Doctoral Dissertations

Jianguo Zhou

Date of Award

5-1996

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Polymer Engineering

Major Professor

Joseph E. Spruiell

Committee Members

Roberto S. Benson, Donald C. Bogue, Marion G. Hansen

Abstract

Hollow fibers are widely used in many chemical operation of exchange and separation, and medical application such as hemodialysis and blood oxygenation. However, little information is available in the literature regard to understanding of the spinning process of hollow filaments.

This research was to establish a mathematical model for the solution dry-spinning process of hollow filaments, using polyacrylonitrile (PAN)-dimethylformamide (DMF) solution as an example material; and to further utilize the model to predict the behavior of the spinline and the properties of spun hollow filaments under a variety of processing conditions.

The investigation included a theoretical modeling part and an experimental part needed to obtain appropriate input data for modeling and for comparison of the model prediction with actual experiment results. The model is based on the fundamental equations of fluid mechanics, heat and mass transfer and spinning dynamics. It includes the equations of continuity, force balance, energy balance as well as mass transfer.

A "thick walled" model was developed, which treats hollow filament as a thick walled cylinder under the triaxial stress state during spinning. A "thin walled" model was also developed on the assumption that the variation in the hoop stress with radial position in the filament could be neglected.

The apparent viscosity of PAN-DMF spinning dope was experimentally obtained at different shear rates. A viscosity equation for PAN-DMF spinning dope was developed based on Modified Cross Model and experimental data. The viscosity equation includes variables such as temperature, shear rate and polymer concentration in the spinning dope.

The experiments of the dry spinning of hollow filaments were conducted to collect data from on-line measurements and the properties of spun hollow fibers. The experiments were designed to vary several important processing parameters such as polymer concentration in the spinning dope, take-up velocity, mass throughput and internal pressure.

Computer modeling under the same processing conditions as the experiments was carried out. A comparison between model predictions and experimental data was made in terms of velocity profiles, spinline tension at the take-up, wall thickness, outer and inner diameters of spun hollow filament, and solvent residue in spun hollow filament. It was demonstrated that the mathematical model for hollow fiber spinning was able to predict the basic effects of processing conditions and material properties on the spinline behavior and properties of spun hollow filaments.

Material and processing variables in dry spinning of hollow filament include polymer concentration in the spinning dope, spinning temperature, mass throughput, internal pressure, quench air temperature and its flow velocity, etc. The complex interaction among these variables requires a lot of experimental trials to investigate the process. In many cases, it is even very difficulty to collect a complete set of experimental data due to the limitation of measurement techniques or other problems. This mathematical model allows us to study the detailed effect of material and processing variables on dry spinning process of hollow fiber and the properties of spun hollow filament. The effects of changing individual processing parameters on the spinline behavior and the properties of as-spun hollow filaments, such as outer and inner diameters, ratio of ID/OD, wall thickness and solvent residue were investigated in detail by modeling.

A comparison between thick-walled and thin-walled models was discussed. It showed that the thick-walled model is able to transform to the thin-walled model when wall thickness of hollow filament becomes small compared to filament radius. The effect of the wall thickness on hoop stress predictions by the two models was also examined. The difference in hoop stress predictions between the two models increased withe increasing wall thickness (or decreasing Ri/Ro ratio). The comparison of the spinline predictions by the two models was conducted. The difference between the predictions of the two models was small enough to be neglected, for most practical cases, when Ri/Ro > 0.75.

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

Share

COinS