Lecture 18 Notes: Euler Equations - Basic Solutions and Forces

18.354J Nonlinear Dynamics II: Continuum Systems 18 Lecture 18 Euler equations: basic solutions and forces In the limit where the flow is irrotational, we just need to find solutions to Laplace’s equation to obtain solutions to the Euler equations. Let’s write down a couple of these to gain some intuition: our aim being to acquire techniques to begin to think about airplane flight. 18.1 Point source We know from electrostatics that a solution of Laplace’s equation is just φ=− c , 4πr (449) where c would be the charge in an electrostatic problem. What does this solution correspond to for us? The velocity field c u= r (450) 4πr3 is a radial source or sink of fluid. 18.2 Uniform flow Another trivial solution is simply uniform flow φ = U · r, (451) which works for any constant velocity vector U . 18.3 Vortex solutions We can also guess solutions by separation of variables φ = f (θ)g(r), where θ and r denote cylindrical coordinates. Laplace’s equation in cylindrical coordinates is   1 ∂ ∂φ 1 ∂2φ r + 2 2 = 0. (452) r ∂r ∂r r ∂θ Plugging this Ansatz into the equation gives that   dg 1 d2 f r d r + = 0. g dr dr f dθ2 (453) Each term in (453) must be a constant, i.e. d2 f = −f k 2 , dθ2   d dg r = gk 2 . r dr dr (454a) (454b) For k = 6 0, f = C sin(kθ) + D cos(kθ), (455a) with continuity of u requiring k to be an integer. Turning to the radial part we guess that g = rα . The radial equation then requires that α = ±k, giving g(r) = Ark + Br−k . (455b) f = C + Dθ. (455c) g(r) = A + B ln r. (456) However, if k = 0 then and the radial part is given by So, the most general solution is φ(r, θ) = (A0 + B0 lnr)(C0 + D0 θ) + ∞ X (Ck sin kθ + Dk cos kθ)(Ak rk + Bk r−k ) (457) k=1 The corresponding velocity field is u = ∇φ = ∂φ 1 ∂φ θ̂. r̂ + ∂r r ∂θ (458) Putting in our general solution with k = 0, we get u= B0 (C0 + D0 θ) D0 r̂ + (A0 + B0 lnr)θ̂ r r (459) Setting B0 = 0 to obtain a continuous flow field, we get a flow with no radial component, and angular component D uθ = , (460) r which is irrotational by virtue of its construction. This is called a point vortex solution. If we consider the circulation about a loop containing the origin Z Z 2π u · dl = uθ rdθ = 2πD = Γ (461) 0 Thus D = Γ/2π, where Γ is the circulation about the point vortex. 18.4 Flow around a cylindrical wing Okay, so this isn’t the true shape of an airplane wing, but it’s a good place to start. Let’s see if we can calculate the lift and drag on a wing of length ` and radius R  ` moving with velocity u0 . In the frame of reference of the wing, the boundary conditions are φ → u0 x as r → ∞, ∂φ = 0 at r = R. ∂r Using the general solution found above, the first boundary condition requires that   D1 B1 φ = u0 r + cos θ + A0 (C0 + D0 θ). r (462a) (462b) (463) The second boundary condition gives us   R2 , φ = D0 θ + u0 cos θ r + r (464) where we have set C0 = 0 since ∇φ is all that matters. Physically we can see that D0 = Γ/2π, where Γ is the circulation about the wing (check this by integrating around a circular loop containing the wing). 18.5 Forces on the circular wing The lift and drag forces on the wing (length `) are respectively given by 2π Z FL = ` p(R, θ) R sin θdθ, (465a) p(R, θ) R cos θdθ. (465b) 0 2π Z FD = ` 0 We can determine the pressure distribution from Bernoulli’s Law ρ p = p0 − (∇φ)2 2 r=R = p0 − ρ 2   Γ 2 2u0 sin θ − . 2πR (466) Putting this into the above relations we find that FD = 0. (467) This result is known as D’Alembert’s paradox, contracting the well-known fact that drag is acting on a moving body even in in low viscosity fluids. Furthermore, the lift on the wing is linearly proportional to the circulation about the wing; FL = Γρ u0 `. (468) Thus, the lift on the cylindrical wing is zero (unless it is spinning, so that Γ 6= 0!). The lift force due to rotation is also known as Magnus force. Our earlier discussion of singular perturbations suggests that D’Alembert’s paradox for inviscid flows arises as a consequence of the fact that we have neglected viscosity in the Euler equations. Historically, this was first realized by Prandtl. Another shortcoming of our above calculations is that wings are not circular and, maybe, if we consider an alternative shape we would find lift. This will be our next avenue of investigation. We could also be worried about the fact that our problem is 2D. However, given that the aspect ratio of a wing is roughly 10:1, it is acceptable to consider 2D flow. In the next part, we will study how things change if we alter the shape of the wing. To do so will require conformal mapping.