CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Non-stopping Capillary Rise (https://www.cfd-online.com/Forums/fluent/228222-non-stopping-capillary-rise.html)

 Fuzzy June 23, 2020 14:22

Non-stopping Capillary Rise

Hello there!

I am currently simulating a capillary rise within a cylindrical tube of 1 mm diameter. I had already managed to setup the model and get the simulation to run, but the problem is that the rise of water level simulated have surpassed the maximum theoretical rise that could occur by a big margin. Moreover, the rate of water rise in the simulation doesn't seem to decline (so it is still ever rising).

Here are the details of my Fluent simulation:
* General - pressure based, absolute velocity, transient, 2d axisymmetric with gravitational acceleration turned on
* Model - Laminar, multiphase VOF consisting of air (primary phase) and water (secondary phase). Courant = 0.25 and volume fraction cutoff = 10^-6. Surface tension modelling and wall adhesion is turned on.
* Boundary condition - pressure inlet and pressure outlet (both have 0 gauge pressure). Water volume fraction of 1 at inlet. Backflow volume fraction of 0 at outlet. Already defined the walls contact angle. Axis is present too.
* Method - PISO
* Residuals - set the absolute criteria to 10^-6
* Initialization - initial water volume fraction of all zones set to 0. Patched the cells near inlet to have some water present.
* Step size has been defined by myself with 100 max iter per time step.

Additionally, in each time step, the scaled residuals observed is still rapidly declining (hasn't flatten out) when the iteration jumps to the next time step. Would anyone kindly help me with this?

Thank you!

 vinerm June 23, 2020 16:09

Setup

What value are you using for operating density? If you have not set it, set it to density of the lighter phase.

Secondly, if there are inlet and outlet, where does the question of capillary rise come from in a 2D axisymmetric scenario? Could you share a picture of the domain?

Do not use more than 40 iterations per time-step. If the simulation requires more than that for convergence, reduce the time-step. But I doubt if the problem is of time-step in your case. It is most likely because of operating density. Second reason could be wrong angle for wall adhesion.

 Fuzzy June 24, 2020 03:06

4 Attachment(s)
Thank you for the reply, really appreciate it!

Indeed, I didn't set the operating density previously. Just now I have ran the simulation with the operating density set, and I can now see the rate of water rise declining over time. However, the water rise still surpassed the maximum theoretical rise by a big margin.

For the second point, the capillary rise comes from the inlet.
The picture of the domain along with the zoomed view of it are attached below. The left and right side are the inlet and outlet respectively. The top side is the wall while the bottom side is the axis. I've also attached a water volume fraction contour to better picture the capillary rise simulation within it. Overall, the length of each cells in the axial direction is approximately 0.075 mm.

As of now, I'm still using a time step size of 5*10^-5. Based on my observation, it takes around 50 to 60 iterations for the scaled residuals to reach 10^-6 before jumping to the next time step (even though it is still rapidly declining). I will try setting the max iteration per time step to 40 and lower the time step size , but I'm afraid that it will require a very low time step size (to make the residuals relatively flat at each time step) and hence making the simulation run very long. Do the scaled residuals really have to be relatively flat (converged) before going to the next time step? And is my understanding of converged residuals correct?

I've included a screenshot of the scaled residuals below (but only for the first few time steps). Maybe it may help you to picture what I have described above.

Lastly, I'm pretty confident with the contact angle that I had set (10 degrees). However, I'm not sure whether the wall roughness would affect the solution since I didn't change it from the default settings. I am also a bit reluctant to refine the mesh more since then I would need to lower the time step size even more and make the simulation run longer.

Would you kindly give some advice or suggestions?

Kind regards,
Fz

 YipBF June 25, 2020 03:29

I am also working on capillary flow. But I faced reversed flow problem in pressure inlet and outlet. May I know how to eliminate this problem?

 Fuzzy June 30, 2020 02:17

Quote:
 Originally Posted by YipBF (Post 775942) I am also working on capillary flow. But I faced reversed flow problem in pressure inlet and outlet. May I know how to eliminate this problem?
Do you have a picture of your domain? Reverse flow in the outlet shouldn't occur if the domain is not fully filled with water

 YipBF July 13, 2020 05:06

My geometry is very simple. Just a vertical straight 2D rectangular channel with 0.1mm width and 10mm length. I set pressure inlet and outlet with zero pressure.

 YipBF July 13, 2020 05:24

Here are the details of my Fluent simulation:
- General - pressure based, absolute velocity, transient, 2D, Gravity enabled
- Model - Laminar, multiphase VOF consisting of air (primary phase) and water (secondary phase). Courant = 0.25
- Surface tension =0.072N/m and wall adhesion = 30
- Boundary condition - pressure inlet and pressure outlet (both have 0 gauge pressure). Water volume fraction of 1 at inlet. Backflow volume fraction of 0 at outlet. Already defined the walls contact angle.
* Method - PISO
* Initialization - initial water volume fraction of all zones set to 0

May I know how to patch the cells near inlet to have some water present?

 All times are GMT -4. The time now is 13:35.