OTC 21781

High-Pressure Gas-Liquid Separation: An Experimental Study on Separator

Performance of Natural Gas Streams at Elevated Pressures
A.S. Bymaster, M.D. Olson, and E.J. Grave, ExxonMobil Upstream Research Company
S.J. Svedeman and F. Viana, Southwest Research Institute®
R. Mikkelsen and R. Akdim, FMC Technologies / CDS Separation Systems

Copyright 2011, Offshore Technology Conference

This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, USA, 2–5 May 2011.

This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect any position of the Offshore Technology Conference, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Offshore Technology Conference is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.

Abstract
In an effort to understand separation phenomena, improve separation technologies, and mitigate the uncertainty and risk
associated with separator design and operation at elevated pressures, high-pressure separation tests have been conducted at
1,500 psig and 2,600 psig over a wide range of gas flow rates and inlet liquid concentrations. The tests were performed at the
Southwest Research Institute® Multiphase Flow Facility using hydrocarbons ― natural gas and a liquid hydrocarbon model
fluid (ExxsolTM D110, with components C12-C17).
The experimental test section used during this study was equipped with compact, high-capacity separation technologies
that accomplish high liquid removal efficiencies while minimizing footprint. The unit consisted of a CDS Separation Systems
inline cyclonic separator for bulk liquid-phase removal and a downstream vertical separator equipped with compact internals
(i.e., a vane-type inlet device, drainable mesh pad coalescer, and demisting cyclones). The design of the test skid provided
significant flexibility to evaluate the performance of the separation devices alone or in combination. In this paper, the high-
pressure separation testing approach, performance trends, and impact of the results are presented and discussed.

Introduction
Efficient gas-liquid separation is essential to the reliability and successful operation of many processes within the oil, gas,
and chemicals industries. Poorly designed or inefficient separators can lead to numerous process-related issues. In gas
processing, separators are used upstream of rotating equipment (e.g., expanders, compressors, turbines), contactors (e.g.,
glycol and amine), mol-sieve dehydrators, fuel systems and the like, and are therefore critical to preventing mechanical
damage, foaming, fouling, or hydrate formation in downstream equipment. In addition, separators are important to meeting
product requirements (e.g., hydrocarbon dew point) or satisfying air emissions and other environmental regulations. The
importance of a properly designed separator is quickly realized as costs associated with the repair or replacement of
equipment and/or gas treating solvents often far exceed the initial cost of the separator.
Gas-liquid separators are designed to handle a mixed-phase inlet stream and use the density difference between the phases
to drive phase separation (e.g., remove entrained liquid droplets from a gas-continuous stream). Gravity-based separators
often employ more conventional demisting devices (i.e., wire mesh pads or vane packs), and therefore rely on gravity and
impingement as the primary separation mechanisms. Unfortunately, conventional gravity separators, and the aforementioned
internals employed within, are not always feasible. This is most evident when de-bottlenecking existing facilities (to increase
throughput and/or prolong the life of the field) and in the development of resources in challenging environments (e.g.,
deepwater, arctic regions, or other remote fields) where vessel diameter, wall thickness, weight, or available footprint may be
restricted. In such cases, more compact separator internals that use centrifugal forces to drive phase separation can be
employed, such as inlet cyclones (bulk phase separation) and demisting cyclones (fine droplet removal). More recently, the
demand for compact, efficient gas-liquid separation in such services has launched the development of inline cyclonic devices
that employ the same separation mechanism (i.e., centrifugal forces) to achieve bulk phase separation [Fantoft et al., 2010].
Inline cyclonic separators can potentially replace larger, more conventional separators or be paired upstream of a separator to
de-bottleneck existing facilities or reduce the footprint in grass-root designs (by reducing the diameter and overall height of
the downstream vessel).
The movement to pursue more remote and marginal oil and gas fields, as well as offshore and deepwater subsea
processing, not only demands the development of more robust and compact separation technologies, but also the ability to
2 OTC 21781

operate at elevated pressures. Operating at higher pressures can have several advantages, especially in reducing capital and
operating costs associated with compression facilities (i.e., optimizing compressor horsepower, thus minimizing compressor
costs and fuel requirements). The benefits of compact separation technologies become especially evident in high-pressure
services where the wall thickness required by pressure vessel and pipe code can significantly impact overall equipment
weight and costs.
Of course, the development of highly efficient separation devices is greatly dependent on high quality, experimental data
collected at conditions similar to the field. Today, much of industry’s knowledge surrounding separator design and
performance is based on air/water systems at ambient conditions. While these experimental investigations have contributed to
our qualitative understanding of separation physics and principles and have been useful for proving concepts of newly
developed technologies, air/water performance data cannot be used to predict reliably the complex fluid dynamics that can
occur within separation devices, especially for hydrocarbon services at elevated pressures. In real hydrocarbon services, the
physical properties of the system (i.e., densities, surface tension, viscosities, etc.) differ significantly from an air/water system
at ambient conditions. Further, the physical properties vary with temperature and pressure and can have a profound, often
unpredictable, influence on the separation characteristics of the system (i.e., droplet size distributions, droplet coalescence,
droplet shearing, re-entrainment, and film stability). As a vessel's operating pressure increases, the density gradient and the
surface tension between the two phases decrease, thus resulting in a more difficult separation and significant uncertainty in
the design and performance of the separator unit. Such uncertainty introduces new process risks, as potential liquid carryover
into downstream equipment can result in considerable process-related disturbances (e.g., damage and critical failure of
compressor units). These factors can lead to either inefficient or highly conservative designs, with higher capital costs and
smaller operating windows. Based on this reasoning, equipment qualification tests should be performed using live
hydrocarbon fluids at the actual design conditions to accurately predict separation performance and optimize design.
Unfortunately, to date, there is a gap in our knowledge of gas-liquid separation at elevated pressures greater than 1,500
psig (103 barg). Such research is challenging and few facilities in the world are equipped to safely operate and conduct high-
pressure separation experiments. As a result, a very limited amount of data is available in the public domain on hydrocarbon
fluids at elevated pressures. Austrheim et al. [2008] represents one of the few groups to study the gas-liquid separation at
high pressures. In their studies, Austrheim et al. conducted experimental investigations on a large-scale scrubber with
hydrocarbon fluids (natural gas and Exxsol D60) at pressures up to 1,640 psig (113 barg). The operating range of the test rig
was 0.01–3 liquid volume percent, therefore covering a sufficient range for most scrubbers in operation today. The results
generated by Austrheim et al. [2008] clearly demonstrate a significant decrease in separation efficiency (for the overall
scrubber and the individual separation internals) at higher operating pressures. Though this data sample is small, the results
demonstrate that different design rules apply for high-pressure separation of hydrocarbon fluids, thus confirming the
importance and need to further our understanding of separation phenomena and modify current design practices at elevated
pressures.
The objective of this work was to conduct high-pressure experimental testing on multiple separation technologies in an
effort to further understand the boundaries, limitations, and risks associated with gas-liquid separations at elevated pressures.
High-pressure qualification tests were conducted at Southwest Research Institute (SwRI®) using compact separation devices
from CDS Separation Systems, an FMC Technologies subsidiary. The experimental test section used during this study was
equipped with an inline cyclonic separator for bulk liquid-phase removal and a downstream vertical separator with compact
internals (i.e., a vane-type inlet device, drainable mesh pad coalescer, and demisting cyclones). The tests were conducted with
different test configurations using methane and Exxsol D110, over a wide range of pressures (1,500-2,600 psig) and
gas/liquid flow rates. The separation devices tested in this study have never been tested or implemented at such elevated
pressures, to the authors’ knowledge. Through this array of test configurations and conditions, a total of 177 data points was
collected that provide useful performance indicators of the combined and individual separation efficiencies of these devices
in larger-scale field installations. In this paper, the overall separation efficiencies of the inline cyclonic separator and the
downstream separator are discussed. The internals used in the separator are described in the sections below; however, the
performance of each individual device will be the subject of a future paper.

Statement of Theory and Definitions

Conventional Gas-Liquid Separation Technologies

Mesh pads are porous blankets of wire used in many applications within the oil and gas industry for removal of entrained
liquid droplets from a gas-continuous stream. In addition, mesh pads can be used to coalesce droplets for downstream
separation equipment, such as demisting cyclones. Mesh pads vary in design and style according to their specified wire
diameters, mesh densities, knitting styles, co-woven materials, and ultimate gas/liquid handling capacities. The underlying
theory supporting mesh pad installations is that the liquid droplet-laden gas is able to flow freely through the interwoven
wires; however, the inertia of the liquid droplets causes the droplets to impinge upon the wire surface and coalesce into larger
droplets that drain down the mesh. The mesh’s ability to drain is a key capacity parameter. In conventional vertical separator
designs, the captured liquid falls directly through a countercurrent gas flow field, where the effects of drag can be quite large;
therefore, larger droplet diameters allow gravity to overcome these drag forces and deliver the captured droplets to the liquid
OTC 21781 3

collection pot of the vessel. Under extremely high gas or liquid rates, a mesh pad installation can exceed its flood point (i.e.,
excessive liquid re-entrainment), resulting in increased liquid carryover to downstream equipment. Special variations of mesh
pads exist where the mesh density increases across the thickness of the screen, thus giving such styles higher gas/liquid
handling capacities.
Vane packs employ the same mechanism as mesh pads, using the inertia of entrained liquid droplets to drive phase
separation. Vane packs contain a vast number of zig-zag baffles, such that the liquid droplet-laden gas flows freely through
the vanes, while the liquid droplets impinge on the surface of the baffles and coalesce into larger droplets. Spacing between
the baffles, turn angles, and the number of passes are designed to satisfy specific droplet removal requirements. In addition,
vane packs can have special liquid collection/removal pockets, designed to increase the overall liquid capacity of the device.
Vane packs are usually more resistant to fouling than mesh pad installations due to the larger physical openings; however,
data suggest that vane pack performance decreases at high pressures for the same capacity. Therefore, mesh pad or high-
capacity demisting internals are typically preferred in high-pressure services.
Separator vessels with mesh pad or vane pack mist eliminators are typically sized using well-known, rule-of-thumb
design criteria that correspond with the empirically quantified K-factor, derived by Sounders and Brown [1934]. The K-factor
is derived from the vertical force balance of a spherical liquid droplet suspended in an upward gas flow, given by,

4 gd ρg
K= =U (1)
3C d (ρ l − ρg )

where the superficial gas velocity is denoted by U, the liquid and gas densities by ρl and ρg, the gravity constant as g, and the
drag coefficient and droplet diameter are represented by Cd and d, respectively. The recommended K-factor range for each
mesh pad style or vane pack geometry is determined experimentally, such that sufficient gas velocity is available to force the
liquid droplets to impinge on the demisting device’s surface (i.e., wires or baffles), while any flooding and/or liquid re-
entrainment effects are minimized. While the K-factor does account for the fundamental principle of lift on a droplet of a
given size, it cannot be used as the sole design parameter for separator design. The K-factor does not account for, nor
describe, variations in droplet size or re-entrainment of liquid, which can result from changing fluid properties (e.g., surface
tension) with temperature, pressure, and composition. In addition, the K-factor does not account for liquid load, which can
have a significant impact on separator performance.

Compact Gas-Liquid Separation Technologies

As the oil and gas industry operates more in remote, offshore, and subsea locations, conventional gravity-based separators
and traditional demisting internals are gradually being replaced with compact, high capacity inline separation devices or
separator internals. Such devices are useful when de-bottlenecking existing facilities or in grass-root designs to minimize
footprint and reduce transportation, installation, and raw material costs. Cyclonic separation devices have been used to
remove liquid droplets from gas streams for numerous processes, appearing in scrubbers, steam generators, coolers, and
separators located downstream of processes that create entrained liquid droplets such as spraying, injection, distillation, etc.
[Hoffmann and Stein, 2008].
Although the basic principles of cyclonic devices have been understood for quite some time, detailed analysis of the two-
phase fluid dynamics involved in gas-liquid separation has only recently become possible. Such advances can be attributed to
the development of equipment to measure the fluid velocities within the main body of the cyclone and the development of
computational fluid dynamics (CFD), which can accurately model the phase behavior of swirling flows [Hoffmann and Stein,
2008]. Through these recent technological advances and the development of more sophisticated experimental methods, two
separate phenomena have been determined to limit the overall liquid droplet capture efficiencies of cyclonic separation
devices: initial droplet capture and film instability (i.e., liquid re-entrainment from the cyclone wall). At high gas and/or
liquid loadings, such as those commonly experienced within scrubbers and separators in the oil and gas processing industry,
the performance of cyclonic separation devices becomes limited by film instability and re-entrainment, rather than initial
droplet capture [Ishii and Grolmes, 1975; Austrheim et al., 2007; Hoffmann and Stein, 2008]. Separation efficiency increases
with higher gas and liquid loads (assuming coalescence effects) if initial droplet capture is the limiting factor. Conversely,
separation performance deteriorates at higher gas and liquid loads if liquid film instability is the limiting factor. The limiting
factor that governs the overall droplet capture efficiencies of cyclonic separation devices can be determined experimentally
by evaluating the cyclone performance over a range of gas and liquid loadings; however, as previously noted, minimal field
or lab performance data are available using hydrocarbon fluids at pressures greater than 1,500 psig.
A more detailed description of the cyclonic separation devices tested in this work is provided in the following sections.
4 OTC 21781

Experimental

Facility Description and Summary of Procedure

The SwRI Multiphase Flow Facility (MFF) in San Antonio, TX is equipped for research and testing in areas such as flow
assurance, gas-liquid separation, wet gas metering, multiphase flow metering, and subsea pipeline design and operation. The
SwRI facility is especially well suited for the high-pressure separator tests of interest for this study, equipped with the high-
pressure multiphase pump, separators, instrumentation, and control system needed to conduct the tests. The facility has the
capability to operate with real hydrocarbon fluids over a wide range of flow rates (0-2.5 actual ft3/s), pressures (100–3600
psig), temperatures (40-120°F), and gas volume fractions (0-100% GVF).
A flow diagram of the MFF as configured for testing is provided below in Figure 1. As illustrated in the figure, the test
facility is a closed loop. Prior to testing, the loop is charged with stabilized condensate (Exxsol D110, see below for a
detailed description of the fluid) and pressurized with natural gas to the desired system pressure. The MFF horizontal
separator acts as the gas and liquid reservoir. The separated gas and liquid streams from the horizontal separator are metered
continuously prior to being mixed and routed to the test section (the gas is in equilibrium with the injected liquid when it
enters the test section). Following the test section, the gas and liquid streams are recombined, and the mixed-phase stream
then flows to the multiphase pump, where it is ultimately discharged back into the MFF horizontal separator. Gas and liquid
bypass lines around the multiphase pump are in place for capacity control and to maintain the minimum 5% liquid flow
volume required by the pump. A heat exchanger located on the liquid bypass line is used to control the temperature of the
system.

Figure 1: Simplified flow diagram of the SwRI Multiphase Flow Facility.

The experimental test skid (see Figure 2) provided by CDS Separations Systems was comprised of an inline cyclonic
separator to remove bulk liquid from the gas-continuous stream and a downstream vertical separator with compact internals
for fine droplet removal. Of course, as with all experimental testing, the quality of the collected data is highly dependent on
the selected instrumentation. Care must be taken in the selection and installation of the instrumentation and meters to
understand the error and limitations associated with such devices. The skid provided by CDS was well equipped with
instrumentation to measure gas and liquid flow rates, pressures, temperatures, phase densities, gas volume fractions, and
liquid levels. During the experimental tests, all liquid removed from the separation skid was continuously measured using
several Coriolis meters (for higher flow rates of captured liquid) or by monitoring the liquid level in the liquid collections
using level radars (for lower rates of capture, where the liquid flow rates were below the range of the Coriolis meters).
Coriolis meters were placed on the liquid boot of the inline cyclonic separator, at the bottom of the vertical separator, and on
the external drains of the separator internals (i.e., demisting cyclones and drainable mesh pad coalescer). The external drain
lines were recombined in the liquid bottoms of the vertical separator.
The inline cyclonic separators (with nominal diameters of 4 inches and 6 inches, respectively) were equipped with a static
mixer upstream of a stationary swirl element, both of which were positioned within an otherwise ordinary pipe spool. The
function of the static mixer is to ensure even distribution of the dispersed liquid in the gas-continuous phase. The stationary
swirl element is used to induce swirling flow and centrifugal phase separation down the length of the inline device. Here, the
liquid droplets, due to their inertia, coalesce into a thin spinning film on the inner wall of the cyclone, while a gas core is
formed in the center of the pipe. The thin liquid film and a small fraction of the gas core are then captured within the annular
OTC 21781 5

space between the gas exit and the inner wall of the cyclone. The liquid proceeds to drain into the liquid collection boot
where it is allowed to degas. The size of boot varies according to the liquid quality requirements, which are typically
specified as less than or equal to 1% gas-in-liquid. The liquid droplet-laden gas exiting the liquid boot is routed back to the
primary separation chamber through a pressure balance-driven recycle line and is re-injected radially into the lower-pressure
gas core via the swirl element. The majority of the gas-dominated core is captured within the main gas outlet and sent
through an anti-swirl element to recover some of the dynamic pressure generated through the cyclone’s main body as static
pressure, thereby decreasing the overall pressure drop associated with the device. A schematic of the CDS Separation
Systems inline cyclonic separator is provided in Figure 3.

(a)

(b)

Coalescer Filter
Vertical Separator Separator

Gas and Liquid

Inlet
Gas Return
Return
Bypass Line

Cyclonic In-line
Device

Liquid Return

Figure 2: (a) Flow diagram of the high-pressure separation skid, and (b) photo of the high-pressure separation skid during testing at
SwRI.

While inline separation technologies are flexible, compact, and relatively easy and inexpensive to install, they are often
more sensitive to changes in process flow than a conventional gravity-based or compact separation vessel. The liquid
collection boot, in combination with the level control valve, enables the inline separator to absorb minor incidental flow
6 OTC 21781

variations and slugging behavior; however, the inline separator is fairly limited in the size of flow variations it can absorb due
to the relatively small amount of buffer volume available. More flexibility can be incorporated into the design by pairing the
inline device with a downstream separation vessel, much like the experimental configuration in this study. Furthermore,
routing the liquid captured by the inline cyclonic separator to the downstream vessel can be considered to enable the
combined separator unit to absorb larger flow variations. Such a setup was not chosen for the test skid in order to accurately
quantify the performance of each individual separation unit independently.

Figure 3: Schematic representation of the CDS Separation Systems inline cyclonic separator.

As mentioned above, the majority of the gas-dominated core captured in the main body of the inline cyclone is routed to
the downstream vertical separator for fine droplet removal. The vertical separator has an internal diameter of 14 inches and
was equipped with a vane-type inlet device, drainable mesh pad coalescer, and two demisting cyclones. A schematic of a
typical compact separator arrangement, much like the vertical separator design used in this study, is provided in Figure 4.

Figure 4: Schematic representation of a typical compact separator with vane-type inlet device, mesh pad coalescer, and demisting
cyclones.

The vane-type inlet device is designed to diffuse the momentum of the incoming mixed-phase feed in a controlled
manner, thus resulting in bulk liquid dropout and reducing potential shearing and shattering of the liquid droplets. This is
especially critical in high-pressure separators, where low liquid surface tensions can lead to much smaller droplet sizes,
which are much more difficult to remove in downstream demisting internals. A vane-type inlet is also useful in promoting
even distribution of the gas phase over the cross-sectional area of the vessel, which helps prevent channeling effects in the
downstream internals. Further, vane-type inlet devices minimize re-entrainment off the liquid surface in the bottom of the
separator.
OTC 21781 7

In a typical compact gas-liquid separator, the next internal that the gas-continuous phase passes through is some form of
coalescing medium (i.e., a mesh pad or vane pack). As previously discussed, a mesh pad encourages droplet coalescence and
phase separation. Further, the mesh pad provides a pressure drop, thereby reducing or eliminating any flow maldistribution
that may be present on the upstream side of the mesh, prior to entering any downstream mist elimination equipment. In this
study, a type 421 mesh pad was chosen, designed to operate under flooding conditions. At lower K-factors, where the
separation performance of downstream, compact demisting internals often deteriorates, the mesh pad coalescer acts as a
conventional mist eliminator. Thus, the turndown capability of the separator vessel is increased. Variations of conventional
mesh pad coalescer designs are also available that incorporate unique design features, such as dedicated troughs for liquid
collection and drainage. The purpose of installing a drainable mesh pad coalescer, much like the one used in this study, is to
remove a portion of the entrained liquid and prevent overloading the downstream demisting cyclones.
Demisting cyclones can achieve high separation efficiencies at higher flow rates by imposing centrifugal forces on the
liquid droplet-laden gas. As a result, the liquid droplets form a spinning film on the inner wall of the cyclone. Axial demisting
cyclones are the most common type and are generally favored due to their relatively low pressure drop, compact design, and
higher liquid separation efficiencies (compared to reverse flow demisting cyclones with tangential inlets). Axial flow
cyclones impose centrifugal forces on the incoming droplet-laden gas stream by passing it through a stationary swirl element
located upstream of the cyclone’s main chamber. Depending upon the design, either straight or helical slits allow the captured
liquid film, along with a small amount of purge gas, to pass into the cyclone housing. Second-generation variations exist
within the axial flow family, including designs where the purge gas is recycled back to the center of the stationary swirl
element and re-injected through the tip in a radial fashion. Such devices, including the CDS Separation Systems second-
generation demisting cyclones used in this study, are commonly referred to as recycling axial flow cyclones. All axial
demisting cyclone bundles are equipped with internal drainage pipes to effectively remove captured liquid from the cyclone
housing and deliver it to the liquid collection pot of the vessel. In this study, the vertical separator was tested with a two-
element demisting cyclone bundle.
The gas and any liquid carryover exiting the vertical test separator was directed to a coalescer filter separator which had
an internal diameter of 18 inches and was equipped with three stages of separation internals. The individual internals within
this separator included demisting cyclones, a vane pack, and a filter coalescer pack in series. The liquid removed from each
stage of separation was collected in separate vessels over the duration of a single test and measured to determine the total
liquid carryover from the high-pressure test skid.
Most of the instruments in the flow loop were connected to the facility data acquisition and control system via a digital
interface. Instruments that did not support this interface were read using either 4-20 mA analog signals or HART digital
protocol, if available. Data from each sensor were logged at two-second intervals. The data acquisition system allowed for
real-time display of all measured and calculated values during the testing.

Experimental Conditions and Test Section Configurations

The high-pressure qualification tests were conducted at two system pressures (1,500 psig and 2,600 psig) over a wide
range of gas and liquid flow rates. All of the experimental tests were conducted at the near-ambient operating temperature of
90°F (32°C). The test skid was designed with a bypass line around the inline cyclonic separator to facilitate testing of the
individual separation components and multiple configurations. The final test matrix was customized for each test
configuration in order to determine the separation efficiencies of the 4-inch and 6-inch inline devices and the downstream
vertical separator (in series and individually) over an appropriate range of flow conditions. For reference, the test
configurations are given below:
1. Flow through the 4-inch inline cyclonic separator and the downstream vertical separator at 2,600 psig.
2. Flow through the 6-inch inline cyclonic separator and the downstream vertical separator at 2,600 psig.
3. Flow through the 6-inch inline cyclonic separator and the downstream vertical separator at 1,500 psig.
4. Bypassed the inline cyclonic separator to isolate the vertical separator at 2,600 psig.
5. Bypassed the inline cyclonic separator to isolate the vertical separator at 1,500 psig.
Configurations 4 and 5 isolate the vertical separator in order to determine the performance of the various separator internals
(i.e., the drainable mesh pad coalescer and demisting cyclones) at specific operating conditions (flow rates and liquid
loadings).

Model Fluid Physical and Transport Property Determination

The flow loop was charged with pure methane gas and a light model fluid, Exxsol D110. Exxsol D110 was selected as the
hydrocarbon liquid phase because of its similar physical properties and characteristics (e.g., molecular weight, density,
dynamic viscosity, and surface tension) to the target field condensate; however, the extremely high pressures at which the
qualification tests were conducted are either beyond or approaching the limits of most thermodynamic equations. To verify
the selection of this test fluid, as well as the fluid property predictions from our models, DBR Technology Center, a division
of Schlumberger (Edmonton, Alberta, Canada), was contracted to measure the fluid properties of the Exxsol D110 and its
vapor-liquid equilibrium with methane at 2,600 psia and 80ºF (the actual test temperature at SwRI was later raised to 90ºF to
reduce the work load on the cooler). The study conducted by DBR Technology Center included compositional analysis and
true boiling point measurements of the Exxsol D110 fluid, as well as the equilibrium measurements of density, viscosity, and
8 OTC 21781

surface tension for methane/Exxsol D110. The compositional results from the Exxsol D110 distillation were reported with
the molecular weight and density of each distilled fraction (the major components of the Exxsol D110 were determined to be
C12 to C17). The primary physical properties of the fluid at 2,600 psig and 80°F are given below in Table 1.

Table 1 – Summary of the measured physical and transport properties of the natural gas/Exxsol D110 mixture at 2,600 psig and 80°F.

GAS PHASE LIQUID PHASE

DENSITY MOLECULAR VISCOSITY DENSITY MOLECULAR VISCOSITY SURFACE TENSION
3 3
(lbm/ft ) WEIGHT (cP) (lbm/ft ) WEIGHT (cP) (dyne/cm)
8.80 16.10 0.02 45.50 110.2 0.84 8.5
Note: The viscosity of the gas phase is not a measured value, but instead a value predicted by the volume translated Peng Robinson
equation of state [Peng and Robinson, 1976; Peneloux and Evelyne, 1982].

The density of the liquid was also measured during the tests using Coriolis flow meters. At the beginning of each trial, the
density of the liquid was measured at the test pressure and temperature with a 1-inch reference Coriolis meter under no-flow
conditions. The baseline densities of the liquid phase, as determined by the 1-inch reference Coriolis meter, were 47.3 lbm/ft3
and 45.9 lbm/ft3 for the test pressures of 1,500 psig and 2,600 psig, respectively. Liquid density measurements taken at the
highest test pressure are in good agreement with the vapor-liquid measurements provided by Schlumberger.

Results and Discussion

The individual separation performance of the inline cyclonic separator and the downstream vertical separator are discussed in
the sections below. The performance of the internals within the vertical separator will be the subject of a future paper.

Inline Cyclonic Separator

The primary objective of test configurations 1-3 (see Experimental Conditions and Test Section Configurations section)
was to quantify the separation performance and capacity of the inline cyclonic separator over a wide range of inlet flow
conditions. Nonetheless, valuable performance data were also recorded for the vertical separator internals for these two-stage
separator configurations. In this section, the individual separation performance and capacity limitations of the inline cyclonic
separator are discussed. Results of the downstream vertical separator are presented in the next section.
The nominal line size of the inline cyclonic separator is often based on the mixed-phase inlet momentum and flow regime
(i.e., stratified, dispersed, mist, etc.); however, these guidelines cannot be used as hard-set rules for the design criteria,
especially at high operating pressures where physical properties such as vapor density, liquid viscosity, and surface tension
are known to negatively impact the separation efficiency. In this study, two inline separators (with nominal diameters of 4
inches and 6 inches) were tested independent of one another in order to cover a wide range of inlet fluid momentums and
identify the operating boundaries of each design. First, the 4-inch inline separator was tested. Following the tests in
configuration 1, the 4-inch inline separator was replaced with the 6-inch inline separator for testing configurations 2 and 3.
Figure 5 below illustrates the separation efficiency of the 4-inch and 6-inch inline cyclonic separators as a function of the
inlet liquid load and the gas flow rate at a system pressure of 2,600 psig. Due to the pressure drop limitations of the MFF test
loop, the higher liquid loads could not be tested at the higher gas rates.
As previously discussed, two separate phenomena exist that limit the liquid removal efficiency of a cyclonic separation
device: initial droplet capture and re-entrainment from the cyclone wall. When the liquid droplets are so small that the liquid
removal efficiency of the cyclone is determined by the droplet size, the separation efficiency increases with increasing gas
flow and liquid loading (assuming coalescence effects). Such behavior is observed for the 6-inch unit, as illustrated in Figure
5(b). However, at higher gas rates, the separation performance of the cyclonic separation device can become limited by re-
entrainment from the cyclone wall. Thus, the efficiency would be expected to decrease with increasing gas flow, as illustrated
by the results for the 4-inch unit in Figure 5(a).
Results for both the 4-inch and 6-inch inline units demonstrate a drop in separation efficiency with decreasing liquid load,
especially at low liquid volume fractions (LVFs). Such a decrease in efficiency can be explained in part since at low liquid
concentrations, a small amount of absolute liquid carryover corresponds to a larger relative amount. In addition, the rate of
droplet coalescence is likely to be reduced at low liquid loadings. Results from Figure 5 show that the 6-inch unit experiences
a significant decrease in separation performance, especially at the lowest gas rate (compared to the 4-inch unit, Y1>Y2). Here,
the flow and corresponding inlet fluid momentum are too low to generate sufficient centrifugal force and ensure the
formation of a stable liquid film on the inside wall of the cyclone. As expected, the efficiency consequently increases with
increasing gas flow rate. Quantifying the turndown capabilities of the cyclone unit is important when considering turndown
requirements in equipment design.
OTC 21781 9

(a)
100

95

Separation Efficiency [%]

91

86

82 Lowest Gas Rate - Design Rate

Low Gas Rate - 150% Design
Medium Gas Rate - 200% Design
High Gas Rate - 250% Design
Highest Gas Rate - 300% Design
Y77
1

2 6 10 14 18
Inlet Liquid Load [vol%]

(b)
100

85
Separation Efficiency [%]

69

54

38 Lowest Gas Rate - 50% Design

Low Gas Rate - 75% Design
Medium Gas Rate - Design Rate
High Gas Rate - 125% Design
Highest Gas Rate - 150% Design
Y2
23
2 6 10 14 18
Inlet Liquid Load [vol%]

Figure 5: Separation efficiency of (a) the 4-inch inline cyclonic separator, and (b) the 6-inch inline cyclonic separator at 2,600 psig
and 90°F. Note the two plots are scaled differently, with Y1>Y2.

To minimize gas carryunder during the qualification tests, sufficient liquid level was maintained within the boot of the
inline separator; however, some carryunder was still evident at high liquid flow rates. Gas carryunder was measured using a
sonar gas volume fraction (GVF) meter and monitored using fluid density readings from the Coriolis flow meter located on
the liquid outlet of the inline boot. Fluid density readings from this 3-inch Coriolis flow meter are provided below in Figure 6
for all data points taken at 2,600 psig. As the plot indicates, significant gas carryunder occurs when the residence time in the
collection boot is not sufficient. When designing a unit, it is important to determine the minimum residence time needed
(based on fluid properties and conditions) to avoid gas carryunder and risk damage to downstream pumps. Routing the liquid
captured by the inline separator to a downstream vertical separator might be a good alternative to absorb larger flow
variations and allow any gas trapped in the liquid to degas.
10 OTC 21781

60

Density of Liquid Exiting Inline Boot [lb/ft3 ]

50

40

Gas Carry-under

30

20
0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 X
100.0

Liquid Residence Time [s]

Figure 6: Observed gas carryunder in the liquid outlet of the inline cyclonic separator for test configurations 1 and 2.

Vertical Separator
Configurations 4 and 5 bypass the inline separator in order to isolate and control the test conditions introduced to the
vertical separator. Figure 7 below provides the separation performance of the vertical separator as a function of pressure and
the liquid and gas loading at the inlet.

100 P=2600 psig

Low K-factor
(below flood)
P=1500 psig

88 P=2600 psig
Separation Efficiency [%]

P=1500 psig Medium K-factor

(above flood)

77
P=2600 psig
High K-factor

P=2600 psig
Highest K-factor
65

54

Y
42
0 2 4 6 8 10 12 14 16
Inlet Liquid Load [vol%]
Figure 7: Separation efficiency of the vertical separator as a function of the inlet liquid load, pressure, and vessel K-factor.

As expected, the separation efficiency is highly dependent on the pressure. The above figure clearly illustrates a decrease
in separation efficiency as the pressure is increased from 1,500 psig to 2,600 psig (for similar liquid loads and vessel K-
factors). This effect is expected as the gas density decreases (from 8.4 lbm/ft3 at 2,600 psig to 4.7 lbm/ft3 at 1,500 psig) and
the surface tension between the gas and liquid phase increases. Both changes have a positive impact on the separation
performance. The lower gas density leads to a higher density difference between the gas and liquid phases, while the higher
surface tension shifts the generated droplet distribution towards larger droplets (via increased coalescence and reduced
droplet/film shearing).
In addition, the results of Figure 7 demonstrate the dependency of the separation efficiency on the liquid and gas loads.
The liquid loading at the inlet was varied over a wide range of liquid volume fractions (0-14.2 % LVF), as was the vessel K-
factor, varying from well below (K<<Kflood) to well above (K>>Kflood) the flood condition of the mesh pad. The best
separation performance was achieved at the lowest vessel K-factor, where the separator was operating well below the flood
point of the mesh pad and well below the normal operation of the demisting cyclones (here the inlet vane and mesh pad act as
the primary separation devices). At the lowest K-factor, nearly all entrained liquid was removed regardless of the inlet liquid
loading. However, as the gas load was increased to higher K-factors (both approaching and exceeding the flood point of the
OTC 21781 11

mesh pad), the efficiency becomes very dependent on the inlet liquid loading to the separator. At higher gas rates, the
separation efficiency initially increases with increasing liquid load; however, a significant drop in removal efficiency occurs
as the liquid load continues to increase. Such behavior correlates directly with the performance of the separator internals,
specifically the demisting cyclones. Under normal operating conditions, demisting cyclones can be very effective in
removing liquid droplets from a gas-continuous stream, capable of maintaining high liquid removal efficiencies even at K-
factors exceeding the flood condition of the mesh. Unfortunately, different design rules apply for high-pressure separations,
as indicated by the data above. The operating range and performance of all internal devices decrease at higher pressures. As
previously mentioned, re-entrainment effects surface at much lower gas and liquid rates due to the decreased surface tension
at high pressure. Increasing gas rates leads to less liquid drop out and more violent droplet break-up at the inlet. At higher K-
factors, less liquid is captured at the inlet and in the mesh as the dense gas flow carries the increased liquid load in the form
of smaller droplets to the downstream demisting cyclones. The demisting cyclones are unable to overcome such deficiencies
and challenges to maintain the same high degree of liquid removal that was achieved at the lowest K-factor. The test data
collected on the demisting cyclones indicate the performance increases initially for increasing incoming liquid loads.
However, eventually the amount of entrained liquid leaving the mesh pad will exceed the liquid handling capacity of the
cyclones. At liquid rates above this point, re-entrainment from the cyclone wall causes the droplet capture efficiencies of the
cyclones and the overall separator to drop significantly.

Combined Separator Unit

Figure 8 below compares the separation performance of the combined separator unit (i.e., the inline cyclonic separator
and the downstream vertical separator) to the separation performance of the stand-alone vertical separator at 2,600 psig. To
provide a good comparison, the 6-inch inline separator under medium (design) gas flow is paired with the separator at the
corresponding high K-factor.
100
Inline + Vertical Scrubber
(Medium gas rate, high K-factor)

88
Separation Efficiency [%]

Medium K-factor
(above flood)

77
High K-factor

Highest K-factor

65 Vertical Separator
(bypass inline)

54

Y
42
0 2 4 6 8 10 12 14 16
Inlet Liquid Load [vol%]
Figure 8: Separation efficiency of combined separator unit (inline cyclonic separator and downstream vertical separator) vs. stand-
alone vertical separator at 2,600 psig.

Illustrated above, the separation efficiency of the stand-alone vertical separator is very sensitive to changes in the inlet
flow conditions, particularly the inlet gas and liquid loading. In contrast, the combined separation unit achieves high
separation efficiency over a larger operating envelope. Such flexibility is important over the life of a field, where the flow
rates and compositions of the produced fluid can vary. By achieving bulk liquid removal upstream of the vertical separator,
the separation efficiencies of the separator internals increase significantly, as less liquid is entrained to the demisting devices.
Through analysis of Figures 5-8, one realizes the turndown flexibility of the combined unit. The separation performance of
the 6-inch inline cyclone deteriorates at low gas rates, where the inlet momentum is not sufficient to form a stable spinning
liquid film on the cyclone wall and drive phase separation; however, at these lower gas rates, the K-factor through the first-
stage vertical separator is sufficiently below the flood point of the drainable mesh pad coalescer. Figure 7 confirms that all of
the entrained liquid is captured at the inlet vane and by the mesh pad at the lowest K-factors (representative of the turndown
in this case), thus providing further support that the inline cyclonic separator can be paired with a downstream separator to
expand the flexibility and operating envelope. It was already mentioned that combining the inline cyclone with a downstream
separator improves the ability of the unit to handle minor slugging transients and reduce any effects of gas carryunder.
12 OTC 21781

Conclusions
In an effort to close the existing gap of gas-liquid separation at elevated pressures, for the first time, high-pressure separation
tests were conducted on inline cyclonic separators and compact separator internals (i.e., inlet vane, drainable mesh pad
coalescer, and demisting cyclones) using natural gas and hydrocarbon liquids at pressures exceeding 2,500 psig. The high-
pressure separation tests were conducted over a wide range of gas flow rates and liquid loadings (0-16% LVF), well beyond
the data currently available in literature.
While only qualitative results and trends are revealed in this paper, the test results can be used to further the
understanding of the boundaries, limitations, and risks associated with gas-liquid separations at elevated pressures. As
expected, the data generated for the inline cyclone and vertical separator show a significant decrease in separation efficiency
at higher operating pressures. Clearly, higher operating pressures introduce new challenges to gas-liquid separations,
including reduced liquid droplet distributions, increased re-entrainment and/or film instability, and the ability of the high
density gas to more easily carry these droplets through separation devices and into downstream equipment. Still, if one
carefully considers and understands the different design rules, as well as the operating boundaries and performance
limitations, high efficient gas-liquid separation is achievable at elevated pressures.

Acknowledgements
The authors gratefully acknowledge Adam Ufford, Adam Hawley, and Becky Owston for their hard work and long hours
operating the SwRI flow loop and collecting the data. The authors also wish to thank Chuck Mart, Barry Boos, Francois
Lajeunesse, and Jerad Naymick for their support throughout the project, and for encouraging us to share this work.

References

Austrheim, T.; Gjertsen, L.; Hoffman, A. “Experimental Investigation of the Performance of a Large-Scale Scrubber
Operating at Elevated Pressure on Live Natural Gas.” Fuel. 87, 1281-1288 (2008).

Austrheim, T.; Gjertsen, L.; Hoffman, A. “Re-entrainment Correlations for Demisting Cyclones Acting at Elevated Pressures
on a Range of Fluids.” Energy and Fuels. 21, 2969-2976 (2007).

Fantoft, R.; Akdim, R.; Mikkelsen, R.; Abdalla, T.; Westra, R.; de Haas, E. “Revolutionizing Offshore Production by Inline
Separation Technology.” SPE Annual Technical Conference and Exhibition, Florence, Italy (2010).

Hoffmann, A.C.; Stein, L.E. Gas Cyclones and Swirl Tubes: Principles, Design, and Operation. Springer, New York (2008).

Peneloux, A.; Evelyne, R. “A consistent correction of the Redlich-Kwong-Soave volumes.” Fluid Phase Equilib. 8, 7-23
(1982).

Peng, D.Y.; Robinson, D.B. “A New Two-Constant Equation of State.” Ind. Eng. Chem. Fund. 15, 59-64 (1976).

Souders, M.; Brown, G. “Design of Fractionating Columns. I. Entrainment and Capacity.” Ind. Eng. Chem. 2, 98-103 (1934).