CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Main CFD Forum

NACA0012 validation using Solidworks 2013

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   May 12, 2015, 09:58
Default NACA0012 validation using Solidworks 2013
  #1
New Member
 
fadli
Join Date: May 2015
Location: Malaysia
Posts: 7
Rep Power: 2
genom92 is on a distinguished road
hey guys, i'm currently doing the validation for NACA0012 using solidworks 2013. The problems I'm getting are the Cl is very low and the Cd is very high as compared to the reference. Even my stall angle is not correct. I'm guessing that my turbulence parameters are wrong but is cant figure them out. can anyone help me with this..i attached together my results.
Attached Images
File Type: png lift curve.PNG (12.4 KB, 16 views)
File Type: png drag curve.PNG (16.7 KB, 14 views)
File Type: jpg NACA 0012.jpg (89.5 KB, 13 views)
genom92 is offline   Reply With Quote

Old   May 12, 2015, 10:35
Default
  #2
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
It looks like your CL-alpha curve is totally incorrect, so the problem is probably more than the turbulence model. If this were the case your low angles of attack (say between +/- 4 degrees) would be pretty spot on with the wind tunnel data.

My first question...are you sure you modeled a NACA 0012 and not something thinner like a NACA 0006?

Second Question...how closely do your model Reynolds numbers match the wind tunnel data. CL-alpha will be lower and much more pronounced when you get below 5*10^5. Given your high CD and low CL your Reynolds number looks like it should be in the 10,000's, which clearly doesn't match the Theory of Wing Sections data at 10^6 and 10^7.
H0T_S0UP is offline   Reply With Quote

Old   May 12, 2015, 11:07
Default
  #3
New Member
 
fadli
Join Date: May 2015
Location: Malaysia
Posts: 7
Rep Power: 2
genom92 is on a distinguished road
Quote:
Originally Posted by H0T_S0UP View Post
It looks like your CL-alpha curve is totally incorrect, so the problem is probably more than the turbulence model. If this were the case your low angles of attack (say between +/- 4 degrees) would be pretty spot on with the wind tunnel data.

My first question...are you sure you modeled a NACA 0012 and not something thinner like a NACA 0006?

Second Question...how closely do your model Reynolds numbers match the wind tunnel data. CL-alpha will be lower and much more pronounced when you get below 5*10^5. Given your high CD and low CL your Reynolds number looks like it should be in the 10,000's, which clearly doesn't match the Theory of Wing Sections data at 10^6 and 10^7.
thanks for the the reply..

for the first question..i did double checked the airfoil and yes im sure its naca0012..

2nd question..i used re=3million and set an equation for the software to calculate by itself..

i'm attaching the the model and parameters..please have a look..
chord length and span are both 1meter.

thanks..
Attached Images
File Type: jpg Capture.jpg (55.2 KB, 12 views)
File Type: png Capture2.PNG (50.8 KB, 11 views)
File Type: png Capture3.PNG (53.8 KB, 11 views)
File Type: png Capture4.PNG (45.6 KB, 10 views)
File Type: jpg Capture5.jpg (51.2 KB, 9 views)
genom92 is offline   Reply With Quote

Old   May 12, 2015, 11:39
Default
  #4
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
A few things:

1) why is your mesh so fine and focused on upstream of the airfoil instead of down? If anything you need maybe 5 chord lengths ahead of the airfoil and 10 behind it, with the more detailed mesh where the wake would be.

2) Why does humidity matter?

3) You don't need to use a time dependent solver unless you want to capture the unsteady effects. This will make your CLs oscillate a little bit.

4) I see you have laminar and turbulent checked off. I would go with turbulent only since the reynolds number is in the millions.

Without knowing the type of solver being used it is hard to comment further. Your CL curves are definitely wrong though and it looks like the flow being modeled is laminar based on the results.
H0T_S0UP is offline   Reply With Quote

Old   May 12, 2015, 11:41
Default
  #5
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
Also, you need to adjust your mesh depending on the angle of attack. Simply modifying the X and Y velocities makes no sense with the current mesh you have.
H0T_S0UP is offline   Reply With Quote

Old   May 12, 2015, 11:56
Default
  #6
New Member
 
fadli
Join Date: May 2015
Location: Malaysia
Posts: 7
Rep Power: 2
genom92 is on a distinguished road
1) the mesh picture doesn't fully capture the mesh..i was actually showing the airfoil thickness..i attached the computational domain and full mesh picture..i used 5chord before, above, below and 15chord after.

2) i'm actually still learning and these CFD software things just come out for the past 2months..i still don't know much..

3) i read somewhere in the web, it says that flow above certain Re is unsteady..thats why i put time dependent at that time i ran the simulation..i don't remember the Re value stated. .

4) ok..i will select turbulent only flow for the next test..

5) the software edit the mesh automatically every time i make any changes..
Attached Images
File Type: png Capture7.PNG (14.7 KB, 5 views)
File Type: jpg Capture8.jpg (93.4 KB, 8 views)
genom92 is offline   Reply With Quote

Old   May 12, 2015, 16:03
Default
  #7
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
So regards to your answer to my second question, you don't know why you selected humidity?

Before you get into any CFD you need to figure out what problem you are solving first and what parameters are important.

Do you have a background in a CFD class that actually requires you to code?
H0T_S0UP is offline   Reply With Quote

Old   May 12, 2015, 16:57
Default
  #8
Senior Member
 
Matt
Join Date: Aug 2014
Posts: 281
Rep Power: 3
MBdonCFD is on a distinguished road
Be advised: Published data sets for NACA 0012 do not agree.

You still have issues unrelated to data incongruity, but you should be aware. I would recommend using a cambered airfoil for validation purposes.
MBdonCFD is offline   Reply With Quote

Old   May 12, 2015, 17:02
Default
  #9
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
From what I understand, data in ToWS was corrected many years ago. If the OP had an old publication I could see it being an issue. I tested a panel code I wrote with NACA data and had close agreement.
H0T_S0UP is offline   Reply With Quote

Old   May 12, 2015, 17:12
Default
  #10
Senior Member
 
Matt
Join Date: Aug 2014
Posts: 281
Rep Power: 3
MBdonCFD is on a distinguished road
How did you come by this info? I was told the exact opposite by someone whom I would put a lot of stock into. Not saying he is never wrong, but he is a well respected wind tunnel test engineer.

He claimed that he spoke to one of the authors of the original paper who claimed that the data collection was flawed. Unless I am mistaken, ToWS uses the original NACA data, no?

Either way, with that kind of uncertainty in the numbers it seems risky to use an 0012 as a validation case when there is a whole universe of airfoils to choose from with more recent and traceable documentation.

Also, not to dog on your code. I am sure its great. I have my own that I am quite proud of, but the assumptions used for boundary layer calculations are very rough approximations. With some airfoils it will be close, with others it may not be. There is some bit of luck involved and using a panel code to verify that experimental data is correct is just as risky as using potentially incorrect data to begin with. Not to mention bad practice.

To be safe, just use an airfoil you are confident with. Otherwise it could be a huge waste of time.
MBdonCFD is offline   Reply With Quote

Old   May 12, 2015, 20:10
Default
  #11
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
I have the Dover edition which states "This Dover edition, first published in 1959, is an unabridged and corrected republication of the first edition published in 1949..."

Now if I was home I'd simply run XFLR5 and compare with the book.
H0T_S0UP is offline   Reply With Quote

Old   May 12, 2015, 22:18
Default
  #12
New Member
 
fadli
Join Date: May 2015
Location: Malaysia
Posts: 7
Rep Power: 2
genom92 is on a distinguished road
Quote:
Originally Posted by H0T_S0UP View Post
So regards to your answer to my second question, you don't know why you selected humidity?

Before you get into any CFD you need to figure out what problem you are solving first and what parameters are important.

Do you have a background in a CFD class that actually requires you to code?
sad to admit, but no. i didnt have any CFD background. the last thing i was taught is the boundary layer theory using prantl eq. the department didnt set the CFD course as compulsory course to be taken. btw, thank you for reminding me to have CFD basics. need to learn it for the next 6months..huh..
genom92 is offline   Reply With Quote

Old   May 12, 2015, 22:23
Default
  #13
New Member
 
fadli
Join Date: May 2015
Location: Malaysia
Posts: 7
Rep Power: 2
genom92 is on a distinguished road
Quote:
Originally Posted by MBdonCFD View Post
Be advised: Published data sets for NACA 0012 do not agree.

You still have issues unrelated to data incongruity, but you should be aware. I would recommend using a cambered airfoil for validation purposes.
what do you mean published data do not agree? i read through the reply but still dont get it..

would you please explain 'issue unrelated to data incongruity'. or at least give an example.

actually this is a task from my final year project supervisor. he said to use naca 0012 for validation..so i'm kind of stuck with this airfoil.
genom92 is offline   Reply With Quote

Old   May 12, 2015, 23:25
Default
  #14
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
For s final project you would get much more out of developing your own code than running one in solid works. Usually NACA00xx series is better for simulation testing or whatever. MB was referring to the fact that the book you are using to check your simulations had some pretty significant errors in the data when it was first published.
H0T_S0UP is offline   Reply With Quote

Old   May 13, 2015, 09:16
Default
  #15
Senior Member
 
Matt
Join Date: Aug 2014
Posts: 281
Rep Power: 3
MBdonCFD is on a distinguished road
what do you mean published data do not agree? i read through the reply but still dont get it..
-I mean exactly what I said. There are multiple data sets in different locations (Theory of Wing Sections, NACA reports, NASA reports). None of them agree on things like drag coefficient, lift curve slope, max lift, etc...

would you please explain 'issue unrelated to data incongruity'. or at least give an example.
-You are using one of the least accurate software packages to solve one of the most complicated problems. I have used this software before and while it has its advantages in terms of useability, it is far from roubust. To get a good solution to an airfoil problem typically requires more control that Solidworks can provide. Not to say you can't get lucky, but k-e is the only turbulence model and it is far from the best choice. Not to mention the issues with controling mesh resolution. Can you even plot your y+ value in that program? (I don't recall) If not it seems like a pointless excersice to run an analysis that is heavily dependent upon boundary layer behavior.

actually this is a task from my final year project supervisor. he said to use naca 0012 for validation..so i'm kind of stuck with this airfoil.
-Well, you gotta do what you gotta do... Maybe just make a statement in the paper that data sets do not agree.

HOT_SOUP
-I have the same eddition of ToWS. Pretty sure that has the bad data in it. I could be wrong, but I don't trust it either way. As for XFLR5, I will reiterate my eariler comment. Panel codes are not a good way to verify experimental data. Usually one would use experimental results to verify numerical results in a validation study. To verify experimental data you need other experimental data. I get that you just want to compare to see if it is close, but how do you know that either is correct to begin with?
MBdonCFD is offline   Reply With Quote

Old   May 13, 2015, 09:19
Default
  #16
Senior Member
 
Matt
Join Date: Aug 2014
Posts: 281
Rep Power: 3
MBdonCFD is on a distinguished road
I definitely agree you will get more out of writing your own code. That is not an easy undertaking though. If this is only an undergrad project, writing your own code is almost certainly beyond the scope of the assignment. That requires numerical analysis, PDE, and other things that you may have not covered yet and are challenging to tackle without any prior exposure to.
MBdonCFD is offline   Reply With Quote

Old   May 13, 2015, 11:47
Default
  #17
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
MBdonCFD, I have designed and tested my own airfoil using XFLR5 and the wind tunnel. I can tell you that agreement is very close for lift for moderate angles of attack, i.e. where separation does not play a significant roll in the behavior of the airfoil. I have also used XFLR5 to do first iteration wing design for UAVs. I can attest that the code is very robust if you know its limits - mainly that it cannot predict drag, any behavior involving significant separation, and any Reynolds number above 10^6. When I ran my single element panel code I got within +/- 5% for ToWS data. Keep in mind that the numbers shouldn't agree exactly; a subtle assumption of potential flow is that the Reynolds number is infinite. Drela's code uses corrections (in the boundary layer I believe) to simulate low Re conditions. Even then you cannot go too low.

Now, mind you, I am just starting to learn CFD myself. I have a very strong background in aeronautics but not in numerical analysis. So I am drawing from my design experience and other projects I have undertaken.

Now for a project I did roughly one year ago (to satisfy my own interests) I extended the Hess-Smith panel method to be capable of conducting 2-element airfoil analysis. Interestingly my code indicated that the wind tunnel tests that were used in the paper I compared where incorrect, as their data had higher numbers than the code! It was a technical report from ~1920.

Also, MB, it is normal that no two wind tunnel tests are perfectly identical. That is what makes experimentation a pain in the ass!
H0T_S0UP is offline   Reply With Quote

Old   May 13, 2015, 13:06
Default
  #18
Senior Member
 
Matt
Join Date: Aug 2014
Posts: 281
Rep Power: 3
MBdonCFD is on a distinguished road
I am not looking to argue this until I die. I never claimed that they should be identical, but there should be reasonable agreement. If I recall I was seeing significant variation between data sets, not something trivial. It was more on the order of 10% or more. I too have done lots of wind tunnel testing, airfoils, aircraft, wind turbines and I know what to expect. This descrepancy was unusual.

It's great that you have that much confidence in your code, but it is misplaced. You cannot say that an experiment was bad based on results of XFLR5 or any other panel code, unless you want people to question your work. Its bad practice.

I am done arguing this. We will just have to agree to disagree.
MBdonCFD is offline   Reply With Quote

Old   May 13, 2015, 16:22
Default
  #19
Member
 
Alex
Join Date: Jan 2014
Posts: 50
Rep Power: 3
H0T_S0UP is on a distinguished road
Quote:
You cannot say that an experiment was bad based on results of XFLR5 or any other panel code, unless you want people to question your work.
I really disagree with this point. If you get WT data that out performs an idealized code, you know something is wrong. For example, if my WT data indicated my Cl-alpha curve was steeper than 2*pi, then there is a good chance something is off.
H0T_S0UP is offline   Reply With Quote

Old   May 13, 2015, 16:25
Default
  #20
Senior Member
 
Matt
Join Date: Aug 2014
Posts: 281
Rep Power: 3
MBdonCFD is on a distinguished road
Your logic is flawed. Why do you need a pannel code to know that your lift curve slope is too high? You can just calculate it form the exp data and say 'something is wrong'. How does your panel code play into that decision at all?
MBdonCFD is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
NACA0012 Validation Accuracy Improvement Alhasan OpenFOAM Running, Solving & CFD 8 February 1, 2015 09:39
Importing Rhinoceros 3d model to SolidWorks george85 FloEFD, FloWorks & FloTHERM 21 March 5, 2014 06:19
CFX problem in ubuntu (linux) Vigneshramaero CFX 0 July 13, 2012 10:22
CFX-Pre problem, pls help!!! cth_yao CFX 0 February 17, 2012 01:52
NACA0012 Validation Case Questions ozzythewise Main CFD Forum 3 August 3, 2010 14:39


All times are GMT -4. The time now is 17:32.