Learning Forum

[alavello]

Hi. I want to simulate 30 sec of the transient heating of a flat plate. The plate is heated by a copper inductor that carries current at 20 kHz. The inductor travels at constant speed of 5 mm/s over the plate’s surface. I know that Maxwell transient simulation allows for motion, but the simulation will use time steps smaller than 50 microseconds (period of a 20 kHz signal). The number of time steps needed to complete a 30 sec simulation is enormous, so the simulation time becomes impractical. 

The logical approach to obtain the heat load on the plate would be to use Eddy current simulations, but Eddy current has no motion capabilities, so it is not an option either. Any suggestion on how to deal with this problem in Ansys?

[Timos]

Hello, the approach you choose will depend on the trade-offs you are willing to make between accuracy and computational cost. At first I would suggest to use HPC options ( Parallel Computing ). Another approach would be to do some approximations to the model based on the desired results. You can perform multiple steady-state simulations with different positions of the inductor relative to the plate (parametric).

[alavello]

Hello. Thanks for your reply. Using HPC is not an option for me because I only have access to a quite limited number of cores. Moreover, this approach is very inefficient and, in my opinion, not reasonable for this kind of simulation.  

Steady state solutions are fine for the electromagnetic field (like in Eddy current simulation), but it is not appropriate for the heat transient solution, which has long settling times. For the thermal part, using a transient solver is a must. 

I am trying the following approach:

1. Set up a parametric model using the motion of the inductor X as parameter. 
2. For X=0, solve Eddy Current simulation and calculate heat losses.
3. Transfer the heat loads to Transient Thermal and solve Transient Thermal between t = 0 s to t = 1 s, keeping constant the heat loads. 
   
4. Modify the value of X (which updates the geometry), find new heat loads using Maxwell Eddy Current, and solve transient between t=1 s and t=2 s, taking the solution of step 3 as initial conditions. 
5. Repeat step 4 until the end of motion. 

When I try to connect Maxwell heat loads to Transient Thermal, I get the error message “No field solution for frequency 25000 Hz”. The pictures below show my Workbench setup, the Transient Thermal model tree and the error message. I wonder if something is wrong with my approach.