Nuclear Technology

ISSN: 0029-5450 (Print) 1943-7471 (Online) Journal homepage: http://www.tandfonline.com/loi/unct20

Analysis of Leaks Through Microchannel Cracks

Using RELAP5-3D

Angelo Frisani & Yassin A. Hassan

To cite this article: Angelo Frisani & Yassin A. Hassan (2009) Analysis of Leaks Through
Microchannel Cracks Using RELAP5-3D, Nuclear Technology, 167:2, 304-312, DOI: 10.13182/
NT09-A8965

To link to this article: http://dx.doi.org/10.13182/NT09-A8965

Published online: 10 Apr 2017.

Submit your article to this journal

Article views: 1

View related articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=unct20

Download by: [University of Florida] Date: 05 November 2017, At: 18:20

 ANALYSIS OF LEAKS THROUGH THERMAL HYDRAULICS

MICROCHANNEL CRACKS KEYWORDS: RELAP5-3D, micro-

channel, leakage
USING RELAP5-3D
ANGELO FRISANI* and YASSIN A. HASSAN Texas A&M University
Department of Nuclear Engineering, 129 Zachry Engineering Center
College Station, Texas 77843-3133

Received June 30, 2008

Accepted for Publication December 3, 2008
Downloaded by [University of Florida] at 18:20 05 November 2017

The purpose of the present work is to study the flow possible to analyze the dependence of channel pressure
leakage through postulated microchannels. In the frame- drop as a function of the Reynolds number and wall rough-
work of the leak before break, it is reasonable to assume ness. For high values of the Reynolds number, the pres-
that a detectable leak develops before a large break oc- sure drop showed very little influence of the Reynolds
curs. A large pressure difference may exist between the number over the fluid conditions inside the microchan-
crack inlet and outlet; the fluid residence time is so brief nel. On the other hand, the wall roughness strongly in-
that thermodynamic equilibrium conditions cannot be fluences the channel pressure drop and, consequently,
reached within the crack. Using REL AP5-3D system code, the critical mass flow rate through the crack. The
the modified Henry’s homogeneous nonequilibrium model REL AP5-3D wall friction correlation was compared with
was adopted to simulate the fluid condition at the choked various available models in the literature, such as John
point. In channels with large L/DH , mechanical equilib- et al. (1987), modified Karman, Nikuradse (1933), and
rium between the phases is usually reached. On the other Button et al. (1978). These correlations predict similar
hand, because of the small residence time, thermal equi- values for the friction factor.
librium may not be achieved. Thus, the critical flow The REL AP5-3D model was also in agreement with
through the crack is kinematically homogeneous, but modified Karman correlation for the studied wall rough-
thermodynamically in nonequilibrium conditions. In this ness values. However, it underestimated the friction fac-
investigation, various channel cross-flow areas were con- tor with respect to John’s formula. This indicates that the
sidered, each having a sensitivity study performed in ref- crack critical mass flow rate predicted by REL AP5-3D is
erence to the wall roughness. In this approach it was larger than that calculated using John’s correlation.

I. INTRODUCTION The reference scenario analyzed was the opening of

a small crack ~on the order of microns! through the wall
One of the accident scenarios in nuclear safety de- of a pressurized water reactor at reference pressure of
sign is the rupture of the high-pressure piping system. almost 15 MPa. At this high pressure, critical flow was
The criterion generally used to analyze this event is the expected to develop inside the crack channel. A wide
instantaneous double-ended guillotine rupture of the pip- hierarchy of models for the critical flow rate prediction
ing. Recent advances in fracture mechanics 1–3 have dem- exists in the literature ~Refs. 4 through 7, among others!.
onstrated that certain pipes under specific conditions The most important physical aspects to consider are non-
cannot fail in a way that results in instantaneous guillo- equilibrium phase-change phenomena, slip between the
tine rupture. Instead, the postulated accident scenario phases, and three-dimensionality of the flow. In general,
may evolve in a leak at a detectable rate. This is known a two-phase critical flow model can be characterized as
as leak before break ~LBB!. The term LBB normally homogeneous equilibrium models ~HEMs!, nonhomo-
refers to nuclear power plant high-pressure components. geneous equilibrium models ~NEMs!, homogeneous non-
equilibrium models ~HNMs!, and nonhomogeneous
*E-mail: [email protected] nonequilibrium model ~NNMs!. Nathwani et al.8 have

304 NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS

extended the homogeneous frozen model ~HFM!, devel- change at the break location. The transition from single-
oped by Henry and Fauske 9 for critical flow of initially to two-phase flow is accompanied by a discontinuous
saturated liquid, to one-dimensional critical flow through change in the fluid bulk modulus, especially for the liquid-
cracks of initially subcooled liquid for LBB analysis. to-liquid0vapor0gas transition.
John et al.10 have developed a modified form of Pana’s 11 Ideally, the two-phase choking criterion can be used
model to study critical flow through rough slits. as a boundary condition for obtaining flow solutions.
In the present work high-energy flow through cracks However, a simplified model 15 is implemented in
was considered. The flow through a pipe is characterized RELAP5-3D, such that the two-phase fluid is assumed
by to be homogeneous. Additionally, choking is assumed
to occur at the narrowest part of the flow channel ~i.e.,
1. a large L0DH ratio
the throat!. Under critical flow conditions, the fluid ve-
2. a relatively small residence time locity at the throat is equal to the two-phase speed of
3. a large relative surface roughness. sound.
For critical flow through ducts with large L0DH II.A. Modified Henry’s HNM
ratios, both mechanical and thermal equilibrium be-
tween the phases are usually established; however, for In the case of a thin orifice plate of 10-mm thickness
flow through cracks and slits, the relatively small resi- with slightly subcooled liquid, the amount of thermal
Downloaded by [University of Florida] at 18:20 05 November 2017

dence time does not allow the attainment of thermal nonequilibrium at the throat calculated employing the
equilibrium—even for large L0DH ratios. Since the time Henry-Fauske model is small compared to actual condi-
required for the fluid to reach thermal equilibrium when tions. The calculated critical flow rates are 40 to 50% less
flowing to regions of lower pressure is of the same than experimental results.15 Henry developed a modified
order of magnitude of the residence time through the version of the original HNM to predict a higher level of
crack, this leads to nonequilibrium vapor generation rates. thermal nonequilibrium at the throat. The model maxi-
The degree of thermal nonequilibrium increases as the mizes flow rate with respect to exit pressure. The process
channel diameter is reduced. Because of the high criti- is assumed isentropic. The expression for the critical flow
cal velocity, the two-phase flow is homogeneous, even rate is

再 冨冎
though the short residence time does not allow a ther-
1 ]Vg ]Vf ]x f low
mal equilibrium. ⫽ x f low ⫹ ~1 ⫺ x f low ! ⫹ ~Vg ⫺ Vf ! ,
In the present work the modified Henry’s HNM was GC2 ]P ]P ]P S t
selected to determine the critical mass flow rate through
a postulated reactor vessel wall crack. ~1!
where
II. RELAP5-3D CHOKED-FLOW MODELS x flow ⫽ flow quality
Vg ⫽ vapor specific volume
Trapp and Ransom developed the standard
RELAP5-3D choked-flow model for calculation of the Vf ⫽ liquid specific volume
mass discharge from the system at a pipe break or a P ⫽ pressure.
nozzle.12 An optional choked-flow model ~modified
Henry-Fauske! is also available in RELAP5-3D In the modified model the following conditions are
~Refs. 13, 14, and 15!. Choking is defined as the con- assumed:
dition wherein the mass flow rate becomes independent
1. homogeneous flow
of the downstream conditions ~i.e., further reduction in
the downstream pressure does not change the mass flow 2. isentropic flow
rate!. Choking occurs when acoustic signals can no lon- 3. liquid is incompressible
ger propagate upstream as a result of the fluid velocity
4. vapor is at equilibrium.
exceeding the acoustic propagation velocity. In these
conditions, no information can propagate into the solu- The assumption that the fluid is homogeneous, i.e.,
tion region from the exterior. The thermal equilibrium the slip ratio is unity, rises from the fact that the effect of
assumption with slip ~i.e., NEM! is used as the basis thermal nonequilibrium at high pressures becomes more
for RELAP5-3D choked-flow criterion when two-phase significant than mechanical nonequilibrium, such that
conditions exist. the effect of slip can be neglected.
In the Henry-Fauske model, the two-phase condi- Moreover, there is a short amount of time for mass
tions are assumed at the break. However, in the early transfer to take place, and it is reasonable to assume that
phase of leakage, the flow approaching the break is sub- the amount of mass transferred during the expansion is
cooled liquid. Therefore, the fluid undergoes a phase negligible. Similarly, the amount of heat transferred

NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009 305

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS

between the phases during the expansion is also negligi- II.B. Calculation Methodology
ble, so that the liquid temperature remains essentially
constant. In Henry’s model the thermodynamic nonequilib-
Since wall shear, heat transfer to the environment, rium effects are considered by introducing an empirical
and interfacial viscous terms are neglected, the system parameter N in the expression for vapor generation rate.
entropy during the expansion is assumed constant. With The nonequilibrium factor N is a function of the equilib-
the hypotheses stated before and the assumption of neg- rium quality and the crack length0hydraulic diameter
ligible amounts of interphase heat and mass transfer, each ratio. The total pressure drop through the crack is the sum
phase expands isentropically. of the following contributions:
This eliminates the need to calculate the liquid spe-
cific volume and the quality at the throat and also pro- DPTOT ⫽ DPe ⫹ DPf ⫹ DPaph ⫹ DPaa , ~5!
vides a relationship for the vapor specific volume in terms where
of the throat pressure and the upstream conditions ~i.e.,
vapor is at equilibrium with pressure at the throat!. With DPe ⫽ entrance pressure drop
assumed negligible heat transfer during expansion, there
is a large temperature difference between the phases at DPf ⫽ friction pressure drop ~the effect of friction
the throat, which means the local rate of heat transfer is significant because of the large values of
Downloaded by [University of Florida] at 18:20 05 November 2017

should be large. Because of the assumptions that the vapor L0DH !

behavior at the throat follows a polytrophic process and
the liquid can be considered incompressible, the only DPaph ⫽ frictionless momentum pressure drop due
term to be evaluated is the derivative of the quality with to phase change
respect to pressure at the throat.
DPaa ⫽ acceleration pressure drop due to area change
The equilibrium quality x eq is defined as
along the crack.
S0 ⫺ Sf, eq Because of the large value of L0DH , the friction pres-
x eq ⫽ , ~2! sure drop is the largest pressure loss contribution in mi-
Sg, eq ⫺ Sf, eq
crochannel configurations. It is calculated by separately
where evaluating the Darcy-Weisbach expression for the liquid
region and the two-phase flow region, and it is is given as
S0 ⫽ entropy at the throat location DPf ⫽ ⫺21 Z f f Gc2 vlo ⫹ ⫺21 ~Z ⫺ Z f ! f Gc2 @vl ⫹ x c ~vgc ⫺ vl !# ,
Sf, eq ⫽ liquid equilibrium entropy
~6!
Sg, eq ⫽ vapor equilibrium entropy.
where Z f ⫽ L f 0DH and L f is the location of the onset of
The effect of thermal nonequilibrium on the mass trans- flashing; Z f is typically set equal to 12.
fer rate in terms of the equilibrium rate is given as The turbulent friction coefficient f calculated by the
Zigrang-Sylvester approximation 14 is expressed as

dP 冨 dP 冨
dx dx eq 1
⫽N , ~3! ⫽ ⫺2 log 10
t t Mf
where the nonequilibrium factor N is assumed to be only
function of the throat equilibrium quantity. The final ex-
pression for the critical value of the mass flow rate can be
⫻ 再 «
3.7D
⫹
2.51
Re
冋
1.14 ⫺ 2 log 10 冉 «
D
⫹
21.25
Re 0.9
冊册冎 ,

approximated as
~7!
1
GC2
再
⫽ x
Vg
gpc
⫹ ~Vg ⫺ Vf, 0 !N
dx eq
dP
冎 t
, ~4! where
« ⫽ wall roughness
where g is the isentropic exponent and pc is the critical D ⫽ hydraulic diameter
pressure. If the thermal nonequilibrium factor N is equal
to one, the prediction of the mass flow rate is close to the Re ⫽ Reynolds number.
HEM. If N is equal to zero, the solution is approximately
the HFM. The quantity N attempts to correlate the partial This turbulent friction factor is an approximation of the
phase change occurring at the throat. Colebrook-White correlation.14 It is valid for Re ⱖ 3000.

306 NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS

III. DESCRIPTION OF THE MICROCHANNEL MODELING The choking condition has been applied to the last
single junction ~i.e., CV-311!. Since the pipes CV401
The RELAP5-3D nodalization model ~Fig. 1! con- through CV410 have a very small volume dimension
sists of a time-dependent volume ~TDV! component CV- ~volume length ⫽ 1.6035 ⫻ 10⫺5 m!, the simulation pre-
201, simulating the inlet fluid boundary condition, i.e., dicts a transient of 0.002 s. A simulation time of 0.02 s
fluid pressure ~14.966 MPa! and temperature ~591.5 K!. was chosen ~i.e., ten times larger than the transient!. The
A single junction ~CV-301! connects CV-201 to a pipe period chosen is sufficient for reaching steady flow con-
component ~CV-401!. The latter represents the micro- ditions. With the dimension set, a maximum time step of
channel inlet. To simulate the crack, ten pipe components 10⫺8 s is necessary because of the very small Courant
~CV-4010CV-410! have been used. Each pipe consists of number, which results in an average computer time per
99 volumes. run of 24 h ~Pentium 4, 2.33 GHz!.
For each pipe volume it is possible to define the
x-coordinate volume flow area Ax ⫽ h ⫻ t, the y-coordinate
volume flow area Ay ⫽ h ⫻ l, and the z-coordinate vol- IV. RELAP5-3D SIMULATIONS
ume flow area Az ⫽ l ⫻ t, where h represents the crack
height, t represents the crack thickness, and l represents
Different crack cross-flow areas were analyzed. For
the length of each pipe volume, respectively ~see Figs. 1
each configuration, different wall roughnesses « were
Downloaded by [University of Florida] at 18:20 05 November 2017

and 2!. The total crack length L is given by the length of

also tested, in order to determine the influence of this
each pipe volume l multiplied by the number of volumes
parameter over the mass flow rate. In turn, if the wall
for each pipe ~99! and the number of pipes ~10!, which is
roughness is changed, the Reynolds number also changes
equal to L ⫽ 10 ⫻ 99 ⫻ l. The reference geometry has
because of a change in the mass flow rate through the
been chosen with a length l of 1.6035 ⫻ 10⫺5 m, a height
system. Table I summarizes all the performed analyses.
of 46.355 mm, and a thickness of 26 mm for all pipe
Figure 2 shows the crack geometry, where t repre-
volumes. The total crack length is 15.875 mm. The pipes
sents the thickness, h represents the height, and L rep-
have been linked to one another with single junctions
resents the total crack length. The pressure drop between
~CV-3020CV-309!. Single junction CV-311 links the last
crack inlet and outlet is set as described in Sec. III
pipe—representing the microchannel outlet ~CV-410!—to
~Pin ⫽ 14.996 MPa, Pout ⫽ 0.101 MPa, and Dp ⫽ 14.895
the TDV CV-202 representing the system outlet ~ p ⫽
MPa!. If the wall roughness is changed, then the pres-
0.101325 MPa, T ⫽ 353 K!. Since the fluid is choked at
sure distribution along the crack will change as a result
the outlet, the boundary condition chosen for CV-202
of the changed friction factor. Consequently, the onset
does not influence the flow conditions upstream.
of two-phase flow shifts to different locations, and the
critical mass flow rate will also change. This allows

TABLE I
Various Investigated Cases
Fig. 1. RELAP5-3D nodalization. t h L Roughness DH
Case ~ mm! ~mm! ~mm! ~ mm! ~ mm!

1 26 0.026 16.5033 0.0 26

2 26 0.026 16.5033 5.0 26
3 26 0.026 16.5033 10.0 26
4 100 0.010 16.5033 20 100
5 100 0.010 16.5033 20 100
6 26 46.335 15.875 0 52
7 26 46.355 15.875 5 52
8 26 46.355 15.875 10 52
9 26 46.355 15.875 20 52
10 26 46.355 15.875 23 52
11 26 46.355 15.875 25 52
12 50 46.355 15.875 30 100
13 26 25.4 15.875 23 52
14 26 25.4 15.875 25 52
15 26 20.32 15.875 23 52
16 26 12.7 15.875 23 52
Fig. 2. Crack geometry.

NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009 307

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS

one to address the influence of the turbulent friction been taken into account by increasing « from zero up to
factor calculated by RELAP5-3D. The friction factor 25 mm, as shown in Fig. 3. Since the inlet0outlet pres-
modeled in RELAP5-3D ~Ref. 14! @Eq. ~7!# was com- sure drop has been specified, increasing the wall rough-
pared to several other correlations as the John et al. ness reduces the mass flow rate. Figure 4 shows the
correlation 10 @Eq. ~8!#, Karman correlation 16 @Eq. ~9!#, change in the pressure trend for the different wall rough-
Nikuradse correlation 17 @Eq. ~10!#, and Button correla- nesses. The case with zero wall roughness exhibits the
tion 18 @Eq. ~11!#: smallest pressure gradient along the pipe, while the case

冉 冊
with the largest wall roughness gives the highest pres-
1 dH sure gradient. In turn, the variation in surface rough-
⫽ 3.39{log 10 ⫹ 0.866 , ~8!
Mf 2R ness determines a different location for the onset of
two-phase flow, since the saturation conditions are
1
Mf 冉
⫽ 2{log 10
dH
2R
⫹ 1.74 冊 , ~9!
reached earlier in the case where the pressure gradient
is larger ~see Fig. 5!. It is also interesting to note in
Fig. 4 the strong pressure gradients close to the choked

冉 冊
point ~i.e., where two-phase flow condition exists! shown
1 dH for all wall roughnesses. It is worthwhile to note that
⫽ 2{log 10 ⫹ 1.14 , ~10!
Mf 2R the maximum wall roughness selected is very close to
Downloaded by [University of Florida] at 18:20 05 November 2017

the channel thickness. The aim was to address the

and RELAP5-3D response when the wall roughness is al-

冉 冊
most the same value of the channel thickness. Figure 4
1 dH also shows a change in the channel pressure inlet for
⫽ 2.25{log 10 ⫹ 1.25 , ~11!
Mf 2R the different wall roughnesses. Since the pressure in the
TDV upstream from the crack has been set to 14.996
where dH represents the hydraulic diameter and R repre- MPa, the differences at the crack inlet are due to the
sents the surface roughness ratio ~i.e., the average rough- pressure gradient obtained by changing «. When the
ness height of the crack surface!. wall roughness is increased, a steeper pressure gradient
is obtained and, therefore, the pressure at the channel
crack inlet is closer to the value set in the TDV.
V. RESULTS AND DISCUSSION If the wall roughness is held constant and different
channel heights are considered, the flashing begins at
In Table I, different cross-flow areas have been an- the same location. This means the pressure gradient is
alyzed. The reference geometry corresponds to h ⫽ not changed, as seen in Fig. 6. This behavior can be
46.355 mm, t ⫽ 26 mm, and L ⫽ 15.875 mm. For this addressed considering the definition of the Reynolds
configuration, the effect of the wall roughness has number:

Fig. 3. Microchannel mass flow rate function of wall roughness.

308 NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS
Downloaded by [University of Florida] at 18:20 05 November 2017

Fig. 4. Microchannel pressure trend function of channel length.

Fig. 5. Void fraction trend function of channel length.

rVD rVAD _
mD m_ 4A 4A
Re ⫽ ⫽ ⫽ ⫽ D⫽ ~13!
m mA mA mA Pw Pw
A ⫽ channel cross-flow area ~A ⫽ h{t !
4 m_ 2 m_ Pw ⫽ wetted perimeter @Pw ⫽ 2~h ⫹ t !# .
⫽ ⫽ , ~12!
mPw m~h ⫹ t ! The mass flow rate changes almost proportionally to the
channel height; therefore, the Reynolds number shows
where almost no dependence from the channel height. If the
wall roughness is not changed, the weak dependence of
r ⫽ fluid density
the turbulent friction coefficient f using the Reynolds
V ⫽ fluid velocity number demonstrates the pattern behavior of Fig. 6; i.e.,
this was an exercise for checking that the correlations are
m ⫽ fluid dynamic viscosity correctly implemented in the code.
As seen in Fig. 7, the fluid flow enters the channel
D ⫽ hydraulic diameter, defined as as subcooled. Once the pressure reaches saturation

NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009 309

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS

Fig. 6. Void fraction trend function of channel length for different heights ~h ⫽ 46.355 mm, h ⫽ 25.4 mm, h ⫽ 20.32 mm, and
Downloaded by [University of Florida] at 18:20 05 November 2017

h ⫽ 12.7 mm!.

Fig. 7. Liquid0vapor0saturation temperature.

conditions for the temperature set, the fluid starts to flash hand, Fig. 8 shows that the friction factor strongly de-
and the void fraction increases. It is worthwhile to note pends on the wall roughness.
that when using the stated assumptions, the choked-flow In Fig. 9 is shown the comparison for the friction fac-
models calculated different temperatures for the liquid tor between RELAP5-3D, John, modified Karman, Ni-
and vapor phases at the throat. On the other hand, both kuradse, and Button correlations. Figure 9 shows that for
the liquid and vapor velocities, though different in the small roughness, all the correlations provide approxi-
adjacent upstream node, were equal to the sonic velocity mately similar friction factors. In particular, the modified
at the choked point. Karman correlation is very close to the Zigrang-Sylvester
Figure 8 presents the RELAP5-3D friction factor func- approximation implemented in RELAP5-3D over the en-
tion of the Reynolds number for different wall rough- tire range of wall roughnesses analyzed. The other corre-
nesses. As stated before, once the wall roughness has lations tend to overpredict the wall friction factor. Since
been set, the friction factor slightly changes with the the John model underestimates the critical mass flow rate
Reynolds number in the region where the Zigrang- with an error of ,20% with respect to the experimental
Sylvester approximation of the Colebrook-White corre- data,10 the modified Henry-Fauske HNM gives a critical
lation is applicable ~i.e., for Re ⱖ 3000!. On the other flow rate closer to the observed data. The friction factor

310 NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS
Downloaded by [University of Florida] at 18:20 05 November 2017

Fig. 8. RELAP5-3D friction factor function of Reynolds number ~h ⫽ 46.355 mm and t ⫽ 26 mm!.

Fig. 9. Comparison of different friction factor models.

estimate is important in crack flow determination. A cor- calculated values using RELAP5-3D are larger than those
relation that overestimates the friction factors inordi- predicted by John’s formula.
nately gives an increase in the pressure gradient along the Also, different analyses were performed using
crack, which means saturation conditions are reached at a RELAP5-Mod3.3 for a crack with a larger cross-flow
point closer to the inlet for a subcooled flow. Therefore, area in the pressure range 4.0 to 9.5 MPa ~Ref. 19!. The
there will be an increased void fraction at the throat, and results have been compared to experimental results.20
the critical mass flow rate will be less. In these analyses the HFM was used. The comparison
In the reference geometry ~h ⫽ 46.355 mm, t ⫽ 26 showed a good agreement at low pressure for the criti-
mm, l ⫽ 15.875 mm, and DH ⫽ 52 mm! using John’s cal mass flow calculated and the experimental value.
correlation, a mass flow rate of 0.01013 kg0s for a wall
roughness of « ⫽ 5 mm is estimated. As can be seen from
Fig. 3, RELAP5-3D calculated mass flow rate decreases VI. CONCLUSIONS
from 0.03571 kg0s for « ⫽ 0.0 mm to 0.01305 kg0s for
« ⫽ 25.0 mm. However, for small wall roughnesses the The analysis performed with RELAP5-3D showed
two wall friction correlations are very close ~Fig. 9!. The that the modified Henry-Fauske HEM gives a critical

NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009 311

 Frisani and Hassan LEAKS THROUGH MICROCHANNEL CRACKS

flow rate slightly larger than that predicted with the John Vol. 1: Light Water Reactors, Atlanta, Georgia, November 27–
model. Since the John model underestimates the critical December 2, 1977, pp. 195–216, American Society of Mechan-
mass flow rate with an error of ,20% with respect to the ical Engineers ~1977!.
experimental data, the modified Henry-Fauske HNM gives 8. J. S. NATHWANI, B. L. KEE, C. S. KIM, and M. L. KOZ-
a critical flow rate closer to the observed data. LUK, “Ontario Hydro’s Leak Before Break Approach: Appli-
Moreover, since choked-flow conditions require fully cation to the Darlington ~CANDU! Nuclear Generating Station
turbulent flow, the RELAP5-3D wall friction assumes A,” Nucl. Eng. Des., 111, 85 ~1989!.
almost no dependence for the friction coefficient from
the Reynolds number. On the other hand, the wall rough- 9. R. E. HENRY, H. K. FAUSKE, “The Two-Phase Critical
Flow of One-Component Mixtures in Nozzles, Orifices, and
ness strongly influences the pressure drop inside the chan- Short Tubes,” J. Heat Transfer, 93, 179 ~1971!.
nel and, therefore, the onset of flashing, the void fraction
at the throat, and the critical mass flow rate. 10. H. JOHN, J. REIMANN, F. WESTPHAL, and L. FRIEDEL,
For a crack with a much larger cross-flow area in the “Critical Two-Phase Flow Through Rough Slits,” Int. J. Multi-
pressure range 4.0 to 9.5 MPa, different analyses were phase Flow, 14, 2, 155 ~1988!.
performed using RELAP5-Mod3.3. The HFM was used 11. P. PANA, “Berechnung der stationaren Massenstrom-
for these analyses. The comparison showed a very good dichte von Wasserdampfgemischen und der auftretender Ruck-
agreement for the critical mass flow calculated and the stobkafte,” Report IRS-W-18 ~1976!.
Downloaded by [University of Florida] at 18:20 05 November 2017

experimental value.
12. J. A. TRAPP and V. H. RANSOM, “A Choked-Flow Cal-
culation Criterion for Non-Homogeneous, Non-Equilibrium,
Two-Phase Flows,” Int. J. Multiphase Flow, 8, 6, 669 ~1982!.
REFERENCES
13. THE RELAP5-3D CODE DEVELOPMENT TEAM,
1. M. L. AGARWAL, M. J. KOZLUK, T. C. LIN, B. W. “Input Requirements,” INEEL-EXT-98-00834, Revision 2.3,
MANNING, and D. K. VIJAY, “A Leak-Before-Break Strategy Idaho National Laboratory ~Apr. 2005!.
for CANDU Primary Piping System,” Int. J. Press. Vessels
Piping, 25, 239 ~1986!. 14. THE RELAP5-3D CODE DEVELOPMENT TEAM,
“RELAP5-3D Code Manual Volume I: Code Structure, System
2. G. BERTHOLOMÉ, R. W. BIESELT, and M. ERVE, “LBB Models, and Solution Methods,” INEEL-EXT-98-00834, Re-
for KWU-Plants,” Nucl. Eng. Des., 111, 3 ~1989!. vision 2.3, Idaho National Laboratory ~Apr. 2005!.

3. G. BERTHOLOMÉ, W. KASTNER, and E. KEIM, “De- 15. THE RELAP5-3D CODE DEVELOPMENT TEAM,
sign and Calibration of Leak Detection Systems by Thermal “RELAP5-3D Code Manual Volume IV: Models and Correla-
Hydraulics and Fracture Mechanics Analyses,” Nucl. Eng. Des., tion,” INEEL-EXT-98-00834, Revision 2.3, Idaho National Lab-
142, 1 ~1993!. oratory ~Apr. 2005!.

4. M. GIOT, “Critical Flows,” Thermohydraulics of Two- 16. B. GHOSH, S. K. BANDYOPADHYAY, S. K. GUPTA,

Phase Systems for Industrial Design and Nuclear Engineering, H. S. KUSHWAHA, and V. VENKAT RAJ, “Leak Rates
Chap. 18, pp. 405– 452, J. M. DELHAYE, M. GIOT, and M. L. Through Cracks and Slits in PHT Pipes for LBB,” Nucl. Eng.
RIRTHMULLER, Eds., Hemisphere Publishing Company, Des., 212, 85 ~2002!.
McGraw Hill Book Company, New York ~1981!. 17. J. NIKURADSE, “Stromungsgesetze in rauhen Rohren,”
VDI Forschungsh., 361 ~1933!.
5. G. B. WALLIS, “Critical Two-Phase Flow,” Int. J. Multi-
phase Flow, 6, 97 ~1980!. 18. B. L. BUTTON, A. F. GROGAN, T. C. CHIVERS, and
P. R. MANNING, “Gas Flow Through Cracks,” J. Fluids Eng.,
6. P. SAHA, “A Review of Two-Phase Steam-Water Critical 100, 453 ~1978!.
Flow Models with Emphasis on Thermal Non-Equilibrium,”
NUREG0CR00417, BNL-NUREG-50907, Brookhaven Na- 19. S. K. KANG, S. HARI, and Y. A. HASSAN, “Flow Char-
tional Laboratory ~1978!. acteristics in Microchannel Driven by High Pressure Differen-
tial,” Trans. Am. Nucl. Soc., 97, 437 ~2007!.
7. J. A. BOURE, “The Critical Flow Phenomena with Refer-
ence to Two-Phase Flow and Nuclear Reactor Systems,” Proc. 20. C. N. AMOS and V. E. SCHROCK, “Two-Phase Critical
Symp. Thermal Hydraulic Aspects of Nuclear Reactor Safety, Flow in Slits,” Nucl. Sci. Eng., 88, 261 ~1984!.

312 NUCLEAR TECHNOLOGY VOL. 167 AUG. 2009