Dear Elmer users,
Coupling of WhitneyAVHarmonicSolver and HeatSolver works fine for simulation of induction heating with a single frequency. In the next step we plan to investigate induction heating under coil currents with higher harmonics, for example signals built by fourier synthesis (like sawtooth, tri- or rectangular pulses) or any other periodic signals containing multiple frequencies.
My first idea would have been using (transient) WhitneyAVSolver, but I have no idea how to deal with the different and independent time scales and also the calculation of time-averaged joule heat.
Another idea is to superpose multiple harmonic solutions with different frequencies in each time step (how to?).
Do you have any suggestions how this can be done using Elmer with either first, second or any other approach?
Any help is kindly appreciated!
Best regards,
Benjamin
Induction heating with higher harmonics
Induction heating with higher harmonics
- Attachments
-
- indHeat_40kHz.zip
- (509.67 KiB) Downloaded 286 times
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Induction heating with higher harmonics
Hi Benjamin
Nice animation!
There is a solver called FourierLoss "Loss estimation using the Fourier series" currently as Ch. 33 in Models Manual. The limitation for this one is that the wake is assumed to be harmonic and the solver then is able to use experimental loss data for higher harmonics of a nonlinear system.
I guess in your case you don't need any experimental models as you have a massive piece of metal? Hence the Joule losses can be readily computed by the CalcFields solver. There is also the FilterTimeSeries module as Ch. 57 that you could use to time-average the computed Joule field. Now the handicap here is that probably you would like to everage the data long enough (at least over one cycle) before going to heat solver? Unfortunately the timestepping strategies of Elmer modules are currently global. You could perhaps use the Solver-specific "Timestep scale" and "Exec Condition" to overcome this limitation.
I also remember the 3rd alternative having been done at least once to combine harmonic syperposition and transient heat equation. I think it involved "scanning modes" for different frequencies and some tailored code to sum them up for the heat source. This would probably require some coding.
So yes everything is possible just tons of details....
-Peter
Nice animation!
There is a solver called FourierLoss "Loss estimation using the Fourier series" currently as Ch. 33 in Models Manual. The limitation for this one is that the wake is assumed to be harmonic and the solver then is able to use experimental loss data for higher harmonics of a nonlinear system.
I guess in your case you don't need any experimental models as you have a massive piece of metal? Hence the Joule losses can be readily computed by the CalcFields solver. There is also the FilterTimeSeries module as Ch. 57 that you could use to time-average the computed Joule field. Now the handicap here is that probably you would like to everage the data long enough (at least over one cycle) before going to heat solver? Unfortunately the timestepping strategies of Elmer modules are currently global. You could perhaps use the Solver-specific "Timestep scale" and "Exec Condition" to overcome this limitation.
I also remember the 3rd alternative having been done at least once to combine harmonic syperposition and transient heat equation. I think it involved "scanning modes" for different frequencies and some tailored code to sum them up for the heat source. This would probably require some coding.
So yes everything is possible just tons of details....
-Peter