DYNAMICS OF PRESSURIZATION AND DEPRESSURIZATION DURING PRESSURE SWINGADSORPTION

Experiments with a single fully instrumented 3.0 m long column packed with either non-adsorbent or adsorbent particles revealed that times r equired for pressurization and depressurization differed substantially . For all the non-adsorbent packings except sand, the column could be filled to pressures of up to 7 bar absolute with air, Ar, O2 or N2 wit hin periods of 1 s. When the column was packed with small particles of sand, pressurization with gas took up to 10 s. The length of time to pressurize with gas was proportional to gas viscosity and the square o f the length of column packing and inversely proportional to bed perme ability and particle size. When the column was packed with a 5A molecu lar sieve and filled with N2 or O2 the time required for the column to reach the final pressure was correspondingly longer than for the non- adsorbent packings except sand. Results obtained when the column was d epressurized showed that blowdown of the column always took a longer t ime than the corresponding pressurization time. The results are interp reted in terms of two quite distinct models. One model consists of a c onventional unsteady-state bed conservation equation coupled with eith er a modified Ergun or Darcy equation to take account of axial pressur e gradients. Results show that application of the steady-state Ergun e quation to the unsteady-state processes of pressurization and depressu rization is unsatisfactory. Step boundary conditions are also shown to be inadequate for non-adsorbing gas-solid systems. For the first mode l it was therefore necessary to impose experimentally determined time- dependent boundary conditions for each system and to assume, additiona lly, instantaneous equilibrium between N2 and also O2 and the 5A zeoli te. Simulations then agreed reasonably well with experimental results, the greatest discrepancies occurring near the gas inlet and at the lo west terminal absolute pressures (3 and 4 bar). The second model consi dered is an electrical analogue representation of the system. The diff iculties in handling boundary conditions (inherent in the conventional approach), and the constraints of instantaneous gas-solid equilibrium , are circumvented by adopting the latter model. It is also shown that determining Ergun coefficients by curve fitting pressure as a functio n of time and distance is a more satisfactory way of proceeding than u tilizing coefficients measured from steady-state experiments. Micropor e diffusion control as a rate determining event is a more satisfactory model for representing the pressurizing and depressurizing of the col umn with N2 or O2 in the presence of the 5A zeolite packing than a mod el based on equilibrium control.