(144a) Data-Mining and Modeling of a Centrifuge | AIChE

(144a) Data-Mining and Modeling of a Centrifuge


Sharma, P. K. - Presenter, Bristol-Myers Squibb Co.
Wilson, A. - Presenter, Howard University
Tabora, J. E. - Presenter, Bristol Myers Squibb

Determining the optimal operating conditions (spin speed and feed rate) to use on a plant centrifuge is a complex scale-up problem often warranting an empirical approach in the plant until a satisfactory filtration performance is achieved. For very compressible cakes, setting the spin speed too high or feeding the slurry too fast can result in poor filtration rates and possible permanent compression of the cake; spinning the basket or feeding the slurry too slow also results in poor filtration rates. The optimal operating conditions for filtration performance are dependent upon both the equipment and the cake properties. Although lab pressure filtration experiments can provide information on cake compressibility, it is difficult to translate this into operating conditions for a centrifuge. A one-dimensional centrifuge model proposed by Chan, Kiang and Brown from Bristol-Myers Squibb was used to develop a simulation tool in Dynochem for centrifuge operations. The model forms an integral part of an overall workflow for scale-up. In conjunction with data-mining strategies from the plant, it is possible to use the model to directly fit cake properties such as compressibility and then run simulations in different centrifuges to find the optimal operating conditions for the equipment. Since poor filtration performance can often be the cause of a significant bottleneck, this type of work can have a significant impact on improving process cycle times.