Quadratic tetrahedral stress projection

[kiakarim]

Dear FEAP community,

Using simple quadratic 10-node tetrahedral elements (v8.6), I do encounter oscillations for the stress projection (e.g. sigma_4).
Are there already any workarounds implemented to that problem?

Kind regards

Kian Karimian

[Prof. R.L. Taylor]

There is the option of using a larger number of quadrature points to force use of the local least squares projection, however, it also produced the same oscillations.

Using local projection schemes has been a problem for triangles and tetrahedrons.  Currently, we use a linear projection over the tet and average to get the mid node values.  The oscillations appear to be  occurring over the element spacing not the nodal ones.

[kiakarim]

Thank you,

in this regard, would you recommend implementing the superconvergent patch recovery SPR-method presented by Zienkiewicz et al.?
Further, on page 58. of pmanual, element averaging is described. Is this already implemented?
However, it does seem like the values oscillate within the element (see. attachment).