Oct-2010

Designing trays for selective gas treating

Field data from eight gas treating plants show how tray design parameters can be selected to improve contactor performances

Ralph Weiland, Nathan Hatcher and Jaime Nava
Optimized Gas Treating

Article Summary

Trays, random packing and structured packing each have their place, uses, advantages and disadvantages, and the reasons for choosing one internal over another are wide and varied. Structurally, trays are robust, they allow easy introduction and removal of mid-tower feeds and side-draws, they can handle the high liquid-to-gas (L/G) ratios common in gas treating and they offer good turndown. If pressure drop must be minimised (vacuum systems, CO2 capture from power plant flue gases) or the columns are extremely small or large in diameter, packing often makes more sense. In systems with small L/G ratios (for instance, glycol dehydration), structured packing is favoured over any other type of internals. In revamps for increased capacity, the higher throughputs possible with structured packing favour it as a replacement for trays.

However, structured packing is almost impossible to clean in situ, is more susceptible to plugging by entrained solids and is inherently more difficult than random packing to remove from a column for cleaning. But these are guidelines, not rules, and it sometimes happens that trays, for example, should be used in a very low L/G ratio acid gas removal application even when conventional thinking would have recommended packing.

As mass transfer or separation devices, by their very nature these three types of tower internals have inherently different mass transfer characteristics. In any separation application, the efficiency or behaviour of a given internal is a function not just of the mechanical construction of the device but also of the hydraulic state of the fluids — gas and liquid — flowing through it and being contacted. In gas treating, mass transfer is nowhere near as simple as just counting the number of equilibrium stages and amending the result by applying some kind of efficiency (trays) or HETP (packing) value. Gas treating, especially using amines, is highly non-ideal and can be predictively simulated only by a mass transfer rate-based model. In the context of reactive absorption, this type of model has been extensively described elsewhere1 and warrants no further discussion beyond the reminder that absorption rates depend on mass transfer coefficients and concentration (or activity) driving forces.

Enhancement factors account for the effect of chemical reaction kinetics and sometimes reaction equilibria on transport rates. The size of mass and heat transfer coefficients depends to a very great extent on the hydraulic state of flow of the fluids being contacted.

This article describes the way absorbers intended for selective H2 S removal behave when operated outside the range common in generic gas treating, in particular at unusually low liquid loads. In 2009 and 2010, we2,3 reported on what appeared to be the impossibly high selectivity observed and validated in two selective treating plants, one in Iowa and the other in New Mexico, USA. In this article, we report on six additional cases, show that all eight cases are completely in line with each other, and then use the results to design a tray capable of producing extraordinary selectivity in a particular application.

Tray hydraulics
The diameter of a tray column is determined by the tray’s jet flood and downcomer choke flood limits. The Souders-Brown factor (CS = uS √ρV/( ρL- ρV) is a measure of the vapour-handling capacity, where uS is superficial vapour velocity and ρ is density). CS is typically below 0.12 m/s (0.4 ft/s). Tray vendors still struggle with reliably calculating downcomer choke flood limits, although progress is being made.
Perhaps the most important factor in deciding the mode of operation of a tray is the liquid weir load. Numerically, the weir load, qL, is the volumetric flow rate, QL, of clear liquid over the weir divided by the physical length of the weir, LW:

qL = QL  / LW                                (1)

Weir loads considerably over 200 gpm/ft have been achieved, although more typically 150 gpm/ft is considered a fairly high weir load. Between roughly 45 to 50 m3/m·hr (60 or 70 gpm/ft) and 110+ m3/m·hr (150+ gpm/ft), the biphase is a well-mixed froth of discontinuous gas dispersed in a more or less continuous liquid (froth regime). As the weir load falls below the lower end of the froth regime, the mode of operation changes gradually from discontinuous gas in continuous liquid to discontinuous liquid in a continuous gas. The froth gives way to a spray. The transition is anything but sharp and takes place over a considerable range of weir loads. However, when the weir load becomes low enough, the flow regime becomes unmistakably a spray of liquid droplets projected across the tray in a rising continuous gas phase, with the drops finally bouncing into the downcomer. From the standpoint of selective gas treating, the question is what effect the flow regime has on the individual gas- and liquid-side mass transfer coefficients.

In the froth regime, where most trays operate in gas treating, the liquid is ripped apart by the gas jetting through the tray perforations and is highly agitated (turbulent) even on a micro level. High turbulence intensity greatly assists the movement of components to and from the interface into the bulk liquid. In other words, liquid-side mass transfer coefficients are much higher than they would be if the liquid were less turbulent or, indeed, were quiescent altogether. Since CO2 absorption is controlled almost entirely by resistance in the liquid, a conventionally operated tray is an excellent contacting device for CO2 removal, but not for CO2 slip. The status of the gas flow is perhaps less clear cut; however, it is the state of turbulence right near the liquid interface that is important. The gas enters the liquid in rather intense bursts and slugs of large bubbles and broad jets. Near the interface its flow is not especially turbulent so H2S transfer is not greatly promoted.

In the spray regime, the liquid flow rate is low relative to the gas, and the liquid on the tray exists in the form of small droplets of about 1mm diameter. The droplets are projected through the vapour space, bounce their way across the tray and find themselves in the downcomer in only a few hops. Within small drops, the liquid is nearly completely stagnant and transport of material takes place almost solely by molecular diffusion — a very slow process compared with turbulent transport. Mass transfer coefficients inside the drops are therefore very small, absorption of CO2 is necessarily severely retarded, and the CO2 slip should improve dramatically. To maximise selectivity, this is exactly what one would like.