An experimental study on flow behaviour of pharmaceutical powders during suction filling.

Die filling is a crucial step in the pharmaceutical tablet manufacturing process. For industrial-scale production using rotary presses, suction filling is typically employed due to its significant efficiency advantages over gravity filling. Despite its widespread use, our understanding of the suction filling process remains limited. Specifically, there is insufficient comprehension of how filling performance is influenced by factors such as suction velocity, filling velocity, and the properties of the powder materials. Building on our previous research, this study aims to further investigate the effects of powder properties and process parameters (e.g., filling velocity, suction velocity, fill depth) on suction filling behaviour. A systematic experimental investigation was conducted using a model suction filling system, considering both cohesive and free-flowing pharmaceutical powders. The effect of fill depth on suction filling of these powders was examined at different filling and suction velocities. The results demonstrate that two distinctive flow regimes for suction filling can be identified: slow filling and fast filling. These regimes are delineated by a critical filling-to-suction velocity ratio. In the slow filling regime, the filling-to-suction velocity ratio is lower than the critical ratio, meaning that the filling phase is slower than the suction phase. Conversely, the fast filling regime occurs when the filling-to-suction velocity ratio exceeds the critical ratio, implying that the filling phase is faster than the suction phase. This study reveals, for the first time, that when the powder flow pattern during suction filling is dominated by plug flow, full die fill (i.e., the fill ratio equals unity) is achieved in the slow filling regime. However, in the fast filling regime, incomplete die fill is obtained. It is also found that when plug flow prevails during fast filling, the fill ratio has an inverse correlation with the filling-to-suction velocity ratio. This study further reveals that when the plug flow assumption is valid, the filling ratio at various fill-to-suction velocity ratios can be well predicted mathematically. Furthermore, it is also found that once the powder flow pattern differs from the ideal plug flow, which could be induced by the filling conditions and powder cohesion, the fill ratio can be overpredicted.