Hello,
I have a series of problems during implementing Pressure inlet (PI) and Outflow in Shan-Chen based LBM. When I use PI and pressure outlet(PO), there is no problem. The velocity inlet and outflow and velocity inlet and PO also work normally.
The problem with PI and outflow is that, after about 40 iterations, the density at the inlet rise up dramatically and divergence will occurs.
I have used MRT-LBM and C-S EOS. The inlet's density must maintain its amount at T/Tcr=0.8 on 0.28. Does anybody know what is the problem?
I will be appreciated if someone helps me.
dear Reza Goharimehr ,
There are many reasons which can result in the problem, and I want to know more information, as follows:
1. What is your boundary of the porous media?
2. Are the density and velocity at the boundary reset at each step of the calculation?
3. Did you use non-equilibrium extrapolation boundary in PI and PO?
Best wishes to you
Da Zhang
Hello, dear Da Zhang,
Thanks for your attention.
1-For porous media I have used an original full-way bounce back. weather I have walls or not, the divergence occurs.
2- Yes, at each step the values of all the directions will update. In fact, I force the density or pressure in each step.
3- I use Zhu and He (1997) non-equilibrium bounce-back at inlet and outlet.
Best regards.
Hello Reza Goharimehr
This problem could be due to many reasons. Let me ask you some questions:
1. How do you implement your force term? Do you update the equilibrium velocity (ueq) or you're using Guo's method? or EDM method? This problem could be due to your way of implementing the force term.
2. Which type of interaction force are you using? Are you using the original single pseudo potential? Or the doubly-attractive pseudo potentials?
3. Are you using the SC MCMP model for immiscible displacement?
4. Do you check your velocity values in each timestep? If the velocity values go higher than 0.2 lu/ts, you are out of Mach number limitation of LBM. In this case, you have to change different parameters in your dimensionless numbers.
Hello Rasa Soleimani
1- I use MRT-EDM based LBM.
2- The interaction force that I used is the standard one that proposed by Shan and Chen.
3- In this simulation, I have used SC SCMP LBM.
4- I tried both standard and CS EOS.
5- I had lots of problems with corner nodes although I had updated them by momentum equilibrium so I replaced the up and down wall BC by a periodic boundary condition.
6- The point that you mentioned about the velocities:
In my code, the velocities do not go higher than 0.2 Lu/ts although I can even run my code with 0.5 Lu/ts or higher.
I have never changed the dimensionless numbers in my code. Actually, I cannot understand what you mean of dimensionless number. The basic formulation of LBM or dimensionless numbers of fluid properties.
Best regards.
The fact that you're using CS EOS along with the MRT would decrease spurious currents on the interface and therefore, that's a very good choice that you made.
In this case, I'd suggest you to use the doubly attractive pseudo potentials for your interaction forces. This way, you'd have more stability and more control over the interfacial tension value as well. You could refer to paper by: Falcucci et al. (2010) or Sbragaglia et al. (2007).
About the dimensionless number:
I meant if your velocity values go higher than 0.2 lu/ts, for a specific Reynolds number you have to set a different value for your relaxation time or change the number of your grids. This problem happens a lot and would result in divergence.
- Badges
- Science topic
- Similar topics
- Thermodynamics
- Pressure