We pulled about 100 of these undamaged links from the system which failed. Every link which tested 45 and higher failed prematurely in terms of elongation. The heat treat certs show they were using the correct process. Those links which were below 45 failed in a nice grouping at almost twice the displacement on average (.003 vs .0055). Failure loads were still well over factor of safty. One of our engineers thinks our failed link could have been scratched on the surface (see photo) and because of notch sensitivity this initiated the failure which then propagated to the d-ferrite layer. Do you think this is a sound conclusion? This part should see very little bending in normal operation.
Maybe you should re-write the spec to lower the hardness slightly. I hope that you are actually testing tensile properties and not hardness. I would go with a minimum yield strength, min and max UTS, and minimum elongation.
This may require test heat treatment of a sample from each master coil. Coil number are important, heat numbers are secondary.
If you need help making the correct contacts let me know.
= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection
This just gets more interesting. What percentage failed? How was the centerline banding distributed between the failed and the good?
I don't think a part with no stress across the centerline can fail there, unless the crack was there prior to the normal loading, so, no, I don't see the need to hypothesize a surface scratch.
They all failed since we were pulling them on our instron machine. I imagine that the sub 45 HRC links were made from one lot of material and the others were made from another. We have comingled stock (no first in first). I haven't looked at those specimens. Great idea! Can you clarifiy this is statement: "I don't think a part with no stress across the centerline can fail there, unless the crack was there prior to the normal loading, so, no, I don't see the need to hypothesize a surface scratch". The engineer isn't saying that the centerline cracked from the scratch but that the scratch was where the failure initiated (normal to the centerline crack)and then propagated along the d-ferrite layer. I haven't found any data which correlates hardness and notch sensitivity for 17-7PH TH1050. Have you?
The stainless steel handbook shows that the notch tensile strength, a good measure of toughness, of 17-7 PH decreases fairly rapidly as tensile strength increases, so I would have to say that notch sensitivity unquestionably increases as hardness increases. This is irrespective of the ferrite layer, which, if your engineer is right, may be an innocent accomplice to the fracture initiated in the PH martensite.
I had an issue with 17-7PH a few years ago, the dog bones were showing correct tensile and elongation, but were splitting perpendicular to the ductial failure. I sent it in for failure analysis, and it was found the cause of the splitting was:
"EDX tests showed increased percentages of Al and Cr, and decreased Ni content."
The certs showed every thing was fine, but I guess mixture of the elements were not homogenious enough.
mighoser,
The heat treat was close, TH1110 (we have to do things a little different). I saw the photo, and it did not look like that, but after reading you had multiple problems, I thought I would throw my experience in for your thoughts. I cannot post the photos of my problem (my company thinks it owns everything). But, the dog bone had a normal ductal failure.
Step 1: require lot control and part marking (laser or acid etch) based on the master coil number for the material. You can trace this back to heat number, but not every coil from the same heat will have the same thermomechanical history.
Is the sheet aged in the coil? After it is slit?
What are the tensile properties? Were any of hte tensile samples RC47? If not someone is cheating.
I still suspect that there is non-uniform aginging and the 'extra hard' zone in some material is cracking during blanking and this results in the early failures. There are a lot of segregation issues in these grades other than just delta ferrite.
If I had samples delaminate in tensile testing I would reject the coil as simply being unsound material.
= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection
Thank you for the sugestions. I'm sure we will implement some form of corrective action to prevent future occurances of this. I don't have tensile prop data from the coil. I still need to look at this. Sheet is received in Contition A and then cut into squares. It is then treated to TH1050 and then fine blanked. During tension tests with actual parts (links) we were able to correlate performance reduction with hardness. But none of the links delaminated the way the one in the posted photo did.
I would be looking for some data on uniformity of properties after the heat treatment. Are the sheets stacked when they are heat treated? Are they unstacked to cool? Are there thermocouples inside the stack?
I can see all kinds of issues. You need a bunch of tensile data on the sheet material after heat treatment.
I would play with bend testing. Maybe even bending samples and then pulling them in tension. You need a way to catch this condition early in the process.
= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection
As aging temperature increases, hardness drops and fracture toughness increases, so there is a strong inverse correlation. Toughness drops off by 50% for a 20% gain in strength, which equals hardness.
Segregation, as monkeydog cites, causes ferrite, but bear in mind that the non-centerline is depleted of Cr and Al, so it will have different heat treat response than a homogeneous alloy.
That photo is incredible. If you ever needed any demonstration of the low toughness of delta ferrite, this is it.
But since it is on the neutral plane of mighoser's part, as with the tensile specimen, it shouldn't be the root cause of failure. I'm looking at the excessive delta ferrite as an indicator of inhomogeneous composition throughout,which could have caused the heat treat response to be abnormal.
To bad we don't have coil tracability. This could proably be solved by looking at coil history and comparing to statistics on other material of the same grade.
It is also possible to have other problems (than retained delta ferrite) in these alloys which result large variations in hardness.
My hunch is that there is a band (or two) of harder material in your strip. When they were blanked you got some cracking in that band.
I have always hated the 'TH1050' type of nomenclature. The inferance that 1050F will always give you specific properties just isn't true. Due to variations in chemistry and thermomechanical history you have have to be prepared to use different temperatures.
We often lab heat treat samples, rather quick and dirty. We compare the results with historical data and decide if that material need a 'little more' or a 'little less'. A 25F change can work wonders some times to reach the required ductility
= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection
