Industries / Food, farming and forestry / Food and beverage processing / Sugar refining
Crystallizing
Boiling sugar syrup under vacuum until crystals grow, spinning the crystals free of the remaining syrup, then drying and conditioning them so they keep.
4
Milestones achieved, from c. 1800 to 1973.
0
To watch for: announced, not yet confirmed outside.
3
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
c. 1800
Boiling in open kettles
Around 1800, plantation sugar was made by boiling cane juice down in a row of four or five open copper kettles of decreasing size, called the "Jamaica train". Enslaved workers skimmed off impurities and moved the liquid from kettle to kettle with long-handled dippers, and the syrup was left in pots to drain into loaves of impure sugar, which were shipped to England for refining.
LessonThe Chemical Engineer records frequent scalds and burns, and gives a rate for 1800 of one enslaved person's life lost for every two tons of sugar made.
Credit
Caribbean plantation owners, using enslaved labor
1813
Boiling under a vacuum
Edward Charles Howard patented a closed pan, heated by steam and held under partial vacuum, in patent No. 3754, "Preparing and refining sugar". Lower pressure makes syrup boil at a lower temperature, so less sugar caramelizes or burns. Boulton and Watt built the equipment, European refiners and American plantations took it up, and Howard's royalties were reported to exceed £60,000 in 1816, according to The Chemical Engineer.
Credit
Edward Charles Howard
1851
Spinning the syrup off
A centrifugal machine "for separating molasses from sugar", made by Manlove and Alliott, was shown at the Great Exhibition in London in 1851. A sugar centrifuge spins the mix of crystals and syrup, called massecuite, in a basket with a screened wall: the syrup is thrown out through the screen and the crystals stay behind. BMA, a German maker, says its first centrifugals of 1860 were belt-driven batch machines emptied by hand, that it launched electrically driven machines around 1900, and that every machine has had a programmable controller since the mid-1980s.
Lessonan article by Robert F. Gibson in Scientific American Supplement says early centrifugals "frequently exploded", and users were made by law to put them below ground.
Credit
Manlove and Alliott (maker); Rotch and Finzel (exhibitor)
1973
Pans that never stop
A batch pan is filled, boiled and emptied one load, or strike, at a time. Fives-Cail Babcock installed a continuous vacuum pan, which takes in syrup and puts out massecuite without stopping, at the Quartier Français sugar factory on Réunion during the 1973 crushing season, as M. Rivière and P. Pithois reported in Sugar y Azúcar in 1976. BMA says it commissioned its first vertical continuous pan in Germany more than 30 years before 2017 and its first for refined-sugar grades in Dubai in 1995; refinery pans are cleaned about weekly, while beet and cane factories run 50 to 60 days between cleanings.
Credit
Fives-Cail Babcock; BMA
Open
01
A public crystal photo benchmark
A public set of microscope photos of sugar crystals, each labeled with crystal size measured by sieving (the mean aperture and the spread around it), with a scoring program and a baseline model that reads both from a photo.
The briefCredit · unclaimed02
Centrifuge camera checked by lab
A set of camera frames from a sugar centrifuge, each matched with the laboratory purity of the sugar it produced, and a model that predicts that purity from the frame, so the wash water could be set from the picture.
The briefCredit · unclaimed03
Will this sugar cake
A bench test and model that tell, before shipping, whether a batch of white sugar will set into lumps in bags or containers (caking), from its moisture uptake curve, grain size and bagging temperature.
The briefCredit · unclaimed
Open 01 · the brief
A public crystal photo benchmark
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public set of microscope photos of sugar crystals, each labeled with crystal size measured by sieving (the mean aperture and the spread around it), with a scoring program and a baseline model that reads both from a photo.
Why it is still open
Camera systems for crystal size are sold to factories, and Lajos Rozsa described image analysis in working pans in 2008, but the published tests cannot be compared. Masoudi, Asadi and Akbarizadeh (2024) scored their methods on a single image of 26 crystals; Ji and colleagues' 2025 food-crystal paper does not name the crystal or publish its data; ParticleTech's 2020 paper compares its own system with sieving. A GitHub project from October 2026 (brighton-li/sugar-crystals-size) sizes crystals in one pan photo with classical image processing, without sieve results to check against. No public labeled set of sugar crystal photos was found.
agrieng.scu.ac.irvaisala.comarxiv.orgapi.crossref.orggithub.com
Built with what exists
a laboratory microscope with a camera, a stack of test sieves, open segmentation models such as Meta's Segment Anything or a U-Net in PyTorch, and a small open vision model fine-tuned on one graphics chip (see the Note "One GPU to fine-tune").
First step
Buy five white sugars of different grain sizes, sieve each, photograph samples spread in vegetable oil under a microscope, and publish the first 50 labeled pairs.
Done looks like
At least 500 labeled photos covering at least five sugar grades, published under an open license, with a baseline whose mean-size estimate is within 5% of the sieve result on photos it was not trained on.
Open 02 · the brief
Centrifuge camera checked by lab
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A set of camera frames from a sugar centrifuge, each matched with the laboratory purity of the sugar it produced, and a model that predicts that purity from the frame, so the wash water could be set from the picture.
Why it is still open
Water is sprayed on the spinning crystals to wash off syrup. Too much dissolves sugar into the molasses; too little leaves syrup in the sugar. South Africa's Sugar Milling Research Institute built a strobe-lit camera over a C-centrifuge, the last, lowest-grade stage, and reported it to the South African Sugar Technologists' Association in 2024. A 2025 University of the Witwatersrand study group improved how the frames are read, and wrote that the image measure still has to be checked against lab-measured purity on a larger set of images.
Built with what exists
recorded frames from the institute's camera and the matching lab results, OpenCV for the k-means method in the Wits report, PyTorch, and a small open vision model fine-tuned on one graphics chip (see the Note "One GPU to fine-tune").
First step
Write to the Sugar Milling Research Institute for recorded frames with matching lab results, and rebuild the Wits k-means method on the sample frames printed in the report.
Done looks like
On at least 300 frames held out from training, the predicted purity tracks lab purity with a correlation of at least 0.8, and the frames, labels and code are published.
Open 03 · the brief
Will this sugar cake
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A bench test and model that tell, before shipping, whether a batch of white sugar will set into lumps in bags or containers (caking), from its moisture uptake curve, grain size and bagging temperature.
Why it is still open
Control Union, a cargo inspection firm, told the 2023 Geneva sugar conference that caking is driven by moisture, from the air or from moisture still bound inside poorly conditioned crystals, and that the usual oven test (ICUMSA loss on drying) measures only free moisture. Models exist: Scott Billings's 2002 Massey University thesis mapped storage limits from the sugar's water activity, and Leaper and colleagues (2002) checked a moisture-migration model in a scaled-down "caking box" standing in for a big bag. Miyazaki and Nagai of Mitsui Sugar (2020) found that moisture uptake predicted caking for 29 Japanese sugars stored in 20-gram paper cups, and wrote that "there are few reports that propose a simplified model or method for predicting the cause of caking of sugar". No quick test scored against many sugars stored in real bags was found.
jstage.jst.go.jpcontrolunion.commro.massey.ac.nzgala.gre.ac.uk
Built with what exists
a dynamic vapor sorption analyzer, or jars of saturated salt solutions for set humidities; a Karl Fischer titrator for total moisture; test sieves; a climate chamber cycling temperature like a shipping container; a texture analyzer to measure lump strength; open statistics software.
First step
Repeat the Mitsui paper-cup test on ten white sugars bought in shops and publish the breaking-energy results.
Done looks like
For at least 30 sugars from several producers, the test predicts caked or free-flowing after six weeks of cycled storage in 25 kg bags, correctly for at least 85% of them, with the data published.