Hard and Tough Tool Steel

[DTM]

I am designing a component (Cam Driver) that has 2 features. One feature that acts as a cam, it slides against another component with heavy surface stress to move the other component.

The 2nd feature of this Cam Driver is a threaded rod that mates with a piston in a pneumatic cylinder, thus it is cyclically loaded as the pneumatic cylinder slams up and down.

So for the cam, we want a hard wear resistant tool steel (Like D3). But for the piston rod, we want a tougher tool steel (Like L6).

We made a prototype from A2 hardened to 55Rc (For legacy reasons, it was similar to some other components we have some experience with).

We broke the piston rod portion in the threaded region.

My ideas were:
1) Case harden this component. Is this expensive? Which steel to use?
2) Make with L6 hardened to 55Rc then anneal just the threaded piston rod portion. Is this possible?
3) Redesign to make 2 different components.

Any other ideas? Any tool steels that are significantly tougher than A2 hardened to 55RC but provide nearly as good wear resistance?

Thanks for any of your thoughts or ideas.

 

[CFDFEAGURU]

Do you have anymore information on the failure? How many cycles did it last for? Did it fail in tension, shear, bending, some other complex mode? What are dimensions of the threaded rod?

Threads are stress raisers and can cause havoc if being used in the wrong situation. Maybe a weld with as smooth a contour as possible could be used?

Anyways, the more information you have the better the forum can help you.

Thanks
Matt

 

[DTM]

The failure appeared to be from fatiugue after 250,000 cycles. It was an M10 thread. It was loaded in pretty much pure tension. When the pneumatic cylinder goes the other way, a shoulder takes the load, not the threads.

So far we've increase the threads as much as we can to M14. For preload, I was taking the threads to about 85% of the yield of annealed A2 (Sy~ 85ksi) The hardened to 55Rc is actually much higher yield, but since it has so little ductility at this hardness, I'm hesatent to preload too high.

This preload is enough to prevent joint separation and minimize cyclical loads.

I'm concerned that the hardened A2 is so brittle that it makes me uneasy. This component must not break in service.

We'd rather not weld in assembly.

 

[Q_Goest]

Hi DTM. I had a similar problem that was discussed breifly https://www.physicsforums.com/showthread.php?t=342256". Can you provide some kind of sketch of your assembly where it is breaking?

 

[FredGarvin]

Can you describe the exact location of failure? Was it in the root (I am assuming this but want to make sure)? Have you tried a finer pitch thread form or is that not possible because of the interface to the cylinder?

Also, did you perform a temper heat treat cycle? If so, when EXACTLY did you do this. I know with A2 it's important to basically remove the part from quench and then right into the temper furnace (not letting the part temperature drop too low before tempering). You can also look into multiple temper cycles to refine the grain structure for better toughness.

 

[DTM]

We didn't do an SEM analysis or anything, but I really think it failed in the root. We could switch to a finer pitch thread, but I don't think that would buy us more than a few % and I'm looking for a lot more than that.

We purchased this part from a manfg. in China. We tested to ensure the hardness is indeed RC 54-56, but that's all I know about the heat treatment.

 

[FredGarvin]

We purchased this part from a manfg. in China. We tested to ensure the hardness is indeed RC 54-56, but that's all I know about the heat treatment.

You might want to look into taking one of the parts and putting it through a temper cycle and see what that gets you.

 

[CFDFEAGURU]

How many cycles do you need to achieve?

Thanks
Matt

 

[DTM]

I need this component to last at least 10 million cycles. I really want to be below the endurance limit so it never fails.

Will a tempering cycle change the hardness? Or just make it a bit tougher while maintaining the hardness?

 

[FredGarvin]

Tempering will not change the hardness since it is a lower temperature operation. IIRC you don not go above the transition temperature. The thing that tempering gives you is a refined grain structure.

Q_Goest's link to a separate thread also lists some other things you could try, i.e. shot peening (surface treatment) to help with fatigue life.

 

[DTM]

Thank you all for all of your comments and ideas.

I'll try to attach a x-section drawing to show a bit more about this component. This is our newer design which has the M14 thread. The one that broke had only an M10 thread. It also had a rubber gasket under the washer which was not good. It made for a low stiffness compression path and much more cyclical loading went throught the stiffer, tension path (the threaded rod). So these design changes should really improve things, but I'm still nervous about the brittleness of the material.

 

[CFDFEAGURU]

Are you going to test the redesign soon?

Here is a paper on the subject.

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V2X-4TS6SSY-2&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&_docanchor=&view=c&_searchStrId=1041469951&_rerunOrigin=google&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=3a6da1366ca2ccd7ff5fa237cf0f5197

Thanks
Matt

 

[DTM]

Thanks for that paper.
We should get the new parts to test in another 2 weeks.

Also, I just tested the hardness of this part and it reads at 63Rc all over the part. It was supposed to be 54-56Rc. Somehow this was missed in our QC. I know harder means more brittle, less fatigue resistant, I'm not sure if going from 55 to 63 would make a huge difference...

 

[Q_Goest]

Are you going to test the redesign soon?

Here is a paper on the subject.

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V2X-4TS6SSY-2&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&_docanchor=&view=c&_searchStrId=1041469951&_rerunOrigin=google&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=3a6da1366ca2ccd7ff5fa237cf0f5197

Thanks
Matt

Nice paper. Thanks for that.

 

[Q_Goest]

Thank you all for all of your comments and ideas.

I'll try to attach a x-section drawing to show a bit more about this component. This is our newer design which has the M14 thread. The one that broke had only an M10 thread. It also had a rubber gasket under the washer which was not good. It made for a low stiffness compression path and much more cyclical loading went throught the stiffer, tension path (the threaded rod). So these design changes should really improve things, but I'm still nervous about the brittleness of the material.

Couple suggestions on the design change, attached. As you've noted, anything that can reduce the spring constant of the 'bolt' and increase the spring constant of the 'joint' will help reduce the alternating stress. To that end, I'd suggest cutting the bolt down to the thread root as far along the part as possible, which looks to be the bottom of the plate where it appears to be sealed. May have to redesign the seal at that point to accommodate this. Also, having a washer that's split and who's ID comes right up to the thread root may help increase the joint constant slightly.

It's interesting that the failure occurred at the root of the second thread and not the first where it would be expected. I wonder if there's significant bending in the plate? That may push stresses farther up the thread by putting a slight bending load on the nut, moving the stress off that first thread slightly. Just a thought...

Other suggestions (generic):
- Increase the height of the nut (per Bickford) this helps improve fatigue life, though I forget exactly why.
- Add a 15 degree chamfer to the nut where the threaded part screws in. This takes some stress off the first thread (puts more bending into first thread instead of direct shear).
- Shot peen threads. Rolling is nice, but it doesn't help where the threads terminate on the shank.
- Increase thread root radius. Talk to a machinest regarding easily obtainable thread forms used for cutting threads.
EDIT: Couple more thoughts
- How are you measuring preload? If just torque, I'd suggest doing something else to confirm actual preload. Pressure sensitive film? Measure bolt stretch?
- How are you preventing loosening of the nut? As I'm sure you're aware, as soon as it loosens up, the joint is done for.

I just tested the hardness of this part and it reads at 63Rc all over the part. It was supposed to be 54-56Rc. Somehow this was missed in our QC. I know harder means more brittle, less fatigue resistant, I'm not sure if going from 55 to 63 would make a huge difference...

I thought harder (stronger) increases fatigue life? Although that's generally true, I think you'll need to discuss this with a metalurgist to see what might be done with your particular material.

Side note: Any possibility of hydrogen embrittlement? Hard chroming processes I've heard can potentially put hydrogen into a material. Any other sources of hydrogen?

 

[DTM]

Q_Goest, Thanks for your ideas. To answer some of your questions...

The nut is chamfered on the ID, so the thread doen't really go all they way to the face of the nut. So the 1st thread is probably not really loaded by the nut. Thus the failure at the 2nd thread.

Just above the nut is a goove in the threaded rod and a snap ring is installed to prevent the nut from backing out. We also use high strength loc-tite. The threaded rod ends about 1 mm above the snap ring.

The seal is indeed between the bottom of the piston and the cam driver shaft (o-ring).

There will be air pressure either above or below the piston, driving it up and down. This will actuate the cams. Depending on where you are on the stroke, the cams are either resisting with an axial load or a radial load. There are 6 symmetric radial loads (6 cams 60 deg. apart) and the main shaft of the cam driver is well supported, so no bending is transferred up to the threaded region. The piston has a clearance fit in the cylinder, so the only bending should be from any dynamic non-uniform pressure distribuiton which I don't think is generally very much in a pneumatic cylinder.

Harder (stronger) is not always good for fatigue, When you get high up on the Rc scale, it also means more brittle. I just got some good info from the metals handbook and it looks like L6 tool steel might be a better choice, it is rated "Very high" toughness and Medium Wear resistance. I could deal with a little more wear on my cams in exchange for a guarantee of not breaking the threaded region which would be very bad in service.

There was no plating processes at all, so I don't think the steel should have any hydrogen, carbon, or other surface impurities.

I will be adding a root radius control to the threads.

Preload is measured just by torque. This will be a fairly high volume production product and we can't mess around with sophistacated techniques for preload measuring. We need the material to be tough and forgiving to a rough assembly process.

Thanks all-

 

[Q_Goest]

Hi DTM.
You mentioned the thread failed after 250,000 hours. Do all of these cycles produce the same stresses on the threaded joint, or do some cycles produce less stress? Can you say how many units you have in service and if any (or what percentage) have exceeded the 250,000 hours? Is this a possible random case (maybe the nut got loose) or is this the most hours you’ve put on a cylinder? And how many hours does 250,000 cycles represent anyway?

I assume you’ve tried to perform a fatigue analysis and plot the operating point on a Goodman diagram. Where do you think you were previously? (where do you think you can get to with the 14 mm thread? Have you compared them?) Can you post your Goodman diagram? I have an Excel spreadsheet that I can do this with if I have the mean and alternating stresses. The inputs have to include various surface factors and stress concentration factors. It calculates fatigue limit for the metal using the same basic parameters used by Bickford and a text I had in college. It’s worth creating some kind of automated spreadsheet every time you have to delve into something like this. You’ll save lots of work down the road by learning it once, then documenting that in a spreadsheet.

Regarding wear resistance, have you considered surface treatments? Consider the basics like nitriding, but also consider various platings. I’ve used a coating by General Magnaplate that was excellent. If you discuss your application with them, they can make a recommendation, but of course there are many other coating suppliers. A little research into what your competitors use might prove useful also.

You mentioned the nut is fixed with locktite, which is good. A high strength version should be sufficient, but are the threads clean and free of oil prior to assembly? Have you ever disassembled a unit and found wet locktite? I have. We were using Permatex Gel Threadlocker and tested another that wasn’t locktite. Both failed to set up. Loctite worked well even with a surface treatment that seemed to make the others fail. As for the clip above the nut, feel free to dispose of it. That will do a great job at keeping the nut from coming all the way off, but won’t do anything to keep it from backing up, and if it backs up, it’s over. Take it out and call it a cost savings.

You mentioned preload is controlled using torque, which is fine for production. But have you actually measured one to see what kind of friction you have in the production environment? I’d suggest taking one and actually measuring the stress in the bolt, either by measuring the stretch, measuring the force, or some other method to determine, given your torque, what the actual preload is. Don’t take friction values from a textbook and apply them. I learned something very interesting recently. I have a situation that’s almost identical to yours. Bolt material is Nitronic 60, annealed. Nut material is 304 SS. All joint materials are 304. I determined spring rates for the bolt and joint using FEA. Then applied those spring rates to a calculator (spreadsheet again). Using canned formulas for bolt load/stretch/stress, etc… I wanted to back out the “coefficient of kinetic friction”. Of course, the parts I have are not nuts and bolts, so there are some assumptions made by the canned equations that don’t exactly fit my geometry. But the error should be linear. So I essentially backed out the coefficient of friction for my situation and found it was as low as 0.05 the very first time I torqued a brand new part. The 2’nd 3’rd and 4’th times I tried it, the coefficient of kinetic friction was 0.10. Exactly double the first test. The subsquent tests were all very consistent, it was just the first test that seemed to stand out. I think the issue was some minor yielding on the contact faces between the threads because stresses were far too low to produce any yielding anywhere else in the joint. So I’d suggest measuring at least one unit from the production floor, prior to it ever being assembled, and try a few cycles of torque while measuring actual force, displacement, etc… so you can correlate the readings to a friction coefficient.

Is there really a correlation between toughness and fatigue strength? I’m not aware of any. I’d be interested in a discussion of how fatigue life is affected by factors other than surface factor, gradient factor, load factor, and stress concentration.