It’s most likely due to extrapolation, maybe you can adjust some results presentation settings in this software.

Just in case, make sure that other output variables are correct.

@FEA way, I'm studying the Kalnin's twice yield method. Since the max alternating stress range is 34 ksi, while the actual yield for SA516-70 is ~38 ksi, is it appropriate to neglect these erroneous values of equivalent plastic strain and assume that the response is wholly elastic?

The calculation of equivalent or effective plastic strain includes squaring the plastic strain (or strain rate) terms. So negative values shouldn't be mathematically possible. Maybe your software has a special meaning for negative values. If not, I would say there's some error and I wouldn't trust the results. I don't like to neglect erroneous values unless I understand where they came from and can convince myself that they don't have any effect on the rest of the answer. Maybe someone with experience using Nastran has more insight.

Maybe mesh refinement will help, it often does in such cases. And yeah, it’s better not to ignore results like that.

"it’s better not to ignore results like that" ... they could be a sign of other problems with the model.

another day in paradise, or is paradise one day closer ?

Ok, I figured it out, Inventor Nastran can only spit out the centroidal value of equivalent plastic strain. Strangely enough, its the only parameter where the software does not store nodal values.

why is a centroid value any different from a nodal value ?

why does -ve plastic strain make sense as a centroidal value ?

does -ve plastic strain mean compression ?

another day in paradise, or is paradise one day closer ?

rb1957,

A centroidal value should be close to a corner nodal value, if a mesh is correctly refined and mesh convergence is verified. However, the codes usually call for nodal values at corners?

It's the first time I'm seeing a negative value of equivalent plastic strain, albeit it is of the order 10^-6. The two explanations for this are: Extrapolation from gauss points inward toward the centroid OR Mathematically, ε_p is positive, like a von Mises value. I suppose that a negative value indicates an elastic response where the stress state lies on/within the VM yield surface.

Does it mean compression? Probably not, as ε_p is a single positive scalar value.

rb1957,

Additionally, Inventor requires the entire stress strain curve (elastic + plastic) . When it comes to evaluating plasticity, this can create problems with smoothening of the curve in the vicinity of the proportional limit.