Industries / Industrial equipment and services / Industrial machinery / Fasteners
Heat treatment
Heat treatment makes a formed bolt strong by heating it until its internal structure changes, quenching it in oil or water, and then tempering it, which means reheating it gently so that it is strong without being brittle.
5
Milestones achieved, from 1967 to 2017.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1967
Japan drops its strongest bolt
Japan's industrial standard for high-strength structural bolts, JIS B 1186, once included a grade called F13T, with a tensile strength of about 1,300 N/mm² (newtons per square millimetre, the same as megapascals). F13T bolts often failed in 1964 from delayed fracture, a crack that appears hours to years after a bolt is tightened and is caused by hydrogen in the steel, and the grade was struck from the standard in 1967. Nippon Steel's engineers say the next grade down, F11T, was "practically banned" in 1979, leaving F10T as the strongest bolt in use; a 2026 paper from Japan's National Institute for Materials Science (NIMS) gives 1981 as the year F11T production stopped.
LessonAbove about 1,200 N/mm², Nippon Steel says, delayed fracture rises sharply, so hydrogen, not the steel's strength, set the upper limit for structural bolts.
Credit
Japanese Industrial Standards, as recounted by Nobuyoshi Uno and colleagues of Nippon Steel (2008)
1999
Hydrogen-tolerant bolts approved in Japan
Nippon Steel and its bolt maker NS Bolten developed SHTB, a structural bolt with a tensile strength of 1,400 N/mm², made from a steel that can hold more hydrogen before it cracks and with a new thread shape meant to stop delayed fracture. By Nippon Steel's account, Japan's Minister of Construction gave SHTB a general approval in 1999 and it was first used in a high-rise building in 2001; the company says each bolt carries 1.5 times the tension of an F10T bolt, so a joint shrinks to about two-thirds of its size. A 2025 paper by NIMS and university authors dates SHTB's practical use to 1996.
Credit
Nippon Steel; NS Bolten
2012
Riser bolts break underwater
On 18 December 2012 the lower marine riser package of the drilling rig Transocean Discoverer India, the top section of the well-control stack on the sea floor, separated from the blowout preventer in the Gulf of Mexico, and about 432 barrels of synthetic drilling fluid were released after bolts in a GE Oil and Gas H4 connector broke. The US offshore regulator, the Bureau of Safety and Environmental Enforcement (BSEE), found the main cause was hydrogen embrittlement: a subcontracted plater had followed the 1998 edition of ASTM B633, the zinc electroplating standard, instead of the 2007 edition, and the older edition did not require a post-plating bake, hours of heating after plating to drive hydrogen out. More than 10,000 bolts were replaced.
LessonA bolt maker's quality system can miss what a second-tier supplier does; here an out-of-date edition of a standard quietly removed the bake.
Credit
BSEE Quality Control-Failure Incident Team (QC-FIT), report of August 2014
2013
Bay Bridge rods crack
In March 2013, 32 of the 96 high-strength anchor rods on two shear keys of Pier E2 of the new East Span of the San Francisco–Oakland Bay Bridge broke after being tightened to their specified tension. They were 3-inch (76 mm) hot-dip galvanized rods of grade ASTM A354 BD. The oversight committee's investigation found hydrogen embrittlement, with the root cause a steel more susceptible than normal: its structure was uneven from centre to edge and very hard near the surface, and the rods made in 2008 had been heat-treated a second time because records of the first treatment could not be produced. The specification had been copied from a project where the rods were only snug-tight, and it set no maximum hardness or strength and no minimum toughness.
LessonA specification that sets only a minimum strength can let through steel too hard for high tension in a wet place.
Credit
Toll Bridge Program Oversight Committee (Caltrans, the Bay Area Toll Authority and the California Transportation Commission), report of 8 July 2013
2017
Subsea bolts made softer
On 13 October 2015, during testing in the Gulf of Mexico, the rig Noble Paul Romano could not close the blind shear ram on its blowout preventer, the pair of blades that cuts the drill pipe and seals a well in an emergency, because six of the eight bolts holding its actuator had broken. BSEE's 2017 report found bolts harder than 35 HRC (Rockwell C, a hardness scale), zinc plated to the 1998 B633 edition with only a 3-hour bake at 190 °C. The maker, National Oilwell Varco, cut the bolts' required yield strength from 167 ksi to 130 ksi (thousand pounds per square inch), held hardness below 35 HRC and lengthened the bake to at least 8 hours, and BSEE agreed with replacing bolts of 37 to 42 HRC with ones of 31 to 34 HRC.
LessonFor a plated bolt in seawater, a weaker and softer bolt can be the safer one.
Credit
National Oilwell Varco; BSEE QC-FIT, report of July 2017
Watch for
Bolts at 1,700 MPa
NIMS researchers are developing quenched and tempered bolts in the 1,700 MPa class from a steel with 0.4% carbon, 2% silicon, 1% chromium and 1% molybdenum, with a shank thinner than the thread and a new thread form. A March 2025 paper said the new thread form improved resistance to delayed fracture in accelerated tests, but a December 2025 paper reports a 10-year outdoor exposure test in Okinawa in which bolts still broke by delayed fracture, at the end of the thread, at the first thread engaged in the nut and under the head.
Achieved whena 1,700 MPa bolt goes through years of outdoor exposure with no delayed fracture and wins a building approval or a place in a standard.
Claimed by
Yuuji Kimura and Tadanobu Inoue, NIMS, December 2025
Open
01
Open record of bolt failures
A public table of fastener failures drawn from published investigations, one row per incident, giving the bolt grade, measured hardness, coating and the edition of its standard, bake time, tightening, service environment, time to fracture and the cause found, with every value linked to the page it came from.
The briefCredit · unclaimed02
Decarburization measured by software
A public set of microscope photos of polished and etched cross-sections of bolt threads, each measured by an accredited lab for decarburization, the carbon lost from the surface during heat treatment that leaves a soft skin at the thread roots, and a model that reads the same depth from the photo.
The briefCredit · unclaimed
Open 01 · the brief
Open record of bolt failures
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of fastener failures drawn from published investigations, one row per incident, giving the bolt grade, measured hardness, coating and the edition of its standard, bake time, tightening, service environment, time to fracture and the cause found, with every value linked to the page it came from.
Why it is still open
A 2018 National Academies study for BSEE found that "little non-proprietary work has been reported" on subsea bolting failures, that companies may hesitate to share fastener failure information because of liability, and listed "collecting and disseminating information about fastener performance, failures, and near misses" among its options. The EU's hydrogen incident database, HIAD, records some bolt failures, such as flange bolts of 48 HRC that broke after about 4,000 hours in a Spanish plant in 2003, but has no fields for grade or coating.
nationalacademies.orgnationalacademies.orgknowledge.clean-hydrogen.europa.eu
Built with what exists
Published reports from BSEE's QC-FIT, the Toll Bridge Program Oversight Committee, HIAD and other investigators; a language model that drafts each field and cites the sentence behind it (see the Notes "Seven more flagged" and "Findings that cite notes"); Python and a spreadsheet; a second person checking a sample.
First step
Code the three public reports above (BSEE 2014, BSEE 2017 and the Bay Bridge report of 2013) and draft the field list from them.
Done looks like
At least 60 incidents from at least four countries coded on at least 10 fields, with a random 10% coded again, blind, by a second person and the agreement reported, and a chart of failures by hardness and coating published with the table.
Open 02 · the brief
Decarburization measured by software
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 polished and etched cross-sections of bolt threads, each measured by an accredited lab for decarburization, the carbon lost from the surface during heat treatment that leaves a soft skin at the thread roots, and a model that reads the same depth from the photo.
Why it is still open
Huang and colleagues (2023) wrote that decarburization depth is still judged by eye or with traditional image analysis, and reported a deep-learning detector with 92.97% detection accuracy on their own images. A search found no public set of fastener thread sections.
Built with what exists
Bolts bought on the open market; a metallurgical lab accredited to ISO/IEC 17025 for sectioning and measurement; a small open vision model fine-tuned on one GPU, such as Liquid AI's LFM3 (see the Note "One GPU to fine-tune"), or a standard segmentation network in PyTorch.
First step
Buy 20 lots of property class 10.9 M10 bolts from five sellers and get sectioning quotes from two accredited labs.
Done looks like
500 labelled thread sections from at least 50 production lots released under an open license, and a model whose depth readings fall within 10% of the lab's on 100 held-out images.