Int. J. Multiphase Flow Vol. 19, No. 4, pp. 639-649, 1993 0301-9322/93 $6.00 + 0.

00
Printed in Great Britain. All rights reserved Copyright 1993 Pergamon Press Ltd
KE L VI N- HE L MHOL T Z S T ABI L I T Y CRI T E RI A F OR
S T RAT I F I E D F L OW: VI S COUS VE RS US
NON- VI S COUS ( I NVI S CI D)
AP P ROACHE S
D. BARNEA and Y. TAITEL
Depar t ment of Fluid Mechanics and Heat Transfer, Faculty of Engineering, Tel-Aviv University,
Ramat -Avi v 69978, Israel
(Received 1 June 1992; in revised f orm 15 March 1993)
Abst r act - - The neut ral stability lines obt ai ned from t he viscous Kel vi n-Hel mhol t z analysis and t he inviscid
analysis are quite different for t he case of low liquid viscosities, whereas they are quite similar for hi gh
viscosity, cont rary to what one would expect. Thi s puzzling result is considered in this work. It is shown
t hat the stability behavi or regarding t he amplification rate is actually al most the same for the two analyses
for a wide range of liquid viscosities and for various pipe inclinations. The results obt ai ned in t he present
work also support Barnea' s i nt erpret at i on of the viscous and inviscid analyses as a means for predicting
various t ransi t i ons from stratified flow.
Key Words: stratified flow, stability, Kel vi n-Hel mhol t z, flow pat t ern
I NT R ODUC T I ON
Kelvin-Helmholtz (KH) linear stability has been used frequently in the past for determining
whether a smoot h stratified flow is stable or unstable. Two types of KH analyses have been used:
(1) The viscous Kelvin-Helmholtz (VKH) analysis, which uses the full two-fluid model and takes
into account the shear stresses (Wallis 1969; Lin & Hanrat t y 1986; Wu et al. 1987; Andritsos et al.
1989; Barnea 1991; Crowley et al. 1992); and (2) the inviscid Kelvin-Helmholtz (IKH) theory, in
which the shear stresses are neglected (Taitel & Dukler 1976; Kordyban 1977; Kordyban & Ranov
1970; Mishima & Ishii 1980), One would expect that the inviscid theory would be a good
approximation for liquid of low viscosity, whereas for high viscosities one will have to use the full
two-fluid model in order to get correct results. Surprisingly the results are j ust the opposite. For
liquid of high viscosity the results of the I KH theory are applicable, while there is a large
discrepancy in the results for the stability criterion between the I KH and VKH theories for low
liquid viscosity.
This fact is demonstrated in figure 1. In this figure the neutral stability lines are plotted on a
ULS VS UGS map to show the regions where stratified flow is stable or unstable. Both theories are
used, the VKH theory and the I KH theory. As can be seen, for high viscosities the stability criteria
using the aforementioned theories are almost identical, whereas for low viscosity they are quite
different.
This dilemma motivated this work.
ANALYSI S
The stability analysis of stratified flow is performed on the "two-fluid model " equations. A brief
outline of this analysis follows.
The continuity equations for the liquid and the gas are:
~t (PLAL) + ~ (pLAL UO = 0 Ill
and
cgt ( P c AG) + ~xx ( P c A c Uo ) = 0.
[21
639
640 D. BARNEA and Y. TAITEL
The moment um equations for each phase are:
O OPiL Oh L
~t (PLAL UL) 4- ~ (PLAL U2) = --'~LSL 4- TiSi -- AL - ~ X -- pLALg COS fl -~X -- pLALg sin fl
and
[3]
~X 0PiG OhL
~( pcA ~ ~) + (p~A~ V ~) = - ~S~ - T~S~- AC- ~x - poA ~g cos/~ ~ x -- -- RGAGg sin ft. [4]
In the above, A is the cross-sectional area, h is the liquid level or gas gap, P is the pressure,
U is the axial average velocity, z is the shear stress, S is the perimeter over which r acts,
p is the phase density and fl is the angle of inclination from the horizontal (positive for
upward flow). The subscripts L and G denote liquid and gas, respectively; the subscript i denotes
interface.
Assuming incompressible flow and combining the two moment um equations by eliminating the
pressure terms using the approximate relation
~2h L
PiG -- PiE = O" OX 2 , [5]
where a is the surface tension, yields the following 3 equations:
~hL A L 0 U L UL 0hL
~ ~ ~:~ Tx + Ox = 0 ,
[6]
and
0hL AGdUG 0hL 0 [7]
0t a [ 0--~- + u~ 0x =
= F ,
where
0 U L 0 U G ~ 0UG ah L 633hL
P L - ~ - - P G - - ~ i - + pL UL Ux - - PG UG--~X-X 4- (PL - - P o ) g c o s f l ~--X - - a OX 3
[s]
F =Z L S L ' ~ GS G ( ! 1 )
- - - - AL 4- ~ 4- TiSi Z ~ - A-G -- (PL -- PG)g sin fl [9]
and A~. is d A L / d h L.
A linearization procedure, which follows the general approach presented by Barnea & Taitel
(1989), yields
- - f ' ; -h'~4 d4~L [ p L U 2 4 O G U ~ --j'rA ] O 2 ~ L + 2 [ O L U L + o G U G l d2~L
"'LO'U~ + l "~'L RG (PL--PG)gCOSfl'~L OX2 L RL RG J O t t ~ x
f f l O L pG-]O2~L [ A O F UGdF ULOF]O]iL [ 1 OF 1 0 F - ] ~ t L
where '~t is the perturbed liquid level and R is the phase holdup. Note that, all the terms in the
square brackets in [10] are evaluated at the steady state (unlike these terms in [1]-[9], where these
values are the local transient values).
Substituting for the perturbed liquid level,
~ L = E e i ( = ` - k x ) , [ 1 1 ]
into [10] yields the following dispersion equation for the angular frequency, co:
oJ 2 -- 2(ak - b i ) t o 4- c k 2 - - dk 4 - e k i = 0, [12]
KH STABILITY CRITERIA FOR STRATIFIED FLOW 641
whe r e
a nd
a = - + [13a]
P
! c3F d F
l _ ( 0 L V t 0 o V A )
c = -P \ - ~ - L + - - R G - - ( P L - - PG)g COS fl~LL [13C1
t r A
d = - - - [13d]
I
pAL
e = -- - [13e]
P ULS. UGS
PL PG [13q
The s ol ut i on f or to is
to = (ak - bi) +_ x/ ( a 2 - c) k 2 - b 2 + dk 4 + (ek - 2abk )i . [14]
The s t e a dy- s t a t e s ol ut i on is uns t a bl e whe ne ve r t he i ma gi na r y p a r t o f co in [14], na me l y tol, is
negat i ve, l eadi ng t o e xpone nt i a l g r o wt h o f t he p e r t u r b e d var i abl e, /~L- The ampl i f i cat i on f a c t or
is - t o l -
F o r t he cas e o f i nvi sci d f l ow, a si mpl e expr es s i on f or to is obt a i ne d:
poU /(P L
PL UL + __ PLPo (UG -- UL) 2
to HL HG PG)g HLHG ak2 [15]
C- - ~= + -- + - - ,
whe r e HE = AL/ A~ a nd HG = AG/A'G; C is t he wa ve vel oci t y a n d k is t he wa v e n u mb e r . As l ong
as t he t er m in t he s q u a r e r o o t is pos i t i ve, t he ampl i f i cat i on f a c t o r in t hi s cas e is 0. Wh e n t he s qua r e
r o o t is negat i ve, t wo c o n j u g a t e s ol ut i ons f or t he i ma gi na r y pa r t s exist. The s e c ond s ol ut i on, na me l y
t he one wi t h t he negat i ve sign, is t he o n e t ha t c o n t r i b u t e s t o t he i nst abi l i t y.
F o r t he vi s cous case, t he s ol ut i on f or to c a n be expr es s ed c onve ni e nt l y in a p o l a r f or m:
( ' ) 1
to, = (ak - bi) + ~ e x p i ~ a r c t a n ~ [16a]
a n d
{ 1 [ ] }
to2 = (ak - bi) + ~ exp i F a r c t a n + 2n [16b]
whe r e 0t = (a 2 - c ) k 2 + dk 4- b 2 a nd ~ = ek - 2abk . The negat i ve val ue o f t he i ma gi na r y pa r t o f
[16a, b] is t he ampl i f i cat i on f act or .
The c o n d i t i o n f or ma r gi na l s t abi l i t y c a n be o b t a i n e d f r o m [12] f or t he speci al cas e whe r e COl, t he
i ma gi na r y par t , e qua l s 0. Thi s l eads t o t he f ol l owi ng s t abi l i t y cr i t er i on f or t he vi s cous case:
e _ a - ( a 2 - c ) - d k 2 < 0 . [17]
Subs t i t ut i ng t he val ue o f a 2 - c f r om [13] i nt o [17] yi el ds
( Cv_Ci v) 2_~ PLPG ( UG_ UL ) 2 pL--p._____._~Ggcosfl A o A k 2 < 0 " [18]
p2RL RG p A'L pAL
642 D. BARNEA and Y. TAITEL
The last t hree t erms on the LHS of [18] can be observed as the wel l -known KH instability of
the i nt erface of one-di mensi onal flow with no viscous effects on the stability. The first t erm is the
addi t i onal effect of t he shear stresses, which t ends to ampl i fy any di st urbance in the film thickness.
Not e t hat the f our t h term, which is the cont r i but i on of t he surface tension, is the onl y term t hat
depends on the wavelength. For l ong waves this t erm appr oach zero and it does not affect the
neut ral stability cri t eri on t hat shoul d appl y to all wavelengths.
The critical wave velocity on the i ncept i on of instability, Cv, obt ai ned f r om [12] f or to, = 0, equals
( e / 2 b ) :
Cv = e = Vts. Vos [19]
The di spersi on equat i on f or the I KH analysis is obt ai ned f r om [12] with e = 0 and b = 0. The
critical wave velocity in this case, Cw, is equal t o [13a], namely:
PL UL RG -Jr" RG UG RL
Cjv = a - [20]
pL RG + pGRL
Thus, t he first t erm in t he KH stability cri t eri on [17], t hat results f r om consi deri ng the shear stresses,
is rel at ed t o t he difference bet ween the wave velocity obt ai ned f r om the VKH t heor y and the wave
velocity f or the inviscid case, on t he i ncept i on of instability.
In this wor k t he shear stress ZL, ZO and zi are eval uat ed as follows:
RL2 U2L [21]
TL =f L -- ,
ZG =f G - - P U~ [22]
2 '
and
where
and
f p o ( UG - UL ) I Uo - ULI
i ~ "
[23]
f L = CL ( D L UL~ -n [24a]
\ V L /
f G = C G ( D G U G ~ " .
\ V G /
D E and DG are t he hydraul i c di amet ers, eval uat ed in the fol l owi ng manner:
[24b1
4AG
Do = SG "[- S- - - - - - - ~ " [25b]
The coefficients Co and CL equal 0.046 f or t ur bul ent flow and 16 for l ami nar flow, n and m t ake
t he values of 0.2 f or t ur bul ent flow and 1.0 for l ami nar flow. The interfacial friction f act or was
assumed t o have a const ant val ue o f f i = 0.014, as suggested by Cohen & Hanr at t y (1968) f or
stratified wavy flow, or f i = f o when f o > 0.014.
and
4AL
DE -- [25a]
SL
KH STABILITY CRITERIA FOR STRATIFIED FLOW 643
RESULTS AND DI SCUSSI ON
Fi gure 1 compares the results of the neutral stability criterion for stratified flow obt ai ned by the
VKH analysis [18] and the I KH approach ([18] wi t hout the first term). The calculations were made
for the case of ai r-l i qui d (of wat er density) in a 5 cm hori zont al pipe. It can be seen t hat for low
viscosities the I KH analysis overpredicts consi derabl y the viscous results. However, as the liquid
viscosity increases the cont ri but i on of the term (Cv - Civ)2 in [18] diminishes and for high viscosity
bot h approaches yield al most the same results.
The fact t hat the results obt ai ned by the VKH analysis, which takes into account the shear
stresses, are different from those using the I KH approach at low viscosities, while the two
approaches yield al most the same results at high viscosities is indeed puzzling.
In order to see the reason for this anomal y, we consider the behavi or of the amplification fact or
at various flow conditions, as obt ai ned by the two analyses.
The rate of amplification is exami ned al ong line a- b in figure 1. Poi nt (a) is the intersection poi nt
with the VKH neut ral stability curve, while poi nt (b) is the intersection with the I KH neutral
stability curve.
The dispersion equat i on [12] yields two solutions for o9i. Positive or zero solutions indicate stable
flow. For the VKH analysis, toll is always positive, while oh2 changes sign from positive (stable flow)
below poi nt (a) to negative (unstable flow) for ULS above poi nt (a). For the I KH analysis, t~l~ and
(/)12 are zero for ULS below poi nt (b). Above poi nt (b) ~Ol2 is negative (unstable) and ~Oll is the positive
conj ugat e sol ut i on for rOlE. The value of -0912 is the rate of amplification of the di st urbance and
is illustrated in figure 2 as a funct i on of the wavelength, for a const ant gas flow rate (Ucs = 5 m/s)
and at various liquid flow rates (along line a- b in figure l). The results obt ai ned for the two analyses
( VKH and I KH) are al most the same and the results of the solutions are indistinguishable in this
figure (al t hough the neut ral stability criterion is quite different). A scaled-up picture of the two
solutions is shown in figure 3 t o show the behavi or at low amplification rates. In this figure the
details ar ound the condi t i ons of neutral stability can be seen. For liquid flow rates below poi nt
(a) (ULs = 0.1 m/s), --0912 is negative for all wavelengths accordi ng to the VKH analysis, namel y
the flow is stable. For ULS between points (a) and (b), 0.15 < ULS < 0.6 m/s, --~Ol2 is positive with
very low absol ut e values and onl y at ULS = 0.6 m/s, where the flow becomes unst abl e also according
to the inviscid analysis, does the rate of amplification become meaningful. Stated differently, under
the flow condi t i ons where the I KH analysis indicates a zero rate of amplification, the VKH analysis
yields a t ransi t i on from a negative to a positive value of -oJi2 with absol ut e values t hat are close
to zero. Onl y at condi t i ons where the I KH analysis indicates unst abl e flow is the rate of
amplification obt ai ned by the VKH analysis substantial.
I 0 . 0
1.0
"~ 0.1
, , . I
::)
0.01
OO01
O O I
\ \ . . . . . I K H
\
" , , - - V K H
I I i ~ l
O.I I.O t O I 0 0
U G S ( m / s )
Figure 1. Effect of liquid viscosity on the VKH and IKH neutral stability criteria. Air-liquid, atmospheric
pressure, horizontal pipe, D = 5 cm.
644
-~ 5 0 O
3
I
D. B A R N E A a nd Y. T A I T E L
I 0 0 0 i I l
UQS= 5 m /s
ULS 2 . 2 ~ 5 I . sm/ S
0
I I I
0 .05 .I .15
X / D
Figure 2. Amplification factor for air-water at UGS = 5 m/s.
.2
A conveni ent met hod f or compar i ng t he ampl i fi cat i on f act or of the t wo sol ut i ons with respect
t o t he flow rates is t o l ook at t he ampl i fi cat i on f act or at a cert ai n wavelength. Thi s wavel engt h
may be chosen arbi t rari l y. In this wor k we chose to compar e the sol ut i ons at the wavel engt h which
yields t he maxi mum rat e of amplification. Fi gure 4 shows t he maxi mal rat e of ampl i fi cat i on at each
liquid flow rat e f or a const ant gas flow rate. The results are given f or a wide range of liquid
viscosities. It can be cl earl y seen, again, t hat f r om this poi nt of view t he stability behavi or of the
system is al most the same accor di ng t o t he t wo KH analyses. For each liquid viscosity under
consi der at i on bot h analyses yield al most the same results and the ampl i fi cat i on f act or f or t he I KH
analysis is, again, i ndi st i ngui shabl e f r om t hat f or the VKH analysis. A careful exami nat i on reveals,
however, t hat t here are differences which are onl y appar ent at a very low ampl i fi cat i on fact or.
For the inviscid case the ampl i fi cat i on f act or is exact l y zero up t o the poi nt where the liquid velo-
city is sufficiently large and t he ampl i fi cat i on curve mai nt ai ns a clearly visible positive slope. Thi s
is t he neut ral stability poi nt f or the I KH analysis. The behavi or of the ampl i fi cat i on f act or f or t he
VKH analysis, al t hough it l ooks t he same in figure 4, is however different. It is not zero for a low
liquid flow rat e but r at her it is negative f or a very low liquid flow rat e and becomes positive for
an increasing liquid flow rate. The poi nt where the ampl i fi cat i on f act or changes sign is the neut ral
stability poi nt f or the VKH analysis. Thi s fact cannot be observed in figure 4, since in t he region
of changi ng sign the ampl i fi cat i on f act or is very small and very close to zero. Ther ef or e the results
f or t he VKH analysis l ook the same as t hose f or the I KH analysis. Thi s poi nt of neut ral stability
I 0
~ 5
3
I
0
0
i I I i
- ~ . . . . . I K H -
-- , \ - - V K H -
. . . . . . . . _ -
- I I " ~ 0 . 0 1 I I
.I .2 .5 .4 .5
X/ D
Figure 3. Scale-up amplification factor for air-water at UGs = 5 m/s.
KH STABILITY CRITERIA FOR STRATIFIED FLOW 645
I 0 0 0 - . . . . . . . ; . . . . . . . . i . . . . . . . . I . , i . . . . . . . I . . . . . . . .
" " / / / / /
neutrol s l o b i t i t y
1 1 121
o ! / 5 o - ~
- o o ~
0 o o - -Y 7 I Y
0
. . . . . . . . I . . . . . . . . I . . . . . . . . I . . . . . . . . I . . . . . . . .
.001 ,01 . I I I0 O0
ULS (m/s)
Fi gur e 4. Effect of l i qui d vi scosi t y on t he ampl i f i cat i on f act or , Uos = 5 m/ s . Ai r - l i qui d, a t mos phe r i c
pr essur e, hor i zont al pi pe, D = 5 cm.
due to the VKH analysis is designated by arrows in figure 4. For example, for a viscosity of 100 cP
the neutral VKH point is at ULS --~ 0.04 m/s, whereas the IKH neutral stability point is at 0.08 m/s.
Thus, although the general behavior of the IKH and VKH analyses is similar with respect to the
value of the amplification factor, they behave differently near the zero amplification factor. As a
result, the neutral stability point for the IKH and VKH analyses is not the same.
An interesting point to observe is the effect of viscosity on the location of the neutral stability
points. One would expect that as the viscosity decreases the VKH analysis would approach the IKH
results. Indeed, for low viscosity, the amplification factor for the low liquid flow rate approaches
the zero value, as in the IKH analysis. But at the same time, the exact point of changing sign, i.e.
the neutral stability point, moves to the "left" compared to the IKH neutral stability point. On
the other hand, for high viscosities the two neutral points are at almost the same location. Thus
(see figure 4), the neutral stability point for 1000 cP for both cases is at ULS ~-- 0.008 m/s. For 100 cP
the neutral stability points are at ULS ~ 0.04 and 0.08 m/s. For viscosity < 1 cP, the VKH neutral
stability point stays at ULS -- 0.15 m/s, while the IKH neutral stability point moves to a higher liquid
flow rate as the viscosity decreases. Thus, it is clearly seen that the neutral stability points for the
VKH and IKH analyses are farther apart for low viscosity than for high viscosity, giving the
impression that the IKH analysis is similar to the VKH analysis for high viscosity but not for low
viscosity.
Phys i cal I nt erpret at i on
The physical interpretation of the results of the aforementioned analysis is not at all straightfor-
ward. The key question is how to interpret the behavior in the unstable regions due to both the
IKH and VKH analyses. Do these instabilities result in a transition to slug flow, annular flow or
just cause the interface to be wavy?!
As is well-known, the IKH instability does not predict well the transition boundary from
stratified flow. The stable region predicted by the IKH analysis is larger compared with the
experimental data. As a result, a few attempts to use this type of analysis required the insertion
of some correction factors to account for this discrepancy (Kordyban & Ranov 1970; Taitel &
Dukler, 1976; Kordyban 1977; Mishima & Ishii 1980).
Lin & Hanratty (1986), Wu et al. (1987) and Crowley et al. (1992) used the VKH analysis for
the prediction of the transition boundaries from stratified flow and found good agreement with
the data for the case of a low void fraction. On the other hand, Hanratty (1983) and Andreussi
et al. (1985) indicated that the VKH analysis can be used successfully to explain the transition to
roll waves for the case of a thin liquid level: namely, the unstable infinitesimal wave grows, due
to this instability to a large-amplitude wave of the order of the liquid film (roll waves). Barnea
(1991) adopted this general appraoch and gave a complete quantitative description for the behavior
M F 1914~H
646 D. BARNEA and Y. TAITEL
of the i nt erface in the vari ous unst abl e regions. It was shown t hat the neut ral stability condi t i on
of the VKH analysis is not directly associated with the t ransi t i on to slug or annul ar flow but rat her
to an unst abl e i nt erface with roll waves. Whenever the liquid suppl y in the film is large enough
(hL/D > 0.5) to pr ovi de the liquid needed to bridge the pipe, the unstable region becomes slug flow.
For hL/D < 0.5, ei t her roll waves or annul ar flow may exist. In or der to obt ai n annul ar flow, the
upper par t of the pipe shoul d be wet t ed also under condi t i ons of high void (low liquid level).
It occurs when the suction effect of the pressure generat ed over the wave by the Bernoulli effect
over comes the stabilizing influence of gravity. Thi s effect is in phase with the wave height and grows
unboundedl y until the upper par t is wetted. Barnea (1991) suggested t hat since this descri pt i on is
consi st ent with the I KH analysis, it is suggested t hat the I KH analysis is also valid when the waves
become finite.
Thus, t wo regions of instability are identified. The region bounded between the VKH and I KH
neut ral stability lines and the region "out s i de" the I KH neut ral stability curve. The first region is
associated with l arge-ampl i t ude roll waves f or hL/D < 0.5 and slug flow for hL/D > 0.5. For the
region out si de the I KH neut ral stability line the I KH instability will result in either slug flow for
hL/D > 0.5 or annul ar flow f or hL/D < 0.5.
The present work suppor t s this general expl anat i on. It f ur t her identifies the region between the
VKH and I KH neut ral stability lines with the region of low ampl i fi cat i on fact or. Within this region
the Bernoulli effect is small and the instability is caused by the viscous effect (the t wo first term
in [18]). In this case the ampl i fi cat i on f act or is of a l ower or der which does not lead to an
unbounded growt h. On t he ot her hand, the region of I KH instability is associated with very high
ampl i fi cat i on f act or (by bot h analyses), which results in exponent i al growt h t hat always leads to
t he t ransi t i on f r om stratified flow.
Referri ng now t o figure 5, the zone bounded by the VKH neut ral stability curve is a zone of
stable stratified flow (stratified smoot h or stratified with smal l -ampl i t ude waves). The region
bet ween the VKH ( ) and t he I KH ( - - - ) curves is a region of either roll waves or slug flow.
The curve hL/D = 0.5, is a dividing line bet ween slug flow to the "l ef t " and l arge-ampl i t ude roll
waves to the "r i ght ". At a relatively high liquid rate these l arge-ampl i t ude waves were t ermed
pseudo-sl ug (Lin & Hanr at t y 1987), wavy annul ar (Barnea et al. 1980) or prot o-sl ug (Ni chol son
et al. 1987). At a low liquid flow rate this is a t ransi t i onal region to annul ar flow (which was
identified as wavy by some researchers and as annul ar by others). The region out si de the I KH curve
is ei t her in slug flow, f or the region hD/D > 0.5, or annul ar flow, f or the region where hL/D < 0.5.
Thus, a compl et e behavi or is det ermi ned by t hree demar cat i on lines: the neut ral VKH line, the
neut ral I KH line and the curve hL/D = 0.5. Not e agai n t hat the I KH neut ral stability line is at the
same l ocat i on as where the VKH analysis shows a sharp increase in the ampl i fi cat i on factor. Thus,
al ong line a- b, the flow changes f r om stable stratified flow to l arge-ampl i t ude roll waves (usually
referred t o as wavy annul ar, pr ot o- or pseudo-sl ug in this region) and t hen above hL/D = 0.5 the
flow pat t er n is slug flow. Al ong the line c~l , the flow changes f r om stable stratified flow to annul ar
flow t hr ough a nar r ow region of l arge-ampl i t ude waves.
i n
, v ~ I i /
,
\ ,, -, SL ,e,,,(o~, /
1.0 - "~'-,..._ IKH bl /
-1
1~ O l KH c'~ \ d
I
0.01 I I I 1
,.01 Ol 1.0 I 0 I 0 0
Ue s ( m/ s l
Figure 5. Flow pattern prediction by the VKH and IKH analyses. Air-water, atmospheric pressure,
horizontal pipe, D = 5 cm. SL--slug; ST--stratified; RW--rol l waves; A--annul ar.
K H S T A B I L I T Y C R I T E R I A F O R S T R A T I F I E D F L O W 647
1.0
S L .
- I
VKH S]
O.01 - - -
i
0 . 0 0 1 i
Ol I O I 0 I 0 0
U G s ( m / s )
Figure 6. Fl ow pattern transition boundaries for upwards
inclined flow, ,8 = 0.25 . Ai r-wat er, atmospheric pressure,
D = 5 cm. SL--slug; ST--stratified; RW- - r ol l waves; A - -
annular.
5 0
4 0
s o
2 o
, i i i i n , I i , 0 H I , I , . , i l l ,
- U o s - l O m / s i
IKH
~VKH
j
I0
0
i , , . , . l = l ! I | | | | | l l i 0 i
OI .I I I 0
U L S ( m/ s )
Figure 7. Amplification factor for upwards inclined flow,
Uos = 10 m/s. Air-water, atmospheric pressure, ~ = 0.25 ,
D = 5cm.
A similar behavi or regardi ng t he stability charact eri st i c of t he system is obt ai ned f or t he case
of inclined stratified flow. For upwar ds inclined flow t he stable area predi ct ed by t he VKH analysis
is bounded by a bell-shaped curve on t he ULS VS UGs pl ot (figure 6), while t he I KH neut ral stability
cur ve shows a t ot al l y different behavi or. The maxi mum ampl i fi cat i on rat es al ong a const ant gas
flow rat e (line a- b) and al ong a const ant liquid flow rat e (line c ~l - e ) are i l l ust rat ed in figures 7
and 8, respectively. It is shown agai n t hat , al t hough the neut ral stability curves obt ai ned by t he
t wo anal yses are qui t e different, t he ampl i fi cat i on curves are very similar. Fi gure 7 shows how t he
maxi mal ampl i fi cat i on f act or changes al ong t he line a- b. The results are similar t o t he hor i zont al
case (figure 4). Fi gure 8 shows t he maxi mal ampl i fi cat i on rat e f or increasing gas flow rate. As can
be seen, t he ampl i fi cat i on f act or is positive and small f or a low gas flow rat e, it becomes negative,
i.e. a stable interface, f or t he region f r om poi nt c t o d and t hereaft er a shar p increase in the
ampl i fi cat i on f act or is shown where t he flow become unst abl e due t o I KH analysis (poi nt e).
Fi gure 9 is a t ypi cal r epr esent at i on of t he stability behavi or of downwar d inclined stratified flow.
The ampl i fi cat i on f act or curves f or t he t wo analyses yield al most the same results. However , al ong
t he region where t he I KH analysis predi ct s zero ampl i fi cat i on (namel y, neut ral stability condi t i ons),
t he flow is unst abl e accor di ng t o t he VKH analysis with a very small ampl i fi cat i on rate. Thi s means
t hat , f or this case, t he neut ral stability line due t o the VKH analysis is absent and the t ransi t i on
5 0
4 0
3 0
I0
0
' ~ ' ~ ' " ' I ' ' ' ' ' ' " I ' ' ' ' ' " I
- U L s = O . O I m l s Z
. . . . . I K H
V K H
Y
i i i l J l , i l L i i , l i i l i j = , , l , , , I
I I0 I 0 0
U o s (m/s)
i i i i i i l l
o

i i , l l l h
0 0 0
Figure 8. Amplification factor for upwards inclined flow, ULS=0.01 m/s. Air-water, a t m o s p h e r i c
pressure, # = 0.25 , D = 5 cm.
648 D. BARNEA and Y. TAITEL
IOO
"x
I
O
-IO
' ' ' ' " 1 ' ' ' ' ' ' " 1 ' ' ' ' ' ' " 1 ' ' ' ' ' ' "
IO
SL / h L / o = O 5 / "
\
RW I g 8 - ~
7
1.0 I 0 I 00
UGS(m/s)
i i , i l l l ] i i . l l i , I I I , ! l i t i i J
I tO I 0 0
I.O
OI
~OOI
OOOI t
O O I OI
ULs=OI m/ s
. . . . . I KH
VKH
= , , , i J ,
I 0 0 0
UGS ( m / s )
Figure 9. Amplification factor for downwards inclined flow, ULs = 0.1 m/s. Ai r-wat er, atmospheric
pressure, / / = - 5 , D = 5 cm. SL--sl ug; ST--stratified; RW- - r ol l waves; A- - annul ar .
from stratified flow is controlled only by the IKH analysis and ht / D = 0.5. The whole region
bounded by the aforementioned lines is, therefore, a region of stratified roll waves.
Note that Andritsos & Hanratty (1987), using a two-dimensional, inviscid analysis for a flat
geometry, also related the wave type to the amplification factor. This approach and interpretation,
however, is quite different from the one presented here.
SUMMARY AND CONCLUSIONS
(1) This work addresses the dilemma that the neutral stability lines based on the viscous
complete two-fluid model equations and the approximate inviscid analysis lead to similar
results for high viscosity but quite different results for low liquid viscosity. This fact seems
to be unreasonable since one would expect the VKH analysis to approach the IKH case for
low viscosity.
(2) It is shown that the results for the amplification factor based on the IKH and VKH analyses
are almost the same.
(3) For increasing liquid flow rates or increasing gas flow rates the amplification factor for the
IKH analysis is zero up to a point, where after it grows very sharply to very high value. This
point is the neutral stability point for the IKH analysis.
(4) In the range where the IKH analysis predicts zero amplification, the VKH analysis predicts
a very low amplification factor. This amplification factor can be positive (unstable) or
negative (stable). The transition from negative to positive amplification is the neutral stability
line for the VKH analysis.
(5) In spite of the fact that the general behavior for the amplification factor for the VKH and
IKH analysis is similar and that the sharp increase in amplification factor occurs almost at
the same conditions, the neutral stability points are quite different for low viscosity fluids.
(6) The results obtained in this analysis confirm the interpretation of Barnea (1991). The region
of low amplification, i.e. the region between the neutral stability line of the VKH analysis
and that of the IKH analysis, is a region of roll waves or slug flow (depending on the liquid
holdup). The region of high amplification, i.e. the region above the IKH neutral stability
line is a region where the flow will be either slug or annular.
REFERENCES
ANDREUSSI, P. , ASALI, J. C. & HANRATTY, T. J. 1985 Initiation of roll waves in gas-liquid flows.
.4IChE Jl 31, 119-126.
ANDRITSOS, N. & HANRATTY, T. J. 1987 Interfacial instability for horizontal gas-liquid flows in
pipelines. Int. J. Multiphase Flow 13, 583-603.
KH STABILITY CRITERIA FOR STRATIFIED FLOW 649
ANDRITSOS, N. , WILLIAMS, L. & HANRATTY, T. J. 1989 Effect of liquid viscosity on the
stratified-slug transition in horizontal pipe flow. Int. J. Multiphase Flow 15, 877-892.
BARNEA, D. 1991. On the effect of viscosity on stability of stratified gas liquid flow--application
to flow pattern transition at various pipe inclination. Chem. Engng Sci. J. 46, 2123-2131.
BARNEA, n. TAITEL, V. 1989 Transient formulation modes and stability of steady state annular
flow. Chem. Engng Sci 44, 325-332.
BARNEA, D., SHOHAM, O., TAITEL, Y. & DUKLER, A. E. 1980 Flow pattern transition for gas-liquid
flow in horizontal and inclined pipes. Comparison of experimental dat a with theory. Int. J.
Multiphase Flow 6, 217-226.
COHEN, S. L. & HANRATTY, T. J. 1968 Effects of waves at a gas-liquid interface on a turbulent
air flow. J. Fluid Mech. 31, 467-469.
CROWLEY, C. J., WALLIS, G. B. & BARRY, J. J. 1992 Validation of a one-dimensional wave model
for the stratified to slug flow regime transition, with consequences for wave growth and slug
frequency. Int. J. Multiphase Flow 18, 249-271.
HANRATTY, T. J. 1983 Interfacial instabilities caused by air flow over a thin liquid layer. In Waves
on Fluid Interfaces, pp. 221-259. Academic Press, New York.
LIN, P. Y. & HANRATTY, T. J. 1986 Prediction of the initiation of slugs with linear stability theory.
Int. J. Multiphase Flow 12, 79-98.
LIN, P. Y. & HANRATTY, T. J. 1987 Effect of pipe diameter on the interfacial configurations for
ai r-wat er flow in horizontal pipes. Int. J. Multiphase Flow 13, 549-563.
KORDYBAN, E. S. 1977 Some characteristics of high waves in closed channels approaching
Kelvin-Helmholtz instability. AS ME Jl Fluids Engng 99, 339-346.
KORDYBAN, E. S. & RANOV, T. 1970 Mechanism of slug formation in horizontal two-phase flow.
J. Bas. Engng 92, 857-864.
MISHIMA, K. & Isun, M. 1980 Theoretical prediction of onset of horizontal slug flow. AS ME J!
Fluids Engng 102, 441~145.
NICHOLSON, M. K., AZIZ, K. & GREGORY, G. A. 1978 Intermittent two phase flow in horizontal
pipes: predictive models. Can. J. Chem. Engng 56, 653~63.
TAITEL, Y. & DUKLER, A. E. 1976 A model for prediction of flow regime transitions in horizontal
and near horizontal gas-liquid flow. ,,fiChE Jl 22, 47-55.
WALLIS, G. B. 1969 One-dimensional Two-phase Flow. McGraw-Hill, New York.
Wu, H. L., POTS, B. F. M., HOLLENBERG, J. F. & MEERHOFF, R. 1987 Fl ow pattern transitions in
two-phase gas/condensate flow at high pressure in an 8-inch horizontal pipe. In Proc. 3rd Int.
Conf. on Multi-Phase Flow, The Hague, The Netherlands, pp. 13-21.