Industries / Food, farming and forestry / Agriculture and farming / Animal feed
Grinding and mixing
Grinding breaks grain into particles of a chosen size, and mixing blends dozens of weighed ingredients, some added at a few grams per tonne, so that every mouthful carries the same recipe.
5
Milestones achieved, from 1961 to 2025.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1961
Coloured tracers test the mix
Micro-Tracers, Inc. of San Francisco, founded by Sylvan Eisenberg, has made tracers for feed since 1961: grains of iron dyed a bright colour are added to a batch, recovered from samples with the company's rotary detector, spread on paper and counted as coloured spots. Even counts across samples show that a batch is well mixed, and spots found in the batch that follows a medicated one show how much medicine carried over. Kansas State University lists counting coloured iron particles among the ways to test a mixer, and a 2023 North Carolina State University broiler study used the company's red and blue tracers beside salt.
Credit
Micro-Tracers, Inc. (Sylvan Eisenberg)
1994
Mixers graded by spread
A mixer is tested by taking 10 samples from one batch and measuring one marker in each, most often salt. The coefficient of variation (CV), the standard deviation of the 10 results as a percentage of their average, shows how evenly the batch is mixed. T. Herrman and K. Behnke of Kansas State University graded a CV under 10% as excellent, 10 to 15% as good, 15 to 20% as fair and over 20% as poor, with a fix for each grade, and Kansas State's 2017 guidance still uses those grades.
LessonMixing longer does not make up for wear: in a 1986 Kansas State test, a 2-ton mixer with a worn outer ribbon never reached a CV under 10% in 3 to 10 minutes of mixing, and after the ribbon was replaced it did in 4 minutes.
Credit
T. Herrman and K. Behnke, Kansas State University (bulletin MF1172, 1994), as reported by Charles Stark and Marut Saensukjaroenphon (2017)
1995
Finer grain, sorer stomachs
Kansas State researchers fed 160 finishing pigs a corn and soybean meal diet with the corn ground to 1,000, 800, 600 or 400 µm (micrometres, thousandths of a millimetre). Grinding from 1,000 to 400 µm raised weight gain per unit of feed by 8% and cut the dry matter in the pigs' manure by 26% and its nitrogen by 27%, but it took more electricity, slowed the mill and caused more stomach lesions. The authors called 600 µm "or slightly less" an acceptable compromise for corn.
LessonThe grind that converts feed best is not the one that is best for the pig's stomach or the mill's power bill, so the target is a compromise.
Credit
K. J. Wondra, J. D. Hancock, K. C. Behnke, R. H. Hines and C. R. Stark, Kansas State University (Journal of Animal Science, 1995)
2015
One sample, several sizes
The size of ground grain is measured by shaking a 100 g sample through a stack of sieves under ANSI/ASAE S319.4, a standard of the American Society of Agricultural and Biological Engineers, and reporting the geometric mean diameter, an average suited to particle sizes that spread over a wide range. The standard lets a laboratory choose the shaking time, whether to put rubber balls and brushes on the sieves, and whether to add a powder that helps particles flow. Running 18 ground samples through 360 analyses, a Kansas State team found that the flow powder alone lowered the reported mean size by 32 to 36 µm.
LessonA standard with options lets two laboratories report different sizes for the same grain; Kansas State's guidance tells mills to know which procedure their laboratory uses.
Credit
J. R. Kalivoda, C. K. Jones and C. R. Stark, Kansas State University
2025
A 1% cap on carry-over
Medicine left in a mill's mixer and conveyors can pass into the next batch, which was meant to carry none. From 20 May 2025 the European Union caps that carry-over at 1% of the antimicrobial in the last medicated batch. For feed for animals producing eggs or milk for people, and for animals close to slaughter, the cap is the limit of quantification, the lowest amount the reference laboratory method can measure: 25 µg/kg for trimethoprim and 10 µg/kg for tiamulin, for example. Feed used to flush medicine out of the equipment may go into other feed only within these limits.
Credit
European Commission, Delegated Regulation (EU) 2024/1229, on a risk assessment by the European Food Safety Authority (EFSA)
Watch for
A hammer mill digital twin
A hammer mill grinds grain with swinging steel hammers until the pieces are small enough to pass through a screen. Iowa State University says it is building a digital twin, a computer model of its own 10-ton-an-hour CPM Champion hammer mill that simulates grain breaking and air flow, to be fed with live sensor data and linked to plant controls for AI-guided settings, with work on that link starting in 2026. Its trials so far show that slowing the motor alone changes the grind: from 100% to 50% speed, corn ground through a 10/64-inch screen went from 259 to 532 µm, and with a 12/64-inch screen energy use fell from 3.76 to 2.62 kWh per ton at steady output.
Achieved whenthe model's predicted particle size and energy use for a commercial mill's runs are published beside the measured values, with the error stated, from a mill other than Iowa State's own.
Claimed by
George Obeng-Akrofi, Dirk Maier and colleagues, Iowa State University, with Kwame Nkrumah University of Science and Technology in Ghana, 2025 progress report and World Grain, August 2026
Open
01
Pool the mixing trials
A meta-analysis, one statistical study that pools every published feeding trial in which pigs or poultry ate the same diet mixed to different evenness (different mixing times or measured CVs), to test whether the 10% CV target predicts growth, feed conversion or how even the animals' weights are, by species and age.
The briefCredit · unclaimed02
Particle size ring test
An interlaboratory comparison: identical portions of ground corn, sorghum and wheat sent to at least 15 laboratories that measure particle size for feed mills, on at least three continents, with each laboratory's reported mean size, its raw sieve weights and its method options published side by side.
The briefCredit · unclaimed
Open 01 · the brief
Pool the mixing trials
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A meta-analysis, one statistical study that pools every published feeding trial in which pigs or poultry ate the same diet mixed to different evenness (different mixing times or measured CVs), to test whether the 10% CV target predicts growth, feed conversion or how even the animals' weights are, by species and age.
Why it is still open
The 10% target grades the feed, not the animal. A 1994 Kansas State trial found that a lower CV helped nursery pigs but did not affect finishing pigs, and in 2023 a North Carolina State trial found that a 30-second mix in place of 4.5 minutes did not change broilers' weight, feed intake or feed conversion; its authors call research on mixing time requirements "sparse". A 2022 Brazilian trial by Rocha and colleagues found that feeds with CVs up to 22.6% had no adverse effect on broilers older than 12 days. Searches of Europe PMC and Crossref in November 2026 found single trials but no pooled analysis.
Built with what exists
The Europe PMC and Crossref search services to find trials; a low-cost AI model to pull designs and results out of the papers, the repetitive extraction work Meta aims its Muse Ember model at (Notes: "The jobs agents repeat"), with every number checked by a person; R with the metafor package for the pooled analysis.
First step
Collect the trials cited in Kansas State's 2017 bulletin and the 2023 North Carolina State paper, add the 2007 broiler study by Clark, Behnke and Poole, the 2015 finishing-pig study by Paulk and colleagues and the 2022 broiler trial by Rocha and colleagues, and code five of them in a spreadsheet.
Done looks like
A public dataset and preprint covering every trial found, at least 10 from at least two species, each with CV, mixing time and performance coded; a second person re-codes a random 20% of trials and the agreement is published; the CV above which performance falls, if there is one, is stated with its confidence interval.
Open 02 · the brief
Particle size ring test
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An interlaboratory comparison: identical portions of ground corn, sorghum and wheat sent to at least 15 laboratories that measure particle size for feed mills, on at least three continents, with each laboratory's reported mean size, its raw sieve weights and its method options published side by side.
Why it is still open
Kansas State's guidance says laboratories "may obtain differing results because they use different procedures", and its 2015 study measured the effect of method options inside one laboratory only. Searches of Crossref and Europe PMC in November 2026 found no comparison of feed particle size results across laboratories.
Built with what exists
A riffle divider to split samples into identical portions; sealed sample bags and courier post; the ANSI/ASAE S319.4 sieving method; a Python or R script that recomputes every laboratory's geometric mean diameter from its raw sieve weights, so that arithmetic slips are separated from differences in sieving.
First step
Riffle-split one 5 kg bag of hammer-milled corn into 20 portions and ask five laboratories, at university feed mills and commercial firms, to measure it.
Done looks like
Four samples measured by at least 15 laboratories in at least 3 countries; for each sample the spread of reported mean sizes is published, with the share of that spread explained by shaking time, sieve agitators and flow powder.