Journal

of NUCLEAR SCIENCE and TECHNOLOGY, 16[5], pp

343-355

(May 1979).

343

Instabilities

in Parallel Boiling

Channel Upflow

of Forced-Convection (III)
Two Channels

System,

System

with

Different

Flow

Conditions

between

Masanori

ARITOMI, Shigebumi

AOKI and Akira INOUE

Research Laboratory Tokyo Institute

for Nuclear Reactors, of Technology*

Received October 20, 1978 Revised February 16, 1979 The instabilities observed in a parallel-channel system carrying boiling fluid in forced upward flow have been studied theoretically and experimentally, using water as test fluid. The system, where the thermo-hydraulic flow conditions in two channels are different from each other, is studied theoretically and experimentally, in order to make clear the effects of the different conditions on the flow instabilities. The different conditions between parallel channels are made artificially by changing the entrance throttlings, the heater lengthes and the heat fluxes. Consequently, the instability in the system where the own characteristic frequencies are approximatelly equal in two channels almost agree with the one obtained under the uniform condition equivalent to the average operating condition in two channels, such as the different entrance throttlings and the different heater lengthes. On the other hand, the system, where the characteristic frequencies differ from two channels, becomes more stable with increasing the difference of flow condition, such as the different heat fluxes. KEYWORDS: boiling flow instability, parallel-channel system, stable region, unstable boundary, different flow condition, two phase flow, flow oscillation, instability, stability, eigenfreguency, heat flux I. INTRODUCTION

The thermo-hydraulic instabilities in parallel boiling channels have gained momentum in the last decade for their importance to the safety of fossil-fuel boilers, boiling water reactors, as well as steam generators of fast breeder reactors and advanced thermal reactors(1)~(5). Especially, the development of LMFBR becomes the pressing problem for the following energy source taking the place of oil energy, and the development of steam generators for LMFBR becomes one of the most important problem. The plants comprising parallel boiling channels are multifarious, and their operating conditions differ from each other, such as operating fluid, heating method, exit flow condition and system pressure etc. It becomes hence difficult to apply a common approach to the understanding of the thermo-hydraulic instabilities in every systems. The present work attempts to classify and to simplify the phenomena and mechanisms of their instabilities to faciliate their understanding. In companion paper(6)(7), the authors have been shown that the instabilities observed in boiling fluid in forced upward flow through twin parallel channels can be classified
* Ookayama, Meguro-ku, Tokyo 152. 39

344

J. Nucl.

Sci. Technol.,

into four types. The instability being occurred through the interaction between channels have been taken up in particular. A mathematical model well explained

the this

phenomenon was proposed, and the instabilities appeared in a system comprising more than three channels showed a behavior quite simular to these in a two-channel system(6). The experimental results in two parallel channels with the uniform flow condition confirmed the analytical ones. In the plant with parallel boiling channels, the flow conditions between channels are not exactly same, and it is not necessarily a good policy to design the plant with simular channel condition. Studies on the instability under the different flow conditions between channels are only found out in Veziroglus' work(8). In their paper, the above mentioned instabilities of four types were discussed using a common approach. Consequently, the mechanisms of these instabilities were not sufficiently understood, and the effects of the difference of flow condition on the instabilities generated through the interaction between parallel channels have not been made clear. The purpose of this paper is to understand these effects. The instabilities under the hydraulic or the thermal different condition are studied experimentally and theoretically, as (1) the entrance throttlings differ between two channels and (2) the heater lengthes or the heat fluxes differ. II. EXPERITENTAL A PPARATUS

An outline of the experimental apparatus is mentioned here, since the details were reported in a companion paper(7). Figure 1 shows the schamatic diagram of the forced-convection water loop operated at atmospheric pressure. It has been deviced to suppress flow oscillations in the loop itself, so as to bring out only the instability caused by the interaction between parallel
channels. Figure mension the inlet of 2 is two the geometry channels plenums. and the between Each di-

parallel the outlet

and has

channel electrodes tary measuring lation. tion to has suppress

a heater, and a rofor oscilsec-

flowmeter flow The been the that flow test

deviced static is,

instability, the tion type, dynamic caused action nels served. Fig. 1 Arrangement of flow loop 40 by head of so that

oscilla-

Ledinegg's only instability the interchanobFig. 2 Test section the

between has been

Vol. 16, No. 5 (May 1979)

345

An orifice (not shown in Fig. 2) was installed at the channel (2) to observe the effects of the different entrance The heated tube is made of thin stainless steel. The are identified, since the resistance per unit length is uniform in a series. In the second condition, the heater in channel with the one in channel (1), so that the heat inputs differ effects of the different heater lengthes on the instability Further, the variable resistence (8) and a magnetic

upstream of the flowineter in throttlings. heat fluxes in two channels and two heater are arranged (2) is shortened as compared between two channels. The

are observed. switch (9) (shown in Fig. 1) are

provided in parallel to the heater of channel (2), in order to get the different heat fluxes between two channels. The effects of the different heat fluxes on the instability are observed. III I The analytical ANALYSIS model used in this paper is essentially the same as the one in a com-

panion paper(6). A revised point and its process are shown below. 1. Initial Condition The continuity and energy equations are solved in each channel and the momentum equations are integrated with the proposal procedure. c114,',/dt is derived from

Two procedures

are examined

to get the initial conditions.

(1) The characteristic curves of the static pressure drop Pin-Pout to the inlet velocity are different between two channels. Solving the equiliblium inlet velocity u(1', and uin), where Pin Pout agrees in two channels through static analysis, the initial conditions are solved statically by substituting UinI), and Uin into the continuity and energy equations in each channel. We selected -5, -2, -1, 1, 2 and 5 m/s2 as du2,)(0)1 . cldd (2) The initial conditions are obtained in the same way as the proposal one using the average inlet velocity air, as the initial inlet velocities in two channels. The resulting velocities re;,) and u)' differ from each other under the different flow contions. Making use of this process, du;,),0,1dt is assumed by substituting the initial conditions into Eq. (1),

From the analytical results of procedures (1) and (2), it was found that the instabilities are not dependent on the initial conditions and the magnitude of initial disturbance like the results under the same parallel channel condition(6)(7). Later initial conditions were adopted in this paper. 2. Phase Analysis From Eq. (1),
41

346

J. Nucl. Sci. Technol.,

Expressing

perturbation

terms, Eq. ( 3 ) becomes

As the pressure drops from the inlet to the outlet in two for the equilibrium inlet velocity C' and u12,)in the steady P,PT=DPTV). Equation ( 4 ) becomes

channels state D

are

equal

to that

(5)

Substituting

for the first term of Eq. ( 6 )

We

obtain

The frequency

response

of Eq. ( 8 ) agrees

with the one of the following

equations

because DPT(i-p/o) is occurred by it'/-;,), and i;6';',' shifted by 180- for up,,. The pulsating is flow with very small amplitude, where the linear is kept, is given z-i;,)in Eq. (14) of Ref. (6). The amplitude ratio and shift of DPT for m(d/dt)itl,' are calculated, and the former is called by gain |1G(i)| and the latter the shift in phase tT). The following equation agrees with the results of Eq.( 8 )

We call "characteristic "characteristic gain" |c|

frequency" fc that which tT of Eq. (10) is equal to 180- and that which corresponds to the characteristic freque ncy. The stable boundary and the period of flow oscillation in the unstable region are agree with the results obtained by the analytical model. That is exemplified in APPENDIX. . EXPERIMENTAL RESULTS IV

1. Effects of Different Entrance Throttlings between Two Channels A orifice was inserted at the upstream of flowmeter in channel (2) only, to observe different entrance throttlings between two channels on the instability. Now, since the pressure drop of the section Lt(in Fig. 2) was proportional to the square of the inlet flow rate, the proportional constant would represent the coefficient of entrance throttle pressure drop. The throttling coefficient in channel (1) was 110 kg,s2/m4, since a orifice was not inserted but the flowmeter and the electrode etc. throttled flow.
42

Vol.

16, No.

5 (May

1979)

347

The experimental and analytical results of the stable boundary are shown in Fig. 3. The quantity adopted for the abscissa is the throttling coefficient in channel (2). The experimental results are agree with the analytical ones. The results in the system with the identical entrance trottling in two channels are reploted from a companion paper(7) in the figure, that is, a dashed line (1) represents the results in the system where the throttling coefficients in two channels are equal and the values adopted for the abscissa and a dashed line (2) represents the one in the system where the throttling coefficients are equal and 110kg,s2/m4 in two channels. To compare with the result of Fig. 3, the average throttling coefficient channels is introduced as in two

In as

this a

case, dashed close

the

stable line in line results.

boundary Fig. 4.

represents It becomes represents

Fig.

Stable boundary in reference different entrance throttlings

to

very the

to

solid

which

experimental With the the view

of

gaining of

a better flow

inoscil-

sight lation, unstable curve.

into

characteristics runs were

further region The flow

conducted by the were

in the stability recorded

indicated oscillations

Fig. Fig. 4 Stable boundary in reference different entrance throttlings to

Variation oscillation different

nature of flow in reference to entrance throttlings

of

43

348

J.

Nucl.

Sci.

Technol.,

by oscillograph

and the averages

of 20~30

cycles were retained

as data.

The typical experimental results are shown in Fig. 5 at C,P=110kg,s2/m4 and C7= 298 kg,s2/m4. The results(7) in the system with the same entrance throttling in two channels are replotted. The amplitudes and periods of flow oscillations lie between the ones at CR=CR'=110 kg,s2/m4 and =298 kg,2s/m4. From the results in Figs. 4 and 5, the instabilities in the system with the different entrance throttlings between two channels show a behavior simular to equal parallel channels with the average throttling coefficient of two channels. 2. Effects of Different Heater Lengthes between Two Channels Shortening the heater in channel (2) only, the different input powers were artificially created between two channels in this case. The stable boundaries obtained experimentally and analytically are shown in Fig. 6, in which the average inlet velocity of both channels are used and LB)=0.4 m and LB)=0.3 m mean the heater lengthes of channels (1) and (2) respectively. Close agreement is obtained between the experimental results and the analytical ones. The stable boundaries obtained in the system with the same heater length are shown in the figure at LB=0.3 m
and =0.4 To the m. consider the results length shown in Fig. 6, to

average

heater

is introduced

The analytical the equal shows Fig. 6. Runs, unstable curve. observed 8 at the heater almost results lengths. results heater good

results lengthes agreement

in the system, of both with channels in Fig. the shown

where are 7. It of

to 0.35 m, are

results

further, region The in the in

were

conducted by and are the the m as with heater The in the

in the stability periods in Fig. well the as was same length present different lies beFig. 6 Stable boundary in reference to different heater lengthes different are where with entrance flow own one throttlings condition

indicated amplitudes region the

shown

LB=0.4 lengthes.

m and LB)=0.3 system on it again the The oscillation condition(7)

period

independent confirm

in equal channel

The oscillation

amplitude system
different to flow the

tween the ones in the =0.3 m and 0 .4 m.

of LB=LB
heater oscillation. instabilities are nearly equivalent in equal, to lengthes Even the

and if

the there

different the the

system agree average

characteristic in the system in two

channel condition

almost the

operating

condition

44

Vol.

16, No. 5 (May

1979)

349

Fig. Fig. 7 Stable boundary heater in reference lengthes to different

Variation oscillation different

of nature of flow in reference to heater lengthes

3. Effects of Different Heat Fluxes between Two Channels By contacting a magnet switch (9) (shown in Fig. 1) arranged in parallel to the heater in channel (2), the heat flux of channel (2) differed from the one of channel (1). The difference of thermal conditions was articially created between two channels. As parameter expressing the magnitude of the different heat fluxes between two channels, a rate Rq of heat flux was introduced, which is defined by

This The The age

value

was

varied boundary results

with

the

variable agree flux.

resistence with the fact

(8) (in Fig. 1). and analytically ones. very are shown case, and figure of more in Fig. the stable useful shows flux 9. analytical may be In this important obtained 10. the The ratio and averwith for to the in-

stable flux the The the in this were tendency

obtained almost heat boundary in the

experimentally increases That and The condition

experimental heat

at the ratio stable

stable of boundary one

and the system in Table 1 was in Fig. with

becomes

increasing design. ascertain same creases Runs oscillations The typical comes flux and clear the

analytically heat

aforementioned as the condition. conducted in the

results in Fig. 9.

is shown stabilization region

unstable

and the

amplitudes

periods

of flow

were observed in various average experimental results of oscillation from ratio the figure flux. that the amplitude are the of heat These

heat fluxes and the amplitude are shown ratio same decreases trends as the with

ratios of heat flux. in Fig. 11. It bethe average boundary. heat stable

45

350

J. Nucl. Sci. Technol.,

All using as

experimental the average in Fig. preheating

results time 12.

of

oscillation for passage induce mainly both

period

fall

along

a single

line

when in two for of the

arranged channels, passage system

required The results affects

through an account the

preheating that the

region time

shown

required

through with the The from is each

region heat required in the fluxes for cases from

characteristic

frequency

different times other

between passage of the the

channels. preheating heat fluxes cases region between with other in two two channels channels. differ This

through different

significant

difference

afore-mentioned

different

conditions.

Fig.

Stable

boundary heat

in reference fluxes of loop operation Fig. 10 Analytical boundary different results of stable in reference to heat fluxes

to different Table 1 Standard

conditions

Fig.

11

Flow

oscillation of heat

with flux

change

in ratio

Fig.

12

Period of flow oscillation in reference to time required for passage through predheating region

46

Vol.

16, No. 5 (May

1979)

351

4.

Transient on as the

Experiment stability in of heat the Fig. curve 13 obtained were of in examined the two inlet in Fig. by 9, using on is two kinds magnet the of transient in response order condiand the to

Based experiment study tion, the when

shown

switch, instability.

effects the

history are equal

velocity

The A"

fluxes

channels,

called "Condition

one, when the heat fluxes differ between two channels, is called "Condition B". The transient response experiments were carried out at Ft,,,= 0.23 m/s, Rq=0.305 and the other conditions in Table 1. The following became clear through two kinds of the transient experiment. (1) The figure of transient inlet velocity change does not reappear, owning to the phase of flow oscillation when step change added and to the heterogeneity of two-phase flow.

(2) Transient time is within 5 s. (3) The instability depends not on the history of the inlet velocity but on the opperating conditions, that is, the stability curve agrees well with the one in Fig. 9. (4) The amplitudes and periods of flow oscillation agree with the ones in Figs. 11 and 12 within the experimental errors after a short interval from the step change added.
The above-mentioned and the the inlet results instability velocity but coincide in only the on with system the the depends opperating papers(6)(7) history of

Fig.

13

Operating transient obtained

condition experiment in

of

conclusion not on

companion and the

the

disturbance

conditions.

V.

DISCUSSION

As a result of studying the effects of three kinds of different thermo-hydraulic conditions between two channels on the instability, the generated phenomena are classified into two types : (1) (2) The generated instability is simular to the one in the system with the same conditions equivalent to the average conditions in two channels.

The system becomes more stable as increasing the difference of the operating conditions between two channels. The system with the different entrance throttlings or with the different heater lengthes corresponds to type (1), and the system with the different heat fluxes does to type (2). To study the system with the different heater lengthes concerning type (1) phase analysis was applied to two operating conditions in Table 2, situated in the unstable region. The characteristic gain and frequency of the system and the shifts in phase of each channel are shown in Table 3. The difference of flow distribution between two channels are about 6%. The gain curves move parallel and the shift curves of phase almost agree for this extent of flow distribution, as shown in Fig. 14 replotted from a companion paper(7). The effects of the heater length on the instability show almost the same trend as the ones of flow distribution, as had discussed in it. Consequenly,
47

352

J. Nucl. Sci. Technol.,

Table

the difference

of the shift in phase is slight

and the characteristic gain of the system almost agree with the arithmetic sum of the gains of two channels. For this reason, the instability in the system with the different entrance throttlings or with the different heater lengthes almost agrees with the one in the system with the same condition equivalent to the average condition in two channels. To study the system with the different heat fluxes concerning type (2), phase analysis is applied to two operating conditions in Table 4 situated in the unstable region. The characteristic gains and frequencies of the system and the shifts were obtained in Table 5. The characteristic gain is 1.053 in case (i) and 1.138 in case (ii). The system with the different heat fluxes is more stable than the one with the same heat fluxes.
Table 4 Operating to differents conditions heat in reference fluxes Table 5 Analytical results adopted for Table 4

Fig.

14

Frequency by gain and

response shift

shown in phase

The reasons the characteristic

for decreasing of the characteristic gain was investigated . Regarding frequency in the system, the following equation is established : G'" Isin .61)+1 Gm sin 71)=0 . (14)

The gain and the own characteristic frequency in channel (2) with lower heat flux are lower than these in channel (1). The shift in phase increases sharply above 180- and the gain decreases with increasing the frequency. Consequently,|1G(2)| becomes fairly lower than the own characteristic gain of channel (2). On the other hand, the shift in phase decreases G(1)| increases slightly slightly from 180- in channel (1), because of Eq. (14) and |1G("I>G(2)|. in comparison with the own characteristic gain of channel (1) . |
48

Vol.

16, No.

5 (May

1979)

353

For these reason, the decrease becomes more stable. Since the following equation G As increasing the difference than the arithmetic average
The gains |G(1)| shown

in |G(2)|

surpasses

the increase

in |G(1)| , and the system gain

is established

concerning

the characteristic

1 cl-=2I G'" Icos z-'71'+I G(2' Icos r7(15) of shift in phase between two channels, IGe I becomes lower of |G(1)| and |G(2)|, and the system becomes stable.
on Table 5 Table 6 Analytical results adopted for Table 4

were obtained by using the average mass of inertia term m (Eq. (9) ), whereas the gains |G'(i)| were obtained by using m(i) at fc, as shown in Table 6. The arithmetic average of |G(i)| is 1.073 and the one of |G'(i)|1.113. The difference term. of boiling boundary in two channels

causes

the

different

masses

of

inertia

Both heat

results flux. From the

are

almost

equal

each

other,

and ZB

becomes

lower

with

increasing

the

afore-mentioned the different

discussion, masses

it

was

found term

that between

the two

different channels

characteristic stabilize the

frequencies system.

and

of inertia

VI.

CONCLUSIONS

Close agreement between the experimental results and the analytical ones is obtained in regard to the stable boundary. The reliability of the analytical model and the generality of the experimental results are confirmed. The results through phase analysis agree with the analytical ones of proposal model as for the stable boundary and the oscillation periods in the unstable region. It is made clear that the phase analysis is applicable to the instability in the system with the different thermo-hydraulic conditions between two channels. In parallel-channel boiling system, a characteristic oscillation is independent of the magnitude and nature of initial disturbance and of the history of inlet velocity without reference to the existence of the different conditions between two channels. In the different flow conditions where the own characteristic frequencies are approximatelly equal in two channel, such as in the different entrance throttlings and heater lengthes, the instability almost agrees with the one in the system with the same flow condition equivalent to the average operating condition of two channels. On the other hand, the system, where the flow conditions as well as the characteristic frequencies and the masses of inertia term differ from two channels, becomes more stable with increasing the difference of flow condition, and the system is most unstable when the flow conditions are equal in two channels, such as the system with the different heat fluxes. This conclusion may be interesting stable, particularly possessing two parallel channels.
49

for designing

the system

to be

354

J. Nucl. Sci. Technol.,

CR: g: G| : Gc|: LB: m: P: PT :

[NOMENCLATURE] Throttling coefficient (kgs2/m4) q": Heat flux (kcal/m2h) Acceleration due to gravity (m/s2) u : Velocity (m/s) Gain ZE: Boiling boundary (m) | Characteristic gain gl: Density of liquid |(kg/m3) Heater length (m) tT Shift in phase Mass of inertia term (kgs2/m3),see Eq. (14) in Ref.(6) Subscripts Pressure (kg/m') in: Inlet, out: Outlet Total pressure drop (kg/m2), see Eq. (14) D in Ref.(6)
REFERENCES -

(1) STENNING, A.H.: Trans. ASME, Ser. D, 86, 213~217 (1964). (2) MEYER, J.E., ROSE, R.P.: ibid., Ser. C, 85, 1~9 (1963). (3) DAVIES, A.L., POTTER, R.: EURATOM Rep., Proc. Symp. on Two-Phase Flow Dynamics, hoven, p. 1255~1266 (1967). (4) D'ARCY, D.F.: ibid., p. 1173~1223. (5) CROWLEY, J.D., et al,: ibid., p. 1132~1171. (6) ARITONII, M., et al.: J. Nucl. Sci. Technol., 14(1), 22~30 (1977). (7) ARITOMI, M., et al.: ibid., 14(2), 88-96 (1977). (8) STENNING, A.H., VEZIROGLC, T.N. : ASME Pap. 64 WA/FE-28, (1964).

Eind-

[APPENDIX] The results obtained through the phase analysis are compared with the ones obtained through the analytical model, in order to examine the reliability of phase analysis. The characteristic gain of the system corresponds to the characteristic frequency, which is shifted by 180- through phase analysis, and were obtained using the average heat flux in two channels as parameter. The results are shown in Fig. A1 as compared with the ones obtained through the analytical model. The characteristic gain increases
with the teristic increasing average gain heat the average flux, where unity, becomes heat the flux, and charac-

is 0.34x106

Fig. Fig. A1 Characteristic in reference gain and frequency to heat flux 50

A2

Comparison between phase analysis and mathematical model for values of oscillation period

Vol.

16, No. 5 (May

1979)

355

kcal/m2,h. On the other hand, the results obtained through the analytical model shown in Fig. 10 indicate that the average heat flux above 0.35 x 10' kcal/m2,h is lain in the unstable region and the one under 0.325 x10' kcal/m2 h in the stable region. The stable boundary obtained through the phase analysis agrees thus with the one obtained through the analytical model. The oscillation periods obtained through the analytical model are compared with the reciprocals of the characteristic frequency based on phase analysis. Figure A2 shows both results and they are agree within 2%.

51