Date of Award


Degree Type


Degree Name

Doctor of Philosophy


Chemical Engineering

Major Professor

John W. Prados

Committee Members

E. Eaves, H. Johnson, E. E. Strausbury



The unsteady state behavior of physical gas absorption in a packed bed was studied by comparing the results of experimental frequency response tests of such a system with theoretical frequency responses determined for several postulated flow models.

Rapidly expanding automatic process control technology has resulted in a great increase in the number of experimental and theoretical studies of process dynamics in the last five years. Countercurrent, diffusional operations in distributed systems, however, have received little attention. Therefore, this study was undertaken to provide a more thorough understanding of the dynamic behavior of such an interphase mass transfer process and to investigate the effects of the system’s physic-chemical and operational variables on this behavior.

The absorption column consisted of a 6-inch ID, Pyrex pipe packed to a depth of 5.12 ft. with 5/8-inch ceramic Raschig rings. The gas phase was a mixture of carbon dioxide and air, while the solvent was water. Both phases were handled in open systems.

The frequency response tests were conducted by introducing a gas mixture with a sinusoidally varying carbon dioxide concentration to the inlet of the column and measuring the resulting steady state concentration wave at the column outlet. The concentration sinusoid was generated by mixing with a constant flow of air stream the carbon dioxide flow from a sinusoidally varying linear valve specifically designed for the application. The sinusoid generator was capable of frequencies of 0.1 to 15 cycles/min.

The gas phase steady state concentration sinusoids at the inlet and outlet of the column were measured continuously with specially designed, high speed thermal conductivity cells and were recorded with a single channel oscillograph. The dynamic response of the column was tested for liquid phase flow rates of 222, 55, and 0 lb. moles/hr-ft2 , at each of the three gas phase flow rates( nominally 1, 10, and 20 lb. moles/hr-ft2 , corresponding to Reynolds numbers of 30, 300, and 600 respectively). The responses of the packing section alone were determined from the total column responses by deducting the effects of the column inlet and outlet sections which were evaluated from frequency response tests of a mock-up of these sections.

The experimental frequency response results obtained for a spectrum of concentration sinusoid frequencies at each set of column operating conditions were expressed graphically in Bode plots (phase shift and logarithm of the amplitude ratio versus logarithm of the sinusoid frequency).

Theoretical frequency responses for the packing sections were calculated for three flow models: a “slug flow” model which presumed no radical velocity gradients, a “mixing cell” and an “axial diffusion” model which attempted quantitative descriptions of the packing section mixing or dispersion phenomena.

The experimental results showed that the greatest portion of the column response was associated with the inlet and outlet sections for all the gas flow tested.

Comparison of the theoretical and experimental packing section frequency responses indicated that, while the “slug flow” model most closely described the absorption dynamics, no model was completely satisfactory in this respect. The deviations of the experimental responses from those of the “slug flow” model were attributed to mixing of the gas phase in the packing section. The presence of these dispersion effects was clearly demonstrated in the non-absorption (dry packing) tests. While neither the “mixing cell” nor the “axial diffusion” models adequately described the observed responses, it was established that the relative degree of mixing was considerably larger than had been previously reported for gas flow in packed beds.

The applicability of a flow model could not be established from comparison of experimental and theoretical phase shifts as all models resulted in similar dad which were, generally, in good agreement with the observed values. The effect of column operating conditions on the degree of mixing was masked by an apparent complex interaction of these variables and by the lack of a satisfactory single mixing parameter.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."