Learning Forum

[justin.nekota2]

Hi all,

I am attempting to extract reaction forces and moments at named selections during random vibration analysis. I am mostly following the method outlined here:

https://www.padtinc.com/2011/06/15/retrieving-accurate-psd-reaction-forces-in-ansys-mechanical/

My problem is that the forces and moments are delivered in the global coordinate system, and since this is a random analysis, they are always positive. So I cannot use a typical transformation matrix to transfer them into another coordinate system. Is there another method I can use to do this?

I have also experimented with using NROTAT to rotate the nodes of interest and then using *GET to extract the force at each node, but I can't extract moments this way with a model composed of 3D elements.

Thanks for your assistance.

[danielshaw]

The PADT article appears to be out of date.  You no longer need to use MAPDL commands to extract the reaction forces from a random vibration analysis.  They are available in the Mechanical UI.   Using *GET and FSUM is still valid, but no longer required.

The reactions are reported in the nodal coordinate system.  You can use NROTAT to rotate the nodal coordinates from global cartesian to a local coordinate system before the solution, but it is not mathematically valid to convert the reactions into another coordinate system after the solution.

[justin.nekota2]

I did see force reaction available as an output but only for boundary conditions and springs. The nodes I am interested in pulling forces on are either attached to beams or shared topology between components. Even with rotating the nodes and using *GET, there doesn't appear to be a way to calculate moments about the local coordinate system.

[dlooman]

You're right, and the only possible way is to rotate the nodes.   Perhaps you could use FSUM to retrieve forces and moments.  Not sure how it is affected by rotated nodes though.

[dlooman]

In a commands object you could issue FSUM,RSYS as shown below.  It would be good to test it on a small model.  I've never used the RSYS option on PSD results myself.

/POST1

SET,3,1

RSYS,12    ! local coordinate system

FSUM,RSYS

[justin.nekota2]

This is exactly what I am trying to accomplish, unfortunately RSYS does not apply for PSD analyses according to the FSUM documentation. I have been partway successful using the SPOINT command to establish a moment summation point prior to FSUM, but the summed values are always delivered in the global coordinate system.

[dlooman]

Sorry.  As I said, I never tried it myself.  The way FSUM is performed for a PSD it actually could support RSYS, but apparently that wasn't enabled.  For a spectrum analysis the FSUM is performed on each mode separately and the modal fsum values are recombined.  I found that FSUM moments aren't in the local coordinate system even if you rotate the node, so that's not a workaround.  You could get the moments in a local coordinate system for an individual mode using RSYS and FSUM,RSYS and a mode factor equal to the square root of the variance.