curious tensile test results

[Tmoose]

we received the attached results from a testing lab for a "pull" on a metallic (some flavor of steel T12) sample. Some folks speculate something slipped during the test. The lab says the results are valid.

Comments ?

Dan T

 

[metengr]

DanT
Results look suspect to me. Looks like a hypersensitive extensometer was used based on the strain scale and when the material began to yield. Something to review;

http://www.asminternational.org/pdf/spotlights/5106_01.pdf

 

[Metalguy]

Find another lab!

"You see, wire telegraph is like a very long cat. You pull his tail in New York and his head is meowing in Los Angeles. Do you understand this? Radio operates the same way: You send signals here, they receive them there. The only difference is there is no cat." A. Einstein

 

[Metalhead97]

This is the strangest tensile test plot I've ever seen and is a reason why multiple tests always need to be done.

Metalhead

 

[redpicker]

Two problems.

One, the extensometer seems to have slipped just prior to full section yielding. The only way for the (apparent) extension to decrease with increasing load is for the extensometer to slip. There are at least 3 places that show a sharp "kink" which makes me wonder about how well the extensometer is tracking.

Two, the specimen slipped in the grips right after yielding, causing a drop in load and corresponding drop in extension.

Problem two is irrelevant; it will not change the results.

Problem one is a bit more complicated. If the requirement was, for example, a minimum 36,000 PIS yield, these issues would not affect the pass/fail. For more precise work, they will affect the result. I've seen this slipping quite a bit, particularly with strip-type specimens. Having too fine a finish on round specimens can cause this, too. Keeping the knife-edges sharp on the extensometer (by replacing them often) will help prevent this. I suspect that it occurs because the cross section of the specimen changes drastically when yielding starts and I've always thought that one of the reasons for the 0.2% offset was to put the load for yield determination out on the flat part, so these minor discrepancies would not affect the result.

If it were my lab and this was routine testing, I would consider this a valid test. If my customer questioned the result, I'd offer to pull another test at no charge. On the other hand, if it were my lab, I would keep the knife-edges sharp and take other measures to prevent this from occurring in the first place.

I do have to say I do not like the axis labels of Stress and Strain, since they are inaccurate. Most likely, the measurements are made in pounds and inches, which is how the data should be reported. The "Engineering Stress" and "Engineering Strain" are calculated based on the specimen dimensions, but the graph suggests that the actual stress and actual strain were measured during the test, which is not true. Yeah, a bit of a nit-pic, but so is the issue with the extensometer slipping. It's just a matter of the size of the nit.

rp

 

[Metalguy]

"----and I've always thought that one of the reasons for the 0.2% offset was to put the load for yield determination out on the flat part, so these minor discrepancies would not affect the result."

No, it was selected as not being enough plastic deformation to affect structural items.

The aircraft industry uses 1/10 that amount, or 0.02%--at least they used to back in the stone age when I did tensile testing.

"You see, wire telegraph is like a very long cat. You pull his tail in New York and his head is meowing in Los Angeles. Do you understand this? Radio operates the same way: You send signals here, they receive them there. The only difference is there is no cat." A. Einstein

 

[redpicker]

The aircraft industry uses 1/10 that amount, or 0.02%--at least they used to back in the stone age when I did tensile testing.

I don't know what to say other than my experience is not the same, although I only go back to the Iron age ;^). I've been involved in manufacturing and testing for Boeing, NASA, and US Military (Army, Navy, and Air Force), among others, and have always used 0.2% when testing with the offset method.

A temperature change of 20C results in 0.02% linear change in steel.

rp

 

[EdStainless]

I have gotten so that I always ask to have the modulus reported as well. If it is too far out of line I use that as a basis to force a retest.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[Metalhead97]

Edstainless,

I do that as well: any problems with the extensometer will show up better in the modulus.

metalhead

 

[redpicker]

I have gotten so that I always ask to have the modulus reported as well. If it is too far out of line I use that as a basis to force a retest.

Just for grins, I had calculated the modulus for the test in the OP. 33 million PSI. I didn't mention it because that's probably closer to 30 than the instruments used are capable of. It is a good check on the test, but I am not sure where you put the limits on what is acceptable vs. unacceptable.

To me +/- 10% is a bit too stringent, particularly if you take the "correct" value as being 29.8 million PSI. What does this forum think the cut-off should be? +/- 25%? Personally, I don't worry too much if it's in the 25-35 million range, but I'd really like to know what others do.

rp

 

[Metalhead97]

redpicker,

Using a video extensometer you can get Youngs Modulus values +/- 3000 ksi (3 msi). At-least this is what I'm used to with our equipment.

metalhead

 

[redpicker]

Thanks, Metalheat97, that is the kind of info I was asking for. Is that number based on round bar specimens, strip specimens, or a generalization over both types?

I've never used a video extensometer, so I have no idea how they work. All the extensometers I've used are of the LVDT type, which is what I had assumed what was used in the OP, Of course, it is not only the extensometer that is important, but also the cross-sectional area (since, despite the axis labels, you are not measuring the stress), and the standard tensile specimen allows variations within the gauge length that can add to errors in the calculation of the modulus.

rp

 

[Metalhead97]

redpicker,

The material that we test is fine wire (0.010"-0.002" in diameter) and strip (as thin as 0.003"). As you would expect, slipping is a huge problem for fine wire and hence an extensometer needs to be used when elongations need tighter control.

I think it call comes down to material being tested. With soft materials like gold you can get away with cross-head displacement as a means to determine strain. However, we have been developing some precious metal alloys that have UTS values over 300 ksi, and thus precise elongations need to be used because 1 or 2% elongation may be the difference in an alloy have 'good' or 'poor' formability.

Video extensometers work by tracking pixels on a specimen and thus there accuracy are determined by resolution of the camera chip.

metalhead

 

[TVP]

Tmoose,

Was the sample a flat tensile specimen (dogbone) with standard dimensions according to ASTM E8, A370, or similar? As the others have stated already, this is an unusual curve, and is not representative of a high-quality test with good accuracy/precision.

 

[EdStainless]

We test tube samples using both LVDT and strain gage extensometers.
I used to use optical on wire testing (with 10" gage length).

I have never seen a test with a modulus reported high that wasn't obviously in error.
Given that all of our alloys have real modulus in the range of 28-30 million then I will question anything outside of 25-30.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube