FE displacments normal to the local surface - HOW TO GET?

[bzzh8c]

Greetings fellow square and triangle slingers...

I was approached to help develop a tool that would post
process finite element displacement data and return back
only that portion of each grid's movement that was normal
to its local surface. (Emmitted noise being the goal).

Any knowledge of a C or FORTRAN routine (or commercial package) to do this existing already?

Todd

 

[brad]

If your FEA solver outputs nodal normals, just input the normal vector for each node and the displacement of each node.

The dot product of these vectors is the normal displacement for the node.

Once the Inputs are sorted out, the calculations are pretty straightforward.

Brad

 

[bzzh8c]

Thanks Brad. Our latest thinking was to average
the normals of the elements attached to the grid
in order to estimate the normal at the grid location.
This avoids a nasty surface fit to the local area which
might give a marginally better answer. We were just
wondering if there was a function out there in some
software that did this already that we could leverage.
We've got the dot product thing working already using
just an arbitrary vector and now we just want to plug
in the local normal calculation.

Todd