Problem with waveforms at finer time step Maxwell

  • Last Post 30 January 2020
Sun1234 posted this 16 December 2019

Dear all, I have designed a synchronous generator using rmxprt and from that I created a Maxwell 2D model...I performed transient analysis ....I used an external circuit created in Maxwell circuit editor...when I ran the generator at a time step of 0.002 sec for a duration of 2 seconds I got a nice sine wave...when I reduced the time step to 0.0001 sec for the same 2 second duration...I got a sine wave with wiggles... (high frequency sine wave superimposed on the actual sinewave)...I wanted a smooth sinewave. .can any one resolve this issue.. thanks in advance....

Order By: Standard | Newest | Votes
Sun1234 posted this 11 January 2020

the picture of the sine wave i got is shown below...what i am worried about is the wiggles that appear...i strongly feel it is due to some setting in the software which i have to do at finer time steps...or may be it is a limitation of the software

pblarsen posted this 30 January 2020

Hello.  Your question is difficult to diagnose without further information.  It looks like a ripple that is either due to a physical cogging process (does this match with the number of teeth?), or there could be a discretization error that is amplified by the model.  The two main types of discretization in these transient simulations are time-discretization and mesh-discretization (compounded with the nonlinear solution process).  Because you are using smaller time-steps, I don't think there is any time-discretization problem.  I would investigate the mesh and the nonlinear residual (and BH curves).  Particularly, I would look at the Band object meshing and use very small segments as a test case (maybe every 1deg in 2D, or every 5deg in 3D).  If the ripples change with that Band segmentation, then you can see what value is acceptable.  Same with the nonlinear residual, try using a very small value to see if it changes the result (maybe 1e-12 in one case), and then you can see what value is acceptable.  When you do the nonlinear residual test, make sure there are enough mesh elements to capture the nonlinear gradients in the magnetic cores.