[msatrustegui]

Good morning to everyone,

I am dealing with a problem which consisted on a solid with a water jacket and all that system surrounded by air. This solid has some losses and it´s cooled by the water jacket.

When it comes to modelling, I´m having problems with energy convergence , the simulation starts well but after few iterations (10-20) the energy starts diverging and cells started to get limit temperatures (5e3) (URF of Energy = 0.2).

I have tried two type of Pressure-Velocity Coupling (SIMPLE and Coupled), but only Coupled seems to work with this type of simulation.

I have also tried to stabilize the simulation by decreasing the URF, but I don´t know if decreasing Energy to 0,1 for example it´s a good idea.

The mesh has skewness under 0,94 but the geometry is so complex that if i reduce the skew, the number of cells gets very high.

Could you please help me with it? Thank´s in advance.

[fluid23]

I am not familiar with Fluent, but I know in Star-CCM it is possible to isolate and display cells based on residual values. If possible, I would try to veryify that your highly skewed cells are not the ones with high energy residuals. If they are, then you might have to improve your mesh quality.

[ugli91]

You should check your boundary conditions. Problem is not about solving method. If you post your geometry picture and boundary conditions values, we can understand problem clearly.

[msatrustegui]

I've tried to simulate the model with all the loss sources set to zero and also al the backflow temperatures to the same temperature, so the energy equation could be easily solved, but it also diverge.

Could it be a problem of boundary conditions? I've stablished a periodic boundary condition, but it does not seem to be the problem. When I take a look on the results, one of the solids is the one with all the temperatures gradients (between -200 a 4000 degrees), but the rest of the model remains with constant temperature.

I think it would be impossible to upload a photo of the model because of confidentiality (Sorry for that).

[vasava]

You have a conjugate heat transfer problem and I think it is completely ok to use coupled approach. Although it is more resource-expensive but it provides more stable solution.

[msatrustegui]

May it be a problem yo have periodic boundary conditions to get energy convergence?

The thing is that I´m modelling an electric motor and to reduce the number of elements I´m only simulating one pole of the machine, having cylindrical periodic boundaries.

Thank you for all the answers.

[vasava]

Here is what I would do to handle the limited temperature warning.

1. Set the temperature limits a range that you expect with Solution Controls --> Limits
2. Run the simulation until you get the limited temperature warning
3. Mark the mesh where the temperature is outside the expected range with iso-value adaption (Adapt-->Iso-Value Adaption).
4. Visualize the 'bad' mesh elements. Click Manage button (this opens a new window). Click Display.

You can change the iso-value range as per your requirement and adapt the mesh.

If your case involves flow then solve the flow equations first. Once you obtain good enough solution then solve the energy equation.

[msatrustegui]

I could improve the mesh to get skewness below 0,8.

When I run the simulation without energy equation, I got residuals stabilized at second order below 1e-3, except from continuity which stablizes at 3e-3. Then, I run the simulation with the energy equation and with small values of energy sources so it can converge easily, but after 3 iterations it starts with temperature limits in some cells (as the energy residual starts growing).

Do you think this must be a boundary problem?

[vasava]

Is it possible for you to upload your case some where? Perhaps it would be more easier.

Yes, there could be problem with boundary condition but without the case it is difficult to make any comments.