THE CONTINUOUS MIXING OF PARTICULATE SOLIDS In such a breakdown the first section may be concerned with volume distribution as a first approximation on the assumption that the convective action is not variably selective. Selective convective action would be even more effective than general convection in smoothing the effect of an irregular malproportion. This feature of the mixer may then be expressed in terms of dispersing characteristics per unit length of the equipment. In the machine used by you the discharge profile of trace feed batches would show the form and 'spread' of the feed variations. The second function would be dependent on the detailed mixing action and in your case of a rotating cylinder, the local mixing effect could be simulated in a short batch type drum mixer operating for a duration equivalent to the residence time. The utility of this approach may show on analysing the effect of long and shallow stream mixer construction, where a relatively high work input is required, e.g. addition of liquids to powders, where segregation is likely to be a factor, i.e. where deep stream, extensive free surface activity or dilate conditions predominate, or in studies of mixer lengths or scale effects. The target for such an approach would be to define the efficiency of a continuous mixer system according to three criteria. (1) The performance under operation with specified conditions and materials assuming completely uniform feed conditions. (2) The capacity of a particular mixer design to smooth variations in feed rates or ratios and backmix, by interchanging quantities of material along the stream length. (3) Definition of the nature of variations of the feed system. MR. WILLIAMS It should in fact be possible to describe the performance of a continuous mixer in a given installation by means of a single figure, the variance reduction ratio, provided the tests are properly designed. The materials handled during the test must be the same as those to be used in the plant and the size of samples collected must be determined by the scale of scrutiny required by the par- ticular process. If there is any tendency for the materials to segregate in the mixer the effect will be included in the test results. The importance of using the appropriate sample size has been referred to in the previous answer to Mr. Pugh. Once the variance reduction ratio under the plant conditions has been established by test this will include your second and third points. I agree that in order to know the quality of mixing at the outlet from the mixer it is necessary to know the fluctuations in the rates from the weigh feeders, and that this must be established by separate tests, again using the same sample size as in the mixer tests. So far, in order to establish techniques, we have used only a very simple type of mixer, and it is possible that with such equipment it would be possible that the fluctuations in composition in the outlet from the mixer could be greater than those in the input, if segregating materials are used. This would indicate that the mixer was not suitable for this particular duty. In the future we propose to examine more complex mixers which would be suitable for segregating materials. The techniques described above will be further tested, but it is expected that they will still be applic- able.

36 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS A MEMBER OF THE AUDIENCE: If yOU have a multi-component mixture is it necess- ary to determine as many variance reduction ratios as there are components? MR. •VILLIAMS: To describe fully the performance of any mixer handling a multi- component system it is necessary to examine each component separately, and see how it is dispersed. In industrial practice this could be very expensive, since it requires the analysis of all samples for each component. In many cases it may be sufficient to pick out what you regard as the most important of the components and examine that. Sometimes one component is more difficult to disperse than the others and it may be reasonable to assume that when that component is adequately dis- persed the others need not be considered.