Mill scale-up: Developing a low-cost, low-risk procedure
P&G successfully used gSOLIDS® for a new, model-based scale-up procedure that reduces the time and cost required for experimentation and minimises the risk associated with the scale-up of an impact mill.
The procedure was initially developed and validated for an inexpensive material before applying the validated procedure to a relatively expensive material.
Mill selection and scale-up still heavily relies on trial-and-error approaches involving extensive experimentation and supplier experience. In general, these approaches are fraught with high financial risks as they often incurs large costs in resources, process shutdowns, equipment fouling and wasted materials.
Development of an effective scale-up procedure required P&G to:
- identify a model that clearly and effectively separates the effect of powder properties and equipment characteristics on milling performance
- design small sets of targeted experiments to accurately estimate the equipment and material related parameters
The Vogel and Peukert model (Chem. Eng. Sci., 60, 5164) was used as it contains both parameters related to the type and scale of the mill and material properties. As the model is highly non-linear, advanced parameter estimation was required to determine the model parameters accurately.
Model implementation, validation and application
gSOLIDS was used to perform all steps of the scale-up procedure:
- implement the Vogel and Peukert model
- estimate the material and equipment parameters
- validate the model against data not used for parameter estimation
- run the model for predictive purposes
The material and equipment parameters were estimated through advanced parameter estimation, using multiple dynamic lab-scale experiments.
Validating the scale-up procedure
To manage cost, the initial lab-scale and pilot-scale experiments used a cheaper powder material to test the scale-up procedure.
A bench top pin mill was used to estimate the relevant material and equipment parameters and the predictive quality of the model was investigated and found to predict the size distribution in the mill output stream well.
Since the same material was used in the pilot-scale as in the lab-scale only the equipment parameters were estimated in these experiments. The predictive capabilities of the model were found to be very good for the larger particles and satisfactory for the fines over a range of mill speeds.
Taking the validated scale-up procedure to the plant scale
Considering the success of the prediction moving from lab-scale to pilot-scale, the lab-scale experiments were repeated using the real, more expensive, material to estimate the real material properties.
The plant-scale mill had four cages with different diameters and varying RPM. The material properties from the lab-scale were used to predict the output of the plant-scale mill without additional estimation, significantly reducing the time and costs required for experimentation.
An average mill diameter for the four cages was used and single collisions were assumed. It was found that the predictions matched the breakage very well over a range of RPM. Further estimates could be made to improve the predictions of the fines, however, this was not considered to be important in this case.
Using gSOLIDS a model-based scale-up procedure was successfully developed and executed. This procedure significantly reduces time and cost required for experimentation and minimises the risk associated with scale-up of mills.