I mentioned this in the general forum but I think it belongs here. Currently, if you need to specify a non-linear elasto-plastic stress-strain curve in Nastran, you have the MATS1 and MATS2 formats available. Neither of these is very convenient if you have tensile test data available. MATS1 works for a bilinear stress-strain model, but if one is interested in the behavior close to the yield point, or if the region between the linear parts of the bilinear model is large, a bi-linear model that intersects at the initial yield point may not be appropriate. Moreover, when entering points into TABLES1, the MATS1 convention dictates that the first point must be the origin and the second point must be the point of initial yield. Most of the time, it doesn't really matter... but it's also not strictly correct. Initial yield is generally defined in terms of a strain offset. A straight line from the origin to the yield point is at the secant slope, but that's not the Young's modulus slope. Yes, it's a nit-pick in most cases, but still... why not make it correct? MATS2 enables one to enter the Ramberg coefficients. This is good, but it seems circuitous and unnecessary to have to compute the Ramberg coefficients to approximate a tensile test curve if one is looking at the actual tensile test curve. My suggestion is that a format (MATS3 perhaps?) be defined that allows direct input of points pulled off the stress-strain curve, along with the method of determining yield point such as 0.1% offset, 0.2% ofset. 0.5% extension under load, etc. , I'm pretty sure that Ansys has exactly this; so should we. It would save time. It's rare that tensile test data actually identifies the onset of plastic deformation, but in case one does have that information, one should be able to put in on the 'MATS3' stateemnt directly. An even better feature would be the ability to put in curve points and yield criteria from tensile test curves in all three axes into TABLESi. Orthotropic properties are very important! I don't think Ansys has this. It would also be very nice to be able to put in some detail about lateral strain, which can easily be obtained with a micrometer during a tensile test, and have Nastran compute Poisson's. I recommend that this data not be required at every point. The user should be able to put in whatever he has, and Nastran can interpolate. Poisson's shouldn't change much in most cases. The Nastran output should be able to give some useful material data that it can calculate from the curve(s), such as Young's modulus, proportional limit, yield point computed based on 0.1% offset, 0.2% offset and 0.5% extension under load, Poissons perhaps, etc. A lot of this might not have anything to do with the current analysis, but it would be useful for the analyst to have.
Show More