Nodal stress, integral point stress or centroid stress for analysis report

[Cc huang]

Hello,

I’m a junior FEA engineer working for phone products. Most of the time, I do dynamic simulations like drop by Abaqus and extract peak stress of parts like display and judge failure risk.

Speaking to extract peak stress, some of my colleagues choose max nodal stress in report, others may choose max value of integral point stress or centroid stress.

My team lead said it’s ok to choose each of them if you think the result is reasonable.
It really confused me and would like to get some suggestions.

PS. I know that stress on the integral point is most accurate. Would like to know if there is a standard in the industry?

 

[FEA way]

Take stresses from integration points if you want the best accuracy. Values in other locations are obtained by extrapolation from integation points - see the documentation chapter "Understanding how results are computed". MISESMAX output variable can be useful here, especially if you use shell elements in your models.

 

[Cc huang]

@FEA way

Thanks you so much for the answer. I have an additional question would like to discuss.

I’m thinking if stress on the integration points are most accurate, why some analysts still choose nodal stress or centroid stress as report output.

Besides, Abaqus MAX/MIN limits in contour is presented as nodal stress. Therefore, most of the time, I started finding the time frame with max stress, snapshot the contour and record the peak stress as max stress showed in Abaqus(It’s nodal stress!) 80% of my colleagues did like that.
If choosing integration point stress, I will need to check elements around and find the max integration point stress manually.

In your opinion, do you think it’s unprofessional for a FEA analyst to observe peak stress by nodal stress?
The question might be stupid but has bothered me for a long time. Would be really grateful if you can share your thoughts.

 

[FEA way]

Nodal stresses are useful for the first assessmet of the results and for contour plots. But after that you should turn off averaging and check stresses at the integration points, especially when examining stress concentrations and trying to find their causes. The differences between nodal and integration point stresses can be really high.