Showing results for 
Search instead for 
Did you mean: 

Negative eigenvalues in SOL 103

Valued Contributor
Valued Contributor

Dear all,

I am using NX Nastran 12 for modal analysis of a model containing contacts.


I followed the guidelines of this discussion, using a static pre-load (e.g. bolt preload) for establishing real contact behavior.


The mode shapes and resonance frequencies are so calculated but when I tried to extract the frequency response function (FRF) of the structure, a warning appears informing that negative eigenvalues occurred in op2 file and check the f06 file for further details, as depicted in the image. This messagge is shown both using a SOL 103 Response Dynamics and a SOL 103 Real Eigenvalues.

Reading the f06 file, however, I find nothing about negative eigenvalues.


Anyhow, the FRF appears but I do not understand if the graph depicted is correct or if it is affected by the problem of negative eigenvalues.


Moreover, when only using SOL103 Real Eigenvalues in particular, an additional messages appears with the following text:


"Solution103 displacment modal results are in the Basic Coordinate System.
This may cause errors with results. To generate modal results consistent for use with Response Dynamics, add the Nastran parameter OUGCORD to the input deck as: PARAM, OUGCORD, GLOBAL".


Reading this topic, an user had this last problem. As suggested, I added the parameter OUGCORD = GLOBAL in the parameter list (PARAM) but the abovementioned message appears again.


How can I fix these two error?






Re: Negative eigenvalues in SOL 103

Valued Contributor
Valued Contributor

I understood because the message concerning the negative eigenvalue appeared. It is due to the tolerance that is entered in the "New Response Dynamics" menù.

Let is take a practical example: Attached you can see the real eigenvalues (rad/s) calculated from a modal analysis. The first rigid mode (1.88e-03 Hz), for numerical resolution is negative but in the view results is shown positive.

In the NX help, Rigid body tolerance is defined as follow:

"Sets the frequency tolerance for determining rigid-body modes. All models have at least six rigid-body modes (three translational and three rotational). In the calculations, the software assumes that any mode below this frequency is a rigid-body mode."

However, entering a value of tolerance equal to 1E-03, the message relative negative eigenvalues appear whereas if we eneter a value of 1E-02, the warning message does not appear.

This is because tolerance is to be interpreted in absolute value (or that its numerical search for rigid modes continues even in the domain of negative frequencies). Attached a picture for a better understanding.
Kind regards,
Ivan GaspariniRealEigen_f06.PNGPost-Processing_results.PNGTolerance_Research.PNG