A centrifuge that runs but doesn't recover is paying its running cost for nothing. On a weighted mud it dumps barite you bought; on dewatering it leaves fines in the system that drive dilution.
Likely causes
Cut point wrong for the duty — G and pond not set for barite vs fines.
Scroll / bowl wear changing the effective cut.
Wrong duty selected — trying to recover barite and strip fines on one setting.
Feed rate too high — short residence time, poor separation.
Under-run — not enough hours on the colloidal load.
Overlapping particle sizes — barite and drilled fines close in size, hard to split.
High viscosity reducing settling in the bowl.
Solids load beyond what the single machine can process in the time available.
How to diagnose it
Confirm which duty the machine is set for vs what's required.
Check G-force and pond against the duty chart.
Trend mud weight / LGS against centrifuge run-hours.
Sample effluent and discard — are you losing barite, or leaving fines?
Run a PSD if the cut point is in question.
The fix — step by step
Set one duty and configure G, pond and differential to it.
On weighted mud, run the recovery cut (lower G) and return barite; send overflow to a second-stage fines machine if available.
Increase run-hours on colloidal load before reaching for dilution.
Reduce feed rate to lengthen residence time and sharpen the cut.
Replace worn scroll / bowl components shifting the effective cut.
Confirm it's fixed
✓ Verify: Mud weight or LGS responding to centrifuge run-hours; a discharge sample confirms the machine is cutting at the intended point.
Field note. “Running” is not “recovering.” The most common waste I see is one centrifuge asked to do two jobs — recover barite and strip fines — on a single setting, so it does neither well. Pick the cut, prove it with a sample, and if you need both jobs, you need two stages.