Analysis of residence time distribution data in horizontal fluidized beds

In chemical, pharmaceutical or food industries many materials are available as solid intermediates and have to be converted or adapted in subsequent processes. In order to decrease labor and downtime, a continuous process is desired, in spite of the disadvantage of a distribution of the particle residence time and hence of the characteristic particle properties. Still because of economic reasons a continuous process is preferred while minimizing dispersion. The general way to determine the dispersion coefficient, indicating the degree of dispersion, is by means of the method of moments. A problem occurs when the measured distribution is not smooth but rather scattered or when the tail of the curve is cut. In this case the calculation of the dispersion coefficient becomes imprecise and other methods have to be applied. For this reason, an one-parametric tank-in-series model was used and fitted to the measured data. The fitted parameter was used to determine the dispersion coefficient. Furthermore a dynamic model based on population balances has been discretized and implemented. Optimization was accomplished by minimizing the difference of measured and simulated data by using the method of least squares. The results of all methods have been compared to each other and depicted graphically. (C) 2015 The Authors. Published by Elsevier Ltd.