BRENNEN

Industries / Mining and materials / Steel and metals / Forging

Heat treatment

Heat treatment heats and cools a forging again under control after it has been shaped, for example quenching it (cooling it fast in water, oil or a polymer solution) to harden it and then tempering it (reheating it gently) to make it tougher, so that it reaches the strength its drawing calls for.

6

Milestones achieved, from 1722 to 2012.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1722

    Quenching tested on samples

    In 1722 the French Academy of Sciences published René-Antoine Ferchault de Réaumur's twelve memoirs on converting wrought iron into steel, one of whose questions was the best way to quench. He judged steels by grain, by hardness after quenching using a scratch test, and by how far a quenched wire bent before breaking, and he heated his samples together in a small bath of molten lead so that all were quenched from the same temperature. J. P. Schotsmans, writing in the Journal of the Historical Metallurgy Society in 1991, calls the study "probably the first scientific approach to solve a metallurgical problem".

    hmsjournal.org

    Credit

    René-Antoine Ferchault de Réaumur

  2. 1930

    Transformation given a clock

    E. S. Davenport and E. C. Bain quenched small steel samples into baths of molten metal held at fixed temperatures and measured how austenite, the form steel takes when red-hot, changed over time. Their 1930 paper added "a time factor" to the iron-carbon phase diagram, creating what is now called the time-temperature-transformation (TTT) diagram, which heat treaters use to plan cooling, and it recorded the structure named bainite in 1934. H. K. D. H. Bhadeshia reviewed the paper's legacy in 2010.

    link.springer.com

    Credit

    E. S. Davenport and E. C. Bain

  3. 1937

    Crankshafts hardened in seconds

    Francis S. Denneen and William C. Dunn of the Ohio Crankshaft Company filed a patent in September 1933, granted in 1939, for hardening a crankshaft pin with induced current, "the entire process being carried out so quickly that the heat is not given time enough to penetrate". By 1937 the company's TOCCO division sold the process: inductor blocks surround a bearing, a high-frequency current heats its surface, and water jets through the blocks quench it to about 600 Brinell, so "hardening, which might in the ordinary course take hours, can be reduced to seconds", in the account of the trade weekly Commercial Motor. Only the areas that must be hard are treated.

    patents.google.comarchive.commercialmotor.com

    Credit

    Francis S. Denneen and William C. Dunn, The Ohio Crankshaft Company

  4. 1970s

    Cooled straight from the forge

    Microalloyed steels carry small additions such as vanadium, so that a forging can reach its strength by controlled cooling in air straight from the forging heat, without a separate quench and temper. P. Skubisz, H. Adrian and J. Sińczak of AGH University in Kraków date the first tests of air-cooling from the hot-working temperature to the 1970s; 49MnVS3, a medium-carbon vanadium steel, became the most common of this first generation and was used for crankshafts. Its hardness was relatively low, so later grades such as 38MnSiV5 added silicon and changed the carbon and manganese. The authors cite an estimate of about 20% savings in making automotive forgings this way, and note that complex shapes are harder to cool evenly.

    LessonThe first grades traded hardness for the saved heat treatment, and the chemistry had to be reworked before the route spread.

    doi.org

    Credit

    European steelmakers and forgers, not named in the source

  5. 1990

    One audit for heat treaters

    Aerospace manufacturers each ran their own audits of suppliers, so a heat treater could face three or four audits of the same process for different customers. In 1990 leading aerospace manufacturers founded Nadcap, then the National Aerospace and Defense Contractors Accreditation Program, to run one standard audit accepted across the industry; heat treating is one of its 24 accreditation programs. Its accreditation decisions are made by industry peers, not by independent agencies.

    p-r-i.org

    Credit

    the founding aerospace manufacturers and the Performance Review Institute, which runs Nadcap

  6. 2012

    Distortion starts before quenching

    Distortion is the change of shape a part undergoes in heat treatment, and with cracking it is among the defects that most often get quenched parts rejected, by the account of Medina and colleagues (2024). In 2012 Th. Lübben and H.-W. Zoch summarized the findings of a German collaborative research center on "Distortion Engineering": the distortion seen after hardening grows from a "distortion potential" carried in from earlier steps, including segregation (uneven chemistry) in the steel and the forging itself, and the center studied it in forged disks of 20MnCr5 steel and bearing rings of 100Cr6.

    LessonDistortion cannot be fixed at the quench alone, because the steel mill, the forge and the machine shop each add to it.

    doi.orgzenodo.org

    Credit

    Th. Lübben and H.-W. Zoch and the Distortion Engineering research center

Watch for

  1. Forgings without quench and temper

    Steve Ooi of the steelmaker Ovako wrote in the May 2026 issue of Materials World that Ovako's direct-quenching steels can give forgings strength, ductility and toughness without a second heat treatment: one concept uses interrupted water cooling for parts up to 80 mm thick, another a salt bath held at a set temperature. Ovako estimates the saving at 126 to 140 kg of CO2-equivalent per tonne of forged component, from about 700 kWh per tonne for a gas-fired quench and temper, and itself notes that direct quenching "has often faced challenges related to meeting toughness requirements". It describes no trials at forges.

    Achieved whena forger reports series production of a part made this way, with strength and Charpy toughness (impact resistance) measured by a laboratory outside Ovako against the quenched and tempered part it replaces.

    ovako.comovako.com

    Claimed by

    Ovako, May 2026

Open

  1. 01

    Quench distortion, scored in public

    A public scorecard of how well an open-source solver predicts the distortion of quenched steel test pieces, run against every published case that gives its full inputs and measured result, starting with C-rings (thick rings with a slot cut in them, whose gap opens or closes when they are quenched).

    The brief
    Credit · unclaimed
  2. 02

    A public hardenability test set

    An open set of Jominy results, the hardness measured along a steel bar quenched from one end under ASTM A255, each with the steel's chemistry, gathered from published papers, theses and forges willing to share, and a blind test of hardenability calculators on it: the calculation method in ASTM A255, published neural networks and new models.

    The brief
    Credit · unclaimed

Open 01 · the brief

Quench distortion, scored in public

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public scorecard of how well an open-source solver predicts the distortion of quenched steel test pieces, run against every published case that gives its full inputs and measured result, starting with C-rings (thick rings with a slot cut in them, whose gap opens or closes when they are quenched).

Why it is still open

Humberto Celleri, Iván Vieitez Portela and Diego Passarella (2014) adapted EDF's open-source Code_Aster to model the quenching of a steel cylinder and wrote that "further work is needed, especially to compare with experimental data". Medina and colleagues (2024) measured the gap of quenched AISI 4340 C-rings, found their simulated gap of the same order of magnitude as the measured one, took their transformation data from the commercial database JMatPro, and noted that little published work validates distortion after quenching for parts of complex shape. Shared tests exist but are closed: in a Japanese benchmark project reported in 2007, Tatsuo Inoue and colleagues ran several programs on a carburized and quenched cylinder, ring and helical gear with common material data and compared them with measurements. Searches of GitHub, Zenodo and Hugging Face in November 2026 found no open quench model scored against measured distortion, and no set of cases with public input files.

amcaonline.org.arzenodo.orgdoi.org

Built with what exists

Code_Aster, which has a built-in model of phase changes in steel during cooling; published C-ring and ring cases with their cooling data; transformation data from open papers; Python to run the cases and score them. No furnace or quench tank is needed.

  1. First step

    Rebuild the Medina et al. (2024) C-ring case in Code_Aster and compare its gap change with their measured value.

  2. Done looks like

    At least ten published cases from at least four laboratories, each rerun with stated inputs, with predicted against measured change in gap or roundness published for every case, a median error, and all input files public.

Open 02 · the brief

A public hardenability test set

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An open set of Jominy results, the hardness measured along a steel bar quenched from one end under ASTM A255, each with the steel's chemistry, gathered from published papers, theses and forges willing to share, and a blind test of hardenability calculators on it: the calculation method in ASTM A255, published neural networks and new models.

Why it is still open

Models that predict Jominy curves from chemistry go back at least to 1995, when P. J. van der Wolk at Delft University of Technology trained a neural network on over 4,000 sets of data from the steelmaker Nedstaal. Marco Vannucci and Valentina Colla (2023) reached a mean error of 0.96 HRC on 1,500 results from one plant, but over 4 HRC when that model was applied without retraining to 250 results from a second plant. Both used plant data that was not released, and searches of GitHub, Zenodo and Hugging Face in November 2026 found small Jominy calculators and simulations but no public dataset of curves with chemistry.

repository.tudelft.nliris.sssup.itastm.org

Built with what exists

Jominy tables and curves from open papers and theses, digitized with WebPlotDigitizer; an AI document reader to pull chemistry and hardness tables from papers and scanned test reports, every value checked by hand (see the Note "Seven more flagged"); Python and scikit-learn; the ASTM A255 calculation as the baseline.

  1. First step

    Collect Jominy curves with full chemistry from open-access papers and theses, and ask two forges whether they would release redacted test reports.

  2. Done looks like

    At least 1,000 curves from at least five sources and three steelmakers released under an open license, with the ASTM A255 calculation and at least two learned models scored on a source held out of training, and the mean error in HRC at each distance from the quenched end published.