Fatigue damage due to compressive load 1

[39minuteman]

I have a question regarding fatigue damage done by compressive-compressive load.

I believe that common approach is to consider only load cycles in which max stress is greater then zero. In other words, assumption is made that compressive-compressive cases (both max and min stresses are compressive) don’t create fatigue damage.

I’m aware that in some cases local compressive yielding could produce residual tensile stress and on that way promote failure of parts which are loaded only in compression.

But, I’m talking here about more fundamental process. If I understand correctly, crack initiation process is caused by local cyclic plastic deformation. As plastic deformation could result from both tensile and compressive cycles, it seems to me that damage due to compressive cycles should be added to.

Actually this post is caused by one remark in MSC.Fatigue manual. In section 14.3 where effect of mean stresses is discussed, there is a note that Smith-Watson-Topper mean stress correction predicts that no fatigue damage can accrue when the max stress becomes zero or negative, which is not strictly true.

Further, in numerical example given in ESDU 95006 document (“Fatigue Life Estimation Under Variable Loading Using Cumulative Damage Calculations”), no difference is made between compressive or tensile cycles. For both, absolute value of local strain range is calculated and used with Strain Life equation to obtain damage increments.

I would really like to hear your opinion on this issue. Do you consider fatigue damage from compressive-compressive load cases and how do you treat them?

 

[saksmito]

I´ve never worked with NASTRAN, but regarding the fundamentals of your question I can tell you that you are right. Compressive loads can create fatigue stress. I can tell you for the fatigue cracks created in engine pistons.
I think it should be more important in pieces which support side buckling stresses.
Hope I´ve been useful.

 

[GregLocock]

"
I believe that common approach is to consider only load cycles in which max stress is greater then zero. In other words, assumption is made that compressive-compressive cases (both max and min stresses are compressive) don’t create fatigue damage.
"

Never seen that.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[rb1957]

i agree with the OP that the common approach (in airplane stcutures) is to neglect the -ve stresses in the spectrum. I agree that -ve stresses do cause fatigue but at a much slower rate than tension stresses. I base this on crack growth curves for -ve R ratios, which show the crack growth rate something like 100x slower than R=0; also fatigue data shows much longer lives with -ve R.

I don't have the ESDU reference in front of me, but i'd be surprised if anyone takes the -ve stresses and turns them into +ve stresses. they'd also have to the zero crossing, otherwise they'd totally mess the spectrum. this sounds like a highly conservative appraoch, possibly in an application quite different to aero-structures, where the spectrum (and the impact of -ve stresses) are different.

I think the answer is that occassional small magnitude -ve stresses are negligible in a spectrum dominated with +ve stresses. Possibly they are not negligible in a spectrum dominated by -ve stresses, with a few tension peaks.

 

[GregLocock]

Yes, that's what I was getting at. Dislocations will move if there is energy available, and there is still energy available in a purely compressive oscillating load.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[der8110]

Crack initiation is actually caused by a geometric discontinuity (flaw). Propagation is due to elastic deformation, that is why the science is called Linear Elastic Fracture Mechanics (LEFM). Plastic deformation actually causes retardation of crack growth, which is used to advantage in cold working holes and shot peening.

Unless you have some really weird geometry, loads that vary entirely on the compression side won't cause fatigue. Probably an unforeseen tensile side of a bending load condition. Not to say you can't have breakage, striations that are indicative of fatigue would be all mashed down, but have to get there by some prior tensile loads.

One thought is you might have a Mode II or Mode III crack growth. Only seen a true Mode II once in practical aviation experience, and never a Mode III.

 

[39minuteman]

First, thank you all for your valuable responses.

der8110
I believe that you are referring to method based on fracture mechanics, in which is common to assume that some initial flaw already exists in the material.

However, my understanding is that fatigue crack could develop in “flawless” material too, at the location of high stress concentration (of whatever nature).

rb1957
Regarding calc in ESDU procedure

They didn’t convert negative stresses to positive one. Hysteresis loop response has been developed for nominal positive/negative stresses and then (absolute) values of local strain ranges are used for evaluation of fatigue life.

Ok, that makes sense in context of your replies. There is probably assumption that energy available for movement of the dislocations is proportional to the strain range.

Thing is that we are using modified stress based approach which neglects negative cycles, for the calculation of crack initiation time. And while there is apparent relation between tensile stresses and crack propagation, as pointed by der8110, I was unsure about effect of neglected compressive stresses on crack initiation.

So, as compressive cycles do have some effect on crack initiation, which is obviously lower then effect of tensile cycles, I would like to hear how you treat them.

Is it enough to apply mean stress correction (as negative mean stress will increase life) or some other factors should be applied too?

 

[Compositepro]

I believe that pure compressive stress will not cause fatigue in a material. However, in reality it is almost impossible to apply a compresive load and not get a complex stress distribution in a material. A cylinder under compression will "barrel", resulting in tensile and shear stresses.

 

[CoryPad]

Compositepro has identified the important consideration. The "ideal" condition of UNIFORM compression-compression is not realistic. Compression almost always leads to non-uniform stress, with shear (and sometimes) tension nearby. While dislocations may move, it is essentially impossible for the material's strain capacity to be exceeded, which is when cracks initiate.

Bearings are an important example here. They are loaded in compression, but there is a subsurface shear stress that results. Eventually, shear strains accumulate, and cracks can initiate and propagate.



Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[prost]

If compressive (negative stresses that is) cycles don't cause damage, then why do landing gears fail from fatigue level stresses (that is, below compressive yield)?

 

[saksmito]

In my career, in the subject "resistance of materials" we were taught about the "core" of the beam (sorry. I studied in Spain and don´t know the name in english).
It´s the part of the beam where you apply your load and have only compressive stress. It used to be a geometrical closed curve within the center of the beam (usually), and if you applied the load in the whole surface you had compressive and tensile strains.

Apart from shear strains (that in several cases, like the one from Corypad in bearings, uses to be the highest one) I knew that appeared tensile strains when applying compressive loads to the whole surface of the beam in some parts.
Sorry but I can´t give you any further details. I´ve just moved and have my books 1200Km away. If anyone knows what I´m talking about maybe can explain it in detail.

Cheers.

 

[rb1957]

prost ...
'cause landing gears see considerable tension loads due to bending, mostly from "spin-up" and "spring-back" drag loads. plus the critical sections are beams, transferring the ground loads into the wing box.

 

[GregLocock]

OK, I had a dig around. IF the Smith Watson Topper equation applies, then there is a slightly complex argument that results in a fatigue life of >10^7 cycles for purely compressive fatigue, in steels etc.

In composites, glass filaments, and other slightly exotic substances there is no such restriction and C-C fatigue is worthy of consideration. The mechanism in composites (eg reinforced concrete) is obvious - axial compression results in a complex stress field between individual fibres and the matrix in which they are held.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Compositepro]

An argument can be made that all failures are tensile failures. For a crack to open molecules in the area must be moving away from each other and that is a tensile failure. Under compression molecule are pushed closer together. Shear stresses can be resolved into compressive and tensile components. It is the tensile component that is the cause of "failure".

 

[swearingen]

I completely agree with Compositpro. Consider a mass of solid steel in orbit around the earth in a liquid filled chamber. If you changed the pressure in the chamber from 100psi to 10000psi and back repeatedly, you would never fail that mass of steel in fatigue. You might fail a composite, however, because the non-uniform nature of composites may create areas of tensile stresses due to shears or bending in the mass.

This also reflects Compositpro's earlier assertion that there are very few applications where variable pure compressive stresses are present during the life of an actual component.



If you "heard" it on the internet, it's guilty until proven innocent. - DCS

 

[CoryPad]

swearingen,

Your case is hydrostatic stress. Deviatoric stresses are needed for plastic strain, and plastic strain is the prerequisite for fatigue.




Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[DavidDDH]

I have been involved in two cases where what appeared on the surface to be a compressive load environment resulted in cracks developing. Both in the area of the landing gear.

The problem was a very high and very local compressive stress resulting in very local plastic deformation that when the load was removed (ie P=0) the elastically strained material adjacent to the plastically strained material returned to its original unstrained condition and the plastic material was put in tension.

Compression stresses do not cause cracks, but plastic compressive strain can cause tension stesses when the load is returned to zero and the resulting tension stress causes cracks.