[shindiregita]

Hello,

I'm analyzing the increase of friction coefficient due to the growth of marine fouling on ship hull. Currently, I've been doing simulations using ANSYS Fluent on a flat plate with five different velocities (3 m/s, 6 m/s, 9 m/s, 12 m/s, and 15 m/s) and four different types of roughness for each velocity. I still need to do simulations for the ship model, but right now I'm still stuck on this problem with grid-independent mesh.

The simulation used a single-phase fluid, the air, as its purpose is to validate my former colleague's experiment of a similar analysis that was done by using a wind tunnel. I've done the grid-independence study for the smallest velocity with the smooth condition of the plate and found that the suitable grid consists of around 1.1 million cells, with a 3.84% error from the experiment result. I've used the k-w SST turbulent model with y+ = 1 for the near-wall treatment. I've used the tetrahedral patch conforming method for the mesh.

The problem is that when I tried to do a simulation for 6 m/s using the same meshing configuration as the previous grid-independent mesh for V = 3 m/s, the result turned out to be 9% different from the wind tunnel experiment result. I only changed the first layer thickness value. Up to this day, I'm still clueless about what has gone wrong. I also tried using cut cell and multizone methods to no avail. I also have asked for help from my seniors and supervisor, but I am still stuck here. If you can help me find out what's wrong with my simulation, it'd be very appreciated.

Thank you so much, I hope you all stay healthy and safe in the middle of this pandemic situation.

[vinerm]

The discrepancy, most likely, is not because of mesh. This is because of some phenomenon not being modeled in the simulation. Either the air properties are different in the wind tunnel or the profile is different. Usually, wind tunnels work with suction fans, implying the flow profile could be different at the inlet than what you might be specifying. Viscosity would not play a big role but density could have its effects. So, instead of looking into mesh, try to look for the physical reason. Other reason could be if there is any flow-separation.

[duri]

Originally Posted by shindiregita

The problem is that when I tried to do a simulation for 6 m/s using the same meshing configuration as the previous grid-independent mesh for V = 3 m/s, the result turned out to be 9% different from the wind tunnel experiment result. I only changed the first layer thickness value.

Check the y+ values for both the mesh and make sure at least 3 to 4 cells are with in y+ = 4. Doubling the velocity can roughly increase the y+ by 1.5 times for same mesh. Also maintain the stretching factor on the first cell.

[shindiregita]

Originally Posted by duri

Check the y+ values for both the mesh and make sure at least 3 to 4 cells are with in y+ = 4. Doubling the velocity can roughly increase the y+ by 1.5 times for same mesh. Also maintain the stretching factor on the first cell.

Hi, thank you for replying. Can you provide any literature for this, please? Thank you