Flow Lab

Syllabus / Module 11: Fluid Dynamics: Navier–Stokes and Beyond

Lesson 11.7

Potential Flow & Stream Functions

Irrotational incompressible flow, velocity potential, stream function, superposing elementary flows, flow past a cylinder, complex potential.

Written byClaude Opus

🌊
Why this matters for simulation: The analytical foundation of airfoil theory.

By the end of this lesson you can…

  • Explain why zero vorticity gives a velocity potential $\phi$ with $\nabla^2\phi = 0$, and why the linearity of Laplace's equation lets flows be added.
  • Derive $\phi$ and $\psi$ for a uniform stream, source, sink, vortex and doublet, and assemble a half-body and a cylinder from them.
  • Locate the stagnation points of a cylinder with circulation as $\Gamma$ grows, including where they leave the body at $G = 2$.
  • Use the complex potential $w(z) = \phi + i\psi$ to get the velocity from $dw/dz = u - iv$, by hand and by finite differences.
  • Check Kutta–Joukowski lift, $\rho U\Gamma = 1148.4$ N/m for the light-aircraft analog, by a pressure integral and by a momentum balance.
  • Explain d'Alembert's zero drag with an energy argument, and say where the missing drag lives.
On this page
·
  1. Irrotational flow has a potential
  2. The stream function, again
  3. Elementary flows
  4. Worked example: a fuselage nose
  5. The cylinder, with and without circulation
  6. Lift two ways
  7. The complex potential
  8. Worked example: stagnation points as a quadratic
  9. d’Alembert’s paradox
  10. Computing it
  11. Concept checks
  12. Problems
  13. Key Takeaways

Last lesson’s wing had a bound circulation of 18.7518.75 m2^2/s, and everything that made it happened in thin places: the boundary layer that made the vorticity, the trailing edge that set how much, the starting and tip vortices that carried it off. Everywhere else the air arrived from far upstream with no vorticity, and Kelvin’s theorem (inviscid, barotropic) says a material loop there keeps zero circulation, so by Stokes the vorticity stays zero. That is almost all the air around an airplane. Zero vorticity makes velocity a gradient, incompressibility turns that into Laplace’s equation, and Laplace’s equation is linear: flows can be added. A stream, a doublet and 11.6’s point vortex assemble into a lifting cylinder whose lift comes out as ρUΓ\rho U\Gamma. Then the honest part: the same theory says every body has zero drag.

Tools: 2.3 Gradient & Directional Derivatives, 2.11 The Laplacian & Vector Identities, 11.2 Conservation of Mass: The Continuity Equation, 11.3 Momentum: Pressure & the Euler Equations and 11.6 Vorticity Dynamics & Circulation.

▶ Video Zero vorticity makes velocity a gradient, elementary flows added like building blocks, a cylinder whose stagnation points slide as circulation grows, and the drag that never appearscaptions only
Transcript
  1. Last lesson ended with a promise. Outside the thin boundary layers and the wake, the air has no vorticity. Zero vorticity makes velocity a gradient: u equals grad phi. Incompressibility turns that into Laplace's equation, and Laplace says each value is the average of its neighbours. The equation is also linear, so flows can simply be added.
  2. The stream function psi is harmonic too, and with phi it obeys the Cauchy-Riemann equations, so streamlines cross equipotentials at right angles, a net of little curved squares. Linearity is the whole trick: any sum of solutions is a solution. One warning. Add velocities, never pressures. Compute pressure once, at the end.
  3. Four building blocks. A uniform stream. A source, phi equal to m over two pi times log r. A vortex, our old point vortex, phi equal to gamma over two pi times theta. And a doublet, a source and a sink squeezed together, with streamlines that are circles through the origin. Each is harmonic, so any sum of them is a flow.
  4. Add a stream to a doublet and the circle of radius a becomes a streamline, so it can be a wall. Add a clockwise vortex and it stays one, for every circulation. For the light-aircraft analog, fifty metres per second, circulation eighteen point seven five, radius zero point three seven five metres, G is zero point one five nine one five five. The stagnation points sit at minus four point five six four and one hundred eighty four point five six four degrees. The pressure coefficient is minus three point six six two on top, and minus two point three eight nine at the bottom.
  5. As circulation grows, the stagnation points slide down the body: minus fourteen point four eight degrees at G equal one half, minus thirty at G equal one. At G equal two they merge at the bottom. At G equal three, one leaves the body, two point six one eight radii below the centre, with air circling round it.
  6. Now the lift. Integrate the pressure around the surface and the force per metre is rho U gamma: one point two two five times fifty times eighteen point seven five, eleven hundred forty eight point four three seven five newtons per metre, twelve thousand six hundred thirty-three over eleven metres. A momentum balance on a circle of radius R agrees. At two radii, seven hundred seventeen point seven seven from pressure plus four hundred thirty point six six from momentum. Far away, half and half. Every sum is the same: the downwash and the low pressure on top are one lift, counted on different boxes.
  7. Complex numbers make this one line. Any formula in z is a potential flow: w equals phi plus i psi, and dw by dz is u minus i v. Note the minus. Stream, source, vortex and doublet become U z, log z and one over z. The lifting cylinder is w equals U times z plus a squared over z, plus i gamma over two pi log z. Set the derivative to zero and you get a quadratic; its roots are the stagnation points, on the circle for G below two.
  8. Now the honest part. Blasius's theorem gives the force on any closed body, and only the one-over-z term survives: lift rho U gamma, and zero drag. Zero, for every shape. This is d'Alembert's paradox. Drag would cost the engine power, D times U, and that energy has nowhere to go: no viscosity, a disturbance that travels with the body unchanged, and no wake.
  9. Real air disagrees. Behind the shoulder, potential flow makes the surface air climb from a pressure coefficient of minus three back to plus one, an adverse pressure gradient. The slow air next to the skin, held back by no-slip, cannot: it stops, reverses and separates, leaving a wide low-pressure wake. The missing drag lives there. Next, 11.8: Prandtl's boundary layer.

Irrotational flow has a potential

In 2D, ω=vx−uy=0\omega = v_x - u_y = 0. By Stokes, the circulation around any loop that bounds a patch of this air is zero, so ∫u⋅dl\int\mathbf u\cdot d\mathbf l between two points does not depend on the path. Then ϕ(x)=∫x0xu⋅dl\phi(\mathbf x) = \int_{\mathbf x_0}^{\mathbf x}\mathbf u\cdot d\mathbf l is a well-defined function with ∇ϕ=u\nabla\phi = \mathbf u, the same construction as potential energy for a conservative force. Conversely, if u=ϕxu = \phi_x and v=ϕyv = \phi_y, then vx−uy=ϕyx−ϕxy=0v_x - u_y = \phi_{yx} - \phi_{xy} = 0.

How can a wing with Γ=18.75\Gamma = 18.75 sit in irrotational air? A loop around the wing does not bound a patch of air: the patch it bounds contains the wing and its boundary layer, where the vorticity is. So ϕ\phi exists but grows by Γ\Gamma every lap around the body; the velocity, its gradient, is single-valued.

Incompressibility, ux+vy=0u_x + v_y = 0, then gives

∇2ϕ=ϕxx+ϕyy=0,\nabla^2\phi = \phi_{xx} + \phi_{yy} = 0 ,

Laplace’s equation. Three facts carry this lesson:

  • Meaning. Taylor expansion gives ϕ(x+h,y)+ϕ(x−h,y)=2ϕ+h2ϕxx+O(h4)\phi(x+h, y) + \phi(x-h, y) = 2\phi + h^2\phi_{xx} + O(h^4); add the yy pair and ∇2ϕ≈(sum of four neighbours−4ϕ)/h2\nabla^2\phi \approx (\text{sum of four neighbours} - 4\phi)/h^2. Laplace says: each value is the average of its neighbours (to O(h4)O(h^4) on a grid). No bumps, no dips inside the flow.
  • Linearity. ∇2(aϕ1+bϕ2)=a∇2ϕ1+b∇2ϕ2=0\nabla^2(a\phi_1 + b\phi_2) = a\nabla^2\phi_1 + b\nabla^2\phi_2 = 0. Any sum of solutions is a solution. This is the whole trick.
  • Boundary conditions. Far away u→(U,0)\mathbf u \to (U, 0); on a body, no flow through it, ∂ϕ/∂n=0\partial\phi/\partial n = 0. Not no-slip: the air slides along the skin, which 11.8 repairs.

The Euler equations’ nonlinearity has not vanished; it moved into the pressure. For irrotational flow Bernoulli holds between any two points, not only along one streamline (stated; 11.3 derived it along streamlines), so

Cp=p−p∞12ρU2=1−∣u∣2U2.C_p = \frac{p - p_\infty}{\tfrac12\rho U^2} = 1 - \frac{|\mathbf u|^2}{U^2} .

The stream function, again

From 11.2 Conservation of Mass: The Continuity Equation: u=ψyu = \psi_y, v=−ψxv = -\psi_x makes ∇⋅u=0\nabla\cdot\mathbf u = 0 automatically, ψ\psi is constant on streamlines, and the difference of ψ\psi between two streamlines is the flow rate between them. Irrotational flow adds ω=vx−uy=−(ψxx+ψyy)=0\omega = v_x - u_y = -(\psi_{xx} + \psi_{yy}) = 0: ψ\psi is harmonic too. Matching the two descriptions of uu and vv,

ϕx=ψy,ϕy=−ψx,\phi_x = \psi_y, \qquad \phi_y = -\psi_x ,

the Cauchy–Riemann equations. Then ∇ϕ⋅∇ψ=ψyψx−ψxψy=0\nabla\phi\cdot\nabla\psi = \psi_y\psi_x - \psi_x\psi_y = 0: equipotentials cross streamlines at right angles, and the two families tile the flow with little curved squares.

Elementary flows

In polar coordinates (chain rule, stated) ur=ϕr=ψθ/ru_r = \phi_r = \psi_\theta/r and uθ=ϕθ/r=−ψru_\theta = \phi_\theta/r = -\psi_r.

  • Uniform stream: ϕ=Ux\phi = Ux, ψ=Uy\psi = Uy.
  • Source of strength mm (volume per second per metre of span): all of it crosses every circle, so 2πr ur=m2\pi r\,u_r = m, ur=m/(2πr)u_r = m/(2\pi r), and ϕ=m2πln⁡r\phi = \frac{m}{2\pi}\ln r, ψ=m2πθ\psi = \frac{m}{2\pi}\theta. A sink has m<0m \lt 0.
  • Vortex (11.6’s point vortex, counterclockwise Γ\Gamma): uθ=Γ/(2πr)u_\theta = \Gamma/(2\pi r), ϕ=Γ2πθ\phi = \frac{\Gamma}{2\pi}\theta, ψ=−Γ2πln⁡r\psi = -\frac{\Gamma}{2\pi}\ln r. The source with ϕ\phi and ψ\psi swapped. ϕ\phi grows by Γ\Gamma per lap: the multivalued potential above.
  • Doublet: a source mm at (−ε,0)(-\varepsilon, 0) and a sink at (ε,0)(\varepsilon, 0), squeezed together with μ=mε/π\mu = m\varepsilon/\pi fixed. Their angles are θ(x±ε,y)≈θ±ε θx\theta(x \pm \varepsilon, y) \approx \theta \pm \varepsilon\,\theta_x with θx=−sin⁡θ/r\theta_x = -\sin\theta/r, so ψ=m2π⋅2ε θx=−μsin⁡θr\psi = \frac{m}{2\pi}\cdot 2\varepsilon\,\theta_x = -\frac{\mu\sin\theta}{r} and ϕ=μcos⁡θr\phi = \frac{\mu\cos\theta}{r}.

Each is harmonic away from its centre (check one: ln⁡r=12ln⁡(x2+y2)\ln r = \tfrac12\ln(x^2 + y^2) has ∂xx=(y2−x2)/r4\partial_{xx} = (y^2 - x^2)/r^4 and ∂yy=(x2−y2)/r4\partial_{yy} = (x^2 - y^2)/r^4, summing to zero for r>0r \gt 0). So any sum of them, plus a stream, is a flow.

Worked example: a fuselage nose

Stream plus source is a Rankine half-body. On the axis upstream the velocity is U+m/(2πx)U + m/(2\pi x), zero at xs=−m/(2πU)x_s = -m/(2\pi U): the stagnation point. Far downstream all of the source’s fluid flows inside the body at speed UU, so the body’s width is m/Um/U. The body is the streamline through the stagnation point, ψ=m/2\psi = m/2.

A light aircraft’s fuselage 1.21.2 m wide at U=50U = 50 m/s needs m=60m = 60 m2^2/s. The stagnation point sits 0.19100.1910 m ahead of the source; the lab finds the fastest surface air, 1.260 U1.260\,U (Cp=−0.587C_p = -0.587), at (0.199,0.390)(0.199, 0.390) m, on the shoulder of the nose. The lab also checks Laplace on this ϕ\phi: at (0.3,0.4)(0.3, 0.4) m, ϕxx=10.695\phi_{xx} = 10.695 s−1^{-1} alone, yet the 5-point Laplacian is 1.3×10−41.3\times10^{-4}, the stencil’s O(h2)O(h^2) error at h=1h = 1 mm.

◆ Interactive Potential-flow builder: drop sources, sinks, vortices and doublets into a stream; streamlines, Cp shading, stagnation points by Newton

Loading simulation…

The half-body preset is this flow scaled to U=1U = 1, m=2πm = 2\pi: stagnation point (−1,0)(-1, 0), width 2π2\pi. The oval (source and sink ±2π\pm2\pi at x=∓1x = \mp1) has stagnation points at (±1.7321,0)(\pm1.7321, 0), which is ±3\pm\sqrt3 (Problem 3). Hover anywhere: central differences of the summed velocity give divergence and vorticity below 10−610^{-6} more than half a unit from any element, growing next to one, where the stencil straddles a steep field.

The cylinder, with and without circulation

Stream plus doublet:

ψ=Ursin⁡θ−μsin⁡θr=Usin⁡θ(r−a2r),a2=μU.\psi = U r\sin\theta - \frac{\mu\sin\theta}{r} = U\sin\theta\Big(r - \frac{a^2}{r}\Big), \qquad a^2 = \frac{\mu}{U} .

On r=ar = a, ψ=0\psi = 0: the circle is a streamline, so it can be a solid wall. This is 11.2’s cylinder, now built.

Add a clockwise vortex Γ\Gamma at the centre (clockwise positive, like the wing’s bound vortex):

ψ=U(r−a2r)sin⁡θ+Γ2πln⁡ra.\psi = U\Big(r - \frac{a^2}{r}\Big)\sin\theta + \frac{\Gamma}{2\pi}\ln\frac{r}{a} .

Still ψ=0\psi = 0 on r=ar = a, for every Γ\Gamma. One body, infinitely many flows: Laplace’s equation outside a body has a solution for each circulation, because a loop around the body can enclose any circulation. Theory alone cannot choose; on a wing the sharp trailing edge does (the Kutta condition, 12.1).

The surface velocity is uθ=−ψr=−(2Usin⁡θ+Γ/(2πa))u_\theta = -\psi_r = -\big(2U\sin\theta + \Gamma/(2\pi a)\big), with θ\theta from the downstream axis. With G=Γ/(2πUa)G = \Gamma/(2\pi U a), Cp=1−(2sin⁡θ+G)2C_p = 1 - (2\sin\theta + G)^2 (11.3), and the stagnation points sit where

sin⁡θ=−G2.\sin\theta = -\frac G2 .

At G=0G = 0: 0∘0^\circ and 180∘180^\circ. At G=0.5G = 0.5: −14.48∘-14.48^\circ and 194.48∘194.48^\circ. At G=1G = 1: −30∘-30^\circ and 210∘210^\circ. As Γ\Gamma grows the two points slide down the back and the front, and at G=2G = 2 they merge at the bottom, 270∘270^\circ. Beyond, sin⁡θ=−G/2\sin\theta = -G/2 has no solution: the stagnation point has left the body (next section).

The light-aircraft analog. Take U=50U = 50 m/s, Γ=18.75\Gamma = 18.75 m2^2/s and a=c/4=0.375a = c/4 = 0.375 m for 11.5’s 1.51.5 m chord (why c/4c/4: the callout below). Then G=0.159155G = 0.159155, which is 1/(2π)1/(2\pi), and the stagnation points sit at −4.564∘-4.564^\circ and 184.564∘184.564^\circ: the rear one has slid just under the back, the front one just under the nose. The top reads Cp=1−(2+G)2=−3.662C_p = 1 - (2 + G)^2 = -3.662, the bottom 1−(2−G)2=−2.3891 - (2 - G)^2 = -2.389.

◆ Interactive Cylinder with circulation at 50 m/s: stagnation points sliding, surface Cp, and lift two ways

Loading simulation…

Lift two ways

Pressure on the surface. L′=−∮(p−p∞) ny ds=−q∞a∫02πCpsin⁡θ dθL' = -\oint(p - p_\infty)\,n_y\,ds = -q_\infty a\int_0^{2\pi}C_p\sin\theta\,d\theta. Expand Cp=1−4sin⁡2θ−4Gsin⁡θ−G2C_p = 1 - 4\sin^2\theta - 4G\sin\theta - G^2: against sin⁡θ\sin\theta over a full turn only the odd term survives, ∫02πsin⁡2θ dθ=π\int_0^{2\pi}\sin^2\theta\,d\theta = \pi, so L′=4πG q∞a=4πΓ2πUa⋅12ρU2a=ρUΓL' = 4\pi G\,q_\infty a = 4\pi\frac{\Gamma}{2\pi U a}\cdot\tfrac12\rho U^2 a = \rho U\Gamma. For our numbers,

L′=1.225⋅50⋅18.75=1148.4375 N/m,L' = 1.225\cdot50\cdot18.75 = 1148.4375 \text{ N/m},

and the sim’s 720-point midpoint sum prints 1148.43751148.4375, drag 2×10−132\times10^{-13}. Over an 1111 m span, as a 2D strip, that is 12 63312\,633 N: 11.6’s lift, recovered from the circulation.

Momentum through a far circle. Put a circle of radius RR around the body. The air between body and circle is steady, so the forces on it balance the momentum it exports: the body pushes it with −F-\mathbf F, the outer air with −∮p n ds-\oint p\,\mathbf n\,ds, and momentum ∮ρu(u⋅n) ds\oint\rho\mathbf u(\mathbf u\cdot\mathbf n)\,ds leaves. Hence

F=−∮R[(p−p∞) n+ρ u (u⋅n)] ds.\mathbf F = -\oint_R\big[(p - p_\infty)\,\mathbf n + \rho\,\mathbf u\,(\mathbf u\cdot\mathbf n)\big]\,ds .

Sim and lab: at R=2aR = 2a, pressure 717.77717.77 plus momentum flux 430.66430.66; at 4a4a, 610.11+538.33610.11 + 538.33; at 100a100a, 574.28+574.16574.28 + 574.16. Every sum is 1148.43751148.4375. On the skin lift is all pressure; far away about half is pressure and half is downward momentum carried out through the circle, the downwash. “The wing pushes air down” and “the pressure is lower on top” are the same lift, counted on different boxes.

The complex potential

Complex numbers, just enough. A complex number z=x+iyz = x + iy, with i2=−1i^2 = -1, is a point of the plane. They add like vectors. In polar form z=reiθz = re^{i\theta}, using Euler’s formula eiθ=cos⁡θ+isin⁡θe^{i\theta} = \cos\theta + i\sin\theta (1.7 Complex Eigenvalues & Rotation); multiplying multiplies lengths and adds angles, so 1/z=e−iθ/r1/z = e^{-i\theta}/r. The logarithm undoes the exponential: log⁡z=ln⁡r+iθ\log z = \ln r + i\theta, with θ\theta defined only up to 2π2\pi, the same multivaluedness as the vortex’s ϕ\phi.

Now the payoff. Let w(z)w(z) be any formula in zz (powers, quotients, log⁡\log). It sees xx and yy only through z=x+iyz = x + iy, so by the chain rule ∂w/∂x=w′(z)\partial w/\partial x = w'(z) and ∂w/∂y=i w′(z)\partial w/\partial y = i\,w'(z). Write w=ϕ+iψw = \phi + i\psi:

ϕy+iψy=i(ϕx+iψx)=−ψx+iϕx.\phi_y + i\psi_y = i(\phi_x + i\psi_x) = -\psi_x + i\phi_x .

Real and imaginary parts: ϕy=−ψx\phi_y = -\psi_x, ψy=ϕx\psi_y = \phi_x. Cauchy–Riemann, for free. Every formula in zz is a potential flow, the real part its potential, the imaginary part its stream function. The velocity is

dwdz=∂w∂x=ϕx+iψx=u−iv(note the minus).\frac{dw}{dz} = \frac{\partial w}{\partial x} = \phi_x + i\psi_x = u - iv \quad(\text{note the minus}).

The catalogue becomes one line each: UzUz; m2πlog⁡z\frac{m}{2\pi}\log z; −iΓ2πlog⁡z-\frac{i\Gamma}{2\pi}\log z (check: −iΓ2π(ln⁡r+iθ)=Γ2πθ−iΓ2πln⁡r-\frac{i\Gamma}{2\pi}(\ln r + i\theta) = \frac{\Gamma}{2\pi}\theta - i\frac{\Gamma}{2\pi}\ln r); μ/z=μr(cos⁡θ−isin⁡θ)\mu/z = \frac{\mu}{r}(\cos\theta - i\sin\theta). The lifting cylinder:

w=U(z+a2z)+iΓ2πlog⁡z,dwdz=U(1−a2z2)+iΓ2πz.w = U\Big(z + \frac{a^2}{z}\Big) + \frac{i\Gamma}{2\pi}\log z, \qquad \frac{dw}{dz} = U\Big(1 - \frac{a^2}{z^2}\Big) + \frac{i\Gamma}{2\pi z} .

Worked example: stagnation points as a quadratic

Set dw/dz=0dw/dz = 0 and multiply by z2/Uz^2/U: z2+iGa z−a2=0z^2 + iGa\,z - a^2 = 0, so

za=−iG±4−G22.\frac za = \frac{-iG \pm\sqrt{4 - G^2}}{2} .

For G≤2G \le 2: ∣z/a∣2=4−G24+G24=1|z/a|^2 = \frac{4 - G^2}{4} + \frac{G^2}{4} = 1, on the circle with sin⁡θ=−G/2\sin\theta = -G/2, as before. For G>2G \gt 2 both roots are on the negative yy axis, z/a=−i(G±G2−4)/2z/a = -i\big(G \pm\sqrt{G^2 - 4}\big)/2, with product of lengths 11: one outside the body, one inside (not flow). At G=3G = 3: r/a=(3+5)/2=2.618034r/a = (3 + \sqrt5)/2 = 2.618034, which the sim’s G=3G = 3 button prints. The streamlines between the body and that point close on themselves (the sim’s nested loops): a ring of air carried round and round.

d’Alembert’s paradox

Both force checks gave drag at round-off, for every Γ\Gamma. Not a cylinder accident. Blasius’ theorem (stated) gives the force on any 2D body as Fx−iFy=iρ2∮(dw/dz)2 dzF_x - iF_y = \frac{i\rho}{2}\oint(dw/dz)^2\,dz around it. Far from a closed body, dw/dz=U+iΓ2πz+O(1/z2)dw/dz = U + \frac{i\Gamma}{2\pi z} + O(1/z^2): no 1/z1/z source term, since a closed body emits no air. In (dw/dz)2=U2+iUΓπz+O(1/z2)(dw/dz)^2 = U^2 + \frac{iU\Gamma}{\pi z} + O(1/z^2) only the 1/z1/z term survives the loop, because on z=Reiθz = Re^{i\theta}, dz=iz dθdz = iz\,d\theta and ∮dz/z=∫02πi dθ=2πi\oint dz/z = \int_0^{2\pi}i\,d\theta = 2\pi i, while ∮z−kdz=0\oint z^{-k}dz = 0 for k≥2k \ge 2. So

Fx−iFy=iρ2⋅2πi⋅iUΓπ=−iρUΓ:F_x - iF_y = \frac{i\rho}{2}\cdot2\pi i\cdot\frac{iU\Gamma}{\pi} = -i\rho U\Gamma :

lift ρUΓ\rho U\Gamma and zero drag, for every shape, in one line. The lab sums the contour at R=2aR = 2a: −1148.4375 i-1148.4375\,i, with a real part at round-off.

Why no drag: the energy has nowhere to go. Fly the plain cylinder through still air at 5050 m/s. A drag DD would cost the engine power DUDU. That energy must become heat (there is no viscosity), or kinetic energy of the air (the disturbance travels with the body unchanged; its energy, 12ρπa2U2=676.49\tfrac12\rho\pi a^2U^2 = 676.49 J/m, is constant), or something left behind (there is no wake: behind the body the air returns to rest exactly as it was disturbed in front). Nowhere: so D=0D = 0. Lift is perpendicular to the motion and does no work, so the theory allows it.

Where real drag comes from. A real 0.750.75 m cylinder at 5050 m/s (Re=2.57×106Re = 2.57\times10^6 with 11.5’s air, lab C) has a drag coefficient of order one (from memory). Behind the shoulder, potential flow demands that the surface air climb from Cp=−3C_p = -3 back to +1+1, an adverse pressure gradient. The slow air next to the skin, held back by no-slip (11.4), cannot climb it: it stops, reverses and separates, and a wide wake of low pressure trails the body. That wake is where vorticity is nonzero, the region we excluded at the start, and it grows as the body moves on: the energy sink. On a streamlined wing the layer stays attached and potential flow gets lift and pressure nearly right.

Computing it

🐍 PythonLab: the cylinder analog's lift by pressure (A), by momentum (B), Blasius and the disturbance energy (C), a fuselage nose (D)
⌘/Ctrl + Enter
🧪 Coding exerciseExercise: velocity from any complex potential by a finite difference, and stagnation points by Newton's method
⌘/Ctrl + Enter

Graded: velocities to 10−610^{-6} (the off-axis point catches the sign of vv), the analog’s rear stagnation point, and the off-body point at G=3G = 3. The solution prints −14.478∘-14.478^\circ and −30.000∘-30.000^\circ for G=0.5G = 0.5 and 11, and r=2.618034r = 2.618034 at G=3G = 3. Keep away from the negative real axis, where np.log jumps by 2πi2\pi i.

Concept checks

✅ Concept check

For the light-wing analog ($U = 50$ m/s, $a = 0.375$ m, $\Gamma = 18.75$ m$^2$/s clockwise), at what angle, in degrees from the downstream axis, is the rear stagnation point?

✅ Concept check

Why can a source, a doublet and a vortex simply be added to a uniform stream?

✅ Concept check

At some point $dw/dz = 3 - 4i$ m/s. What is the velocity?

✅ Concept check

Why does steady potential flow put zero drag on any body?

Problems

✏️ Pen & paperA fuselage nose●○○

A uniform stream UU plus a source mm at the origin. (a) Find the stagnation point. (b) Find the body’s width far downstream. (c) Evaluate for U=50U = 50 m/s and a body 1.21.2 m wide.

Show worked solution

(a) On the axis upstream, u=U+m/(2πx)=0u = U + m/(2\pi x) = 0 at xs=−m/(2πU)x_s = -m/(2\pi U). (b) Far downstream the source’s whole output mm flows inside the body at speed UU, so the width is m/Um/U (the body is the streamline ψ=m/2\psi = m/2, approaching y=±m/(2U)y = \pm m/(2U)). (c) m=1.2⋅50=60m = 1.2\cdot50 = 60 m2^2/s and xs=−60/(100π)=−0.1910x_s = -60/(100\pi) = -0.1910 m. The lab adds the peak surface speed, 1.260 U1.260\,U.

✏️ Pen & paperStagnation points with circulation●●○

For w=U(z+a2/z)+iΓ2πlog⁡zw = U(z + a^2/z) + \frac{i\Gamma}{2\pi}\log z, find all stagnation points for G=Γ/(2πUa)=0.5G = \Gamma/(2\pi Ua) = 0.5 and G=3G = 3.

Show worked solution

dw/dz=0dw/dz = 0 gives z2+iGa z−a2=0z^2 + iGa\,z - a^2 = 0, so z/a=(−iG±4−G2)/2z/a = (-iG \pm\sqrt{4 - G^2})/2. G=0.5G = 0.5: z/a=±0.968246−0.25iz/a = \pm0.968246 - 0.25i, on the circle at θ=−14.48∘\theta = -14.48^\circ and 194.48∘194.48^\circ (sin⁡θ=−0.25\sin\theta = -0.25). G=3G = 3: 4−9=i5\sqrt{4 - 9} = i\sqrt5, so z/a=−i(3∓5)/2z/a = -i(3 \mp\sqrt5)/2: −2.618034 i-2.618034\,i in the flow and −0.381966 i-0.381966\,i inside the body, which is not flow.

✏️ Pen & paperThe Rankine oval●●○

A source mm at (−b,0)(-b, 0) and a sink −m-m at (b,0)(b, 0) in a stream UU. Find the stagnation points, and evaluate for U=b=1U = b = 1, m=2πm = 2\pi (the builder’s oval preset).

Show worked solution

On the axis outside the pair, u=U+m2π(1x+b−1x−b)=U−m2π⋅2bx2−b2u = U + \frac{m}{2\pi}\Big(\frac{1}{x + b} - \frac{1}{x - b}\Big) = U - \frac{m}{2\pi}\cdot\frac{2b}{x^2 - b^2}. Setting it to zero: x2=b2+mbπUx^2 = b^2 + \frac{mb}{\pi U}. With the numbers, x2=1+2=3x^2 = 1 + 2 = 3, x=±1.7321x = \pm1.7321, what the sim prints.

Key Takeaways

  • Zero vorticity makes u=∇ϕ\mathbf u = \nabla\phi; incompressibility makes ∇2ϕ=0\nabla^2\phi = 0, each value the average of its neighbours. Around a body with circulation ϕ\phi grows by Γ\Gamma per lap.
  • Laplace’s equation is linear, so stream, source (m2πln⁡r\frac{m}{2\pi}\ln r), vortex (Γ2πθ\frac{\Gamma}{2\pi}\theta) and doublet (μcos⁡θ/r\mu\cos\theta/r) superpose. Pressure does not: compute it from Bernoulli at the end.
  • ψ\psi is harmonic too, and ϕx=ψy\phi_x = \psi_y, ϕy=−ψx\phi_y = -\psi_x: streamlines cross equipotentials at right angles.
  • Stream plus doublet is a cylinder; a clockwise vortex adds lift. Stagnation points sit at sin⁡θ=−G/2\sin\theta = -G/2, merge at G=2G = 2, then leave the body (r=2.618ar = 2.618a at G=3G = 3).
  • Any formula w(z)w(z) is a flow, with dw/dz=u−ivdw/dz = u - iv.
  • Kutta–Joukowski: ρUΓ=1148.4375\rho U\Gamma = 1148.4375 N/m for the light-wing analog, from the surface pressure and from any momentum circle (half pressure, half downwash far away).
  • d’Alembert: zero drag for every body, because drag’s work would have nowhere to go.
  • Next, 11.8: Prandtl’s answer to the paradox. The missing drag lives in a boundary layer, a sheet of air so thin that potential flow is right everywhere outside it, while inside it no-slip drags the speed down to zero at the skin; we shrink Navier–Stokes to that layer, solve its profile by shooting, and find the adverse pressure gradient that tears it off the surface.
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.

Where this leads

These later lessons build directly on this one: