Dear all

I have just gone into final year at university and I intend on doing my final year project and thesis on CFD and airfoil/hydrofoils. I have at my disposal at the university a small wind tunnel (unfortunately no cavitation tunnel), and a Joukowsky airfoil with pressure tapings on the upper surface. My question to you guys, is can you think of something really neat or new and original that I could do in the wind tunnel and then use CFD to analyse. At present all I can think of is to run a series of angles of attack and then try to reproduce them in CFD. This is not only boring but the is nothing new about this. I would also like to try an link in some work with hydrofoils that I have been doing over the summer.

I am open to all suggestions, I really want to do something interesting and new. Please do not think that I am trying to ask you for a title for my these and project, it is just that two heads are better than one, and there are a lot more than two of you out there.

Thanks again

Andy

Andrew Parker a.j.parker@hw.ac.uk

Here is a way to add some excitement to your project:

Connect each of the pressure taps to a gasoline supply, set the whole thing on fire (while the tunnel is running), and then see if you can reproduce that flame propagation using CFD.

Warning: Have your local fire department standing by just in case...

By "final year at University" do you mean the first degree program (3 years in the UK?), or are you finishing an advanced degree?

When you mention "CFD" are you planning on using a canned code, modifying a public domain code, writing your own, or ... ?

Will this project (experiment pluc cfd) be all that you're doing in this final year, or will you have tutorials and exam preparation as well?

The point is, if this is an undergraduate project, it could easily grow to exceed the time you have available. Just the job of running those "unexciting" experiments and simulating them with CFD could well be a lot more exciting than you can imagine right now. And you may learn a great deal more than you expect!

Sorry, this does not speak to your question directly, but it needs to be said.

This is the final year project and thesis, of a four year undergraduate honours course in Mechanical Engineering. I will have tutorials and exams to do as well. My point was that for my project to be of any worth and to be of interest, has to contain some form of originality, and I thought that if I reproduced lab results using a commercial CFD code (Fluent), then there would not be that much of originality behind it. All I was looking for was an interesting spin-off or tangent to this work to add some flavour. Again any suggestions would be gratefully received. In response to Axel, I have already tried what you suggested and it is not as interesting as you might think.

Thanks again, I will be sure to respond your questions with an equal cander.

Andy

Andy,

Sorry, that was my pyromaniac side responding. But all joking aside, I don't think it would be boring to simply measure a pressure distribution as well as lift and drag for different angles of attack and compare that with numerical results. Considering that you already have a 'full plate' with your courses and exams that would be a very worthwhile project, and I think you would learn a lot about the inaccuracies of both experimental and numerical work.

Axel

Well, I don't really know whats been done before but how about changing the velocity profiles on the airfoil. I mean the main flow be left to right with different angles of attack but what about in addition have some sort of periodic blowing onto the top or bottom of the airfoil. I guess this might be like a turbulent burst which can cause planes to crash.

(btw. whats the top speed of the wind tunnel?)

The top speed of the tunnel is slow about 25-35 m/s I think, it is something I need to find out myself. In response to Axel, I have an ulterior motive behind me wanting to do an original and good final year project. I am and have been for a while, very interested in doing a PhD in CFD with an interest in turbulence, transition, and inverse design. I see my final year project as a real chance to prove that I could do something good. I really fancy the CFD Lab at Cambridge University in Britain, and it will be hard to get in and get the funding so this is why I want to do some good work. It is not because I am some sort of book boffin or CFD nurd, just that I see this as a good stepping stone.

Cheers

Andy

Ok, so you are looking for something a little more 'kick-ass'. How about this:

Put the airfoil at some critical angle of attack (near stalling), and while keeping the airspeed constant, change the free stream (i.e. upstream) turbulence intensity. Measure lift and drag as well as pressure distribution over time, which should be unsteady, if you operate right around stall/reattachment. Then try to recreate the same scenario on your computer, varying the upstream turbulence intensity in the same manner, and find out which turbulence model best matches your experimental results.

Axel

This is a really nice idea and is exactly the kind of thing I was looking for. It could be that this could produce some nice results and hence a good project – I am glad you 'caught my drift'. I wondered if I could ask your permission to further investigate this idea and may be even use it as my project – I don't want to step on anybody's toes. I am not quit sure how I would alter the upstream properties like the turbulence intensity without altering to many variables, I guess I would need to investigate this further, but that's what it is all about.

Cheers , any further suggestions would be gratefully received – again!

Andy

I would be very careful with the suggested project if I were you. It is the CFD equivalent of walking on land mines.

Massive separation/stall is a very difficult problem to simulate for many traditional CFD codes.

Before you jump into a multi-year PhD level work for a 1-year undergraduate final project, I would recommend that you

1) check the massive literature on airfoils for this particular problem to see if others have done it (or something similar)

2) understand (from literature) whether, for the particular Reynolds number that you will be testing, the flow characteristics are a function of turbulence intensity strongly or not. For many problems, there is a critical Reynolds number beyond which the global (and even local) features of the flow become independent of the Reynolds number (or other turbulence properties). If you are in that range, then the proposed project does not make sense, does it?

3) focus on the CFD aspect and find out whether the code that you'll be using has been tested on something remotely similar. There has to be an element of massive separation in their study; otherwise the airfoil problem is easy to solve!

4) focus on experimental data and find out whether you really need to conduct experiments.

5) If your advisor lets you do both experiment and CFD, question his/her motive(s) and/or experience on similar problems ;-))) (experimental setup alone will take you at least a couple months, even if everything appears to be "all" set)

Lastly, if your interest is in doing CFD for PhD, then I question the value of the experimental portion of your project as far as admission to the university is concerned. Of course, I personally believe that every CFD student must be required to have some level of training with experiments (and thus hands-on appreciation for the physics)

Adrin Gharakhani

Andrew,

I agree fully with Adrin.

I myself tried to do both experiments and CFD involving transonic wind tunnel testing of wedge airfoils for my undergraduate thesis. In course of 8 months, the experimental equipment broke down twice. I had to wait for 42 days for some one qualified to come over and fix the stuff the second time. Finally when I finished the experiments, I had 3 days to write up the thesis. To prevent the leading edge shock reflection (reflected shock can return and hit the airfoil since transonic shocks are near normal), the walls of the wind tunnel were porous. In the end, the BC I used to model wall porosity did not quite work and I had to abandon the CFD part completely.

Faculty are often not frank or realistic about undergraduate thesis projects. In the end, an undergraduate thesis does not matter as much to them as a Masters or a Ph.D thesis. Some just want to get some work done my enthusiastic students while others are probably too nice to sound discouraging. I have also come across a lot of faculty members (my undergraduate adviser being one of them) who get carried away by the idea as much as the student does.

Finally, I did not mean to doubt your abilities or dampen your interest but you have to realistic (or have a back up plan atleast).

Dear all

Cheers guys I appreciate all your input, I would now tend to agree with Adrin and Kalyan. I guess I don't know enough about the field to appreciate all the problems that could come up. I think I am back to plan one, and I should just do the experiments in the wind tunnel and then try to model them in CFD-I hope this will be enough!!!!!

Cheers again

Andy

Andy,

I understand entirely your motives and desires on this project, having just recently been a student with lofty aspirations. Let me say that it VERY easy to overload yourself when dealing with aerodynamics studies as a student. With the end goal of getting into a good doctoral program, I would say that people will be more interested in your methodology and the quality of your work than they will the title of it. If you have the coolest project ever but end up somehow doing a half-assed job (time constraints, experimental equipment failure, etc.), it will not be as valuable to your future as if you had chosen a more "boring" project and turned out some very high quality work. As a side note, while people tend to accept any data labeled experimental nowadays, it is in fact very hard to get good data experimentally. One could devote a great deal of time simply to developing a robust and accurate setup in a wind tunnel, especially for a "new" type of flow problem.

I agree with Keith's sentiments.

In my earlier post I discouraged you from attempting to do everything, because it will be impossible, and as Keith pointed out very well, it will end up backfiring on you rather than help you towards your PhD aspirations.

Now, let me recommend one approach. It seems to me that you already know where you wish to do your PhD. That is already a very big step. Next, you need to figure out who you want to work with AND whether _they_ would/will have the funding or the interest to take you in as a student. This is much much much more important than _anything_ you will do for your project. Note the number of "much"s in the previous sentence - the emphasis is intentional and the message has been validated time and again. That is, you could be applying to the same university with the same qualifications in different years and get different responses - qualification alone is not sufficient reason for you to get into a university UNLESS you are indepedently wealthy (in which case you should not do CFD!!)

Ok, once you have found _the_ professor and got him(/her) interested in you, you can ask _him_ what he thinks is a good project for you to conduct. He will most probably suggest a project that will help him measure your capabilities based on _his_ understanding of what is required to work on _his_ projects. If you do well, then there is a better chance that you're in. If you don't do well, then don't expect that any other project that you might have done well might/could have caught his interest. This is because every professor is focused on his/her specialty. You could be an excellent reseacher in CFD but you may not get a "job" with someone who is looking for an experimentalist! I hope you understand where I'm going with this.

In case of failure of the above approach, I would recommend that you think carefully what you mean by "I want to do CFD". Do you want to develop methodology or just run codes to study physics? I personally frown on anyone who thinks the latter has anything to do with CFD. It's like saying that I'm a mathematician because I can use a pocket calculator to punch 2+2 to get 4!

If you decide that you want to develop methodology, then the best thing to do is to solve some "simple" CFD-related mathematical problem, or develop some simplified numerical methodology to solve some interesting problem. The objective, at your level, is not to demonstrate that you are brilliant but that you are an independent thinker and can bring to the table the "minimum" required to carry out bigger projects.

Good luck

Adrin Gharakhani

FYI

About five years ago a researcher (female) did a study on who graduates from postgraduate programs and who "flunks out". The major problem was not "brains" or "the project" but the personalities of the professor(s) and the student - did they "like" or at least tolerate each other or did they "not get along".

Does anyone have references to this report or book.

Ya Ya - I know this is "off topic".

Dear Keith, Adrin, and the rest

You are all really top people, and I thank you very much for your detailed and VERY informative responses, that seem to based on your experience which is invaluable. You guys might remember that about six week ago there was a post that caused quit a lot of response, from me included, called 'I post a question and all I get is gibberish'. I think that you guys have just demonstrated that this forum does work and when you post a question you do get a good response, a very good response.

I have decided to do the best possible experiments that the time and equipment allows, and then to try to model this in Flunet, or CFX not sure yet. I can see that this will teach me a lot and should not be that "boring". I think I will try to paint the surface of the airfoil with an oil/paint die such that when the tunnel is run and the foil is at a set angle of attack, I get a clear indication of the point of separation (I have been reading up). I will use this to check my results from CFD amount other variables.

"I want to do CFD", I guess what I meant by this is that this is the area, and I know that I need to be more defined. I have been doing work in the summer using an inverse design routine and I think I would in the future like to develop the methodology behind this and develop better codes. I am also greatly interested in turbulence, transition, and the ideas behind separation, and therefore these areas again I feel will be involved in what I will try to do.

Cheers again, and thanks again!!

Andy

> Ya Ya - I know this is "off topic".

Absolutely not! This is very much related to what I posted earlier. That is why I suggested that prior to coming up with a project (to qulify for a PhD program) it is better to find a professor who will take you.

Adrin Gharakhani

Some advice, if tey are needed :

1) before making the first experiment, try to simulate it with your CFD code, whatever it is. This will teach you the quantities you have to look at in the experiment (turbulence intensity, turbulence length scale, flow velocity ...). And keep in mind that if you miss one of this quantities, you won't be able to correctly compare CFD and experiment.

2) what you want to do with the CFD code is very difficult and I won't be surprised if the answer is "none of the CFD codes/Turbulence model used give the correct behaviour". In that case, you will have to show that the CFD codes/Turbulence model were used the "best way" in term of mesh, boundary condition...

3) at last, if you have enought time, I think it could be interesting if you make a computation of a case showing a clear separation zone with an unsteady time scheme and compare the result with a steady state one.

Good luck

Sylvain