Capture of carbon dioxide from post-combustion flue gases by vacuum swing adsorption : influence of water vapour

VSA (Vacuum Swing Adsorption) is a promising technology for capturing CO2 which is known to contribute to global warming. Capture of CO2 from flue gas streams using adsorption processes must deal with the prospect of high humidity streams containing bulk CO2 as well as other impurities such as SOx, NOx, etc. However, most studies to date have ignored this aspect of CO2 capture. The major problem caused by water vapour is that water is a much stronger adsorbate than CO2 on most of the polar adsorbents thus drastically reducing the CO2 adsorption capacity. Although the water problem may be tackled by adding a pretreatment drier before the CO2VSA unit, this will result in a large increase of capital and operational cost. Therefore, there is a strong economic motivation to integrate the drying and CO2 recovery in a single VSA process for commercialization of CO2VSA technology. The main purpose of this project is to study the influence of water vapour on the adsorption of CO2 both in the light of fundamentals of adsorption and in the application of post-combustion carbon capture by vacuum swing adsorption experimentally and theoretically. Adsorption equilibria of a CO2/H2O binary mixture were measured on activated alumina F-200 at several temperatures and over a wide range of concentrations from 4% to around 90% relative humidity. In comparison with the single component data, the loading of CO2 was not reduced in the presence of H2O whereas at low relative humidity the adsorption of H2O was depressed. The binary system was described by a competitive/cooperative adsorption model where the readily adsorbed water layers acted as secondary sites for further CO2 adsorption via hydrogen bonding or hydration reactions. The combination of kinetic models namely a Langmuir isotherm for characterizing pure CO2 adsorption and a BET isotherm for H2O was extended to derive a binary adsorption equilibrium model for the CO2/H2O mixture. Models based on the ideal adsorbed solution theory of Myers and Prausnitz failed to characterize the data over the whole composition range and a large deviation of binary CO2/H2O equilibrium from ideal solution behavior was observed. The extended Langmuir-BET (LBET) isotherm, analogous to the extended Langmuir equation, drastically underestimated the CO2 loading. By incorporating the interactions between CO2 and H2O molecules on the adsorbent surface and taking into account the effect of nonideality, the realistic interactive LBET (R-LBET) model was found to be in very good agreement with the experimental data. In contrast, CO2 adsorption on zeolite 13X was entirely depressed at higher water humidity. Direct modification of 13X by silanes increased the hydrophobicity of the adsorbent but also reduced CO2 uptake. A laboratory-scale VSA apparatus was constructed and used to experimentally examine the capture of CO2 from a 10–12% synthetic flue gas stream over a range of water relative humidity. Breakthrough experiments with a binary CO2/H2O mixture in a near-adiabatic double layered 3A/13X column showed a peculiar dual roll-up phenomenon. Water adsorption generated a pure thermal wave which traveled ahead of the water concentration front and swept off the readily adsorbed CO2 leading to a thermal induced roll-up; the slow propagation of the water concentration wave displaced the CO2 by competitive adsorption resulting in n equilibrium induced roll-up. Cyclic VSA experiments with single layered 13X column and multilayered Al2O3/13X column configurations were conducted. The migration of the water and its subsequent impact on capture performance was evaluated. The formation of a water zone creates a “cold spot” which has implications for the system performance. Although the concentration of water leaving the bed under vacuum was high, the low vacuum pressure prevented condensation of this stream. The vacuum pump acted as a condenser and separator to remove bulk water. An important consequence of the presence of a water zone was to elevate the vacuum level thereby reducing CO2 working capacity. On the other hand, the internal purge of CO2 was found to be of critical importance to lower the water partial pressure during evacuation. The penetration of water in the column could be managed by keeping an appropriate volumetric purge-to-feed ratio or a higher vacuum level. This effect was predicted by our axial adiabatic working capacity model. At relatively high water content (> 4% v/v) in the feed, the use of a water prelayer was essential to prevent failure of the system. The overall performance of the VSA with wet feed decreased slightly compared with the performance for dry feed. Reasonable results have been achieved for a triple layered single column VSA in the case with the highest feed humidity of 8.5% v/v, with a product CO2 recovery of 58.2%, purity 52.4% and productivity 0.128 kg CO2/h/L adsorbent. Further scale-up of this process by using multi-columns and a more sophisticated cycle design is expected to further improve the performance. Thus although there is a detrimental effect of water on CO2 capture, long term recovery of CO2 is still possible in a single VSA process.