Temperature swing adsorption (TSA) processes for the removal of urea during the regeneration of renal dialysate were investigated. The adsorption potential theory, the local equilibrium theory, the numerical method of lines and process experiments were used in the investigation. It was concluded that it is possible to regenerate renal dialysate using a process that includes conventional adsorption, ion exchange and recycle temperature swing adsorption. This process could reduce the amount of dialysis fluid needed for a typical patient to about 1 liter/day.

The adsorption potential theory was used to represent the adsorption isotherms for urea and glucose on an activated carbon and a carbonaceous adsorbent. The adsorption potential theory allowed data from multiple temperatures to be analyzed together. The adsorption potential theory provided an excellent fit to the adsorption data at all temperatures. The adsorption potential theory provided a systematic way in which to interpolate adsorption isotherms to intermediate temperatures. This type of interpolation was useful in the analysis, modeling and design of TSA processes.

The local equilibrium theory was used to development an analytic model of TSA process operation. This model was used to development new TSA process operating cycles. Some of these cycles involved both concentrated and dilute recycle to improve the urea separation factors. The cycles investigated in this work differed from those used by previous workers in that the concentrated recycle was not mixed with the fresh feed before being returned to the adsorbent bed.

The numerical method of lines was used in the two numerical models of the urea removal process. Both models accounted for film mass transfer resistance and axial dispersion. Uniform intraparticle concentrations, corresponding to infinite intraparticle diffusivities, were assumed in one of the numerical models. An adjustable parameter had to be included in this model to make it agree with the results of the process experiments. This parameter was interpreted as representing lumped intraparticle diffusivity and concentration gradients. The other numerical model did not involve the assumption of uniform intraparticle concentration. It treated intraparticle diffusion rigorously and did not include any adjustable parameters. It was in reasonable agreement with the experimental results.