Hello,
I am currently working on an electromagnetic simulation using a transient magnetic analysis with thermal coupling. I am trying to obtain simulation results for a total simulated time of 60–300 seconds, using a time step of 0.001 s and a write interval of 1000 time steps.
The simulation runs successfully for shorter simulation times (for example, 0.05 s) and with smaller write intervals. However, when I increase the total simulated time to 60 s or more, an unknown error occurs and the simulation terminates unexpectedly. Interestingly, the simulation consistently stops at approximately 4 seconds of simulated time.
I would greatly appreciate your assistance in identifying the cause of this issue and finding a suitable solution. In particular, I would like to understand whether the problem could be related to the time-stepping settings, write interval, memory requirements, solver settings, or the thermal-electromagnetic coupling.
If required, I can provide the simulation setup, solver configuration, error message/log file, or relevant screenshots for further investigation.
Thank you in advance for your support.
Best regards,
Aziz
Hi,
The issue is related to the timestep definition itself. If you have e.g. a 300 seconds end time with a timestep of 0.001 seconds, this gives you a total of 300000 steps to calculate.
Even though the mesh is a bit coarse, 300000 steps is still quite a bit, and it will take some time.
Suggestion: think about the transient physical behavior that you are looking to see. How long does it take to develop? How long does a full cycle takes to happen?
With those in mind, set up appropriate end time and timestep values. Just as an example, in pump CFD studies with sliding meshes, a typical starting point is an end time = 3 full rotations of the pump and a timestep size equivalent to less than 3 degrees of rotation.
Hi,
Thank you for your reply.
The issue I am facing is that I would like to observe the temperature evolution in my system over a period of approximately 2–3 minutes. However, my excitation signal has a frequency of 50 Hz. Therefore, using a larger time step may cause the simulation to lose the characteristics of the excitation signal and produce physically inconsistent or illogical results. I have already tested this approach and observed such behavior.
What I am therefore looking for is a way to extend the simulation to the desired time period while still maintaining a sufficiently small time step to accurately capture the 50 Hz excitation signal.
For context, I have included a link to my project below:
In this project, my goal is to capture the evolution of the magnetic field, magnetic flux, and temperature in the plate over the required time period.
Here is a link to the project :
https://www.simscale.com/workbench/?pid=4807844169153574994&mi=spec:c71fd748-f997-4ba9-9119-23b681848f4e%2Cservice:SIMULATION%2Cstrategy:4
Could you please advise me on the best way to set up the simulation so that I can achieve the required 2–3 minute simulation time without losing the characteristics of the 50 Hz excitation?
As described in the previous response, the issue arises when you have two very different time scales in the same simulation. Attempting to solve a 50 Hz signal from the Emag side of things and, at the same time, solve another thermal factor that takes minutes of transient runtime in the same study will lead you to thousands of steps.
In short, my suggestion is to split this study in 2: one focusing on the higher frequency aspect (i.e. Emag). You can also gain an understanding as far as heat sources go from this project.
The heat source information would then be taken to a second study, now focusing on the thermal side of things, probably with a conjugate heat transfer study. The idea here is to push the timestep further so you can actually solve the few minutes of transient runtime in a timely manner.
If you attempt to have the two things done in the same simulation, I foresee you running into problem, because you have two time scales that are several orders of magnitude apart.
I hope this helps.
