on the Impact of Hydrogen on Water Reactor Safety, Albuquerque, N. M., Jan.

25-28, 1981.

4.

Lee, J.H.S., "Some Recent Large Scale Experiments on Gas Explosions, " and " Flame Acceleration Mechanisms in Closed Ve s s els. "

Both papers presented at the Workshop on the Impact j

of Hydrogen on Water Reactor Safety, Albuquerque, N. M.

Jan.

25 28, 1981.

5.

Memorandum from W. Butler, Chief, Containment Systems Branch, DSI to L. Rubenstein, Assistant Director for Core and Containment Systems, DSL

Subject:

. Meeting with Dr. J. Lee re: H Combustion, April 7,1981.

2 6.

Tennesseee Valley Authority, Sequoyah Nuclear Plant, Core Degradation Program, Volume 2, " Report on the Safety Evaluation of the Interim Distributed Ignition System, " December 15, 1980.

7.

Sichel, M., " Simplified Calculation of Detonation Induced Impulse,"

paper presented at The Workshop on the Impact of Hydrogen on Water Reactor Safety, Albuquerque, N. M., Jan. 25 28, 1981.

Lewis, B., and von Elbe, G., ; Combustion, Flames and Explosions 8.

of Gases, Academic Press, New York,1961.

Sherman, M.P., et al., "The Behavior of Hydrogen During Accidents 9.

in Light Water Reactors," Sandia National Laboratory and Energy Incorporated, NOREG/ CR-1561 53, August 1980.

10.

Strehlow, R. A., " Evaluation of the Glow Plug Igniter Concept for use in the Sequoyah Nuclear Plant. " Prepared for Mr. J.

Milhoan, P. E., Office of Policy Evaluation, NRC, Jan. 9,1981.

G. L. Schott, letter to Dr. J. Carson Mark, Chairman, Sequoyah 11.

Reactor Subcommittee, Advisory Committee on Reactor Safeguards, NRC, May 7,1981.

w

~

w