I am frustrated all week with this problem. It appears that there is some obvious mistake in my model but I just dont seem to detect it - the model runs for about 3-4 days of simulation time with time steps of 360 seconds. But then continuity residual loses its convergence and keeps oscilating around 0.9 to 12

Model is for gas diffusion problem in stored grain - Stored grain is modeled as porous medium. (viscous resistance: 2.857143e8 inertial resistance: 20620.81 porosity: 0.38 )

Following are the details of my model

1. 2D, axisymmetric, unsteady 2. species transport model - (CO2 + O2 + N2) 3. Gravity is ON - in the x direction -9.81 m/s2 4. Ideal gas is used for density formulation for all the gases and mixture template 5. I use k epsilon model with default parameters for the headspace of the tank. LAminar option is turned ON for the porous zone - Since velocities in the porous zone are very small of the order of e-7, flow there can not be turbulent. 6. since all the elements in my model are quad, I use PRESTO scheme, all under relax and other schemes are kept at default. 7. While creating the geometry I use split face method to create the interface between grain (porous zone) and headspace - As a result the axis of my model is also split - Thus I have to name to different axis for the model - They are the same line but split - I am sure there is nothing wrong there since my model runs for 3-4 days before losing its convergence.

please help me - If you would like to take a look at my case file, please let me know and I will email you.

I do not have any specific idea which could be a solution to the problem. But I want to comment on two points. First, I never come across to the idea that if the elements are quad PRESTO should be chosen. I know that a choice of any of them can affect greatly convergency. I suggest you to choose all the default schemes at first and try different of them if the default failed to converge. Second, the time step is too large in my opinion. How about trying smaller time steps in the order of 0.0001 s.

Good Luck

solomon:

thanks for your note. I will try the default schemes and let you know -- I will also try to reduce time step size.

Thanks, Co2