Industries / Mining and materials / Glass and ceramics
Forming
Forming is the work of shaping hot glass or soft clay into bottles, sheets, tiles, slabs and other parts before they set.
7
Milestones achieved, from 1st century BCE to c. 2014.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1st century BCE
Glassblowing
Glassmakers learned to inflate hot glass on a hollow pipe, which made glass vessels cheap and possible in great variety. The Encyclopaedia Judaica places the invention toward the end of the 1st century BCE, in the reign of Augustus, somewhere on the Phoenician coast, perhaps at Sidon. Blowing into metal molds, which shapes and patterns a vessel in one step, followed, and mold-blown jugs were later made in Jerusalem.
Credit
unnamed glassmakers of the eastern Mediterranean coast
1840
Tiles pressed from dust
Richard Prosser, a Birmingham engineer, patented a way to press buttons from dry clay powder between metal dies, sealed on 17 June 1840. The potter Herbert Minton took a share in the patent and applied it to tiles, and Prosser's 1854 obituary credits the process with Minton's colored mosaic tiles. A 2013 history of Belgian tiles, quoted by Grace's Guide, says dust-pressed wall tiles were made at Mettlach in Germany from 1852.
Credit
Richard Prosser; Herbert Minton
1903
Bottles made by machine
Michael J. Owens's machine made bottles with no glassblower, and the American Society of Mechanical Engineers (ASME) calls it the first commercially successful, fully automatic bottle-making machine. Its AR model of 1912 averaged 50,400 bottles a day, and the last Owens machines ran in Gas City, Indiana, until December 1982.
LessonIn 1913 the National Child Labor Committee said the machine did more to end child labor in glass plants than laws had.
Credit
Michael J. Owens, with financing from Edward D. Libbey
1924
The individual section machine
Henry W. Ingle of the Hartford-Empire Company built a bottle machine from several identical sections, all fed from one feeder that cuts the molten glass into gobs, measured lumps of glass for one bottle each. Bucher Emhart Glass, which traces its history to Hartford-Empire, dates the first machine to 1924 and the first four in operation to 1925. A 1995 Glass International article instead dates Ingle's first ideas to 1922 and the first commercial four-section machine to 1927, at Maryland Glass in Baltimore, and says 16-section machines making three bottles at a time were then in use. Ingle's US patent, filed in 1928, was granted in 1933.
Credit
Henry W. Ingle, Hartford-Empire Company
1938
Glass spun into wool
In 1931 Games Slayter persuaded the Owens-Illinois Glass Company to back research into new uses for glass. Jack Thomas and Dale Kleist worked out a way of blowing molten glass into fibers with steam, which Slayter turned into mass production, and the patent Slayter and Thomas filed in 1933, granted on 11 October 1938, covers drawing a stream of molten glass into fine filaments with a fast jet of gas. The fibers went into wool insulation for homes and, by the 1940s, into Owens-Corning's reinforcements for plastic laminates used in aircraft and boats, according to the National Inventors Hall of Fame.
Credit
Games Slayter, Jack Thomas and Dale Kleist, Owens-Illinois Glass Company
1952
Glass floated on tin
In the float process, molten glass flows onto a bath of molten tin, spreads into a level ribbon and leaves the bath solid at about 600 °C, with a fire-polished finish that needs no grinding. Pilkington dates the invention by Alastair Pilkington to 1952; the consultancy glassglobal dates it to 1953 to 1957 and names Kenneth Bickerstaff as co-inventor. Pilkington licensed the process to more than 40 manufacturers in 30 countries and counts about 370 float lines operating, being built or planned, and glassglobal says float makes more than 85% of the world's flat glass.
Credit
Alastair Pilkington and Kenneth Bickerstaff, Pilkington
c. 2014
Slabs pressed on a belt
SACMI's Continua+ line presses ceramic powder in a belt-type compactor and decorates it digitally on the line, making tile up to 1,600 mm wide and slabs up to 3,200 mm long, by Ceramic Industry's account. Ceramic Industry reported in October 2015 that 13 plants had installed the line a year after its launch. SACMI said in June 2021 that it had sold more than 100 such lines on five continents, and that one model makes slabs up to 20 mm thick and 120 cm wide in lengths limited only by cutting.
Credit
SACMI
Watch for
Learned settings for gob forming
A neural network trained on historical data from several bottle plants predicts how changes to the feeder settings alter the gob, and a search over its inputs picks the settings expected to give a target gob. The authors report "promising" results on historical datasets from several production lines, not on a running line.
Achieved whena plant runs the method on a live line and someone outside the team publishes waste or defect rates before and after.
Claimed by
Mattia Pujatti and colleagues at Fondazione Bruno Kessler and GLASSFORM.AI, October 2025
Printed cores for turbine blades
Boom Supersonic bought three Lithoz CeraFab S65 printers to make ceramic casting cores, the parts that form the channels and cavities inside a cast single-crystal turbine blade. The printers use lithography-based ceramic manufacturing (LCM), a 3D printing method that hardens each layer with a light projector. Boom's chief engineer says a series of prototypes gave "repeatable, reliable results".
Achieved whenblades cast around printed cores run in a turbine and a party other than Boom or Lithoz reports their yield or test results.
Claimed by
Lithoz and Boom Supersonic, November 2026
Open
01
An open bottle-forming simulator
free, open-source software that simulates the two blowing steps of bottle making and predicts the wall thickness of the finished bottle from the gob's temperature, the shape of the first-stage blank (the parison) and the machine timing, published with its test cases.
The briefCredit · unclaimed02
Robot finishing of green sanitaryware
a low-cost robot arm that learns from a few human demonstrations to sponge and trim the seam lines on leather-hard slip-cast pieces, and then finishes a shape it was not trained on.
The briefCredit · unclaimed
Open 01 · the brief
An open bottle-forming simulator
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
free, open-source software that simulates the two blowing steps of bottle making and predicts the wall thickness of the finished bottle from the gob's temperature, the shape of the first-stage blank (the parison) and the machine timing, published with its test cases.
Why it is still open
Glassmakers rely on dedicated simulators that a 2025 Fondazione Bruno Kessler and GLASSFORM.AI paper calls computationally and economically expensive and complex to adapt. Eindhoven University of Technology researchers checked a blow-blow model against industrial thickness measurements in 2011, but searches of GitHub and Crossref found no public code for it or any other glass container forming model.
Built with what exists
FEniCSx or OpenFOAM, free solvers for flow and heat problems; Gmsh for meshes; ParaView for viewing results; published viscosity-temperature data for soda-lime glass.
First step
read the 2011 paper and its 2008 predecessor, then set up the first blowing step as a round, axisymmetric model with uniform glass temperature and match the 2008 test problems.
Done looks like
the code reproduces the wall-thickness profiles that Giannopapa and Groot published in 2011 within the error they report, runs one bottle on a laptop in under an hour, and is public with its inputs.
Open 02 · the brief
Robot finishing of green sanitaryware
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
a low-cost robot arm that learns from a few human demonstrations to sponge and trim the seam lines on leather-hard slip-cast pieces, and then finishes a shape it was not trained on.
Why it is still open
SACMI sells robot cells that fettle (trim) and sponge green sanitaryware, but publishes no figure for how long a new model takes to set up. Zhangguo Yu and Hsien-I Lin (2021) built a robot that fettles clay pots with a hand-tuned force controller, noting that robot polishing usually assumes a CAD model, which clay pots lack. Learned robot policies have shaped clay: RoPotter (2024) made clay bowls on a wheel after 40 demonstrations. No published work found applies learning from demonstration to finishing cast ware.
Built with what exists
the SO-101 arm and the LeRobot open-source robot learning library (see the Note "No download"); a policy model that learns from few demonstrations, such as NVIDIA's GR00T (see the Note "Five demonstrations per task"); an RGB-D camera; two-part plaster molds and casting slip from a pottery supplier. Work only on wet clay, since dry clay dust contains crystalline silica.
First step
cast ten small test pieces with a seam line in a two-part plaster mold and record five teleoperated sponging demonstrations on each.
Done looks like
20 filmed attempts on three cast shapes outside the training set, the seam height measured after each, and a published tally of pieces finished to a seam under 0.5 mm with no cracks or gouges.