Convective Extra Reading 2 Blasius Solution of BL
Convective Extra Reading 2 Blasius Solution of BL
The Blasius solution of the momentum equation for the Laminar Boundary Layer:
∂p x
• Flat plate, ∂ y = 0, no streamline curvature.
• Boundary layer flow, only diffusion in cross-stream direction.
• Steady state and steady flow.
• No significant streamwise acceleration (indirectly goes with the statement "flat plate," which distinguishes this
from a "wedge flow").
By continuity:
∂u ∂v
∂x + ∂y = 0
From the streamwise momentum equation:
1 dp ∂ 2u
u ∂u ∂x + v ∂u ∂y = − + ν 2 , where the pressure gradient term is zero. We can come to this
ρ dx ∂y
1 dp
conclusion by writing the momentum equation for y > δ :
du ∞
u∞ = − .
dx ρ dx
du ∞ dp
If dx = 0, then = 0. We can solve this set with the boundary conditions:
dx
u → u ∞ as y → ∞
•
• u = 0 @ y = 0 for all x
∂u ⎛ A ⎞ ∂u
∂x = ⎜⎝ P ⎟⎠ ∂ x , etc. Substituting into the boundary layer equations:
Continuity: Momentum:
⎛ BP ⎞ ⎛ BP ⎞ ⎛ νP ⎞ 2
∂u + ⎜⎜ ⎟ ∂v =0
∂ x ⎝ AQ ⎟⎠ ∂ y u ∂u + ⎜⎜ ⎟v ∂u
⎟ = ⎜⎜ ⎟∂ u
∂ x ⎝ AQ ⎠ ∂ y ⎝ AQ 2 ⎟⎠ ∂y
2
C A =1, u
which gives invariant
u∞
The above terms are invariant. We have transformed the problem in new variables without any of the old variables "left
over" and without changing the variables (no left over stretching factors). The transformation was successful. But, any
combination of the new variables is also invariant. Lets look for a more suitable form. Looking for a form where y is
singled out:
⎛ uy 2 ⎞ u∞ u∞ y
⎜ ⎟ = y = Re x which is the canonical (simplest) form.
⎜ νx ⎟ u νx x
⎝ ⎠
vx ∞ ⎛u ⎞ u v u∞x v u
Singling out v: uy y ⎜⎜ ⎟⎟ = = Re x and singling out u: . Now that we have
⎝ νx ⎠ u ∞ u∞ ν u∞ u∞
this grouping, assign new variable names to these quantities:
η = y ∞ ; f1 (η ) = ; and f 2 (η ) =
u u v u∞x
νx u∞ u∞ ν
Now look at the boundary conditions:
u = 0 @ y = 0, x > 0 transforms to f1 (0 ) = 0
v = 0 @ y = 0, x > 0 transforms to f 2 (0 ) = 0
u = u ∞ when x = 0, y > 0 to f1 (∞ ) = 1 .
Note that the last two boundary conditions collapsed into one. This suggests that we can reduce the number of
dependent variables.
We now have two dependent variables f1 (η ) and f 2 (η ) and one independent variable, η .
The original variables have been replaced by the new -- they remained invariant.
We could reduce f1 (η ) and f 2 (η ) to one dependent variable if we introduce the stream function, ϕ , such that;
∂ϕ ∂ϕ
u = ∂y and v = − ∂x .
The continuity eqn. becomes:
∂ 2ϕ ∂ 2ϕ
∂x∂y − ∂y∂x = 0
. Thus, it is always satisfied.
u → u ∞ as y → ∞ ; ϕ y → u ∞ as y → ∞ x > 0
ϕ y = f ′ (η )u ∞ ϕ yy = f ′′ (η )u ∞ ϕ yyy = f ′′′ (η )
u ∞2
u∞
νx νx
f (η ) ν u
ϕ x = − f ′ (η ) ϕ xy = − f ′′ (η )
yu ∞ y u∞ u∞
2x +
∞
2 x 2 x νx
Substituting into the momentum equation gives:
1
f ′′′ + ff ′′ = 0 The Blasius Equation (Continuity and Momentum)
2
With boundary conditions:
u=0 @ y=0 x > 0; ()
f′ 0 = 0
u → u ∞ as y → ∞ , x > 0 f ′(∞ ) = 1
u = u∞ at x = 0 , y > 0 f ′ (∞ ) = 1 Note that the last two collapse into one.
Also, from the momentum equation:
∂ 2u ∂u ∂u
ν = u +v
∂y 2 ∂x ∂y
at the wall, u = v = 0
∂2u
=0 f ′′′ (0 ) = 0 . Or f ′′ 0 ( ) is a constant (constant shear stress).
∂y 2
f ′′′ (0 ) + f (0 )f ′′ (0 ) = 0
1
From the Blasius equation at the wall,
2
f (0 )f ′′ (0 ) = 0 but,
⎛ ⎞
f ′′ (0 ) = ⎜ ∂u ⎟
1
This requires that: for all x.
⎝ ∂y ⎠ 0 u∞
u∞
νx
()
Thus, f ′′ 0 could not be zero (otherwise the shear stress would be zero for all x).
()
Therefore, it must be that f 0 = 0 . Since f (0 ) = ⎜
⎜
⎛
ϕ ⎟ ,
⎞
⎟
ϕ 0 =0.
⎝ νxu ∞ ⎠ 0
We can find the velocity profile if we can solve the Blasius Eqn.
1
f ′′′ + () ( )
ff ′′ = 0 f ′ 0 = 0 , f ′ ∞ = 1 , and f 0 = 0 . ()
2
One way to solve this is to nodalize the range of η into discrete ∆η , write the equations (below) for each and
solve them simultaneously, matching the f ′ (∞ ) = 1 boundary condition at the outermost node. Another way is the
"shoot and correct method." Here, one first guesses the value of the wall shear stress f ′′ 0 , then integrates the ()
following discretized set of equations stepwise from the wall to the edge of the boundary layer:
= f ′ (η ) = f ′ (y ),
⎛⎛ y ⎞ ⎞
f ′′⎜⎜ ⎜ ⎟ Re x ⎟⎟
u Re x τ f ′′ (η ) ⎝ ⎝ x ⎠ ⎠ ,
u∞ x = f ′′ (0 ) = f ′′ (0 )
τ0
v
u∞ Re x =
1
[ηf ′(η )− f (η )] , τ = µ
∂ 2ϕ
=
ρν 2
Re x
3
2 f ′′ (η ) ,
2 ∂y 2
x2
ν2 3 f ′′ (0 ) ,
cf
2 =
τ0
= Re x 2 f ′′ (0 ) =
ρu ∞ 2 2
x u∞ 2
Re x
∞
θ
( )
∞
ρu ⎛⎜ u ⎞ 1
θ = ∫ 1−
ρ∞u∞ ⎜⎝ u∞
⎟ dy ,
⎟
⎠
x
=
Re x ∫ f ′ (η ) 1− f ′ (η ) d η ,
0 0
∞
∞
δ∗ 1
δ
∗ ⎛
∫
= ⎜⎜ 1−
ρu
ρ ∞u∞
⎞
⎟⎟ dy , and
x
=
Re x ∫ (1− f ′(η ))dη .
0⎝ ⎠ 0
Using the tabulated results of Table 5-1.
f"(0) = 0.332
1
[ηf ′−f ]η→∞ = 0.86
2
∞ ∞
δ99
x =
4.6 = 0.017 , δ 99 = 2 mm ( ≈ 1/16 inch at x = 10 cm)
Re x
v∞ 0.86
u∞ = ,v = 0.05m / sec
Re x ∞
u∞θ
Reθ = ν = 180 Probably ready for transition, if it were in a disturbed flow (called "engineering flows").