Non exactly zero stress at unload (elastic model)?
Non exactly zero stress at unload (elastic model)?
(OP)
Hello,
For a linear isotropic elastic material, I applied a load defined by an amplitude (0,0;0.25,1;0.5,0;0.75,-1;1,0) to obtain uniaxial tension on a block.
At steptime 0.25 and 0.75, the stress is OK (constant=to the applied load); at time 0.5 and 1, there is a very small negligible stress (1.10^-13 MPa compared to 10 MPa) distribution in the part. I am very surprised with this result at unload (see attachment) especially in linear elasticity where you should come back to the initial state when you unload. I tried it in ANSYS, and one obtains exactly zero!!!!
Is this normal? Is it a problem known with Abaqus? Is there a way to overcome this? It is perturbing when teaching to students.
Thank you for your help
Rgx
For a linear isotropic elastic material, I applied a load defined by an amplitude (0,0;0.25,1;0.5,0;0.75,-1;1,0) to obtain uniaxial tension on a block.
At steptime 0.25 and 0.75, the stress is OK (constant=to the applied load); at time 0.5 and 1, there is a very small negligible stress (1.10^-13 MPa compared to 10 MPa) distribution in the part. I am very surprised with this result at unload (see attachment) especially in linear elasticity where you should come back to the initial state when you unload. I tried it in ANSYS, and one obtains exactly zero!!!!
Is this normal? Is it a problem known with Abaqus? Is there a way to overcome this? It is perturbing when teaching to students.
Thank you for your help
Rgx





RE: Non exactly zero stress at unload (elastic model)?
Also, as a teacher you should know that there is no such thing as absolute zero in numerics. A computer code does not know or care about it; a developer does. Odds are ANSYS (by default, perhaps) decides to round it off for you and Abaqus does not (1.10^-13 in the context of the model is likely to be as good as zero). What you might want to pay attention to is that ANSYS users tend to rely on units, whereas there is no such thing as a unit in Abaqus - the user is responsible to choose a consistent unit system appropriate for the physics they are simulating.
*********************************************************
Are you new to this forum? If so, please read these FAQs:
http://www.eng-tips.com/faqs.cfm?fid=376
http://www.eng-tips.com/faqs.cfm?fid=1083
RE: Non exactly zero stress at unload (elastic model)?
Find attached the inp file that was run to obtain the previous result.
According to the Field output lines, the end result is different (10^-13 in the present case).
If you comment the first line and uncomment the second one, the end result is EXACTLY zero.
*Output, field, variable=PRESELECT, number interval=4
***Output, field, variable=PRESELECT
I am just impressed that for elasticity, we obtain two different results according to the number of intervals that are requested.
I agree that 10^-13 is zero but it does not explain why it depends on the number of intervals.
Thank you for your help.
rgx
RE: Non exactly zero stress at unload (elastic model)?
Absolute zero is only possible when you are doing an analytical analysis (i.e., working with mathematical equations), not a numerical one. Again, in certain situations, a developer may choose to display an exact zero on the screen but there is no exact zero in a computer - it can not be. For instance, it is possible that 10^-20 is an insignificant number in one application whereas the same number can mean a LOT in another, depending on the physical quantity/unit system/ .. Without context, numbers have no meaning.
*********************************************************
Are you new to this forum? If so, please read these FAQs:
http://www.eng-tips.com/faqs.cfm?fid=376
http://www.eng-tips.com/faqs.cfm?fid=1083
RE: Non exactly zero stress at unload (elastic model)?
Zero stress is displayed with some "output field" options and not for others.
Zero stress is displayed for the tensile direction at unload but not for other components.
Of course I am going through the documentation, but so far I did not get a clear answer about the "output field" options and their effect on the result.
Regards