Available online at www.sciencedirect.

com

Nuclear Engineering and Design 237 (2007) 2302–2310

New algorithm for the numerical simulation of two-phase stratified

gas–liquid flow and its application for analyzing the Kelvin–Helmholtz
instability criterion with respect to wavelength effect
M.R. Ansari ∗ , V. Shokri
Mechanical Engineering Department, Tarbiat Modarres University, P.O. Box 14115-143, Tehran, Islamic Republic of Iran
Received 15 December 2006; received in revised form 8 April 2007; accepted 12 April 2007

Abstract
In this article, the transient condition of two-phase stratified gas–liquid flow was investigated using numerical simulation. The basis of the method
involves the one-space dimensional transient solution of the governing equations using the two-fluid model. In this paper, an analogy between the
SIMPLE algorithm in two-space dimensional single-phase flow and one-space dimensional two-phase stratified flow is established through the
application of a special algorithm created to solve the conservation equations. After the modeling was established and justified, wave growth was
examined in two-phase stratified flow in a horizontal duct. The results were then compared with the results of the previously published articles. The
results show that the classical criterion for the Kelvin–Helmholtz (K–H) instability is consistent when the long wavelength with small amplitude is
considered. In this case (of the K–H instability criterion), the wavelength effect on this instability and pressure variation on the two phases interface
was consistent with prior researchers’ correlations. However, as the wavelengths decreased, the results indicated that the K–H instability criterion
is over-predicted and must be modified. The application of the present numerical simulation method improved the results, and the consistency with
the analytical solution is higher in comparison with other well-known computer codes.
© 2007 Elsevier B.V. All rights reserved.

1. Introduction nuclear engineering. However, such transient flow phenom-

ena are particularly difficult to analyze, except under certain
1.1. Literature survey simple conditions. Many articles on two-phase flow models
are available in the extant literature, such as Ramshaw and
The subject of two-phase gas–liquid flow, or in general, Trapp (1978), Bergles et al. (1981), Ishii (1975), Ransom and
multiphase flow, has become increasingly important in engi- Scofield (1976), Ransom and Trapp (1983), Wendroff (1980),
neering design and technology, particularly for oil and gas Ardron (1980), Banerjee (1980), Banerjee and Chan (1980),
pipelines, the process industry, oil platforms, and nuclear tech- Safety Code Development Group (1981), Akimoto et al. (1986)
nology, especially in nuclear steam generators and evaporators and Hirano et al. (1986). Both mathematical and experimental
of power stations. Furthermore, applications of two-phase flow investigations of two-phase flow instability criteria during wave
are relevant not only to engineering and modern scientific prob- initiation have been carried out by several researchers (see, e.g.,
lems, but also to natural phenomena, and hence, they merit Ansari et al., 1988; Ansari, 1998, 2000; Wallis and Dobson,
further investigation. The ability to understand and to model 1973; Kordyban, 1977; Kordyban and Ranov, 1979; Taitel and
the principal physical mechanisms associated with the transient Dukler, 1976). As empirical methods involve a tremendous vari-
response characteristics of these flows is of considerable impor- ety of factors, such as apparatus geometry and the physical
tance as they relate to system design and control including properties of fluids, it is essential that engineers understand
the theoretical modeling underlying the relevant applications
in order to design equipment correctly (especially in nuclear
∗ Corresponding author. Tel.: +98 21 8801 1001; fax: +98 21 8800 5040.
power stations). In addition, two-phase flow modeling, which is
E-mail addresses: mra [email protected] (M.R. Ansari), of the utmost importance for nuclear reactor system simulations,
[email protected] (V. Shokri). involves a high computational resource requirement. In recent

0029-5493/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2007.04.005
 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310 2303

Thomson (1963) as
Nomenclature  0.5
ρl − ρg
a velocity coefficient u g − ul ≥ ghg (1)
ρg
g gravitational acceleration
h height of phase Taitel and Dukler (1976) proposed a theoretical model, where
hc distance from wave crest to top wall the flow regime transitions are defined as
H height of duct  0.5
k wave number ρl − ρg
ug > K ghg (2)
K constant as defined on Eq. (3) ρg
P pressure
where
t time
 0.5
u phase velocity 2
x axial distance K= (3)
(hg / hc )(hg / hc + 1)
Greek letters Kordyban and Ranov (1979) considered waves of finite ampli-
α void fraction tude and arrived at an approximate criterion as
δ operator (represents a small perturbation from
gρl /ρg k
equilibrium) ug − ul > (4)
λ wavelength coth(khg − 0.9) + 0.45 coth2 (khg − 0.9)
ρ density
Some theoretical, but primarily experimental, research has
Subscripts and superscripts been carried out on wave initiation by Lin and Hanratty (1986,
e staggered east 1987) and Hihara and Saito (1983). Wallis and Dobson (1973)
E east presented the K–H instability criterion from the observation of
g gas phase experimental data as
i interface  0.5
ρl − ρg
k phase index (g or l) ug − ul > 0.5 ghg (5)
ρg
l liquid phase
n new time step Hancox et al. (1980) defined the instability limit by the fol-
o old time step lowing equation as
P pole
ref reference ρ∗ (ρl − ρg )
(ug − ul )2 < gH (6)
w staggered west ρl ρg
W west
where
* estimated value
ρ∗ = αl ρg + αg ρl (7)

He declared his model valid when Eq. (6) is satisfied. The

years, a number of computer programs have been developed for problem with his model is that he proposed a constant pressure
the simulation and analysis of thermo hydraulics in two-phase on the interface of two phases. However, he emphasized that a
flow, including CATHARE, RELAP, MINCS, PHOENICS, and constant pressure must be substituted correctly in future work,
TRAC. These have been discussed by Wulff (1981) and Physical as was further documented by Hirano and Watanabe (1988).
Benchmarking Exercise (1987), among others. Such computer They omitted the pressure term from the momentum equation
codes not only predict the optimized criteria, but also fill in the after some mathematical manipulations. Their model worked
gaps between the limited results from experiments conducted properly, but the effect of interface pressure of two-phases was
under substantially different conditions. These computer codes not observed in the solution. Later, this point was readdressed
have also been used widely for consideration in nuclear engineer- in the two-fluid model, which is used for modeling slug flow
ing applications, particularly when thermo hydraulic instability in a horizontal and nearly horizontal pipe, by Issa and Kempf
occurs. During thermo hydraulic instability occurrence, gas or (2002). The conservation equations used in the Issa model are
steam velocity is one of the important parameters. If the gas or the same as those used in the Hirano et al. model; in addition
steam velocity becomes high enough to overcome the hydro- Issa included the interfacial pressure term in his model. Mishima
static forces, then the K–H instability will be generated. The and Ishii (1980) estimated the instability limit when the densities
required gas velocity for the K–H instability, investigated by difference of two phases is high, while neglecting the interfacial
prior researchers, will be presented in this section. surface tension as
Two-phase flow, studied in light of the K–H instability cri-  0.5
ρl − ρ g
terion (as will be explained later) for long waves of small ug − ul > 0.487 ghg (8)
amplitude, is given theoretically by Lamb (1945) and Milne- ρg
2304 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310

Ishii, 1975; Issa and Kempf, 2002). Since the two-phase air water
flow model used in this attempt was considered at atmospheric
conditions, the thermal energy transferred between phases would
be negligible. Thus, the energy equation was not considered
in the calculation. Under isothermal conditions with no phase
change, ρk = f(Pk ); (k = g for air and k = l for water). However, if
the change in pressure or isothermal compressibility is small, the
Fig. 1. Schematic of two-phase flow in horizontal duct.
flow can be considered to be incompressible, i.e., ρk = constant.
The general field equations considered in the mathematical
Ansari and Sani (2007) tested a two-fluid model with surface model are
tension and obtained the uniqueness of results.
Ransom and Hicks (1984) established a hyperbolic two- • gas mass equation:
fluid model. Their assumptions included hydrodynamic pressure
rather than hydrostatic pressure. Ansari (2004) used the Ransom ∂αg ρg ∂αg ρg ug
+ =0 (9)
and Hicks (1984) model for K–H instability limit consideration. ∂t ∂x
The results showed that this model is well posed in the physical • liquid mass equation:
and numerical sense, although the calculation time increased.
∂αl ρl ∂αl ρl ul
In the present article, the Issa model is improved by adding + =0 (10)
gas pressure to the equations, which was neglected in prior ∂t ∂x
research. This model is applied for a transient two-phase air– • gas momentum equation:
water stratified flow at atmospheric condition in a horizontal ∂αg ρg ug ∂αg ρg ug ug ∂P
duct. The conservation equations were solved using the finite + = −αg (11)
∂t ∂x ∂x
volume method. Considerable work can be found in the liter-
ature on two-dimensional single-phase flow modeling showing • liquid momentum equation:
that the governing equations were solved by the SIMPLE algo- ∂αl ρl ul ∂αl ρl ul ul ∂P ∂αl
rithm. This paper is the first to investigate an analogy between + = −αl + ρgHαl (12)
∂t ∂x ∂x ∂x
the SIMPLE algorithm in two-space dimensional single-phase
flow and one-space dimensional two-phase stratified flow by For the interfacial pressure difference (Pk − P), a similar form
applying a special algorithm to solve the conservation equa- to that proposed by Ardron (1980) and Hancox et al. (1980) are
tions. The calculation procedure is explained in details in Section assumed to be
3.3. The two-phase interface was disturbed by a sinus perturba-  
1
tion and the growth or decline of wave was examined through Pl = P + Hαl ρl g (13)
2
the development of a computer code. The two-phase stratified  
flow instability criterion was considered and the results showed 1
Pg = P − Hαg ρg g (14)
that the instability limit is consistent with the classical based 2
K–H prediction. In this paper, the effect of wavelength on the
where
instability limit was also investigated.
αg + αl = 1 (15)
1.2. The basic wave growth mechanism (K–H instability)
The assumptions which have been neglected on the above
equations are: frictions between wall to phases, frictions between
When air flows over a wavy liquid surface at sufficiently high
phases to interface, heat transfer rate between phases to wall,
velocities, the aerodynamic lift due to the pressure variation over
mass transfer between phases, and the vertical components of
a wave as a destabilizing factor becomes sufficient to overcome
the velocity. Since the overall behavior of system is consid-
the hydrostatic forces as a stabilizing factor, referred to as “K–H
ered, surface tension may be neglected, as per Mishima and
instability”. In a closed channel, the Bernoulli effect helps to
Ishii (1980).
increase the destabilizing term. Therefore, the enhanced aero-
dynamic lift will make a particular wave grow with time until
2.2. Initial and boundary conditions
it reaches a limiting amplitude. The instability relation for long
waves of small amplitude was declared as Eq. (1) (see Fig. 1).
Periodic boundary conditions for the entrance and exit of a
duct were studied (i.e., the flow condition at the exit is fed back to
2. Modeling the entrance). Air is traveling in a horizontal duct with the water
lying below. In the initial state, phases are stratified and the
2.1. Field equations velocity of air is uniform, as indicated in Fig. 2. The departure
from uniformity caused by a small sinusoidal perturbation is
The one-space dimensional time representation field equa- defined by
tions were considered for the two-fluid non-equilibrium model
using mass and momentum equations for each phase (see e.g., αl = αinitial + δα (16)
 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310 2305

Fig. 2. Initial perturbation for stratified flow.

Fig. 5. Staggered mesh arrangement.

Fig. 3. Gas velocity variation with respect to time.

αg = αinitial − δα (17) 3. Governing equations solution

ulinitial αinitial
ul = (18) 3.1. Numerical simulation technique
αl
uginitial αinitial For solving Eqs. (9)–(12), the finite volume technique was
ug = (19)
αg applied using a staggered mesh. Fluid properties such as den-
sity, pressure and void fraction have their principal points at
δα = 0.005 sin 2πx/λ (20) the centers and the values of these properties at cell edges are
obtained by interpolation. Velocities are determined at the cell
Various relations for instability criteria have been proposed edges. Fig. 5 shows a staggered mesh arrangement for the one-
by prior researchers. In order to clarify the required criteria, dimensional domain, with control volumes given for pressure
the non-linear conservative equations were employed with the and velocity cells. The symbol P refers to the centers of the
input of three different wavelengths, 100, 50 and 25 cm. The control volumes (whether continuity or momentum), and E and
periodic boundary conditions were equal for all three cases and W refer to the neighboring nodes. Symbols e and w denote cell
the numerical damping effect was eliminated by selecting a con- face values for the control volume surrounding node P. Finally,
stant number for t/x. The periodic boundary condition is a set of algebraic equations was solved using three diagonal
valid when the phases’ velocities remain almost constant or the matrix algorithm (TDMA). The SIMPLE algorithm was applied
system remains without acceleration (Galilian transform). This to solve the field equations. An analogy between the SIMPLE
verification is shown in Figs. 3 and 4. algorithm in two-space dimensional single-phase flow and one-
space-dimensional two-phase stratified flow was incorporated
into the analysis by applying a special algorithm to solve the
conservation equations. To obtain the pressure correction equa-
tion in the two-phase flow in the SIMPLE algorithm, two mass
conservation equations were summed up, and then the resultant
equation was integrated.

3.2. Differentiation of equations

We used Euller implicit differentiation for time dependent

terms and the up-wind scheme for space differentiation terms.
Mass equations on staggered control volume and momentum
equations on scalar control volume were integrated. The finite
Fig. 4. Liquid velocity variation with respect to time. volume formulation of the continuity equation for time step new
2306 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310

(n) for either the gas or the liquid phase is given by For velocities of two phases the following equations can be
declared as
x n n
(ρ α − ρPo αoP ) + (ρe ue − ρw uw )n αnP + (max(ρw uw , 0)n
t P P ug = u∗g + ug (26)
− min(ρe ue , 0)n )αnP + (min(ρe ue , 0)n )αne
ul = u∗l + ul (27)
− (max(ρw uw , 0) n
)αnw =0 (21)
where ug and ul are the velocities correctors of the two phases
Similarly, the discretized momentum equation for either the gas and u∗g and u∗l are the velocities that can be obtained when P* is
or the liquid is given by substituted in the gas and liquid momentum equations.
The main components of the calculation procedure follow as
x n n n
(ρ α u − ρPo αoP uop ) + (αe ρe ue − αw ρw uw )n unP
t P P p 1- The pressure value P* is estimated.
+ (max(αw ρw uw , 0)n − min (αe ρe ue , 0)n )unP 2- By substituting the P* value in the liquid momentum Eq.
(22), the u∗l value is obtained.
= (min (αe ρe ue , 0)n )une − (max (αw ρw uw , 0)n )unw
3- By substituting P* value in the gas momentum Eq. (22), the
+ αnP (Pe − Pw )n + ρgHαnlP (αne − αnw ) (22) u∗g value is obtained.
4- The pressure correction Eq. (23) is solved in order for P to
where, again, the upwind difference and Euller implicit scheme be obtained.
are employed. 5- The gas mass Eq. (21) is solved in order for the void fraction,
A pressure equation is derived by combining the two conti- αg , to be calculated (αl is calculated from αg + αl = 1).
nuity Eq. (21) together with the momentum Eq. (22) to obtain: 6- P is obtained by adding P to P*.
 n n n d + αn ρ n d  7- By using the velocity corrector values, ug and ul are calcu-
αle ρle dle + αnlw ρlwn d
lw αnge ρge ge gw gw gw
+ lated.
ρlref ρgref 8- The correct pressure value P is substituted for the new pre-
 n n n d  dicted pressure value P* and the calculation procedure is
 αlw ρlw dlw αngw ρgw gw 
PP = + PW repeated starting from step 2 for the results to converge.
ρlref ρgref
 n n n d 
αle ρle dle αnge ρge ge 4. Results
+ + PE
ρlref ρgref
  Wave growth was studied for different air velocities under
x 1 following conditions: liquid density (water) = 1000 kg/m3 , gas
− (αn ρn − α0lP ρlP 0
)
t ρlref lP lP density (air) = 1 kg/m3 , wavelength = 1 m, Duct length = 1 m,

1 Duct height = 0.2 m, initial wave amplitude = 5 mm, initial
+ (αngp ρgp
n
− α0gp ρgp
0
) velocity of liquid phase = 1 m/s. Wave growth is defined as
ρgref
(␣maximum − ␣minumum ). Fig. 6 presents that the wave will grow
1 for the velocities higher than 29 m/s. This value is consistent with
+ (αnle ρle ule − α0lw ρlw
n n∗ 0 n∗
ulw )
ρlref the value obtained by K–H Eq. (1) that is evaluated at 31 m/s. As
 can be seen from this figure, the wave growth decreased early
1
+ αnge ρge
n n∗
uge − α0gw ρgw
0 n∗
ugw (23) in time, but increases as time passed with respect to required
ρgref
gas velocity. The reason for the decline of the wave growth at
where
αn
d= (24)
a
If the gas and liquid momentum equations are written in gen-
eral form of finite volume formulation, “a” will be the velocity
coefficient at the new time step.

3.3. Arrangement of calculation procedure

P is the corrected value of pressure and is calculated as

P = P∗ + P (25)

where P* is the predictor of the pressure value and P is the

pressure corrector. Fig. 6. K–H instability limit for λ = 1 m.
 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310 2307

Fig. 7. Effect of higher gas velocity on wave growth. Fig. 9. Distribution of calculated pressure value over a single wave (present
work).

Fig. 8. Distribution of measured pressure value over a single wave (by Kordy-
ban). Fig. 10. Wave growth for λ = 1 m, ug = 29 m/s and ul = 1 m/s.

an early time is that the built-up gas pressure is low and insuffi- the instability limit (20 m/s); as the gas velocity is lower, the
cient to overcome the weight force of the wave from the initial waves are being damped.
condition, in order to establish wave growth. As the gas veloc- For the wavelength effect on the instability criterion, two sets
ity increases, the pressure build-up increases and the required of results for wavelengths of 0.5 and 0.25 m are also presented
time for wave growth decreases. At much higher gas veloci- in Figs. 12–17. The periodic boundary and initial conditions
ties, the wave growth decline at early time is negligible, and the (and all other conditions) remain the same, except that wave-
wave growth increases from the beginning of the wave initia-
tion, see Fig. 7. For wave growth result verification, the related
literature was investigated. The only comparable data is the pres-
sure measurement over a wave in a closed duct produced by
Kordyban (1977). See Figs. 8 and 9 for a qualitative comparison
of pressure distributions over a wave. Clearly, the results agree
qualitatively. The results are not compared quantitatively, since
Kordyban (1977) used a wave generator in addition to flowing
air. For this reason, the required pressure for wave growth was
less than what is found in the current analysis. Hence, the two
figures are shown separately (Fig. 8 has been redrawn because
of poor quality). Fig. 10 presents the interface of two-phase flow
for velocities higher than instability criterion (29 m/s) at three
different times, 1, 1.5 and 2 s. The assumed wavelength is 1 m.
As expected, the waves are growing over time because of gas
velocities that exceed the instability limit. Fig. 11 shows the
interface for the same condition, but for velocities lower than Fig. 11. Decline of wave for λ = 1 m, ug = 20 m/s and ul = 1 m/s.
2308 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310

Fig. 15. K–H instability limit for λ = 0.25 m and ul = 1 m/s.

Fig. 12. K–H instability limit for λ = 0.5 m and ul = 1 m/s.

Fig. 16. Wave growth for λ = 0.25 m, ug = 26 m/s and ul = 1 m/s.

Fig. 13. Wave growth for λ = 0.5 m, ug = 27 m/s and ul = 1 m/s.

higher than the stability limit, wave growth will grow over time.
Fig. 14 shows same condition, but for gas velocity lower than
lengths are 1, 0.5 and 0.25 m. Fig. 12 presents wave growth for instability initiation value (18 m/s). The wave growth declined
wavelengths of 0.5 m, where the gas velocity for the instability as time increased. Fig. 15 shows the result for a wavelength
limit is 26 m. Further, gas velocities higher than the instabil- of 0.25 m. The required gas velocity for instability initiation
ity initiation required velocity are examined. See Fig. 13 for gas is around 25 m/s. Fig. 16 shows same conditions, but with gas
velocity of 27 m/s at 0.4, 0.6 and 0.8 s. As these gas velocities are velocity exceeding the instability limit, at 0.4, 0.6 and 0.8 s.

Fig. 14. Decline of wave for λ = 0.5 m, ug = 18 m/s and ul = 1 m/s. Fig. 17. Decline of wave for λ = 0.25 m, ug = 16 m/s and ul = 1 m/s.
 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310 2309

Fig. 20. Comparison of present result with the well-known two-phase flow
codes.

the results from well-known two-phase flow computer codes,

Fig. 18. Effect of wave number on onset of instability initiation. Physical Benchmarking Exercise (1987). See Fig. 20, which
shows that the wave growth result is consistent with the analyt-
Fig. 16 shows higher values than the instability limit (26 m/s) ical results in comparison with the other well-known two-phase
and Fig. 17 demonstrates a velocity lower than the instability flow computer codes.
limit (16 m/s). The wave growth was increased with a gas veloc-
ity exceeding the instability value and was decreased for the 5. Conclusion
gas velocity below the instability limit as time increased. The
analysis shows that the required gas velocity decreases as the The wave growth and hydro-dynamical instability criterion in
wavelength decreases, the gas velocity of around 28 m/s for 1m two-phase two-fluid flow was investigated using numerical sim-
wavelength, the gas velocity of 26 m/s for 0.5 m wavelength, ulation. Transient from stratified regime to wavy gas–liquid flow
and the gas velocity of around 25 m/s for 0.25 m wavelength. was modeled in a horizontal duct. The transient non-linear one-
These results also show that the instability limit is consistent space dimensional conservation equations were solved by the
with Eq. (1) only for long wavelengths, and agreement lessens application of a special algorithm and the finite volume method.
as wavelength decreases. This means that short wavelengths are A wide range of gas velocities at the transient condition were
not consistent with Eq. (1). Thus, the wavelength effect must considered for wave growth and the hydro dynamical instability
be included in Eq. (1), or in the K–H instability criterion. See limit. The results obtained by numerical analysis showed that:
Fig. 18 for the wavelength effect on the instability initiation gas
velocity. 1- The K–H instability criterion is correct only for long waves
For verification of the results, the mesh independency was with small amplitude.
examined for different mesh numbers (10, 20, 40, 80, 100, 2- The required gas velocity for the initiation of instability with
150 and 160). Fig. 19 shows that as mesh numbers increase short wavelengths is below than that defined by the K–H
to more than 100, the difference between the results becomes instability relation. In other words, the K–H instability limit
negligible. In order to prevent high CPU time consumption, decreases as the wavelength decreases. Thus, the criterion is
all the results presented have been conducted with 100 mesh invalid, and should be modified.
numbers (x = 1 cm). The result of the investigation of wave 3- In solving the conservation equations, an analogy between
growth from the present computer code was also compared with the SIMPLE algorithm in two-space dimensional single-
phase flow and one-space dimensional two-phase stratified
flow was applied. The results’ consistency with the analyti-
cal solution is higher in comparison with other well-known
computer codes.

References

Akimoto, M., et al., 1986. Development of two-phase flow analyzer: MINCS.

In: Proceedings of the 2nd International Topical Meeting on Nuclear Power
Plant Thermal Hydraulics and Operation, pp. 72–79.
Ansari, M.R., 1998. Numerical analysis for slugging of steam-water stratified
two-phase flow in horizontal duct. Fluid Dynam. Res. 22, 329–344.
Ansari, M.R., 2000. Wavelength effect on Kelvin–Helmholtz instability criteria
in two-phase stratified flow. Iran. J. Sci. Technol. 24 (3), 259–267.
Ansari, M.R., 2004. Effect of pressure on two-phase stratified flow modeling. J.
Nucl. Sci. Technol. 41 (7), 709–714.
Ansari, M.R., Sani, E., 2007. Surface tension effect on stability of two-phase
Fig. 19. Effect of total mesh number. stratified flow. J. Fluid Dynam. Res. 39, 279–291.
2310 M.R. Ansari, V. Shokri / Nuclear Engineering and Design 237 (2007) 2302–2310

Ansari, M.R., et al., 1988. Numerical analysis on slugging of air–water strati- Lamb, H., 1945. Hydrodynamics. Dover, New York.
fied flow in horizontal duct using MINCS code. In: Proceedings of the 3rd Lin, P.Y., Hanratty, T.J., 1986. Prediction of the ignition of slugs with linear
International Topical Meeting on Nuclear Power Plant Thermal Hydraulics stability theory. Int. J. Multiphase Flow 12 (1).
and Operation, Seoul, Korea, A1-115-122. Lin, P.Y., Hanratty, T.Y., 1987. Effect of pipe diameter on flow patterns for
Ardron, K.H., 1980. One-dimensional two-fluid equations for horizontal strati- air–water in horizontal pipes. Int. J. Multiphase Flow 13 (4).
fied two-phase flow. Int. J. Multiphase Flow 6, 295–304. Milne-Thomson, L.M., 1963. Theoretical Hydrodynamics. MacMillan, New
Banerjee, S., 1980. Separated flow models-II, higher dispersion effects in the York.
averaged formulation. Int. J. Multiphase Flow 6, 241–248. Mishima, K., Ishii, M., 1980. Theoretical prediction of onset of horizontal slug
Banerjee, S., Chan, M.C., 1980. Separated flow models-I. Int. J. Multiphase flow. J. Fluids Eng., 102.
Flow 6, 1–24. Physical Benchmarking Exercise, 1987. DOE/EPRI, 2nd International Work-
Bergles, A.E., Collier, J.G., Delhaye, J.M., Hewitt, G.F., Mayinger, F., 1981. shop of Two-Phase Flow Fundamentals. Rensselar Polytechnic Institute,
Two-phase flow and heat transfer in the power and process industries. Hemi- Troy, NY, USA.
sphere Publ. Corp.. Ramshaw, J.D., Trapp, J.A., 1978. Characteristics, stability, and short-
Hancox, W.H., Frech, R.L., Liu, W.S., Nieman, R.E., 1980. One-dimensional wavelength phenomena in two-phase flow equation systems. Nucl. Sci. Eng.
models for transient gas–liquid flows in ducts. Int. J. Multiphase Flow, 6. 66, 93.
Hihara, E., Saito, T., 1983. Slug flow transition in a horizontal tube. In: ASME- Ransom, V.H., Hicks, D.L., 1984. Hyperbolic two-pressure models for two-
JSME Thermal Engineering Joint Conference. phase flow. J. Comput. Phys. 53, 124–151.
Hirano, M., Watanabe, T., 1988. Numerical study on shock phenomena and void Ransom, V.H., Scofield, M.P., 1976. Two-Pressure Hydrodynamic Model for
wave propagation in horizontal stratified flow. In: Proceedings of the 3rd Two Phase Separated Flow. SRD-50-76. Idaho National Engineering labo-
International Topical Meeting on Nuclear Power Plant Thermal Hydraulics ratory.
and Operations, Seoul, Korea. Ransom, V.H., Trapp, J.A., 1983. Applied mathematical methods in nuclear
Hirano, M., et al., 1986. Evaluation of interfacial shear model for bubbly flow thermal hydraulic. Trans. Am. Nucl. Soc..
regime with MINCS code. In: Proceecdings of 2nd International Topical Safety Code Development Group, 1981. TRAC-PD2, An Advanced Best Esti-
Meeting on Nuclear Power Plant Thermal Hydraulics and Operation, pp. mate Computer Program for Pressurized Water reactor Loss of Coolant
80–87. Accident Analysis. NUREG/CR-2054, LA-8709-MS.
Ishii, M., 1975. Thermo-Fluid Dynamic Theory of Two-Phase Flow. Eyrolles, Taitel, Y., Dukler, A.E., 1976. A Model for predicting flow regime transitions in
Paris. horizontal and near horizontal gas–liquid flow. AIChE J. 22, 47–55.
Issa, R.I., Kempf, M.H.W., 2002. Simulation of slug flow in horizontal and Wallis, G.B., Dobson, J.E., 1973. The onset of slugging in horizontal stratified
nearly horizontal pipes with the two-fluid model. Int. J. Multiphase Flow 29, air–water flow. Int. J. Multiphase Flow 1, 173.
69–95. Wendroff, B., 1980. Two-fluid models: a critical survey. In: Proceedings of the
Kordyban, E., 1977. Some characteristics of high waves in closed channels EPRI Workshop, Basic Two-phase Flow Modeling in Reactor Safety and
approaching Kelvin–Helmholtz instability. J. Fluid Eng., 339. performance. WS78-143, vol. 2.
Kordyban, E.S., Ranov, T., 1979. Mechanism of slug formation in horizontal Wulff, W., 1981. Major Systems, Codes Capabilities and Limitations. EPRI
two-phase flow. ASME J. Basic Eng., 92. WS-81-212.