Washing curves

If it were possible to obtain a perfect sing displacement wash, the frac tion remaining would be numerically equal to I minus the wash ratio. This ideal condition is represented by the maximum theoretical line as shown in Fig. 18-103. Since it represents the best that can be done, no data point should fall to the left of this cni ve. Most, but not all, cake-washing curves tend to fall along the heavy solid line shown. In the absence of ac tnal data, one may estimate washing results by using this cni ve. 

Table 8-5. Single column (3.9 g ChiraLig ) washing curve after loading 25 mM D-methyl ester valine versus 25 mM L-methyl ester valine using a 3 M LiClO and 0.1 M HCIO wash at a flowrate of 0.4 ml...

Program FILTWASH models the dimensionless filtration wash curves for the above case of a filter cake with constant porosity, axial dispersion in the liquid flow and desorption of solute from the solid particles of the filter bed (Boyd, 1993). 

Figure 5.210. Variation of wash curve with dispersion coefficient, DL = 8.5 10, lO- and 10-5.

Figure 5.212. Variation of wash curve with reciprocal order of adsorption isotherm. N = 1, 2 and 3 (curves A, B and C for C3 and D, E, and F for C9).

Experimental wash curves represented as fraction of solute remaining versus the wash ratio j (ratio of wash to void volume of cake) can be plotted semilogarithmically as in Fig. 14.11 (the solid line). No experimental point will fall on the left of the maximum theoretical curve (the dotted line), which represents perfect displacement. 

Washing curves are used to describe washing efhciency. Frequently washing curves are represented by plots of CJCg as functions of wash ratio j as shown in Figure 22.58. The ratio of in mass to C indicating wash efhciency is constant for a portion of the wash cycle and drops rapidly after the wash ratio j increases beyond 0.6. The drop starts to slow down when the wash ratio j reaches about 1.5 when diffusion washing becomes the dominate factor. Solute is not uniformly distributed... 

The washing curves can be obtained by experimental data of Q, as the function of time, and instantaneous volume of washing hquid with time. Interpretation of experimental data must be approached with caution. At the beginning of the washing cycle, the conduits leading from the filter unit will be full of filtrate. Any curves that are drawn to show the percentage of solute remaining in the cake must take into account the solute in the body of the filter as the well as solute in the cake. 

Figure 1.15 Concentration-time washing curve Solute concraitration vs. time...

Figure 8.38 laical washing curve. Curve (a), flooded cake curve (bX dewatered cake... 

Values of ko and y were assumed in order to fit the numerical solution to their e erimental results. For ko = 0.15 and y = 7.5 a distinct tail was obtained in the predicted washing curve which then fitted the eT erimental results. This tail is indicative of a wash Equor distribution effect on the cake sur ce afifecting about 2.5% of the cake depth. 

Conq>aiison of these models with experimratal results has shown a match over part of the washing curve on some occadons however both are difficult to appfy and for these reasons are not preferred to the di erdon modd. 

Figure 113 Filter cake washing curves pAfakeman, 1981]...

Figure 21. Op nuan wash curves for a dynamic thickener Wash at feed concentration (top) wash at Aickened concentration (middle) wos A at semithickened concentration A bottom), is a function of rate yvian inverse function of solids concentration.

By repeating the experiment several times but using a different volume of wash liquid each time, the variation of the amount of soluble species in the cake can be plotted as a function of the amount of wash liquid permeated through the cake (i.e. a wash curve). The amount of wash liquid used is estimated from a knowledge of the amount of residual liquid in the cake just before washing is started which can, for instance, be determined by measurements in accordance with Step 9 in Section 4.1.1. The wash liquid amount should be varied from about 0.5 up to 3.5 times the residual amount (i.e. between 0.5 and 3.5 wash ratios). 

The aim of experimental washing curve determinations is to gain a knowledge of the amount of wash liquor, number of wash ratios (W) and washing time (0 required to remove a given quantity of solute from a filter cake. It is usually necessary to calculate the amount of solute to be removed from a specification of the allowable solute concentration remaining in the cake at the end of the wash. When typical data are recorded, the results can be plotted in a number of ways as the ... 

At the end of the experiment the cake was removed from the apparatus and weighed, the mass was 412.7 g. To determine the residual quantity of salts, a 137 cm aliquot of distilled water was used to reslurry the cake. The resulhng suspension was filtered and analysis of the filtrate showed a concentration of (j) = 0.9 ppm. The reslurrying/filtration process was performed a fiorther three times with the filtrate being analysed each time to respectively give Thermal drying of the final cake yielded a mass of 276.0 g. Plot the wash curve for the experiment. 

To plot a wash curve it is necessary to maifipulate the data recorded in the experiment. The mass of liquid in the cake M = 412.7 - 276 = 136.7 g equates to a volume of 136.7/0.998 = 137 cm and represents a wash ratio W = 1. The complete list of wash ratios shown in Column (3) of Table 4.3 is obtained by dividing each value in Column (1) by 137. 

Figure 4.7 Washing curve data derived for Example 4.3. (a) instantaneous solute concentration in the wash liquid discharged from the cake, (b) fraction of solute removed from the cake, f (c) fraction solute remaining in the cake, ft...

For the purpose of calculations, the family of curves in Figure 6.7 can also be derived from sequences of coefficients that describe plots of dimensionless solute concentration ( ) v. wash ratio for a range of dispersion number (see Appendix B and Wakeman and Tarleton (2005a)). Their numerical integration allows F to be evaluated as required. As an alternative to using the dispersion model a wash curve can be specified after experimental work, for instance in terms of F vs. W or F vs. t . In this case a dispersion number does not need to be calculated and the fractional recovery is interpolated directly from the wash curve. 

Hermia J. and Letesson Ph., 1982. The universal washing curve and its application to multi-staged counter current washing processes, Proc. 3rd World Filtration Congress, 426-435, Downingtown. 

---