The tubular filter press is a novel tubular configured filter press for the filtration or dewatering of sludges. The unique features of the filter press are: (i) during the cake deposition cycle, cake is deposited on the internal walls of a self-supporting array of horizontal collapsible porous fabric tubes; (ii) during the cake removal cycle, cake is dislodged from the tube walls by means of a roller cleaning device and the dislodged flakes of cake are hydraulically transported out of the tubes by the feed sludge which is simultaneously re-circulated at a high flow rate through the tubes. The two main problems experienced on a prototype tubular filter press, which was erected at a water treatment plant to dewater the sludge from the clarifier underflow, were: (i) tube blockage problems during the filtration cycle; (ii) low cake recoveries (high cake losses) during the cake removal cycle. The following objectives which were defined for this study, were regarded as fundamental prerequisites for any solution of the two main problems: (i) to develop a predictive dead-end internal cylindrical model for compressible cake filtration inside a porous tube; (ii) to investigate the cake losses during the cake removal cycle of the tubular filter press; (iii) to develop a predictive unsteady-state internal cylindrical cross-flow microfiltration model for a non-Newtonian sludge which, when filtered, produces a very compressible cake. (An alternative to dead-end filtration during the filtration cycle of a tubular filter press is low axial velocity cross-flow filtration). On the basis of the objectives the study was divided into three separate investigations. To date no one has developed a model which incorporates the cylindrical configuration of the filter medium for dead-end compressible cake filtration inside a porous tube. The most comprehensive model for dead-end external cylindrical compressible cake filtration is that of Tiller and Yeh (1985). This model was adapted for internal cylindrical compressible cake filtration. In essence the model by Tiller and Yeh (1985) requires the solution of a system of two ordinary differential equations in order to calculate the radial variation of solids compressive and liquid pressures in a compressible filter cake deposited externally on a cylindrical surface. The relevant equations were derived for internal cylindrical compressible cake filtration and it was found that one of the differential equations changes from: dPl/dr = H1/2nrK (external cylindrical) to dPl/dr = H12nrK (internal cylindrical). The other differential equation remains unaltered for internal cylindrical compressible cake filtration. A batch of waterworks clarifier sludge from the prototype tubular filter press was used for experiments to evaluate the performance of the internal cylindrical filtration model. The cake produced by the filtration of this sludge had to be characterized for the model. Compression-permeability data were obtained over a wide solids compressive pressure range. A Compression-Permeability (C-P) cell was used for high solids compressive pressures (10 kPa<_ ps<_400 kPa) and settling tests were used for low solids compressive pressures (0,0065 Pa <_ ps < 525,6 Pa). The cake was found to be very compressible (compressibility coefficient = 0,989). Empirical equations of the form, K' = Fps - b and (1 - E) = B pbs , were derived from the C-P cell and settling tests to relate permeability and porosity to solids compressive pressure. The equations were slightly different to those proposed by Tiller and Cooper (1962). The predictions by the internal cylindrical compressible cake filtration model were compared to the results of constant pressure internal cylindrical filtration experiments, at filtration pressures of 100 kPa, 200 kPa and 300 kPa, using the waterworks clarifier sludge. The internal diameter of the filter tube which was used for the experiments was 26,25 mm. The model accurately described the results of the filtration experiments in terms of volume of filtrate, average cake dry solids concentration, filtrate flux and internal cake diameter. The differences between external cylindrical, internal cylindrical and planar compressible cake filtration were highlighted. Since the tubular filter press is a novel process, the cake losses during the cake removal cycle have not been investigated before. An investigation was therefore conducted into the cake losses which occur during the cake removal cycle. The same batch of clarifier sludge was also used for the investigation of cake losses during the cake removal cycle at filtration pressures of 100 kPa and 300 kPa. It was found that significant cake losses occurred due to: (i) the shear of the cleaning fluid prior to the action of the rollers (losses varied between 10 % to 20 % of the deposited cake dry solids); (ii) the combined action of the rollers when dislodging the cake and the hydraulic conveyance of the dislodged flakes of cake (losses varied between 30 % to 40 % of deposited cake dry solids). A new shear model, which was developed, accurately predicted the cake losses and increase in internal cake diameter and average cake dry solids concentration, which occurred due to the shear of the cleaning fluid. For the shear model the sludge (cake) rheology was determined using a capillary-tube viscometer. It was found that the sludges exhibited Bingham plastic behaviour in the solids concentration range: 3,58 % m/m <_Cs <_16,71 % m/m. The cake losses due to the action of the rollers and hydraulic conveyance of the dislodged flakesof cake decreased markedly as filtration pressure and filtration time were increased, while a decrease in path length for hydraulic conveyance of dislodged cake flakes resulted in a mild decrease in these cake losses. A literature review revealed that to date only one mathematical model (Pearson and Sherwood, 1988) is available for the unsteady-state cross-flow microfiltration of a non-Newtonian sludge which, when filtered, produces a compressible cake. A new unsteady-state internal cylindrical axial convection shear model (for laminar flow of the feed sludge) was developed for cross-flow microfiltration of a Bingham plastic sludge which, when filtered, produces a very compressible cake. Similar to the approach by Pearson and Sherwood (1988) this model is a combination of the dead-end internal cylindrical compressible cake filtration model and the "cleaning fluid" shear model. The major difference between the new model and the model by Pearson and Sherwood (1988) is that unlike the convection-diffusion model of Pearson and Sherwood (1988), diffusive and shear induced diffusive back-mixing of particles were assumed to be negligible. The existence of a shear plane within the cake forms the basis of the model. Those cake layers with a yield stress less than the shear stress exerted by the flowing feed sludge at the inner cake wall are convected along the shear plane. It was assumed that the axial convection of the solids in the moving cake layer along the shear plane is the sole mechanism for removal of solids deposited at the cake surface. The model was compared to the results of cross-flow microfiltration experiments at one filtration pressure (300 kPa) and cross-flow flow rates of 0,84 l / min; 1,58 l / min; 2,43 l / min and 4,44 l /min. The model accurately described the variation of filtrate flux, internal cake diameter and average cake dry solids concentration during the unsteady-state time period. The model, however, had to be "extended" by incorporating empirical equations for changes in permeability and porosity (due to further cake compaction) to obtain a good fit between the model and experimental results during the pseudo steady-state time period. The results of all three investigations provide a greater understanding of the cake deposition process (during both dead-end and cross-flow filtration modes) and the cake removal process for the tubular filter press. This should assist in finding solutions to the two main problems which were experienced on the prototype tubular filter press. / Thesis (Ph.D.)-University of Natal, Durban, 1992.