Professional Documents
Culture Documents
Choking in 2-Phase Flow - Kelkar
Choking in 2-Phase Flow - Kelkar
Predictions of Two-Phase Critical Flow Boundary and Mass Flow Rate across
Chokes
E. M. Al-safran, SPE, Kuwait University, and M. Kelkar, SPE, The University of Tulsa
approaches, basically derived from mass, momentum reasonable accuracy and is considered, in this stage, a
and energy balances such as Sachdeva et al.1, qualitative tool only.
Perkins2 and Selmer-Olsen et al.9 The theoretical The above literature review on theoretical choke
models are mostly used by the industry because of studies demonstrates that modeling the slippage
their ability to simulate the physical phenomena, thus phenomenon across chokes improves the accuracy of
they are considered more accurate. The following is a mass flow rate predictions. Furthermore, the
brief description of these theoretical models. literature review shows the need for a simple
Sachdeva et al.1 acquired experimental data for theoretical, yet accurate slip model to predict the
critical, sub-critical flows and the boundary between critical/sub-critical flow boundary and the mass flow
them using air/water and air/kerosene system for rate which is the aim of this study.
different sizes of choke diameters ranging from 0.25
to 0.5 in. A theoretical model is developed from the Model
1D mass, momentum and energy balance equations The present slip model is developed on the basis of
for a two-phase gas liquid mixture to predict the Sachdeva et al.1 and Perkins2 models. The model
critical and sub-critical mass fluxes, which is accounts for conservation of mass, momentum and
considered comprehensive since fluid properties are energy for two-phase flow across restriction with the
also involved. Perkins2 presented theoretical following assumptions:
equations that describe isentropic multiphase flow Flow is one dimensional.
across a choke valid for critical sub-critical flows. Acceleration is the predominant pressure term.
The model incorporates the oil, gas and water
Quality is constant frozen flow.
properties correlations to predict the critical/sub-
Liquid phase is incompressible.
critical boundary. An expression for total mass flow
rate across the choke was developed through a Gas phase expands polytropically.
combined equation of conservation of mass and Slippage exists at choke throat.
isentropic expansion of a homogeneous multiphase The present slip model is capable of calculating
mixture. Selmer-Olsen et al.10 acquired experimental the critical/sub-critical flow boundary and the mass
data and developed the Hydro model which uses a flow rate for critical and sub-critical flow behaviors.
control volume approach for the choke orifice and its The complete model derivation is provided in the
downstream to mechanistically account for Appendix. In the Appendix, we also demonstrate
irreversible loss process instead of using discharge that our model can be reduced to either Perkins or
coefficient only. The model is derived from the local Sachadeva et al.1 models if some simplifying
cross-sectional averaged balance equations of mass, assumptions are made.
momentum and energy for steady-state flow of a
multiphase mixture. Contrary to Sachdeva et al.1 and Critical Flow Boundary. The final form of the
Perkins2 models, the Hydro model accounts for critical/sub-critical flow boundary equation is
slippage between the phases. Although the Hydro presented as follows.
model is more accurate, it is complex in formulation
(1 rc ) +
and solution for the boundary. n
In a recent comprehensive study, Schuller et al.11 n 1
(rc )
1 1
{ }
acquired experimental large scale oil/gas/water sub-
n = 2
(1)
n n 1
critical flow data across chokes where they measured + 1 + rc n
the mass flow rate and pressure drop across chokes. n 1 2
The slip Hydro model was evaluated against the Where
experimental data and compared with Sachdeva et
al.1 and Perkins2 models. Schuller et al.11 study xg kCvg + (1 xg ) CL
n=
xg Cvg + (1 xg ) CL
revealed the accuracy of their improved slip Hydro . (2)
model over the no-slip Sachdeva et al.1 and Perkins2
models in predicting the sub-critical mass flow rate.
In another resent study by Schuller et al.12, critical R(1 xg )vL
mass flow rate data were acquired from the same = (3)
experimental facility against which the three models xg vg1
were evaluated. The improved slip Hydro model
outperformed Sachdeva et al.1 and Perkins2 model in which rc=critical pressure ratio (p2/p1),
with high accuracy. On the other hand, Schuller et n=polytropic gas expansion exponent, k=gas specific
al.12 showed that the Hydro model is unable to heat ratio, Cv=gas specific heat at constant volume,
predict the critical/sub-critical flow boundary with a CL=liquid specific heat constant, xg= gas quality,
109243 3
vg=gas specific volume, vL=liquid specific volume value that is more reasonable for the sub-critical flow
and R=slip ratio (ug/uL). Eq. (1) is an implicit behavior. Grolmes and Leung20 slip model is the
equation that requires an iterative procedure to solve selected slip model for sub-critical flow behavior
for rc. which is presented as follows.
Critical and Sub-critical mass flow rate. The
complete derivation of the total mass flow rate ( a1 1) ( a 2 +1) a3
1 xg L L (6)
equation is presented in the Appendix. The final R = a0
x
form of the equation is as follows. g g g
in which =phase viscosity and the constants value of
n
n 1
a0=1, a1=1, a2=-0.83 and a3=0 are given by Simpson
CA22 p1 (1 r ) + 1 r n et al.21
n 1
.. (4) The analysis of different slip models in this study
m 2 =
x g v g1 r n + x g + (1 x g )
1 2
1 suggests that the selection of specific slip model
R should partially depend on the fluids viscosity ratio.
In the case of high liquid viscosity, the gas will flow
past the liquid droplets leading to high slippage
in which A2=choke cross sectional area, p1 upstream
between the two phases. This was tested for when
pressure and C is a constant and will change
the Schuller et al. oil water mixture (high viscosity)
depending on the units used (C=2000 CD for SI units
mass flow rate data is compared with water (low
and C=2*CD2*gc*144). In this study, a value of
viscosity) mass flow rate data, different slip model
CD=0.75 was found appropriate to calibrate for the
models were found appropriate for each case.
model imperfections and irreversible losses. In the
case of critical flow (r<rc), rc is used in Eq. (4) to
calculate the critical total mass flow rate. On the Model Evaluation
other hand, when flow is predicted as sub-critical The present slip model is evaluated on its predictions
(r>rc), r is used in Eq. (4) to calculate the sub-critical of critical/sub-critical flow boundary as well sub-
total mass flow rate. In this study and due to sub- critical mass flow rate. The evaluation results are
critical flow data availability, the model was presented below.
evaluated only for sub-critical flow. Critical/sub-critical flow boundary. There is a
Slip model. Due to high acceleration at choke throat, very limited published data on the critical flow
gas phase flows past liquid phase resulting in boundary. The present slip model is validated against
slippage between the two phases. To complete the Sachdeva et al.22 critical flow boundary (110 points)
present slip model an algebraic expression is required experimental data set. The present model predictions
for the slip ratio. Several studies provided an were compared with Sachdeva et al.1 and Perkins2
expression for the slip ratio with different models. Fig. 1 shows the present slip model critical
assumptions. These include Henry13, Tangren et flow boundary prediction as a function of upstream
al.14, Henry and Fauske15, Chishom16, McNeil and GLR against the critical and sub-critical flow data.
Morris17, Leung18,19, Grolmes and Leung20 and The model predictions separate the two flow
Schuller et al.11. In this study, several slip models behaviors very well identifying the boundary
were tested to improve the accuracy of the present between them.
slip model critical flow boundary predictions and
sub-critical mass flow rate. Among all, Schuller et 1.0
0.4
(3-4) which is reasonable for critical flow conditions.
Schuller et al.11 slip ratio expression is as follows. 0.3
0.2
0.1
[ ]
R = 1 + x g L 1 1 + 0.6e g . (5)
5.0 x 0.0
0.1 1 3
Insitu GLR (m /m )
3 10 100
mass flow rate predictions, a different slip model was Fig. 1- Comparison of predicted critical flow boundary
found more appropriate which provided lower slip with experimental data
4 109243
The present slip, Sachdeva et al.1 and Perkins2 Mass flow rate. Laboratory validation. A validation
models were compared with the experimental critical study is carried to test the model against sub-critical
flow boundary. Statistical error analysis was carried mass flow rate and compare it with Sachdeva et al.1
out to calculate the average percent error (E1), and Perkins2 predictions. In this validation study two
absolute average percent error (E2) and standard sub-critical data bases were used, namely Sachdeva
deviation (E3) for each model. The statistical et al.22 and Schuller et al.11 Fig. 2, 3 and 4 are cross
parameters are defined in Eqs. (8-10). The results of plots of the present slip, Sachdeva et al.1 and Perkins2
the error analysis are presented in Table 1. models, respectively.
1 rc rc i ( m )
E1 = i ( c )
2.5
x100 (8)
n r c i(m) 2
0.5
2
rc i ( c ) rc i ( m )
E1 0
rc i ( m )
0 0.5 1 1.5 2 2.5
(10)
n
E3 =
Measured mass flow rate (kg/s)
Perry22 as follows,
0.5
( p p ) .. (7)
"
p2' = p1
[1 (d d ) ]
1 2 0
1.85 0 0.5 1 1.5 2 2.5
c dp Measured mass flow rate (kg/s)
in which dc=choke diameter and ddp=downstream Fig. 4 Perkins predictions vs. measured
pipe diameter. In the case of boundary flow, p2=p2, mass flow rate
thus rc=p2/p1=p2/p1.
109243 5
0
Table 3: Models Error Analysis 0 5 10 15 20 25
Model E1 (%) E2 (%) E3 (%) Measured mass flow rate (kg/s)
Present slip model 2.7 9.7 12.5
Sachdeva et al. -10.7 13.4 14.1 Fig. 6 Sachdeva et al. predictions vs. field measured
mass flow rate
Perkins -17.6 18.8 14.0
5
Field validation. A field validation study is carried
out in this study to test and compare the new model 0
slightly lower standard deviation and abosolute error E1 = Average percent error (%)
than the present model. E2 = Absolute average perecent error
The lab and field verification study above shows (%)
that, overall, the proposed new model outperforms E3 = Standard deviation (%)
both the Sachadeva et al. and Perkins models. k = Heat capacity ratio = Cp/Cv (-)
Sachdeva et al.1 model perform better in sub-critical m = mass flow rate (kg/s)
condition (lab validation) than in the sub-critical n = Polytropic gas expansion
conditions (field validation). Conversely, the present coefficient (-) or number of data
slip and Perkins2 models are more accurate in points
predicting the critical mass flow rate (field p = Pressure (Pa)
validation) than the sub-critical mass flow rate. This p = Pressure just downstream of choke
result illustrates that the present model is the most (Pa)
accurate model in both cases of critical and sub- P = Recovered downstream pressure
critical. (Pa)
rc = Critical pressure ratio = p2/p1 (-)
Conclusions R = Slip ratio (-)
Several conclusions can be derived based on this v = Specific volume = 1/ (m3/kg)
study as follows: x = Quality
A new model accounting for slip between the Greek letters
two phases is derived on the basis of Sachdeva et = Viscosity (Pa.s)
al. and Perkins models of 1D balance equation of = Density (kg/m3)
mass, momentum and energy with the
assumptions of constant quality and Subscripts
incompressible liquid phase. c = choke, calculated, critical
The present slip model is capable of predicting g
i
=
=
Gas
elimental data point
the critical/sub-critical flow boundary and the
mass flow rate of critical and sub-critical flow L = Liquid
behaviors. m = Mixture, measured
dp = Downstream Pipe
A validation study demonstrated the capability of 1 = Upstream
the new model to predict the critical/sub-critical 2 = At choke throat
flow boundary with average error of 5.2% and
standard deviation of 15.5% outperforming Acknowledgement
Sachdeva et al. and Perkins models. The first author wishes to express his thanks to
A validation study on sub-critical mass flow rate Kuwait University for supporting this research. Both
showed that the present slip model accurately authors acknowledge the University of Tulsa and
predicted the mass flow rate with average error Tulsa University Fluid Flow Projects for making the
of 2.7% and standard deviation of 12.5% experimental data available. The authors also thank
outperforming Sachdeva et al.1 and Perkins2 Dr. Ciro Perez for checking the accurary of the
models. equations.
Slippage phenomenon between the two phases at
References
the choke entrance and throat is an important
parameter that should be accounted for in
modeling flow across the choke. Furthermore, 1. Sachdeva, R., Schmidt, Z., Brill, J. P., and Blais,
the selection of slip model depends on the type R.: Two-Phase Flow Through Chokes, SPE
of flow, quality range as well as fluids viscosity 15657 presented at the 1986 SPE Annual
ratio. Technical Conference and Exhibition, New
Orleans, 5-8 October.
Nomenclature
A = Cross sectional area (m2) 2. Perkins, T. K.: Critical and Sub-Critical Flow
C = Heat capacity (KJ/kg/K) of Multiphase Mixtures Through Chokes, SPE
CD = Discharge coefficient 20633, SPE Drilling and Completion (December
Cv = Heat capacity at constant volume 1993), 271.
(KJ/kg/K)
d = Diameter (m)
109243 7
3. Gilbert, W. E.: Flowing and Gas-Lift Well 14. Tangren, r., Dodge, C, and Seifert, H.:
Performance, Drilling and Production Prac. Compressibility Effects in Two-Phase Flow, J.
(1954) 126. Applied Physics, 1949, vol. 20, No. 7, 637-645.
4. Ros, N. C. J.: An Analysis of Critical 15. Henry, R. and Fauske, H.: The Two-Phase
Simultaneous Gas-Liquid Flow Through a Critical Flow of two-Component Mixtures in
Restriction and its Application to Flow Nozzles, Orifices and Short Tubes, ASME J.
Metering, Appl. Sci. Res. (1960) 9. Heat Transfer, vol. 93, May 1971, 179-187.
5. Achong, I.: Revised Bean Performance Formula 16. Chisholm, D.: Two-Phase Flow in Pipelines and
for Lake Maracaibo Wells, Internal Company Heat Exchangers, George Godwin (Longman
Report, Shell Oil Company, Houston, TX Group Ltd.) and IChemE, London (1983).
(October 1961).
17. McNeil, D and Morris, S.: Simple Explicit
6. Pilehavri, A.A.: Experimental Study of Critical Method for Estimating Gas/Liquid Flow
Two-Phase Flow Through Wellhead Chokes, Conditions in Pipeline Restrictions,
M.Sc. Thesis, U. of Tulsa, Tulsa OK 1981. Proceedings of 2nd UK National Heat Transfer
Conference, Glasgow 1988, vol. 2, 1234-1256.
7. Ashford, F.E. and Pierce, P.E.: Determining
Multiphase Pressure Drop and Flow Capacities 18. Leung, J.: A Generalized Correlation for One-
in Downhole Safety Valves, JPT (September component homogenous Equilibrium Flashing
1975) 1145. Coked Flow, AIChE Journal, vol 32, No. 10,
1986, 1743.
8. Osman, M. E. and Dokla, M. E.: Gas
Condensate Flow through Chokes, SPE 20988, 19. Leung, J.: Size Safety Relief Valves for
Richardson, TX (1990). Flashing Liquids, Chemical Engineering
Progress, February 1992, 70-75 (Corrections in
9. Omana, R., Houssiere, C., Brown, K., Brill, J., March 1992, 108)
and Thompson, R.: Multiphase Flow Through
Chokes, paper SPE 2682 presented at the 1969 20. Grolmes, M. and Leung, J.: Chemical
SPE Annual Fall Meeting, Denver, Colorado 28 Engineering Progress, 1985, vol 81, No. 8, 47.
September-1 October, 1969.
21. Simpson, H., Rooney, d., and Grattan, E.: Two-
10. Selmer-Oslen, S., Holm, H., Haugen, K., Nilsen, phase Flow through Gate Valves and Orifice
P., and Sandbegr, R.: Subsea Choke Flow Plates, paper presented at the Int. conference on
Characteristics, paper presented in BHRG the Physical Modeling of Multiphase Flow,
Multiphase Production Conference, Cannes, Coventry, England, 19-21 April 1983.
France (1995) 441.
22. Sachdeva, R.: Two-Phase Flow Through
11. Schller, R.B., Munaweera, S., Selmer-Olsen, S., Chokes, M.Sc. Thesis, U. of Tulsa, Tulsa, OK,
and Solbakken, T.: Critical and Subcritical 1984.
Oil/Gas/Water Mass Flow Rate Experiments and
Predictions for Chokes, SPE Production and 23. Chemical Engineers Handbook, third edition,
Facilities (August 2006) 372. John H. Perry, McGraw-Hill Book Co. Inc., New
York City, 1950, 404.
12. Schller, R.B., Solbakken, T., and Selmer-Olsen,
S.: Evaluation of Multiphase Flow Rate
Models for Chokes Under Sub-Critical Metric Conversion Factors
o
Oil/Gas/Water Flow Conditions, SPE F (oF-32)/1.8 x 1.0 E+00 = K
Production and Facilities (August 2003) 170. ft x 3.048 E-01 = m
ft2 x 9.29 E-02 = m2
3
13. Henry, E.: Calculation Techniques for Two- ft x 2.831 E-02 = m3
Phase Critical Flows, Two-Phase Flow in. x 2.54 E+01 = m
Dynamics, 1981, Hemisphere, 415-436. kg/m3 x 1.198 264 E+02 = lbm/gal
m3/s x 5.803 036 E+01 = bbl/D
8 109243
=
m 2
[ ]
xg vg 2 + R (1 xg )vL xg + (1 xg )
2 1
x R (1 xg )
xg + (1 xg )
2
1 1 2 A2 R
= g +
m 2 g 2 L 2 R x v + R(1 x )v 2 A 2
x 1 g g1 g l
2
xg vg 2 + R(1 xg )vL A1
(A.2)
R(1 xg )vL
restriction, for mixture, we can define
=
xg kCvg + (1 xg ) CL
... (A.7)
xg vg1
n=
xg Cvg + (1 xg ) CL
.. (A.3)
d m x g v g1
2 parts of the equation. The reason for this discrepancy
= 0 . (A.9) is the inconsistency in the development of the
dr 2 A22 p1 equation by Sachdeva, et al. In writing the
momentum balance equation, they assume that
Taking the derivative and rearranging, we obtain pvgn = constant [see equation A.17 in the Appendix
of Sachdeva, et al.s paper]. However, in integrating
(1 rc ) +
n energy balance equation, they assume that gas
(rc )1 n n 1 expands isotropically, or pv g = constant. Due to the
1 k
=
{
}
1 + A2
2
2
n n
+ 1 + rc
1
n
1 1
inconsistency in their equations, both k and n appear
n 1 2 rc n + A1 in their equations. If we will replace k with n in
equation A.14, we will obtain identical expressions as
(A.10) in A.12.
This equation requires a trial-and-error solution Rewriting equation A.8, we can write the mass
since rc is on both sides of the equation. If we flow rate equation as
assume that
n
n 1
A2 CA22 p1 (1 r ) + 1 r n
<< 1 .. (A.11) n 1
m 2 =
A1 2
1 + A2
2 2
x g v g1 r + 1 1 x g + (1 x g )
1 1
n
A
r + 1
n R
we can simplify the equation as
(A.15)
(1 rc ) +
n
n 1 A2
(rc )1 1n = << 1 , we can simplify
{ }
.. (A.12) If we assume
n n 1 2 A1
+ 1 + rc n
n 1 2
n
n 1
CA22 p1 (1 r ) + 1 r n
n 1
For comparison purposes, let us consider an
equation provided by Sachdeva, et al.:
m 2 =
x g v g1 r n + x g + (1 x g )
1 2
1
(1 rc ) +
k R
r
1 1
n
= k 1 .. (A.13)
( )
c 2
k n 1 x g L
v (A.16)
+ 1 +
k 1 2 xg vg 2 In the two equations above, C is a constant and
will change depending on the units used. For
This is the same as equation A.2 in the original comparison purposes, Sachdeva, et al. provides the
paper. If we assume slippage and substitute following equation for mass flow rate.
1
vg 2 = vg1rc k
, we can rewrite equation A.13 as
k
k 1
CA22 p1 (1 r ) + 1 r k
k 1
(1 rc ) +
k m =
2
2
k 1 x g v g1 r k +
1 1 1
{ }
r k
= .. (A.14)
c
k n 1 2
+ 1 + rc k
k 1 2
(A.17)
When we compare equations A.12 and A.14, we
observe that the two equations are similar except that If we substitute n for k, and account for slippage,
instead of using n, Sachdeva, et al. has k in some equation A.17 will become A.16.