Oct-2014

Structured packing in a CO2 absorber

Solvent regeneration is the biggest energy consumer in an amine treating unit but efforts to minimise energy consumption should be made cautiously

RALPH WEILAND and NATHAN HATCHER
Optimized Gas Treating, Inc.

Article Summary

Structured packing in deep CO2 removal applications such as LNG production can offer significant advantages, including higher throughput, over other internals. Piperazine promoted MDEA is the most common choice of solvent for this application. However, tight designs, as typically specified in offshore situations, can make such units hard to operate. This article presents a case study in which absorber performance seemed to be disproportionately affected by the reboiler duty of the amine regenerator.

Liquefying natural gas enables it to be transported economically over immense distances to end users remote from the gas source. Shale gas is a large resource for natural gas liquids (NGLs), but it is also the source of enormous amounts of gas. The so-called shale gas revolution is what is in large part responsible for driving the construction of new LNG plants and specialised shipping.

Most shale gas is sweet in that it contains little or only nuisance amounts of hydrogen sulphide and other sulphur compounds. However, significant concentrations of CO2 are normal. Before gas can enter liquefaction, CO2 must be removed to less than 50 ppmv (and sometimes even lower), referred to as deep CO2 removal, and it must then be dehydrated to an almost moisture-free state, usually using molecular sieves. This article is concerned with the CO2 removal step and focuses on amine treating as the most commonly used process.

Although a number of amines are used in the CO2 removal step in LNG production, by far the most common solvent is based on N-methyldiethanolamine (MDEA) with the CO2 reaction kinetics promoted through use of modest concentrations of the activator, piperazine. The reason is that, as a tertiary amine, MDEA does not form a carbamate by reacting with CO2 and therefore it has a much lower heat of absorption than primary and secondary amines. This translates into lower solvent regeneration energy consumption; however, the down side is that MDEA by itself is unsuited to deep CO2 removal. The CO2 absorption rate is too slow and the phase equilibrium with CO2 is often unfavourable.

To get to 50 ppmv CO2 in a column of reasonable physical height it is necessary to speed up the absorption process. This is done using a few weight percent piperazine (typically 3-9 wt%). The rate constant for the piperazine-CO2 reaction has the extraordinarily high value of 50 000 L.gmol–1.s–1 even at room temperature, making this amine an obvious choice as a CO2 absorption rate promoter. All major solvent vendors offer at least one or two formulations of piperazine with MDEA for deep CO2 removal applications, and some include such solvents as part of a licensed process.

Case study
The study is of an LNG related CO2 removal unit using piperazine promoted MDEA in which the absorber contains structured packing and the regenerator is trayed. The packing specific (dry physical) area is nominally 250 m2/m3. Table 1 is a simplified summary of the conditions and composition of the raw gas. The solvent is 50 wt% of a proprietary, piperazine promoted MDEA contaminated with about ¼ mol% of heat stable salts. The gas is a typical LNG unit feed.

The treating process uses a completely conventional gas treating flow sheet with absorber and regenerator tied together through the usual rich amine flash for hydrocarbon recovery, cross exchanger, trim cooler, and pumps. Before embarking on a plant optimisation, it was decided to determine sensible operating ranges for various parameters and to find out how sensitive overall treating performance was to the most important ones. This exercise uncovered a rather surprising sensitivity. It also demonstrated the importance of simulating the entire treating plant, not just the absorber or regenerator as isolated pieces of equipment, and of validating data before an optimisation study is started.

The first simulation run was based on what was thought to be good process information regarding compositions, flows, thermal duties, and so on. A thermal imaging scan of the absorber was also available, taken on the same day the process information was recorded on the plant’s DCS. The big surprise was that the simulated peak temperature in the absorber was found to be 40°C hotter than the thermal scan indicated, and the simulated bulge temperature covered most of the tower’s packed height. The simulated temperature profile could not have been further away from the thermally imaged one! But the main red flag was that the simulated solvent lean loading was nearly 0.07 moles CO2 per mole of total amine. This is a ridiculously high value obtained by simulating the entire plant (less than 0.01–0.02 is more normal for a well maintained solvent). These observations indicated that bad data had likely been input into the simulation. However, the broad temperature bulge made it worthwhile first to isolate the absorber to assess the sensitivity to lean loading.

Figure 1 shows that the residual CO2 in the treated gas (blue line in the figure) steadily climbs with increasing lean loading, but it suddenly escalates explosively as the loading passes 0.033 mol/mol. On the other hand, the CO2 concentration in equilibrium with the lean solvent climbs continually and steadily throughout the loading range (as it should). Below the transition point at 0.033 mole loading, the CO2 leak closely follows the lowest achievable level consistent with a given lean solvent. Because treating very closely tracks CO2 equilibrium over the lean solvent, it can be said that treating in this region is thoroughly lean end pinched. As will soon become evident, this term means that the CO2 concentration in the treated gas is determined entirely by the solvent lean loading. Nevertheless, the transition itself is abrupt. The reason is subtle but revealing.

Figure 2 shows a series of absorber gas-phase temperature profiles for several solvent lean loading values. The curves at the low loading end are what one might expect for CO2 absorption by a fast reacting solvent such as piperazine promoted MDEA; the curves at the high loading end, however, are not. The observation that the peak temperature at the bulge increases with increasing loading provides the first clue to an explanation. There are at least two reasons the bulge temperature itself increases with loading: heat capacity decreases as loading goes up – that is, the same heat release results in higher temperature; and the heat of absorption itself increases with temperature and this exacerbates the effect.

Somewhere between a lean loading of 0.032 and 0.033 mol/mol, a bulge temperature is reached at which the partial pressure of CO2 in the gas right at the bulge is equal to the partial pressure in equilibrium with the solvent there. In other words, at the transition loading, the driving force for absorption becomes zero. At a lean loading only slightly above this point the zero driving force explodes across most of the upper part of the column until the cold lean solvent draws the temperature down near the top of the packing and absorption resumes. The curves represented by the solid lines all correspond to lean-end pinch conditions. The dashed curves are all bulge pinched, meaning that along the flat part of these curves there is essentially no driving force for CO2 transfer in either direction. As the bulge spreads further across the column at higher lean loadings, less and less gas is absorbed and this causes the peak temperature to decrease slowly.