ANSYS modelling question

[mgage]

I am currently modelling a metallic seal made of 316l stainless steel. The outputs I'm most concerned with are the load to compress the seal 0.01" and the resulting springback after unloading. I've tried to match physical testing data without much luck. The biso and miso hardening rules overpredict load and greatly underpredict springback. The bkin and mkin models overpredict load as well as springback. I've also looked at the chaboche model but have no idea where to find the material constants for such a model. I was wondering if anyone has run into this problem and/or has any other modelling ideas.

 

[brad]

You state that it overpredicts the load and underpredicts the springback. I just want to clarify--by 'underpredict the springback', do you instead mean that it underpredicts the permanent set? What is your tested permanent deflection, and what is your predicted permanent deflection?

My reason for the request for clarification is the following--an overprediction of the load would be consistent with a particular error; that same error would also result in an underprediction of permanent set.

I will overclarify just to make sure that we are using the same terms--
Let's say my load point has initial position of x0=0. I deflect it to position x1=0.01. It then 'springs back' to position x2=0.002. In this case, my permanent deflection is 0.002 (x2-x0), and my springback is 0.008 (x1-x2).

Now-what may be happening--frequently when people describe elastic-plastic material models, they describe them directly from test data, rather than converting the test data into equivalent mechanics data (nominal stress-strain in the former, true stress-plastic strain in the latter). If your test data for this material was not converted into a correct format, this would typically result in an overly-stiff material definition. In your situation, this would manifest itself as a higher load than is correct, and a lower permanent than is correct. Frankly, I never trust material data unless I know personally the source of that data (and I trust their level of competence). My first check for an error source would be this.

Brad

 

[mgage]

In our testing, the seals are compressed to a height of 0.051". The load is removed and the final height of the seal is 0.0535", giving a springback of 0.0025". The load to get this compression was 330lbs.

I was given true stress-true strain data for my model. My biso model predicts a load of 500lbs and springback of 0.0002", while the miso model gives 800lbs and 0.0003". I also tried the mkin model which gave 449lbs and springback of 0.0022", however the deformed shape does not reflect the actual deformed shape.

 

[brad]

This could be indicative of stiffer material properties being used than those used in the test sample. Even if it is the same material on paper, and your test data is correct, they could still be off due to differences in processing, forming, etc. between your material test sample (for the stress-strain data) and your gasket material.
A test which may help determine this--slightly modify the properties in your analysis for this material--lower the yield stress to 90% of that which you are currently using, and lower the post-yield stresses to 90% also. Re-run the analysis, and verify whether the changes in results are directionally correct.

Another thought--are you using nonlinear geometry in your analysis. This could also have a significant effect on these results.

Brad

 

[mgage]

Thanks for the tip. I actually took it a step further and reduced the yield stress and the stresses along the curve by 75 percent. The results matched well with the physical data. The test now is wether the model predicts the correct deformed shape (which I haven't seen yet) and if it can predict load and recovery when they make changes to the seal.

thanks for your help.

 

[brad]

mgage-
Glad to be of help. Let me know how it ultimately turns out, if you get a chance.
Brad