1. What a D90 Shift Actually Means
Particle-size distribution describes the range and relative distribution of particle sizes in a cement sample. D10, D50 and D90 are percentile diameters on the cumulative distribution: they describe different parts of the measured distribution and should be interpreted together with the full curve. Blaine fineness and sieve residue measure different properties, so two cements can show similar Blaine values while having materially different PSD shapes.
A verified D90 increase usually means the coarse tail has moved to larger reported particle sizes. That may reflect a real grinding or classification change, but it can also be created by sampling, agglomeration, dispersion or method changes. Treat D90 as a diagnostic signal rather than a direct instruction to adjust the separator.
2. Verify the Measurement Chain Before the Process
ISO 13320:2020 provides the current general framework for laser-diffraction particle-size analysis, including sample preparation, dispersion, measurement, repeatability and instrument qualification. Plant laboratories should still follow the validated method established for their specific cement and instrument.
3. Use Complementary Evidence
| Observed pattern | First interpretation | Next check |
|---|---|---|
| D90 changes, repeated preparations disagree | Measurement/preparation variation is still plausible | Sampling, dispersion, cleanliness and method |
| D90 and coarse sieve residue both rise | A real coarse-fraction process shift becomes more plausible | Separator and grinding trends |
| D50 and D90 move together | Broader PSD shift | Grinding/classification and product composition |
| PSD changes but Blaine is similar | Distribution shape may have changed without a similar change in specific surface | Review full PSD and residue rather than forcing agreement |
4. D90 Troubleshooting Sequence
- Confirm sample identity, cement type and sampling time.
- Repeat the same prepared portion to check short-term measurement repeatability.
- Prepare a second portion from the same laboratory sample to test preparation consistency.
- Review the full cumulative and differential PSD curves, not D90 alone.
- Compare D10, D50 and D90 with relevant sieve-residue and Blaine trends.
- Check instrument cleanliness, background and the saved analytical method.
- Confirm the dispersion procedure and sample concentration were unchanged.
- If uncertainty remains, obtain a fresh representative process sample.
- Only after the PSD shift is confirmed, correlate it with separator settings, mill feed, ventilation, circulating-load indicators, grinding-component condition, grindability and product changes.