In process industries, time-dependent random fluctuations of raw material characteristics and mass flow rates in bulk material or pulp flows cannot be always prevented for different reasons, eg because of the geological situation (no uniform formation of the deposit) or/and because of technological reasons (segregation in bunkers or on stockpiles). Since a long time, there has been practical experiences that such fluctuations can have negative consequences, eg: fluctuations of the final products characteristics, higher expenditure for process control, and worsening of technological characteristics (recovery, sharpness of separation etc). In the literature, there can be found a lot of details as to, eg the improvement of the sharpness of separation in a jig in dependence on the raw material blending (Kubitza, 1981; Gathen, 1983) or the decreasing flotation recovery with increasing raw material characteristics fluctuations in iron and potash ore processing plants (Bubnova, 1978). Up to now, however, these consequences of random fluctuations, eg on the separation efficiency were not predictable theoretically. It could be shown by means of the general model for mechanical macroprocesses (Graichen, 1989), that the reduction of input fluctuations passing a blending facility can be calculated in this way. In this example, the steady state of the process is the wanted one, but, in general, the model is able to describe also non-steady states of the process regarded. On the other hand, the general model of mechanical macroprocesses has been used successfully for modelling separation processes, eg cross-flow and air classification (Molerus, 1969; Neesse, 1991; Senden, 1979; Tichonov, 1973).
That is why it was tried to model the influence of input fluctuations on the separation efficiency, that means, the impact of non-steady states - as a consequence of random fluctuations - on a separation process. The models discussed in the literature are related nearly exclusive to the steady state of the process. The reasons for that lie in the (relative) simplicity of the models and the fact that the steady state is aimed at by means of blending processes under practical conditions - because of the advantages mentioned above. These models cited have proved to be superior to other models, since the basic general model for mechanical macropocesses makes possible the consideration of turbulence acting in the process unit and influencing the separation process essentially. This is valid also for modelling the turbulent cross flow classification, applied to mechanical classifiers and hydrocyclones, where the correspondence between the separation characteristics (cut size, separation sharpness) calculated and determined experimentally is much better than for other models.