BRENNEN

Industries / Industrial equipment and services / Industrial machinery / Boilers and pressure vessels

Testing before shipping

This work proves a finished vessel before it leaves the shop: testing the toughness of its steel, filling it with water above its working pressure and having an independent inspector witness the result and stamp the vessel.

5

Milestones achieved, from 1921 to 2012.

0

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1921

    One code, one inspector, one stamp

    Carl O. Myers, Ohio's chief boiler inspector, and six other chief inspectors founded the National Board of Boiler and Pressure Vessel Inspectors on 2 December 1919 (ASME's chronology says February 1919). In 1921 Myers set the goals of "one code, one inspector, and one stamp"; that year manufacturers began registering boilers and filing data reports with the Board, and the first inspector commission was issued.

    LessonA code is only as good as its enforcement, and enforcement had to look the same in every state for a boiler built in one to be accepted in another.

    nationalboard.org

    Credit

    Carl O. Myers and the National Board

  2. 1965

    A vessel breaks on test

    In December 1965 an ammonia converter built by John Thompson (Wolverhampton) Ltd for ICI at Immingham, England, 18.2 m long with 150 mm steel walls, broke apart during its hydraulic test at 34 N/mm², below its required test pressure of 48 N/mm², with the water below 10 °C. One piece of about 2 t flew about 46 m and went through a workshop wall. TWI's account traces the failure to hydrogen cracking at a weld, a heat treatment that ran below its specified temperature, and weld metal far less tough than required.

    LessonThick high-strength steel must be tested above the temperature at which it turns brittle, and toughness must be set by fracture mechanics, the study of when a crack of a given size will grow.

    twi-global.com

    Credit

    British Welding Research Association, which investigated

  3. 1985

    Listening during the test

    Acoustic emission testing places sensors on a vessel to pick up the bursts of sound that growing cracks give off while the pressure rises. An article by Dennis A. White of the National Board, as reposted online, says ASME issued Section V Article 11, for fibre-reinforced plastic vessels, in June 1985, with Article 12 for metal vessels to follow. MISTRAS now sells acoustic emission testing of new vessels during their first hydrotest under Article 12.

    materials-welding.blogspot.commistrasgroup.com

    Credit

    ASME Section V committee

  4. 1987

    Toughness tied to thickness

    A Charpy test breaks a small notched bar with a swinging hammer to measure how much energy the steel absorbs, a measure of toughness. Sumio Yukawa's 1990 report for NIST and OSHA says the 1987 Addenda to Section VIII Division 1 added exemption curves, which set by thickness and steel group when Charpy tests can be skipped, and required impact tests for steels not covered. The same report notes that brittle fracture can still occur in service after a successful hydrotest when the service temperature is lower than the test temperature.

    nvlpubs.nist.gov

    Credit

    ASME; rationale credited by Yukawa to a paper by Selz

  5. 2012

    Test with water, not gas

    Compressed gas stores far more energy than water at the same pressure, so a vessel that fails under a gas test throws fragments. The UK Health and Safety Executive's guidance GS4, in its fourth edition, tells testers to "pressurise with water (or other fluid) rather than a gas wherever possible", to test remote from people or inside an enclosure, and gives at least 3 mm mild steel plate as a guide for enclosures.

    hse.gov.uk

    Credit

    Health and Safety Executive

Open

  1. 01

    Toughness fractures graded by camera

    An open set of photographs of broken Charpy specimens, each labeled with its percent shear (the share of the broken face that tore in a ductile way) by experts, and a model that measures percent shear from a photograph.

    The brief
    Credit · unclaimed
  2. 02

    Vessels that failed on test

    An open, sourced list of vessels that broke or leaked during their pressure test, with the test fluid, test temperature, steel, its toughness, the pressure at failure and the cause, each row linked to a primary source.

    The brief
    Credit · unclaimed

Open 01 · the brief

Toughness fractures graded by camera

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 photographs of broken Charpy specimens, each labeled with its percent shear (the share of the broken face that tore in a ductile way) by experts, and a model that measures percent shear from a photograph.

Why it is still open

Lucon and colleagues at NIST and ORNL compared methods on 129 instrumented Charpy tests and found that two operators measuring the same faces optically differed by -7% to +10%; ASTM E23 still allows methods from hand measurement to comparison charts. A search in November 2026 found a 2025 machine learning paper on the problem but no labeled image set that anyone can download.

osti.gov

Built with what exists

Broken specimens from a university or test laboratory, used with permission; a camera on a copy stand with fixed lighting; CVAT for outlining; a segmentation model such as a U-Net trained on one graphics card.

  1. First step

    Ask a university materials laboratory for 50 broken specimens and photograph them under one fixed setup.

  2. Done looks like

    At least 200 labeled photographs, each rated by two experts, with the model's error against the experts reported and within ±5 percentage points on held-out specimens.

Open 02 · the brief

Vessels that failed on test

Credit · unclaimed

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

What to build

An open, sourced list of vessels that broke or leaked during their pressure test, with the test fluid, test temperature, steel, its toughness, the pressure at failure and the cause, each row linked to a primary source.

Why it is still open

O'Donnell, Phillips and Winter searched ten incident databases covering more than 50,000 events for brittle failures of storage vessels and wrote that no incident database is complete. Their work and NIST's 1990 report cite cases, but no public table brings them together.

icheme.orgnvlpubs.nist.gov

Built with what exists

TWI case studies, NIST and HSE reports, the IChemE paper's references; an AI model to find and tabulate cases, every value checked by hand against the source; a CSV file on Zenodo.

  1. First step

    Extract the cases cited in NIST SP 780 and TWI's case study series into the first rows.

  2. Done looks like

    At least 50 incidents, each with a primary source link and at least six of the fields above filled in, published under an open license.