Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations TugboatEng on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Unnotched S-N Curve for Higher Kt
- Thread starter Lee.Conti
- Start date
Kt = 1 is "unnotched".
maybe "for" should be "from" … in which case no, you cannot use the Kt = 1 data for a corner radius.
does "corner radius" mean a hole (Kt = 3) or the corner of a cut-out, in which case you have to determine the Kt from the cut-out geometry (ESDU or Petersen), most likely > 3.
I don't think there is any reliable method to calculate a Kt curve from unnotched data. If there were then no one would bother with notched s/n curves.
what material ?
another day in paradise, or is paradise one day closer ?
maybe "for" should be "from" … in which case no, you cannot use the Kt = 1 data for a corner radius.
does "corner radius" mean a hole (Kt = 3) or the corner of a cut-out, in which case you have to determine the Kt from the cut-out geometry (ESDU or Petersen), most likely > 3.
I don't think there is any reliable method to calculate a Kt curve from unnotched data. If there were then no one would bother with notched s/n curves.
what material ?
another day in paradise, or is paradise one day closer ?
- Thread starter
- #3
The point is to have s/n data that is somewhat representative of your geometry.
If your geometry is Kt (or Kf) = 2.4, then you could interpolated between Kt = 2 and Kt = 3 s/n curves. You could possibly "fudge" at Kt 2 curve (if that's all you have) You could use Kt = 3 as being conservative or "fudge" it to get something a little less conservative.
"plastic" ? If this is a composite part, then I think you should test if you Need something "accurate". It may be ok to say the fatigue stresses are so low that fatigue is not a problem. Is it solid "plastic" or laminate ?
another day in paradise, or is paradise one day closer ?
If your geometry is Kt (or Kf) = 2.4, then you could interpolated between Kt = 2 and Kt = 3 s/n curves. You could possibly "fudge" at Kt 2 curve (if that's all you have) You could use Kt = 3 as being conservative or "fudge" it to get something a little less conservative.
"plastic" ? If this is a composite part, then I think you should test if you Need something "accurate". It may be ok to say the fatigue stresses are so low that fatigue is not a problem. Is it solid "plastic" or laminate ?
another day in paradise, or is paradise one day closer ?
If I'm working with a material with well-known fatigue test data, I usually use the next higher Kt chart. If the Kt of the feature is 2.4, and have charts for Kt=2.0 and Kt=3.0, I use the Kt=3 chart.
There is a way to deal with this. Schijve goes into the subject at great depth. Unfortunately all that proves is that you need to have a great depth of understanding to know what you're doing.
What material are you using?
There is a way to deal with this. Schijve goes into the subject at great depth. Unfortunately all that proves is that you need to have a great depth of understanding to know what you're doing.
What material are you using?
- Thread starter
- #6
Hi rb1957 & SparWeb,
I understand. I also use higher Kt chart for the calculated kt or interpolation.
However, I am working on a PC plastic (not prepreg composite). It seems the fatigue information is not common. Only kt =1 S-N curve is available. So, I was wondering if I use unnotched S-N curve for the concentrated stress location, it seems anti-conservative.
Alternatively, I extract nominal stress as I am using solid elements to avoid peak stress. Then use it to compare with unnotched curve.
What do you think?
I understand. I also use higher Kt chart for the calculated kt or interpolation.
However, I am working on a PC plastic (not prepreg composite). It seems the fatigue information is not common. Only kt =1 S-N curve is available. So, I was wondering if I use unnotched S-N curve for the concentrated stress location, it seems anti-conservative.
Alternatively, I extract nominal stress as I am using solid elements to avoid peak stress. Then use it to compare with unnotched curve.
What do you think?
SWComposites
Aerospace
You need to either get fatigue data for test specimens with a representative Kt, or you need to fatigue test the actual part.
L.C...
The physical/mechanical/chemical/fracture properties of metals... and thermosetting or thermoplastic fiber reinforced composites... and homogeneous plastics/plastic-alloys... and other non-metallic materials [glass, ceramics, etc] have to have [1] element in common... specified/rigorous material compositions and fabrication processes... otherwise the product made from/per these is '1-off'... or garbage.
IF You have a 1-off part in a unique configuration, then every aspect has to be quantified/stabilized before quantification testing can begin... much-less the validation testing a production proof part.
Please advise...
PC = ? = [monolithic] polycarbonate? Per 'what' material specification? In what 'Form': sheet? extrusion? injection-mold/cast? machined [drilled, etc]? additive-manufacturing?
NOTE. The 'simplicity' of additive manufacturing is often heavily counter-balanced by the complexity of material/processes validation/certification... especially for safety critical parts.
Regards, Wil Taylor
o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]
The physical/mechanical/chemical/fracture properties of metals... and thermosetting or thermoplastic fiber reinforced composites... and homogeneous plastics/plastic-alloys... and other non-metallic materials [glass, ceramics, etc] have to have [1] element in common... specified/rigorous material compositions and fabrication processes... otherwise the product made from/per these is '1-off'... or garbage.
IF You have a 1-off part in a unique configuration, then every aspect has to be quantified/stabilized before quantification testing can begin... much-less the validation testing a production proof part.
Please advise...
PC = ? = [monolithic] polycarbonate? Per 'what' material specification? In what 'Form': sheet? extrusion? injection-mold/cast? machined [drilled, etc]? additive-manufacturing?
NOTE. The 'simplicity' of additive manufacturing is often heavily counter-balanced by the complexity of material/processes validation/certification... especially for safety critical parts.
Regards, Wil Taylor
o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]
- Thread starter
- #10
Hi SWComposites, rb1957,
Unfortunately, we do not have test data yet existing practice is using Yield Strength for evaluation which I find it not suitable as it is random vibration.
It is not structural part but because it is under random vibration condition, performing cumulative damage would be more appropriate to evaluate the failure.
Hi WK Taylor,
It is Polycarbonate and injection molding.
Unfortunately, we do not have test data yet existing practice is using Yield Strength for evaluation which I find it not suitable as it is random vibration.
It is not structural part but because it is under random vibration condition, performing cumulative damage would be more appropriate to evaluate the failure.
Hi WK Taylor,
It is Polycarbonate and injection molding.
I'd do a coupon test. I'd use a shoulder stepped shaft with a Kt like what you have in design. Support the shaft on two bearings and apply a static load; different loads or different positions to have different stresses (due to bending) at the Kt. then spin the shaft, maybe 1000 rpm, maybe 10000 to get a cycle count. nice little research project.
But it sounds like you don't have that type of condition. "random vibration" brings DO160 to mind … test full sized parts ? but if not structural, then you cares if it fails … what are the consequences of failure ? I expect that answer is "contract requirement".
I guess being non-structural is an opening for an out … if you have some Kt data and a Kt of your install you "could" scale the unnotched s/n to get a good result (!) and "rationalise" your scaling. Sure, someone could call BS on this (and quite rightly) but it could be a way to complete your contractual requirement ?
another day in paradise, or is paradise one day closer ?
But it sounds like you don't have that type of condition. "random vibration" brings DO160 to mind … test full sized parts ? but if not structural, then you cares if it fails … what are the consequences of failure ? I expect that answer is "contract requirement".
I guess being non-structural is an opening for an out … if you have some Kt data and a Kt of your install you "could" scale the unnotched s/n to get a good result (!) and "rationalise" your scaling. Sure, someone could call BS on this (and quite rightly) but it could be a way to complete your contractual requirement ?
another day in paradise, or is paradise one day closer ?
SWComposites
Aerospace
I doubt there is a valid cumulative damage model for polycarbonate. Just test it.
- Thread starter
- #13
Hi rb1957,
There is a testing for the product under random vibration but because it is a non-structural product so the failure commonly observed at the mounting area. In analysis, we only compare the statistical result with YS only, not even with Fatigue Strength.
Hi SWComposites,
The reason I think of cumulative damage model as it is random vibration. Mind to tell me your doubt? I agree that perhaps we shall perform a fatigue test under random vibration.
There is a testing for the product under random vibration but because it is a non-structural product so the failure commonly observed at the mounting area. In analysis, we only compare the statistical result with YS only, not even with Fatigue Strength.
Hi SWComposites,
The reason I think of cumulative damage model as it is random vibration. Mind to tell me your doubt? I agree that perhaps we shall perform a fatigue test under random vibration.
no what I think SWC is telling you is that no one has developed a cumulative damage model for PC.
If you have unnotched s/n data, and the part is non-structural (failure is "benign", not a hazard to safe and continued flight, and readily detectable) then there is a case for factoring the unnotched s/n curve (or your applied stress) for your Kt. This should show that fatigue is not an issue no matter (almost) how conservative you are. This is not a rigorous analytical approach so you can get called out for it … "BS".
The only alternative I can see is a simple coupon test (as I described above).
another day in paradise, or is paradise one day closer ?
If you have unnotched s/n data, and the part is non-structural (failure is "benign", not a hazard to safe and continued flight, and readily detectable) then there is a case for factoring the unnotched s/n curve (or your applied stress) for your Kt. This should show that fatigue is not an issue no matter (almost) how conservative you are. This is not a rigorous analytical approach so you can get called out for it … "BS".
The only alternative I can see is a simple coupon test (as I described above).
another day in paradise, or is paradise one day closer ?
- Thread starter
- #15
Hi rb1957
Thanks for clarifying it. I wonder if it is metal, we can use Miner rule, for instance, as accumulative damage method for random vibration fatigue? Why not PC?
I agree that if it is not structural part, it may not need to perform fatigue analysis or perhaps just use the statistical stress from random analysis against fatigue strength will do.
What do you think?
Thanks for clarifying it. I wonder if it is metal, we can use Miner rule, for instance, as accumulative damage method for random vibration fatigue? Why not PC?
I agree that if it is not structural part, it may not need to perform fatigue analysis or perhaps just use the statistical stress from random analysis against fatigue strength will do.
What do you think?
SWComposites
Aerospace
Why not PC? well, the material behavior is likely very different. Most metals exhibit large amounts of plasticity. PC probably is a brittle material with little to no non-linear stress-strain response. Miner's rule is an empirical method that has been validated with metallic fatigue data; it is not physics based. If you don't have fatigue data for PC validating Miner's rule method or any other empirical method, you cannot just make a wild assumption that the methods are valid for the material.
- Thread starter
- #17
Hi SWComposites,
Thanks again for responding to my question.
Correct me if I get it wrongly that if we have S-N curve for PC, it is fine to do Miner's rule or other cumulative damage model for random vibration fatigue, is that right?
I read the this Altair Blog about getting notched S-N curve from unnotched curve and in prior experience, we also use Kf for such purpose but these are for metal materials.
Once again, thank you everyone for sharing your thoughts here!
Thanks again for responding to my question.
Correct me if I get it wrongly that if we have S-N curve for PC, it is fine to do Miner's rule or other cumulative damage model for random vibration fatigue, is that right?
I read the this Altair Blog about getting notched S-N curve from unnotched curve and in prior experience, we also use Kf for such purpose but these are for metal materials.
Once again, thank you everyone for sharing your thoughts here!
- Status
Similar threads
- Question
