R 091825471

Proceedings World Geothermal Congress 2000
Kyushu - Tohoku, Japan, May 28 - June 10, 2000
GEOTHERMAL TWO-PHASE FLOW IN HORIZONTAL PIPES

H.D. Zhao, K.C. Lee and D.H. Freeston
Geothermal Institute, The University of Auckland, Private Bag 92019, Auckland, New Zealand
Key Words: geothermal, steam-water flow, two-phase flow.
Harrisons (1975) correlation, based on geothermal data, is the

most suitable.
ABSTRACT
Freeston et al. (1983) studied geothermal two-phase pressure
drops in 100 mm diameter straight pipe and fittings, collected
255 sets of valuable data. However, attempts to correlate twophase pressure gradients in a straight pipe as a function of
liquid phase velocity give divergent result (see Figure 1).
This paper reanalysed the 255 sets of geothermal two-phase

flow data of Freeston et al (1983).
A new void fraction correlation is proposed in this paper. The
new correlation is derived from the analysis of two-phase flow
velocity distribution using the Seventh Power Law as:
7 /8
1
g f
= ( 1)( )( )
f g
x
y = 0.0537x + 0.2598x
2
R = 0.6709
7
Measured dp/dz, kPa/m
7 /8
To predict the two-phase pressure drop, an equivalent pseudosingle-phase flow having the same boundary layer velocity
distribution is assumed. The average velocity of the equivalent
single-phase flow is used to determine the wall friction factor
and hence the two-phase pressure drop. This method gives
very good agreement with the experimental data. The average
velocity of the equivalent single-phase flow is also a very good
correlating parameter for the prediction of geothermal twophase pressure drops in a horizontal straight pipe.
6
5
4
3
2
1
0
INTRODUCTION
Geothermal resources suitable for power production are mainly

of the wet type that requires the separation process for use in
conventional steam turbines. The separation process can take
place anywhere between the wellhead and the powerhouse.
Separation at the wellhead allows high separation pressure but
lower steam output due to higher wellhead pressure (WHP).
On the other hand, separator near the turbine has low steam
pressure but more quantity of steam. Hence the separation
pressure determines the turbine inlet pressure and the
utilisation efficiency of the resource. Since the two-phase well
fluid pressure drop in the transmission line between the
wellhead and the separator affects these two important
parameters, it is important to design the two-phase pipelines to
transport the two-phase fluid as efficiently as possible.
Liquid phase velocity, m/s
Figure 1: Measured pressure gradient versus liquid phase

velocity (prior to the introduction of PC) of Freeston et al
(1983).
It has to be pointed out that these results were computed prior
to the availability of powerful Personal Computer (PC) and
spreadsheet software. Steam-water properties were correlated
from steam tables.
Mundakir (1997) reevaluated the data on a PC using a
spreadsheet software, and steam-water properties were
obtained from the computer software Engineering Equation
Solver (EES) by Klein and Alvarado (1994). The correlation
with liquid phase velocity improved significantly as shown in
Figure 2. However, attempts to correlate pressure drop across
pipe fittings showed no consistent results with liquid phase
velocity.
James (1968) and Takahashi et al. (1970) showed that there

were economic advantages using long two-phase transmission
pipelines. From experience at Wairakei, this has the added
advantage of solving the steam pipe corrosion problem due to
the steam condensate. However, there is a minimum length of
steam pipe required upstream of the steam turbines to scrub the
steam of salts carryover from the separation process and for
steam pressure control.
Mundakirs (1997) attempt to correlate the measured pressure

gradient of the straight pipe as a function of a two-phase
pressure coefficient [(1/2V2)TP.] gave divergent results similar
that of Figure 1.
Although there are many theories and correlations available for

predicting two-phase pressure drop, most of them were based
on small pipe diameter (<60 mm) and high heat transfer rate
conditions for single-component flows else they are twocomponent, two-phase flows. For geothermal application
(large diameter, low heat transfer rate and single-component),
This paper is an attempt to find a better correlating parameter

for the two-phase pressure drop data of Freeston et al. (1983).
3349
Zhao et al
where,
u = Velocity at radius r,
U = Central line velocity,
r = Radius,
R = Pipe radius.
9
2
y = 0.0778x + 0.055x
2
R = 0.8942
7
6
If the void fraction for a two-phase flow is , then the gaseous

phase inside core has the radius of 1/2R and the liquid phase
flows within the outer layer of the radius from 1/2R to R.
5
4
By applying the Seventh-Power Law to the liquid phase layer,

we get the liquid phase volume flow rate:
3
2
0
0
Liquid phase velocity, m/s
Figure 2: Measured pressure gradient versus liquid phase

velocity (results of analysis by a PC and EES software) of
Mundakir (1997).
2rdr
8
49
)
Uf R 2 (1 ) 8 / 7 (1 +
7
60
mf = fVf =
(2)
8
49
fUf R 2 (1 ) 8 / 7 (1 +
)
7
60
(3)
where,
Vf = liquid phase volume flow rate,
mf = liquid phase mass flow rate,
Uf = liquid phase centreline velocity assuming liquid
phase only flowing through the pipe with the same
boundary velocity distribution as that of the twophase liquid layer.
The original data of Freeston et al (1983) were contained in

Lee et al (1979). The data were gathered from an experimental
rig installed at geothermal well WK207 at Wairakei (see
Figure 3). Although there were a total of 255 sets of data, only
189 sets were good. The rest was either incomplete trial runs,
incorrect steam orifice plate installation, or flows not at critical
flow for the James Lip Pressure method to be applicable.
From above equation, we also get:
The data presented in Lee et al (1979) were mainly measured

gauge pressures and pressure drops. The original computed
results were presented in Lee and Freeston (1979).
49
8
)
Uf R 2 (1 ) 8 / 7 (1 +
Vf
60
7
=
Vf =
Af
(1 )R 2
This paper reanalysed the horizontal straight pipe data in Lee

et al (1979).
METHOD OF ANALYSIS
49
=
Uf
60
In two-phase flow pressure drop prediction by the separated

flow model, void fraction () is the most important
fundamental parameter.
It determines other two-phase
parameters such as the liquid phase velocity, V f , and the
(1 ) 8 / 7 (1 +
8
)
7
(1 )
(4)
and
r
0 Uf (1 R )
V =
These in turn determine the two-phase
so
The Seventh-Power Law velocity distribution is used to

describe many fluid flow situations. It is used to derive the
void fraction correlation for the two-phase flow pressure drop
prediction in this paper.
Vf
=
V
where,
V
The Seventh-Power Law velocity distribution for a turbulent

flow can be expressed as:
u
r
= 1
U
R
1/ 7
So the liquid phase mass flow rate is:
DATA REVIEW
mean density, .
pressure drop.
R Uf (1 R )
Vf =
1/ 7
(1)
3350
1/ 7
2rdr
(1 ) 8 / 7 (1 +
49
Uf
60
8
)
7
(1 )
(5)
(6)
average liquid phase film velocity,
average velocity of the equivalent singlephase flow as defined for Uf

velocity ratio, a parameter which will be used
for predicting pressure drop.
/V =
Zhao et al
Inclined
pipe
Straight pipe section
Tee
LR mitred
bend
Two-phase & steam bypass
Horizontal
Silencer
Water bypass
Water
Sudden
contraction
Vertical
Silencer
Steam
U bend
Well
Separator
S bend
45S bend
LR elbow
Rupture
disc
Figure 3: Schematic experimental layout of Freeston et al (1983).
By applying the Seventh-Power Law to gaseous phase, the

velocity distribution of the gaseous core is:
where,
u
r
u = u IP + Ug 1
R
So gaseous phase mass flow rate is:

49
(9)
gUgR 2
60
After getting the mass flow rates for two phases, we can relate
them to the steam quality of the two-phase flow system:
mg
x=
mg + mf
1/ 7
mg =
(7)
= Velocity at radius r,
u IP
Ug
r
R
= Inter-phase velocity,
= Gaseous phase centreline velocity,
= Radius,
= Pipe radius.
Here, u IP is only approximately 7% of Ug. In order to

simplify the problem, it is considered to be negligible. It is
found later that the assumption only introduces 3% error to the
final value of void fraction, . Furthermore, the accuracy is
always of the same trend, so that it can be corrected by a factor
later. So the gaseous phase volume flow rate can be expressed
as:
Vg =
Ug (1
49
Ug R 2
60
1/ 7
g
8
Uf
g + ( )(1 ) 8 / 7 (1 +
) f
7
Ug
(10)
Here assuming Uf/Ug = g/f, which is a reasonable

assumption for a turbulent flow in this situation, then after
simplification the formula becomes:
1
g f
( 1)( )( )
f g
x
2rdr
7 /8
(1 )(1 +
7 /8
According to Wallis (1969),

(8)
(1 )(1 +
3351
8
)7 /8
7
8
)7 / 8 1
7
(11)
Zhao et al
It is found later that this approximation only introduces an

error of <1% to void fraction, > 0.88, and <5% to . > 0.61.
SAMPLE CALCULATION
The final void fraction correlation is:
Measured data (Run 99)

Pipe inner diameter, D
Pipe wall roughness,
Pressure, p
Water flow, mf
Steam flow, mg
Pressure gradient,dp/dz
= 0.1023 m
= 0.00015 m
= 2.925 bar abs
= 3.868 kg/s
= 1.625 kg/s
= 4.867 kPa/m
Predicted pressure gradient

Pipe cross-sectional area, A
Total mass flow, m
Dryness, x
Fluid specific enthalpy, h
Void fraction, , (Eq 12)
Liquid phase velocity, Vf (Eq 13)
Equivalent single phase velocity,V (Eq 6)
Re based on V
Friction factor,
Liquid density, f
Specific steam volume, vg
= 0.008215 m2
= 5.493 kg/s
= 0.2958
= 1198 kJ/kg
= 0.9164
= 4.676 m/s
= 6.857 m/s
= 3.14E6
= .0217
= 932 kg/m3
= 0.620 m3/kg
7 /8
1
g f
= ( 1)( )( )
f g
x
7 /8
(12)
The assumptions made in deriving the void fraction correlation

can be summarised as:
1.
2.
3.
Gaseous phase centreline velocity Ug is not sensitive to

flow diameter, i.e. for flow diameter of 2R and 21/2R, the
Ug value is the same.
Inter-phase velocity effect on gaseous phase is negligible.
No liquid is in gaseous phase core, i.e. no entrainment.
PREDICTION OF PRESSURE DROP

Harrison (1975) used average liquid phase velocity V f and
superficial liquid phase wall friction factor (Cf)f,s to predict
two-phase pressure gradient. However, the idea of using
friction factor and velocity to predict pressure gradient comes
from single-phase flow system. Therefore, it seems a better
idea to relate two-phase system to a pseudo-single-phase
system which has the same pressure gradient. Then two-phase
pressure gradient is evaluated through the equivalent singlephase flow system. Such an equivalent single-phase system is
considered to be the one which has the same boundary layer
velocity distribution as the two-phase flow liquid layer.
Wall shear stress, w = fV2/8

Pressure gradient, dp/dz = 4w/D
(neglecting acceleration effect)
= 118.9 Pa/m2
= 4.649 kPa/m
Acceleration correction, AC = mg2vg/(pA2)
= 0.0905
Pressure gradient, dp/dz = 4w/[D(1-AC)]
= 5.112 kPa/m
RESULTS AND DISCUSSION
Although equation (12) is not in the simplest form, it can still

be solved by trial and error. The equation is solved by
Engineering Equation Solver (EES) computer program.
Figure 4 shows the result of the new method of calculation.

Comparing to Figure 5 of Harrisons prediction method, Figure
4 gives less scatter of data points.
The average liquid phase film velocity can be expressed as:

V f = 1.1 (1 x)
(13)
= Total mass flow rate,

= Steam quality,
= Liquid phase specific volume,
= Correction factor mainly for the entrainment.
7
where,
W
x
f
1.1(1-x)
W (1 x)f
(1 ) A
At this stage, a correction factor is introduced to account for

the entrainment effect and the simplification made in deriving
void fraction correction. It can be explained as 1.1(1-x)
fraction of the liquid phase is left in the liquid phase boundary
layer. The other fraction is entrained inside the gaseous phase
as water droplets. When the steam quality decreases, the
gaseous phase can carry less liquid. This means a higher
percentage of the liquid is left in the boundary layer. The
choice of factor is mainly to give a good result rather than
having a rigorous theoretical justification.
6
5
4
3
2
1
0
0
Equation (6) shows the relationship between the average liquid

phase film velocity and the average velocity of the equivalent
single-phase flow. The average velocity of the equivalent
flow, V , can then be calculated. A sample calculation is given
below.
Predicted dp/dz (new method), kPa/m
Figure 4: Measured versus predicted geothermal two-phase

flow pressure gradient by the new method.
3352
Zhao et al
CONCLUSIONS
9
A new void fraction correlation derived from the Seventh

Power Law gives good agreement between the measured and
the predicted two-phase pressure drops in geothermal steamwater flow in large pipes.
7
6
The average velocity of the equivalent single-phase flow, V ,

is a good correlating parameter of the geothermal two-phase
pressure drops in straight horizontal pipes.
5
4
REFERENCES
Freeston DH, Lee KC and Hole H (1983): Utilisation of

geothermal energy two-phase flow studies. NZERDC Rep
#95, The University of Auckland, Sept 1983.
2
1
0
0
Harrison RF (1975): Method for the analysis of geothermal

two-phase flow. ME thesis, The Univ of Auckland.
Predicted dp/dz (Harrison), kPa/m
James CR (1968): Second generation geothermal power. NZ

Engineering, vol 23, pp230-236.
Figure 5: Measured versus predicted geothermal two-phase
flow pressure gradients by Harrisons method (1975).
Klein SA and Alvarado FL (1994). Engineering Equation

Solver (EES) for the Microsoft Windows Operation System.
F-Chart Software, 4406 Fox Bluff Rd, Middleton, Wisconsin
53562, USA.
In Figure 6, the measured pressure drops are plotted as a

function of the average velocity of the equivalent single-phase
flow, V . It shows a marginally improved correlating
parameter than the liquid phase velocity in Figure 2 of
Mundakir (1997). However, it also appears to be a better
correlating parameter for pressure drop across pipe-fittings
than liquid phase velocity but this is inconclusive at this point
in time.
Lee KC and Freeston DH (1979): Experimental results of

geothermal steam-water flow I a 10-cm-pipe rig at Wairakei.
Dept of Mech Engng, Rep ME R79/3, The University of
Auckland, May 1979.
Lee KC, Hole HM and Freeston DH (1979): Pressure-drop
data on geothermal steam-water flow in a 10-cm-pipe
experimental rig. Dept of Mech Engng, Rep ME R79/1, The
University of Auckland, March 1979.
9
2
y = 0.0855x + 0.1044x
2
R = 0.8952
Measured pressure gradient, kPa/m
Mundakir A (1997): Two-phase flow studies: Reevaluation of

pressure losses in pipe fittings. Geothermal Institute report
#97.15, The University of Auckland, Oct 1997.
7
6
Takahashi Y, Hayashida T, Soezima S, Aramaki S, and Soda

M (1970): An experiment on pipeline transportation of steamwater mixtures at Otaki geothermal field. UN Symp on the
development and utilisation of geothermal resources, Pisa,
Geothermics special issue 2, pp 882-891.
5
4
3
2
Wallis GB (1969): One-dimensional two-phase flow. McGraw

Hill, New York.
1
0
0
Equivalent single-phase average velocity, m/s
Figure 6: Measured pressure gradient versus equivalent singlephase average velocity.
3353

R 091825471

Uploaded by

Copyright:

Available Formats

R 091825471

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

R 091825471

Uploaded by

Copyright:

Available Formats

Proceedings World Geothermal Congress 2000

Kyushu - Tohoku, Japan, May 28 - June 10, 2000

GEOTHERMAL TWO-PHASE FLOW IN HORIZONTAL PIPES

Key Words: geothermal, steam-water flow, two-phase flow.

Harrisons (1975) correlation, based on geothermal data, is the

This paper reanalysed the 255 sets of geothermal two-phase

Geothermal resources suitable for power production are mainly

Liquid phase velocity, m/s

Figure 1: Measured pressure gradient versus liquid phase

James (1968) and Takahashi et al. (1970) showed that there

Mundakirs (1997) attempt to correlate the measured pressure

Although there are many theories and correlations available for

This paper is an attempt to find a better correlating parameter

Measured dp/dz, kPa/m

If the void fraction for a two-phase flow is , then the gaseous

By applying the Seventh-Power Law to the liquid phase layer,

Liquid phase velocity, m/s

Figure 2: Measured pressure gradient versus liquid phase

The original data of Freeston et al (1983) were contained in

From above equation, we also get:

The data presented in Lee et al (1979) were mainly measured

This paper reanalysed the horizontal straight pipe data in Lee

In two-phase flow pressure drop prediction by the separated

These in turn determine the two-phase

The Seventh-Power Law velocity distribution is used to

The Seventh-Power Law velocity distribution for a turbulent

So the liquid phase mass flow rate is:

average liquid phase film velocity,

average velocity of the equivalent singlephase flow as defined for Uf

Two-phase & steam bypass

By applying the Seventh-Power Law to gaseous phase, the

So gaseous phase mass flow rate is:

Here, u IP is only approximately 7% of Ug. In order to

Here assuming Uf/Ug = g/f, which is a reasonable

According to Wallis (1969),

It is found later that this approximation only introduces an

The final void fraction correlation is:

Measured data (Run 99)

Predicted pressure gradient

The assumptions made in deriving the void fraction correlation

Gaseous phase centreline velocity Ug is not sensitive to

PREDICTION OF PRESSURE DROP

Wall shear stress, w = fV2/8

Acceleration correction, AC = mg2vg/(pA2)

Pressure gradient, dp/dz = 4w/[D(1-AC)]

RESULTS AND DISCUSSION

Although equation (12) is not in the simplest form, it can still

Figure 4 shows the result of the new method of calculation.

The average liquid phase film velocity can be expressed as:

= Total mass flow rate,

At this stage, a correction factor is introduced to account for

Equation (6) shows the relationship between the average liquid

Predicted dp/dz (new method), kPa/m

Figure 4: Measured versus predicted geothermal two-phase

A new void fraction correlation derived from the Seventh

Measured dp/dz, kPa/m

The average velocity of the equivalent single-phase flow, V ,

Freeston DH, Lee KC and Hole H (1983): Utilisation of