# Large Model Simulation

fatbmaxwell1
Member

in Electronics

Hi everyone

I want to simulate an Induction furnace and I am forced to simulate it at real world dimensions to investigate skin effect, proximity effect and so on (am I?). I don't no whether I can scale it down or not because for example if I scale my model what will happen to skin effect and penetration depth? On the other hand my computational resources are not adequate for real size simulation.

Now I wonder what is the solution for this problem and what do you offer.

really thanks

## Answers

If that is the case, can you not make certain assumptions and make the model 2D/2D- Axisymmetric? I believe, that it would help you achieve first cut results and based on the deviations from the experimental results, you can refine your model further. Considering your limited computational resources, I think this should be one of the possible plan of action. You can always go for a coarse mesh on the 3D case, but that would not be a wise option!

Thanks for helping Subhamdas

Can I ask what do you exactly mean by "certain assumptions"? I should say the model is roughly symmetric around a plane but cutting a simulation area (which is almost 8 m3 at the time) in half will not help much. I think I should find a way to scale down the simulation area (the object) at least a million times to let my PC have the analyze done. The only thing I wonder is that as the induction furnace works through induction effect and is closely bound to skin effect and son on, what will happen to the fields and currents in comparison to real world size. For example the thickness of the copper tube (from which the coil is constructed) is designed considering penetration depth and skin effect. If I scale it down to micron scale how this skin effect will act on that thickness? The problem I discussed can be applicable to various problems of the scaled model.

Now how can I approach this problem?

Take my case for example. I had to perform combustion modeling in a boiler tube, but the domain size was quite large, so I considered only a section of the tube where ignition is actually happening and simulated the case. For the boundary conditions, I ran a cold flow simulation over the entire domain, and supplied it to the reduced section. To further simplify the model, I had enabled the axisymmetric with swirl tab and considered a 2D section of the same. The aim was to compare the difference in results(in my case it was the temperature at different locations) obtained when the model is simplified. But in your case I see that your parameters of interest are different.

But are you sure that scaling down the model would reduce the mesh count? I doubt so.