Friday stumper: is the coefficient of friction between threads in this situation? 2

[cbrf23]

I'm running myself in circles trying to solve my own problem. Looking for some smarter minds than mind to slap me straight.


I have a nut made of stainless steel.
I have a bolt made of alloy steel which has been completely coated in a CVD coating which has a .05 CoF.

If I apply moly/graphite lube to the threads, which has a .075 CoF, what then is the CoF between the threads?

Originally, I was thinking it would just be the minimum value (.05), but I've gone and over-thought it and no I'm not so sure that is a safe assumption.

I think this is much easier to think of in another way:
-------------------------------------------

The CoF numbers provided are determined from an ASME standard that measures each material's CoF against plain dry steel.
In this standardized test, the lubricant (e.g. moly-lube or PVD coating) is applied to some object which is set on a plain dry steel plate, the plate is then slowly tilted until the object starts to slide.
The angle at which the object begins to slide is measured, and the CoF is derived from this angle.

Performing this test on a plain steel object with no lubricant, tells us that the steel has a .80 static CoF.

If I were to replace the plain steel plate in this test, with one that had the CVD coating (static CoF=.05) and I were to run the Moly-lube test again, at what angle would the moly-lube slide?

Is it going to hold on to the same angle, corresponding to .075 CoF?
Or will it slide at an angle corresponding to the lowest static CoF, meaning .05 is the result?
Or, will slide at something less than that, some product of .05 and .075?

Is there anyway to calculate this other than experimental testing? (I need to use a number for theoretical modeling and I can't do experimental testing)

 

[kingnero]

Empirical testing will probably yield the best results...

Performing this test on a plain steel object with no lubricant, tells us that the steel has a .80 static CoF.


That isn't a realistic value...

Https://youtu.be/558Htmux6UE

 

[cbrf23]

@kingnero

I don't have parts to test - I need to determine the numbers theoretically for a calculation.

I have not done any empirical testing myself; the steel CoF number provided was just for reference.
Most of the resources I looked at listed the static CoF of plain (dry) steel on plain (dry) steel between .74 and .78.
I just rounded to .80 in my question above because I felt it was easier to work with for explaining the problem and (hopefully) explaining a solution.

Regardless of the actual CoF of steel, do you know any way to determine the CoF of two materials measured against a common third material, without empirical testing?

 

[kingnero]

No, I do not.

 

[Tmoose]

" [[[ THE ]]] CoF of two materials measured against a common third material..... "

What is the unit load of that ASME test?
The under head loading of a fastener designed to work for a living can be pretty high, enough to embed in soft steel or iron.
I would not dare presume the CoF is particularly linear with load.

Here are some test results from as well respected fastener company of the results of fastener clamp loads variation resulting from repeatedly tightening a fastener using various a few different lubricants.

http://arp-bolts.com/i/UT/UT_ARP8pp.004.jpg

The presumption is the act of tightening smooths or buffs down high spots, so the CoF generally gets lower except when it gets higher.

========
From the Unbrako engineering guide.

VARIABLES IN TORQUE
Coefficient of Friction
Since the torque applied to a fastener must overcome all friction before any loading takes place, the amount of friction present is important.
In a standard unlubricated assembly, the friction to be overcome is the head bearing area and the thread-to-thread
friction.
Approximately 50% of the torque applied will be used to overcome this head-bearing friction and
approximately 35% to overcome the thread friction. So 85% of the torque is overcoming friction and only 15% is
available to produce bolt load.
If these interfaces are lubricated (cadmium plate, molybdenum disulfide, anti-seize compounds, etc.), the
friction is reduced and thus greater preload is produced with the same torque.
The change in the coefficient of friction for different conditions can have a very significant effect on the
slope of the torque tension curve. If this is not taken into consideration, the proper torque specified for a
plain unlubricated bolt may be sufficient to yield or break a lubricated fastener.

 

[EdStainless]

You will have to do empirical tests.
The values that you have are not mated to SS at high loads, so neither of them gives you the correct answer.
Or you could make a series of assumptions (SS on hard alloy steel dry, then lubed) and see what range of values you get. If it is a fairly small range then use the assumptions. But if it results in a large variation you need to plan some tests.


= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[racookpe1978]

You WILL face not only a friction (static, then dynamic as it loosens force) but a very real potential for galling (static, low temperature welding) of the stainless to steel threads.

These tiny thread areas are under extreme pressures at the thread-to-thread contact point, and those thousands of psi will squeeze out the lubricants differently than on a bench test of two "simple" flat plates at the tested "low" pressures when separated by the same lubricant.

 

[tbuelna]

The question you should be asking is what would be acceptable static/dynamic coefficient of friction values in my industry for this particular case that I can use in an analysis? Friction coefficient values can vary greatly, especially with the contact conditions in screw thread connections. Depending on what characteristic you are analyzing the threaded connection for (does prevailing friction provide an advantage or disadvantage in your situation?) you would select a CoF value to ensure your analysis covers all realistic conditions.

Your other option would be to require an in-process verification of the prevailing torque for each installation of this component.

 

[cbrf23]

@tbulena, this CVD coating is new - I'm not aware of anyone else in the industry using this coating.


Im disappointed that no one can offer any suggestions towards rough estimation. I realize that empirical testing will be required to obtain any modicum of accuracy, but as I stated in the original post, I can't do experimental testing right now. I don't have parts to do so.

I was hoping to get a ballpark number, it doesn't have to be super accurate, but just some way to roughly estimate the CoF without doing experimental testing.

Surely I can't be the only person in history looking to estimate a CoF before parts are made.


For what it's worth, the CVD coating is only a few millionths of an inch thick, and doesn't change the surface finish at all, or at least not in a measurable way. While dry SS threads do have galling issues, this CVD coating and the moly-lube each individually have proven to prevent galling entirely in similar applications.

Just to clarify, this assembly is permanent, the threads are one-time use. We do not have to concern ourselves with changing CoF due to removal/re-installation as that is outside of the scope of use/service for this particular product.

 

[Tmoose]

"I'm disappointed that no one can offer any suggestions towards rough estimation."

I think that everyone did !

 

[racookpe1978]

If you are not willing to test a "first-time application" of a "new lubricant" that will be used "only once" because "I do not have to ever take the assembly apart" then you are designing for failure.

Darn near guaranteed failure. Your assumptions WILL BE wrong. Either your assumptions will be too tight (you will assume a coef of friction - up to 85% of the torque goes into friction per the above! - that assures you do NOT get enough pre-load on the bolt to clamp the assembly together); or your assumptions will be too loose, and you will never the assembly apart. It will be destructively removed, perhaps catastrophically removed or failed.

Either way, the assembly fails. Make 12 bolts and nuts. TEST THEM.

 

[Screwman1]

I had a lab manager who used to say "without data, all you have is hope". That is pretty much your case here; you have no data and are just hoping that everything will work the way you 'hope'.

 

[cbrf23]

@Tmoose,

I'm paraphrasing, but it seems the only advice I've received is to do empirical testing.
If you look at my original question, the advice I was looking for was how to estimate without testing.

Is there anyway to calculate this other than experimental testing?

It seems that this is not possible, however I find it fascinating that there's no way to predict how the slide test would be affected by adding the CVD coating - assuming all other variables remain the same.

I do have lots of empirical data on these threaded components without the coating.
I know what torque is needed to achieve the preload we want without lubricant, and what torque is needed to achieve preload with the the moly-lube.
I also know the CoF specified for the moly-lube and for the CVD coating.
Someone mentioned surface finish earlier, and I don't know the exact surface finish on the threads (if you know a good way to measure surface finish on the flanks of a 1-32 thread please share) but I know our machining process is consistent, and this is a variable that will not change due to the coating.


All I was hoping to do was roughly estimate what kind of reduction in torque I could achieve by adding the coating in addition to lubricating the threads.
Would it reduce the torque at all, or would adding lubricant to a surface that is more lubricious than the lubricant have no effect.
That's all I'm trying to get answers on - roughly what behavior to expect.

These parts are undergoing final validation testing now - that's why I don't have any extras to run experiments on.
As we get closer to release, I'm looking at ways of simplifying the production process and possibly eliminate some specialty equipment that is currently necessary to achieve the required torque without coating.
If I could drop it by 50-100 ft-lbs, we could potentially assemble the tool with existing standard equipment, rather than having to develop special fixtures and procure new (expensive) equipment.
Before wasting money on ordering more prototypes that could possibly offset the cost savings of the production benefits, I wanted to determine if this idea (of adding moly-lube in conjunction with the CVD coating to lower CoF even further) had any legs.





@Racookpe1978

I think you are jumping to conclusions. If I were to go forward with this experiment, I would thoroughly test the product to determine the proper torque required to achieve the desired preload.
That said, I usually put forth my best efforts to determine if the idea has any merit before I waste money and time on experiments that could be avoided.

I was hoping to be able to answer <YES or NO> to the question "would moly-lube + the CVD coating have a lower CoF than either alone, on the same product, in the same application, under the same loads and all other conditions being the same."

I didn't realize that would be an impossible question to answer

 

[Tmoose]

Hi CBRF23,

Are you tightening the nut or the bolt?
Are you lubing the nut or bolt bearing face?
Note the Holokrome info I provided mentions thread friction can only be 35% of the total torque requirement, and the under-head friction contribution can be around 50%.
========================
" I know what torque is needed to achieve the preload we want without lubricant."

How does that compare to the .80 static CoF that I think you measured using the ASME method?

What are the conditions of the test from which .05 CoF for the CVD coating was established?

If this is an install-once application, and you know the unlubed torque required, why is the expense of a fancy coating being considered?

regards,

Dan T

 

[cbrf23]

@tmoose

I wish I never would have mentioned the nut and bolt, simply because it keeps leading astray from the core of the inquiry: Which is what affect adding a lubricant to an already extremely lubricious (in fact more lubricious than the lubricant) surface would have?


Although in the interest of candor:
I am tightening the nut.

I've done tests with the threads "dry" (i.e. as machinined) as well as "dry and clean" (i.e. cleaned using acetone to remove trace oil residues from machining) as well as with the moly-lubricant.
In the moly-lubed tests, the lubricant was applied to the threads and bearing surface of the nut, as well as the threads of the shaft.

The COF of .80 for steel I think I stated was a generalization provided solely for the purposes of explaining the problem, it is not an actual number that I measured. It was provided for reference only.

The CVD coating test and moly-lube test were performed to the same ASTM standard. I'd have to look up what it was, to look up the test conditions required.

The CVD coating provides corrosion resistance. In this application, other traditional corrosion inhibiting coatings have not held up to the wear of moving parts in service. Other wear resistance coatings examined did not meet other application requirements, such as resistance to bending moments seen in service as well as fatigue life with regards to damage accumulation and crack propagation. This CVD coating promises both excellent corrosion resistance as well as the durability and lubriciousness needed to hold up to the moving parts.

While I'm able to roughly calculate the torque that should be required by replacing the moly-lube (or dry) threads with this coating, it was unclear to me what could be achieved by combining the coating with the moly-lube.

 

[EdStainless]

The addition of the lube should help reduce friction 'at very high loads'.
The problem is that most friction measurement is actually done at rather low loads, nothing like those seen in a threaded fitting.
How the CVD preforms as the substrate begins to deform is the real question.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[tbuelna]

cbrf23-

I think you're missing the point. You stated you need a CoF value to use in an analysis of a thread connection. What you need to consider is that someone will likely review and sign-off on your analysis. And they will want to know what basis do you have for the CoF values used in your analysis. I don't think they would accept an answer such as "my best guess" or "an internet forum". The fact that the coating you are considering is new and has limited test data available for friction properties makes things much more difficult. Friction characteristics can vary greatly, and to get a reliable result you would likely need to test many sample under carefully controlled conditions.

Take a look at table IV of this NASA document. It lists static and sliding CoF values for various combinations of thread materials/lubrication that would be accepted for use in an analysis by the aerospace industry. Since you don't have this type of basis available for your material/coating combination, and you don't have the ability to perform testing of the material/coating combination, then your best remaining option would be to ask the person that will review/approve your analysis what value they would accept for static/dynamic CoF.

 

[cbrf23]

@edstainless

Yes, lube typically helps, but when the CoF of the substrate is actually less than that of the lubricant - does it still help then?
(I've never ran into that scenario before now, and I can't imagine it's very common)

Aslo, excellent point in testing at low loads vs application at high.
Given the tribological nature of the coating, which is that it reduces friction and increases wear (and corrosion) resistance without changing the substrate material's properties, I would expect the threads to behave (deform) in much the same way as without the coating. Therefore, I would expect the behavior to be similar to what I have tested, with the only change being to friction at the thread and bearing surface interfaces. I have no way of predicting the exact nature of this performance - I simply want to propose the best hypotheses for my best-case analysis.

As far as how the coating will perform at high loads, I can't give a definitive answer.
I take what the coating engineers who designed it have told me, which is that the coating should maintain all of its properties at these loads, and I use that for my assumptions.
I'm just looking to figure out what assumptions to use for some best case/worst case analysis, which serves no purpose further than to allow me to use my best judgment to determine whether or not there is likely benefit to the application and whether it's worth pursuing further (e.g. testing, or spending more time on it in any other manner)


@Tbuelna

I see your point, but I think you are making assumptions regarding the nature of my work.
Nobody has to sign-off on this work.
I'm doing rough calculations to determine whether or not it's worth spending time/money on further studies on the nature of this material/lube combination, that's all I'm doing.

I love reading research papers, so thank you for sharing that!
Again though, don't assume I am not using a reliable source for my analyses.
I've been using tables from the Machinery's handbook for most of my calculations - which I believe also to be as reliable a source as needed for this analysis.
Also, don't assume I would ever put a part into production based on theory.

I don't know how everyone else works, but I tend to try to understand things the best I can before jumping in to making physical parts and testing them.

With all the recommendations to go do testing I've received, I don't think anybody has even bothered to ask what a part costs.
You don't know whether my parts coast $.30 or $30,000 a piece. Given they are closer to the latter (though not quite that high) I think it's worth trying to find answers before making parts


That said, I do think I have an answer, and that is that there is no good way to assume what will happen.
With that in mind, I will just run more calculations about what could happen, and go from there.
I find these thought experiments are useful as it helps identify potential failure modes and gaps in analysis. When it comes time to move into testing, it means I have a better idea of what to look for.

And for that, I thank everyone for their participation and efforts in guidance on this topic.

 

[3DDave]

What is the basis for your calculations? I don't really follow the tribology community, but I would think they would be crowing about a calculation method for accurately determining friction based solely on theory.

 

[cloa]

What about a rough method that gets a ball-park result which what the OP is seeking in order to commence testwork. I don't think anyone in tribology would desperate for that- they do actual tests such Prof Gwidon Stackovich (guessing the name's spelling).

http://www.eng-tips.com/threadminder.cfm?pid=1529

Use translation assistance for Engineers forum

Note the rules include No Student posting