A decanter does not have one magic setting. It has a coupled operating window: bowl speed changes settling force; flow changes residence time and mass loading; differential speed changes solids transport; torque reveals conveyor resistance; and pond depth changes clarification volume and beach exposure. Moving one variable forces the others to respond.
The five-variable operating system
| Variable | Primary physical effect | Typical consequence if increased |
|---|---|---|
| Bowl speed | RCF and compaction pressure | Faster radial settling, more wear and stress |
| Feed flow | Hydraulic residence time | More throughput, less time to clarify |
| Feed solids concentration | Solids mass load | More conveyor duty and torque |
| Differential speed | Conveyor transport rate | More solids transport, shorter solids residence |
| Pond depth | Liquid volume and dry beach | More clarification volume, potentially wetter cake |
These effects are directional, not universal promises. For example, lowering differential speed commonly increases solids residence and cake dryness, but only until the bed becomes too deep, torque rises or transport becomes unstable. Increasing bowl speed may improve clarification, but may also compact a cohesive bed so strongly that the conveyor struggles to move it.
Separate hydraulic load from solids load
Two feeds at 400 gpm are not the same duty. If one contains 2% solids by volume and the other 6%, the second delivers roughly three times the volumetric solids load before density corrections. The centrifuge must process both the liquid flow and the solids mass. Hydraulic overload shows as poor clarification because residence time is short. Solids overload shows as rising torque, unstable differential speed, heavy cake discharge, vibration risk or protective trips. Both may occur together.
Volumetric solids load = Q × Cv
Nominal solids mass load = Q × Cv × ρs
This requires Q as total slurry volumetric flow, Cv as true solids volume fraction and ρs as representative particle density. For mixed HGS/LGS, sum Q × Cvi × ρi. Captured conveyor load is lower when recovery is below 100%.
The correct order of optimization
- Define the objective. LGS rejection, barite recovery and dewatering require different outlets and different definitions of success.
- Stabilize feed source, mixing and pump delivery. Optimization on a surging feed is not repeatable.
- Set bowl speed inside the rated range to create the required RCF, not the largest available number.
- Set a conservative feed rate and observe centrate, cake, torque, vibration and motor load.
- Adjust differential speed in small steps while protecting torque margin.
- Adjust pond depth only with the correct hardware/procedure and after considering dry-beach length.
- Run long enough to reach a new steady condition before sampling.
Why single-variable tests fail
If bowl speed is raised while the feed pump simultaneously changes flow, the result cannot be attributed to speed. If differential speed is lowered but the incoming solids concentration doubles, rising torque does not prove the setting was wrong. A defensible test changes one controlled variable, holds the others as stable as practicable, waits for transport delay and bowl inventory to respond, then samples all relevant streams.
A practical test matrix
| Test | Hold constant | Measure | Decision |
|---|---|---|---|
| Flow sweep | Bowl and differential speed | Feed/centrate solids, torque, cake rate | Hydraulic capacity |
| Bowl-speed sweep | Flow and solids source | RCF, recovery, cake, vibration | Useful acceleration window |
| Differential sweep | Flow and bowl speed | Torque, cake dryness, centrate | Transport window |
| Mass-balance run | Stable full setting | Flow and solids in all streams | True recovery and loss |
Interactions that reverse simple rules
Simple operating rules fail when another variable becomes limiting. Lowering differential speed can dry cake at steady load, but at higher incoming solids load it can build inventory until torque control accelerates the scroll, producing a wetter and less stable discharge. Raising bowl speed can clean centrate at moderate load, but at high load it can capture and compact more solids than the conveyor can remove. Deepening the pond can increase clarification volume, yet reduce exposed beach and retain more liquid in cake.
Process delay and steady state
Controller values respond quickly; material inside the bowl does not. After a change, the liquid flow field, bed thickness, cake transport and outlet composition need time to establish a new condition. Sampling immediately after a change mixes old and new regimes. Record the time of every change, estimate feed-to-outlet delay, watch torque and discharge stabilize, and use a defined composite sampling window.
Optimization scorecard
A useful scorecard gives no single metric veto power except safety limits. Track centrate solids, cake liquid, solids recovery, liquid recovery, HGS/LGS split, stable throughput, torque margin, vibration, energy and wear exposure. Weight those metrics according to the declared objective. This prevents a visually clean centrate or dry cake from disguising an uneconomic total result.
Common questions
Which centrifuge variable should be adjusted first?
First define the treatment objective, stabilize the feed and establish the OEM-approved baseline. The test sequence for bowl speed, feed and differential speed depends on machine design, control mode and the limiting process mechanism.
Can one setting work for every hole section?
No. Particle-size distribution, solids concentration, mud density, viscosity, barite content and treatment objective change, so the operating window must be revalidated.
How long should a test setting run?
Long enough for liquid flow, solids inventory, conveyor transport and outlet composition to become representative. Confirm stability from trends and a defined sampling window rather than a universal time.
