I obtained the complex potential for flow past a flat plate at angle of attack by applying the Joukowski transform $z = \zeta + 1/\zeta$ (centered at the origin, so $\zeta = e^{i\theta}$) to flow past a cylinder.

For reference, the complex potential for flow past a cylinder (at angle of attack $\alpha$, of unit radius, with freestream velocity $U_\infty$, and with circulation $\Gamma$) is $$ F(\zeta) = U_\infty \left( \zeta e^{-i \alpha} + \frac{e^{i \alpha}} {\zeta} \right) -\frac{i \Gamma}{2 \pi} \log \zeta $$ and after taking the Joukowski transform, I get $$ F(z) = U_\infty \left( z \cos \alpha - i (z^2 - 4)^{1/2} \sin \alpha \right) - \frac{i \Gamma}{2 \pi} \log \left[\frac{1}{2}\left( z + (z^2-4)^{1/2} \right)\right] $$

Then we should be able to calculate lift around the flat plate with the Blasius formula, $$ F_x - i F_y = \frac{i \rho}{2} \oint \left( \frac{dF(z)}{dz} \right)^2 dz $$ where the integral is around the surface (flat plate from $z = -2$ to $z = 2$, or the cylinder $\zeta = e^{i \theta}$ if we integrate in the $\zeta$-plane).

What's the best way to evaluate this contour integral? I tried changing the integral to the $\zeta$-plane with something like $$ F_x - i F_y = \frac{i \rho}{2} \oint \left( \frac{dF(\zeta)}{d\zeta} \frac{d\zeta}{dz} \right)^2 \frac{dz}{d\zeta} d\zeta = \frac{i \rho}{2} \oint \left( \frac{dF(\zeta)}{d\zeta} \right)^2 \frac{1}{1-1/\zeta^2} d\zeta $$ but the $\frac{1}{1-1/\zeta^2}$ factor seems to mess things up by adding poles at $z=\pm1$ and making the residue of the pole at $z=0$ go to $0$. (I know what the answer should come out to by the Kutta-Joukowski theorem, but I want to be able to verify it.) Do I need to evaluate this integral in the $z$-plane?

Thanks in advance, any help is appreciated! My complex analysis and conformal mapping is definitely a bit rusty.

EDIT 1: I decided to try and generate interest by taking the calculations further to show where the problem is, and found out that continuing as I was doing does indeed give the right answer! Valuable lesson for me there that I shouldn't claim to be stuck just because it doesn't "seem" like I will get the right answer.

In the end, you do indeed have poles at $\zeta=\pm 1$ which have nonzero residues, and the pole at $\zeta = 0$ still has a nonzero residue as well. They add up perfectly to give the same sum of residues as in the non-transformed case, so that the Kutta-Joukowski theorem holds as we expect. (I guess so it's no longer unanswered, I'll answer my own question.)

  • $\begingroup$ Welcome to MSE =) $\endgroup$ – Patrick Da Silva Dec 7 '11 at 6:40
  • $\begingroup$ Is $U_\infty$ a function, or a constant ? $\endgroup$ – Sasha Dec 7 '11 at 13:43
  • $\begingroup$ Thanks! And $U_\infty$ is a constant, the freestream velocity. (I'll edit my original post.) $\endgroup$ – Frecka Dec 7 '11 at 16:00

I feel a little silly answering my own question, but that's alright.

As I said in edit 1, the approach in the $\zeta$-plane is perfectly valid. I'll show the math here real quick too. (I have to write it up since it's part of the solution set for a class I'm TAing anyway!)

Starting from where I left off in the question, $$ F_x - i F_y = \frac{i \rho}{2} \oint \left( \frac{dF(\zeta)}{d\zeta} \right)^2 \frac{1}{1-1/\zeta^2} d\zeta = \frac{i \rho}{2} \oint \left( \frac{dF(\zeta)}{d\zeta} \right)^2 \frac{\zeta^2}{\zeta^2-1} d\zeta $$ We can evaluate the derivative of the complex potential in the $\zeta$ plane as $$ \frac{dF(\zeta)}{d\zeta} = U_\infty \left( e^{-i \alpha} - \frac{e^{i \alpha}} {\zeta^2} \right) -\frac{i \Gamma}{2 \pi \zeta} $$ Squaring it gives $$ \left( \frac{dF(\zeta)}{d\zeta} \right)^2 = U_\infty^2 \left( e^{-2 i \alpha} - \frac{2}{\zeta^2} +\frac{e^{2 i \alpha}} {\zeta^4} \right) -\frac{i U_\infty \Gamma}{\pi} \left( \frac{e^{-i \alpha}}{\zeta} - \frac{e^{i \alpha}}{\zeta^3} \right) -\frac{\Gamma^2}{4 \pi \zeta^2} = A(\zeta) + B(\zeta) $$ where we group all the terms with even powers of $\zeta$ in $A(\zeta)$, and all the terms with odd powers of $\zeta$ in $B(\zeta)$. That is, $$ A(\zeta) = U_\infty^2 \left( e^{-2 i \alpha} - \frac{2}{\zeta^2} +\frac{e^{2 i \alpha}} {\zeta^4} \right) -\frac{\Gamma^2}{4 \pi \zeta^2}, \qquad B(\zeta) = -\frac{i U_\infty \Gamma}{\pi} \left( \frac{e^{-i \alpha}}{\zeta} - \frac{e^{i \alpha}}{\zeta^3} \right) $$ so $A(\zeta) = A(-\zeta)$, and $B(\zeta) = -B(-\zeta)$.

Now we're ready to evaluate our contour integral. $$ F_x - i F_y = \frac{i \rho}{2} \oint \left( \frac{dF(\zeta)}{d\zeta} \right)^2 \frac{\zeta^2}{\zeta^2-1} d\zeta = \frac{i \rho}{2} \oint \frac{(A(\zeta) + B(\zeta))\zeta^2}{(\zeta+1)(\zeta-1)} d\zeta $$ Applying the Residue theorem, $$ F_x - i F_y = \frac{i \rho}{2} (2 \pi i) \sum_k \text{Res} \left( \frac{(A(\zeta) + B(\zeta))\zeta^2}{(\zeta+1)(\zeta-1)},a_k \right) $$ Now we just need to calculate the residue of all the poles of that function inside the unit circle. There are 3 poles, $\zeta = 0, \pm 1$.

For $\zeta = 0$, the only term with $1/\zeta$ is $\frac{i U_\infty \Gamma}{\pi} \left(\frac{e^{i \alpha}}{{\zeta (\zeta+1) (\zeta-1)}} \right)$, which gives us $$ \text{Res}(0) = -\frac{i U_\infty \Gamma e^{i \alpha}}{\pi} $$ All terms have $\zeta = \pm 1$ as a pole, so $$ \text{Res}(1) = \frac{A(1) + B(1)}{2} $$ $$ \text{Res}(-1) = \frac{A(-1) + B(-1)}{-2} = \frac{-A(1) + B(1)}{2} $$ Summing the residues, we find $$ \sum_k \text{Res} \left( a_k \right) = B(1) - \frac{i U_\infty \Gamma e^{i \alpha}}{\pi} = -\frac{i U_\infty \Gamma}{\pi} \left( e^{-i \alpha} - e^{i \alpha} \right) - \frac{i U_\infty \Gamma e^{i \alpha}}{\pi} = -\frac{i U_\infty \Gamma e^{-i \alpha}}{\pi} $$ This is the exactly same as Res$(0)$ for the case of flow past a cylinder (without the $\zeta^2/(\zeta^2-1)$ factor), which is the only residue in that case, and so we know our lift will be the same. $$ F_x - i F_y = \frac{i \rho}{2} (2 \pi i) \left(-\frac{i U_\infty \Gamma e^{-i \alpha}}{\pi}\right) = i \rho \Gamma U_\infty e^{-i \alpha} $$ so $$ F_x = \rho \Gamma U_\infty \sin \alpha = \rho \Gamma U_{\infty,y}, \qquad F_y = -\rho \Gamma U_\infty \cos \alpha = -\rho \Gamma U_{\infty,x} $$ give the (lift-induced) drag and lift, respectively.

The agreement with the Kutta-Joukowski theorem is easier to see if we notice that the freestream isn't in the $x$-direction and compute $F_\perp$ and $F_\parallel$ rather than $F_x$ and $F_y$. Multiplying both sides by $e^{i \alpha}$ gives $$ e^{i \alpha} (F_x - i F_y) = (F_x \cos \alpha + F_y \sin \alpha) - i(-F_x \sin \alpha + F_y \cos \alpha ) = F_\parallel - i F_\perp = i \rho \Gamma U_\infty $$ Thus we have $F_\parallel = 0$ and $F_\perp = -\rho \Gamma U_\infty$ as desired.

  • 1
    $\begingroup$ You shouldn't feel silly for answering your own question; it's perfectly alright here... $\endgroup$ – J. M. is a poor mathematician Dec 9 '11 at 0:30

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.