Apr-2002

Column design using mass transfer rate simulation

This article looks at examples of commercial columns used in CO2 removal and selective H2S treating, comparing performance test data to demonstrate benefits of modelling actual column internals using mass transfer rate process simulation

Ralph H Weiland and John C Dingman
Optimized Gas Treating

Article Summary

At least three major advances in gas sweetening at the process level have been made in recent years: more easily regenerable solvents developed, solvents having greater selectivity have been commercialised, and desirable properties of individual solvents have been exploited by creating solvents with two or three amines. These approaches have led to significant reductions in process energy requirements, primarily by increasing CO2 rejection and tailoring CO2 slip. Reduced energy requirements are mainly a consequence of lowered solvent circulation rates.

There has also been a new advance in process simulation capability, the development of a completely new approach to column modelling and analysis in the form of the ProTreat mass transfer rate-based tower model. This models towers in full detail as real physical equipment as opposed to a theoretical-stage idealisation of reality. The key attribute is that the mass transfer rate model uses actual trays and packing for simulation of columns.

Thus, if a column contains 2-pass trays in one section and 4-pass trays in another, or trays in one section and packing in another, it is modelled exactly that way. Attention to this variety of design detail within a column is obviously important to actual column performance, and equally important to column simulation. Avoiding the use of calculations based on theoretical stages is of the utmost importance when dealing with applications involving either selective H2S removal or customised CO2 slip, because the internals and their mass transfer characteristics determine to a major extent, the actual H2S and CO2 treating levels achieved.

Absorption and stripping
In processes for total acid-gas removal, treated gas quality is completely determined by phase equilibrium, provided the column contains enough trays or packed depth. This is not the case in selective treating. Here, the extent to which each acid gas is removed is related directly to its mass transfer rate, as well as to the mass transfer rates of each of the other absorbing acid-gas species. The separation is a rate process rather than one dominated by phase equilibrium.

An appreciation of the fact that all alkaline solvents are thermodynamically selective for CO2 but kinetically selective for H2S is vital to understanding the importance of mass transfer rates to contactor performance. For a given lean-solvent acid-gas loadings, a high enough tray count or a deep enough packed bed guarantees that the treated gas leaves the contactor in equilibrium with the lean solvent (or for low solvent rates, that the rich solvent leaves in equilibrium with the sour gas).

However, as the tray count is reduced (or the bed shortened) the treated gas becomes further and further removed from equilibrium. The thermodynamics of acid gas-amine systems is such that CO2 is the preferred solute because it absorbs by forming a fairly stable chemical bond with the amine. But the CO2-amine reaction is of finite rate and, in fact, is quite slow in MDEA, for example. On the other hand, H2S ionises instantaneously (to bisulphide ion); it does not react with the amine at all, it forms no chemical bonds, and the ionisation reaction is immediately reversible. Thus, the chemical reaction kinetics are much faster for H2S; therefore, CO2 absorbs more slowly.

At short contact times (read low interfacial areas, small tray counts, short packed beds) H2S absorbs at a higher rate than CO2, and so H2S is preferentially absorbed. At long contact times (high interfacial areas, many trays, deep beds), CO2 absorbs more completely, albeit more slowly, and CO2 is preferentially absorbed.

Thus, control over selectivity can be achieved by choosing an amine (or a multiple-amine mixture) with the right reactivity toward CO2, allowing contact in a column with the right number of trays or the right depth of packing, and choosing the kind of column internals that favour either CO2 or H2S absorption.

Selectivity depends on rates – not just reaction rates, but mass transfer rates – which implies dependence on all the factors that affect the mass transfer characteristics and mass transfer performance of the actual physical hardware in which the process is carried out. Equilibrium stage models simply cannot capture these effects.

The currency of equilibrium stage models is the number of theoretical stages—the currency of internals vendors and gas processors is actual tray counts, types, and passes and volumes and depths of packing of specified size, type and material. With ideal stages, translation between the two is forever an open question. A true mass transfer rate model, on the other hand, always deals in real trays and real packing – there is never a question about how many trays are needed or what depth of packing to install.

It is equally important to be able to model solvent regeneration accurately if for no other reason than the fact that the loading of the lean solvent produced by the stripper directly and significantly affects contactor performance. Not only does it affect its ability to meet treated gas specifications, but also the actual treated gas composition.

Equilibrium stage models do not work very well here either, because the reactions, the tower internals type, and details affect mass transfer in ways just as important as in absorption. None of the trays in a stripper come even close to an equilibrium stage, and the desorption rate of each acid gas affects the rate of the other.

From a technical standpoint, the ProTreat mass transfer rate-based stripper model treats regenerators as rigorously as absorbers and produces the best possible predictions of regenerator performance without the need for empirical adjustment. When absorber and stripper models are tied together in a recycle flowsheet, the best possible prediction of treating-plant performance is obtained without applying user-supplied or internally-generated empirical corrections of any kind. This complete freedom from empiricism allows the engineer to design and predict the performance of new facilities for which absolutely no operating data or field experience exists.

Case studies
The case studies that follow compare mass transfer rate-based simulation with performance test data collected from three separate treating facilities run by different operating companies. The first data set is for the Worland, Wyoming, plant of Texas Gulf Sulphur, as reported in an article by Estep et al[Sulfur from natural and refinery gases; Advances in Petroleum Chemistry, 1962].