Heated air at 190 deg F enters the room from opposite walls and exits out of the top of the room. A mass of material sits in the middle of the room. The ambient temperature outside of the room (and initial temperature inside it) is 80 deg F. I am trying to do a laminar steady state CHT analysis to show the temperature that the room and material will be maintained at. I am also incorporating the losses through the walls and floor. The simulation titled CHT-With-Wall-Flux is my attempt to model this.
I am getting a solver fatal error for maximum iterations exceeded when I try to run this. In the solver log I notice the minimum and maximum temperatures listed are impossibly high and low, and the solid enthalpy solver uses 1000 iterations. I have tried running this with zero gradient walls, lower inlet velocities, and different meshes and get similar errors each time. I have looked at similar CHT projects and haven’t been able to figure out what the problem may be. Any help is appreciated.
No luck. I reran “Refined_Try1” simulation with the outlet at 0 psi and got the same error. The solver log shows several lines for the solid enthalpy during time step one, mostly with “No Iterations 1000”, and the temperature for both fluid and solid regions blows up in the second time step. I’m not sure why so many lines for the solid region enthalpy are listed in the solver log either.
I’ve checked through your simulation and don’t think I’ve spotted any glaring issues. Compared it with other projects as well and everything so far checks out.
Sorry buddy but I don’t think I have much more to input. Hopefully the other powerusers with more experience in this will be able to help you out!
If anyone else has any ideas for this, please let me know. I still haven’t been able to figure out what is causing the max iterations exceeded error.
I would suggest to go with the very basic (even a quick run to only 100 iterations) setup, i.e. velocity/flow rate inlet and pressure outlet boundary condition with smaller temperature differences initially. This can help us eliminate mesh issues out of the consideration.
By default, the numerics condition are pretty strict. We can make it lenient by changing the relative tolerance values for all solvers (GAMG and Smooth) to 1e-2. This should help during the initial iterations and solve the problem.
I made a new simulation as you suggested. Under initial conditions, I set the default temperature at 67 F and made subdomains for the fluid and solid. I made the fluid region 67 F and the solid region 80 F. I set the inlets as velocity inlets at 67 F and the outlet is a pressure outlet at 14.7. I made the walls no slip / zero gradient. I also changed all of the relative tolerances to 0.01. The simulation run still failed due to maximum iterations exceeded. It made it to time step 3, which is farther than the previous runs. I still notice in the solver log that the fluid temperature range explodes in the second time step. Here is a section from the log:
Time = 2
Solving for fluid region solid_0
smoothSolver: Solving for Ux, Initial residual = 0.193054301571, Final residual = 0.000352077328641, No Iterations 3
smoothSolver: Solving for Uy, Initial residual = 0.198772106341, Final residual = 0.000342300454845, No Iterations 3
smoothSolver: Solving for Uz, Initial residual = 0.0614444122772, Final residual = 0.000101502240098, No Iterations 3
smoothSolver: Solving for h, Initial residual = 0.99999999983, Final residual = 0.00671447003664, No Iterations 5
Min/max T:-12474261098.1 22330384809.8
I have read that temperature blowing up like this can be caused by unreal boundary conditions, such as 0 pressure. I don’t see this in my model though.
I made a new project with metric units, and meshed and simulated this again just to see if it would make any difference. I got the same error. I also made the temperature of everything ambient, so there should be no heat transfer, and the max iterations error still occurred.
Maybe there is an issue with the geometry I’m not seeing? The geometry seems pretty simple - just a box within a box. The meshes I have tried state no errors in the mesh log and list the correct faces for interfaces. I’m not sure what else it could be besides geometry or meshing though.
I have an update concerning this problem. I made a bit of progress. I noticed that other similar CHT projects, including the tutorial, have geometries that are physically much smaller than what I am modeling. I scaled down my geometry by a factor of 0.05 and meshed it again.
The laminar steady state CHT run then finished when I used all zero gradient walls. When I add heat flux to the walls, I get the max iteration error after several time steps. I’m still working on that part, but at least the basic model will run if I scale it down. That makes me think the error was probably due to the mesh not being fine enough with the full scale model. I can’t test this though because the full scale geometry mesh would require too many elements.
I did make some very good progress. I was able to get the CHT analysis to run with heat flux through the walls and floor. I had to scale down the CAD geometry like I said above. I also had to further refine the mesh, especially around the solid region. I ran the analysis for 900 iterations. This took several hours with 32 cores. The analysis was certainly not yet finished. I would need to run this at many more iterations to get valid steady state results. That won’t be necessary though, as I was able to get the flow information I was looking for with a simpler convection analysis.
Sounds very good! Do you mind creating a post-processing documentation of your results and (if available) compare it with analytical calculations? Would be very interesting!
