[homay95]

well in fact Im doing right same as that, I mean Im doing ok and i get the same valid results when I apply boundary conditions as : 1.no-slip and 2. slip with 1 micrometer slip length, but all problem i am dealing with is when i try to use slip length over 1 micrometer
can you help me in any way please??

[vinerm]

Could you share the reference that you are validating against?

[homay95]

sorry for delay, and yes of course
this is the article title:
Geometric optimization of an enhanced microchannel heat sink with superhydrophobic
walls
by :Hamidreza Ermagan, Roohollah Rafee
and this is the doi:
https://doi.org/10.1016/j.applthermaleng.2017.11.039

[vinerm]

Which velocity component are you hooking this UDF to? It has to be the velocity tangential to the boundary. Other velocities must be 0.

[homay95]

i hook this UDF as x-component of velocity and put y & z components equal to zero, because as you can see the flow is along with x axis

[vinerm]

Where do I see that? Did you post a picture?

[homay95]

yeah actually i sent a picture. but i think it hasnt been uploaded. nevermind, i already did just like u said (applying udf to tangential component to the boundary) and i dont think my problem have any thing to do with basis settings. i just need to lower my residuals for the case slip length=20 micrometer. how could i do that?

[vinerm]

You can't stabilize a simulation if the physics is wrong. However, assuming that the physical setup is correct, you can try to bring in the changes slowly. Run the simulation with no-slip condition and then, after the case has achieved a stable state, hook the UDF.

[homay95]

i did it yesterday, i run it with no slip, after it got converged, i hooked the UDF and then i run the calculation, it was stable and didnt diverged at all, but the only problem was that my residuals were too high, i mean about 1e03

[vinerm]

Residuals higher than 1 imply divergence.

Possibly, there is some other setting that is wrong. As you stated earlier, the gradient field does not look good. In that case, use reconstructed gradient.

[homay95]

i used _RG instead of _G too, as u advised me to before, but it didnt work either. i even lower under relaxation factors for pressure and momentum, but still nothing, do i have to do something with advanced option in under relaxation part??

[vinerm]

If you are aware of multigrid methods, then you can change those options, otherwise, prefer not to touch those. Such kind of behavior is anyway not pointing towards numerics but physics itself. Create a Custom-Field-Function with 20 micrometer multiplied by the normal gradient. Plot its contour and see the numbers. If the velocities are much much higher than the average flow velocity, then you know the reason.

[homay95]

ok Im done with velocity slip. but when I initialize the case and then try to apply temperature jump UDF, there occurs an error that says :"signal SIGSEGV". what sould I do to get rid of that?

[vinerm]

You have to correct the code. SIGSEGV implies that the UDF is trying to access something that is unavailable, most likely C_T_G in your case.

[homay95]

yes you are right. I have used C_T_G in my code. what is wrong with this? what should I use instead?

[vinerm]

If you use DEFINE_SOURCE for energy, then C_T_G automatically becomes available, otherwise you need to use following commands

Alloc_Storage_Vars(domain, SV_T_RG, SV_T_G, SV_NULL);
T_derivatives(domain);
Free_Storage_Vars(domain, SV_T_RG, SV_NULL);

Ensure that the commands are in used in same sequence. domain should be pointing to domain of your case.

[homay95]

actually i want to apply temperature jump for a wall. so should I write a code with "Define Profile" or with a "Define Source"?

[vinerm]

Temperature jump is to be applied as a profile. So, you can use DEFINE_PROFILE.

[homay95]

ok. so you are saying i should use Define Profile Macro and also use the following expressions within my code, right?
Alloc_Storage_Vars(domain, SV_T_RG, SV_T_G, SV_NULL);
T_derivatives(domain);
Free_Storage_Vars(domain, SV_T_RG, SV_NULL);

[vinerm]

Yes, that's correct. But DEFINE_PROFILE and temperature derivative have nothing to do with each other. Those can be used exclusive of each other as well.