I'm a new user to Comsol, having used other FE and CFD packages and I'm having difficulty with getting a test model with natural convection to run successfully. I've followed the examples in the help file which work perfectly, but any model I've built myself either gives errors before the model even starts running ('Not a number' results in various nodes in the model) or fails to converge.
The test model is a heated aluminium block 30 mm x 20 mm x 1 mm high generating 1 W power, centered on the floor of a 100 mm edge cube filled with air (I know this can be done in 2D or as a quarter model, but the follow on model I want to run has no symmetry and will have the same overall boundary conditions, so getting this running would be most of the way towards the final model). The floor is insulated and has no airflow through it and the remaining boundaries of the air volume are open to flow (Walls are Inlets (I've tried with them as open boundaries too with even worse convergence), ceiling is outlet and outflow).
The results from the model that ran but remained unconverged show the whole model at a uniform temperature of 3e5 DegC, with the maximum air velocity at 5e-7 m/s (the directions of the velocity vectors show some convection movement), so it looks like the model is almost working but no energy is leaving the system and no air is flowing into or out of the system. This suggests that I'm almost definnitely using Inlet, outlet, outflow, periodic, open and wall boundaries incorrectly.
What I've found in the help files about these boundaries doesn't really help me to work out which is appropriate for my model. If there's a document that defines all these in terms of heat and flow that could well answer my questions.
The formulae I've put in for the volume force and initial temperature and pressure as per the example models.heat.circuit_board_nat_3d.pdf are all in with no errors returned from the program.
The test model is a heated aluminium block 30 mm x 20 mm x 1 mm high generating 1 W power, centered on the floor of a 100 mm edge cube filled with air (I know this can be done in 2D or as a quarter model, but the follow on model I want to run has no symmetry and will have the same overall boundary conditions, so getting this running would be most of the way towards the final model). The floor is insulated and has no airflow through it and the remaining boundaries of the air volume are open to flow (Walls are Inlets (I've tried with them as open boundaries too with even worse convergence), ceiling is outlet and outflow).
The results from the model that ran but remained unconverged show the whole model at a uniform temperature of 3e5 DegC, with the maximum air velocity at 5e-7 m/s (the directions of the velocity vectors show some convection movement), so it looks like the model is almost working but no energy is leaving the system and no air is flowing into or out of the system. This suggests that I'm almost definnitely using Inlet, outlet, outflow, periodic, open and wall boundaries incorrectly.
What I've found in the help files about these boundaries doesn't really help me to work out which is appropriate for my model. If there's a document that defines all these in terms of heat and flow that could well answer my questions.
The formulae I've put in for the volume force and initial temperature and pressure as per the example models.heat.circuit_board_nat_3d.pdf are all in with no errors returned from the program.