(652a) Process Scaling and Design for Large Scale CO2 Capture By PSA

Authors: 
Jiang, H., University of South Carolina
Ebner, A. D., University of South Carolina
Ritter, J. A., University of South Carolina
Extremely large gas flow rates, like that associated with flue gas produced from a 550 MW coal fired power plant, can exceed 25,000,000 SLPM. Even larger flue gas flow rates can be produced from steel mills. The flue gas produced from a 550 MW coal fired power plant consists primarily of 15 vol% CO2, 65 vol% N2, 8 vol% O2, 12 vol% H2O and ppm levels of SO2 and NOx at a pressure slightly above about 100 kPa and a temperature between ambient and 125 oC. According to the Department of Energy (DOE), the goal is to capture and concentrate the CO2 and produce it in a stream containing at least 95 vol% CO2 with at least 90 % recovery and with less than 10 ppm of O2. Pressure and vacuum swing adsorption processes are being touted for this application. However, an issue arises with these low pressure flue gas streams when fed to a pressure swing adsorption (PSA) process.

The issue is that the feed cannot be compressed very much because of the cost of compression. Hence, the feed must be compressed only slightly with a blower to about 120 kPa, thereby limiting the axial pressure drop along the bed to be less than about 20 kPa atm differential. This pressure drop limit/issue does not exist for PSA processes that operate with a feed pressure well above atmospheric pressure, like from 80 psia to 800 psia, the higher the better. Moreover, because the feed pressure is so low, to effectively regenerate the beds a vacuum must be applied, typically down to absolute pressures of around 5 kPa to 10 kPa. This is commonly referred to as a vacuum swing adsorption (VSA) process. To make matters even worse, the regeneration steps, typically consisting of countercurrent depressurization (CnD) and light reflux (LR) steps, also have a limit on the magnitude of the axial pressure drop, for if it is too large, then the beds are not effectively regenerated because the light end of the bed never experiences the lowest vacuum pressure experienced by the heavy or feed end of the bed. These limits on the axial pressure drop during the feed and regeneration steps of a VSA cycle are dictated by the interstitial velocity in the bed which varies with time and position.

These axial pressure drop limits give rise to two constraints: 1) the longer the bed the higher the axial pressure drop and 2) the higher the interstitial velocity the higher the axial pressure drop. When limited to bed aspect ratios of no less than one, these constraints limit both the height of the bed and the diameter of the bed, which, in turn, limits the amount of gas that can be processed in a bed during the feed step. A unique scaling procedure was developed to meet these constraints while minimizing the total number of beds required in the VSA system by operating one or more identical VSA units in parallel. Each of these VSA units contains a minimum number of identical beds. Many of these trains of multibed VSA units meet the required process performance, including the pressure drop constraints, but not the limit on the O2 concentration in the CO2 product. The latest results of this study PSA scaling and design will be presented.