Gravimetric Feeding in Pharmaceutical Production

Feeding in a pharmaceutical plant is either a batch charge or, increasingly, a continuous process where the feeders define the formulation in real time. In the second case the feeder is no longer a utility; a rate deviation becomes a composition deviation in product that is already being made.

The quantities are small and the materials are difficult. An active at a low percentage is often the most cohesive component in the formulation, prone to bridging and to a bulk density that changes with every handling step, and it is exactly the one whose rate matters most.

Containment complicates the mechanics. Charging a potent powder into a hopper, refilling it during operation and cleaning the machine afterwards are three operations where the operator meets the compound, and the enclosures that solve that add weight, stiffness and connections to a machine whose accuracy depends on being mechanically free.

And everything has to be demonstrable. An accuracy figure from a catalogue is not evidence; a catch test on the installed machine with the actual material is.

Feed rate accuracy defined at the actual rate and at a time scale that matches the process, demonstrated by catch tests on the installed equipment with the real material.

Containment through the whole cycle, including refill and cleaning, achieved without adding force paths that corrupt the weighing.

A refill strategy that keeps the blind period short and infrequent, with the behaviour during refill characterised rather than hoped for.

In plastics compounding the feeder has to survive a bulk density range and a fibre length, and a light filler dropped into the throat entrains air before it is ever weighed.

The accuracy requirement is derived from the process, not from the supplier. A continuous direct compression line has a residence time and a mixing capability that will smooth variation over some period and not over a shorter one, and the feeder specification follows from that: a figure at a stated rate over a stated sample interval.

Discharge devices are then chosen for the material. Twin screws for cohesive actives and for most excipients; a single screw where the material is genuinely free flowing. Hopper agitation is kept to the minimum that maintains flow, since a vertical agitator that works the powder changes its bulk density and therefore the feed factor continuously.

Containment is designed around the three contact operations. Charging through a split butterfly valve or from a closed liner, refill without opening the hopper to the room, and a cleaning route that reaches everything. The key mechanical rule is that the containment must not restrain the weighed assembly: flexible connections stay flexible, and nothing bridges them.

Performance is then characterised and documented. Catch tests across the rate range, feed factor behaviour through a refill cycle, and the response to a deliberate disturbance. That body of evidence is what makes the feeder part of a validated process rather than a machine with a datasheet.

• Accuracy requirement derived from process residence time and the mixing available downstream.
• Discharge device matched to the material, with twin screws for cohesive actives.
• Agitation limited to what maintains flow, to keep the feed factor stable.
• Containment designed for charging, refill and cleaning without restraining the weighed assembly.
• Refill short and infrequent, with behaviour through the refill characterised.
• Catch tests on the installed machine with real material, documented as evidence.

• Rate and tolerance for each component, and the sample time over which the tolerance applies.
• Process residence time, which decides whether short term variation matters or is smoothed out.
• Flow behaviour of the active, measured, since it is usually the most cohesive and most critical component.
• Containment band, which decides charging, refill and cleaning arrangements.
• Cleaning validation and whether every internal surface can be reached and swabbed.
• Evidence requirement, which decides how much characterisation testing is planned from the start.