Non linear analysis using optistruct
Altair Employee Hello folks,
I am doing geometric non linear static analysis on a structure and got stuck into some elemental problems
under the material section,
material - matx02
some parameters are given such as
A- plastic yield stress
B - plastic hardening parameter
n - plastic hardening coefficient
C - strain rate coeff.
EPSmax. max possible stress
my questions (for titanium ti6al4v)
1. value of A: as per wiki it should be in 90-130 MPa but when I saw some papers value for yield stress was >900MPa
2. B- nothing got from google. what it actually signifies in reality. Please explain.
3. EPSMAX: does its value should be ultimate tensile stress.
is there any website available where one can find the values of any material( specially - ti6al4v) in exactly these parameters?
Has anyone got a research article or paper regarding this topic: 'Geometric Non linear compressive analysis of Ti alloy using optistruct'
Thanks n regards
Answers
-
Hi Ashu,
MATX02 follows Johnson Cook material model which is an approximate method to represent the plasticity of the material.
The field A represents the first or initial yield stress of the material.
The yield stress will not be the same for different materials. In fact, yield stress is a point beyond which the material starts behaving as plastic.
For most of the metals, the yield stress would range between 90-150 or so as you have pointed out but the alloys and high strength metals would have more yield stress value.
You could get this value either from literature or from a simple tensile test from the lab.
For th Titanium alloy who have mentioned you can just google out to get an approximate value of Yield Stress.
The Parameter 'B' represents the slope in the plastic region which is basically the stress over strain. Any slope which is taken up to the Yeild point would fetch you the Young's Modulus and the slope taken beyond the Yield stress would fetch you the Hardening Modulus.
An approximate way to arrive at this value is to manually interpolate a point in between the Yield and the UTS stress and divide it by corresponding strain value.
This value will actually help the MATX02 material model to approximately follow th stress strain behaviour of the material.
This material model though captures the plasticity reasonably, we strongly suggest you to go with MATX36 if you have the full stress-strain curve.
MATX 02 tries to follow the 36 with the limited parameters the user inputs which makes it an approximate method.
EPSMAX is the UTS of the material.
I hope this helps. Do let us know if you need any further information
Thanks,
Vikas
0 -
Thank you @vikas kenchapur for this helping reply.
i will be looking forward to work acc. to ur guideline0 -
Hello Vikas,
i have encountered a fatal error in solving the file. #4002 Radioss engine failedAnd .h3d file hasn't formed after result?
Can you elaborate the meaning of fatal error in result section.
I am performing geometric non linear analysis on a structure.
0 -
Hi Ashu,
Could you please attach the complete .out files.
You will have 3 .out files, the starter, engine and main out. Please attach all the 3 output files
Thanks
Vikas
0 -
Hi Ashu,
Further to MATX02, please find attached spread sheet which helps you to get the 'B' and 'n' parameters.
I hope this helps.
Thanks,
Vikas
0 -
Altair Forum User said:
Hi Ashu,
Further to MATX02, please find attached spread sheet which helps you to get the 'B' and 'n' parameters.
I hope this helps.
Thanks,
Vikas
Thanks for the trouble Vikas...it really means a lot to me....
I wanted to ask some things:
1. When I use MATX02, it requires the value of Yield Stress of material....now since I am doing a compressive analysis I will use the compressive yield strength. But my confusion is this - its theoretical value which I usually found in many websites was around 800-950 MPa, which is the value in case of a standard specimen, but experimental values from some research papers suggest it to be only around 80-120 MPa.
Now I have this understanding that the curve will start yielding from a point coming from this value of 'A' I provide to the solver. Is this true? This is the first confusion I am unable to clear.
Also since I am performing the analysis on multiple strucutures, I will have to compare the results based on their yielding but what I wanted was to find out the yield point through the analysis, but if I myself am putting the value of yield stress than whats the difference?
Can you give me some conceptual explaination regarding the actual relation of the parameter A with the output result? I am using a porous lattice structure but the values are of standardized solid specimen. So, is yield stress dependent on these factors? If yes, what value should I use?......
2. In your previous replies, you told me that MATX02 is an approximation of the Johnson-Cook model and that MATX36 was accurate in these terms.
But I wanted to know, which parameters do I need to input and how to I decide the values. Actually, I am not able to understand all the terms inside MATX36, and cant identify which ones to use in my case.
Is there any tutorial for usage of MATX36 if yes please share it with me it will be really helpful.
Sorry for asking so many things at once. Please do reply, especially regarding point one.
BIG THANKS FOR HELPING ME OUT.
0 -
Hi Ashu,
I got your doubt and I normally get this question from the Industry too. Let me explain it in simple words.
Any FE solver will solve the mathematical equations using the inputs given by the user. The solver would have no idea or is not smart enough to predict the behavior of the material.
The user should tell to the solver how is his material gonna behave and using this material behavior how it affects a component or a structure would be estimated by the solver.
Please note that, you are giving only the material behavior as an input to the solver and the same material can be used on a simple structure like beam or a complex one like Aircraft.
Based on the geometry of the structure and the loading conditions the stress values may differ but the material behavior should be same. For example the Yield stress of the material would not change whether you use this material in a beam or an Aircraft but if the magnitude of load between this two structures is different you may see one of them yielding and the other in elastic zone or one of them would reach UTS the other just cross the Yield or the behavior can lie at any point on the stress-strain curve.
Now, to accurately capture this material behavior, the solver should follow every point on the stress-strain curve. Without a proper input curve, no solver can trace/follow the material points and thus you need to input the stress-strain curve as an Input.
To cross check the robustness of the solver, we always plot the stress-strain curve in the output and see if it matches to the Input stress-strain curve. The solver is accurate only when it follows every point on the Input stress-strain curve.
For MATX36, you can refer the the below tutorial from the help manual
RD-3595: Three Point Bending with HyperMesh
RD-3500: Tensile Test Setup using HyperMesh
Let me know if this helps!
Thanks,
Vikas
0 -
Thanks a lot Vikas.....
You cleared my doubt regarding the concepts, which in my opinion is rare on the forum.
I am gonna use MATX02 as it follows the Johnson-Cook Material Model and most of the papers have this model used for defining the plasticity behaviour of material.
Values of A, B and n : I am taking these from a paper in which they performed analysis on some porous structures of same Ti alloy.
Right now I am facing these problems: the 2 errors which usually show up
1. #error 4001: Running RADIOSS starter, that's it.....this is the description. And it shows quite quickly. (Disabled the anti-virus, some say it works, but not here) No .h3d file is made...basically no analysis done.
2. #error 4079: Non-convergent error....its not the whole description....but it shows up after some time.....so in the results , analysis is done upto only a few substeps, the result cant converge.
These two errors are bugging me on daily basis. Please tell what to usually check or do when these errors show up? And more important, what they actually mean.
Parmaeters I am using:
NLPARM: NINC=20
MAXITER=25
KSTEP(6) DEFAUT
CONV: UPW and Values of EPSU, EPSP, EPSW ALL BY DEFAULT 0.001
rest are also default.
I am guessing 2nd error has something to do with NLPARM. Absolutly no idea about 1st error. Please give suggestions. AND yes, if any more info regarding other parameters is required please do tell me.
THANKS A LOT FOR HELPING.
0 -
Hi Ashu,
Glad the answers helped you with the concepts.
I prefer looking at the .out files to suggest you further on the debugging steps.
Could you please send me the 3 .out files ( Starter,Engine and main)
Thanks
Vikas
0 -
PFA all the .out files
(There were 4 of them, i have attached all)
Actually i dont know about the starter, engine and main files.....i hope you can figure that out....
Thanks
0 -
Hi Ashu,
Could you please attach the input deck or the .hm file
Thanks
Vikas
0 -
OK,
PFA the .hm file of the model.
One request, since now you have the complete model, please just check out all the parameters we have taken and clarify whether we are doing it the right way or not, ONCE AND FOR ALL.
WE ARE PERFORMING - GEOMETRIC NON-LINEAR IMPLICIT STATIC ANALYSIS,
LOAD=1100N, MATERIAL= Ti-6Al-4V Grade 5 Alloy
Hyperworks 13.0 Full VERSION
and also, I wanted to plot the stress-strain curve for this strucutre......the stress and the strain(displacement in fact) values which I want to take on the plot are the peak values at each time-substep of the analysis. Can you tell me if it is possible to take out the tabular data with precisely these values so that i can export them to EXCEL. Direct plotting of GRAPH will also do, but i want it between these peak values of each substep. I tried through QUERY DATA in HyperView, but it required manual input of each time substep by selecting. Since, I will be making 10-12 structures more, I want to know if there is a direct method possible for this?
THANKS A LOT FOR GOING THROUGH ALL THIS TROUBLE. IT REALLY MEANS A LOT TO OUR GROUP.
0 -
Sure, I shall pass on this model internally within the team.
We shall get back to you shortly.
Thanks,
Vikas
0 -
Hello Vikas,
have you come up with any solution.
I am looking for it0 -
Hello @Ashu
Looks like you are running the analysis on cluster. Since analysis involves B2B coversion, RADIOSS is not started to proceed with analysis.
As a workaround, you can use .rad files to perform analysis. In the engine file add below commands and run the analysis.
/IMPL/SOLVER/7- Switches solver automatically based on performance
0