A study of compressibility and scale and their influence in dead-end pressure filtration
A systematic study has been made of the factors which influence compressibility in cake filtration. A fully automated dead-end pressure filtration rig was designed, constructed and assembled at Loughborough and used to provide accurate data for an experimental matrix. The parameters investigated included pressure, feed concentration, time, surface charge, size and shape of the particulate material and scale of filtration. A proven electrical resistance measuring technique was used to determine transient solids concentrations through a filtering cake/suspension and subsequently interpret filtration performance. The particulate materials calcite and zinc sulphide dispersed in water were used to span a range of filter cake compressibility. The initial results from experiments using calcite were found to be reproducible and generally followed the expected trends. More compressible materials, such as zinc sulphide, were subsequently filtered at different pHs to investigate the influence of surface charge. Filter cells of different filter sizes were used to examine the influence of scale on filtration. Several curious observations were made such as sudden increases in filtrate rate at apparently stable process conditions and lower concentration measurements near the base of the cake. The importance of considering all scale-up parameters and their relation to each other is highlighted in the work. Further points such as the definition of compressibility over a narrow range of pressure and the structure of the filter cake have been discussed. The results were analysed with the use of conventional and new filtration theories. The conventional and modem theories generally gave good predictions of cake height for all the materials tested but for unstable/compressible suspensions the prediction of the concentration profile was less accurate. The use of the maximum solids concentration value at the base of the cake has been suggested to reduce this error and so help prevent filter under sizing. Results indicating retarded packing compressibility have been discussed along with volume-time data which suggest an influence of scale on filtration. Scale-up constants were found to vary considerably in some cases. Large changes in cake resistance were seen for small changes in cake concentration suggesting a need to use other structural characterising parameters as well as macroscopic cake concentration values. The Koenders and Wakeman model gave good predictions of the first linear part of the volume-time relationships for stable suspensions. Conclusions are drawn on the influence of each of the studied parameters on the filtration of compressible materials.