Cryogenic Separation

This takes a deeper dive into the main cryogenic separation technology - namely distillation.

Cryogenic Separation

This article provides a deeper dive into distillation and is focussed on cryogenic applications.

It is often necessary to separate natural or industrial gas mixtures into the individual constituents. A common example is the separation of air into oxygen, nitrogen and argon all of which have value as pure components.

The chief cryogenic low temperature separation techniques are distillation, adsorption and absorption.  The last two are dealt with in separate articles.

Distillation is by far the most widely used technology, which segregates two or more species in a mixture by virtue of their different volatilities (relative tendency to vaporise or condense).

Distillation takes place in a vertical column containing a cascade of separation steps or stages.  The lightest most volatile component(s) leaves the top of the column, and the least volatile species leave the bottom.  The feed stream is supplied to the column at some point / stage between top and bottom where its composition matches that of the local internal streams as closely as possible.  For a liquid feed, the liquid entering the feed stage from the stage above should match the feed composition.

The Theoretical Equilibrium Stage

Ideally the gas and the liquid leaving each stage are at the same temperature and are in thermodynamic equilibrium.  When this  is the case, it is called an equilibrium stage or  theoretical plate or stage.  A well-designed system will have trays that approach the performance of the theoretical stage.

In the cryogenic sector, two types of equipment are used for the separation stages – trays or packing.

Distillation Trays.

Each stage comprises a horizontal tray or plate with liquid flowing across and perforations to allow gas to bubble through the liquid, where the mass transfer takes place. The vapour bubbles leaving the liquid surface combine and pass upward to the next tray.  At the same time liquid that has passed across the tray and been contacted by the gas, leaves the tray via a pipe or channel called a down-comer and is fed to the tray below.  Other features include a weir at the inlet to the downcomer to establish a required level of liquid on the tray, and a liquid seal at the bottom of the downcomer to prevent gas bypass.

  Structured Packing.

As in a trayed column, the vapour and liquid pass counter-current but not in discrete steps but through a cylindrical bed of structured packing – perforated and corrugated slanted sheets that ensure intimate contact between gas and liquid.  The height of packed bed needed to reach equilibrium between the phases is termed the HETP (Height Equivalent to a Theoretical Plate).  To obtain satisfactory performance, uniform distribution of vapour and especially liquid throughout each bed is necessary.
A bed will comprise several theoretical stages, but after a certain bed height, liquid may tend to track towards the wall and is therefore collected beneath the bed and redistributed evenly over the bed beneath.  The maximum height of each bed and the correct design of the re-distributors is critical especially when producing a high purity product.

A related concept to HETP which is sometimes used in packed columns is the height of a transfer unit, HTU. The required bed height is then HTU x NTU, where NTU is the number of transfer units. The expressions for NTU reflect the change in concentration for a key component / driving force for mass transfer, both expressed as mol fractions. Expressions for both HTU and NTU are chosen based on the main resistance to mass transfer - either the liquid film or the gas film. Smaller packing dimensions improve area for mass transfer and reduce HTU or HETP but increase gas phase pressure drop and hence column diameter. Several packing vendors publish their expressions or graphical data for both HTU and pressure gradient as a function of liquid or gas mass velocity.

Stage efficiency

For structured packing the ‘stage efficiency’ is built into the HETP. 

Intimate local mixing of vapour and liquid before the phases go their separate ways results in a high mass transfer efficiency and low HETP for packing and for trays a close approach to an ideal theoretical stage.

For trays, a semi empirical tray efficiency is often used to increase the actual required number of trays above the theoretical number.

N-actual = N-theoretical / tray efficiency.

There is much published data on estimation of the local ‘point’ efficiency, the stage (tray) efficiency and the overall column efficiency. (eg Drickamer & Bradford, O’Connell and AICHE correlations).  Liquid viscosity, surface tension and relative volatility all impact mass transfer and hence stage efficiency.  Formation of foam above the tray has been shown to improve mass transfer by increasing interfacial area.  (This should not be confused with foaming in amine CO2 absorption columns due to contaminant buildup -which is clearly not beneficial).

Zuiderweg and Harmens produced a seminal work showing that for systems where liquid surface tension increases going down the column the liquid film is stabilised and mass transfer improves.  The surface tension is a function of both temperature and local composition, so the point efficiency in a column can vary from zone to zone.

The normal flow path on distillation trays is ‘cross-flow’.  Because the liquid flowing across a tray varies in composition as the light component is stripped out the composition of vapour leaving the tray at different points on the liquid flow path will vary.  If the vapour gets well mixed before it reaches the next tray, and if point efficiencies are high the vapour composition will be roughly what would be in equilibrium with the average liquid composition on the tray.  This vapour will have a higher (better) content of the More Volatile Component (MVC), than vapour in equilibrium with the liquid leaving the tray.  This would give a tray efficiency of > 100%.  In practice, factors including less than complete local phase equilibrium, liquid recycling on the tray, liquid droplet entrainment, and un-mixed vapour would reduce the efficiency where the liquid flow direction alternates on successive trays.

In cryogenic separations such as ASU a type of tray called a ‘ring tray’ or a racetrack tray is sometimes used which can in practice achieve efficiencies above 100 percent.  Here the liquid flow rotates always in the same direction about a central cylinder   Successive downcomer pipes are slightly offset from the previous one.

Reboil and Reflux

Distillation columns generally require a source of vapour, typically produced at The base of the column in a reboiler that vaporises part of the liquid leaving the trays or packing and returns to the bottom tray/stage.

The column also normally has a condenser which liquefies at least part of the vapour leaving the top stage and returns it to the column as ‘reflux’.
Together the reboiler and condenser generate sufficient counter-current flows of respectively vapour and liquid to achieve the desired net separation by the cascade of theoretical stages.

There is a trade-off between the amount of heat added and removed and the number of stages (hence column height) to achieve the desired separation.  However, increasing the reboil heat and therefore vapour flow requires a corresponding increase in condenser ‘duty’ and liquid flow.  Increase in both phase flows and especially vapour will lead to an increase in column diameter to avoid ‘flooding’.  This is a limiting phenomenon when one or other of the dimensions of the column internals becomes hydraulically unable to allow the liquid and vapour phases to flow smoothly in counter-current at all points in the column.  As a result, liquid cannot get out and the column becomes flooded with liquid.  Other phenomena are liquid droplet entrainment from the tray at excessive vapour velocity and ‘weeping’ of liquid through the vapour perforations in the trays at low vapour velocity for example on plant turndown

The optimal hydraulic design requires consideration of many interconnected factors.

Heat supply and removal

Unlike conventional distillation at above ambient temperature, which often uses steam as the reboiler heat source, in cryogenic distillation the reboiler typically uses a process stream that benefits from being cooled and/or condensed.  For the overhead condenser which is the coldest location in the distillation unit the source of refrigeration may be a cold process stream that is vapourised.

To operate at steady state any distillation  process must balance the heat added and removed including heat added in feed and removed in the products. 
This is simply the 1st law of thermodynamics.

The supply and removal of heat essentially ‘drive’ the process.  Like heat supplied to and removed from a heat engine to deliver mechanical work, the supply and removal of heat to/from a distillation unit produce the work to separate the light from the heavier components.

Separation Work and Efficiency

The theoretical minimum work or energy to effect a separation is related to the reduction in total entropy from a mixed feed to two relatively pure products. (See article on Exergy).

This minimum required work Wo = Sum(Exergy of products) – Exergy of feed.
This amount relates to a perfect reversible separation device – with no losses, and negligibly small driving forces.

The Exergy or work supplied by means of heat to the reboiler (Qb) at the bottom temperature Tb and heat Qc removed in the condenser at Tc is the maximum work that could be obtained from a reversible Carnot engine.

Wa = Qc. (Tb – Tc)/Tc

The actual work supplied Wa > Wo because the actual distillation process is not a reversible process.  Its main irreversibilities arise because the driving forces for mass transfer on each stage are finite.  For the column to be almost reversible would require an infinite number of theoretical stages.  This in turn implies the reboil heat would be supplied not just at the bottom of the column but in infinitesimal amounts on each successive stage below the feed.  Similarly, the cooling would be applied in infinitesimal steps on each stage above the feed. 

Referring to the McCabe Thiele diagrams for a binary in the distillation article (Link) the result would be curved operating lines that were essentially coincident with the equilibrium curve.

Also, for a reversible distillation the overhead and bottom products would be respectively warmed and cooled to leave at the feed temperature and there would be no frictional pressure loss or environmental heat ingress.

Clearly this reversible distillation is a perfect ideal that cannot be even approached in practice, but it provides some direction for practical features that may reduce the energy needed for an actual cryogenic distillation, and gives an absolute benchmark for the thermodynamic efficiency of the distillation system.