I want to do CFD Verification and Validation In low speed range.

[Haoliang Yu]

Hi, 

I want to compare the aero analysis results between vsp and wind tunnel test.
CL CD and Cp,etc.

So ,I am trying to find out some data about wing wind tunnel test result,which airfoil can be NACA0012, NACA2412 or any other low speed airfoil.

I have just find the data about ONERA M6 wing 

https://www.grc.nasa.gov/WWW/wind/valid/m6wing/m6wing.html

but Ma=0.8395 which is not fit for my needs. So where can I get some useful data for my work?

Yours sincerely, 

Knight

[Brandon Litherland]

There was a full-scale study of a low-speed NLF airfoil on a Cessna 210 in the 30x60 wind tunnel.

https://ntrs.nasa.gov/citations/19900003225 

It depends on what you are trying to validate.  If you're just looking to try and match 2D airfoil data, then make a very high AR constant chord wing and look at the results near the root.  Take the sectional information.

[Haoliang Yu]

I am sorry I did not say what I wanted to do clearly.

Here is a picture show what I want to do. A 3D wing  wind tunnel test  of the aircraft for general aviation. I focus on 3D wing but not airfoils.

I want to do a V&V about a 3D wing at low speed. 

And I see the research about  ' full-scale study of a low-speed NLF airfoil on a Cessna 210 in the 30x60 wind tunnel' .

there only two picture about the 3D result. So can I use these data for my V&V ?

Or is there any other better choice?

Yours sincerely, 

Knight

[Brandon Litherland]

There are LOTS of Technical Notes/Reports from the Langley Full-Scale Tunnel.  You should search the NASA Technical Reports Server (NTRS) to locate more.

Two excellent TNs on a twin engine and single engine full-scale aircraft are included below as a start.

https://ntrs.nasa.gov/api/citations/19690007971/downloads/19690007971.pdf

https://ntrs.nasa.gov/api/citations/19700013310/downloads/19700013310.pdf

From the introduction of these papers, you'll find that there was an entire campaign of testing full-scale, light, general aviation aircraft at Langley over several years.  A summary paper of one such campaign is Reference #1 in these reports.  See if you can locate this paper on NTRS as practice. (It's there)

  1. Barber, Marvin R.; Jones, Charles K.; Sisk, Thomas R.; and Haise, Fred W.: An Evaluation of the Handling Qualities of Seven General-Aviation Aircraft. NASA TN D-3726, 1966.    

Such resources and experiments provide a fantastic way of building confidence and experience in tools and methods.  Start simple with a wing and tail in VLM and then build up your approach.  Eventually, you should be able to accurately recreate each of these plots with a combination of the Parasite Drag tool, stall modeling, VLM or Panel mode potential methods, and a few other corrections.

[Inseo Choi]

Hi.

VLM cannot solve  precisely  about transonic flow(mach number = 0.8~1.3), because effect of the viscousity.

So, you should have to compare the experiment results from subsonic range.

You can compare the wind tunnel test result from Veldhuis's PROWIM with VLM using VSPAERO.

The freestream velocity of the wing is 49.5m/s.

[PROWIM Link] https://repository.tudelft.nl/islandora/object/uuid%3A8ffbde9c-b483-40de-90e0-97095202fbe3

Inseo.

2023년 1월 24일 화요일 오후 11시 28분 13초 UTC+9에 28894...@qq.com님이 작성:

[Rob McDonald]

On Tuesday, January 24, 2023 at 6:09:13 PM UTC-8 wrote:

I am sorry I did not say what I wanted to do clearly.

Here is a picture show what I want to do. A 3D wing  wind tunnel test  of the aircraft for general aviation. I focus on 3D wing but not airfoils.

I want to do a V&V about a 3D wing at low speed. 

For the case pictured above, you need more spanwise resolution --and perhaps less chordwise resolution.

You can use OpenVSP and VSPAERO to do whatever you want -- and you can compare the results to whatever results you want.  Doing a V&V study is a good idea.  It will help you gain experience with the tools and also prove to you that they are good to use for your application.

The rest of the questions you ask are not appropriate for the OpenVSP Group.  Ask them to your advisor, supervisor, or boss.

We're here to answer questions about OpenVSP and VSPAERO.  We aren't here to plan your school project, find your validation cases for you, or tell you what to do.

Rob

 

And I see the research about  ' full-scale study of a low-speed NLF airfoil on a Cessna 210 in the 30x60 wind tunnel' .

there only two picture about the 3D result. So can I use these data for my V&V ?

[Kun Drew]

I think my question is very consistent with the thread, so I ask the question directly here.

I 'm comparing the VSPAERO results（VLM) for the E387 airfoil with the wind tunnel test results from NASA Langley Research Center.( Experimental results for the Eppler 387 airfoil at low Reynolds numbers in the Langley low-turbulence pressure tunnel - NASA Technical Reports Server (NTRS))

Previously I have considered how much aspect ratio can be used to simulate two-dimensional results as much as possible with three-dimensional calculations.The chord length of my rectangular straight wing is 0.15 m. The following is the calculation results of the lift coefficient CL of different span lengths.

Therefore, I used a 150m span ( i.e., an aspect ratio of 1000 ) to obtain a three-dimensional wing model for calculation and comparison with experimental results.According to the experimental conditions, the Reynolds number is 200000 and the Mach number is 0.06.The following is the comparison of the results of lift coefficient and drag coefficient.

The results show that the lift coefficients of the two results are in good agreement before the angle of attack of 8 °, but the deviation of the drag coefficients is large ( about 30 % -50 % ).However, this set of results is performed at the default spanwise and chordwise resolutions.Subsequently, I set the span resolution, SectTess _ U, to 60 and 600 for calculation and compare it with the default SectTess _ U of 6.Unfortunately, changing the spanwise resolution has little effect on the drag coefficient results.

I want to know whether the reason for the large deviation of the resistance coefficient is my model setting, my calculation setting, or the inherent defect of VLM.

I ' ll attach my model file in the hope that someone will answer my doubts.

I will be grateful！

[Brandon Litherland]

Thank you for providing the information and model. This helps a great deal in diagnosing your potential issues.

Let's have a look at what's going on.

I ran your model as-is and obtained the same drag results.  I assume that you are using CDtot for your CD value.  Showing the components of that coefficient as CDi and CDo, very nearly all of the total is coming from CDo which is a flat plate parasite drag approximation with some viscous correction for a NACA 0012 thrown in to give some lift-dependent behavior. You have to be very skeptical of attempting to capture 2D airfoil performance with a 3D code.  It seems like you're doing a great job of capturing the lift (ignoring separation and stall) so your induced drag should be fairly trustworthy.  This means that the profile/viscous/parasite components are off.  You will have to experiment with some of the parameters to capture this effectively but again I don't really recommend trying to catch 2D behavior with a 3D code.  If you want to see where the differences are coming from, you can always run Panel mode and capture a Cp slice and compare that profile to the Cp distribution published.  From what I saw for the AoA = 5 deg result, they aren't very far apart.

[Kun Drew]

Hi Brandon,

Thank you for your response. I would like to address or raise questions regarding the following :

(1)Yes, I used CDtot​​ for comparison and found that the deviation primarily originates from CD0​​. However, as per your previous response, when using a 3D wing to simulate a 2D airfoil, I have employed a very high aspect ratio. I further examined the parameters near the wing root and found CD​ in the lod file, which refers to CDi​​. Although the CDi near the wing root is indeed smaller than the CDi of the whole wing, its effect on CDtot is almost negligible. In the history file, I found CD0​​, but it is not related to the spanwise position. Therefore, I still have doubts about the deviation in the CDtot​​ data.

(2)Following your suggestion, I ran the panel method and captured a Cp​ slice. However, as shown in the figure below, it is clear that this solution is problematic. Considering that the low spanwise resolution might be the cause, I attempted to increase the num-u to 1000. However, this led to extremely slow computational iterations, and I ultimately killed the solver.

(3)For cp, the panel method obtains cp, while vlm obtains dcp, what is the difference between them, I can find the upper surface cp and lower surface cp in the NASA report, if I try to compare, should I use the panel method.

(4)As you mentioned, it is not recommended to use a 3D wing to simulate a 2D airfoil. Therefore, I would like to know if there are any classic and general 3D validation cases within the Reynolds number range of 200,000 to 500,000. Ultimately, my objective is to validate the fidelity of the Vortex Lattice Method (VLM). I am aware that its fidelity is slightly lower than that of CFD methods. However, the current results of CDtot show that it is not only low fidelity but appears to be distorted.

(5)If there are no 3D validation cases available, do you have any other suggestions for validating the E387 airfoil?

(6)Is it really the inherent defect of VLM that causes the deviation of CDtot？

Best regards

[Brandon Litherland]

1) Leveraging a near-infinite span section to simulate 2D will only get you so far and I don't recommend taking this approach. 

2) Running panel mode on the large span section won't matter here.  Back it off to something much shorter, like 2 in your case, and run again.  Greatly increasing NumU isn't the solution here.  At most, you could use 15-20.

3) Cp is the pressure coefficient around the wing. dCp is the delta (or difference) in pressure across the VLM surface.  You could get dCp from Cp by finding the difference between the two pressures along the chord.  I recommend looking at the Cp distribution of a panel run and comparing that to the Cp distributions in the paper.

4) In the posts above here, there are several high-quality NASA Technical Notes and additional references to many more in summary papers.  I recommend starting there.

5) VSPAERO is not a 2D airfoil code but there are several out there that you can use to compare to this data.  I suggest starting with XFOIL.

6) The CDtot defect you've obtained is not a result of VLM but rather how you are trying to model the wing.

As a general rule, we should test what we model and model what we test.  Attempting to recreate the wind tunnel "2D" solution by making the span near-infinite is a significant deviation from how the results were tested.  Try modeling the test section in OpenVSP and run VSPAERO then isolate the loads on the wing. (see example below of a 6 inch by 36 inch representation of the model with walls)

I think that you might find your results making more sense.