Quick Visit to Bernoulli Land Although we have seen the Bernoulli equation and seen it derived before, this next note shows its derivation for an uncompressible & inviscid flow. The derivation follows that of Kuethe &Chow most closely (I like it better than Anderson).1 Start from inviscid, incompressible momentum equation ∂u 1 + u ⋅ ∇u = − ∇ p ∂t ρ There is a vector calculus identity: 1 2 u • ∇u = ∇ u − u × ( ∇ × u ) 2 ω ,vorticity ⇒ ∂u 1 2 1 + ∇ u + ∇p = u × ω ∂t 2 ρ From here, we can make the final re-arrangement: 1 ∂u 2 ∇ p + ρ u = ρu × ω − ρ 2 ∂t Two common applications: 1. Steady irrotational flow ∂u ω=0 =0 ∂t Irrotational Steady 1 2 ⇒ ∇ p + ρ u = 0 2 ⇒ p+ 1 1 2 ρ u = const. for entire flow 2 Kuethe and Chow, 5th Ed. Sec 3.3-3.5 Quick Visit to Bernoulli Land 2. Steady but rotational flow ∂u ω≠0 =0 ∂t Rotational Steady 1 2 ⇒ ∇ p + ρ u = ρu × ω 2 This is a vector equation. If we dot product this into the streamwise direction: u s ≡ ← streamwise direction u 1 2 ⇒ s • ∇ p + ρ u = ρ s • (u × ω ) 2 =0, ( u ×ω ) ⊥u ⇒ 1 d 2 p+ ρ u =0 2 ds ⇒ p+ 1 2 ρ u = const. along streamline 2 Vortex Panel Methods2 Step#1: Replace airfoil surface with panels m i+1 i Original airfoil i-1 4 3 2 m+1 1 m-panels (m+1 nodes) Step #2: Distribute singularities on each panel with unknown strengths In our case we will use vortices distributed such that their strength varies linearly from node to node: Recall a point vortex at the origin is: φ=− Γ Γ y θ =− tan −1 2π 2π x 2 Kuethe and Chow, 5th Ed. Sec. 5.10 16.100 2002 2 Quick Visit to Bernoulli Land A point vortex at xˆ , yˆ is: φ=− Γ y − yˆ tan −1 2π x − xˆ Next, consider an arbitrary panel: (xj,yj) sj j Vortex of strength γ(sj)ds Sj expanded view γ(sj) dsj Midpoints will be (xj,yj) j+1 (xj+1,yj+1) At any s j , we will place a vortex with strength γ ( s j ) ds : ⇒ dφ ( x, y ) = − γ ( s j )ds −1 y − yˆ j tan x − xˆ j 2π where xˆ j ≡ x j + ( x j +1 − x j ) sj Sj yˆ j ≡ y j + ( y j +1 − y j ) sj Sj Thus, the potential at any ( x, y ) due to the entire panel j is: Sj φ j ( x, y ) ≡ − ∫ 0 γ (sj ) y − yˆ j ds tan −1 x − xˆ j 2π We will assume linear varying γ on each panel: γ ( s j ) = γ j + (γ j +1 − γ j ) 16.100 2002 sj Sj 3 Quick Visit to Bernoulli Land With this type of panel, we have m+1 unknowns = γ 1, γ 2, γ 3...γ m −1,γ m, γ m +1 , so we need m+1 equations. Step#3: Enforce Flow Tangency at Panel Midpoints The next step is to enforce some approximation of the boundary conditions at the airfoil surface. To do this, we will enforce flow tangency at the midpoint of each panel. Panel method lingo: control point is a location where u • n = o is enforced. To do this, we need to find the potential and the velocity at each control point. The potential has the following form: freestream individual panel + ∑ φ = potential potential # panels Suppose freestream has angle α : m Sj φ ( x, y ) = V∞ ( x cos α + y sin α ) − ∑ ∫ γ (sj ) j =1 0 The required boundary condition is y − yˆ j ds tan −1 x − xˆ j 2π ∂φ ( xi , yi ) = 0 for all i = 1 → m ∂ni So, let’s carry this out a little further: m S j γ (s ) ∂φ j ∂ xi , yi ) = V∞ ( cos α i + sin α j ) • ni − ∑ ∫ ( ∂n j j =0 0 2π ∂ni component of freestream normal to surface of panel i And recall γ (s j ) = γ j + (γ j +1 − γ j ) −1 y − yˆ j tan ds = 0 x − xˆ j xi , yi normal velocity due to panel j at control point of panel i sj . Sj We can re-write these integrals in a compact notation: γ ( s j ) ∂ ∫0 2π ∂ni Sj −1 y − yˆ j tan ds = Cn1ij γ j + Cn 2ij γ j +1 x − xˆ j xi , yi Cn 1ij = Influence of panel j due to xi , y i node j on control point of panel i i.e. Cn1ij γ j = normal velocity from panel j due to node j on control point of panel i . 16.100 2002 4 Quick Visit to Bernoulli Land Cn 2ij = Influence of panel j due to node j + 1 on control point at panel i ⇒ Total normal velocity at control point of panel i due to panel j = Cn1ij γ j + Cn 2ij γ j +1 So, let’s look at the control point normal velocity So, for panel i , flow tangency looks like: ∑ (C m j =1 n1 ij ) γ j + Cn 2 γ j +1 = V∞ ( cos α i + sin α j )ini , for all i = 1 → m ij V∞ ni We can write this as a set of m equations for m+1 unknowns. Question: What can we do for one more equation? Step#4: Apply Kutta condition We need to relate Kutta condition to the unknown vortex strengths γ j . To do this, consider a portion of a vortex panel. γ ds 16.100 2002 5 Quick Visit to Bernoulli Land Put a contour about differential element ds U2 dn γds V1 V2 U1 ds Γ = − ∫ u ⋅ ds Recall: ⇒ γ ds = − [V2dn − U 2ds − V1dn + U1ds ] = (V1 − V2 ) dn − (U1 − U 2 ) ds Now let dn & ds → 0 : dn → O , γ ds = − (U1 − U 2 ) ds γ = U 2 − U1 , or γ = U top − U bottom , in general So, since the Kutta condition requires U top = U bottom at TE: γ t .e. = 0, Kutta condition For the vortex panel method, this means: γ 1 + γ m +1 = 0 Step#5: Set-up System of Equations & Solve 16.100 2002 6 Quick Visit to Bernoulli Land I11 I 21 I31 I41 I51 I61 I 71 I81 1 uin = 0@i = 1 uin = 0@i = 2 uin = 0@i = m Kutta I12 I22 I32 0 I33 0 0 V∞ n1 I19 γ 1 γ V∞n 2 2 . γ 3 . . . = − . . V∞n8 γ 8 1 γ 9 0 γ I V∞n Where I ij = total influence of node j at control point i I 37 = C n137 + C n 236 For example: 6 5 7 Cn236 4 8 Cn137 3 9 2 Nodes Control points 1 The problem thus reduces to Ax = b , or, using our notation Iγ = V∞n , which we solve to find the vector of γ ’s! 16.100 2002 7 Quick Visit to Bernoulli Land Step #6: Post-processing The final step is to post-process the results to find the pressures and the lift acting on the airfoil. L′ = ρV∞ Γ = ρV∞ ∫ γ ds airfoil So, for our method, this reduces to: L′ = m 1 ρV∞ ∑ (γ i + γ i +1 )Si 2 i =1 ⇒ Cl = m γ γ S = ∑ i + i +1 i 1 ρV∞2c i =1 V∞ V∞ c 2 L′ Vortex Panel Method Summary In practice, the vortex panel method used for airfoil flows is a little different than the strategy used in the windy city problem. Here’s a summary: Step #1: Replace airfoil surface with panels m 4 3 2 m+1 1 Note: the trailing edge is double-numbered ⇒ m + 1 points, m panels . Step #2: Distribute vortex singularities with linear strength variables on each panel γ(sj) sj γ ( s j ) = γ j + (γ j +1 − γ j ) Sj γ(j) γ γ(j+1) j γds sj ds 16.100 2002 8 Sj j+1 Quick Visit to Bernoulli Land This means we have m+1 unknowns: γ 1, γ 2, γ 3, ............γ m, γ m +1 Step #3: Enforce flow tangency at panel midpoints u • n = 0 at midpoint of every panel ⇒ m equations Step#4: Apply Kutta condition Kutta condition becomes: γ t .e. = 0 ⇒ γ 1 + γ m+1 = 0 ⇒ m + 1 equations & m + 1 unknowns 16.100 2002 9