** Error**

I'm carrying out the electromagnetic transient analysis of a magnetic damper which is installed on the high speed compressor on ANSYS APDL . While doing the analysis an error occurs stated as -

A large negative pivot value ( -1239682.55 ) has been encountered in the global assembled matrix. This may be caused by a bad temperature-dependent material property used in the model.

Please give your suggestion in order to remove this error. It is meshing issue or analysis issue. Model of dampers is, a conducting plate (copper) is installed at the end of the rotor of compressor and halbach permanent magnet setup magnetically suspended against to it (stator). Rotor rotates at 11280 rpm (188Hz)


  • wrbulatwrbulat Posts: 73Ansys Employee


    Are you using

    1) new technology PLANE233/SOLID226/227 element types or

    2) legacy PLAN53/SOLID97 element types?

    If you are using option (1), then, in the case of a 3D model with SOLID236/237 element types, you must ensure that the mesh is conformal. That is, the nodes on interfaces between bodies must be shared by the elements on both side of the interfaces. There is no companion contact element for SOLID236/237 that can be used to create a "magnetic bond" between the bodies. If you have not done this, that might explain the large negative pivot error.

    Another possibility might be not constraining the VOLT DOF on at least one node in the Cu conductor mesh.

    It isn't clear to me from your description what the configuration is. I would think that you should to have both the Halbach array and the Cu plate attached to either the rotor or the stator, but not one on the rotor and the other on the stator. I say this because if the Cu plate is on the rotating rotor and the PMs are on the stationary stator, the Lorentz forces attributable to the eddy currents induced in Cu plate will act as a brake, inhibiting the rotation of the rotor (such as in the hypothetical system depicted in is presentation):

    If your objective is to simulate vibration damping due to motion effects, maybe you should use coupled field SOLID226/227 with keyopt(1)=10001 for non-conducting regions and keyopt(1)=10101 for conducting regions. These will automatically couple structural, magnetic, and electric DOFs and may be used in full transient analyses. Lorentz forces associated with the eddy currents induced by the relative motion between the PMs and the Cu plate will automatically be transferred to the calculation of the structural response, allowing you to observe the reduction in motion. The coupling is bi-directional... equilibrium iterations are performed until results of all three field calculations are self-consistent.

    Kind regards,


  • D_DD_D Posts: 22Member


    Thank You Bill for your guidance!

    Yours Sincerely

Sign In or Register to comment.