Design and Selection of Gas-Liquid Separators in Chemical Processes: Entrainment Control and Flash Separation Optimization

Gas-liquid separator design is not simply a matter of calculating vessel diameter and installing a mist eliminator. Reliable separation depends on vapor-liquid equilibrium, inlet momentum, droplet size distribution, gas velocity, liquid drainage, foaming behavior, and process dynamics.

A separator may satisfy a conventional sizing equation yet still allow liquid carryover during pressure changes, feed surges, or turndown. Effective design must connect flash thermodynamics with separator hydraulics, process vessel internals, control strategy, and the complete operating envelope.

Flash Calculation Does Not Predict Entrainment

A process simulator can estimate how much vapor and liquid should exist at equilibrium after a change in pressure or temperature. It does not automatically predict how much liquid will be mechanically carried out with the vapor.

These are two different engineering questions:

• Thermodynamic flash separation: How much of each component enters the vapor and liquid phases?
• Mechanical phase separation: Can droplets and gas bubbles disengage before reaching the wrong outlet?

This distinction is critical in flash drum design. The equilibrium model may predict vapor fraction accurately, while excessive inlet shear, an overloaded mist extractor, or unstable liquid level still causes downstream carryover.

Flash calculations determine the expected phase quantities. Separator hydraulics determine whether those phases leave through the correct outlets.

Define the Operating Envelope Before Sizing

Sizing a two-phase separator at one normal operating point creates hidden risk. Gas volume, liquid loading, density, viscosity, surface tension, and foaming tendency can change substantially during startup, shutdown, feed transitions, or pressure disturbances.

A practical design basis should include:

• Normal, minimum, and maximum gas flow
• Normal and peak liquid flow
• Operating temperature and pressure range
• Feed composition variability
• Flashing across valves or restrictions
• Slug or intermittent liquid arrival
• Foaming and emulsion tendencies
• Solids, salts, polymers, or fouling species
• Permissible liquid carryover and gas carry-under
• Downstream equipment sensitivity

Maximum gas flow may determine vapor disengagement area, while a short liquid surge may control vessel hold-up. Minimum flow can become critical when cyclonic internals fall below their efficient operating range.

The controlling case is therefore not always the case with the highest total feed rate.

Couple Flash Thermodynamics With Hydraulic Design

For a multicomponent feed, vapor fraction can be obtained through a flash calculation using component equilibrium ratios and an appropriate thermodynamic model. The Rachford-Rice relationship is commonly used:$$\sum_{i=1}^{N}\frac{z_i(K_i-1)}{1+\beta(K_i-1)}=0$$

Here, $z_i$ is the feed composition, $K_i$ is the vapor-liquid equilibrium ratio, and $\beta$ is the vapor fraction.

Model selection should reflect the actual mixture. Cubic equations of state such as Peng-Robinson may suit many hydrocarbon and high-pressure systems. Strongly nonideal liquid mixtures may require an activity-coefficient model such as NRTL, sometimes within a combined modeling framework.

The simulation provides phase flow rates, compositions, densities, viscosity, and other design properties. These outputs become hydraulic inputs, but they do not define droplet removal efficiency.

Reducing flash pressure may increase volatile-component recovery, but it also expands vapor volume and increases the required separator area. Higher temperature may increase vaporization while raising cooling demand, downstream condensation, or degradation risk for heat-sensitive materials.

Flash optimization must therefore balance recovery, purity, vessel size, energy demand, and downstream processing.

Design for a Droplet Population, Not One Diameter

Simplified separator sizing often assumes one target droplet size. Industrial feeds contain a distribution ranging from coarse droplets that settle quickly to fine mist that remains suspended in the gas.

That distribution changes through:

• Flash nucleation and vapor expansion
• Shear across control valves
• Turbulence in feed piping
• Impact against inlet devices
• Droplet collision and coalescence
• Breakup in high-velocity regions
• Foam-film rupture
• Re-entrainment from wet internals

Population balance modeling provides a useful framework because it treats breakup and coalescence as competing processes. Although correlations developed for liquid-liquid dispersions cannot be transferred directly to vapor-liquid separators, the underlying lesson remains valuable: the separator does not receive a fixed droplet size.

The Sauter mean diameter, $d_{32}$, can characterize interfacial area but does not describe the fine-droplet tail. Two feeds with the same $d_{32}$ may impose very different loads on the mist eliminator.

Average droplet size is not necessarily the most difficult droplet size that must be removed.

Inlet Hydraulics Shape the Separation Load

The inlet device must reduce momentum, initiate bulk separation, and distribute the gas without creating a finer dispersion than the downstream internals can handle.

A simple diverter plate may work in relatively stable, clean service. Under severe inlet momentum, direct impact can break liquid into smaller droplets, disturb the liquid surface, and create local recirculation.

More controlled inlet arrangements can provide:

• Initial bulk phase separation
• Uniform gas distribution
• Lower downstream mist loading
• Reduced liquid-surface disturbance
• Better handling of intermittent flow
• More predictable vapor disengagement

Nozzle orientation and clearance above the liquid level also matter. A concentrated liquid jet striking the pool can entrain gas below the surface through a plunging-jet effect, increasing gas carry-under even when nominal residence time appears adequate.

The inlet should therefore be evaluated as a droplet-generation and flow-distribution device, not merely a mechanical impact component.

From Gas Velocity to Vessel Configuration

The gravity separation zone must provide enough area and travel distance for droplets to move out of the gas stream. Settling behavior depends on droplet diameter, gas viscosity, phase densities, droplet deformation, and the applicable drag regime.

For small spherical droplets within the Stokes regime, terminal velocity can be approximated by:$$U_t=\frac{g d^2(\rho_L-\rho_G)}{18\mu_G}$$

Stokes’ law does not apply universally. Larger or deformed droplets, turbulent wakes, and high-pressure systems require more suitable drag correlations.

The Souders-Brown equation remains a common screening method for allowable gas velocity:$$U_{\text{allow}}=K\sqrt{\frac{\rho_L-\rho_G}{\rho_G}}$$

The $K$-factor is not universal. It depends on vessel geometry, internal construction, liquid load, operating pressure, fluid properties, droplet-removal target, and acceptable carryover.

Catalog values may be based on low-pressure air-water tests that do not reproduce high-pressure, low-surface-tension, viscous, or foaming service. Gas-liquid separator design should therefore treat Souders-Brown as one component of hydraulic evaluation rather than a complete sizing method.

Vessel orientation should follow the controlling process constraint.

Design condition	Vertical configuration	Horizontal configuration
Restricted plot space	Often favorable	Larger footprint
Significant liquid surge	Limited flexibility	Usually favorable
Large liquid hold-up	May require extra height	Easier to accommodate
Liquid-level interface area	Smaller	Larger
Three-phase service	Possible but constrained	Usually easier to arrange
Solids collection	Bottom withdrawal may help	Requires planned drains
Gas distribution	Compact flow path	Greater risk of maldistribution
Maintenance	Height can restrict access	Long internals may be easier to access

A horizontal vessel may provide greater surge volume and separation length, but maldistribution can leave part of its area ineffective. A vertical vapor-liquid separator has a smaller footprint, yet added height does not always improve bubble disengagement.

Installed volume is not the same as effective separation volume.

Match Mist Extraction to the Service

Mist extractor selection should consider capture efficiency, gas capacity, liquid drainage, turndown, fouling, pressure drop, and maintenance together.

Internal	Main strength	Main limitation	Suitable tendency
Wire-mesh demister	High efficiency for relatively fine droplets	Flooding, fouling, and drainage sensitivity	Clean service with fine mist
Vane pack	Higher liquid capacity and open flow paths	Lower efficiency for very fine droplets	Moderate fouling or high liquid load
Demisting cyclone	Compact, high-capacity separation	Pressure drop and weak low-flow performance	High gas load with stable flow
Staged system	Combines coalescence and high-capacity capture	More complexity and drainage requirements	Wide operating range or difficult mist

Wire mesh captures droplets through impaction, interception, coalescence, and drainage. Once overloaded, it may re-entrain liquid, although it can still enlarge droplets that a downstream vane or cyclone stage can remove.

Vane packs tolerate higher liquid loads but depend on effective drainage from their surfaces. Cyclonic elements can provide strong capacity within their design range, yet solids may cause erosion, deposition, or plugging.

Adding more internals does not automatically improve performance. Poor spacing, uneven flow distribution, or restricted drainage can reduce the benefit of a staged arrangement.

Integrate Hold-Up, Degassing, and Dynamic Control

Liquid residence time is useful for preliminary sizing, but it does not fully describe degassing or control performance.

The liquid section may need to provide:

• Gas-bubble disengagement
• Stable level measurement
• Surge absorption
• A minimum liquid seal
• Alarm-to-shutdown response time
• Operator intervention time
• Separation between operating and emergency levels

In a vertical vessel, increasing liquid height adds hold-up volume but may have limited effect on downward liquid velocity. Increasing diameter can reduce liquid velocity more directly and improve bubble separation.

Dissolved gas may also flash after the level-control valve as pressure decreases. Enlarging the upstream knockout drum to remove very fine free bubbles may offer limited benefit if the downstream pressure drop produces a much larger volume of flash gas.

Dynamic behavior often reveals the real limitation:

Process signal	Likely limitation	Engineering response
Carryover rises after pressure reduction	Increased flash vapor	Recalculate phase load and gas area
Carryover appears only at peak gas flow	Mist extractor overload	Review face velocity and drainage
Performance deteriorates gradually	Fouling or blocked drains	Inspect internals and revise maintenance
Liquid level oscillates	Control instability	Review control valve and level range
Gas carry-under increases	Inlet jet entrainment	Modify inlet distribution
Failure occurs during slugs	Insufficient surge volume	Reassess hold-up and vessel orientation
Low-flow efficiency declines	Cyclone below operating range	Consider staged or alternative internals

Carryover does not always mean that the vessel diameter is too small. Blocked drainage, foaming, inlet maldistribution, or unstable level control may require a different engineering response.

Address Foaming and Re-Entrainment Explicitly

Foam occupies vapor disengagement space, continuously wets mist extraction internals, and can transport liquid as unstable films rather than discrete droplets.

Risk depends on composition, surfactants, contaminants, viscosity, pressure reduction, and gas-generation rate. Laboratory observation, plant history, pilot testing, or targeted vendor testing may be more useful than a generic derating factor.

Re-entrainment can originate from excessive gas velocity, mist eliminator flooding, poor drainage, liquid-wall impact, surface waves, high liquid level, or sudden pressure changes. Entrainment control must therefore cover the entire separator, not only the outlet demister.

Validate Scale-Up and Convert Results Into a Process Package

Gas-liquid separation does not scale through vessel volume alone. Diameter, inlet momentum, residence-time distribution, support structures, velocity uniformity, and drainage paths change at different rates.

Pilot testing or representative validation can examine:

• Droplet and foam behavior
• Inlet device performance
• Mist extractor drainage
• Pressure-drop response
• Fouling tendency
• Turndown limits
• Control-system dynamics
• Carryover during transients

CFD can identify recirculation, short-circuiting, maldistribution, and local high velocities. Its conclusions still depend on realistic droplet distributions, multiphase assumptions, turbulence models, and validation data.

DODGEN integrates thermodynamic modeling, separator sizing, process validation, and pilot-to-commercial scale-up. The results can be converted into equipment specifications, operating envelopes, control philosophies, process vessel internal requirements, and an integrated process package for industrial implementation.

Conclusion

Reliable gas-liquid separator design requires more than equilibrium calculations, fixed residence times, or a single Souders-Brown velocity. It must connect flash vapor generation with droplet behavior, inlet hydraulics, gravity disengagement, mist extraction, drainage, liquid-level control, fouling, and transient operation.

The most suitable separator is not necessarily the largest vessel or the design with the most internals. It is the system whose thermodynamics, hydraulics, controls, and operating range remain aligned with downstream purity, recovery, and equipment-protection requirements.

For processes involving variable flashing, fine mist, foaming, or scale-up uncertainty, an integrated engineering approach can reduce carryover risk and improve long-term process stability.