Extrapolating results available at Gauss or quadrature points to nodal points.

[ha.ba...@gmail.com]

Hi

Results are first calculated at certain locations, called Gauss or quadrature points, located inside each element. It is traditional to save results at nodal point (not quadrature points). Is there a function to extrapolate the calculated quadrature point values to nodes (such as stress values for an elastic solid model)? this function must have ability to calculate averaged values for a node common to several elements.

Thanks,

H.B

[Wolfgang Bangerth]

On 03/15/2016 12:07 PM, ha.ba...@gmail.com wrote:
>
> Results are first calculated at certain locations, called Gauss or quadrature
> points, located inside each element. It is traditional to save results at

> /nodal point (not quadrature points)./ Is there a function to extrapolate the
> calculated /quadrature point values/ to /nodes //(such as stress values for an

> elastic solid model)? this function must have ability to calculate averaged

> values for a node common to several elements./

There are some ideas and links to functions here:

https://dealii.org/developer/doxygen/deal.II/step_18.html#Refinementduringtimesteps

The approach taken there is to project the data from quadrature points to a
*discontinuous* element, so you don't have to do the averaging.

Best
W.

--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@math.tamu.edu
www: http://www.math.tamu.edu/~bangerth/

[ha.ba...@gmail.com]

Dear Professor Bangerth

Thank you for the response.

In fact I have a problem during adaptivity and the aforementioned method will also help me in mesh refinement. I am working on a coupled thermo-mechanical damage model. For history dependent constitutive material models, such as plasticity and damage models, it is necessary to transfer quadrature point data from old mesh to new one, because the new stress and other state variables are calculated based on the previous state.

Best Regards,

H. Badnava

[Denis Davydov]

On Tuesday, March 15, 2016 at 6:07:39 PM UTC+1, ha.ba...@gmail.com wrote:

Hi

Results are first calculated at certain locations, called Gauss or quadrature points, located inside each element. It is traditional to save results at nodal point (not quadrature points). Is there a function to extrapolate the calculated quadrature point values to nodes (such as stress values for an elastic solid model)? this function must have ability to calculate averaged values for a node common to several elements.

These questions were also already discussed here, please see https://groups.google.com/forum/?fromgroups#!searchin/dealii/transfer$20quadrature$20point$20data/dealii/OSaxZJR19W8/500-t0xg0ZsJ

and also some related corner cases related to distributed Tria: 

https://groups.google.com/forum/?fromgroups#!searchin/dealii/transfer$20quadrature$20point$20data/dealii/K4aenk1Nc3A/t3qfUnFDUekJ

Regards,

Denis.