VSP AERO Parasite drag accuracy

300 views
Skip to first unread message

José Alfredo Rosas

unread,
Feb 25, 2021, 11:02:05 AM2/25/21
to OpenVSP
Hello, everyone.

I am currently looking into using genetic algorithm to carry out optimum wing design, my design approach has two steps: planform design and airfoil design, I have succesfully developed an algorithm  that approximates an eliptic planform given a set of constraints and now I'd like to modify the airfoil at diferent wing sections to achieve the desired C_L at cruise speed while minimizing drag. My question is the following: VSPAERO outputs a parasite drag result when a simulaiton is carried out, are these results reliable for this kind of analysis or would it be bettyer to get 2D data from Xfoil and then integrate taht along the wing span.

Thanks in advanced for the response :-)

Rob McDonald

unread,
Feb 25, 2021, 12:14:13 PM2/25/21
to ope...@googlegroups.com
The parasite drag calculation in VSPAERO is not suitable for this kind of thing.  It is a very simple model based on NACA 0012 data.  As you change the airfoil, the drag model will not change.

That said, I am confident that a genetic algorithm will be very inefficient at designing a wing.

Families of airfoils are often parameterized in terms of design lift coefficient.  You should first choose a family based on your needs (supercritical, laminar flow, etc).  Then, consider the cl distribution required to obtain your target cl*c distribution given your planform c distribution.  (If that sentence doesn't make sense, you need to study wing design until it does).  Next, choose the airfoil design cl distribution to match the target cl distribution.  Finally, adjust the twist distribution to get any final adjustments to your lift distribution.  There are techniques to do this that will give an answer much faster than optimization.  The result will be a wing that is very difficult to beat -- for almost zero computational cost.

Rob


--
You received this message because you are subscribed to the Google Groups "OpenVSP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openvsp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/cc183cf8-d855-4494-8dfd-bfc2905fec0an%40googlegroups.com.
Message has been deleted

José Alfredo Rosas

unread,
Feb 25, 2021, 2:30:19 PM2/25/21
to OpenVSP
Hello, Mr. McDnald.

Thanks for the quick response I think I did not explain myself as clearly as I thought. I used the GA algorithm to fit a wing of N sections with linear taper as close as possible to the eliptic planform given  a minimum allowable chordlenght  and a target surface area (the figure below is the result for a wing divided into 6 sections along its span and a minimum chord of 0.10). As you suggested with that chord distribution I was planning to look for a local c_l distribution that helped me in aproxximating a full eliptical lift distribution. For now I am limiting myselft to NACA 4-series as they are the simplest case of a parametric airfoil geometry, I am planning on iterating the airfoil sections at each of the span stations that divide the wing to get the required c_l while minimizing drag, then the airfoil section across each of the partitions would be an interpolation of the initial and final one. Using VSP  to analyze the aerodyanmic performance of each of thegenerated wings would be convinient as I am alredy using it to visualize the planform optimization results (I appended an image of the wing corresponding to the plot mentioned above), sadly its not possible so I will stick to using Xfoil to analyze the individual sections and then try to integrate them along the wingspan.

Thanks again for the quick answer and valuable recomendations :-)
3section.png
Reply all
Reply to author
Forward
0 new messages