• 未标题-1

Pellets for Pigs in a Highland Maize Economy: Grinding, Cooling and Aflatoxin Discipline at a Rwandan Feed Mill

Executive Summary

A feed mill in Rwanda’s Southern Province runs a single Hongyang SZLH420 ring die pellet mill: a 420 mm die, 120 mm working width, 287 rpm die speed, a 110 kW main motor and a published band of 3 to 14 t/h, fed by an SFSP112×40 hammer mill. Its ration is highland maize, wheat bran and imported soybean meal. Its constraint was water, not compression.

The company sold mash for eleven years and failed at pelleting on the first attempt. Ground through a 4 mm screen, the meal was too coarse and too widely spread to bind; pellets left the press at 60 to 70 °C into a cooler sized for a smaller machine; bags reached retailers at 13.8 to 14.2 percent moisture and moulded within a month. Hongyang re-specified grinding to a 600 to 800 micron band, rebuilt the die around a bran-heavy mash, and matched cooling to the press.

Over twelve months durability held 92 to 94 percent, fines at the bagging scale fell below 3 percent, bagged moisture settled at 12.2 to 12.8 percent, and none of the mill’s sixty pig-farm customers reported mould across two rainy seasons, with grower throughput at 6.5 to 7.5 t/h.

Market Context

Rwanda’s pig herd roughly doubled in a decade, from 684,708 head at the 2010 livestock census to about 1.5 million by 2022. The 2022 census found 14.9 percent of households keeping pigs, rising to 27.0 percent in the Southern Province where this mill sits. Carcass output has not followed: FAOSTAT records pigmeat at 9,000 tonnes in 2023 against 15,000 tonnes at the 2016 and 2019 peaks.

Industrial capacity is thin and underused: MINAGRI counts seven registered feed companies with about 163,000 tonnes a year installed, running near 53 percent of it. Maize, close to half of most recipes, is short, with production of 450,000 to 508,000 tonnes against demand above 850,000, so imports set the local price. Soybean meal comes largely from India at around Rwf 1,000 per kilogram. Pig feed quoted near Rwf 600 per kilogram in recent trade reporting, against about Rwf 270 in 2022, and feed takes 60 to 70 percent of a producer’s cost.

Underneath lies a safety problem pelleting cannot solve. Nishimwe and colleagues, sampling maize flour in seven Kigali markets over two rounds, measured mean aflatoxin B1 of 8.0 to 24.7 and 10.4 to 25.7 micrograms per kilogram, and found most animal feed samples above 100. A 2024 agro-ecological assessment put mean total aflatoxin in stored maize at 40.45 micrograms per kilogram in the humid volcanic highlands against 4.66 in the Eastern Savannah, because grain is hard to dry where the air is cool and saturated. Compounded pig feed answers to RS EAS 55; RS 98 sets the code of practice for production and storage.

Customer Profile

The customer is a family-owned mill in Muhanga district: twenty-two employees, one grinding and pelleting line, roughly sixty pig farms within a two-hour drive, mostly farrow-to-finish units of 20 to 120 sows plus a few operations finishing 300 head at a time for the Butare and Kigali pork trade. Volume runs 40 to 50 tonnes a day on one extended shift.

Three products matter: a 3.0 mm creep feed for weaners, a 4.0 mm grower pellet, and a 4.0 mm finisher pellet that takes oil in a post-pellet coating drum rather than in the mixer. Formulations use what the market offers: maize at 55 to 60 percent, wheat bran at 10 to 15, soybean meal at 15 to 18, with limestone, phosphate, lysine, salt and a toxin binder.

The company sold mash for eleven years, until two of its largest customers asked for pellets: mash sorted by the animal, dusted into the air and caked in the trough when humidity climbed. A 2024 trial on an unbranded die bought on price produced the opposite of a reference case. Throughput reached 5.5 to 6 t/h only with the feeder held back, the starter tripped on current peaks, durability sat at 82 to 85 percent on the plant’s tumbling can, fines ran 12 to 15 percent of finished output, and each die lasted 400 to 500 hours. The complaint that reached Hongyang was the bags, leaving at 13.8 to 14.2 percent moisture.

Solution

Hongyang’s engineers asked for three months of production records, a sample of each ingredient and the local weather record before quoting. The diagnosis separated two problems the plant had treated as one.

Grinding came first. A 4 mm screen produced a wide particle distribution, with too little fine material to bind and enough coarse grit to resist compression, so the press compensated with compression ratio and paid in motor load, discharge temperature and die wear. The specification moved to a 2.0 mm screen, median particle size 650 to 750 microns, but stopped short of the finest setting: Wondra and colleagues showed that finer grinding raises gastric keratinisation in finishing pigs, so the mill checks a sieve stack monthly and holds the sub-400 micron fraction in bounds. Hammer mill throughput fell from roughly 9 t/h to 5.5 or 6, absorbed by a night grinding shift on off-peak tariff rather than a looser screen.

The die was rebuilt around the mash. Grower and finisher pellets both run 4.0 mm holes at compression ratios near 1:9 and 1:8.5; the finisher die is shorter because post-pellet oil keeps free fat low and a longer channel would add only back-pressure. The counterbore was deepened and its lead-in widened so a bran-heavy mash finds the hole instead of smearing across the die face, and die bodies were ordered in hardened stainless at HRC 54 to 56 for corrosion resistance against a moist, mildly acidic mash.

Conditioning holds mash 45 to 60 seconds at 0.25 to 0.35 MPa and 80 to 85 °C, discharge moisture 15.5 to 16.5 percent. Temperature was not pushed higher: at 10 to 15 percent bran, extra heat buys little gelatinisation and costs moisture the cooler must remove.

Cooling, the actual failure, was re-engineered. A counterflow cooler sized for 8 t/h with six to eight minutes of retention replaced the undersized unit, and the operating rule changed with it: in the long rains ambient air at high humidity re-wets a pellet rather than drying it, so product is cooled to within 4 to 5 °C of ambient and never below, and bagging waits until a probe confirms 12.5 percent moisture or less. Bag temperature is checked at the spout, because a warm bag condenses on the inside of the polypropylene.

Aflatoxin discipline sits above all of it. Pelleting does not destroy aflatoxin, and the plant does not pretend a temperature setpoint solves a grain problem: every maize delivery is probed and screened, loads above an internal ceiling of 20 micrograms per kilogram are rejected, and a binder is carried as insurance rather than licence.

Commissioning and Training

Two Hongyang engineers spent ten days on site in February, the short dry season. Three findings mattered.

Steam was wet. The boiler line was short, uninsulated over its last run and without a proper separator, so the conditioner reached temperature by adding water the die then had to press out. Re-piping, insulation, a filter, reducing valve and trap turned the same boiler output into dry saturated steam and lifted discharge temperature without lifting mash moisture.

The cooler was fighting itself. Its plenum was half-blocked with fines, its fan blades worn, and the air balance never set against press output. Cleaning, new blades and a damper setting matched to 7 t/h changed the drying curve more than any other intervention.

Nothing was recorded: with no die log, sieve or moisture record, the plant could not see a problem until a customer reported it. Eight operators and two supervisors were trained to read motor current as the primary control signal, targeting 85 to 90 percent of rated load, to check die temperature and pellet appearance before touching the steam valve, and to log tonnes per die, screen size and moisture every shift. The first die was bedded in for 45 minutes on a bran-heavy mix at reduced feed rate with roller gap at 0.1 to 0.3 mm, and the sequence was posted beside the press.

Results

Throughput on the 4.0 mm grower pellet held 6.5 to 7.5 t/h, inside the SZLH420′s published 3 to 14 t/h band and above the 5.5 to 6 t/h the plant had coaxed from its first die. Durability settled at 92 to 94 percent, fines at the bagging scale fell below 3 percent from 12 to 15, and bagged moisture held 12.2 to 12.8 percent through two rainy seasons without a mould claim.

Dies now run 700 to 900 hours, roughly 4,500 to 6,000 tonnes, against 400 to 500 hours on the original specification, and die cost per tonne of finished feed has fallen by more than a third. Specific energy across the pelleting circuit runs 24 to 29 kWh per tonne; fine grinding adds 9 to 12 kWh per tonne at the hammer mill, accepted because it buys durability and less trough waste.

Producers describe cleaner troughs and less feed scratched to the floor, and the mill’s two largest customers now order the 4.0 mm pellet by specification rather than by name. Feed conversion across the customer base improved by a low single-digit percentage, consistent with published work on fines and pig performance rather than attributable to the die alone. The plant has added a night grinding shift and is discussing a second press, which is a more honest measure of satisfaction than any questionnaire.

Why It Matters

Rwanda’s pig sector is growing faster than its feed industry can supply it, and the arithmetic is unforgiving: a herd near 1.5 million head, seven industrial mills just over half utilised, a maize crop several hundred thousand tonnes short of demand, and a feed bill taking two-thirds of a producer’s budget. In that setting a mill’s position is decided by how efficiently it converts scarce, variable, sometimes contaminated grain into a pellet a farmer will pay for.

The transferable lesson is that pellet quality in a humid highland is a water problem before it is a pressure problem. Compression ratio, hole geometry and metallurgy all matter, but no die can rescue a pellet leaving the cooler at 14 percent moisture. Grinding to a controlled particle band, cooling to a defined moisture rather than to a temperature, and verifying both with instruments rather than with a hand in the bag turned a failing line into a production asset.

For Hongyang the useful part was not the sale, but that a customer with a mould complaint and no vocabulary for it received an answer in microns, ratios, hardness and relative humidity, and could verify the result on its own instruments within a month.

References

1. FAO, FAOSTAT: Rwanda pigmeat production series, 2010-2023.

2. MINAGRI, Rwanda livestock census 2010 and herd estimates to 2022; NISR, Fifth Population and Housing Census 2022.

3. MINAGRI, registered feed mill capacity and utilisation, 2024-2026.

4. RDB and MINAGRI, maize production, demand and import reports; soybean meal import pricing, 2026.

5. Nishimwe, K. et al., Aflatoxin B1 in maize flour and animal feed, Kigali markets, Food Control, 2016-2017.

6. Agro-ecological assessment of aflatoxin in stored Rwandan maize, volcanic highland versus Eastern Savannah, 2024.

7. Rwanda Standards Board, RS EAS 55 Compounded pig feeds – Specification; RS 98 Animal feed code of practice.

8. Wondra, K.J. et al., Effects of particle size and pelleting on growth performance, nutrient digestibility and stomach morphology in finishing pigs, Journal of Animal Science, 1995.

9. Thomas, M. and van der Poel, A.F.B., Physical quality of pelleted animal feed, Animal Feed Science and Technology 61: 89-112, 1996.

10. Hongyang Feed Machinery, SZLH420 pellet mill and SFSP112×40 hammer mill specifications, ringdies.com.


Post time: Nov-16-2026
  • Previous:
  • Next: