Estimated bounds considering effect of discretization

[gurmeet2003]

Greetings to everybody,

I use ANSYS software for FEA. I have a question regarding interpretation of results. If I plot the results of the analysis (equivalent stress SEQV)of the compressor cylinder head mentioned in my last post, the maximum stress value SMX is 123000 psi. This run was made at twice the design load and has nonlinear material properties (BISO). However the plot also displays a value called SMXB and the value of same is 155000. If I list the value of Percent Engergy Norm it is only 7%. According to vendor literature a value less than 10% is acceptable. According to vendor literature SMXB represents ESTIMATED BOUND CONSIDERING THE EFFECT OF DISCRETIZATION ERROR.

I am wondering if those of you who use other FEA software have come across something similar.

I am also puzzled by the fact that if my mesh is ok (PERR 7%) why do I have such a high deviation (23%)between the predicted value and the estimated bound of stress.

One final question: Do I need to consider the Estimated bound for calculating Factor of Safety or can I ignore it and use 123000 psi?

Thanks,

Gurmeet