A concentrator measures slurry density at half a dozen places, and the reason differs at every one. Treating them as one measurement problem is the usual route to an instrument that reads correctly and controls nothing useful.
Grinding and classification
Mill discharge / cyclone feed. Density here sets classification efficiency. A cyclone separates by settling velocity, and settling velocity depends on the density and viscosity of the medium the particles are falling through. Feed density drifting up shifts the cut point coarser at constant pressure; drifting down sends fines to the underflow and recirculates them. This is the highest-leverage density point in the plant, because grinding is where the energy goes.
Cyclone overflow. The product density that flotation will see. Usually the tightest specification in the circuit.
Cyclone underflow. Very high solids, coarse, extremely abrasive. Hard on any wetted instrument, and the classic place to find a meter that was specified on accuracy and selected without regard to wear.
Flotation
Feed density affects residence time, reagent dosing per tonne, and froth stability. Note the complication specific to flotation circuits: the process deliberately introduces air. Any density instrument in or near a flotation circuit has to be judged on how it behaves with entrained gas, not merely on its accuracy on degassed slurry. An instrument that reports a gas index alongside density lets an operator separate "more solids" from "more bubbles" instead of guessing — a distinction that matters because both push a naive density reading in the same direction.
Thickening and dewatering
Thickener feed drives flocculant dosing. Dosing on a measured feed density instead of a fixed rate is one of the more reliable reagent savings available in a plant.
Thickener underflow controls the underflow pump and the water balance. High solids, and the pump can be damaged by running against material that has thickened past design.
Tailings
Density on the tailings line governs deposition behaviour, water recovery and — with paste and thickened tailings — whether the material behaves as designed when placed. It also flags an upset upstream faster than most other signals do.
What is different about mining service
- Abrasion is the constraint, not accuracy. Most principles can reach the accuracy a concentrator needs. Far fewer survive the duty. Selecting on the accuracy line of a datasheet is the standard error.
- Grind changes. Ore hardness varies by blend and by depth. Methods whose calibration depends on particle size distribution can drift when nothing in the mass balance has changed — see the attenuation coefficient explainer.
- Scale and coating. High-pH circuits scale. A coated sensing surface biases the reading slowly enough that it is trusted long after it should not be.
- Access. Some of the most valuable measuring points are on lines that cannot be shut down for a tapping. Clamp-on and externally mounted instruments earn their place here.
The installation mistakes that keep recurring
- Horizontal pipe at low velocity. Solids stratify and the meter reads a layer, not the line. A vertical rising section is the default answer, and it fixes more field complaints than any calibration adjustment.
- Immediately downstream of a pump. Turbulence and entrained air, both worst right at the discharge.
- On a line that runs empty or part-full. Every principle assumes a full pipe. Interlocking the reading to a pump-running signal costs nothing and prevents a whole category of false alarms.
- Calibrating against a badly taken sample. If the sampling point is far from the instrument, the comparison measures transport lag as much as instrument error. Sample as close to the sensor as the piping allows, and record the time.