Industries / Food, farming and forestry / Agriculture and farming / Animal feed
Delivery to farms
Delivery is getting finished feed from the mill into a farm's bins or bags, by truck and route, on time and without carrying disease from farm to farm.
4
Milestones achieved, from 1956 to 2022.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1956
From sacks to bulk
George Rogers and Harry Woodworth of the University of New Hampshire reported that feed dealers had begun delivering in bulk, with special trucks that unload into farm bins, on the US West Coast shortly after the Second World War, and that the practice had spread east; many trucks blew the feed into the bin by air. On 60 delivery routes they studied in southeastern New Hampshire, bulk routes averaged 1.3 stops against 8.1 on bagged routes, and farmers buying bulk saved about $3 a ton in early 1954.
LessonBulk paid only where farms were large: the authors found it "could be neither universally recommended nor adopted", and small farms kept handling 100-pound bags.
Credit
George B. Rogers and Harry C. Woodworth (UNH Station Bulletins 427 and 431, 1956)
1977
Routes planned by computer
Suhas Sathe of Virginia Tech took a week of truck records from a bulk poultry feed mill in Salisbury, Maryland, that served farms within about 50 miles (80 km), and used Gillett and Miller's "sweep" algorithm to plan the routes on a computer. Against the dispatcher's routes, the computer plans showed a potential saving of at least 0.20 miles per ton delivered, or about 200 miles a day for a mill hauling 1,000 tons. The report notes that more than 60% of US poultry feed was by then delivered in bulk.
LessonPlanning needed records first: the mill had to set up forms to log each truck's trips before any route could be computed, and the study could not compare costs.
Credit
Suhas G. Sathe, advised by Marvin Agee (Virginia Tech report, dated December 1977)
2019
Virus found on trucks
After African swine fever reached Vietnam, a pig company swabbed its feed mill, ingredients and delivery trucks for the virus's DNA. In the first round, in August 2019, all 4 positives among 78 samples came from the cabs of feed trucks, which carried 25 kg bags hand-stacked on tarp-covered flatbeds, and drivers sometimes left the cab to help farm staff unload. The company began cleaning truck cabs and pressure-washing every returning truck before disinfection.
LessonThe weak point was the truck cab and the driver's movements at the farm, not the feed.
Credit
Jordan Gebhardt and colleagues, Kansas State University (Transboundary and Emerging Diseases, 2022)
2022
Feed trucks link farms
Using a year of GPS records from one US pig company's vehicles in 2020, Jason Galvis, Cesar Corzo and Gustavo Machado found that feed trucks linked farms more densely than any other network they measured, including the movement of live pigs. Their model attributed about 12% of PRRS (porcine reproductive and respiratory syndrome) infections on sow farms to feed vehicles, with a wide 95% range of 0.5% to 32%.
LessonFeed must move even when animals cannot; in a 2026 simulation of ASF by the same group, letting feed delivery continue during a movement standstill added a median of 11.3 infected farms, 6.4% more than a full standstill.
Credit
Galvis, Corzo and Machado, North Carolina State University and University of Minnesota (Transboundary and Emerging Diseases, 2022)
Watch for
Hydrogen for feed trucks
Dutch feed maker De Heus Voeders is buying a truck with a hydrogen combustion engine, which it presents as a CO2-free alternative to diesel transport, with a filling station to be built by Fieten Olie in Steenwijk. Delivery is planned for 2026, followed by extensive testing; a consortium including De Heus won a Dutch hydrogen mobility subsidy (SWIM) on 16 September 2025.
Achieved whenthe truck has made feed deliveries for a year and its fuel use, availability and emissions per tonne delivered, compared with diesel trucks on similar routes, are published and checked by someone outside De Heus.
Claimed by
De Heus Voeders, October 2025
Rerouting trucks around outbreaks
Jason Galvis and Gustavo Machado of North Carolina State University, with Cesar Corzo of the University of Minnesota, simulated a scoring system that reassigns the trucks carrying feed and pigs among 1,609 commercial pig farms, favouring trucks cleaned after their last trip and keeping them away from outbreak farms. Even if truck cleaning did not work at all, they report, rerouting cut contacts between infected and uninfected farms by up to 42%; it also added up to 81% more cleaning stops and up to 54% more distance per truck.
Achieved whena pig or feed company uses such rerouting during a real outbreak and publishes, with outside review, the contacts, distance and infected farms against its usual routing.
Claimed by
Galvis and Machado (Preventive Veterinary Medicine), August 2025
Open
01
Bin sensors against the scale
An independent test of feed bin level sensors. For farm bins fitted with sensors from at least two makers, compare each sensor's reading of feed delivered and used with the weight on the delivery truck's scale ticket, fill after fill, and publish the error by sensor type, feed form and bin.
The briefCredit · unclaimed02
Rerouting scored in infections
An open test that turns truck rerouting into infected farms avoided and kilometres added. In an open disease-spread model of pig farms, plan feed deliveries three ways (shortest distance; each truck kept inside one zone and visiting farms in a fixed order; a ranking like Galvis and Machado's that favours cleaned trucks and avoids outbreak farms), and count infected farms, distance driven and cleaning stops over many simulated outbreaks, including a movement standstill during which feed must still arrive.
The briefCredit · unclaimed
Open 01 · the brief
Bin sensors against the scale
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An independent test of feed bin level sensors. For farm bins fitted with sensors from at least two makers, compare each sensor's reading of feed delivered and used with the weight on the delivery truck's scale ticket, fill after fill, and publish the error by sensor type, feed form and bin.
Why it is still open
The figures in circulation are the makers' own. BinSentry says it monitors more than 25,000 feed bins and that in a year-long study with the pork producer HANOR across 100 barns nearly 80% of out-of-feed events came from mistakes with bin slides, not empty bins; its release calls manual bin checks "unreliable and dangerous", since they mean climbing the bin. The nearest outside test found, by Olds College in Alberta in 2024, checked a radar inventory sensor in grain bins for its maker, North Star Systems, and its project page gives no error figures. Searches of Europe PMC and Crossref in November 2026 found no independent test of sensor accuracy in feed bins.
Built with what exists
Scale tickets from the feed mill's delivery records; commercially sold bin sensors (camera, radar or load-cell types) and their data exports; Python or R for the error tables. Nobody climbs a bin: every reading comes from the sensor or the scale.
First step
Ask one feed mill or integrator to share three months of scale tickets and sensor readings for 10 bins, and compute a first error table.
Done looks like
At least 40 bins on at least 5 farms followed for 6 months; for each sensor type, the median and 90th-percentile error in tonnes per fill, for pellets and for mash, published with the anonymised data.
Open 02 · the brief
Rerouting scored in infections
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open test that turns truck rerouting into infected farms avoided and kilometres added. In an open disease-spread model of pig farms, plan feed deliveries three ways (shortest distance; each truck kept inside one zone and visiting farms in a fixed order; a ranking like Galvis and Machado's that favours cleaned trucks and avoids outbreak farms), and count infected farms, distance driven and cleaning stops over many simulated outbreaks, including a movement standstill during which feed must still arrive.
Why it is still open
Galvis and Machado's 2025 rerouting study counted contacts between farms, not infections, and named the cost of rerouting and the effect of truck cleaning as its next questions. A 2026 study from the same group (Sanchez and colleagues) counted the extra infected farms when feed delivery continues through a standstill, but compared delivery on and off, not ways of planning it. Searches of Crossref, arXiv, Europe PMC and the web in November 2026 found no published comparison of delivery plans by infections avoided.
Built with what exists
Google's OR-Tools routing solver; the MIT-licensed PigSpread-ASF model from Gustavo Machado's group; a synthetic farm network built from public farm locations; Python.
First step
Reproduce the feed-only standstill scenario in PigSpread-ASF on a small synthetic network, then replace its truck contacts with OR-Tools routes that keep each truck inside one zone.
Done looks like
On a public test network of at least 500 farms, every farm is fed within its bin capacity under each plan; across 1,000 simulated outbreaks, the infected farms, kilometres and cleaning stops for each plan are published with their ranges, along with the kilometres added for each infected farm avoided; code and test networks public.