Computing Performance with High Loading

6 Computing the Performance of a Cyclone with High Loading 

We wish to conclude this discussion on the entrance solids-loading effect by simply stating that the effects of solids loading cannot be ignored if performance is to be accurately predicted. This applies not only to cyclone simulation models, but also to the use of the scaling laws developed in Chap. 8. 

There are two possible strategies for predicting the performance of a cyclone at elevated solids loading. One is to  

The other is to use a comprehensive model that accounts for the effect of solids loading directly, such as the Muschelknautz model given in Chap. 6. 

If one uses test data taken on a model cyclone to help predict the performance of a full-sized unit, it is important not to use a higher solids loading in the model than that, which the full-sized unit is expected to experience. It is best to test the model over a range of known solids loadings. 

---

Most applications require standardization and involve the potential hazard of distortion of the objects. The recognized similarities must, therefore, be proved with statistical methods. It is very dangerous to cut single features or to reduce the data set. This leads to a change of the data basis on which the standardization will be performed. A comparison with previously computed stars thus becomes impossible. Fig. 5-7 demonstrates a star plot with feature standardization for the example of the PAH in river sediments. Highly loaded sampling points are characterized by big stars and vice versa and similar patterns by similar stars. Similarities in the PAH patterns are particularly apparent for the last five objects. 

It is impossible to derive a general model that applies to all application domains and all computer systems." In theory, one can predict performance from first principles, but this would require a detailed understanding of every part of the computation and how its memory access (data communication) patterns interact with the computer s hardware and operating system. Except for small computational kernels, it is not feasible to acquire such understanding. In practice, a more satisfactory approach is to construct a fairly high level model using approximate functional forms for the amount of computation, load balance, and overheads. 

---