exploding convergence with every boundary condition

 Register Blogs Members List Search Today's Posts Mark Forums Read

 June 26, 2023, 13:32 exploding convergence with every boundary condition #1 New Member   Lynn S. Join Date: Mar 2023 Posts: 4 Rep Power: 2 Dear friends of CFD-online, for a project, I am trying to convert a steady-state hotflow to a transient hotflow. Second-order discretization didn't meet the convergence requirements, so I switched to first-order to see the convergence potential. Now, in every simulation, the continuity, k and omega residuals explode at some point. I have tried different solvers (coupled and PISO) in combination with different turbulent boundary conditions (k-omega, intensity & viscosity, intensity & hydraulic diameter). Some explode later than others, but eventually, they all do. Do you have any suggestions on how to solve this problem? Thankyou so much in advance, Lynn

 June 26, 2023, 15:42 #2 Senior Member     Kareem Join Date: Nov 2022 Location: New York Posts: 101 Rep Power: 3 Divergence in transient simulations are usually the result of using too large of a time step or having a mesh with poor quality. If the same mesh/set-up solved well in steady state then I would suspect your timestep for the transient is too large. LynnCFD likes this. __________________ Please like the answer if it helped! Video Tutorials and Tips: https://www.youtube.com/@cfdkareem/featured

 June 27, 2023, 12:48 #3 New Member   Lynn S. Join Date: Mar 2023 Posts: 4 Rep Power: 2 Hi Kareem, thank you for your reply. I reduced the time step and up till now, it has not exploded. I was thinking, might a CFL-based (CFL of 0.8 for example) timestep be more efficient, rather than arbitrarily reducing the timestep? Thank you again, Lynn

June 28, 2023, 12:52
#4
Senior Member

Kareem
Join Date: Nov 2022
Location: New York
Posts: 101
Rep Power: 3
Quote:
 Originally Posted by LynnCFD Hi Kareem, thank you for your reply. I reduced the time step and up till now, it has not exploded. I was thinking, might a CFL-based (CFL of 0.8 for example) timestep be more efficient, rather than arbitrarily reducing the timestep? Thank you again, Lynn
Yes, trying to match your timestep to keep the CFL number under 1 is the most efficient technique. You can use the automatic time stepping which lets you set a CFL number and the timestep will update accordingly. This will allow the simulation to run at the highest possible timestep at each point and speed up your simulation time.
__________________

 June 29, 2023, 11:04 #5 New Member   Lynn S. Join Date: Mar 2023 Posts: 4 Rep Power: 2 Thank you again. The smaller fixed time (reduced with factor 4) has only delayed the explosion. It is always the continuity, k, and omega that explode. Before the explosion, the simulation converges. https://ibb.co/278KS8V Isn't it more likely that there is something unstable in the model itself, especially regarding the turbulence (based on the k and omega involved in the explosion)? I have tried different turbulence boundary settings, but they all explode at some point. Can a change in the viscous model resolve this? Or any other ideas? Thank you.

June 29, 2023, 13:15
#6
Senior Member

Kareem
Join Date: Nov 2022
Location: New York
Posts: 101
Rep Power: 3
Quote:
 Originally Posted by LynnCFD Thank you again. The smaller fixed time (reduced with factor 4) has only delayed the explosion. It is always the continuity, k, and omega that explode. Before the explosion, the simulation converges. https://ibb.co/278KS8V Isn't it more likely that there is something unstable in the model itself, especially regarding the turbulence (based on the k and omega involved in the explosion)? I have tried different turbulence boundary settings, but they all explode at some point. Can a change in the viscous model resolve this? Or any other ideas? Thank you.
It looks like almost all of your quantities are diverging. Could you give a bit more detail on what your model setup looks like? I see you are doing some species transport as well. The two things I would check is first that your mesh quality is okay and that it is fine enough for capturing the physics you are trying to solve. Second, I would double check your model and material property settings. Specifically with the species transport it is easy to make a mistake in the setup of material properties.
__________________

 Tags boundary conditions, convergence, explosion, turbulence