Documente Academic
Documente Profesional
Documente Cultură
ARTICLE IN PRESS
Journal of Computational Science xxx (2013) xxxxxx
J.L. Balino
Departmento de Engenharia Mecnica, Escola Politcnica, Universidade de So Paulo, Av. Prof. Mello Moraes, 2231, CEP 05508-900, Cidade Universitria,
So Paulo, SP, Brazil
a r t i c l e
i n f o
Article history:
Received 31 January 2013
Received in revised form 29 July 2013
Accepted 25 August 2013
Available online xxx
Keywords:
Severe slugging
Pipeline-riser system
Air-water ow
Stability
Lumped and distributed parameter systems
Switched systems
Petroleum production technology
a b s t r a c t
A mathematical model and numerical simulations corresponding to severe slugging in air-water pipelineriser systems are presented. The mathematical model considers continuity equations for liquid and gas
phases, with a simplied momentum equation for the mixture. A drift-ux model, evaluated for the local
conditions in the riser, is used as a closure law. In many models appearing in the literature, propagation
of pressure waves is neglected both in the pipeline and in the riser. Besides, variations of void fraction in
the stratied ow in the pipeline are also neglected and the void fraction obtained from the stationary
state is used in the simulations. This paper shows an improvement in a model previously published by the
author, including inertial effects. In the riser, inertial terms are taken into account by using the rigid waterhammer approximation. In the pipeline, the local acceleration of the water and gas phases are included in
the momentum equations for stratied ow, allowing to calculate the instantaneous values of pressure
drop and void fraction. The developed model predicts the location of the liquid accumulation front in
the pipeline and the liquid level in the riser, so it is possible to determine which type of severe slugging
occurs in the system. A comparison is made with experimental results published in literature including a
choke valve and gas injection at the bottom of the riser, showing very good results for slugging cycle and
stability maps. Simulations were also made assessing the effect of different strategies to mitigate severe
slugging, such as choking, gas injection and increase in separation pressure, showing correct trends.
2013 Elsevier B.V. All rights reserved.
1. Introduction
Severe slugging is a terrain dominated phenomenon, characterized by the formation and cyclical production of long liquid slugs
and fast gas blowdown. Severe slugging may appear for low gas and
liquid ow rates when a section with downward inclination angle
(pipeline) is followed by another section with an upward inclination (riser). This conguration is common in off-shore petroleum
production systems. Main issues related to severe slugging are:
(a) High average back pressure at well head, causing tremendous
production losses, (b) High instantaneous ow rates, causing instabilities in the liquid control system of the separators and eventually
shutdown, and (c) Reservoir ow oscillations.
For steady state and low ow rates, the ow pattern in the
pipeline may be stratied, while it may be intermittent in the riser,
as shown in Fig. 1(a).
A cycle of severe slugging can be described as taking place
according to the following stages [1]. Once the system destabilizes and gas passage is blocked at the bottom of the riser, liquid
1877-7503/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jocs.2013.08.006
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
no / Journal of Computational Science xxx (2013) xxxxxx
J.L. Bali
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
(1)
g0 Rg Tg
m
=
Ppt A
(2)
g0 and Ql0 are respectively the gas mass ow and the liquid
where m
volumetric ow injected in the pipeline and Rg and Tg are respectively the gas constant and temperature.
2.1.1. Condition x > 0
For this condition there is no gas owing out the pipeline, resulting:
jgb = 0
(3)
Applying mass conservation equation for the liquid and gas phases,
it is obtained:
Ql0
A
jlb
+ p
dp
=
Lx
dt
dP g
=
dt
P g
jlb
Ql0
A
dx
dt
(4)
+
Rg Tg
A
g0
m
(5)
(L x) p + Le
where L is the pipeline length and t is time. A mass balance for the
liquid in the stratied region yields:
dp
Q
(L x)
+ jls l0 = 0
A
dt
(6)
jgs = jgb
(10)
jls = jlb
(11)
Pps = Ppb
(12)
dx
dt
(8)
2f
dj
= Pps + l x g sin ll jlb |jlb | lb
D
dt
(9)
where fll is the Fanning friction factor (assuming that only liquid
is lling the cross sectional area), g is the gravity acceleration constant, l is the liquid density and is the pipeline inclination angle
jlb
P g
Ql0
A
jgb + jlb
(13)
Ql0
A
Lp + Le
Rg Tg
A
g0
m
(14)
dx
=0
dt
(15)
jgb
=0
(16)
(7)
Ppb
2.1.2. Condition x = 0
For this condition there is no liquid penetration front, resulting:
+
jgb
= p
dx
jls = jlb + p
dt
jgs = p
(positive when downwards). The Fanning friction factor is calculated by using the correlation from [12].
dx
dt
jlb
= jlb
+ p
(17)
dx
dt
(18)
+
jgb
= jgb
=0
(19)
jlb
= jlb
(20)
dx
=0
dt
(21)
In Eqs. (15)(21) the superscripts and + denote variables evaluated respectively at t0 and t0+ . In this way, in any commutation
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
ARTICLE IN PRESS
wg
ug =
l
1
1
+
1 p
p
dug
dul
g
dt
dt
jg =
jg
(24)
Pg
Rg Tg
(25)
1
(jgt + jgs )
2
Q
l0
+ jls
(26)
(27)
g sin
dul
dt
Ppt = Pps
0
Lx
jl
+
=0
t
s
(29)
In Eqs. (22)(29) the wetted and interfacial perimeters are determined considering a stratied geometry, while the shear stresses
are related to the supercial velocities of the phases through friction factors based on the hydraulic diameters for each phase [13].
(31)
where jg , jl and j are respectively the gas, liquid and total supercial
velocities, P is the pressure, s is the position along the riser, is the
void fraction and ug and ul are respectively the gas and liquid phase
velocities.
The supercial velocities for the phases are determined by using
a drift ux correlation, assumed to be locally valid:
jg = ug = (Cd j + Ud )
(32)
jl = j jg = ul (1 ) = (1 Cd )j Ud
(33)
(34)
(35)
Ud = 0.35
gD sin
(37)
P
P
ds = Pps
(L x)
s
s
(30)
(P) +
(Pjg ) = 0
t
s
Cd = 1.2
(28)
2.2. Riser
(23)
1
jl =
2
(22)
=0
jl
ul =
1 p
g =
(38)
gD
j
+ (Cd j + Ud )
=0
Dt
s
s
(39)
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
Dg P
j
+P
=0
Dt
s
(40)
where:
Dg
=
+ ug
Dt
t
s
(41)
Dg ug
fm
P
Du
(1 )l l l
= m g sin + 2 j
j
g
D
Dt
Dt
s
m = l (1 ) +
Rg Tg
(42)
(43)
Dl
=
+ ul
Dt
t
s
(44)
Rg Tg
A
gl + P jg
m
P+
P+ 1
P =
1
Rg Tg
(45)
j 2 +
g
1
+ =
2
l jl
1
Rg Tg
1
1+
j 2
1
1
(46)
jg+
(Cd j + Ud )
(47)
(48)
If su sl , we get:
dP gr
P gr
=
st su
dt
jgt ju
Rg Tg
P gr A
gl
m
(49)
where jgt is the gas supercial velocity at the top of the riser, ju is
the total supercial velocity at the liquid level and st is the position
of the top of the riser.
The momentum conservation equation results, for both cases:
Pu = Pt + g g(zt zu ) +
jgr =
1
(ju + jgt )
2
djgr
2fgg
jgr
jgr
D
dt
(st su )
(50)
(51)
where fgg is the Fanning friction factor considering only gas owing,
jgr and g are respectively the mean gas supercial velocity and the
mean gas density at the gas region; zt and zu are respectively the
vertical positions at the top of the riser and at the liquid level.
2.2.3. Riser geometry
The riser geometry is characterized by the coordinates X and Z
corresponding to the top of the riser and a set of functions furnishing the ordinate z and local inclination angle as a function of the
local position s along the riser (see Fig. 4). For a constant angle riser,
for instance, it results:
= arctan
Z
X
st = (X 2 + Z 2 )
z = s sin
1/2
(52)
(53)
(54)
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
ARTICLE IN PRESS
1
Pt Ps = Kv mt jt
jt
(55)
(56)
Pt Ps = Cjlt
jlt
and total supercial velocity. The pipeline imposes the supercial velocities for the gas and liquid phases at the bottom of the
riser, while the riser imposes the pressure to the pipeline; these
variables are the boundary conditions for the corresponding subsystems. Additional boundary conditions are the liquid volumetric
ow rate and the gas mass ow rate at the pipeline, as well as the
gas lift mass ow rate and separation pressure at the riser.
As initial conditions, the stationary conditions were chosen, this
is, the solution of the system of equations obtained after setting
equal to zero the time derivatives.
3. Discretization and numerical implementation
(57)
where jlt is the liquid supercial velocity at the top of the riser and
C is a dimensional valve constant. According to this relationship, it
is neglected the contribution of the gas phase to pressure drop.
2.3. Coupling between pipeline and riser
Assuming the same ow passage area for the pipeline and riser,
the pressure and supercial velocities at the bottom of the riser are
continuous:
P (s = 0, t) = Pb (t)
(58)
jg (s = 0, t) = jgb (t)
(59)
jl (s = 0, t) = jlb (t)
(60)
jgb
Cdb jb + Udb
(61)
4. Simulations
Fig. 7 shows the state variables and the coupling between the
subsystems. State variables for the pipeline are the average gas
pressure, void fraction and position of the liquid accumulation
front, while for the riser they are the local pressure, void fraction
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
Fig. 9. Simulation results for jg0 = 0.063 m/s and jl0 = 0.124 m/s (case 1, Table 1).
g0
Rg T0 m
P0 A
(62)
jl0 =
Ql0
A
(63)
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
ARTICLE IN PRESS
Table 1
Comparison with experimental results [3].
Case
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Experiment
Simulation
Simulation [5]
Simulation [3]
jg0 (m/s)
jl0 (m/s)
Texp (s)
Tsim (s)
Error (%)
Tsim (s)
Error (%)
Tsim (s)
Error (%)
0.063
0.064
0.123
0.124
0.062
0.063
0.063
0.064
0.065
0.122
0.123
0.126
0.187
0.188
0.188
0.19
0.058
0.063
0.122
0.126
0.126
0.184
0.185
0.187
0.188
0.19
0.313
0.314
0.319
0.321
0.43
0.433
0.124
0.209
0.183
0.212
0.679
0.367
0.679
0.535
0.226
0.374
0.621
0.228
0.226
0.466
0.502
0.312
0.705
0.698
0.730
0.673
0.085
0.127
0.161
0.551
0.755
0.685
0.433
0.347
0.614
0.744
0.604
0.701
24
20
15
14
6
13
9
10
19
11
8
13
11
8
7
10
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
25.4
19.5
15.5
14.6
7.36
13.3
7.35
9.27
18.7
10.8
7.36
14.0
11.9
8.22
7.86
10.1
7.19
7.21
6.45
14.0
18.2
4.24
12.3
7.36
Stable
6.24
14.0
Stable
Stable
Stable
Stable
Stable
5.8
2.5
3.3
4.3
22.7
2.3
18.3
7.3
1.6
1.8
8.0
7.7
8.2
2.8
12.3
1.0
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
24.2
18.4
14.7
13.7
6.6
12.0
6.5
8.3
17.3
9.8
6.8
13.0
10.8
7.3
7.0
9.2
6.4
6.5
Stable
Stable
18.2
13.0
12.0
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
0.8
8.0
2.0
2.1
10
7.7
27.8
17.0
8.9
10.9
15.0
0.0
1.8
8.8
0.0
8.0
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
22
17
14
13
6
11
6
7
16
9
Stable
12
10
7
Stable
8
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
8.3
15
6.7
7.1
0.0
15.4
33.3
30.0
15.8
18.1
NA
7.7
9.1
12.5
NA
20.0
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
instability condition and the uctuations associated to the intermittent ow based only on a visualization. Besides, the experimental
determination of the stability curve requires a very careful control
of variables such as separator pressure and input ows; cases 5
and 7 in Table 1, for instance, show a large discrepancy in experimental period for almost the same values of supercial velocities,
indicating that these or other variables were not kept constant in
the experiment.
Fig. 10. Numerically generated stability map and data from [3].
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
10
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
Table 2
Comparison with experimental results, choke valve, C = 1.2 105 Pa s2 /m2 [16].
Case
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Experiment
Simulation
Simulation [16]
jg0 (m/s)
jl0 (m/s)
Texp (s)
Tsim (s)
Error (%)
Tsim (s)
Error (%)
0.0753
0.1147
0.1739
0.0781
0.1181
0.0809
0.1209
0.1713
0.1209
0.1734
0.2493
0.1698
0.2474
0.2502
0.251
0.0959
0.0949
0.0959
0.0497
0.0487
0.1704
0.1693
0.0497
0.2365
0.2354
0.2386
0.1693
0.1704
0.0959
0.0497
46.6
37.6
31.8
47.5
38.5
45
39.9
31.5
Stable
Stable
Stable
Stable
Stable
Stable
Stable
54.9
43.7
32.8
56.2
43.7
50.4
42.4
31.8
43.0
36.7
30.6
35.5
28.5
25.0
22.0
17.8
16.2
3.1
18.3
13.5
12.0
6.3
1.0
NA
NA
NA
NA
NA
NA
NA
44.9
38.1
Stable
50.7
Stable
35
31
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
3.6
1.3
NA
6.7
NA
22.0
22.3
NA
NA
NA
NA
NA
NA
NA
NA
gl
Rg T0 m
(64)
P0 A
Table 3 shows a comparison of experimental and model simulated severe slugging periods for different gas and liquid supercial
velocities. Compared to the former simulations, errors are higher
for the cases with low gas supercial velocities, comparable to the
injection gas supercial velocity (cases 13, 5 and 6). Again, there
are some stable cases close to the stability line (1417) in which
the model prediction is unstable. In general, the agreement with
the experimental data is pretty good.
In Table 3 are also shown simulation results using the model of
[16], which is a modication of the model of [3] (without inertia
and friction) to take into account the gas injection. It can also be
observed higher errors for low gas supercial velocities. Besides,
there are some cases in which a unstable ow is predicted as stable
(cases 713).
Table 3
Comparison with experimental results, gas injection, jg0gl = 0.091 m/s [16].
Case
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Experiment
Simulation
Simulation [16]
jg0 (m/s)
jl0 (m/s)
Texp (s)
Tsim (s)
Error (%)
Tsim (s)
Error (%)
0.0808
0.1153
0.1702
0.2515
0.0791
0.1147
0.3125
0.1695
0.248
0.1129
0.1737
0.3215
0.2489
0.366
0.4115
0.369
0.3141
0.2528
0.2549
0.2571
0.2582
0.152
0.152
0.1542
0.1013
0.0949
0.0487
0.0476
0.0916
0.0465
0.1552
0.1552
0.0981
0.0444
21.5
18.9
14.9
13.4
24.1
20.5
10.8
16.8
13.2
27
18.8
9.5
14
Stable
Stable
Stable
Stable
29.8
23.8
18.1
13.2
36.9
27.3
11.0
19.4
13.9
28.0
19.7
11.1
13.6
9.82
8.96
9.41
10.4
38.6
25.9
21.5
1.5
53.1
33.2
1.9
15.5
5.3
3.7
4.8
16.8
2.9
NA
NA
NA
NA
29.1
21.8
16.1
11.6
33.8
24.4
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
Stable
35.3
15.3
8.1
13.4
40.2
19.0
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
ARTICLE IN PRESS
11
Fig. 11. Simulation results for jg0 = 0.063 m/s, jl0 = 0.124 m/s and kv = 100.
Fig. 12. Simulation results for jg0 = 0.063 m/s, jl0 = 0.124 m/s and kv = 150.
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
12
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
Fig. 13. Simulation results for jg0 = 0.063 m/s, jl0 = 0.124 m/s and jg0gl = 0.1 m/s.
Fig. 14. Simulation results for jg0 = 0.063 m/s, jl0 = 0.124 m/s and jg0gl = 0.3 m/s.
Fig. 15. Simulation results for jg0 = 0.063 m/s, jl0 = 0.124 m/s and Ps = 1.3 bara.
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
no / Journal of Computational Science xxx (2013) xxxxxx
J.L. Bali
13
Fig. 16. Simulation results for jg0 = 0.063 m/s, jl0 = 0.124 m/s and Ps = 1.7 bara.
Acknowledgements
This work was supported by Petrleo Brasileiro S. A. (Petrobras).
The nancial support of Fundaco de Amparo Pesquisa do Estado
de So Paulo (FAPESP) to attend IMAACA 2012 (where a shorter
version of this paper received the Best Paper Award) is deeply
acknowledged.
References
[1] Y. Taitel, Stability of severe slugging, International Journal of Multiphase Flow
12 (2) (1986) 203217.
[2] Z. Schmidt, Experimental study of two-phase slug ow in a pipeline-riser system, The University of Tulsa, Tulsa, USA, 1977 (Ph.D. thesis).
[3] Y. Taitel, S. Vierkand, O. Shoham, J.P. Brill, Severe slugging in a riser system:
experiments and modeling, International Journal of Multiphase Flow 16 (1)
(1990) 5768.
[4] C. Sarica, O. Shoham, A simplied transient model for pipeline-riser systems,
Chemical Engineering Science 46 (9) (1991) 21672179.
K.P. Burr, R.H. Nemoto, Modeling and simulation of severe slugging
[5] J.L. Balino,
in air-water pipeline-riser systems, International Journal of Multiphase Flow
36 (2010) 643660.
[6] C. Wordsworth, I. Das, W.L. Loh, G. McNulty, P.C. Lima, F. Barbuto, Multiphase
Flow Behavior in a Catenary Shaped Riser. CALTEC Report No.: CR 6820, Bedford,
UK, 1998.
Modeling and simulation of severe slugging with mass
[7] R.H. Nemoto, J.L. Balino,
transfer effects, International Journal of Multiphase Flow 40 (2012) 144157.
[8] S. Mokhatab, Severe Slugging in Offshore Production Systems, Nova Science
Publishers, Inc., New York, 2010.
[9] J.M. Masella, Q.H. Tran, D. Ferre, C. Pauchon, Transient simulation of two-phase
ows in pipes, International Journal of Multiphase Flow 24 (1998) 739755.
[10] Y. Taitel, A.E. Dukler, A model for predicting ow regime transitions in horizontal and near horizontal gasliquid ow, American Institute of Chemical
Engineers Journal 22 (1) (1976) 4755.
[11] M.H. Chaudhry, Applied Hydraulic Transients, Van Nostrand Reinhold, New
York, USA, 1987.
[12] N.H. Chen, An explicit equation for friction factor in pipe, Industrial and Engineering Chemical Fundamentals 18 (1979) 296297.
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006
G Model
JOCS-225; No. of Pages 14
ARTICLE IN PRESS
14
J.L. Bali
no / Journal of Computational Science xxx (2013) xxxxxx
Modeling and simulation of severe slugging in air-water systems including inertial effects,
Please cite this article in press as: J.L. Balino,
J. Comput. Sci. (2013), http://dx.doi.org/10.1016/j.jocs.2013.08.006