Flour Quality Control: Ash, Protein, Moisture, Granulation and Farinograph

Quality control in a flour mill has one job: to detect a process drift before the customer does.

The measurement that does most of that work is ash — cheap, fast, and diagnostically powerful. Everything else in this handbook is secondary to getting ash right, though the other tests matter for specific decisions.

Ash: the master measurement

Ash is what remains after burning a flour sample at 550–600 °C. It is mineral residue: bran, germ, and the mineral content of the endosperm itself.

Because bran contains roughly 20× the mineral content of endosperm, ash is a direct proxy for bran contamination. A flour at 0.55% ash and one at 0.70% ash may look identical in the bag, but the second carries 30% more bran-derived material, which changes colour, water absorption and baking behaviour.

Product	Typical ash (% dry basis)
Patent / special grade 1	0.45–0.60
Special grade 2	0.60–0.75
Standard / clear flour	0.75–1.00
Atta / whole wheat	1.20–1.80
Bran (by-product)	4.50–6.50

Run ash every shift on every finished stream. The test takes about 90 minutes including cooling, costs almost nothing, and will catch:

• A worn roll that is polishing rather than cutting bran
• A torn sifter screen leaking oversize into flour
• A destoner that has lost air volume and is passing stones
• A purifier whose air balance has drifted
• A scourer with worn beaters that has stopped scouring

Any of these can degrade ash by 0.05–0.20 points without any other visible symptom. Ash is the only measurement that catches all five.

Moisture: the margin measurement

Flour moisture is both a legal and a commercial number.

Legal: most jurisdictions specify a maximum moisture for flour sold by weight. Exceeding it means selling water as flour, which is a labelling offence.

Commercial: flour is sold by weight. Every 0.1% of excess moisture in a 50 kg bag is 50 grams of water you sold at flour price. On 60 t/day that is 60 kg of water per day — 20 tonnes a year.

Target moisture varies by market but 13.5–14.2% is typical for shelf-stable flour. Hold it within ±0.15%.

Drift almost always traces to one of four sources:

1. Tempering water not reaching target — check the dampener calibration against an oven-dry reference monthly
2. Ambient humidity change with an unsealed pneumatic conveying system
3. Wheat moisture variation that the automatic dampener has not compensated for
4. Packaging line not compensating for ambient conditions

Protein: the specification measurement

Protein determines what the flour can be used for, and it is set by wheat selection far more than by milling.

Application	Protein (N × 5.7, 14% moisture basis)
Biscuit / cookie	8.0–9.5%
Cake / pastry	8.5–10.0%
Noodle	10.0–11.5%
Steamed bread	10.5–12.0%
Pan bread	11.5–13.0%
High-protein / hearth bread	12.5–14.0%

Milling can shift protein slightly by stream selection — patent streams run higher protein than clear streams because the aleurone layer is protein-rich. But you cannot mill low-protein wheat into bread flour. If the wheat is 9% protein, the flour will be 9% protein.

Test protein on wheat intake and on finished flour. The delta tells you whether your stream selection is working.

Granulation: the under-tested dimension

Two flours at identical ash and protein can behave completely differently in a bakery if their particle size distributions differ.

A flour with a narrow distribution around 100 µm behaves very differently from one with 30% passing 60 µm and 20% over 150 µm. The fine fraction hydrates fast and feeds yeast; the coarse fraction hydrates slowly and contributes to crust colour.

Granulation is controlled by:

• Roll gap and differential on reduction passages
• Sifter screen aperture on the flour passages
• Impact detacher settings (flaking and de-flaking)

Test with a sieve stack (typically 100, 150, 200, 250 mesh) on every significant product change, and whenever a customer reports a behaviour change.

Farinograph and extensograph: for decisions, not for shifts

A farinograph measures dough resistance over time as water is added and the dough is mixed. It gives you water absorption, development time, stability and degree of softening.

Parameter	Bread flour target	Biscuit flour target
Water absorption	58–62%	50–54%
Development time	4–8 min	1–3 min
Stability	8–15 min	1–3 min
Degree of softening	30–60 BU	100–150 BU

An extensograph measures resistance and extensibility, which predicts how dough behaves during shaping and proofing.

These are decision instruments, not production instruments. Run farinograph when:

• Wheat origin or blend changes
• A customer complains about dough behaviour
• You are developing a new product specification
• You are qualifying a new wheat supplier

Running farinograph every shift tells you nothing you did not already know and consumes laboratory time better spent on ash.

Sampling: where quality control actually fails

Most quality failures are sampling failures, not measurement failures.

Stream	Frequency	Sample point
Wheat intake	Every truck	After weighing, before cleaning
Cleaned wheat	Every shift	After final aspiration, before 1B
Tempered wheat	Every shift	Bin outlet
Each break flour	Every shift	Sifter outlet
Each reduction flour	Every shift	Sifter outlet
Finished flour, per grade	Every hour	Before packer
Bran	Every shift	Bran finisher outlet

The reason to sample each stream separately rather than only the blend is diagnostic. If total flour ash rises but only the 3M stream is out of specification, you know exactly which roll or screen to inspect. If you only test the blend, you know there is a problem and nothing about where it is.

Complaint investigation: a worked example

A bakery reports that the last three deliveries produce bread with a grey crumb and poor volume.

Step 1 — Retrieve retained samples. Every mill should retain a 500 g sample of each production lot for 30 days. If you have them, compare the complained lot against a known-good lot.

Step 2 — Measure ash. If the complained lot is 0.68% versus a specification of 0.55%, you have bran contamination. If ash is within specification, the problem is not bran.

Step 3 — Measure protein. If protein has dropped from 12.2% to 10.8%, the problem is wheat selection or blending, not milling. Check the intake records for that period.

Step 4 — Measure granulation. If ash and protein are correct but granulation is coarser, check the sifter screens and the detacher.

Step 5 — Check stream data. If the shift records show ash rising only on the 3M stream, inspect the 3M roll for wear or damage.

The whole investigation takes about four hours if the records exist, and about four weeks of arguing if they do not.

Laboratory equipment: minimum viable and full

Tier	Equipment	Approximate cost	Covers
Minimum	Moisture oven, muffle furnace, analytical balance, sieve shaker	USD 4,000–7,000	Moisture, ash, granulation
Standard	+ Kjeldahl or NIR protein, falling number	USD 12,000–20,000	+ Protein, sprout damage
Full	+ Farinograph, extensograph, gluten washer	USD 45,000–80,000	+ Rheology, product development

An NIR analyser is worth considering above 100 t/d: it measures moisture, protein and ash in 30 seconds with no consumables, which lets you test every hour instead of every shift.

The quality system, summarised

1. Sample every stream, every shift. Diagnostic resolution comes from stream-level data.
2. Measure ash first, always. It is the cheapest and most informative test in the mill.
3. Hold moisture to ±0.15%. It is both a legal and a margin issue.
4. Test protein on wheat intake. You cannot fix in the mill what arrived in the truck.
5. Test granulation on change, not on schedule. It matters for customer disputes.
6. Retain samples for 30 days. This is the difference between a four-hour investigation and a four-week argument.
7. Trace every deviation to a machine. A quality number without a machine attribution is just anxiety.