7
$\begingroup$

Will the lift coefficient (Cl) and drag coefficient (Cd) I obtain by using software simulations or subsonic wind-tunnel tests be the same as on an actual aircraft while cruising? Or will there be some error?

$\endgroup$
  • 4
    $\begingroup$ There will be some error. I have plenty of experience with computational mechanics, and there is always an inherent error. The question is always whether the error is acceptable given the application, not whether there is an error at all. If unacceptable, you will need a more sophisticated model, either by a more sophisticated model or by obtaining experimental data. $\endgroup$ – Sanchises Nov 11 '15 at 14:10
  • $\begingroup$ Wind tunnel tests usually are "real-world conditions" for all practical purposes. That's why we have them and don't just use CFD for everything. $\endgroup$ – reirab Nov 11 '15 at 16:26
  • $\begingroup$ Also with wind-tunnel testing there will be differences coming from (possibly) different Reynolds numbers, different Mach numbers, different geometrical scaling, blockage effects, support effects etc. That's why there's still full-scale flight testing :) $\endgroup$ – ROIMaison Nov 11 '15 at 18:17
  • $\begingroup$ There will be lots of errors due to the technics involved (method itself), and then there are the errors caused by configuration differences (aero configuration of the real flying thing (that are not modeled or are modeled with different geometries), and weight of the real thing, etc), and finally from the time the aircraft design is frozen till entry into service there are design changes which may influence the aero characteristics. That's why you fine tune the approximations during flight test phase with a prototype aircraft. $\endgroup$ – Gürkan Çetin Nov 11 '15 at 19:12
  • $\begingroup$ As mentioned, there are differences between CFD, wind tunnel and the real world. This is very evident in Formula 1, where the first practice session is used by the teams to do real world testing of aerodynamic pieces that they think will work based on the CFD & WT. Some teams spend an entire season struggling to figure out why what works in the lab doesn't work on the track. $\endgroup$ – FreeMan Nov 12 '15 at 19:22
8
$\begingroup$

It depends.

In computational fluid dynamics (CFD) there are limits for the calculation of turbulent flow. This means that large flow separation cannot be modeled precisely; especially its time-dependent characteristics are computationally much too complex to be modeled. Instead, statistical methods must be used.

For attached and far-field characteristics the methods are refined enough to yield very accurate results, so much that what had historically been studied in wind tunnel tests or only on the finished aircraft can now be calculated with confidence up to hypersonic speeds. Errors still exist - after all, if the geometric model does not include the manufacturing irregularities of the real aircraft, the results will not be in total agreement, but close enough for engineering purposes.

In wind tunnel testing you cannot expect the same results as long as the Reynolds number and Mach number are different from the real thing. This is adjusted with correction factors, and again the results become good enough for engineering purposes. Cryogenic wind tunnels even remove this restriction, but are very expensive to build and operate, because they run with pure nitrogen at very low temperature and several bars of pressure. Small errors remain, because all the details of the real aircraft cannot be modeled in a wind tunnel model and their ratio between dynamic pressure and stiffness cannot be adjusted for all flow conditions, so aeroelastic effects will remain different.

Windtunnel tests are still valuable for predicting forces in post-stall conditions and for calibrating new CFD codes. Most of the operational envelope can then be covered with CFD. The combination of both is enough to predict the force coefficients with confidence, and only aeroelastic effects and details of the stall behavior might turn out slightly differently than anticipated from tests.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.