ME 563 - Intermediate Fluid Dynamics - Su

Lecture 9 - Even more viscous flow examples

Reading: Acheson, §2.4–2.5.

In the last lecture, we began to consider flows with circular streamlines, for which the velocity fields
are given by

u = uθ (r, t) êθ .

For such a flow, the Navier-Stokes equation describing incompressibility, ∇ · u = 0, is obviously

satisfied. The Navier-Stokes equation for the time evolution of uθ is
 2 
∂uθ ∂ uθ 1 ∂uθ uθ
=ν + − 2 . (1)
∂t ∂r 2 r ∂r r

Considering flows with circular streamlines is a convenient way to illustrate the viscous diffusion of
vorticity. We’ll work through one of the book examples, that of the viscous decay of a line vortex.

1 Viscous decay of a line vortex

Define a line vortex as
Γ0
u= êθ , (2)
2πr
with Γ0 constant. For a flow with circular streamlines, the vorticity is computed as

1 ∂(ruθ )
ω =∇×u= .
r ∂r
For the line vortex flow of (2), then, the vorticity works out as
(
1 ∂(Γ0 /2π) undefined if r = 0
ω= =
r ∂r 0 if r > 0.

That is, there is a vorticity singularity at r = 0. If we set this line vortex as the initial condition,
we want to know what happens to the flow as time proceeds.
Instead of proceeding by looking at the velocity, uθ , the book chooses to proceed using the
circulation:

Γ(r, t) = 2πruθ (r, t), (3)

where, obviously,

Γ(r, t)
uθ (r, t) = .
2πr
Expressed as an evolution equation for the circulation, the Navier-Stokes equation (1) is thus
 2 
∂Γ ∂ Γ 1 ∂Γ
=ν − , (4)
∂t ∂r 2 r ∂r

1
using
∂uθ 1 ∂Γ
= ,
∂t 2πr ∂t
∂uθ 1 ∂Γ Γ
= − , and
∂r 2πr ∂r 2πr 2
∂ 2 uθ 1 ∂2Γ 1 ∂Γ Γ
2
= 2
− 2 + 3.
∂r 2πr ∂r πr ∂r πr
The initial condition is clear from the definition of the line vortex (2) and the expression for
the circulation (3) –

Γ(r, t = 0) = Γ0 . (5)

We also have the boundary condition that uθ be finite at r = 0 and t > 0, namely

Γ(r = 0, t) = 0 for t > 0. (6)

This problem is very similar in form to the one about the impulsively moved plane boundary
we talked about in the last lecture. As we did there, we will look for a similarity solution, of the
form
√
Γ = Γ0 f (η), where η = r/ νt. (7)

Plugging (7) into (4), we get

"   #
2 2η
∂η ∂η ∂ 1 ∂η
f0 = ν f 00 + f0 2 − f0 , (8)
∂t ∂r ∂r r ∂r

using the notation df /dη = f 0 . For our definition of η, we have

∂η r
= − 1/2 3/2 ,
∂t 2ν t
∂η 1
= √ , and
∂r νt
2
∂ η
= 0.
∂r 2
Inserted in (8):
 
0 r 1 1 0 1
00
−f =ν f − f √ .
2ν 1/2 t3/2 νt r νt
r
√
Multiplying through by ν νt on both sides:

r2 r
−f 0 = f 00 √ − f 0 , giving
2νt νt
η 2
−f 0 = f 00 η − f 0 , or
2 
0 η 1
−f − = f 00 . (9)
2 η

2
As in the last lecture, we approach (9) first as a first-order differential equation, by using g(η) =
f 0 (η). This gives
 
η 1 dg
−g − = g0 = .
2 η dη

By separation of variables,
 
dg 1 η
= − dη.
g η 2

Integrating both sides, we get

η2
ln g = ln η − + C,
4
where C is an arbitrary constant. Taking the exponential of both sides,
2 /4 df
g = η eC e−η = .
dη
Integrating again to get f , we have
2 /4
f = −2eC e−η + A. (10)

Now, we need to apply the initial and boundary conditions. By the initial condition (5), at
t = 0, r > 0 ⇒ η = ∞, f = 1. Plugged into (10), this gives us
2 /4
f (∞) = −2eC e−(∞) + A = 1, so A = 1.

The boundary condition (6) requires f = 0 for r = 0 and t > 0, i.e. f (η = 0) = 0. This gives
1
f (0) = −2eC + 1 = 0, i.e. eC = .
2
With these conditions, our solution for the circulation, Γ, is
 2

Γ = Γ0 1 − e−r /4νt ,

giving the velocity field as

Γ0  2

uθ = 1 − e−r /4νt .
2πr
The velocity, uθ , at different times (different values of νt), plotted against r, is shown in Fig. 1.

3
Figure 1: Plot of uθ vs. r for νt = 0, 0.5, 1, 2 and 4. From Batchelor, An Introduction to Fluid
Dynamics.