Particle Size Distribution
Specification for Particulates Feed, recycle, and product from size reduction operations are defined in terms of the sizes involved. It is also important to have an understanding of the degree of aggregation or agglomeration that exists in the measured distribution. The fullest description of a powder is given by its particle-size distribution.
This can be presented in tabular or graphical form. The simplest presentation is in linear form with equal size intervals. The significance of the distribution is more easily grasped when the data are presented pictorially, the simplest form of which is the histogram. More usually the plot is of cumulative percentage oversize or undersized against particle diameters, or percentage frequency against particle diameters. It is common to use a weight basis for percentage but surface or number may, in some cases, be more relevant.
The basis of percentage; weight, surface, or volume should be specified, together with the basis of diameter; sieve, Stokes, or otherwise. The measuring procedure should also be noted. In order to smooth out experimental errors it is best to generate the frequency curve from the slope of the cumulative curve, to use wide-size intervals or a data-smoothing computer program.
The advantage of this method of presenting frequency data is that the area under the frequency curve equals 100 percent, hence, it is easy to visually compare similar data. A typical title for such a presentation would be: Relative and cumulative mass distributions of quartz powder by pipet sedimentation.
In this case the sizes on the abscissa are in a logarithmic progression [log (x)] and the frequency is [dP/d ln (x)] so that the area under the frequency curve is, again, 100. This form of presentation is useful for wide-size distributions: Many instrument software programs generate data in a logarithmic-size interval and information is compressed in the finer-size intervals if an arithmetic-size progression is used.
It is always preferable to plot data so that the area under the frequency curve is normalized to 100 percent since this facilitates data comparison.