Level Measurement and Inventory Control in Food Plants

A level sensor in an ingredient silo has to do three things at once, and plants usually specify only the first. It has to measure, it has to be cleanable, and it has to keep working while the silo is being filled.

Cleanability is the one that gets missed. Any sensor that intrudes into the vessel creates a ledge, and a ledge in a flour or milk powder silo is a place product sits, ages and eventually falls into a batch. Threaded fittings, protruding horns and sensor cables all do this.

The filling condition is the second trap. Pneumatic filling puts a dense dust cloud in the headspace, and ultrasonic measurement, which works perfectly on a quiet silo, loses its echo completely. The reading goes away at exactly the moment somebody wants it.

And the inventory question is separate from the level question. A powder surface is a cone, a crater or something stranger, so a single point reading is a poor estimate of mass. If the number is used for stock control or for a batch, weighing is a different and better answer.

In a food plant the sensor is also a product contact part, which rules out several of the arrangements that would otherwise read best.

A sensor with a flush, hygienic mounting that creates no ledge inside the vessel and can be cleaned with the silo.

A measurement that works during pneumatic filling, so the reading is available when it matters.

Overfill protection from an independent point level device, and inventory derived from weighing where the figure is used for stock or for a batch.

For a pharmaceutical plant the question is what the number is used for, because a level converted to mass with an assumed bulk density and entered into a batch record is a measurement nobody made.

Technology is matched to the filling condition first. Radar works through dust far better than ultrasonic, which is why it dominates in pneumatically filled silos; the antenna is then selected for the dielectric constant of the powder, since flour and starch reflect weakly and a small antenna on a low dielectric material gives a poor signal.

The mounting is then designed. A flush hygienic flange with no internal protrusion, a nozzle short enough not to produce false echoes, and a position that avoids the filling stream, because a sensor looking into a falling column measures the column.

Overfill protection comes from a separate device. A continuous measurement is an operational signal; a safety function needs an independent point level switch with its own path to the trip, because a single sensor that fails silently takes both functions with it.

Inventory is then decided. Where the number goes into stock control or a batch record, load cells under the silo give a defensible figure; a level reading converted to mass with an assumed bulk density gives a number whose error nobody can state.

• Radar rather than ultrasonic where filling is pneumatic, with the antenna matched to the dielectric constant.
• Flush hygienic mounting with no internal ledge and a short nozzle.
• Sensor positioned away from the filling stream.
• Independent point level switch for overfill protection, with its own trip path.
• Weighing used where the inventory figure goes into stock control or a batch record.
• Sensor and mounting included in the silo cleaning routine.

• Filling method, since pneumatic filling rules out some technologies during the one period that matters.
• Dielectric constant of the powder, which decides antenna selection for radar.
• Hygienic standard and whether the mounting can be cleaned with the vessel.
• Use of the number: operational signal, stock control or batch record, which decides level versus weight.
• Overfill protection requirement and the independence it needs.
• Surface profile of the powder, which limits how good any single point reading can be.