You are currently viewing: Articles


Predict amine plant performance with mass transfer rate-based simulation

Alkanolamines have been successfully used for nearly 80 years for sweetening hydrocarbon products. Yet despite the longevity and track record of amine treating as a process, it is common to find the engineering behind the process still being done with calculation methods that rely heavily on rules-of-thumb and the now-antiquated approximation of ideal or theoretical stages.

Ralph H Weiland, Nathan A Hatcher and Jaime L Nava
Optimized Gas Treating
Viewed : 1712
Article Summary
The advent 30 years ago of high-speed, desktop computing power makes this approach today not just severely limiting, but unnecessary. Heat transfer equipment design has enjoyed the benefits of the transfer rate approach for at least as long as amines have been used to remove sour components from gases. Compared with mass separations however, heat transfer is a relatively simple operation involving the transfer of the single entity, heat. Mass transfer, on the other hand, involves the transport of possibly many components as well as heat, and it involves sometimes complex phase-equilibrium thermodynamics, as well as chemical reaction equilibrium and kinetics. Conceptually, however, the difference between heat and mass transfer is computational rather than fundamental. The beauty of mass transfer rate-based simulation is that separations equipment can be designed and analysed completely without recourse to theoretical stages, tray efficiencies and transfer unit heights. The separation is calculated directly without appealing to such artificialities, using only equipment parameters that can be measured with a ruler. Heat exchangers are never treated as equilibrium stages, and mass transfer calculations do not have to be done that way either. After 25 years of successfully applying mass transfer rate-based simulation tools to such diverse operations as azeotropic and extractive distillation, three-phase distillation, distillation with catalytic chemical reaction and reactive amine-based gas absorption, the need for efficiencies and residence-times on theoretical stages is past. So what does the term mass transfer rate-based really mean and what distinguishes it from other approaches?

Instead of efficiencies and HETPs, mass transfer rate models use mass (and heat) transfer coefficients, gas-liquid contact areas (equivalent to heat transfer surface area), and concentration difference driving forces (just like temperature differences in heat transfer). Figure 1 shows a magnified view of the gas-liquid interfacial region of the liquid film flowing on the surface of structured packing. Just as temperature differences drive the flow of heat, concentration differences drive the diffusion of material species. Mass transfer coefficients play the role of heat transfer coefficients, and now the interface across which chemical components transfer is a flexible moving boundary rather than the fixed boundary of a heat exchanger tube or plate. The advantage in heat transfer is that individual shell and tube-side coefficients for heat transfer can be readily correlated with thermal transport properties (density, heat capacity, thermal conductivity) and equipment geometry (tube diameter, plate spacing, baffle placement). This allows heat transfer calculations to be generalized, and even better, it makes them fully predictive. If we have good correlations for the individual heat transfer coefficients and the equipment geometry has been characterised by physical measurement, we can predict the performance of a given heat exchanger. The same is true for mass transfer. Good correlations for film coefficients and measured equipment geometries (weir heights, passes, open area on trays, or random packing size and brand, or crimp angle and crimp height of a structured packing) allow the performance of a real tray or a given depth of real packing to be predicted. The mass transfer rate model does not use artificial parameters such as residence time per theoretical stage, or HETP, and it does not rely on engineer-supplied estimates of efficiency any more than heat transfer calculations do.

A mass transfer rate-based model should not be confused with one called simply “rate-based” or with one that estimates a tray efficiency, then applies it to a few equilibrium stages. There seems to be disinformation, at least in the gas treating arena, as to what constitutes a rate model. In a nutshell, any column simulation based on equilibrium stage calculations no matter how modified is patently not a rate model. The need for a mass-transfer-rate basis for simulation is most evident when selectivity for H2S over CO2 is a concern, commonly when aqueous MDEA is the solvent. MDEA absorbs CO2 only slowly because the main reaction, CO2 hydrolysis to bicarbonate, is slow. Nevertheless, MDEA is capable of absorbing a lot of CO2, so CO2 absorption cannot be ignored. In fact, it determines the CO2 slip through the absorber, influences the equilibrium pressure of H2S and, therefore, the H2S content of the treated gas. Getting the CO2 slip right is critically important to getting the treating right.

One approach to modeling CO2 absorption by MDEA which retains the equilibrium stage concept while attempting to account for reaction rates is to conceptualise the liquid volume held up on a theoretical stage as a stirred tank reactor. This conceptualisation is shown in Figure 2, where several trays or a large volume of random or structured packing (or indeed whatever else might be in the column) is represented by a single ideal stage of contact. Dissolved molecular CO2 reacts with MDEA at a rate dictated by the reaction kinetics. For an ideal stage whose liquid holdup volume is , (equal to volumetric liquid flow rate, L, multiplied by the ideal-stage residence time, θ), CO2 disappears by reaction on the stage at the rate (mol/s) shown in Figure 2. Bear in mind that physically an ideal stage has no tie to anything that is actually inside the column. Because it is purely an idealisation, the very meaning of ideal-stage residence time defies reason.

In principle, this calculated reaction rate (Figure 2) can be used to compute the increase in total flow rate of dissolved CO2 between the liquid inlet and outlet, i.e., the separation, and it allows the outlet liquid not to be in equilibrium with the outlet gas (equilibrium is the primary approximation of the equilibrium stage model). So is the problem solved? Well, not quite! There are two issues with the whole approach. The first is how to assign a value to the ethereal residence time on a theoretical stage. The second is that there is no way of knowing the concentration of dissolved but unreacted CO2 in the bulk liquid phase. If the reaction were fast,  could be zero, if it were very slow it could be the value in equilibrium with the gas.

Absorption is a two-step process: (1) dissolve, then diffuse through the liquid and (2) react. The simplest assumption is that the reaction is slow enough and mass transfer is fast enough for  always to have its equilibrium value, i.e., there is no mass transfer resistance at all and the rate-limiting step is reaction. But, the reaction model focuses on the wrong process as the rate-limiting step. In fact, CO2 absorption is never reaction rate controlled — it is always mass transfer rate controlled. Disregarding this fact results in the highest, most optimistic possible calculated amount of absorption. Although simulation can be matched to plant performance data through the adjustable parameter, θ, the residence time per theoretical stage is disconnected from anything physical, and the right value for a new situation is just as unknown as a stage efficiency is. If you guess too small a value, you remove too little CO2; too large and you remove too much. Yes, there is an element of “rate” through reaction kinetics, but to call it rate-based is disingenuous — it is still an equilibrium stage model, but now containing an adjustable parameter.
Current Rating :  1

Add your rating:

Your rate: 1 2 3 4 5