You are on page 1of 6

Simulation of Gas/Liquid Flow

in Slug Catchers
A. 80S, Koninklijke/Shell Laboratorium
J.G. du Chatinier, Koninklijke/Shell Laboratorium

Summary. For the optimization of existing multiple-pipe-type slug catchers and the development of novel
configurations, small-scale model tests are used that u.se a two-liquid s~stem with a density r~tio repr~se.ntative
of that of the gas/liquid phases in practice. The techmque has been valIdated by field tests WIth an eXIstmg
slug catcher.

Introduction
Trunklines transporting natural gas often operate in the a suitable laboratory technique available to test a design
two-phase-flow mode because of liquid injection and/or concept. In this paper, a laboratory-scale modeling tech-
additional liquid formation by retrograde condensation. nique is presented, together with its application for the
Slip in velocity between the gas and liquid phases leads improvement of an existing mUltiple-pipe slug catcher.
to accumulation of liquid in the pipeline (liquid holdup). First, though, a more detailed description of multiple-pipe
The total amount of liquid present at any time in a long- slug catchers is given.
distance two-phase pipeline can be significant. When oper-
ating conditions are changed, large volumes of liquid may The Multiple-Pipe Slug Catcher
emerge from the pipeline, maybe as a result of a change In slug-catcher design, the mUltiple-pipe concept finds
in volume flow-i.e., in velocity-or a sphere (or pig) wide application. A multiple-pipe slug catcher consists
run through the line. The largest slug that can ever occur of an entrance section, where liquid/gas separation occurs,
is that caused by sphering. The occasionally very large and an array of parallel downward-sloping bottles (of stan-
volumes of ("live") liquids encountered must be handled dard line-pipe size) for liquid storage.
and stored onshore as they emerge from the pipeline, Fig. 1 shows the layout of an existing mUltiple-pipe slug
preferably without any reduction in velocity, which would catcher (in operation in The Netherlands). The names of
be reflected in the gas production. For this reason, a the various parts are indicated. An incoming liquid slug
liquid-receiving facility known as a slug catcher is always flows through the splitter into the inlet manifold and then
connected to a two-phase pipeline. through downcomers into and down the sloping bottles.
A slug catcher consists essentially of two parts: a sepa- The downward-flowing slug displaces the gas present in
rator, which separates the liquid from the mixed stream the bottles up through the risers mounted on the bottles;
arriving under normal (steady) flow conditions, and from the risers, the gas flows through the outlet header
storage, which receives and stores the incoming liquid slug into the gas-treating plant. The liquid/gas exchange that
created by upset conditions (such as running a sphere takes place in the bottles ensures that the gas supply to
through the pipeline). When a more-or-less continuous the downstream facilities remains uninterrupted during liq-
slug ofliquid arrives, the liquid displaces the gas present uid slug arrival, provided that excessive liquid carry-over
in the slug catcher, ensuring an uninterrupted supply of can be avoided.
gas to the downstream facilities (compressor, treating In this particular slug catcher, only primary bottles are
plant, and liquefied natural gas plant). Gas lines general- used. The liquid from the inlet header flows directly into
ly operate at velocities of up to 12 m/s [39 ft/sec], and the bottles, which have both a separation and a storage
large slugs will take only a matter of minutes to arrive. function. Fig. 2 shows the layout of another slug catcher
Therefore, the holding capacity of the slug catcher must (in operation in the U.K.), which has nine primary and
essentially be as great as the volume of the largest slug. four secondary bottles. The secondary bottles have only
Although liquid carry-over must be limited, a slug catch- a storage function and are filled from the bottom through
er is not meant to replace a high-efficiency separator. the primary bottles and bottom header.
A great number of slug catchers are in operation The decision of how many primary and secondary bot-
throughout the world. \-3 They may have a vessel, tles should be used in a given multiple-pipe slug catcher
mUltiple-pipe, or parking-loop configuration and vary depends on several factors: the gas flow rate in the pipe-
widely in geometry, mainly because of the many degrees line; the liquid storage capacity required; the size of plot
of freedom possible in the design. It is very difficult to available (length available for bottles); and the diameter
produce a reliable design on a purely theoretical basis, and slope of the bottles, which determine the maximum
especially when the geometry is as complex as in the case slug flow rate that the primary bottles can accommodate
of a multiple-pipe slug catcher. Then, it is useful to have without liquid carry-over occurring. It is desirable to keep
the number of primary bottles as low as possible for eco-
Copyright 1987 Society of Petroleum Engineers nomic and operational reasons; a lower number will en-
178 SPE Production Engineering, August 1987
INLET HEAOER
FROM GAS LINE / DOWNCOMERS
----7 G~~~~~LET HEADERS

BOTTOM HEADER

~MARY BOTTLES

~ECONDARY BOTTLES

L1QU!O OUTLET

Fig. 1-Geometry of a mUltiple-pipe slug catcher with only Fig. 2-Geometry of a multiple-pipe slug catcher with
primary bottles. primary and secondary bottles.

sure that the number of T -connections and hence the is not yet completely understood for two-phase flow
chances of liquid-slug maldistribution are minimized. through a simple tee ,5; flow through a 3D network is
A major advantage of the multiple-pipe slug catcher is even less understood. At Koninklijke/Shell Laboratori-
its ease of operation. Apart from level control, no other um, Amsterdam, the laboratory-scale modeling approach
instrumentation is required. There are, however, a num- has therefore been adopted.
ber of disadvantages. In this approach, a scaled-down version of the proto-
1. It is most important that the incoming liquid be dis- type slug catcher is built from Perspex TM. To create a
tributed evenly over the primary bottles. If the distribu- realistic ratio between the densities of the heavy and light
tion is uneven, excessive liquid carry-over may occur phases, the gas phase is simulated with kerosene and the
because of overloading of certain bottles. condensate with 55 wt % zinc chloride concentrate in
2. The incoming liquid flows into the primary bottles water. Table 1 shows that as far as the density ratio is
and drives out the gas in countercurrent flow. Should the concerned, a much closer similarity is obtained than if
velocity of the gas relative to the liquid rise beyond a crit- a water/air system were used. The mismatch as regards
icallevel, the gaslliquid interface will become unstable. the viscosity is of little importance, provided that, in the
In this case, excessive wave growth or even choking of full-scale and model slug catchers (if the scale is not too
the bottle may take place, which can result in excessive small), all fluids are in turbulent flow. In the simulation
liquid carry-over. The slope of the bottles has a marked studies, Froude-number scaling is applied; i.e., the flow
influence on this. conditions in the model slug catcher are considered rep-
3. It is not surprising that, given the nature of the flow resentative of those in the full-scale slug catcher charac-
[two-phase flow in a three-dimensional (3D) network], terized by the same densimetric Froude number, N Fr :
the configuration of a multiple-pipe slug catcher is some-
times very complex.

Setup of Model Studies


Research is being carried out to optimize the perform-
ance of existing slug catchers and to develop novel types. when:l
Particularly in the case of multiple-pipe slug catchers, it VSg =
superficial gas velocity in the pipeline,
is very difficult to design a reliable slug catcher on a purely mis,
theoretical basis because of the complex problem of two- PL,Pg = liquid and gas densities, respectively,
phase flow in a 3D network. The distribution mechanism kg/m3,

TABLE 1-COMPARISON OF PHYSICAL PROPERTIES

Natural Gas/Condensate
The Netherlands U.K. Kerosene/Zinc Chloride
Slug Catcher Slug Catcher Concentrate
System Physical Properties (70 bar) (110 bar) (55 wt%) Air/Water
Density of light phase, kg/m 3 65 158 755 1.3
Density of heavy phase, kg/m 3 590 418 1660 1000
Density ratio 9.1 2.6 2.2 769
Dynamic viscosity of light
phase, 10 -5 Ns/m2 1.35 1.8 140 1.8
Dynamic visosity of heavy
phase, 10 - 5 Ns/m 2 15.5 8.5 660 100
Interfacial surface
tension, 10 -3 N/m 9 0.65 12 72

SPE Production Engineering, August 1987 179


DISTRIBUTION OF
SLUG OVER BOTTLES, 0/0

20
CARRY-OVER
VESSEL

10
PERSPEX MODEL

o
B

A
P LOCATION OF THE4
INLETS OF THE
INLET HEADER

Fig. 4-Distribution of the slug over the bottles in the non-


modified slu~-catcher model. D = 1.7 m 3th (equivalent to
Fig. 3-The two-liquid test facility. 0=ZnC1 2 con- 15x10 6 Nm td); 0=2.5 m 3 th (equivalent to 23x10 6
centrate. Nm 3 td).

d = pipeline diameter, m, and Strictly speaking, this approach is justified only in the first
g = acceleration caused by gravity, m/s 2 . stage of the slug-catching operation. In a full-scale slug
catcher, the bottles act in this first stage as if they are
The reasoning behind this approach is that the physical closed because they are never completely drained off; the
mechanism involved in the liquid carry-over in slug catch- condensate left in the bottom of the bottles then acts as
ers is wave formation on the liquid/gas interface and slug- a seal. Eventually, when the descending condensate stream
ging in the bottles. These phenomena are governed in the bottles has made contact with the condensate re-
primarily by the Froude number. 6, 7 maining in the bottom of the bottles, the situation will
Fig. 3 shows schematically the experimental setup. The change. Condensate will flow from the more highly loaded
two-liquid test facility consists of a Perspex model of the bottles to the less highly loaded ones through the bottom
slug catcher under study and the slug generator unit. The header. This will affect the ascending gas stream and even-
main components of the slug generator unit are the storage tually the distribution of the slug over the bottles.
vessel (containing kerosene and zinc chloride concentrate)
and the slug vessel (containing zinc chloride concentrate).
A test cycle consists of the following steps. Model Studies for the Multiple-Pipe
1. Kerosene is recirculated through the model and the Slug Catcher
storage vessel [with Valves A and C closed, Valve B open, Model studies with the two-liquid test facility, which has
and Valve D (the three-way valve) in the "up" position]. only primary storage bottles, have been carried out in The
This simulates the normal flow situation (i.e., no slug Netherlands for the existing multiple-pipe slug catcher.
arrival). This offered an ideal opportunity to verify the two-liquid
2. The kerosene stream is diverted through the slug ves- model test approach. In its original version, the slug catch-
sel by switching Valve D to the "down" position, which er could accommodate slugs pigged at a gas flow rate in
forces a slug of zinc chloride concentrate into the slug the range of 18 to 20 million Nm 3 /d [670 to 750
catcher. To simulate a certain "gas" content, the slug may MMscf/D] without liquid carry-over. Above this gas flow
be mixed with kerosene. rate, however, the slug catcher started to give liquid carry-
3. After the slug-catching operation is completed, the over (0.3% of the slug volume at 20x 10 6 Nm 3 /d [750
normal kerosene flow circulation is re-established and the MMscf/DD, which hampered the operations of the gas-
zinc chloride concentrate is drained from the Perspex treating facilities downstream. Because the gas flow rate
model. was expected to increase to up to 30 X 10 6 N m 3 / d [1, 120
4. The slug vessel is refilled with zinc chloride con- MMscf/D], the performance of the slug catcher had to
centrate (by opening Valves A and C and closing Valve be improved.
B). As a first step, a team carried out neutron back-
The zinc chloride carried over during a slug-catching scattering (NBS) measurements 8 to measure the local liq-
operation is collected in the carry-over vessel downstream uid holdup in the slug catcher during slug arrival. These
of the riser system, and its amount is measured when the measurements revealed that the slug was unevenly dis-
test is completed. The distribution of the slug over the tributed over the inlet header, leading to an uneven dis-
bottles was found from tests in which the bottles were iso- tribution of the slug over the bottles and hence to
lated from each other at the.1ower end by means of an overloading of several bottles.
insert in the bottom header, the "condensate" contents Next, model tests were set up with the two-liquid test
of the individual bottles being determined after the test. facility to simulate the real-life situation and to develop

180 SPE Production Engineering, August 1987


CARRY- OVER, 0/0

10
DISTRIBUTION OF
20~LUG OVER BOTTLES, %

AVERAGE LEVEL
-- ------

5 10t-

OL...

~
LOCATION OF THE~
INLETS OF THE
0.3 ------- INLET HEADER
o 1 2 3 4 5 6
SLUG FLOW RATE IN MODEL, m3 /h Fig. 6-Effect of downcomer constrictions on the slug dis-
tribution in the model slug catcher; flow rate is 2.5 m 3 /h
Fig. 5-Effect of eccentric 15-mm downcomer constric- (equivalent to 23 x 10 6 Nm 3/d). 0 = no constrictions;
tions on the performance of the model slug catcher. ~ = 15-mm eccentric constrictions.

ways of improving the distribution of the slug over the The left curve in Fig. 5 represents the condensate carry-
bottles to avoid carry-over. over in the model as a function of the slug flow rate. It
AI: 20 Perspex model of the slug catcher was built. can be seen that, even at the lowest flow rate, liquid carry-
Froude scaling implies for the velocity, v, over occurred. In the model, the slug flow rate at which
the liquid carry-over reached a value of 0.3% is about
vfS=11.6 Vm
1.7 m 3 Jh [10.7 bbl/hr], corresponding to a gas flow rate
of 15 X 10 6 Nm 3 Jd [562 MMscflD]. This is in the same
or, for the flow rate, q, order of magnitude as the flow rate of 20 x 10 6 Nm 3 Jd
[750 MMscfJD], at which the same amount of carry-over
qfs=9qm' occurred in practice.
where qm is in m 3 Jh and qjS is in 10 6 Nm 3 Jd (at7,OOO Therefore, we may conclude that the model qualitatively
kPa [70 bar] and lO o e [50°F]). simulates well the liquid distribution of the slug over the
First the performance of the full-scale slug catcher was inlet header and bottles and the liquid carry-over mecha-
simulated. The slug flow rate in the model varied between nism of a full-scale slug catcher.
1.2 and 5.1 m 3 Jh [7.5 and 32 bbl/hr], corresponding to Both the field measurements (NBS) and laboratory ex-
a gas flow rate of 11 X 10 6 to 46 X 10 6 Nm 3 Jd (at 7,000 periments indicated that the slug catcher did not function
kPa and lO o e [70 bar and 50°F]) in the full-scale slug optimally because of the maldistribution of the slug over
catcher. Uneven liquid distribution was observed to occur the bottles. A straightforward remedy is to introduce a
at flow rates below 3.5 m 3 Jh [22 bbl/hr]. This effect was flow resistance for the liquid by installing constrictions
more pronounced for lower flow rates. in the downcomer portholes; the extra pressure drop
Fig. 4 shows the distribution of the slug over the bot- created by this measure is conducive to better distribu-
tles for two flow rates. Most of the slug flows into the tion. For this purpose, a close-fitting pipe was mounted
bottles closest to the inlets of the inlet header. A few bot- in the inlet header of the model. The pipe had holes of
tles farther downstream receive hardly any slug. A simi- equal size in the place of the downcomers. These holes
lar maldistribution pattern has also been observed in the functioned as constrictions. In the model experiments,
full-scale slug catcher. It was further observed in the pipes were used with holes ranging in diameter from 8
model tests that, in the risers of the less highly loaded to 22 mm [0.3 to 0.9 in.]. The constrictions minimized
bottles, the "gas" flowed downward rather than upward. the maldistribution of the slug and therefore improved the
This gas flowed back to the inlet header, thereby proba- slug-catcher performance. If the constrictions were too
bly reinforcing the maldistribution of the slug in the inlet small, however, the increased turbulence in the inlet sec-
header. The liquid-carry-over mechanism was the same tion of the bottle generated foam, which adversely affected
as observed by NBS in the full-scale slug catcher. High the performance of the slug catcher. Furthermore, foam-
waves of "condensate" built up in the bottles just up- ing was markedly reduced if the constrictions were lo-
stream of the riser, and condensate whipped off the wave cated eccentrically, because the condensate was then
crests was entrained in the gas stream escaping through guided along the downcomer wall rather than flowing as
the riser. jets into the bottles. Fig. 6 shows the beneficial effect of
SPE Production Engineering, August 1987 181
VSg~
the optimum downcomer constrictions (I5-mm [0.6-in.]
diameter, eccentric) on the slug distribution. Now each N Fr = Froude number, Pg
bottle receives an amount ofliquid deviating less than 20% (PL -Pg)dg
from the average. q = gas flow rate, m 3 /h [bbl/hr] or Nm 3 /d
The effect of the constrictions on the condensate-carry- v = velocity, m/s [ft/sec]
over point is demonstrated in Fig. 5. It is clear that the
slug-catcher performance has improved markedly. The
P = density, kg/m3 [Ibm/gal]
onset of carry-over has shifted to a flow rate of 3.5 m 3 /h
[22 bbl/hr], which is equivalent to a "gas" flow rate of Subscripts
32 X 10 6 Nm 3 /d [1,200 MMscf/D] in practice. Applica- fs = full-scale
tion of the scaling rule that the ratio of the resistance fac- g = gas
tor of the constrictions and the Fanning friction factor L = liquid
associated with the condensate flow in the inlet header m = model
should be the same for both the model and the full-scale S = superficial
slug catcher results in an optimum of 38 cm [15 in.] for
the diameter of the downcomer constrictions in the real- References
life situation.
I. Huntley, A.R. and Silvester, R.S.: "Hydrodynamic Analysis Aids
The optimum downcomer constrictions were installed Slug Catcher Design," Oil & Gas 1. (Sept. 19, 1983) 81, 95.
in the full-scale slug catcher. Follow-up NBS measure- 2. Oranje, L.: "Handling Two-Phase Gas/Condensate Flow in
ments showed that the slug distribution over the inlet head- Offshore Pipeline Systems," Oil & Gas 1. (April 18, 1983) 81, 129.
er and bottles had improved substantially. All eight bottles 3. Martin, R.E.: "Handling Liquids in Offshore Gas Lines Gets New
now received the same amount of slug to within about Approach," Oil & Gas J. (April 27, 1981) 79, 143.
4. Azzopardi, B.I. and Whalley, P.B.: "The Effect of Flow Patterns
20 %. The onset of carry-over was shifted to a gas flow on Two-Phase Flow in a T-junction," Inti. 1. Multiphase Flow
rate of 34x10 6 Nm 3/d [1,275 MMscf/d], which is in (1982) 8, 491.
good agreement with the prediction based on the model 5. Saba, N. and Lahey, R. T.: "The Analysis of Phase Separation Phe-
experiments (32X10 6 Nm 3/d [1,200 MMscfIDD. nomena in Branch Conduits," IntI. 1. Multiphase Flow (1984) 10, I.
6. Wallis, G.B. and Dobson, D.E.: "The Onset of Slugging in
Horizontal Stratified Flow," Inti. 1. Multiphase Flow (1973) 1, 173.
Conclusions 7. Taitel, Y. and Dukler, A.E.: "A Model for Predicting Flow Regime
1. Two-phase flow behavior in a high-pressure multiple- Transitions in Horizontal and· Near-Horizontal Gas/Liquid Flow,"
AIChE 1. (1976) 22, 47.
pipe slug catcher (i.e., slug distribution and liquid-carry-
8. De Monchy, A.R., Kok, C.A., and van Klaveren, I.A.: "Appli-
over mechanism) has been successfully simulated in a two- cations of Californium-252 Neutron Scattering Gauges in the
liquid laboratory model at atmospheric conditions. Petrochemical Industry," Proc., IntI. Symposium on Califor-
2. The model studies showed that the performance of nium-252 Utilization, Paris (April 26-28, 1976).
an existing slug catcher could be improved by correcting
maldistribution of the liquid slug. This was achieved in SI Metric Conversion Factors
the model by installing downcomer constrictions.
bar X 1.0* E+05 Pa
3. The effectiveness of appropriately sized constrictions,
located eccentrically, has been confirmed in the full-scale
bbl X 1.589873 E-Ol m3
in. X 2.54* E+OO cm
slug catcher. The new, much higher onset-of-carry-over
flow rate was predicted well by the model experiments.
* Conversion factor is exact. SPEPE
Nomenclature
Original manuscript received in the Society of Petroleum Engineers office March 6,
d = diameter, m [ft] 1985. Paper accepted for publicalion May 28,1986. Revised manuscript received Sept.
5,1986. Paper (SPE 13724) first presented at the 1985 SPE Middle East Oil Technical
g = gravity constant, m/s 2 [ft/sec 2 ] Conference and Exhibition held in Bahrain, March 11-14.

182 SPE Production Engineering, August 1987


Discussion of Simulation of Gas/Liquid
Flow in Slug Catchers
Douglas F. North, SPE, North & Assocs.

Bos and du Chatinier's paper "Simulation of Gas/Liquid Flow in carry-over from the system. The restriction of liquid flow, how-
Slug Catchers" (Aug. 1987 SPEPE, Pages 178-82) was of interest ever, would increase the pressure fluctuations when the slugs are
to me because of my experience in the design of slug-catcher facil- received. Therefore, the advantage of minimal pressure disturbance
ities. Slug catchers provide a number of advantages in two-phase may be compromised with the better liquid distribution in the head-
pipeline systems. The primary advantages are (I) a reduced pressure ers. I would appreciate Bos and du Chatinier's comments on the
drop oh the pipeline system that can approach single-phase condi- pressure variations in the simulation studies, as well as in the oper-
tions with sufficient pigging; (2) the ability to separate gas and liquid ating facility, before and after the downcomer constrictions were
slugs effectively and to handle large quantities of liquid in a timely installed.
and controlled manner; (3) with systematic pigging, a reduction in I am not aware of any published laboratory investigations or field
the accumulation of "free water" in the lower sections of the pipe- measurements of operating slug catchers that recorded the magnitude
line, which reduces the possibility of internal corrosion; and of pressure variations when liquids are being received. If the authors
(4) minimal pressure fluctuations downstream of the facility when could provide their comments on this aspect of slug-catcher opera-
slugs are received, thus reducing the possibility of upsets in com- tion, I am sure it would provide insight for those involved in the
pression or processing equipment. operation or design of these facilities.
Downcomer constrictions on the slug catcher were used to dis-
tribute the liquids in the bottles better and thereby to reduce liquid (SPE 17173) SPEPE

Author~ Reply to Discussion of Simulation of


Gas/Liquid Flow in Slug Catchers
A. Bos, Koninklijke/Sheli Laboratorium

We agree that one of the important advantages of using a slug catcher order of 20 kPa [0.2 bar] are generated. Because of the buffering
at the end of a two-phase pipeline is that it will minimize pressure effect of the slug catcher volume, the pressure effects in the down-
fluctuations in the downstream gas-treating facilities during slug stream gas-treating facilities will be less. For the slug catcher simu-
arrival. The reason is that, because the incoming slug displaces the lated in our studies, we calculated that when the slug passes tlie
gas present in the slug catcher, the gas flow to the gas plant is not downcomer constrictions, the pressure drop over the constrictions
interrupted. In the simulation studies, the pressure effects associated increases from 2 to 20 kPa [0.02 to 0.2 bar] (at a gas flow rate
with the slug-catching operation have not been measured. of 30 X 10 6 normal m 3 /d [1,117 x 10 6 riormal ft3/D]). This pres-
As far as the situation in the field is concerned, in a few cases sure drop is on the same order of magnitude as the "natural" pres-
we measured the pressure upstream rather than downstream from sure fluctuations generated by the slug while it is still in the pipeline.
a slug catcher to detect an approaching slug. The philosophy be- Because these pressure effects are relatively small, they will not
hind this was that a slug traveling through a pipeline in hilly ter- affect the downstream equipment.
rain will generate pressure fluctuations. Indeed, it was noticed that
just before the slug enters the slug catcher, pressure effects on the (SPE 17465) SPEPE

132 SPE Production Engineering, February 1988

You might also like