Two Phase Flow

Analysis of two phase flow systems is complicated by the appearance and disappearance of phases, discontinuous
fluid properties at phase boundaries, the relative motion of the phases and mass, energy and momentum transfer
between the phases. Of particular interest is the void distribution in a two phase channel. An example of a
subcooled boiling channel is illustrated below.

Figure 1: Void Distribution in a Heated Channel (from Rust, Nuclear Power Plant Engineering)

Prior to the incipient boiling point (or nucleation point) zn, no voids exist and the flow characteristics are those of a
single phase liquid. Between the incipient boiling point and the bubble departure point (zd), voids appear on the
heated surface but remain attached to the wall. As there is no flow of the vapor phase, the flow quality is zero and
the void fraction is related to the mass fraction of the vapor phase by the static quality. After the bubble departure
point, the flow quality is greater than zero and two phase flow exist with the liquid phase subcooled and the vapor
phase either superheated or saturated. Below the bulk boiling point (zbulk), the average fluid enthalpy is below that
of a saturated liquid. The equilibrium quality is less than zero and models based on equilibrium quality alone would
fail to predict the existence of voids in the channel. At the bulk boiling point, the average fluid enthalpy equals that
of a saturated liquid, even though the liquid phase may still be subcooled. The equilibrium quality is zero and
models based on equilibrium quality would begin to predict the existence of vapor in the channel. At the
equilibrium point (zeq), the liquid and vapor phases are both at saturation and the flow quality and equilibrium
quality are the same. Note, the phases reach thermodynamic equilibrium when both phases are at saturation, not
when the equilibrium quality is zero. To accurately describe the mass, energy and momentum distributions in a two
phase system requires conservation equations for each phase.

214
 Conservation Equations in Two-Phase Flow

Consider the arbitrary two-phase control volume illustrated below. For ease of representation we have assumed an
annular flow regime, though the multi-phase flow equations we derive will be valid regardless of the flow regime.

z +∆ z
liq. . liq.
δ 'i
vapor

Figure 1: Two-Phase Control Volume

In our derivation of the single phase flow equations, we considered two types of surfaces: (a) Flow or convective
surfaces (i.e. those at z and z+∆z) and (b) fixed or wall surfaces. The convective surfaces provided for the
transport of mass, energy and momentum through the normal convective flow of the fluid. The fixed or wall
surfaces provided surface sources of heat and momentum (drag forces). In two-phase flow, the interface between
the liquid and vapor phases provides an additional surface for both convective and surface sources of mass, energy
and momentum transfer. In this derivation we assume phase change can occur at heated surfaces ( δ&w′ ) and at the
interface between the liquid and vapor ( δ& ′ ), where δ& ′ represents the phase change rate per unit length. We define
i

a total effective phase change rate δ& ′ = δ&i′ + δ&w′ which is the sum of the interfacial and wall components and adopt a
sign convention such that δ& ′ is positive for evaporation and negative for condensation. No assumptions are made
regarding the state of the individual phases. For example, the liquid may be subcooled and the vapor superheated.
Under these conditions, heat transfer from the superheated vapor to the liquid/vapor interface could lead to
evaporation or condensation, depending on the relative states of the two phases. As a result, δ&w′ and δ&i′ can have
different signs, such as in subcooled boiling where evaporation occurs at the wall, but condensation occurs once the
bubbles have been injected into the flow stream.

Mass Conservation (Phasic Continuity Equation)

The two-phase conservation equations are obtained by performing mass, energy and momentum balances on the
individual phases. Consider first a mass balance on the liquid phase

dM l
dt
= ∑ m& − ∑ m&
in
l in
exit
l exit (1)

215
Note: M l = ρlVl = ρlα lV = ρlα l Ax ∆z
m& l = ( ρl vl Al ) = ( ρl vlαl Ax )

d (αl ρl )
∴ Ax ∆z = ( ρl vlαl Ax ) − ( ρl vlαl Ax )
z + ∆z
− δ&′∆z (2)
dt z

To obtain the differential form, divide by ∆z and take the limit as ∆z goes to zero

∂ (αl ρl ) ∂
Ax
∂t
+ (
α ρ v A = −δ&′
∂z l l l x
) (3)

which is the one-dimensional form of the liquid phase mass conservation equation. It is relatively easy to show that
a similar treatment of the vapor phase gives for the vapor phase mass equation

∂ (α g ρg ) ∂
Ax
∂t
+ (
α ρ v A = δ&′
∂z g g g x ) (4)

Energy Conservation

Recall, that energy conservation is simply a restatement of the first law

Q& + ∑ m& (e
in
in in + Pinυin = ) dMe
dt
+ ∑ m& exit
exit (eexit + Pexitυexit ) (5)

v2
where e ≡ u + + gH
2

and we have neglected the external work terms. We again consider a control volume around the liquid phase such
that the first law is

dM l el
dt
+ ∑ m&
exit
l exit (el + Plυl )exit − ∑ m& lin (el + Plυl )in = Q& l
in
(6)

For: M l = α l ρl Ax ∆z
m& l (el + Plυl ) exit = (αl ρl el vl Ax ) + (αl Pl vl Ax ) + δ&′∆z(el + Plυl ) i
z + ∆z z + ∆z

m& l (el + Plυl ) in = (αl ρl el vl Ax ) + (αl Pl vl Ax )

z z
Q& l = q& w′ ,l ∆z + q&i′,l ∆z
123 123
Wall Interface

the liquid phase energy equation is

d (α l ρl el )
A x ∆z + (α l ρl el v l Ax ) − (α l ρl el v l Ax ) + (α l Pl v l Ax ) − (α l Pl v l Ax ) = −δ&′∆z (el + Plυl ) i + q& w′ ,l ∆z + q& i′,l ∆z (7)
dt z + ∆z z z + ∆z z

If we again divide by ∆z and take the limit as ∆z goes to zero we obtain the one-dimensional, differential form of
the liquid phase energy equation

216
 ∂ (αl ρl el ) ∂ ∂
Ax
∂t
+
∂z
(
αl ρl el vl Ax +
∂z l l l x
) ( )
α P v A = −δ&′(el + Plυl ) i + q& w′ ,l + q&i′,l (8)

and similarly for the vapor phase

∂ (α g ρg e g ) ∂
Ax
∂t
+ (
α ρ e v A +
∂
∂z g g g g x ∂z g g g x ) (
α P v A = δ&′(e g + Pgυg ) i + q& w′ ,g + q&i′, g ) (9)

Momentum Conservation

We consider the arbitrary two-phase control volume illustrated below.

liq
.
vap
or
.
δ' liq
z+
∆z
.
τi

Figure 2: Momentum Control Volume

A momentum balance on the liquid phase gives

1 d ( M l vl )
gc dt
=
1
gc ∑ (m& v )
in
l l in −
1
gc ∑ (m& v ) + ∑ ( F )
exit
l l exit l z (10)

For: M l = α l ρl Ax ∆z
(m& l vl ) exit = (αl ρl vl vl Ax ) + δ&′∆z (vl ) i
z + ∆z

(m& l vl ) in = (α l ρl vl vl Ax )
z

The liquid phase momentum equation is

Ax
gc
∆z
d (α l ρl v l )
dt
+
1
(α ρ v v A ) −
1
g c l l l l x z + ∆z g c l l l l x z
1
(α ρ v v A ) = − δ&′∆z ( v l ) i +
gc ∑ (F )
l
z
(11)

As in the single phase case, the forces acting on the fluid are (1) pressure forces, (2) viscous or drag forces, (3)
weight forces and (4) mechanical forces (if present).

Pressure Forces

The pressure forces in the two-phase equations are similar to those in the single-phase equations and may be written

217
 Pl Al z − Pl Al z + ∆z (
+ Pl Al z + ∆z
− Al z ) (12)

Viscous Forces

In addition to wall sheer, an additional interfacial sheer force acts at the liquid/vapor interface. We will assume a
sign convention, such that the vapor phase is exerting a positive force on the liquid phase (and conversely, a
negative force on the vapor phase). The sheer force is then

− (τ w Pw ) l ∆z + τ i Pi ∆z (13)

where Pi is the interfacial perimeter and a function of the flow regime.

Weight Force

The weight force in two phase flow is again similar to that in single phase flow and given as

g
− α l ρl Ax ∆z sin θ (14)
gc

Mechanical Forces

If pumps are present in the flow system, then the pump force acting on the liquid phase is

∆Pp′l Al ∆z (15)
The liquid phase momentum equation is then

Ax ∆z ∂α l ρ l vl
gc ∂t
=
1
gc
(m& v) l z −
1
gc
(m& v) l z + ∆z
−
1 &
gc
δ ′∆z (vl ) i + Pl Al z − Pl Al z + ∆z
(
+ Pl Al z + ∆z
− Al z
)
(16)
g
− (τ w Pw ) l ∆z + (τ i Pi )∆z − α l ρ l Ax ∆z sinθ + ∆Pp′l Al ∆z
gc

We again divide by ∆z and take the limit as ∆z → 0 to give

Ax ∂α l ρ l vl
+
1 ∂
(α l ρ l vl vl Ax ) = − 1 δ& ′(vl )i − ∂ (Pl Al ) + Pl ∂Al
gc ∂t g c ∂z gc ∂z ∂z
(17)
g
− (τ w Pw ) l + (τ i Pi ) − α l ρ l Ax sin θ + ∆Pp′ l Al
gc

or by expanding the pressure derivative and recognizing Al = α l Ax

Ax ∂α l ρ l vl
+
1 ∂
(α l ρ l vl vl Ax ) = − 1 δ&′(vl ) i − α l Ax ∂Pl
gc ∂t g c ∂z gc ∂z
(18)
g
− (τ w Pw ) l + (τ i Pi ) − α l ρ l Ax sin θ + ∆Pp′ lα l Ax
gc

Similarly for the vapor phase

218
 Ax ∂α g ρ g v g 1 ∂ ∂Pg
gc ∂t
+
g c ∂z
( )
α g ρ g v g v g Ax =
1 &
gc
δ ′(v g ) i − α g Ax
∂z
(19)
g
− (τ w Pw ) g − (τ i Pi ) − α g ρ g Ax ′ α g Ax
sin θ + ∆Ppg
gc

Equations 18 and 19 are the one-dimensional form of the phasic momentum equations.

Closure

The phasic conservation equations include a number of phase dependent interfacial terms, i.e. δ& ′ , τ i Pi , q&i′,g , q&i′, l ,
(el + Plυl ) i , (eg + Pgυ g ) i , (vl ) i and (v g ) i . In addition, we have phase/wall interaction terms such as q& w′ ,l , q& w′ ,g ,
(τ w Pw ) l and (τ w Pw ) g . If we assume that these terms can be correlated in terms of the remaining phasic variables,
then the phasic conservation equations contain as unknowns

Unknowns ⇒ α k , ρ k , uk , v k and Pk ( k = l, g ) = 10

The phasic conservation equations provide six equations, implying four more are required for closure. As in the
single phase equations, equations of state are required for each phase, i.e.

State

ρ k = ρ k (uk , Pk ) k = l, g (20)

which gives then an total of eight equations. In addition, we have the volume constraint

∑α
k = l ,g
k =1 (21)

leaving one equation required for closure. The last equation relates the phase pressures, i.e. Pl = Pl ( Pg ) , the
simplest of which is simply Pl = Pg = P .

The phasic equations are the most general form of the multi-phase conservation equations and represent the current
state of the art in reactor safety analysis. Since there exist 6 equations ( two each for mass, energy and momentum)
models based on the phasic conservation equations are often called six equation models. The phasic equations
explicitly represent the motion and thermodynamic state of each phase and allow for the modeling of non-
homogenous (unequal phase velocities) and non-equilibrium systems. While considered to be the most general
form of the multi-phase equations, the phasic equations present computational difficulties and contain interfacial
interaction terms which are in general not well known. As a result, we will concentrate on a simpler set of multi-
phase equations which eliminate some of these concerns.

219
 Mixture Equations

Mixture equations represent global balances of mass, energy and momentum in a two-phase system, and are
obtained by adding the phasic balance equations.

Mixture Mass

To obtain a total mass balance on the control volume, add the phasic mass balance equations to yield

Ax
∂
∂t (
αl ρl + α g ρg +
∂
) (
α ρ v + α g ρg v g Ax = 0
∂z l l l ) (22)

Note the cancellation of the interfacial mass transfer term, as mass leaving the liquid phase due to evaporation
appears in the vapor phase. The term α l ρl + α g ρ g is the effective two-phase density ρ. In addition, from previous
discussions α l ρl vl + α g ρ g v g is the total mass flux G, such the mass balance equation can be written as

∂ρ ∂GAx
Ax + = 0. (23)
∂t ∂z

Notice, that the mixture mass equation is identical in form to the single phase mass equation.

Mixture Energy

As with the total or mixture mass equation, we can obtain a mixture energy equation by adding the phasic equations.

∂ ∂ ∂
Ax
∂t
(α l ρl el + α g ρ g e g ) + (α l ρl el v l + α g ρ g e g v g ) Ax + (α l Pl v l + α g Pg v g ) Ax =
∂z ∂z (24)
δ& ′(e g + Pg υ g − el − Plυl ) i + q& w′ , g + q& i′, g + q& w′ ,l + q& i′,l

The sum of the wall heat transfer terms is simply q& w′ = q& w′ ,l + q& w′ , g or the total two-phase linear heat rate. The
treatment of wall heat transfer under two-phase conditions was considered in our discussion of boiling regimes and
boiling heat transfer and can be relatively easily correlated in terms of mixture parameters. Equation 24 contains
interfacial terms that at first glance do not immediately cancel. To address these additional terms, consider an
energy balance on the interface illustrated below. Note, that since the interface has no volume, we need not
consider volumetric or accumulation (time derivative terms). An energy balance on the interface gives

q& g′ ,i + δ&′(el + Plυl ) = q&i′,l + δ&′(eg + Pgυg ) (25)

or

q& g′ ,i = q&i′,l + δ&′(eg + Pgυg − el − Plυl ) (26)

Since q& g′ ,i = − q&i′, g the interfacial terms cancel in the mixture energy equation leaving

Ax
∂
(
α ρ e + α g ρ g eg +
∂t l l l
∂
) (
α ρ e v + α g ρ g eg v g Ax +
∂z l l l l
∂
) (
α P v + α g Pg v g Ax = q& w′
∂z l l l ) (27)

220
 Vapor Liquid

. .
'
qg,i q i,' l

. .
δ' ( e g + Pg νg ) δ' ( el + Pl νl )

Interface

Figure 2: Two-Phase Interface

To further simplify the energy equation, recall that

α l ρl vl = G (1 − x )
α g ρ g v g = Gx

such that the convective terms may be written

α l ρl el vl + α g ρ g eg v g + α l Pl vl + α g Pg v g = G (1 − x )(el + Plυ l ) + Gx (eg + Pgυ g )

⎧⎪ ⎛ vl2 ⎞ ⎛ v g2 ⎞ ⎫⎪
G (1 − x )(el + Plυl ) + Gx (eg + Pgυg ) = G ⎨(1 − x )⎜ hl + ⎜
+ gH ⎟ + x ⎜ hg + + gH ⎟⎟ ⎬
⎪⎩ ⎝ 2 ⎠ ⎝ 2 ⎠ ⎪⎭
(28)
⎧⎪ ⎛ vl2 ⎞ ⎛ v g2 ⎞ ⎫⎪
= G ⎨h + (1 − x ) ⎜ ⎟ + x ⎜⎜ ⎟⎟ + gH ⎬
⎪⎩ ⎝ 2⎠ ⎝ 2⎠ ⎪⎭

where we have defined a general mixture enthalpy

h = (1 − x )hl + xhg (29)

which is valid whether the phases are saturated or not. In this formulation, x is the flow quality, and from Equation
29 may be expressed as

h − hl
x= (30)
hg − hl

Note, that if the phases are at equilibrium, then this reduces to

h − hf
x= (31)
h fg

221
which is by definition the equilibrium quality. This is consistent with earlier statements that the equilibrium and
flow qualities are equal under saturation conditions.

The energy equation may then be written as

∂ ⎧⎪ ⎛ ⎞ ⎫⎪
( )
2
∂ v2 vg
Ax α l ρ l el + α g ρ g eg + ⎨G ⎜⎜ h + (1 − x ) l + x + gH ⎟⎟ ⎬ Ax = q& w′ (32)
∂t ∂z ⎪ ⎝ 2 2 ⎠ ⎪⎭
⎩

If we neglect kinetic and potential energy terms, then ek ≅ uk and the energy equation is

∂ ∂
Ax ( ρu) + { Gh} Ax = q& w′ (33)
∂t ∂z

where we have defined ρu ≡ α l ρ l ul + α g ρ g ug . Equation 33 is identical in form to energy equations we have used
in the past. Again, no assumptions have been made relative to the fluid state, such that the enthalpy in Equation 33
is our general mixture enthalpy.

Mixture Momentum

To obtain the mixture momentum equation, we add Equations 18 and 19 to get

Ax ∂ 1 ∂ ∂P
g c ∂t
( )
α l ρ l vl + α g ρ g v g +
g c ∂z
( )
α l ρ l vl vl + α g ρ g v g v g Ax =
1 &
gc
δ ′(v g − vl ) i − α l Ax l
∂z
(34)
∂Pg
sin θ + (α l ∆Pp′ l + α g ∆Ppg
′ )Ax
g
− α g Ax − (τ w Pw ) l − (τ w Pw ) g − (α l ρ l + α g ρ g ) Ax
∂z gc

For G = α l ρ l vl + α g ρ g v g
ρ = α l ρl + α g ρ g

and assuming equal phase pressures and equal phase velocities at the interface, the mixture momentum equation
reduces to

Ax ∂
(G ) + 1 ∂ α l ρ l vl vl + α g ρ g v g v g Ax = − Ax ∂P
( )
g c ∂t g c ∂z ∂z
(35)
− (τ w Pw ) l − (τ w Pw ) g − ρAx
g
gc
(
sin θ + α l ∆Pp′ l + α g ∆Ppg)
′ Ax

We define the total two-phase viscous loss as

(τ w Pw ) 2φ ≡ (τ w Pw ) l + (τ w Pw ) g (36)

and the total two-phase pump Delta P as

(∆Pp′ )2φ ≡ α l ∆Pp′l + α g ∆Ppg′

222
 Ax ∂
(G ) + 1 ∂ α l ρ l vl vl + α g ρ g v g v g Ax = − Ax ∂P
( )
g c ∂t g c ∂z ∂z
(37)
− (τ w Pw ) 2φ − ρAx
g
gc
( )2φ Ax
sin θ + ∆Pp′

The convective terms can not be written solely in terms of mixture density and mass flux due to the unequal phase
velocities. We can eliminate the phase velocities by noting

G (1 − x )
α l ρl vl = G (1 − x ) ⇒ vl = (38)
α l ρl

Gx
α g ρ g v g = Gx ⇒ v g = (39)
α g ρg

giving the final form of the mixture momentum equation

Ax ∂ ⎛ 2 ⎞
(G ) + 1 ∂ ⎜⎜ G (1 − x) + G x ⎟ Ax = − Ax ∂P
2 2 2

g c ∂t g c ∂z ⎝ α l ρ l α g ρg ⎟ ∂z
⎠ (40)
− (τ w Pw ) 2φ − ρAx
g
gc
( )
sin θ + ∆Pp′ 2φ Ax

The mixture mass, energy and momentum equations will serve as the basis for further discussions of two-phase
systems.

Solution of the Mixture Equations

For simplicity, consider the steady-state form of the mixture equations

Mass

∂GAx
=0 (41)
∂z

Energy

∂
{ Gh} Ax = q& w′ (42)
∂z

Momentum

1 ∂ ⎛⎜ G 2 (1 − x ) 2 G 2 x 2 ⎞⎟ ∂P
+ Ax = − Ax
⎜
g c ∂z ⎝ α l ρ l αg ρg ⎠⎟ ∂z
(43)
− (τ w Pw ) 2φ − ρAx
g
gc
( )
sin θ + ∆Pp′ 2φ Ax

supplemented by equations of state for each phase

223
State

ρ k = ρ k ( hk , P) k = l, g

where for convenience the state equation has been expressed in terms of enthalpy instead of internal energy. Again
assuming the two-phase heat transfer and frictional terms can be correlated in terms of the remaining system
variables, the number of unknowns is

Unknowns ⇒ α k , ρ k , hk , ρ , h, G , x , P ( k = l, g ) = 11

Since we now have only three conservation equations plus the two state equations, 6 additional equations or
constraints on the fluid must be found. The volume constraint

a) ∑α
k = l,g
k =1

and definitions

b)
∑α ρ
k = l,g
k k =ρ

c) h = (1 − x )hl + xhg

provide three additional equations, leaving three still required. At this point however, no additional equations are
available which requires that we resort to assumptions regarding the motion and/or state of the individual phases.
This information was lost when we added the phasic equations, which provided total balances on the fluid, but no
information regarding the distribution of mass, energy and momentum among the phases.

Void-Quality Relations

Recall in our earlier discussions of two-phase flow parameters, we derived the fundamental void-quality-slip
relationship

1
α= (44)
1 − x ρg
1+ S
x ρl

vg
where the slip ratio S = represents the ratio of the vapor to liquid velocities. The void fraction is then implicitly
vl
a function of the phase velocities. If we assume homogeneous flow ( S = 1 ), then Equation 44 reduces to

1
α= (45)
1 − x ρg
1+
x ρl

providing the final equation necessary for solution of the mixture equations. The homogeneous flow assumption is
overly restrictive however and does not provide a realistic representation of phase motion. As an alternative, semi-
empirical void-quality relationships have been developed which implicitly account for the phase motion. One
popular model of this type is the Zuber-Findlay model

224
 1
α= (46)
⎪⎧ ⎡ 1 − x ρg ⎤ ρgVgj ⎪⎫
⎨Co ⎢1 + ⎥+ ⎬
⎪⎩ ⎣ x ρl ⎦ Gx ⎪⎭

where Co is the Concentration Parameter and Vgj is the Drift Velocity. Co and Vgj are correlated parameters and
are in general functions of the flow regime. However, for high pressure steam water flows

Co = 1.13
1
⎛ σggc ( ρl − ρ g ) ⎞ 4

. ⎜
Vgj = 141 ⎟ (47)
⎝ ρl2 ⎠

can be used regardless of flow regime. Note, that for Co = 1 and Vgj = 0 the Zuber-Findlay model reduces to the
homogeneous void-quality-slip model.

Profile Fit Models

Since no information is available regarding the relative states of the phases, we could assume the phases are at
equilibrium, i.e.

hl = h f ( P )

hg = hg ( P )

Note, that this specification of the phase enthalpies will automatically produce saturation densities through the
equations of state and forces the flow quality and equilibrium quality to be equal over the entire channel. The
restriction on the phase states prevents modeling of subcooled boiling (from a fluid dynamics, not heat transfer
standpoint) and other non-equilibrium effects and limits the applicability of the mixture model.

A number of empirical models have been developed to determine the flow quality in heated channels under
nonequilibrium conditions. These “profile fit” models are generally functions of the equilibrium quality and fluid
conditions at the bubble departure point. One such model is by Levy and has the form

⎛ x ⎞
x = xe − ( xe ) d exp⎜⎜ e − 1⎟⎟
⎝ e d
( x ) ⎠

where xe is the local equilibrium quality and ( xe ) d is the equilibrium quality at the bubble departure point, i.e.

hld − h f
( xe ) d =
h fg

where hld is the liquid enthalpy at the bubble departure point and in single heated channels can be found from the
simple energy balance

∫
zd
1
hld = hin + q′′( z )πDdz
m& 0

225
Note, that while x will always be greater than zero, the equilibrium quality will be negative for h < h f . The profile
fit model allows for negative equilibrium quality.

The bubble departure point is also a correlated property. One correlation for the bubble departure point is the Saha-
Zuber correlation

⎧ DeC p
⎪0.0022 × q′′( zd ) × Pe < 70,000
⎪ k
h f − hld = ⎨
⎪ q′′( zd )
⎪ 154 × Pe > 70,000
⎩ G

GDeC p
where Pe ≡ = Re × Pr is the Peclet Number.
k

Since the heat flux is normally a nonlinear function of position, solution for the bubble departure point is in general
iterative.

The void-quality relationship and the profile fit model add two of the three necessary equations for closure. Since
no additional information is available, it is normally assumed that the vapor phase is saturated providing the final
equation necessary for solution of the Mixture Equations.

226