ASTM E139

What the test does

A machined metal bar is installed in a load train running vertically through a split furnace, brought to temperature and held until the whole gauge length is at heat. A constant force is then applied smoothly, without shock — usually through a lever arm multiplying a hanging dead weight — and left in place. Either the extension is tracked continuously, or the bar is left until it breaks and the elapsed time recorded. Runs of tens of hours to 10 000 h are routine; frames are built for durations approaching 100 000 h.

What it measures, and why it matters

A creep test yields extension against time at one stress and one temperature: the decaying primary stage, the steady secondary stage that gives the minimum creep rate, and the accelerating tertiary stage ending in fracture. Times to specified strains come off the same curve. A stress-rupture test drops strain measurement and reports rupture life alone, with elongation and reduction of area from the broken halves.

These numbers set allowable stresses for anything that runs hot under load — turbine blading, boiler and superheater tubing, pressure-vessel internals, hot fasteners. Minimum creep rate governs how far a component distorts before replacement, which for a rotating part is a clearance question, not a strength one. Rupture life across stresses and temperatures is what time-temperature parameter methods extrapolate from, so a design life is only as sound as these runs.

Specimen

Specimen geometry is drawn from Test Methods E8/E8M rather than defined here. Round bars are usual, machined to a smooth reduced section and carried by threaded, shouldered or pinned ends; flat specimens serve where product form demands. Round-bar gauge lengths are commonly a multiple of diameter, four or five being familiar choices, but no single value is published for this method, so the test plan should state what was used. One specimen gives one point, so a design curve needs a matrix of bars at several stresses per temperature — hence banks of frames. Notched-bar rupture is not covered here; that is Practice E292.

What the machine must be capable of

The method prescribes an accuracy class, not a capacity: force is governed by Practices E4. Sizing follows the specimen. A 6 mm bar of about 28 mm² at creep stresses between 50 and 400 MPa needs roughly 1.4 to 11 kN, a 10 mm bar at 300 MPa about 24 kN. Lever creep frames for this work are commonly rated 20 to 50 kN, ratios of 20:1 or 50:1 keeping the dead weights manageable.

No crosshead speed or strain rate is programmed. The frame must hold a force absolutely steady for the whole run and keep the load train straight: bending superimposed on the axial stress shortens rupture life without warning, and alignment is verified separately under Practice E1012.

Strain measurement is performance-based rather than tied to a named class. Practice E83 is referenced for extensometer classification and the class used must be reported, but the requirement is that the system resolve creep strain finely enough for the intended use of the data, and that it sense the specimen rather than the parts attached to it. Extension pieces bring the sensing element out of the hot zone.

Temperature is the other half of the machine: a multi-zone furnace holding the set point for the entire run, thermocouples on the reduced section, permitted deviation banded by temperature in the standard's own text. No bounds are set — furnaces sold against the method typically cover roughly 200 °C to 1 200 °C, which is equipment capability, not a method limit. Humidity is not controlled.

What goes wrong in practice

Off-axis loading is the classic invisible error. A load train fractionally out of line puts bending on one side of the gauge section; the test looks normal, the rupture time comes out short, and nothing in the record explains it.

Thermocouple drift is the long-run equivalent. Over thousands of hours at temperature a thermocouple decalibrates, and a set point that has quietly slipped a few degrees moves rupture life far more than the same error would move a tensile result.

Oxidation of the gauge section in air masquerades as creep: a thin bar loses section to scale, true stress climbs during the run, and the specimen fails early — worst on small diameters and long durations. A fracture at or near a shoulder, from pull-out or thread relaxation, is no result at all whatever the elapsed time says.

Related and equivalent standards

ISO 204 is the nearest counterpart, covering uniaxial creep in tension for metallic materials. The practical difference lies in how requirements are stated: ISO 204 names verification classes outright — force to class 1 of ISO 7500-2, extensometry to class 1 of ISO 9513 — where this method routes force verification through Practices E4 and leaves strain-measurement adequacy to be shown against intended use.

Several ASTM neighbours are regularly confused with it. Test Methods E21 covers short-duration elevated-temperature tension — strength on the day, not deformation over years — and Practice E292 covers notched-specimen rupture, explicitly outside this scope. Creep crack growth is E1457.