Case study: Milling overload reduced by process understanding

Problems of variable product quality were being experienced during pharmaceutical milling. Material was fed to the mill by a rotary valve, and ground using turbulent air jets. Ground material exited via the dynamic classifier.

Figure 1: The Milling Process

Periodic sampling of the milling product took place at a fixed period of D t. Samples measured in this way showed the milling process to be acceptably close to the setpoint.

Figure 2: Manual Milling Samples

However, when real-time particle size measurements were undertaken, using an Insitec on the outlet of the mill, the milling process was revealed to be much less stable.

Figure 3: Milling Process with Real-Time Particle Size Measurement

The graph below shows that the median size (Dv 50) (red line) fluctuates significantly during milling. This corresponds to a periodic fluctuation in the measured transmission (blue line), which is related to the milling load. 100% transmission corresponds to there being no particles present, and 0% means that so many are present that no light is reaching the detector.

Figure 4: Milling Process Size Fluctuations

Periodic fluctuations that are the consequence of forcing material through the milling process, lead to "short-circuiting" of the classifier. This affects the Dv 50 of the particles. The higher the loading, the more pronounced is the effect, resulting in a larger (coarser) median size. As more and more particles are forced through the classifier, the classifier motor draws more and more current (green line). 

The remedy to this situation was to use the measured transmission value to drive the feed rotary valve.

Figure 5: Milling Process with Measured Transmission Valve

As a result transmission oscillates around a mean value (indicative of a simple control scheme) and critically the median size is continually close to the setpoint.

Figure 6: Milling Overload Reduced