What's new
What's new

4340 parts failing magnetic particle inspection

IAmAlex

Aluminum
Joined
Sep 11, 2015
Hi all,

Hoping that someone can provide some insight here as I am having an issue that I cannot seem to get solved. I have been making some parts from 4340 steel bars that are sawn from plate. I then cut those bars down, mill, heat treat, and they go to mag particle inspection. I have made three lots of these parts over the past year with the following results:

Lot #1
Material Heat: #1
Machining Process: #1
Heat Treat Lot: #1
MPI Results: 85 pass/0 fail

Lot #2
Material Heat: #1
Machining Process #1
Heat Treat Lot: #2
MPI Results: 7 pass/16 fail

Lot #3
Material Heat: #1
Machining Process: #1
Heat Treat Lot: #3
MPI Results: 1 pass/25 fail

To summarize, my material supplier is reporting all material as coming from the same heat (note that I did NOT order all of the material at one time. The material came in three different shipments, but they are reporting it all from the same heat). The machining process has not changed since the first lot. The material is being heat treated to 50-53 Rockwell C.

I am not at the shop to take a picture of a failed part, but here is a mock up of the failure cracks that my MPI vendor is reporting:

drawing.JPG

The top and bottom bold black lines are the where the faces of the plate would be pre-machining. The red lines are the cracks. As you can see, they are nearly perfectly horizontal cracks that run constantly parallel to the surface of the plate, with very few exceptions. Nearly all of the cracks occur at the exact same height in the plate as well. That height is the same on every part. This continues around the entire 2D profile of the part. The parts are about 1/2'' thick.

My material vendor came in and took one of my parts and sent it back to the mill for testing. I just received their test report and they are claiming no issues with the material. They are faulting heat treat or machining (surprise surprise). Despite their results, I am skeptical of that for two main reasons:

1. There is NO WAY I am putting enough stress into this part with a 1/2'' endmill, 1/8'' endmill, and 1/8'' drill to crack it.
2. The location, shape, and consistency of the cracks screams to me that there is a constant layer in the material where the melt is not bonded properly, which is then worsening/splitting during heat treat.
(Why this is happening in lots 2 and 3 but not on 1, I don't know. Separate plates from the same heat or a bad section of the same plate?)

If anyone taking the time to read this has some wisdom to offer, it would be much appreciated. I am not ruling out heat treat, but given the physical evidence that I am seeing I find it unlikely.
 
It is a material fault: it might only show up after heat treat but heat treat cannot cause it, rather, it exposes the fault. I think they call it 'piping', where porosity in the ingot gets rolled flat but doesn't weld together when rolled. So it's not exactly cracking, but just delaminating. Still doesn't help you, though, you need new stock cut from another heat.
 
Could be the edge of the plate VS the middle of the plate...

The equal height crack says (to me) that it probably happened in rolling.

How about an ultra sonic inspection prior to you even setting a cutter to the
material.. The place I buy plate from has a company down the street that will do
it, they run it down there and pick it up the next day, and I get certs... And its
not that expensive.

An ultra sonic inspection should easily pick up a layered crack like that, actually
I think that is what it is meant to pick up.
 
You say they are cracks, but are they really. Could be material seams, folds or quench cracks. You will need to take a section through the " indication" mount it and take a look under an electron microscope. Seams and folds are from the mill. Seams will contain crud and folds will have a decarb layer. Quench cracks have an obvious high temp iron oxide in the crack. It is an unusual place for a quench crack. Not a fold on a cut face, so I'm betting an seams from the mill. This is related to steel cleanliness and overall reduction ratio from cast thickness to rolled thickness. You normally cannot use the term clean steel and Asia in the same sentence.
 
It is a material fault: it might only show up after heat treat but heat treat cannot cause it, rather, it exposes the fault. I think they call it 'piping', where porosity in the ingot gets rolled flat but doesn't weld together when rolled. So it's not exactly cracking, but just delaminating. Still doesn't help you, though, you need new stock cut from another heat.

This was my thought exactly, as well as my heat treater's. Glad to hear someone else came to the same conclusion with the same evidence.

Send a coupon of the original material for heat treat without any machining to see if there is an issue? What is the temper of the material as supplied?

I do have leftover pieces of short stock from lots 1 and 3. I may end up doing this.

Could be the edge of the plate VS the middle of the plate...

The equal height crack says (to me) that it probably happened in rolling.

How about an ultra sonic inspection prior to you even setting a cutter to the
material.. The place I buy plate from has a company down the street that will do
it, they run it down there and pick it up the next day, and I get certs... And its
not that expensive.

An ultra sonic inspection should easily pick up a layered crack like that, actually
I think that is what it is meant to pick up.

Good idea. The lab that the mill had test the material is showing use of ASTM E45 on their report. This is a microscopy inclusion check, so am I crazy in thinking that of course they did not find anything with this... because whatever they are checking for (sulfide, alumina, oxide) would have escaped when the layer delaminated? This test is not actually checking anything past the surface of the material, which is what I've already paid my MPI vendor for twice. :nutter:
 
You say they are cracks, but are they really. Could be material seams, folds or quench cracks. You will need to take a section through the " indication" mount it and take a look under an electron microscope. Seams and folds are from the mill. Seams will contain crud and folds will have a decarb layer. Quench cracks have an obvious high temp iron oxide in the crack. It is an unusual place for a quench crack. Not a fold on a cut face, so I'm betting an seams from the mill. This is related to steel cleanliness and overall reduction ratio from cast thickness to rolled thickness. You normally cannot use the term clean steel and Asia in the same sentence.

Ricky, can you take this out a step further? What does each different type of failure mean?? and how the
heck do you fix it (its already broken).. But how do you catch each one of these methods of failure before
it becomes an issue, like a failed mag particle??? and a whole bunch of time and money down the shitter.
 
You say they are cracks, but are they really. Could be material seams, folds or quench cracks. You will need to take a section through the " indication" mount it and take a look under an electron microscope. Seams and folds are from the mill. Seams will contain crud and folds will have a decarb layer. Quench cracks have an obvious high temp iron oxide in the crack. It is an unusual place for a quench crack. Not a fold on a cut face, so I'm betting an seams from the mill. This is related to steel cleanliness and overall reduction ratio from cast thickness to rolled thickness. You normally cannot use the term clean steel and Asia in the same sentence.

I am calling them cracks but I think you are likely right in what the correct terminology should be. I do not think they are quench cracks either as they are too clean and consistent, and not located around something like a weak feature on the part.
 
Heat treating can definately cause 4340 to crack, especially a poorly designed part, or a part that is incorrectly heat treated.

Carefull design is as important as choosing a heat treater who can correctly heat treat 4340, not all of them can.

I'm in the aerospace prototype world, and have worked at 2 large companies where the use of 4340 was verboten.

Did the parts fail dye pen inspection as well?
 
Ricky, can you take this out a step further? What does each different type of failure mean?? and how the
heck do you fix it (its already broken).. But how do you catch each one of these methods of failure before
it becomes an issue, like a failed mag particle??? and a whole bunch of time and money down the shitter.
Buy E4340 or 300M grade for starters :rolleyes5:
 
Last edited:
Good idea. The lab that the mill had test the material is showing use of ASTM E45 on their report. This is a microscopy inclusion check, so am I crazy in thinking that of course they did not find anything with this... because whatever they are checking for (sulfide, alumina, oxide) would have escaped when the layer delaminated? This test is not actually checking anything past the surface of the material, which is what I've already paid my MPI vendor for twice. :nutter:

The inclusion test is not looking at the surface. It is done on a cut section of the material, and is looking at a surface that used to be inside the bar. E45 specifies where the sample needs to be cut from the bar. The sample is then mounted, polished and etched. Under a specified magnification the non-metallic inclusions are counted and classified. There are some pictures of the result in a broken stud analysis here: Premature failure of re-sulphurized steel studs

The inclusions won't escape during preparation.
 
The inclusion test is not looking at the surface. It is done on a cut section of the material, and is looking at a surface that used to be inside the bar. E45 specifies where the sample needs to be cut from the bar. The sample is then mounted, polished and etched. Under a specified magnification the non-metallic inclusions are counted and classified. There are some pictures of the result in a broken stud analysis here: Premature failure of re-sulphurized steel studs

The inclusions won't escape during preparation.

That makes sense then. If that is the case though, how are these seams not showing up on the inclusion test?
 
I think replies 2,4 and 5 have it. The defects happened during rolling; they usually result from contamination but can also be caused by rolling at too low a temperature. I have uncovered not just cracks but one-inch-long x half-inch-wide x .060 deep voids in 304 stainless. There is virtually nothing you can do to avoid this except use "aircraft quality" vacuum arc remelt 4340 (MIL-S-8844 if I remember correctly). Metal purity, or rather the lack thereof, is at the bottom of just about all material faults, whence the growing resistance to Chinese steels.
 
Its a definate rolling defect, have seen it more than a few times in rolled flat strips too, plate wise, if you can get them to cut the strip off the end of the plate you will get good ones in the middle in all likely-hood and more bad ones as you get near the edge. Strips cut off the edge of such a plate will probably all be full of defects.
 
and how the
heck do you fix it (its already broken).. But how do you catch each one of these methods of failure before
it becomes an issue, like a failed mag particle??? and a whole bunch of time and money down the shitter.

4340 and other high strength alloys are available in VAR (vacuum arc remelted) material, usually in bar form. This material is double the price of air melt, but clean as a whistle. A small price to pay on mission critical parts with lots of machine time.

OP's blanks were cut from plate. Not sure if VAR material is available in plate form. Forge shops typically have 4340 VAR ingots on hand and can make blanks to just about any size.
 
Its a definate rolling defect, have seen it more than a few times in rolled flat strips too, plate wise, if you can get them to cut the strip off the end of the plate you will get good ones in the middle in all likely-hood and more bad ones as you get near the edge. Strips cut off the edge of such a plate will probably all be full of defects.

This, and an earlier comment, plus the fact that the "cracks" are showing up at the same place in the failed parts, leads me to agree.
 
That makes sense then. If that is the case though, how are these seams not showing up on the inclusion test?

The test specifies a minimum area of only 0.25 in^2 per sample, and it needs to be parallel to the rolling direction. Multiple samples are taken per heat, but it generally isn't a good test for defects like you have. The samples are typically taken close to the middle of the bar, but your schematic show the defects close to the surface of the bar. The samples are small relative to the volume of metal they represent. There are better tests to look for rolling or forging defects that can sample more of the volume of the material.
 
In addition to adama's info, it has always been standard practice at the places I've worked to request material from a ways from the edges of burnouts from plate products whenever possible - especially on critical parts that need to be crack or mag tested. If not possible we always had them add some stock so a good bit past the edges could be machined away. Rolling defects tend to be near the edges of rolled plate since this cools faster than the center.
 
The test specifies a minimum area of only 0.25 in^2 per sample, and it needs to be parallel to the rolling direction. Multiple samples are taken per heat, but it generally isn't a good test for defects like you have. The samples are typically taken close to the middle of the bar, but your schematic show the defects close to the surface of the bar. The samples are small relative to the volume of metal they represent. There are better tests to look for rolling or forging defects that can sample more of the volume of the material.

They performed this test on one of my failed parts that had been machined and heat treated. They did not perform it on raw coupons from the material heat. The seams show up on both sides of the part, parallel to the rolling direction of the plate. The two sides of the part are 3.00'' inches apart at the most and about 1'' at the least. If the seams are showing on both sides of my part, I would not think that you could take a section through the part and get good results. I am trying to understand how they reported acceptable results from a part that very clearly has these defects. Maybe they cut a section through my part that just so happened to be free of seams?
 








 
Back
Top