Feb-2007

Troubleshooting amine plants 
using mass transfer rate-based simulation tools

Computer simulation models are indispensable tools for designing and troubleshooting amine gas treating plants. These models, especially ones based on mass transfer rate calculations, allow users quickly and accurately to design and rate equipment, to identify equipment malfunctions, and to recognise process limitations.

Jenny Seagraves, INEOS Oxide LLC
Ralph Weiland, Optimized Gas Treating, Inc

Article Summary

Simulation models can also be used to predict areas of performance shortfalls and corrosion concerns, as well as provide considerable physical insight into plant operations and how process parameters interact in sometimes non-obvious ways.

This paper gives some important criteria and strategies for designing and evaluating amine plants. The discussion centres around three commercial case studies using detailed field data from a plant removing CO2 from synthesis gas, from an off-shore plant processing natural gas, and from a plant removing CO2 from a high pressure natural gas. Analysis of field data is used to demonstrate effective ways of applying both commercial and proprietary amine simulators to design and operating problems.

Introduction
For most of the time since gases first started to be treated to remove CO2, H2S, and other sulphur compounds, treating plants have been built with a 20-tray absorber and 20 tray regenerator, sometimes with two or three water wash trays added to the top of the regenerator. However, the last 20 years or so have seen tremendous improvements to treating technology. These have included (a) the more prevalent use of packing, mostly random but, of late, structured as well and (b) the introduction of a host of new amine chemicals, some highly reactive towards CO2, and others not reactive at all, yet still quite alkaline. Other solvent advances encompass (c) the use of mixed amines or solvent blends. At present, packing’s primary application is tail gas and liquid treating, although there are instances of structured packing being used in fairly high pressure ammonia synthesis gas contactors, as well as low fouling service. In future, the issues of global warming and reduction of greenhouse gases may drive large-scale CO2 removal from power plant flue gases, and this will require the low pressure drops characteristic of structured packing.

The introduction and acceptance of new amines and solvent blends has been energy driven. For example, the ability to removal selectively H2S from natural gas allows substantial amounts of CO2 to remain in the treated gas, thereby reducing the energy required for solvent regeneration (about 80% of the cost of operating an amine plant). Another example is the use of MDEA-based solvents promoted by certain amounts of more reactive amines to remove CO2 from gases while at the same time using much less regeneration energy and lower solvent circulation rates than the older amines. The technology of new amine treating chemicals has even incorporated partial neutralisation of the amine with an acid to achieve the counter-intuitive result of much more complete solvent regeneration and higher purity treated gas. These improvements include the use of high efficiency packings (random and structured), use of alternative process configurations, and acceptance of specialty amine solvents that can provide specific treating characteristics. These new approaches to gas treating all depend crucially not just on different rates of reaction, but also on the effect of reaction rates on relative rates of mass transfer. Advanced mass-transfer rate-based simulation models are inherently aligned with, and completely at home in, this world of new gas treating technologies.

Mass-transfer rate-based simulation models use all the basic building blocks of other, more-traditional models, including material balances and phase equilibrium. But what sets them apart is simply that they use the mass transfer rate characteristics of the equipment and its internals to calculate the separation. Mass transfer rate models use the actual number of real trays supplied by the tray vendor, or the actual depth of the particular physical packing fabricated by the packing vendor and installed in the column. One expects a column containing 40 feet of a 1Y-style structured packing to give different mass transfer performance from the same column packed with 40 feet of 4X structured packing or 40 feet of IMTP-25 random packing, or 20 conventional valve trays. And one expects a reliable process simulator to give different answers in each of these circumstances, too. A mass transfer rate-based simulation model marries material balances, phase equilibrium, chemical kinetics, and the mass transfer characteristics of tower internals into a comprehensive, reliable, predictive tool.

The rest of this paper examines three case studies, each of which points out different aspects of gas treating and how mass-transfer rate-based simulation can be used to understand the process, to suggest better ways of operation, and to allow more efficient operation under changing circumstances.

Case Study 1
This study shows how simulation was used in a tricky retrofit situation. A client asked INEOS to assist in an engineering study of an ammonia plant, in which the plant was to be revamped to accommodate an MDEA-based solvent. The processing scheme is unusual in that absorber temperature is not controlled by heat exchangers on streams feeding the absorber, but by directly controlling reboiler energy flow in the regenerator.

Figure 1 shows the process flow diagram of the existing plant. Processing is done in a split flow configuration with a substantial semi-lean stream being drawn from an intermediate point in the regenerator and sent directly to a point near the middle of the absorber, without modulating its temperature. The remaining solvent continues down the regeneration column where it is further stripped to a low acid gas loading. The fully-stripped lean amine is cooled and then sent to the top of the absorber to provide CO2 polishing. The raw syngas is described in Table 1. The target treated gas is less than 1000 ppmv CO2.

As is true in all process plants, this unit has equipment limits such as maximum attainable solvent flows, maximum available reboiler energy, and limited cooler capacities. However, the most challenging and interesting aspect of this particular plant is the fact that there is no direct control on the temperature of the semi-lean stream feeding the absorber. The flow of this stream is 2.9–3.3 times greater than the fully-lean stream, so its temperature plays a dominant role in setting the absorber temperature profile. It determines the performance of the absorber and, indeed, the entire unit. So it must be controlled, but how?

It turns out that if this were a conventional regenerator, it would be called severely under boiled at any point within the range of operating conditions of the process. The regenerator is so lightly reboiled relative to the total solvent flow to the column, that the steam flow though the regenerator completely collapses within the first few feet from the bottom. In essence, the rich solvent flashes upon entering the column, and the semi-lean stream has a composition equivalent to the liquid leaving a single-stage nearly-adiabatic flash. However, the reboiler energy flow is certainly adequate to strip sufficiently the 25% of the total solvent represented by the fully-lean amine. As will be shown, the lean solvent is roughly what one would expect from a two-stage flash: the first flash stage is the same flash that produces the semi-lean, while the second stage is essentially a QP flash where Q is the energy flow to the reboiler. Admittedly, there is only a rough equivalence between the regenerator and a two-stage flash, but the comparison is made to guide how we think about the situation.