SPE

Society of Petroleum Engineers

SPE 15657

Two-Phase Flow Through Chokes

by A. Sachdeva, Z. Schmidt, J.P. Brill, and A.M. Blais, U. of Tulsa
SPE Members

Copyright 1986, Society of Petroleum Engineers

This paper was prepared for presentation at the 61st Annual Technical Conference and Exhibition of the Society of Petroleum Engineers held in New
Orleans, LA October 5-8, 1986.

This paper was selected for presentation by an SPE Program Commillee following review of information contained in an abstract submilled by the
author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the
author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers
presented at SPE meetings are subject to publication review by Editorial Commillees of the Society of Petroleum Engineers. Permission to copy is
restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment of
where and by whom the paper is presented. Write Publications Manager, SPE, P.O. Box 833836, Richardson, TX 75083-3836. Telex, 730989 SPEDAl.

ABSTRACT THEORY

Two-phase flow through wellhead chokes, including both For the purpose of modeling, a wellhead choke can be
critical and subcritical flow and the boundary between treated as a restriction in a pipe. Two types of two-
them, was studied. Data were gathered for air-water phase flow can exist in a choke: critical and subcri-
and air-kerosene flows through five choke diameters tical flow. During critical flow, the flow rate
from 1/4 in. (6.35 rom) to 1/2 in. (12.7 rom), and through the choke reaches a maximum value with respect
results were compared to published correlations. A to the prevailing upstream conditions. The velocity
new theoretical model for predicting flow rates and of the fluids flowing through the restriction reaches
the critical-subcritical flow boundary was tested the sonic or pressure wave propagation velocity for
against these data, as well as data from two published the two-phase fluid. This implies that the flow is
studies. The new model substantially improves the "choked" because downstream disturbances cannot propa-
existing methods for predicting choke behavior in two- gate upstream. Therefore, decreasing the downstream
phase flow. pressure does not increase the flow rate. If the
downstream pressure is gradually increased, there will
INTRODUCTION be no change in either the flow rate or the upstream
pressure until the critical-subcritical flow boundary
Chokes are widely used in the petroleum industry to is reached. If the downstream pressure is increased
protect surface processing equipment from slugging, to slightly beyond the boundary conditions, both flow
control flow rates from wells, to provide the rate and upstream pressure are affected. The veloci-
necessary backpressure to a reservoir to avoid for- ties of fluids passing through the choke drop below
mation damage from excessive drawdown, to maintain the sonic velocity of the upstream fluids. Here, the
stable pressure downstream from the choke and dampen flow rate depends on the pressure differential and
large pressure fluctuations. changes in the downstream pressure affect the upstream
pressure. This behavior characterizes subcritical
Either critical or subcritical flow may exist. Since flow.
different methods apply for predicting choke behavior
in these regimes, the prediction of the critical- Although it is often desirable to operate wells under
subcritical flow boundary is also important. The critical flow conditions with ~niform flow rate and
majority of correlations available apply to critical downstream pressure, Fortunati reports that a
flow only. Pressure drops through chokes can be majority of wells in the field operate under subcriti-
substantial. For example, in critical flow the cal conditions. However, most of the correlations
pressure downstream from the choke may be as low as available to petroleum industry are for critical flow.
50% or even 5% of the upstream pressure. Modern tech-
niques, like Nodal* Analysis, of analyzing the entire Existing Methods A complete model for two-phase flow
production system require two-phase models of com- through chokes should define the boundary between the
parable accuracy for each system component. Thus, to critical and subcritical flow regimes and predict the
optimize the performance of the entire production functional relationships of flow rate through the
system, an improved two-phase choke model is required. choke and the pressure differenial across the choke
for a given set of fluid properties and flow con-
*Nodal Analysis is a trademark of Flopetrol-Johnston- ditions. Most existing methods model critical flow
Schlumberger only and a few even attempt to define the critical-
subcritical flow boundary. These models are surveyed.
References and illustrations at end of paper.
2 TWO-PHASE FLOW THROUGH CHOKES SPE 15657
Tangeren
19
did the first significant study on two- conditions recommends CD = 0.85 for liquid and
0.90 for gas in absence of prior knowledge of Co.
en
phase flow through restrictions. He assumed polytro-
pic expansion of gas that is dispersed uniformly in
the mixture having liquid as the gontinuous phase. He EXPERIMENTAL PROGRAM
1
studied only critical flow. Ros extended Tangeren's
work by assuming liquid phase is homogeneously Two-phase data were gathered for critical, boundary
dispersed as droplets in a continous gas phase. He and subcritical regions for five choke sizes: 16, 20,
showed that accelerational pressure drop completely 24, 28 and 32 sixty-fourths of an inch (6.35 mm to
dominates choke behavior and slippage effects are 12.7 mm). Kerosene and water were used to cover the
negligible. approximate range of liquid densities encountered in
the field. The gas was air.
7. . 10
Fortunati , draW1ng on Guzhov and Medv1ediev , deve-
loped correlations for both critical and sub critical Test Facility Fig. 1 is a schematic diagram of the
flow and the boundary between these regimes. His test facility. A two-stage compressoj supplies air at
model is valid if downstream pressure exceeds 1.5 a maximum rate of 0.6 MMscf/D (708 m /h) at gauge
atmospheres (152 kPa). His correlation relates the pressure of 120 psi (827 kPa). Air is metered by 2
upstream and downstream pressure to a mixture velocity in. (51 mm) or 4 in. (102 mm) orifice meters.
at 19.8 psia (136 kPa). Fluid properties are calcu- Kerose~e and water are stored in separate 2000 gallon
lated at downstream conditions. (320 m ) tanks. Liquid is provided by a centrifugal
pump with a capacity of 200 GPM (12.6 dm 3 Is) at a
2
gauge pressure of 125 psi (862 kPa). A 4 in. (102 mm)
Ashford developed a model for two-phas critical r6
choke flow by extending the work of Ros • He assumed orifice meter measures liquid rates. For low liquid
conditions and a critical-subcritical boundary defini- rates, a rotameter is used. To ensure fully developed
tion similar to Tangeren's. A similar approach for two-phase flow, the gas-liquid mixing tee is placed
subsurface ~afety v~lves was also presented by Ashford 200 ft (60 m) upstream from the choke.
and Pierce. Gould plotted the Ashford boundary,
showing that different values of the polytropic expo- Figure 2 is a section of the choke. The choke is
nent yield different boundaries. installed horizontally to eliminate the perturbing
effects of elbows or choke housing. The two-phase
In addition to these theoretical approaches, numerous mixture flows through the choke and then into the
empirical methods also exist to predict choke beha- separator kept at low working gauge pressure (8 psi
vior. A popular form of correlations for critical (55 kPa» to ensure that the separated liquid automa-
flow is of the form: tically drains into the storage tank.

Temperatures are measured directly by thermometers.

, B
A qL R Barton recorders monitor gas and liquid rates through
P • • • • • • • • • • • • • • • • • • • • • (1 ) orifice meters and the rotameter is read directly •
Calibrated pressure gauges help maintain a rough check
on the pressure recording system shown in Fig. 3.
Pressure upstream from the choke is obtained from a
4 1 17 13
Baxendell , Achong , Ros and Pilehvari have deve- single pressure tap. Seven pressure taps downstream
loped varia~ions of this equation originally proposed from the choke ensure that fully recovered pressures
by Gilbert. Table 1 shows the coefficients A, Band are recorded. Pressures from the test pipe are
C suggested by various investigators. Note that the transmitted to a transducer manifold by 1/4 in. (6.4
critical flow rate is independent of the downstream mm) 200 psi (1.4 MPa) flexible tubing. At any time,
pressure which, therefore, does not figure in the only one of the three transducers (maximum gauge
Gilbert-type equations. Also,' fluid properties are pressure readings 32 psi (221 kPa), 75 psi (517 kPa)
neglected. and 125 psi (862 kPa» is used. Each transducer is
connected to a Validyne demodulator for shaping the
15 output signal. These outputs are transmitted to a
Poettmann and Beck used the work of Ros to construct
curves for 20°, 30° and 40° API crudes. This work is strip-chart recorder via a three-way electrical switch
also limited to critical flow and1ihe presence of to select readings from the appropriate transducer. A
water invalidates results. Omana , on the basis of filter consisting of a capacitor and a variable
47 two-phase tests on choke diameters from 4/64 in. resistor connected in parallel smooths the signal into
(1.59 mm) to 14/64 in. (5.56 mm), arrived at a corre- the recorder so that time-averaged pressure values are
lation valid for sm~ll choke sizes and flow rates less recorded. Care is taken so that all the recording
than 800 BPD (127 m~D). He arbitarily deemed the flow system components are calibrated properly.
to be critical when the ratio of downstream to
upstream pressure is less than 0.546 and when super- Test Procedure Initially, the system is stabilized
ficial gas velocity exceeds superficial liquid velo- with the valve downstream from the choke fully open
city. under two-phase conditions. Flow is considered stable
if both average pressures and flow rates are stable.
Of the above methods, only those of Ashford and Flow is usually critical at this stage since
Pierce, Fortunati and Pilehvari can be used for increasing the downstream pressure and allowing the
subcritical two-phase choke flow. Although the system to stabilise does not affect either the
methods of Ashford and Fortunati alone are derived upstream pressure or the flow rate. The valve
from basic principles, they both recommended discharge downstream from the choke is slowly closed and system
coefficients (Cn ) exceeding unity. This violates the stabilized each time until the critical-subcritical
first law of thermodynamics. For comparison, 5he flow boundary is reached. The flow condition is
value of en for liquid flow through an orifice is deemed to be at the boundary when a very slight
approximately 0.62. In addition, the API 14 RP method increase in downstream pressure produces a slight
for subsurfae safety valve performance in two-phase change in the average upstream pressure and a slight
SPE 15657 R. SACHDEVA. Z. SCHMIDT. J. P. BRILL & R. N. BLAIS 3
decrease in flow rates. Further closure of the valve meters being equal, the critical pressure ratio for
yields subcritical flow data. Each run typically PI = 800 psi (5.52 MPa) 1S lower than that for PI =
yields two critical. one boundary. and two subcritical 400 psi (2.76 MPa) or PI = 80 psi (552 kPa). This
data points.11Fig. 4 shows the location of the data on agrees with predictions of the model as evidenced by
the Mandhane flow pattern map. The orifice meters curves A, Band C. Similarly. if temperature were to
and the pressure recording system are properly increase while other parameters remained constant, gas
calibrated and operated. density would decrease. This would thus cause higher
critical pressure ratios to exist as shown by com-
Data Summary paring curve D with curve A.

Choke sizes: 16/64. 20/64. 24/64, 28/64 and 32/64 in. Changing the specific gravity of the liquid phase
(6.35 mm to 12.7 mm) while holding other parameters constant yields a pre-
Test fluids: Air-kerosene and air-wate~ mixtures dicted boundary given by curve E. Comparing curve E
Maximum liquid rate: 1340 bbl/D (213 m ~D) with A shows that liquid density hardly effects the
Maximum gas rate: 136.6 Mscf/D (161.2 m /hr) critical-subcritical boundary.
Maximum upstream pressure: 105.5 psia (700 kPa)
Number of critical points: 223 Note that at x = 1 (pure gas case). the critical
Number of subcritical points: 220 pressu:e satio equals that obtained by the
Number of boundary points: 110 folloW1ng well known boundary equation for an all gas
case:
PRESENT MODEL
k
Equations for conservation of mass. momentum and
energy for two-phase flow through chokes assume: (_2_)k-l ••••••••••••••••• (3)
k+1
• flow is one dimensional
• phase velocities are equal at the throat
• the predominant pressure term is accelerational Also note that after about x = 0.7. the boundary cur-
• the quality is constant for high speed process ves dip downward because of high values of qualities
• the liquid phase is incompressible (Xl) being used in Eq (2). Such high qualities are
seldom encountered in practice.
The final equations are capable of both finding the
boundary between critical-subcritical two-phase flow Critical and Subcritical Flow Rates Once the boundary
and calculating the flow rate through the choke for has been determined, flow rates through the choke are
critical and subcritical flow. Model development obtained from the following equation:
appears in the Appendix.

Boundary The critical-subcritical boundary is obtained

by iterating and converging on y from the following G
equation: 2

O-x ) V (l-y) k
k l L
k-l + k=l
Xl VGl
y where.
k n nO-Xl )VL +. !! CI-Xl)vg 2
+ 2'+ x V M +
k-l
I G2
2 x V
I G2 G
G2 ~2
2 •••••••••••••••••••• (5)
A
c
••••••••••••••••••• (2)
1
-'k
V V y ••••••••••••••••••••• (6)
Note that the converged value of y from the above G2 Gl
equation is the critical pressure ratio (Yc). A y >
Yc implies subcritical flow whereas y ~ Yc indicates and,
critical flow.

Figure 5 shows the critical-subcritical boundary under ••••••••• (7 )

various conditions for a hypothetical case. For
illustration, curve A is used as a reference. Curve A
is plotted using the following values:
If Yc ~ Yactual (Yc being obtained from Eq(2», criti-
k = 1.4. Sp.Gr. (liq) = 0.9. PI = 80 psi (552 kPa). cal flow exists. and the appropriate value of y is
T1 = 100°F (37.8°C). Cp =0.24, ~ = 0.8. given by y = Yc. If Yactual > Yc' sub critical flow
exists and y = Yactual should be used in Eq (4). Once
The effect of increasing upstream pressure (PI) is G2 is known. liquid and gas rates can be easily calcu-
shown by curves Band C. As PI increases, gas becomes lated since Xl' PI' and the area of choke are known.
more dense. and the sonic velocity of the mixture
increases. Thus higher flow rates through the choke For the sake of simplicity, no effort was made to
are needed to attain critical flow. This necessitates distinguish between free and dissolved gas and a mix-
larger pressure differentials. which are reflected by ture of liquids. In this discussion. the terms gas
lower critical pressure ratios. Thus. all other para- and quality pertain to free gas. Also. if more than
4 TWO-PHASE FLOW THROUGH CHOKES SPE 15657
one liquid is present in the two-phase mixture, the Because of added turbulence caused by the elbow, the
common method of using the average weighted properties value of CD would be lower than 0.85, since the flow
applies. Similarly, the value of liquid density used is less similar to what the model assumes. CD = 0.75
should account for effects of dissolved gas and the gave the best results. Note that more emphasis was
gas compressibility factor should be used to obtain placed on the absolute average error and standard
correct in situ volume of the free gas. deviation than on the average error since the former
two are better parameters for error analysis. The
EVALUATION OF PRESENT MODEL results of the analysis are shown in Table 5 for two-
phase data comprising a total of 630 points. This
As with any model, the best test of the hypotheses table shows that with CD = 0.75, the present model
involves the application of the model to empirical yields substantially better results with Pilehvari's
data. This was done using: data for both critical and subcritical flow than his
own correlations gave.
• the data from the authors' experimental program,
which used air-kerosene and air-water flows: a Model Analysis with Ashford's data Ashford collected
total of 553 two-phase test points. field data and his choke configuration is similar to
• the two-phase air-water measurements reported by the one used by Pilehvari. Again, the present model,
Pilehvari: a total of 189 critical and 441 subcri- with Co = 0.75 gives substantially better results
tical data points. (Table 6) than Ashford had obtained using his
• the 27 two-phase field data points published by suggested value of CD = 1.
Ashford.
Note that both Pilehvari and Ashford had a different
Major existing models were also evaluated against the choke configuration from that used in the present
data gathered. study. Their chokes were mounted with an elbow
upstream. The present model yields good results with
Discharge coefficient selection For fluid flow through CD = 0.85 for a choke free of upstream disturbances
restrict~ons, ~t ~s common to use discharge coef- and with Co = 0.75 for a choke configuration involving
ficients (CD) as a final modifying factor in the flow- an elbow upstream from the choke. Choke configuration
rate equation. It is usually hoped that its use would thus seems to play a definite role on the value of CD.
absorb errors due to assumptions made while developing The analysis above leads to the following recommen-
a model. Thus, the values of Co depend on the assump- dations on the use of the present model:
tions made during model development. A "perfect"
model will have CD = 1.0, i.e., the use of Co will be • use CD = 0.75 when the choke is installed in a
redundant if all flow processes are accurately housing (as is common in the field),
accounted for. Laws of Thermodynamics imply that • use CD = 0.85 for a choke where the flow is
values of CD less than unity should result. Often, undisturbed by the effects of an elbow immediately
CD is also correlated with factors such as Reynolds upstream from it.
number, pressure differential, gas expansion factor
etc. Obviously, if a model adequately accounts for Model Application to Sub-Surface safet* Valves The
such variables, such dependance should not exist. The model was derived for two-phase flow t rough a pipe
present model was developed for one-dimensional, two- restriction. The term restriction implies that flow
phase flow through a restriction. Choke installations velocities are substantially increased in the restric-
involve elbows upstream from the choke. Effects of tion such that the accelerational pressure drop term
presence of these fittings that perturb the flow imm- dominates. Usually, the ratio of diameter of the
diately upstream from the choke were not accounted for safety valve to the tubing is above 0.8 and thus the
during model development. As will be seen later, the use of the model will not be correct since the acce-
manner in which a choke is installed becomes the only lerational pressure drop term will not be predominant.
factor on which the CD of the present model depends. However, should a ratio of 0.5 or less exist (lesser
Two values of CD are suggested depending on the choke ratios increasing the model's validity) this method
installation. These values are constants, and, based may be used. A CD of 0.85 should be used since the
on the results from error analysis, no further corre- valve configuration will be similar to the choke con-
lation for Co (in terms of other parameters) was figuration in the experimental program of this study.
necessary.
CONCLUSIONS
Evaluation of existing models and proposed model with
data gathered As mentioned earlier, it was hoped that (i) The proposed model performs better than the
the use of a discharge coefficient (CD) would absorb existing models in predicting two-phase flow
errors resulting from various assumptions, especially through chokes.
that flow is one-dimensional. A value of Co = 0.85
yields optimal answers when applied to data gathered (ii) For chokes installed in a housing at the
with the choke free of effect of the upstream elbow. wellhead, where effects of elbow (or choke
Table 4 shows the present model, when use with a CD housing) are present, a discharge coefficient
0.85, gave better results than other models (Tables 2 (CD) of 0.75 should be used. Such con-
& 3). This value is recommended whenever the choke figurations are common in field practice. For
configuration used is similar to the one in the test chokes where such effects of choke housing are
facility, i.e., whenever the flow is undisturbed by not present, CD = 0.85 is recommended.
elbow upstream from the choke.
(iii) Since the model is derived from dynamic prin-
Model Analysis with Pilehvari's data Pilehvari's ciples, and was tested sucessfully with three
choke was installed in a choke housing with elbow different sets of data, it is expected to work
upstream from the choke. The assumption of unper- well in the field.
turbed, one-dimensional flow is less valid here.
SPE15657 R. SACHDEVA Z. SCHMIDT J. P. BRILL & R. N. RUT<:

ACKNOWLEDGEMENTS Subscripts

The authors would like to thank the University of 1 Upstream conditions

Tulsa and the Tulsa University Fluid Flow Projects
(TUFFP) for support. Helpful suggestions from Dr. O. 2 Downstream conditions
Shoham and Dr. B.M. Kelkar of The University of Tulsa
are sincerely appreciated. c Choke, critical

NOMENCLATURE G Gas

Description L Liquid

m Mixture
A Area (ft 2 ), coefficient in Eq. (1)
p Producing, pipe
B Coefficient in Eq. (1)
s Superficial
C Coefficient in Eq. (1)
REFERENCES
Discharge coefficient
1. Achong, I.: "Revised Bean Performance Formula
Specific heat of liquid for Lake Maracaibo Wells", Internal Report
(October, 1961).
Specific heat of gas at constant pressure
2. Ashford, F.E: "An Evaluation of Critical
Specific heat of gas at constant volume Multiphase Flow Performance Through Well Head
Chokes", J. Pet. Tech., (Aug. 1974) pp. 863-850.
d Diameter, ft
3. Ashford F.E., and Pierce, P.E.: "The
Choke diameter, 1/64ths of in. Determination of Multiphase Pressure Drops and
Flow Capacities in Downhole Safey Valves (Storm
Percent Error chokes)", SPE 5161, Presented at SPE Annual Fall
Meeting, Houston, Texas, (Oct. 1976).
E Average pecent error
4. Baxendell, P.B.: "Bean Performance-Lake Wells",
Absolute average percent error Internal Report (Oct. 1957).

Gravitational constant 5. Bird, B.B., Stewart, W.E. and Lightfort, E.N.:

Transport Phenomena, Published by J. Wiley &
2
G Mass flux, lbm/ft /sec Sons, New York (1960), Page 482.

k Ratio of specific heats, C /C 6. Brill, J.P. and Beggs, H.D.: Two-Phase Flow in
p v Pipes, Tulsa, OK (1981).
L Length of choke, ft
c 7. Fortunati, F.: "Two-Phase Flow Through Wellhead
M Mass flow rate, Ibm/sec Chokes", SPE 3742, presented at SPE European
Meeting, (1972).
n Polytropic exponent for gas
8. Gilbert, W.E.: "Flowing and Gas-Lift Well
P Pressure, psia Performance", API D. and P.P. (l954) , 126.

q' Volumertric flow rate at standard conditions, 9. Gould, T.L.: Discussion of paper "An Evaluation
STB/D of Critical Multiphase Flow Performance Through
Wellhead Chokes", by Ashford, F.E., J. Pet.
SD Standard deviation from average percent error Tech., (Aug. 1976), 843.

v Velocity, ft/sec 10. Guzhov, A.J. and Medeiediev, V.F.: "Critical

Flow of Two-Phase Fluids Through Wellhead
3
V Specific volume ft /lbm Chokes", Nieftianoie Xoziastva-Moskva No. 11,
(1962), (in Russian).
x Free gas quality
11. Mandhane, J.M., Gregory, G.A. and Aziz, K.: "A
y Downstream to upstream pressure ratio Flow Pattern Map for Gas-Liquid Flow in
Horizontal Pipes", Intern. J. of Multiphase Flow,
1., pp. 537-553, (1974).
Greek Symbols
12. Omana, R. et al: "Multiphase Flow Through
a Void fraction Chokes", SPE 2682, Presented at 44th Annual Fall
Meeting, Colorado, (1969).
3
p Density, lbm/ft

T Wall shear stress

w
6 TWO-PHASE FLOW THROUGH CHOKES SPE 15657
13. Pilehvari. A. A.: "Experimental Study of Critical The phase continuity equations are:
Two-phase Flow Through Wellhead Chokes." U. of
Tulsa Fluid Flow Projects Report. (June 1981).
• •••••••••••• (A.3)
14. Pilehvari. A.A.: "Experimental Study of
Subcritical Two-Phase Flow Through Wellhead
Chokes." U. of Tulsa Fluid Flow Projects Report. M =A v P • •••••••••••• (A. 4)
G G G G
(Sept. 1980).

15. Poettmann. F.H. and Beck. R.L.: "New Charts and

Developed to Predict Gas-Liquid Flow Through
Chokes." World Oil. (March 1963). 95-101. • •••••••••••• (A. 5)

16. Ros. N.C.J.: "An Analysis of Critical Also.

Simultaneous Gas-Liquid Flow Through a
Restriction and its Application To Flow O-x ) G
2 2
Metering." Applied Sci. Research (960). !. 0-<:1 ) P
••••••• (A.6)
Section A. 374. 2 L2

17. Ros. N. C. J.: "Theoretical Approach to the Study

of Critical Gas-Liquid Flow Through Beans." since. G2 = (~2 + ML2)/A2 •
Internal Report. (Feb. 1959).
and similarly.
18. Sachdeva. R.: "Two-Phase Flow Through Chokes."
M.S. Thesis. U. of Tulsa. 1984.
• •••••••• (A. 7 )
19. Tangeren. R.E. et al: "Compressibility Effects of
Two-Phase Flow." J. App. Phys •• 20. number 7.
(949). Elimination of <:12 between Eqs. (A.6) and (A.7) and
assuming vL2 = vG2 = v2 gives:
20. Wallis. G.B.: One Dimensional Two-Phase Flow.
McGraw Hill. (1969).

APPENDIX A • •••••••••••• (A. 8)

MODEL DEVELOPMENT
where.
Critical-Subcritical Boundary
.!..... = ~ + (I-x) ••••••••••• (A. 9)
Equations describing conservation of mass. momentum Pm PG PL
and energy were used to determine relationships for
critical and subcritical flow and the boundary between
them. For horizontal two-phase separated flow. the Expanding Eq. (A.2) yields:
following momentum equation at the throat can be writ-
ten:
•••••• (A.10)

••.••••••••••••• • (A. 1) For a fixed set of upstream conditions. during criti-

cal flow. the mass flux reaches a maximum with respect
Ros has shown that in two-phase flow through chokes. to the downstream (throat) pressure.
the acceleration term predominates and that the wall
shear forces can be safely neglected. He also showed
Since G2 = (~2 + ML2 )!Ac ' the boundary can be defined
that there is practically no slippage at the throat as:
and hence we will assume equal phasic velocities at
the throat: (vL2 = vG2 = v2)' Hence Eq. (A.1) redu-
ces to:
o ••••••••••••• (A.11)

During critical flow. conditions described by Eq.

(A. 11) hold; hence Eq. (A.10) reduces to:

2 dV m2
- 144 gc = G2 ~ (from Eqs. (A.8) and (A. 11»
2
••••••••••••••• (A.12)

••••••• (A. 2)
SPE 15657 R. SACHDEVA. Z. SCHMIDT. J. P. BRILL & R. N. BLAIS 7
Diffferentiating Eq. (A.9) we have: Usually, v2 2 » vI 2 (since d c = d2 « dpip ) for a
choke. Noting also that v2 = G2 /Pm2' Eq. tA.2I) redu-
ces to:
••••• (A.I3)

The liquid phase can be assumed to be incompressible:

••••••••••••• (A.I4)
•••••••••••••• (A.22)
Typical velocities of mixtures flowing through chokes
are high (approximately 50-150 ft/s). Thus, there is
virtually no time for mass transfer at the throat. Elimination of G2 2 between Eqs. (A.I9) and (A.22)
Thus, we have: yields:
••••••••••••••••• (A.I5)

Combining Eqs. (A.I3), (A.I4) and (A.I5) gives:

••••••• (A.I6)
•••••••••••• (A.23)

During gas expansion at the throat, a temperature gra-

dient exists between the phases, resulting in fast
heat transfer between them. We also know that this Denoting,
process is in-between the extremes of isothermal and
adiabatic processes (for which the value of n would
respectively be unity and ep/~). Thus, the heat flow •••••••••••••••••• (A. 24)
in the gas-liquid mixture is approximated by a
polytropic process such as
and rearranging Eq. (A.23) yields:

( c=constant) ••••••••• (A.I7)

k (I-xl) V O-y) k
L
--+
k-I
k=I
where, the polytropic exponent given by Ros is xl VGI
y
k n

k-I
+ -2 +
n = 1 •••••••••••• (A.I8)
••••••••••••••• (A. 25)

Equations (A.I6) and (A.I7) give: While developing Eq. (A.25), we had assumed in Eq.
(A.II) that the flow condition was such that the mass
flux did not change with respect to the downstream
•••••••••••.••••••.•• (A.19) pressure (P2)' that is, flow was at the critical-
subcritical boundary. Thus the downstream-upstream
pressure ratio at the critical-subcritical boundary
Now, proceeding with the previous assumptions of hori- (Yc) can be obtained from Eq. (A.25) by solving for y
zontal flow, negligible friction and equal phase velo- iteratively. Once upstream conditions are known, Eq.
cities, the energy equation reduces to: (A.25) can be solved using
2
dG J ••••••••••••••• (A.20)
•••••••••••• (A.26)
or,
2
d (v 2) Let the value of y from Eq. (A.25) be Yc' This is the
••••••• (A.2I)
, critical pressure ratio that determines the boundary.
If actual conditions are such that Yactual > Yc' then
the flow is subcritical and if Yactual ~ Yc' the flow
We will integrate Eq. (A.2I) between PI and P2 , noting is critical.
that liquid density and gas quality remain constant,
and that the gas expansion is adiabatic.
8 TWO-PHASE FLOW THROUGH CHOKES SPE 15657
Flow Rate Calculation
Critical Flow: Incorporating a discharge coefficient
and rearranging Eq. (A.22) we get:

where,

1
V Y-
Gl
k ••••••••••• (A.28)

and

1
_1_ = xl VGl Y k + (I-xl) VL •••••••••••• (A.29)
Pm2

The first step is to determine Yc from Eq. (A.25). If

Yactual i Yc ' the flow is critical. Thus Y =
Yc should be used in Eq. (A.27) to calculate the cri-
tical flow rate.

Subcritical Flow: While deriving Eq. (A.27), no

assumption was made as to the nature of the flow.
Again, the first step is to determine Yc from Eq.
(A.25). If actual conditions are such that Yactual >
Yc ' the flow is subcritical. For subcritical flow, Y
= Yactual should be used in Eq. (A.27) to determine
the subcritical flow rates.

Discharge Coefficient (CD)

As explained in the paper, CD = 0.75 should be used

when a choke is installed in a housing i.e., has an
elbow upstream from it (as is common in the field).
If the choke is installed such that there are no flow-
perturbing effects due to an ,elbow immediately
upstream from it, then, Co = 0.85 should be used.

SI Metric Conversion Factors

atm x 1.013 25* E+o2 = qa

bbl x 1. 589 873 E-ol = m
feet x 3.048* E-ol m
OF (OF-32)/1.8 E+OO o~
gal x 3.785 412 E-03 m
in x 2.54* E+OO em
lbm x 4.535 924 E-ol kg
psi x 6.894 757 E+OO kPa
*Conversion factor is exact
 TABLE 1

Coefficients for the Gilbert-type Equation

for TWo-Phase Critical Flow

Correlation A B C

Gilbert 10.00 0.546 1.84

Ros 17.40 0.500 2.00

Baxendell 9.56 0.546 1.93

Achong 3.82 0.650 1.88

Pilehvari 46.67 0.313 2.11

TABLE 2

Evaluation of Correlations Using Air-Water Data

c critical Standard Average Absolute

s subcritical deviation error average
(%) (%) error (%)
Correlation

Gilbert (c) 22.2 -71.7 71.7

Ros ( c) 13.8 -25.2 26.2
Achong ( c) 19.0 -5.8 17.0
Pilehvari (c) 16.5 -18.5 22.2
Ashford ( c) 7.7 36.7 36.7
Omana (c) 6.4 59.2 59.2
Pilehvari (s) 40.9 -16.7 24.7

TABLE 3

Evaluation of Correlations Using Air-Kerosene Data

c = critical Standard Average Absolute

s subcritical deviation error average
(%) (%) error (%)
Correlation

Gilbert (c) 15.5 15.9 17 .5

Ros (c) 11.6 43.6 43.3
Achong ( c) 10.1 48.3 48.3
Pilehvari (c) 14.0 18.4 19.6
Ashford (c) 25.1 -25.8 27.6
Omana (c) 7.5 52.1 52.1
Pilehvari (s) 31.1 -5.6 20.6
 TABLE 4

Analysis of Data with Model Developed

Flow Type Fluids Standard Average Average

Deviation Error Absolute
(%) (%) Error
(%)

Critical Air-Water 6.5 -13.7 14.7

Critical Air-Kerosene 13.7 -5.6 14.9
Subcritical Air-Water 8.3 5.2 8.4
Subcritical Air-Kerosene 9.1 2.2 9.1
Boundary Air-Water 10.1 14.1 14.9
Boundary Air-Kerosene 13.4 8.9 15.0

TABLE 5

Analysis of Pilehvari Data

Average Average Standard Remarks

Error Absolute Deviation
(%) Error (%) (%)

Pilehvari Correlation 5.2 19.6 24.7 Critical

data,
Present Method 9.6 U.S 10.8 189
(CD = 0.75) points

Pilehvari Correlation 3.0 15.5 19.3 Sub-critical

data,
Present Method 0.3 8.0 12.8 441
(CD = 0.75) points

TABLE 6

Analysis of Ashford Data

Average Average Standard Remarks

Error Absolute Deviation
(%) Error (%) (%)

Ashford Method 1.1 14.6 23.2 27 points,

critical
Present Method 12.5 15.1 13.6 flow
(CD = 0.75)
 41N. LINE

AIR FROM
COMPRESSOR 2 IN. LINE
PUMP BYPASS
MIXING t -__..2
_ IN_._L_IN_E.,
TO TANK 41N. LINE TEE
LIQUID
FROM
STORAGE 21N. LINE
TANK TWO-PHASE
FLOW
LIQUID TO DEVELOPING
AIR PRESSURE
STORAGE
TANK VENTED ,---------- .... GAUGE

: i
I ~ ~ ~ aa~~..
1..--
CHOKE LEGEND

{fl\.-1I4 IN. PLASTIC

-/ I- ORIFICE METERS
'j-( TUBING TO RECORDING -{J- ROTAMETER
SYSTEM
Q GOULD PUMP
-(><J-VALVES

Fig. 1-Schematic diagram of test facility.

1-2/3 IN.
-I-.f--1-I/4 IN. CHOKE I.O·l
....I-~ I.. GIN.· ... 1
Fig. 2-Choke sec1lon.

TRANSDUCERS

UPSTREAM
PRESSURE
DOWNSTREAM
PRESSURE I I I
I I I

/oL;Jo
1/4 IN.
PLASTIC DEAD WT.
TUBING TESTER
L_ I J
DEMODULATORS ..... ........::i-ELECTRIC
WIRES

.....--....-~ RESISTANCE-CAPACITANCE
FILTER
STRIP CHART
RECORDER

Fig. 3-Pressure recording system.

 IO~

V SL ANNULAR.
ANNULAR
eFT/S) MIST
FLOW

~RATIFI~D FL~
0.1-

,\, I

0.010L.-
1
--_.L...._-_....10-......;;;;;;;................'---IO--loo
VSG (FT/S)

Fig. 4-Locatlon 01 data on mandhane map.

0.6

0.4
CRITICAL
PRESSURE
RATIO LIQ. TI PI
SP. GR (OF) (PSIA)
0.2 •
+ 8
A 0.9 80 100
0.9 400 100
l::1 C 0.9 800 100
0 0 0.9 80 200
0 E 0.8 80 100
OL.-------L. """- --''--- -I

0.0001 0.01 1.0

FREE GAS QUALITY
Fig. 5-Crltical pressure ratio YB. quality lor K =1.4.