unsteady scalar diffusion stopping before steady state
Posted: Thu Jun 21, 2018 6:12 pm
Hello, Yvan, I move my previous post here as sugested by you. I attached the setup programmes. My mesh is msh and it can't be uploaded to the forum apparently. It is a structured uniform 1000 hexa elements mesh along 10 mm in x. In the BC programme uploaded, concetration is set at 1e3. If you change that value to 1, you will find that the solver stops as if it had reached stationnary, far too soon.
I was making some test because I want to perform some calculations of mixing in small channels and I found the following result:
Imagine a unidimensional diffusion problem say a structured mesh on x direction of 10 mm, with 1000 elements. On one end I set a dirichlet condition for my salar, and at the other end I set homogeneus Neuman. Velocity field is frozen at 0.
I set the diffusion coefficient at 5e(-10)*rho (with rho=998 kg/m3).
If I set the boundary condition with a scalar value of 1 (a concetration of 1/kg) the calculation stops way before the concentration spreads along x.
If I set the boundary condition with a scalar value of 1e3 (a concetration of 1e3/kg) the calculation reaches steady state OK.
If I set the boundary condition with a scalar value of 1e6 (a concetration of 1e6/kg) the calculation reaches steady state OK.
If I plot the concentration profile divided by the scalar value imposed (1e3 and 1e6 respectively, both profiles are superimposed at all time steps. The profile for a scalar value of 1, is also superimposed with the other two but only for the first time steps and then it frozes while the other two continue diffusing properly.
I guess there is probably a code variable that tells the code that steady state has been reached and it stops calculating but I don't know which is that variable, which is its default value etc.
Could you please help me with this issue?
Thanks
I was making some test because I want to perform some calculations of mixing in small channels and I found the following result:
Imagine a unidimensional diffusion problem say a structured mesh on x direction of 10 mm, with 1000 elements. On one end I set a dirichlet condition for my salar, and at the other end I set homogeneus Neuman. Velocity field is frozen at 0.
I set the diffusion coefficient at 5e(-10)*rho (with rho=998 kg/m3).
If I set the boundary condition with a scalar value of 1 (a concetration of 1/kg) the calculation stops way before the concentration spreads along x.
If I set the boundary condition with a scalar value of 1e3 (a concetration of 1e3/kg) the calculation reaches steady state OK.
If I set the boundary condition with a scalar value of 1e6 (a concetration of 1e6/kg) the calculation reaches steady state OK.
If I plot the concentration profile divided by the scalar value imposed (1e3 and 1e6 respectively, both profiles are superimposed at all time steps. The profile for a scalar value of 1, is also superimposed with the other two but only for the first time steps and then it frozes while the other two continue diffusing properly.
I guess there is probably a code variable that tells the code that steady state has been reached and it stops calculating but I don't know which is that variable, which is its default value etc.
Could you please help me with this issue?
Thanks