Track Flour Mill Yield Per Batch, Not Per Year

Extraction rate is the simplest number in a flour mill, and it is the one most small mills cannot state for last Tuesday.

It is flour produced divided by wheat milled, expressed as a percentage. Mill 1,000 kg of wheat, get 740 kg of flour, and your extraction rate is 74 per cent. That is the entire calculation. The difficulty is never the arithmetic — it is that most mills do it once a year, on a yearly total, and a yearly total cannot tell you what went wrong.

Where people get the denominator wrong

BAKERpedia defines extraction rate as the amount of white flour extracted from a given weight of clean and conditioned wheat, expressed as a percentage of the wheat entering the first break rolls. Two words there decide whether your number means anything: clean and conditioned. Measure against wheat as it arrived on the truck and you are charging screenings, stones and dust against your milling performance, so a dirty consignment looks like a bad run.

Neither basis is wrong, but pick one and never switch, because switching quietly moves your yield by a percentage point or two and nobody will remember why. Record both weights anyway: the gap between them is your cleaning loss.

Typical ranges, and why they are only a starting point

Published figures give a rough sense of where a mill sits. Treat them as reference points rather than targets — your equipment, wheat and product mix may put you legitimately outside them.

BAKERpedia notes that in efficient roller mills extraction rates of 72 to 76 per cent are normally obtained, and that theoretical yields approaching 83 per cent are never achieved. That ceiling is not arbitrary: the Wheat Foods Council puts the endosperm at about 83 per cent of kernel weight, bran around 14.5 per cent and germ around 2.5 per cent. You cannot extract white flour that was never there.

Whole wheat milling is a different exercise. Extraction approaches 100 per cent by definition, so the useful question becomes total loss — dust, spillage and what stays in the machine.

Why the annual figure lies to you

Suppose you milled 1,200 tonnes of wheat last year and produced 888 tonnes of flour. That is 74 per cent. It looks respectable, so you stop thinking about it.

Now split the year into 240 runs. You may well find that one supplier consistently delivers 71 per cent while another delivers 76. That the night shift runs two points below the day shift. That yield fell in March and never recovered, which is roughly when the rolls were last adjusted.

The annual average contains all of that and shows none of it. You are not measuring to produce a figure for the accountant, you are measuring to generate comparisons. Three of them pay for the effort almost immediately:

• Supplier against supplier. Same mill, same settings, different wheat. A two-point yield gap between suppliers is a real price difference your purchase price does not show.
• Shift against shift. Same wheat, same machine, different people. Consistent gaps point at technique, at how the mill is fed, or at how carefully sacks are weighed off.
• This month against last. Same wheat, same people, drifting numbers. That is usually the machine — roll wear, sieve blinding, a worn brush.

By-products are stock, not waste

Most small mills track flour and ignore everything else. Bran, fine offal, semolina and germ are physical stock with a market value. Bran sells to feed dealers and cattle owners, and semolina and coarse fractions often sell at a premium over ordinary flour.

If your records show wheat in and flour out only, every kilogram that did not become flour looks like loss. Some of it is product you sold and never posted against the batch. That distorts margin both ways: you understate the run's profit by ignoring by-product revenue, and you understate the losses too, because by-product sold for cash and never entered into stock looks the same in your books as by-product that walked out of the yard.

The fix is to make each output a stock item with its own weight on the run record. Wheat in at the top; flour, bran, semolina and any other fraction underneath. Then the line nobody enjoys: wheat in minus everything out, your unaccounted loss. In a well-run mill it is small and stable, and its value is not the number but the day it changes.

By-product sold to feed dealers is a sale like any other and should leave a proper invoice behind it — our guide to Saudi VAT invoice requirements lists the fields a tax invoice has to carry. Confirm the VAT treatment of animal feed with your accountant rather than assuming it matches your flour sales. Mills with a retail counter usually run that side through POS software so sack sales land in the same stock records.

Moisture will make you chase problems that do not exist

Conditioning adds water deliberately, so the bran becomes tough, the endosperm becomes brittle and the two separate cleanly. That added water has weight. Wheat tempered to a higher moisture produces a different apparent yield than the same wheat tempered less, with nothing about the milling having changed. The same applies at the front door: wheat arriving at 10 per cent moisture and wheat arriving at 13 per cent are not the same purchase per kilogram.

So record incoming moisture, water added and rest time. Conditioning time varies with grain hardness, and hard wheat generally needs a longer rest than soft, so one procedure applied to two consignments is not the same treatment. Without those fields you will spend a week adjusting rolls that were fine.

The run record

Keep it short enough that an operator will actually complete it at the end of every run.

• Date and run number
• Supplier and consignment reference
• Wheat in — gross received weight and cleaned weight
• Moisture — incoming, and water added at conditioning
• Each output out — flour by grade, bran, semolina, fine offal, weighed separately
• Operator and shift
• Notes — stoppages, roll adjustments, anything unusual about the wheat

That is perhaps two minutes. A run without a record is a run you cannot learn from later, so an incomplete sheet is a real problem rather than a small untidiness.

What three months of this gives you

One run tells you nothing. Three months of runs tells you almost everything.

Average the extraction by supplier. Then by operator. Then plot the figures in date order and look for the step change rather than the wobble — day-to-day variation is mostly noise, but a level that shifts and stays shifted is a machine or a procedure, and it has a date on it.

Then do the calculation that actually matters, which is not extraction rate at all: total value out per tonne of wheat in, flour value plus by-product value against wheat cost. A supplier with lower flour extraction but a heavier, cleaner bran fraction may be the better buy, and the extraction percentage alone will never tell you that.

Weigh it, do not estimate it

Every argument above collapses if the weights are guesses. Sack counts multiplied by an assumed sack weight will hide a two per cent yield problem indefinitely, because the assumption absorbs the error. Sacks are not consistent, and fill levels vary by operator.

So weigh the wheat in and weigh each output. Check the scale against a known weight on a fixed schedule, and record actual weights rather than nominal ones. If a sack is nominally 50 kg and weighs 48.6 kg, write 48.6.

Yield differences worth acting on are frequently one or two percentage points, exactly the size of error estimating introduces. Estimate, and you will only ever see problems too large to have needed records at all.

If keeping this in a book is more than you want to maintain, our flour mill management software records production runs with wheat in, multiple outputs and by-products as separate stock items, and reports extraction by supplier, operator and period. It comes with a 7-day free trial and is paid after that; the invoice generator and CV Studio are free to use.