ASTM D790

What the test does

A rectangular bar of plastic is laid flatwise across two parallel cylindrical supports; nothing is gripped or clamped. A loading nose descends onto the midpoint of the span and bends the bar at a constant rate while force and midspan deflection are recorded. The run ends at rupture or at 5.0 % outer-fibre strain, whichever comes first, so a ductile grade may finish the test unbroken.

What it measures, and why it matters

The method reports flexural strength, flexural stress at any chosen point on the curve, maximum outer-fibre strain and flexural modulus, taken from the initial straight portion of the curve. Modulus is the stiffness figure that feeds wall-thickness calculations and finite-element modelling, and it separates unfilled grades from filled ones. Flexural strength serves lot release on moulding compounds and electrical insulating materials. Because the outer fibre carries the peak stress, the result is sensitive to moulded skin quality, which makes it useful in moulding investigations.

Specimen and span

Nothing is gripped in this test, so the specimen and the span between the supports are what define it. Get the span wrong and every number in the batch is wrong with it.

Reference moulded bar

127 × 12.7 × 3.2 mmThe recommended specimen for moulding materials, tested flatwise.

Span-to-depth ratio

16 : 1The default. 32:1 or 40:1 for reinforced materials that would otherwise fail in shear; up to 60:1 where modulus is the object.

Support span, 3.2 mm bar

51.2 mm16 × the measured depth — not the nominal depth off the drawing.

Overhang beyond each support

At least 10 % of the spanSo the bar cannot walk off a support as it deflects.

Specimen width, deep bars

Not more than ¼ of the spanApplies to specimens deeper than 3.2 mm, where a wide bar would stiffen anticlastically.

Specimens per sample

5 minimumA set in each principal direction for anisotropic materials, so ten in total.

Conditioning

23 ± 2 °C, 50 ± 10 % RH, 40 hPractice D618, for specimens 7 mm thick or less. Test in the same atmosphere.

Surface condition

Discard flawed barsPracticeA void, sink mark or tool mark on the tension face decides where the crack starts, so the flaw is measured instead of the material.

Measure every bar

Width and depth, at the midspanDakDepth enters the stress equation squared and the modulus equation cubed. A 2 % error in depth is a 6 % error in modulus.

Depth drives the span, the rate and both results. Measure the bar you are about to test rather than trusting the nominal thickness, and reset the span whenever the thickness changes.

Test speed

D790 does not name a crosshead speed. It names a strain rate, and the speed is calculated from the specimen you actually have — which is why two laboratories testing the same grade at different thicknesses correctly run at different speeds.

Procedure A — strain rate

0.01 mm/mm/minFor materials that break at comparatively small deflections. This is the procedure for modulus.

Procedure B — strain rate

0.10 mm/mm/minTen times faster, for materials that undergo large deflections before breaking.

Procedure A, 3.2 mm bar on 51.2 mm span

≈ 1.4 mm/minWorked from the formula below, not quoted from a table.

Procedure B, same bar

≈ 13.7 mm/min

End of test

Rupture, or 5.0 % outer-fibre strainWhichever comes first. A ductile grade often reaches 5 % intact, and then there is no flexural strength to report.

State the procedure in the report

AlwaysDakA flexural strength quoted without A or B beside it cannot be compared with anything.

The rate follows the specimen. Changing the thickness or the span ratio changes the correct speed, and a speed carried over from the previous job is one of the most common sources of a wrong modulus.

Calculations

Three equations carry the whole method. They are reproduced here with every term defined, so the page is usable without the standard open beside it — but the governing text is always the current edition.

How the test runs

The sequence below is the method as practised. It follows the standard's requirements but the ordering and the checks are ours.

1. 01Condition the specimens to Practice D618 and keep them in that atmosphere for the test.
2. 02Measure width and depth of each bar at the midspan and record them individually — these values, not the nominal ones, go into every calculation.
3. 03Set the support span to 16 times the measured depth, or to the higher ratio the material calls for, and verify it against the graduations rather than by eye.
4. 04Check that the supports and the loading nose are parallel and that the nose is centred on the span.
5. 05Calculate the crosshead rate from R = ZL²/6d for the procedure you are running.
6. 06Place the bar flatwise across the supports, centred, with the moulded surface orientation recorded.
7. 07Fit the deflectometer if flexural modulus is required; crosshead position alone is accepted for the other reported values.
8. 08Zero force and deflection with the nose just in contact, taking up any slack without preloading the bar.
9. 09Run at the calculated rate, recording force against midspan deflection continuously.
10. 10Stop at rupture or at 5.0 % outer-fibre strain, whichever comes first.
11. 11Take the modulus slope from the initial straight portion of the curve, not from a fixed pair of points chosen for convenience.
12. 12Repeat across at least five specimens and report the mean with the standard deviation.

The two steps most often skipped are measuring each bar and resetting the span. Both are silent: the test runs, a curve appears, and the numbers are wrong in a way nothing on the screen reveals.

Watch the test

The fixture this method needs

Adjustable spanTJ-124

Three Point Bend Fixture

Two cylindrical supports on a graduated beam with a central loading nose, and a span that resets from 25 to 300 mm as the specimen depth changes.

Specifications

What the report has to contain

A flexural result travels badly without its conditions. These are the entries that make a number reproducible by somebody else.

• Full designation and edition of the standard, and whether Procedure A or Procedure B was used
• Complete material identification, including grade, source, and any filler or reinforcement
• Specimen preparation route — moulded to size, or machined from sheet or plate
• Measured width and depth of each specimen, and the direction tested for anisotropic materials
• Support span and the span-to-depth ratio used
• Radii of the supports and of the loading nose
• Conditioning procedure, and the temperature and humidity at test
• Crosshead rate, and the strain rate it was calculated from
• Flexural strength, or flexural stress at 5 % strain where the specimen did not break
• Flexural modulus, and how the slope was taken
• Number of specimens, the mean, and the standard deviation
• Any specimen discarded, and why

What the machine must be capable of

Force demand is low. Most moulded bars fail between roughly 20 N and 1 kN, so a frame of around 2 kN carries almost all D790 work, provided the load cell resolves cleanly down to about 10 N rather than merely reaching the top of its range. Thick laminated thermoset bars run at 32:1 or 40:1 can call for several kilonewtons. Force indication must be verified to ASTM E4.

Rate is defined as a strain rate and converted to crosshead speed by R = ZL²/6d. Procedure A runs at 0.01 mm/mm/min and Procedure B ten times faster at 0.10 mm/mm/min, giving about 1.4 and 13.7 mm/min for the 3.2 mm bar on its 51.2 mm span. Supports must be span-adjustable: 16:1 as standard, 32:1 or 40:1 for high-strength reinforced composites that would otherwise shear, and up to 60:1 where modulus is the object.

Modulus requires a Type II deflectometer classified to ASTM E83 Class B-2; Class C is accepted for the other reported values, and Type I crosshead-position measurement is also permitted. The fixture is a three-point rig — two cylindrical supports and a central loading nose, 5.0 ±0.1 mm radii unless otherwise specified. For bars 3.2 mm deep or more, the support radius may be raised to 1.6 times the depth and the nose radius to 4 times, so the sides of the nose do not foul the specimen. Supports and nose must be parallel, with the bar centred between them. Testing takes place in the D618 atmosphere; other temperatures apply only where the material specification calls for them.

What goes wrong in practice

Four failures recur. A loading nose of small radius indents a soft or thin bar, so local crush is counted as deflection and modulus reads low. A reinforced laminate run at 16:1 fails in shear between the plies instead of in tension, and the number recorded is not a flexural strength at all. Ductile grades reach the 5.0 % strain cut-off intact, leaving no strength value to report. And a span left at a previous setting, or taken from nominal rather than measured depth, biases every result in the batch.

Procedure A or Procedure B

One method, two rates, and the choice belongs to the material rather than to the schedule.

Results from the two procedures are not interchangeable. Plastics are rate-sensitive, and the faster run generally returns the higher strength.

How it differs from the standards nearest to it

These four all bend a bar, and none of their results may be pooled with another's.

ISO 178 is the one mistaken for D790 most often. The geometry looks identical, but the specimen, the rate basis and the modulus definition all differ — and the support radius is the detail most often missed when a D790 rig is reused for ISO work. Numbers from the two cannot be compared, and must never be averaged together.

Questions we are asked about this test