Relation between Stress Intensity Factors and J-integra via CINT calculation - please help

[Fanatik]

Hello, I am using CINT command in Ansys v.14 to evaluate values of Stress Intensity Factors and J-integral. What is different between these two "types" of CINT command? Only change in input is in CINT,TYPE,JINT / SIFS... My problem is that theoretically there is relation between J and K (sifs) that J=K^2/E. It is for elastic problem because sifs are not defined in elastic-plastic problem. But my question is: if I create simulation with plasticity and I order Ansys to evaluate Sifs by CINT command, the relation J=K^2/E should work? I tried to create this simulation and for small-scale loading it works, but for higher loading J-integral calculated directly from CINT command does not equal with J-integral calculated from K (which was evaluated by CINT). And i guess why? Just dont be "angry" because i am using stress intensity factors in plastic simulation, i know it is not correct but i though that "numerically" the relation between J and K from CINT should work in any type of loading/materials and obviously it is not. Can anyone help me and tell me what is different between CINT for J-integral and for K-factors? Some explanation for Ansys inner routines should be helpful... Thanks a lot for everything.

 

[rb1957]

K is for Linear Elastic Fracture Meachanics.

Jintegral is for Elastic-Plastic Fracture Machanics. the J = K^2/E relationship is valid for elastic loading; if your model has large scale plasticity, then i don't think you can use K and need to use a Jintegral approach. this'll be ok if you're looking for residual strength, but if you're looking for crack growth i think you're in for "a world of hurt".

Quando Omni Flunkus Moritati

 

[Fanatik]

Thanks for answer, but i know that K is not valid for plasticity... i just thought that CINT command works similarly in elastic and plastic and the number value for CINT,SIFS should be evaluated to J even if it is not valid and doesnt mean the right value for plastic problem... i wanted to know what exactly changes in integration behind the CINT command for plasticity and i am not discussing the validity of K for plastic problem

 

[rb1957]

"if I create simulation with plasticity and I order Ansys to evaluate Sifs by CINT command, the relation J=K^2/E should work?" sounds like you are trying to calc K from J.

i would expect that the J integral calc knows nothing of K; that they're working from internal stresses along a path.

i would have thought that we use the realtioanship to calc equivalent K from a J integral result.

Quando Omni Flunkus Moritati

 

[Fanatik]

Well, as far as i know, CINT integration evaluates contour integral (in ansys help it is called Interaction integral) which is very similar to theory of J-integral. From this integration Ansys calculates J or K. KI, KII and KIII are separated with use of some auxiliary K-functions which are =1 or =0... At least this is how i understand it. I got this from sciencedirect.com, article: Interaction integral procedures for 3-D curved cracks including surface tractions, written by Matthew C. Walters, Glaucio H. Paulino, Robert H. Dodds Jr.

From this i was expecting that J=K^2/E should work anytime, obviously it is not and i want to know why.

And rb1957 can you explain what do you mean by "equivalent K"? I am not from english speaking country and sometimes i have problems with understanding english technical expressions. Thanks.

 

[rb1957]

i suspect that your model, at least your integral path, has a high degree of plasticity (stresses above yield). i think J = K^2/E isn't valid with the plastic stresses.

I expect you're running NL ANSYS ?

If you're running linear ANSYS, then you'd expect the J = K^2/E relationship to hold (since your model is elastic) ... however the results are not real (because reality is plastic).

"what is different between CINT for J-integral and for K-factors" ... have you looked at the ANSYS manuals ? the developer manual might go into this detail ? have you asked the help desk ? have you tried your model is 1% loading (ie, drive it to be elastic) ?

Quando Omni Flunkus Moritati