Hi guys,

My name is Julie and I am an engineer student at Ohio State University (Proudly one of the few female engineer students!). I am doing a project that involves setting up an actual wind tunnel model and then simulating the wind tunnel with Fluent. I have come a long way, but my Fluent results just aren't as expected. It's been really hard for me to get help from my professor because I have cheerleading practice during his office hours and we aren't allowed to skip. All the guys in my class are eager to help me, but when it comes down to it, they dont know what they are doing and I think they are just using it as an excuse to get to know me (boys will be boys I guess). Anway, if someone could help me I would really appriciate it. My email is jb321201@ohio.edu I also have no problem if we comunicate on the message board, but I don't know if this would be as effective (so much easier to just sign on my email).

Thanks, Julie B.

Hi,

if your results are strange, it is allway helpfull to check the reference values (->report->reference values). A lot of results are related to them (like Tu, Re, Drag, Nu,...)

Ralf

If people are willing to work through e-mail, then I'd take them up on that. I recommend using the forum though. You should search the forum to see if someone's already answered your question. Then there's the FAQ, which has a pretty good amount of information already put together (http://www.cfd-online.com/Wiki/Fluent_FAQ). If you still can't find an answer, then post your problem. Instead of getting one response from one person through e-mail, you may get a few responses from different people. And if someone gives you bad advice, then the other people on the forum will see it and set it straight, and other people can learn from it (which isn't possible if you're having a one-on-one through e-mail).

Post more info about your problem (what you're modeling, and what's coming out that's not expected... plus solver settings, any turbulence models, flow properties, etc) and I'm sure you'll get responses.

Ralf's right... if you're looking at it and your output coefficients are coming out wrong, then the reference values probably aren't set (common mistake).

Good luck, Jason

Hi guys,

Thanks for the feedback. I guess my main problem is that although the 'pressure' lift and drag forces are coming out correct, the viscous lift and drag forces are extremely small. They are about 20 to 100 times smaller than I was expecting.

Although my mesh isn't perfect, I believe it isn't causing this problem. It has 2 Million cells and other than a worst case cell with a Skewsize of .89, most cells are in between .3-.5. I think I am just setting something up wrong in Fluent. I am leaving most settings at Default (this includes leaving a 'laminar' viscous model).

One more thing I wanted to ask was do you feel I should use 2nd order? I was told it gave more precise results than 1st order. In the 'solve/solution' menu I leave the pressure to 'standard'. However, I have been told to set the pressure to Presto! for airfoils. Do you suggest I use Presto or standard? I have tried both and got slightly different results. What exactly does Presto! do?

Thanks for the help, Julie B.

Well, there are several things that can throw off viscous forces. First of all, CFD typically doesn't do very well at calculating viscous forces. If you look at typical results for a wing, the lift is right on, but the drag is usually off (but of the same order of magnitude). But that's due to simplifications that are made to the model, and better approximations for drag can be made.

I would definitely recommend second order. I'm used to seeing Presto for natural convection type problems, but I've seen some notes on this forum about using Presto for airfoils. I would try the second order solver for all variables, and then try switching pressure from Second Order to Presto and see how much of a change you get. I never report 1st order results. I only use them as a way to get to a 2nd order solution (but with the use of the fmg initializer [in the TUI type 's i fmg yes' to run the fmg initializer], I usually don't even bother with 1st order solutions any longer).

The number of cells in your model doesn't mean anything if they're in the wrong places or if your domain is too large. Even with a laminar model, you want a few elements in the BL to properly define the BL. Remember, Fluent is cell centered and although it makes an approximation for the BL, it doesn't store the wall derivatives (which are directly proportional to the viscous shear foces), therefore you need cells close to the wall to refine the approximation of the wall shear. The normal force typically dominates a wing, and it's much less sensitive to shear forces than pressure forces, so it's common to use a single element to represent the BL, but then you have to take the viscous shear forces (and drag force) with a grain of salt. But this all depends on what you're trying to accomplish.

Where are you getting your initial estimates for your viscous forces? If it's literature or experimental data, be careful that the data you're looking at isn't turbulent, because that will be very different than the viscous forces in a laminar regime.

Hope this helps, and good luck, Jason

Hey Jason,

Thanks for all the help.

To answer your questions 'Where are you getting your initial estimates for your viscous forces?', I am using an actual model in a wind tunnel and finding the max L/D. Once I have found this, I am using the priciple that at max L/D the viscous and pressure forces are equal. Therefore, my viscous drag should be half of the total drag. Isn't this a correct way to do this?

Thanks, Julie

I haven't heard that before. If someone else is following this conversation, can they chime in and comment on viscous and pressure forces being equal at max L/D? Are there any assumptions in this that can be a trap (i.e. shape dependency, RE dependency, etc)?

Again, if you're concerned with viscous forces, then I'd pay attention to making sure you have enough elements in the boundary layer.

Best of luck, Jason

Thanks Jason,

How do you check if you have enough elements in the boundary layer? Is this something I just can visually see?

Thanks, Julie

Usually, you need at least 4 elements in your boundary layer if you use wall functions. Do you have a prism layer on the airfoil? For an airfoil I would use a turbulent model (e.g. k-epsilon) and wall functions. A laminar model is only useful when you know from experiment that your flow is completely laminar. You can use first order to get an approximate solution, wich you can use as initial condition for second order simulation. Can you give some additional information on your model (speed, compressibility, boundary conditions)?

In response to Ralf: I've read the conversation and I have to say that, like you, I've never heard about, at max L/D, viscous and pressure forces are equal.

Julie, I recommend you to tell us something about boundary conditions. I've got a lot of wrong results because of the boundary conditions. Use turbulent model, as Vincent says. With some information about it we can guide you a little bit better, specially to get a well set turbulent model.

greetings