CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   3D simulation in Fluent (https://www.cfd-online.com/Forums/fluent/44615-3d-simulation-fluent.html)

 Amer May 6, 2007 16:51

3D simulation in Fluent

Hi, I did a 3D simulation for a marine propeller to study cavitation. I used a single phase, steady, and turbulent flow (J=0.2 and a cavitation number of 2). I considered the flow to be rotating and the propeller to be fixed. Could you please answer the following:

1- My solution is converging with only 200 iterations to 0.0001. (I heard that since it's 3D then it should take at least 2000 iterations for the solution to converge). what could be the problem (if there is any)?

2- The value of Y-star that I got is 10^4 and a professor told me that it shouldn't be more than 500. So what should I do (given I've tried the simulation with normalizing and then without normalizing in my solution and it gave me almost the same value for Y-star). how can I decrease the Y-star value)?

Thank you

 Phil May 6, 2007 17:47

Re: 3D simulation in Fluent

It's the number of cells that determines the number of iterations - and 3d meshes generally have alot more cells than 2d.

How many cells do you have. Without combustion a 3d mesh with under 100000 cells could converge in 200 iterations.

 Amer May 7, 2007 06:20

Re: 3D simulation in Fluent

thx alot! the number of cells I have is 450000 cells. does it make sense to converge in 200 iterations! and do u anything about the Y-star value?

 Phil May 7, 2007 07:32

Re: 3D simulation in Fluent

450000 is alot to converge in that number of iterations. Are you sure Fluent read the full 450000 cells?

Sorry, don't know anything about y-star.

YOU SHOULD CHECK THE CONVERGED RESULTS AND SEE IF THEY LOOK SENSIBLE.

 Amer May 7, 2007 16:43

Re: 3D simulation in Fluent

thanks alot! I've been told that maybe it's because I'm checking the box scale (I'm activating scale) near normalize in my solution. So, do u think it's a good practice to do that or my results will not be accurate? should I use normalize instead or is it better not to use any of them?

 Phil May 7, 2007 17:55

Re: 3D simulation in Fluent

If you mean in the residual monitors window, the default in the normalization box is to have scale checked and normalize unchecked. I wouldn't have expected it to affect convergence that way.

One thing may be that you have the convergence criterion too low. it should usually be 0.001 (0.000001 for energy).

Make sure all boundary conditions and models are correct and if necessary re-read the mesh into a fresh run of fluent(with defaults) and try again to run the simulation.

 davidmurphy26 May 16, 2012 16:13

Quote:
 Originally Posted by Amer ;141347 Hi, I did a 3D simulation for a marine propeller to study cavitation. I used a single phase, steady, and turbulent flow (J=0.2 and a cavitation number of 2). I considered the flow to be rotating and the propeller to be fixed. Could you please answer the following: 1- My solution is converging with only 200 iterations to 0.0001. (I heard that since it's 3D then it should take at least 2000 iterations for the solution to converge). what could be the problem (if there is any)? 2- The value of Y-star that I got is 10^4 and a professor told me that it shouldn't be more than 500. So what should I do (given I've tried the simulation with normalizing and then without normalizing in my solution and it gave me almost the same value for Y-star). how can I decrease the Y-star value)? Please help me in both or any of the above questions. Thank you
hi
perhaps your professor told you y-plus must be below 500 that you should decreaseing the first layer of wall in the computational domain(e.g. first layer on the hydrofoil, ...)

 foamcfd May 17, 2012 07:22

Your professor is right.Ystar should be less than 500 and more than 15 if you are using standard wall function. Fluent uses Ystar instead of Yplus values for their own reasons. Refine the mesh near wall to get Y plus around 50-60 this should fix your problem. Please read fluent help on Near wall treatment.

 davidmurphy26 May 18, 2012 07:13

Quote:
 Originally Posted by foamcfd (Post 361641) Your professor is right.Ystar should be less than 500 and more than 15 if you are using standard wall function. Fluent uses Ystar instead of Yplus values for their own reasons. Refine the mesh near wall to get Y plus around 50-60 this should fix your problem. Please read fluent help on Near wall treatment.
hi
i think Yplus is right no Ystar. of coure you know in about propeller this parameter isnot more important and pressure coeffienet and thrust and torque coeffient are important that more discussed in literature. wall function isn't more important than mesh near the wall if you see appropriated articles and books.

all the best
david

 All times are GMT -4. The time now is 08:08.