In simulating a fan for airflow performance, I've obtained a good and converged solution. In comparison with the experimental results, the pressure rise predicted from the fan is exactly on the money (within 2-3%). However, the moment is much higher than expected. As a result efficiency calculated from the pressure rise and HP becomes non-sense. Any thoughts or suggesations?

Hi Turkali

I think your problem is mesh. If you would like to calculate moment or drag accurately, you should use a mesh that is suitable to capture boundary layer correctly so Fluent can easily calculate the moment, integrating pressure and skin friction around the moment center.

I hope this will be helpful.

Regards, Mustafa Gelisli

Thanks for your recommendation. However, due to complexity of my fan blade geomtery, Gambit fails to build boundary layer around the blae. What you think my alternatives are? How about adaptation scheme. Have you had any experiences with adaptation?

Hi Turkali,

Gambit has boundary layer tool to create a boundary layer around faces easily. You can use that tool for boundary layer creation. To do that, you should mesh your faces first then use the tool and then determine your boundary layer parameters(first row height,ratio between adjecent cells and number of rows) for that faces(blade faces-and please be carefull when you create mesh around the blade tips). When you build volume mesh, Gambit automatically create a boundary layer around these faces. If this is a little bit confusing, you can use tgrid to create boundary layer.For more details please read the third tgrid tutorial about viscous hybrid mesh.

However, even if you create a boundary layer around the blades, that does not mean you obtain an accurate data in your first attempt. There are a few other parameters to specify carefully such as turbulence model. I suggest you to use RNG K-epsilon model or R -K-epsilon model with standart wall functions. And you must be carefull about Y plus value after the solution. On your blades, this value must be between 11,5-80 if you use standart wall functions.

Adaptation is a very useful tool but it sometimes make the convergence a little bit harder if you use Y plus adaptaion, because it creates new smaller cells around boundaries and these new cells may have a volume that is much smaller that adjecent cells, so be careful using that tool.

As I mentioned before, I have a very complex blade geometry and for that reason Gambit fails to create B.L. around the blade surfaces. The blade under investigation is composed of two/three layers of different materials on top of each other with different chord-lengths. On the other hand, blade surface has obstruction on the lower mid-portion and looks like step duct if viewed from radial direction. I used 1st element height of 0.03" for Reymold No. of 2.4E+5 and Chord-length 0f 6.0". The Y+ plot of the blade zone shows values between 10 to 500 with 150 being the populated and averaged. By the way, What's the best way to calculate the 1st element height for the B.L.? Thanks again.

Hi Turkali,

There is not an easy answer to your question. In Fluent manual some formula is given to calculate the first row height but it was derived from flat plate calculation so I dont think it is reasonable for your problem.In my opinion,the first row height should differ from root to tip because linear velocity differs. However, maybe you can find an avarage value for the first hight (there is no problem as long as y plus is between 12-80).

In adition, I insist on meshing the geometry using Gambit. Maybe you need a few face merging to overcome the problem.

And a last thing, If you reveal where are you from, I will be happy.

Mustafa, I'm originally from Iran and have been residing in the USA for 27 years. My mother is from Turkey. How about yourself.

I am from Istanbul, Turkey and I am 24 years old. I am a postgraduate student at university. Meanwhile I work for a engineering company where I use Fluent and Gambit as tool.

I tought that you could be a Turkish because of your name. It is very rare to see a Turkish name in such kind of forums. That is why I asked where you were from.