You are on page 1of 11

Slug-Catcher Design for Dynamic

Slugging in an Offshore
Production Facility
M. Miyoshi, SPE, JGC Corp.
D.R. Doty, U. of Tulsa
Z. Schmidt, SPE, U. of Tulsa

Summary. A hydrodynamic model is developed for a typical offshore gathering system that consists of a pipeline, riser pipe,
and slug-catching separator. The dynamic model is capable of determining the dynamic interaction between the slug-catching
separator and the pipeline slug flow. In particular, a procedure is incorporated into the model for directly predicting the separator
foam layer associated with the crude-oil stream. The model is capable of predicting all variables necessary for slug-catcher design.
The influence of several variables-including the riser-pipe height, separator foam volume, and incoming slug size-on separator
performance is examined. The analysis shows that each is critical to proper sizing of a slug-catching separator.

Introduction
Oil and gas development in hostile environments, such as the Arc- ence ofthe riser pipe and unequal pressure losses in the gas bubble
tic area and offshore production, often necessitates transportation and liquid slug combine to alter the dynamics of how the liquid
of the produced fluids in a single pipeline. When such a pipeline enters the slug catcher. Therefore, it is the objective of this study
is sized economically for a well's initial production, it usually to generate a dynamic model that can adequately size a slug-catching
produces slug flow for a good portion of its later operational life. separator that processes a foamy crude-oil stream in slug flow.
Slug flow is characterized as the alternate flow of liquid slugs and
gas bubbles through the pipeline.
Separator Principles for a Horizontal
A liquid slug can be described as a long plug ofliquid, with much
SlugCatching Separator
smaller gas bubbles entrained throughout. A gas bubble can be de-
scribed as a long continuous gas bubble with a layer of liquid film Fig. 1 illustrates the basic construction of a slug-catching separa-
at the bottom of the pipeline. As the liquid slug moves through the tor. The fundamental purpose of such a separator is to remove free
pipeline, it tends to scoop up additional liquid from the slower- gas from the liquid phase and to deliver a relatively even supply
moving liquid film while shedding liquid from behind. In addition, of liquid to the rest of the production facility. 5.6
the turbulence caused by the scooping action of the liquid slug tends Initial separation occurs as the pipeline fluids strike the momen-
to entrain small bubbles into the liquid slug from the preceding gas tum absorber, which acts to dissipate the incoming momentum. The
bubble. These small bubbles will eventually be shed, along with liquid stream, along with its entrained gas bubbles, falls to the lower
slug liquid, into the following gas bubble. portion of the separator, where it moves horizontally with greatly
It is the nature of slug flow that slug lengths tend to grow as the reduced speed to the liquid discharge line. The section of the sepa-
slugs move down the pipeline in that shorter slugs tend to coalesce rator containing the liquid is therefore designed to avoid turbulence
into longer slugs. If the pipeline is sufficiently long, it is possible and to allow sufficient time for the entrained gas bubbles to evolve
for the liquid slug to reach a length of several thousand feet. 1 To out of the liquid stream. Also, a vortex breaker is commonly in-
accommodate slugging, it is common to install a large separator stalled at the outlet of the separator to prevent gas from being re-
(or slug catcher) at the exit of the pipeline. Because the slug catch- entrained in the exiting liquid stream.
er is the first element in the processing facility, determining its prop- As the gas stream enters the separator, it flows over the inlet baffle
er size is vital to the operation of the entire facility. into the upper portion of the separator. There it moves with great-
There are basically two types of slug catchers,2.3 the vessel and ly decreased speed, allowing liquid droplets to settle out as it moves
the mUltiple-pipe types; mUltiple-pipe separators have been widely to the gas discharge line. Enough space should be allocated for the
applied in facilities processing a gas-condensate stream. In gas- gas stream so that liquid droplets larger than a predetermined size
condensate facilities, the incoming gas is relatively clean and the have sufficient time to settle out by gravitational forces. In addi-
liquid production is low. It is common practice, however, to pig tion to gravitational separation, an internal component called a mist
such lines, with the result that a single large liquid slug is produced. extractor (or mist eliminator) is often installed to increase the effi-
Because of the extreme difference in liquid flow rates during nor- ciency of separating small liquid droplets from the gas stream. It
mal operation and pigging, the primary consideration in the design can be of the vane, wire mesh, or plate type. The mist extractor
of a slug catcher is including adequate liquid surge volume. How- functions by providing a greatly increased surface area on which
ever, the dynamic nature of pigging is different from that encoun- the liquid droplets impinge and/or a tortuous path that uses inertia
tered in normal pipeline slug flow and is therefore outside the scope to increase the likelihood of a liquid droplet impinging on the mist
of this study. extractor's surface.
For crude-oil streams, where foaming sometimes emerges as a In addition to processing and separating the gas and liquid streams,
major problem, 4 a vessel-type slug catcher is normally used. Such a separator must also be capable of processing any foam that may
a slug-catching separator must be designed to process the incom- enter. Foam is often a major problem in crude-oil systems 4 (e.g.,
ing liquid slugs produced by normal pipeline slug flow and to per- it is not uncommon for the foam to occupy half the separator's
form an adequate initial separation of the gas and liquid phases. volume). Foam starts to form immediately after the pressure in the
When the slug catcher is installed on an offshore platform, the in- wellbore or reservoir falls below the bubblepoint pressure and con-
coming pipeline usually terminates in a vertical riser pipe. The pres- tinues to form as the two-phase mixture is transported up the well-
bore and along the pipeline. After the foam enters the separator,
it begins to decay by the following three processes 4 : (1) growth
Copyright 1988 Society of Petroleum Engineers of bubbles by coalescence and gas diffusion across a common bub-

SPE Production Engineering. November 1988 563


10.000 TO 150,000 BOPD
LIGHT OIL
MIST
EXTRACTOR r-..I-.-1>Ic!-........:...----!I

PIP
.
0
0
III
NORMALGOR

."'
0:

l-
f/)
0
<>
...J 1000 pal;
w
f/)
f/)
w
>
w
>
;::

"'---
Fig. 1-Slug-catching separator. <t
...J
W
0:
100pai;

ble wall, (2) bubble rupture at the foam/gas interface, and (3) oil 0 1 2 3 4 5 6 7 8 9 10
drainage through the foam. LID RATIO
At constant flow rate, temperature, and pressure, the volume of
foam in the separator is constant and is obtained by dynamic equi- Fig. 2-Vessel cost vs. length-to-dlameter ratio (courtesy
librium between the rate of the foam decay and the rate of incom- CaE f-:latco).
ing foam. Therefore, an adequate method of predicting separator
foam volume is required to size a separator properly.
Finally, a separator must also be able to accommodate large fluc- Typical choices for retention time vary from 1.0 to 5.0 minutes,
tuations in the rate at which the liquid enters the separator. 7 The depending on the fluid physical properties, operating conditions,
fluctuations result from the presence of slug flow in the pipeline and type of slug catcher.
or from the dynamics of a slug traveling up the riser pipe into the
separator. In either case, a procedure is needed to predict the re- Liquid Surge Volume. As a liquid slug enters the slug catcher,
quired increase in separator volume. the liquid level rises in response to the input rate exceeding the dis-
charge rate. Conversely, as the gas bubble enters the separator,
Standard Nondynamlc Procedure for Sizing a the liquid level tends to fall. The additional volume required to ac-
Slug Catcher for a Foamy Crude-Oil Stream commodate the fluctuation in the liquid level is called the surge
volume. 7
Although many variations 8 ,9 in the procedure for sizing a separa- When specific information is not available on expected slug sizes,
tor exist that incorporate individual requirements and experience, then API RP 14EII lists typical surge factors for offshore instal-
we use the following procedure as our standard. lations that increase the design liquid and gas volumes. However,
After the designer has made a proper selection of the separator surge volume can be predicted more accurately with the slug char-
internals for primary separation and mist extraction and has dealt acteristics indicated by the Prudhoe Bay correlation. 1 If steady
with any miscellaneous problems, such as the presence of paraffins, state is assumed, the maximum anticipated surge volume is
wax, and sand, he can size the separator on the basis of step input
flow rates and a constant discharge rate. The procedure is to deter- Vsurge=[(UL +Ug)YsL -uLlAptL' ...................... (3)
mine the gas capacity, liquid capacity, surge volume, and foam
volume required under the current operating conditions. It should where the slug liquid holdup and liquid-slug residence time are cal-
be observed that this nondynamic procedure does not incorporate culated from the Prudhoe Bay correlation. Slugs ranging in size
any additional separator volume that might be required by the dy- from a few to several thousand feet in length have been observed
namic action of the slug flow. in the Prudhoe Bay study, where slug sizes were found to obey
a log-normal distribution about the mean. It is therefore necessary
Gas Capacity. It is possible to calculate the gas capacity using the to choose a slug of "maximum" expected size to be used in sizing
Sounders and Brown 10 correlation, which determines the maxi- the slug catcher. Experience gained from the Prudhoe Bay tests in-
mum gas velocity in a separator above which the gas stream can dicates that the largest expected slug out of 1,000 slugs corresponds
pick up foam and/or liquid: approximately to the largest slugs actually observed. Therefore,
it seems reasonable to choose a slug of this length to size the slug
catcher. The correlations associated with the Prudhoe Bay study
V max =KJ PL -Pg . ................................ (1) can be found in the Appendix.
Pg
Foam Volume. By testing about 100 crude-oil samples gathered
from around the world in a plant that simulates field pipeline/sepa-
The empirical constant, K, is called the K factor and depends on
rator systems, CaE Natco* was able to generate the following cor-
the fluid physical properties, operational conditions, and the type
relation for separator foam volume 4 :
of separator used. Typical values for the K factor range from 0.4
to 0.6 for a horizontal separator. In the absence of a procedure for Vf=FbfFpdFsFfhqo,
estimating the foam volume, however, it has been standard prac-
tice to derate the K factor by multiplying by a number between 0.06 Fjh=Zj-64 . ....................................... (4)
and 0.5 to size a vessel when foam is expected. The best value to
use within this range has often been difficult to determine. Current The basic foaminess factor, F bf' which depends on the fluid phys-
expertise 4 has made it possible to estimate the foam volume direct- ical properties of the crude-oil stream, typically ranges from 0.5
ly and therefore to eliminate the need to derate the K factor. * to 10.0; for the crude oil used in this study, it is 1.70. The pipe-
diameter factor, F pd ' usually has little influence on the separator
Liquid Capacity. The minimum liquid volume required to ensure foam volume for the light-oil and high pipeline velocities consid-
sufficient time for the gas entrained in the liquid to escape is calcu- ered in this study and is taken to be 1.0. Likewise, the actual GOR
lated from the desired retention time. The retention time is related factor, Fs ' usually has only moderate influence unless the GOR is
to liquid retention volume by very high or very low; in this study, it is taken to be 1.0. The foam-
height factor, F fh , is a function of the foam height in the separa-
V ret =t ret 60qiL' .................................... (2) tor. Foam height is an important variable in determining the decay
rate of the separator foam layer. It should be observed that for
'Personal communication with M.L. Rooker, R&D, CE Natco, Tulsa, OK (1986). 'Personal communication with M.L. Rooker, R&D. C-E Natco, Tulsa. OK (1986).

564 SPE Production Engineering, November 1988


LOW HIGH NORMAL
LEVELS LEVELS LEVELS
Fig. 3-Separator liquid and foam levels.

horizontal cylindrical separators, the elevation of the foam layer TABLE 1-DATA FOR THE STANDARD SEPARATOR
within the separator can significantly alter its thickness and there- DESIGN PROCEDURE
fore its decay rate. The correlation for Ffh is generally valid for
an average 35 API [0.85-g/cm 3 ] oil at 100 to 1,000 Scf/STB [18 q iL, 1t3/sec 3.133
to 180 std m 3 /stock-tank m 3]. Finally, separator foam volume is q ig. 1t3/sec 14.838
found to be directly proportional to the incoming oil flow rate. U L. ftlsec 4.03
ug ftlsec 19.09
Vessel Shape. For a cylindrical slug catcher, the length-to-diameter YsL 0.49
ratio determines the shape of the separator. A proper choice of the tL. seconds 83.14
vessel's length-to-diameter ratio is important because it determines Lsi. It 2,446
the fabrication costs. 12 The design considerations defined above PL. Ibm/lt3 50.53
P g , Ibm/lt3 1.39
can also change as the vessel's shape changes. For example, if a 400
Psp' psi a
cylindrical vessel's shape changes without changing its internal T sp ' OF 140
volume, then the equilibrium foam volume predicted by Eq. 4 in- Bo 1.07
creases as the diameter increases. 4 Some studies 13 have been per- t ret minutes 3
formed to determine the optimal length-to-diameter ratio, where K factor. fUsee 0.6
a ratio of four seems to be a reasonable value. Likewise, research Separator length-to-diameter ratio 4
indicates that a ratio of four is a good choice for the separator pres- Ap. 1t2 0.777
sure cOllsidered in this study (see Fig. 2).

Allowances for Internals. Allowances must be made for the in- out of 1,000 slugs; its properties are listed in Table 1. Substitution
stallation of any internals in the separator. For example, Sigales 14 of these quantities into Eq. 3 determines the liquid surge volume
assigns 20% of the separator's diameter (or at least 10 in. [25.4 to be 471. 7 ft3 [13.36 m 3 ]. If this volume is combined with the
cm)) as the minimum space for the vapor phase. He also assigns liquid capacity required by the retention time and the result is divided
10% of the separator's diameter (or at least 5 in. [12.7 cm as the by the separator's length, then the resultant area corresponds to
minimum space for the liquid. In addition, the liquid height should the high liquid level in Fig. 3.
be sufficient to prevent the gas from being entrained in a liquid
vortex at the inlet to the liquid discharge line. 15 Foam Volume. Eq. 4 gives the foam volume for an average 35 API
[0.85-g/cm 3 ] oil at 140F [60C]. The equation is generally valid
Sample Calculation With the Nondynamic for GOR's ranging from 100 to 1,000 Scf/STB [18 to 180 std
Standard Design Procedure ft 3/stock-tank m 3].
The above procedure is implemented here for a hypothetical ex-
ample; Table 1 contains the relevant information. The slug catcher Vessel Shape. Although a number of different vessel shapes could
is to be designed to accommodate an inflow of 45,000 STBID [7154 be tried, only the single length-to-diameter ratio of four for the cylin-
stock-tank: m3 /d] with a 35API [0.85-g/cm 3 ] gravity, and a GOR drical separator will be examined.
of 800 Scf/STB [144 std m 3 /stock-tank: m 3 ] with a 0.75 specific
gas gravity. The separator is to operate at 400 psia and 140F [2.8 Calculation Procedure. The desired vessel has a minimum cross-
MPa and 60C]. sectional area while satisfying Eqs. I through 4 and also a length-
to-diaIl1eter ratio of four. If a guess of 8.63 ft [2.63 m] for the ves-
Gas Capacity. The maximum allowable gas velocity in the sepa- sel diameter is made (length equals 34.52 ft [10.52 m)), then the
rator is 3.567 ft/sec [1.087 m/s], calculated with a K factor of 0.6 cross-sectional area occupied by the retention and surge volumes
in Eq. 1. Dividing this velocity into the gas flow rate produces the is 30 ft2 [2.787 m 2]. With the minimum area required for the gas
minimum allowable cross-sectional area for the gas phase, which phase, the remaining area and height for the foam can be calculat-
is 4.160 ft2 [0.3865 m 2]. Referring to Fig. 3, this area corresponds ed (see Fig. 3). This results in a foam height of 37.9 inches (96.3
to the gas during high liquid and foam levels. cm] and a foam volume of 840 ft 3 [23.8 m 3]. When this foam
height is substituted into Eq. 4, it yields 820 ft3 [23.2 m 3], which
Liquid Capacity. The choice of 3 minutes for the retention time is close enough to be acceptable. If the foam volume calculated with
requires a minimum retention volume of 563.9 ft3 [15.97 m 3], the guessed vessel's diameter does not agree with the foam volume
where Eq. 2 was used for the calculation. Referring to Fig. 3, if calculated with Eq. 4, then the vessel diameter is corrected and
this volume is divided by the separator's length, then the resultant the process repeated.
area corresponds to the low liquid level. This procedure does not consider the dynamic nature of normal
pipeline slug flow. In addition, the existence of a riser pipe com-
Liquid Surge Volume. The Prudhoe Bay correlations, used to monly associated with offshore production facilities tends to am-
predict the properties of the design slug corresponding to a pipe- plify the dynamics of the slugs entering the slug catcher. Therefore,
line of the above size and flow conditions, can be found in the Ap- to understand the dynamic nature of slug catcher performance bet-
pendix. The design slug corresponds to the largest expected slug ter, the following dynamic model was developed.
SPE Production Engineering. November 1988 565
r
~ SYSTEM CONSIDERED SYSTEM CONSIDERED ~~_ _ _ _ _ _ _ -=...,
, i
i
I,
I
I IIh SLUG UNIT
TO
SLUG
CATCHER

bI ~~~~Jj)o:z::Jf+. L j _ l -_
p..:'----Lj--- j' _-
--.l.
'USER-PIPE

I GAS SCRUBBER
2 IP Sf PAR ATOR
3 FLASH DRUM
TO SURGE TANK

TO WATER TREATMENT
UNIT

i,....-,_....-----L
\4--------Lj+1 - - - " -
-,
4
5
6
TREATER
COMPRESSOR
HP SEPARATOR ISLUG CATCHER)
p
PIP!UN

Fig. 4-81ug flow in a pipeline/riser-pipe system. Fig. 5-Production facility.

Description of Dynamic Normal Slug Flow in a Dynamic Normal Slug Flow Mode1 16,17
Pipeline/Riser. Pipe System Connected A pipeline containing a number of liquid slugs and gas bubbles is
to a Separator shown in Fig. 4, where the bubble-front velocities, distance to the
Figs. 1 and 4 illustrate normal slug flow in a pipeline/riser-pipe slugs, pressure, and volumes of the bubbles are also shown. The
system connected to a separator. The precise dynamic action ofthe following assumptions were made in the construction of the slug
liquid slugs and gas bubbles is a consequence of the configuration flow model.
of the total system, including inlet flow rates, fluid physical prop- I. The pipeline operating conditions produce slug flow.
erties, pipeline geometry, riser pipe, slug-catching separator, con- 2. Only a portion of the pipeline needs to be considered.
trol system, and any secondary-separation facilities. The 3. Physical quantities, such as velocity and pressure, for each
secondary-separation facilities (Fig. 5) might include an intermediate gas bubble and liquid slug can be represented by their average
pressure separator, treater, gas scrubber, and additional control sys- values.
tems. However, the objective of this paper is to study the dynamic 4. The errors introduced in sizing a slug-catching separator as-
influence of large liquid slugs on the performance and design of suming instantaneous transmission of pressure information along
a slug-catching separator. Therefore, it is reasonable to limit the the length of the pipeline are negligible.
dynamic system to include only the last portion of the pipeline, riser 5. Pressure drop in the gas bubble is assumed to be negligible
pipe, slug-catching separator, and associated control system (see when compared with that of the liquid slug and therefore is not con-
Figs. 4 and 5). sidered.
Figs. I and 4 indicate the dynamic behavior of a liquid slug and 6. The gas flow is assumed to be isothermal.
associated gas bubble as they travel along the pipeline up the riser 7. The mass of a gas bubble remains unchanged as it travels along
pipe and into the separator. When the liquid slug enters the base the length of the pipeline, although its volume changes in response
of the riser pipe, it is expected to decelerate in response to increas- to pressure changes.
ing gravitational forces, resulting in the compression of the trail- 8. For the pressure fluctuations encountered, the compression
ing gas bubble. This continues until the bubble pressure becomes of the liquid slug is negligible compared with that of the gas bub-
sufficient to overcome the gravitational forces, allowing the liquid ble; therefore, the length of the liquid slug is assumed to be con-
slug to continue up the riser pipe into the separator. The trailing stant in both the pipeline and riser pipe.
gas bubble eventually enters the base of the riser pipe, decreasing 9. The physical properties and motions of the gas bubbles and
the gravitational forces, causing the tail of the liquid slug to ac- liquid slugs are assumed to be represented by the correlation de-
celerate up the riser pipe and shoot into the slug-catching separa- veloped in the Prudhoe Bay tests (found in the Appendix) and are
tor with a significantly higher-than-average velocity. It is at this therefore correlated with bubble-front velocity.
time that the liquid level in the slug-catching separator is expected Consider the ith slug unit in Fig. 4. The gas bubble-front veloci-
to be at its highest. Correspondingly, the volume remaining for the ty is related to bubble-front position by
gas flow and foam is at a minimum.
As the bubble enters the slug-catching separator, the separator dL;
pressure rapidly increases in response to the higher bubble pres- vbi=-' ........................................ (5)
sure. This effect combines with the minimum available gas volume dt
to produce the highest gas velocity through the separator. It is dur-
The tests performed at Prudhoe Bay I indicate a simple linear rela-
ing this critical time that foam carry-over into the gas line can occur
if the slug-catching separator is insufficiently sized. As the gas bub- tionship between bubble-front velocity and liquid-slug mixture ve-
ble is produced, liquid production from the liquid film associated locity:
with the bubble is low. Thus, the liquid level in the slug-catching
Vbi=CoVmi, ...................................... (6)
separator falls, eventually reaching its minimum. When the next
liquid slug enters the base of the riser pipe, the whole process repeats
where the constant Co was taken to be 1.25. Conservation of linear
itself, changing in magnitude in response to the size of the next
momentum for the liquid slug yields
liquid slug and gas bubble.
As displayed in Fig. I, the response of the slug-catching separa-
tor to incoming slugs is determined by the valves on the gas and d fLPLYsLLsiV~i
liquid flowlines exiting the separator. The gas valve is usually con- PLY sLLsi-(Vmi) =g c l44(Pi -Pi-I) - ....:::.'--'----
dt 2dp
trolled by a pressure regulator, which is intended to keep the sepa-
rator pressure nearly constant. The liquid dump valve is usually -PLYsLgZi, ...................................... (7)
operated by a level in the separator. Proper sizing of both valves
and proper adjustment of their respective control systems are re- where Zi is the difference between the height of the nose and tail
quired to ensure that both the slug-catching separator and any ad- of the ith liquid slug. Differentiating the ideal gas law applied to
ditional downstream separation facilities function properly. the gas bubble gives the change in pressure with respect to time:
A hydrodynamic model was developed in which the variables of
interest were related to each other through a system of ordinary d vb,i+l-vb,i
differential equations based on fundamental physical principles. A -PI= Pi- ......................... (8)
description of the model follows. dt Li+I-Li-Ls,i+l
566 SPE Production Engineering, November 1988
Separator Inlet Flow Rate. where it is assumed that the bubble of the ith slug unit is entering
As the ith slug unit exits the pipeline and riser pipe, it becomes the separator.
the input for the slug-catching separator. When the liquid slug is In contrast to the gas volume, the dynamic behavior of the foam
being produced, assuming that all the small bubbles are moving volume is much more difficult to specify. As described earlier, the
with the same velocity as the liquid, the separator inlet flow rates are average decay rate of the foam, and therefore the resultant steady-
state foam volume, is influenced by several different parameters,
qiL =VmiApYsL .................................... (9) including the incoming flow rate and foam height. The foam height
in turn depends on the geometry of the separator and the liquid level.
and Likewise, the liquid level is related to the liquid volume and the
geometry of the separator by
qig =VmiAp(1-YsL)' ............................... (10)
VL =Lsp(dsp)2[cos -I (1- 2fLs)
In addition, when the gas bubble is entering, continuity considera-
tions for the slug unit gives the film velocity and the gas velocity: -2(1-2fLs).JfLs(I-fLs) ]/4, ....................... (19)

where fLs is the liquid height as a fraction of separator diameter.


Therefore, in this study, we will assume that the behavior of the
and foam volume is adequately described by the process defined earli-
er in Eq. 4.

Separator DI.charge Flow Rate.


Therefore, the inlet flow rates during bubble production are Each discharge line exiting the separator consists of a control valve
and a segment of line. 18,19 Any other restrictions, such as piping
qiL =vLjiApYbL ................................... (13) components or heat exchanger, can be included by use of equiva-
lent length.
and For the liquid discharge line, the relationship between the liquid
flow rate and pressure loss across the liquid control valve is given
qig=VgiAp(1-YbL)' ............................... (14) by 20

Dynamic SlugCatchlng Separator Model


Fig. I illustrates the configuration of a typical single-vessel slug-
catching separator. The separator contains a number of internal com-
qL=2.228xlO -3 FLv J ilpVL
--FL(fVL),
'YL
.............. (20)

ponents, each of which is designed to increase the efficiency of the


separation and slug-catching processes. For example, the separa- where FLv is the valve flow coefficient and FL(fVL) relates the
tor may contain a momentum absorber, mist extractor, and vortex valve flow characteristics to the valve stem position. In addition,
breaker. To generate a dynamic model, however, we assume that the pressure loss across the liquid discharge line can be calculated
the influence of each internal component is contained in the value from
of specific parameters defined by the model. For example, the ap-
propriate choice of a maximum allowable K factor for mist extrac-
tion from the gas phase would be influenced by the type of mist .......... (21)
extractor installed.
The dynamic behavior of the separator therefore reduces to de-
termining changes in the liquid and foam volumes, along with where the gravitational term assumes that the elevation of the dis-
changes in separator pressure. To accomplish this, the conserva- charge line is identical to the elevation of the bottom of the separa-
tion of mass for the liquid phase in the separator, coupled with the tor. The flow rate through the valve is limited by the maximum
assumption of incompressibility, results in flow rate, which occurs at critical flow. The criterion for critical
flow is 21
dVL
- =qiL -qL, .................................. (15)
dt

where qiL and qL are the inlet and discharge liquid volumetric flow
rates, respectively. The conservation of mass for the gas phase in For the gas discharge line, the relationship between the gas flow
the separator and the ideal gas law yield rate and pressure loss across the gas control valve is given by22

dVg Vg dpsp
--+---=qig-qg' ......................... (16)
dt Psp dt

where qig and qg are the inlet and discharge gas in-situ volumet- y= 1-{3/(3fp Fh), ................................ (23b)
ric flow rates, respectively. The gas volume during liquid slug pro-
duction is defined as and

Vg=Vsp-VL-Vj, ................................ (17) (3=ilpvg/Pu, ................................... (23c)

where. Vsp is the total volume of the ~eparator and Vj is the volume where Fgv is the val~e. flow coeffi~ient and Fg(fVg) represents the
OCCUPIed by the foam. However, dUrIng gas bubble production, the valve flow characterIstIcs as a functIon of valve stem position. Also,
gas volume is augmented to include the gas volume contained in the pressure loss across the gas discharge line is given by
the incoming bubble:
l6fgP gL gqrg
2 5' ................. (24)
'If 2g c l44dDg

SPE Production Engineering, November 1988 567


TABLE 2-DATA FOR HYPOTHETICAL EXAMPLE

Pipeline length, It 10,000


Pipeline ID, in. 11.938
VALVE FLOW Riser-pipe height, ft 150
LEVEL CONTROLLER CHARACTERISTICS Oil production, ST81D 45,000
GOR 800
Z ~ 1.0.------..-----. 1.01.---------,. Oil gravity, API 35
~~ Specific gas gravity 0.75
Separator pressure set point, psia 400
~~
"- .... Separator temperature, OF 140
2 ....
~O
Liquid discharge line length, It 2,000
Cl)Z Liquid discharge line ID, in. 10.02
",9
~t; Liquid discharge line end pressure, psia 100
<I'" Gas discharge line length, It 3,000
::> ~ 0.0.~..L--.1-.J~-......J
- 0.0 LL HL 1.0 1.0
Gas discharge line ID, in. 10.02
LIOUID LEVEL LlOUID FLOW Gas discharge line end pressure, psia 380
(FRACTION OF VESSEL DIAMETER) (FRACTION OF MAXIMUM FLOW) Separator retention time, minutes 3
K factor, ftlsec 0.6
Fig. 6-Level controller and valve flow characteristics. Separator length-to-diameter ratio 4

As before, the maximum flow through the gas control valve is limit- 5) downstream of the slug-catching separator, such as an intermedi-
ed by the critical flow condition: ate pressure separator, treater, or gas compressor, will impose ad-
ditional constraints on an acceptable valve and control system.
i1PVg ?:'!pFhpsp' ................................. (25) However, the scope of this study does not directly include the in-
fluence of the downstream equipment; therefore, only typical choices
Eqs. 20 and 23 require the definition of the function that relates will be made in sizing the valves and adjusting the control system.
valve flow characteristics to valve stem position. Several commonly For a good range of control, the liquid-control valve usually has
used characteristics include the following: linear, a valve flow coefficient 1.25 to 2.0 times larger than that required
to pass the average flow. In this study, the valve flow coefficient
F(x)=x, was adjusted so that the valve would pass 130% of the average flow
when fully open. This choice was made to ensure that the down-
square root, stream facilities are not subjected to excessively large fluctuations
in the liquid flow rate.
F(x)=Vx, The liquid-control system consists of defining the valve flow char-
acteristics and the liquid-level controller set point and gain. As Fig.
and equal percentage, 6 illustrates, both curves are assumed to obey a linear relationship,
and together they define the relationship between the separator liq-
F(x)=<5x - 1 , ................................... (26) uid level and the liquid flow rate through the control valve. In an
actual field facility, the set point and gain of the level controller
where the constant <5 typically ranges from 20 to 50. For this study, are usually adjusted in response to experience with the perform-
a linear relationship was assumed to describe the valve flow char- ance of the total system. For this study, however, it is convenient
acteristics as a function of valve stem position. to fix the procedure for setting the liquid-level controller so that
comparisons can be made with the dynamic model.
Control Systems for the Gas and Fig. 3 shows the fluctuations of liquid level expected in a sepa-
liquid Control Valves rator designed with the standard procedure and a slug of maximum
A controller compares the process signal sent by the transmitter expected size. Because the liquid level should not fall below the
with a set point and produces a control signal for the control valve, low liquid level (LL), it is natural to take LL as the set point where
which in turn regulates the flow. For liquid-level control, a propor- the valve closes. Also, because the liquid should not rise above the
tional controller 20 is often sufficient. Disregarding valve dynam- high liquid level (HL), it is convenient to adjust the gain of the liquid-
ics, a proportional controller is defined by level controller so that the control valve is fully open at this level.
When the dynamic model is run with this procedure to set the liquid-
\evel control system, it may be necessary to increase the size of
the separator to compensate for the influence of the dynamic effects.
where KLc is the gain of the control system. The values of the gain The separator pressure controller and valve are intended to main-
and the set point are adjusted to ensure that the liquid-control valve tain a nearly constant separator pressure. The gas-valve flow coeffi-
closes and opens in response to specific liquid levels in the separator. cient was chosen to give a maximum flow of 130 % of the average
Separator pressure is often controlled with a proportional integral flow. Also, the gas-valve flow characteristic was assumed to be
controller 20 : linear. In addition, the set point was taken as the desired separator
pressure, while the gain and reset time were adjusted to ensure that
pressure fluctuations did not exceed 10 % of the separator pressure.

Initial Conditions
The dynamic model consists of a system of algebraic and differen-
where Kgc is the gain and fIg is the reset time. Proper choices for tiill equations. It is therefore necessary to initialize the variables
the gain and reset times are required to ensure that pressure fluctu- that appear in differential form in some reasonable fashion.
ations within the separator remain relatively small. The pipeline and riser pipe contain slugs for which size and phys-
ical properties must be specified. The Prudhoe Bay tests indicate
Determination of ControlValve Sizes and that slug sizes follow a log-normal distribution. For the current
ControlSystem Adjustments model, it is reasonable to initialize the slug sizes and physical prop-
The proper choice and adjustment of a control valve and associat- erties in terms of the correlations developed in that study. In par-
ed control system are critical to the successful operation of a pro- ticular, the study gives a correlation for the ratio of bubble length
duction facility. In particular, any additional equipment (see Fig. to liquid-slug length, R bL , for liquid holdup, and for the various

568 SPE Production Engineering, November 1988


30

25
~
~
0.8
~20
j 0,7
FOAM LEVEL
iii ~i
.... ~-;
0.6
~ 15
~.~ o.~
~
~
~~ 0.4
LIQUID LEVEL.

J ~g 0.3
IL
'00
0 10 cZ
5
0
5~ 0.2 80~
0", Z
::; :i 0.1 60 w
5
.., .......... -- -------- -. '" .. ~ISCHARGEI:LOW RATE
----- -------- ---. ----._._-. _.
0.0 40~t
0
0 50 100 150
INLET FLOW RATE

200 250
'"

300 350
'"

- 400
0~1---5~6---'16-0--~I~~--200~I---2~;O--~3~~--3r~---.T~~O
1
_ 20 ~

TIME, (sec)
TIME. (c)

Fig. 7-Separator liquid flow ratf3s for a maximum slug. Fig. a-Separator liquid and foam levels for a maximum slug.

component velocities. Therefore, at the initial time, the bubble pres- stock-tank m 3 /d] with a GOR of 800 scf/STB [144 std m 3 /stock-
sures are assumed to be calculated from Eq. 7, where the time tank m3 ] into a separator. The slug-catching separator was sized
derivative of the mixture velocity is assumed to be zero, and the with the dynamic model, with a K factor of 0.6 and a minimum
bubble-front velocity is calculated to correspond to the average pipe- liquid retention time of 3 minutes. Also, to ensure that the dynam-
line mass flow rates. In addition, the distances from the separator ic liquid surge volume was adequately predicted, a slug of maxi-
to the bubble front can be initialized by mum expected size was included along with smaller slugs in the
pipeline. The characteristics of the maximum expected slug can be
Li=L i - 1 +RbLLs.i-l +L si , ........................ (29) found in Table 1. Figs. 7 through 10 display the results.
Fig. 7 shows the dynamic inlet and discharge liquid flow rates
where L S i is the length of the ith liquid slug. The initial length of
for the separator. Observe that it takes 83 seconds for the maxi-
each slug in the pipeline can be chosen randomly according to the
mum liquid slug 2,446 ft [746 m] long to be produced up the riser
distribution given in the Prudhoe Bay correlation. To illustrate the
pipe into the separator, while bubble production requires 225 sec-
range of behavior produced in a slug-catching separator better, how-
onds. During liquid-slug production, the liquid inlet flow rate in-
ever, some of the initial slugs were chosen to have an average or
creases from about three times the average rate to a peak of nine
maximum expected length.
times. The peak inlet liquid flow rate coincides with the slug tail
exiting the riser pipe and shooting into the separator. As shown
Boundary Conditions in Fig. 8, this also coincides with the maximum separator liquid
The bubble-front velocity of the slug unit currently entering the and foam levels.
pipeline inlet is assumed to be constant. As each slug is produced When the bubble is produced, the inlet liquid flow rate falls dra-
into the separator, another slug is introduced into the pipeline inlet matically, allowing time for the separator to purge the excess liq-
that has gas and liquid masses identical to those of the slug being uid. As the liquid level falls, the liquid dump valve restricts the
produced. Also, at the other end of the system, the pressures at liquid flow, preventing the separator liquid volume from falling
the end of the gas and liquid discharge lines are assumed to be below that required to maintain a 3-minute retention time. When
constant. bubble production is complete, the liquid and foam are at their lowest
levels and the whole cycle repeats.
Evaluation 01 the Dynamic Model's Predictions Fig. 9 shows the gas inlet and discharge separator flow rates for
A hypothetical example allows analysis of the dynamic interaction the same slug cycle. Observe that both the inlet and discharge flow
of slug flow in a pipeline/riser-pipe system producing into a slug- rates fluctuate dramatically in response to the dynamic nature of
catching separator. The parameters defining the example are listed the slugs entering the separator. As expected, the critical time for
in Table 2. In particular, a 1O,000-ft [3048-m] pipeline terminat- gas flow happens just as the bubble enters the separator, which
ing in a 150-ft [45.72-m] riser pipe produces 45,000 STBID [7154 occurs just after the peak liquid level and peak inlet liquid flow

30

SEPARATOR PRESSURE

h
\~
0,6
u INLET FLOW RATE ~ 0.5
~
~
15

.......
:
.......... /\ ~
, \i
...............................
DISCHARGE FLOW RATE :
~ : ;

~0.3
0.4

100
'"
!;( 10
\.f '.' L., ~ 0.2 K-FACTOR
aog
'" ...........r .. ..........
z
~
~ 0.1
t.; :"i 60 ~it
IL
(/) ::
0,0 ,i
ii
r VALVE
OPENING
40 ~
>
~ ~
""<:> O+---~----~--~----~---r----r----r--~ ~--~--'----r---.----r---~--~---+O
\J 20

o 50 100 150 200 250 300 350 400 o 50 100 150 200 250 300 350 400
TIME. < e)
TIME. (sec)

Fig. 9-Separator gas flow rates for a maximum slug. Fig. 1O-Separator pressure and K factor for a maximum slug.

SPE Production Engineering, November 1988 569


30
08

25 .:; 07 FOAM LEVEL

~
u ~i
.... ;;

0.6 -~
~
~20 ::E 05

~~
...
..j

0:: 15 5ET FLOW RATE


~
0'
g
04
03
.r--
~
LIQUID LEVEL

100
~ :; ~ 02

~~~ tf
0
....
...o 10
0",
:. 0.1
80

60 ~
'5 0.0 Q.-
~t

If """-1 "J (
":::; 5 FLOW RATE
40
20 ~
.... - .. -._--- ~
'''J''''
",

6 sO I
100 1;'0 200 2;'0 300 350 4X,o
0 TIME. (sec)
0 50 100 150 200 250 300 350 400
TIME, (seC)
Fig. 12-Separator liquid and foam levels for average-size
Fig. 11-Separator liquid flow rates for average-size slugs. slugs.

rate, As Fig. 10 shows, the gas K factor also peaks at this time, derating factor tends to give unreliable results compared with direct
as the separator pressure increases to equalize with that of the in- prediction of foam volume. 4
coming gas bubble, The dynamic model can also be compared with the standard de-
To understand better the dynamic effect of large slugs on sepa- sign procedure in terms of separator-size predictions, Table 3 con-
rator performance, the above example is contrasted with the case tains four separators, each sized with the standard design procedure
where only average-size slugs are in the pipeline. The results of assuming different combinations of liquid-slug surge volumes and
this case are presented in Figs, 11 through 14, where the average separator foam volumes. The slug and foam volumes were calcu-
liquid slug length is predicted by the Prudhoe Bay correlations to lated with the information in Table 2. As Table 3 shows, separator
be 441 ft [134 m]. In Fig, 11, peak inlet liquid flow rate is com- volume must increase as slug and foam volumes are incorporated
parable to that predicted when a maximum slug enters the separa- into the design. In particular, if a separator is designed with both
tor. In Figs, 12 through 14, however, the dynamic fluctuation in liquid-slug surge and foam volumes ignored, then it will be signif-
separator liquid level, pressure, and K factor are all significantly icantly undersized (having only one-third of the required capacity).
reduced, The separator volume predicted with the dynamic model was
The magnitude of the dynamic fluctuations predicted by the model 2,069 ft3 [58.6 m 3] with a foam volume of755 ft3 [21.4 m 3]. This
depends not only on slug size but also on the presence of the riser volume is only slightly larger than the 2,019 ft3 [57,2 m 3] pre-
pipe at the entrance to the separator, It is expected that the liquid dicted by the standard design procedure. However. it should be
slug will deacce1erate as it enters the riser pipe in response to gravita- noted that a number of opposing dynamic effects combine to result
tional forces, This is confirmed by the model's predictions, which in such close predictions. In particular, the dynamic model pre-
indicate an increase of 12% in the time required to produce the dicts a peak gas flow' rate 90 % higher than that predicted by the
liquid slug. As a result, for identical incoming slugs, the liquid level standard design procedure. Therefore, the minimum separator area
in the separator will tend to rise higher when no riser pipe is pres- allocated for the gas flow predicted by the dynamic model must
ent. However, this effect is countered by the fact that the trailing be 90% higher-as well. As observed above, however. the high sepa-
gas bubble does not pressurize in the absence of a riser pipe. There- rator liquid and foam levels predicted by the dynamic model were
fore, the peak inlet liquid flow rate is predicted to be lower by 49%, 12 % lower than the standard design procedure. The net effect was
while the peak gas K factor is predicted to be lower by only 3 %. for the change in separator size to be small relative to the separa-
Consequently, the separator size required with or without a riser tor's total capacity.
pipe is about the same, but the dynamics is significantly different. Finally, as Fig. 15 illustrates, a two-train separator was consid-
The required separator foam volume is also important; foam ered. The purpose of this example was to determine whether the
volume for the above example is predicted to occupy 37% of the two-vessel design has any substantial advantage over the single-
total vessel volume, If foam is not incorporated into the dynamic vessel design. The dynamic model was used to size the two ves-
model, then the gas K factor would have to be derated from 0,6 sels, resulting in a predicted diameter of 6,65 ft [2,03 m] for each
to 0,13 to predict a vessel of the same size, This corresponds to vesseL Compared with a single-vessel separator, the total volume
a derating factor of about 0.22. However, guessing an appropriate required for two vessels is smaller by 12.0%. However, the total

~
30
420
410 UJ
g:'O
400 Q:"Wj:

2 o.s
~
o
:- 20 1 ..
390 ~
380
0::
fu
U)
c;
.
u
~
IL
15

~ 100
<5
~ 10 80 z
Z
0:: 60 w
~ '" ,,:"'1 ""-" ~~
40 w
~ 5 V ~./ ~~FNa--\ jj \/
20
:;
CI)
" ~
"" O+----r----.---~--~----r_--~--_,--~
a 50 100 150 200 250 300 350 400
0

TIME, (S.C)
o 50 100 150 200 250 300 350 400
TIME. (sec)
Fig. 14-Separator pressure and K factor for average-size
Fig. 13-Separator gas flow rates for average-size slugs. slugs.

570 SPE Production Engineering, November 1988


TABLE 3-VESSEL SIZES CALCULATED BY THE STANDARD DESIGN PROCEDURE

Slug Foam Vessel Vessel Normal Normal


Volume Volume Diameter Volume Liquid Level Foam Level
~ (ft 3 ) (tt) ~ (fraction of d se) (fraction of d se)
0 0 5.95 662 0.793
473 0 7.17 1,158 0.490
0 980 8.11 1,676 0.370 0.865
473 839 8.63 2,019 0.323 0.511

weight, which corresponds to fabrication costs, is only about 7.6%


smaller. Therefore, once the additional cost of instrumentation is
considered, any potential advantage of a two-vessel slug-catching
separator is small.

Accuracy of Predictions
The dynamic model is capable of predicting all parameters required
in the design of a slug-catching separator. The accuracy of these
predictions depends not only on the dynamic behavior of the sys-
tem but also on the nature of the slug flow entering the separator.
Given an accurate prediction of the slug characteristics generated
within the pipeline, we feel that the dynamic model can do a good
Fig. 15-Two-vessel slug-catching separator.
job in predicting separator performance. The Prudhoe Bay corre-
lations were used as a basis for the model's slug-characteristic pre-
dictions because it was the most appropriate correlation available i g = gas friction factor
at the time of this study. In addition, the criterion that the largest fL = liquid friction factor
slug expected out of 1,000 slugs predicted by that correlation be iLs = separator liquid level, fraction of diameter
used as the design slug was based solely on the experience gained ip = rate pressure-drop ratio
in the Prudhoe Bay study. Because ofthe sensitivity of the model's
predictions on the basis of the design slug, if superior techniques
ivg = gas-valve stem position, fraction of fully open
iVL = liquid-valve stem position, fraction of fully open
become available, they should be incorporated into the dynamic
model to enhance its accuracy. Fbf = separator basic foaminess factor
Ffh = separator foam-height factor
Conclusions Fgv = gas-valve flow coefficient
1. A dynamic model was developed that is capable of sizing a Fh = separator-heat-ratio factor
slug-catching separator and determining the dynamic interaction be- FLv = liquid-valve flow coefficient
tween the separator and the incoming slug flow. Fp = pressure-recovery factor
2. A reasonable estimate of the maximum slug size is critical in Fpd = separator pipe-diameter factor
adequately sizing a separator. Fs = separator actual GOR factor
3. A procedure is described that directly predicts separator equi- g = acceleration of gravity, ftl sec 2 [ml s 2]
librium foam volume, eliminating the guesswork in derating the gc = units conversion factor
gas K factor. K = separator gas K factor
4. Large dynamic fluctuations in separator inlet gas and liquid
K gc = gas-control gain
flow rates, liquid and foam levels, and the gas K factor are pre-
dicted by the dynamic model when a slug-catching separator must KLc = liquid-control gain
process large incoming slugs traveling up a riser pipe. Lg = length of gas discharge line, ft [m]
S. As riser-pipe height increases, the magnitude of the dynamic Li = distance from separator to ith slug bubble front,
fluctuations also increases, and proper consideration must be given ft [m]
to determining adequate separator area for the gas flow. However, LL = length of liquid discharge line, ft [m]
as riser-pipe height increases, the liquid slug also decelerates in Lp = length of pipeline, ft [m]
the riser pipe, diminishing the required increase in separator size. Lsi = ith liquid slug length, ft [m]
6. As predicted slug size increases, the associated dynamic fluc- Lsmax = maximum expected liquid-slug length, ft [m]
tuation also increases, along with the required separator capacity. Lsp = separator length, ft [m]
7. The standard design procedure is adequate for sizing slug-
Pc = critical pressure, psia [kPa]
catching separators as long as incoming slug sizes and separator
PDg = gas discharge line end pressure, psia [kPa]
foam volume are properly considered.
8. From a process standpoint, a two-vessel slug-catching sepa- PDL = liquid discharge line end pressure, psia [kPa]
rator does not show a significant advantage over a single-vessel Pi = bubble pressure in ith slug, psia [kPa]
separator for the single case examined. Psp = separator pressure, psia [kPa]
Pu = upstream pressure, psia [kPa]
Nomenclature Pv = vapor pressure, psia [kPa]
QO,1,2 = regression coefficients !::.PVg = gas-valve pressure drop, psi [kPa]
Ap = pipe cross-sectional area, ft2 [m 2] !::.PVL = liquid-valve pressure drop, psi [kPa]
Bo = oil FVF qg = separator discharge gas flow rate, ft 3 /sec [m 3 Is]
Co = constant qig = separator inlet gas flow rate, ft 3 /sec [m 3 Is]
d = pipeline JD, in. [cm] qiL = separator inlet liquid flow rate, ft 3 /sec [m 3 Is]
d Dg = diameter of gas discharge line, ft [m] qL = separator discharge liquid flow rate, ft 3 /sec [m 3 Is]
dDL = diameter of liquid discharge line, ft [m] q0 = oil production, 10 3 BID [m 3 I d]
dp = pipeline diameter, ft [m] RbL = ratio of slug-bubble length to liquid-slug length
d sp = separator diameter, ft [m] Rv = ratio of slug bubble-front velocity to mixture
e = error velocity
SPE Production Engineering, November 1988 571
t = time, seconds 14. Sigales, B.: "How to Design Reflux Drums," Chem. Eng. (March 3,
tIg = gas control reset time, min/repeat 1975) 82, No.5, 157-60.
tL = liquid slug residence time, seconds 15. Moduffie, N.G.: "Vortex Free Downflow in Vertical Drains," AIChE
1. (Jan. 1977) 23, No.1, 37-40.
tret = separator liquid retention time, minutes 16. Furukawa, H.: "Slug Catcher Design and Simulation of its Behavior
T = temperature, of [0C] During Slug Flow," U. of Tulsa, Tulsa, OK (1981).
Tsp = separator temperature, of rOC] 17. Giozza, W. F.: "Simulation of Gas-Oil Separator Behavior under Slug
ug = superficial gas velocity, flux rate of gas phase Flow Conditions," MS thesis, U. of Tulsa, Tulsa, OK (1983).
18. Kent, G.R.: "Preliminary Pipeline Sizing," Chem. Eng. (Sept. 25,
flowing in pipe, ft/sec [m/s] 1978) 85, No. 21, 119-20.
UL = superficial liquid velocity, flux rate of liquid phase 19. Kern, R.: "Control Valves in Process Plant," Chem. Eng. (April 14,
flowing in pipe, ft/sec [m/s] 1975) 82, No.8, 85-93.
Vbi = ith slug bubble-front velocity, ft/sec [m/s] 20. Lugben, W.L.: Process Modeling Simulation and Control for Chemi-
cal Engineers, McGraw-HilI Book Co. Inc., New York City (1973).
Vgi = gas bubble velocity in ith slug, ft/sec [m/s]
21. Monsen, J.F.: "Program Sizes Control Valves for Liquids," Chem.
vLfi = liquid film velocity, ft/sec [m/s] Eng. (May 18, 1981) 88, No. 10, 159-63.
vm = mixture velocity, ft/sec [m/s] 22. Monsen, J.F.: "Program Sizes Control Valves for Gas and Vapor,"
v max = maximum separator gas velocity, ft/sec [m/s] Chem. Eng. (Oct. 1983) 90, No. 22, 45-49.
vmi = ith slug mixture velocity, ft/sec [m/s]
Vf = separator foam volume, ft3 [m 3] Appendix-Prudhoe Bay Correlations for
Vg = separator gas volume, ft3 [m 3] LlquldSlug Lengths
VL = separator liquid volume, ft3 [m 3] Two-phase flow tests conducted in two 3-mile [4.8-km] -long
Vret = separator retention volume, ft3 [m 3] flowlines in the Prudhoe Bay field indicated that the liquid slug
Vsp = separator volume, ft3 [m 3] lengths are log-normally distributed. A regression analysis approach
Vsurge = separator liquid surge volume, ft3 [m 3] was used for estimating the mean logarithmic value of the liquid-
slug length. The resultant correlation is
YbL = gas-bubble liquid holdup
YsL = liquid-slug liquid holdup
z = gas compressibility factor
Zf = separator foam height, in. [cm]
Zi = difference between height of nose and tail of ith
liquid slug, ft [m] where
ZR = riser-pipe height, ft [m] \ = In Ls,
'Y g = specific gas gravity d = pipeline ID,
Vm = mixture velocity calculated by dividing the combined
'YL = specific liquid gravity
gas and liquid in-situ volumetric flow rates by the
Pg = gas density, Ibm/ft 3 [kg/m3]
PL = liquid density, Ibm/ft 3 [kg/m3] cross-sectional area of the pipe,
Ls = liquid-slug length, and
Acknowledgments e= error, with the variance al.
Technical information supporting this research was supplied by C-
E Natco. In particular, we also thank Mitchel L. Rooker, R&D The regression coefficients are ao=2.663, a, =5.441, a2=0.059,
separation engineer at C-E Natco, for his valuable advice and as- and a; =0.309 (a=0.556). The maximum liquid-slug length that
sistance. was observed in the Prudhoe Bay tests did not exceed the
O.OOl-probability slug length predicted by the standard probability
References tables as
I. Brill, J.P. et al.: "Analysis of Two-Phase Tests in Large-Diameter Flow-
Lines in Prudhoe Bay Field," SPEl (June 1981) 363-78.
2. Huntley, A.R. and Silvester, R.S.: "Hydrodynamic Analysis Aids Slug Lsmax =exp[(0.556)(3.08)+ I],
Catcher Design," Oil & GasJ. (Sept. 19,1983) 81, No. 38,95-100.
3. Bos, A. and du Chatinier, J.G.: "Simulation of Gas/Liquid Flow in
Slug Catchers," SPEPE (Aug. 1987) 182-289; Trans., AIME, 178.
4. Rooker, M. L.: "Crude Oil Separator Foam Problems, " Proc., South- where Lsmax is the maximum expected liquid slug length in feet,
western Petroleum Short Course, Lubbock, TX (April 27-28, 1983) 0.556 is the standard deviation, and the 3.08 is the standard nor-
400-06.
mal variate corresponding to the 0.001 percentile.
5. Arnold, K. and Stewart, M. Jr.: "Designing Oil and Gas Production
Systems," World Oil (Nov. 1984) 199, No.6, 74-78. The liquid-slug length, predicted by the above equation, is thus
6. Katapodis, L.: "Oil and Gas Separation Theory, Application and De- assumed to be the expected maximum liquid slug length.
sign," paper SPE 6470 presented at the 1977 SPE Oklahoma City The empirical correlations for the liquid holdup in the liquid slug
Regional Meeting, Oklahoma City, Feb. 21-22. and the gas bubble' were obtained from 1.5- and 2.0-in. [3.8- and
7. Mehra, Y.R.: "Liquid Surge Capacity in Horizontal and Vertical Ves- 5.1-em] -diameter test facilities in which kerosene and air were used
sels," Chem. Eng. (July 2, 1979) 86, No. 14, 87-88. as test fluids. The liquid-holdup predictions from these correlations
8. Gerunda, A.: "How to Size Liquid-Vapor Separators," Chem. Eng. were found to agree favorably with the limited data on liquid hold-
(May 4, 1981) 88, No.9, 81-84.
ups gathered in Prudhoe Bay.
9. Watkins, R.N.: "Sizing Separators and Accumulators," Hydro. Proc.
(Nov. 1967) 46, No. 11, 353-56. The empirical correlations are given as
10. Sounders, M. Jr. and Brown, G.G.: "Design of Fractioning Columns,
1, Entrainment and Capacity," Ind. Eng. Chem. (Jan. 1934) 26, No.
1,98-103.
11. API RP 14E, Recommended Practice For Design and Installation of
Qlfshore Production Platform Piping Systems, fourth edition, API, Dallas
(April 15, 1984). where
12. Mulet, A., Corripio, A.B., and Lawrence, B.E.: "Estimate Costs of a = 4.47108-0.13691 uL,
Pressure Vessels via Correlations," Chem. Eng. (Oct. 5, 1981) 88, No.
20, 145-50. b = -0.05831+0.08070 UL,
13. Younger, A.H.: "How to Size Future Process Vessels," Chem. Eng. c = 0.02124-0.01169 uL,
(May 1955) 62, No.5, 201-02. u g = superficial gas velocity,

572 SPE Production Engineering, November 1988


UL = superficial liquid velocity, and SI Metric Conversion Factors
YbL = gas-bubble liquid holdup, API 141.5/(131.5 + API) 0 g/cm 3
bbl x 1.589873 E-Ol m3
and also ft x 3.048* E-Ol m
ft3 X 2.831 685 E-02 m3
OF (OF-32)/1.8 c
in. x 2.54* E+OO cm
where Ibm/ft3 x 1.601 846 E+01 kg/m3
a = -0.52728+0.43839 UL. psi x 6.894757 E+OO kPa
b = 2.01451-0.17878 UL, scf/bbl x 1.801 175 E-Ol std m 3/bbl
c = -0.20271 +0.01819 UL, and
YsL = liquid-slug liquid holdup. *' Conversion factor is exact. SPEPE
The following values for the liquid slugs and holdups were cal- Original SPE manuscript received for review March 14. 1986. Paper accepted for publica
tion Feb. 22, 1988. Revised manuscript received Aug. 3, 1987. Paper (SPE 14124) first
culated with the data from Table 1: Ls=441 ft [134.4 m], presented at the 1986 SPE IntI. Meeting on Petroleum Engineering held in Beijing, March
Lsmax =2,446 ft [745.5 m], YsL =0.49, and YbL =0.12. 17-20.

SPE Production Engineering, November 1988 573

You might also like