Chapter 7

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Chapter 7

Unsteady flows
Transient pressure flow
Assume: laminar flow at low Reynolds number, little effect of entrance region,
isothermal incompressible Newtonian flow, the only velocity component u z (r , t )

∂u z ∂u z ∂p 1 ∂ ⎛ ∂u z ⎞
0= ρ =− +μ ⎜r ⎟
∂z ∂t ∂z r ∂r ⎝ ∂r ⎠

Steady solution of Hagen-Poiseuille flow


⎡ ⎛ r ⎞2 ⎤ ∂p Δp 4u zmax μ
u =u
s max
⎢1 − ⎜ ⎟ ⎥ − = = constant =
z z
⎢⎣ ⎝ R ⎠ ⎥⎦ ∂z L R2

Assume:
1.solution is the sum of steady solution and an unknown transient function U(r,t)
2.transient pressure is identical to the steady profile even for the unsteady flow
∂p Δp ΔP − ρ g z L
u z = u zs + U (r , t ) − = = = constant
∂z L L
∂U 1 ∂ ⎛ ∂U ⎞ r U μt
ρ =μ ⎜r ⎟ s= Φ= τ=
∂t r ∂r ⎝ ∂r ⎠ R u zmax ρR 2

∂Φ 1 ∂ ⎛ ∂Φ ⎞
= ⎜s ⎟
∂τ s ∂s ⎝ ∂s ⎠
∂Φ
= 0 along s = 0 Φ = −(1 − s 2 ) at τ = 0 Φ = 0 on s =1
∂s

J o (λn s)
Φ = −8∑ exp(−λ2nτ )
n =1 λn J1 (λn )
3

Transient disappears in a
dimensionless time of the
order of τ = 1
ρR2
t∞ =
μ
R: equivalent radius
Quasi-steady flows
- draining of a tank through a capillary
Time dependence comes from the time
dependent pressure that drives the flow
p = ρ gH (t ) at z=0
Pressure source is hydrostatic head

∂p ρ gH (t ) ∂u z ρ gH (t ) 1 ∂ ⎛ ∂u z ⎞
− = C (t ) = ρ = +μ ⎜r ⎟
∂z L ∂t L r ∂r ⎝ ∂r ⎠
dV dH R
Q=− = − AT = ∫ 2π ru z dr
dt dt 0

Coupled integro-differential equation


If the flow is slow enough, we use a steady state model for one particular
feature of an unsteady but slowly varing flow

C (t ) R 2 ⎡ ⎛ r ⎞2 ⎤ π R 4 ρgH (t ) π R 4 Δp
uz = − ⎢1 − ⎜ ⎟ ⎥ Q= = − AT
dH Q=
4μ ⎢⎣ ⎝ R ⎠ ⎥⎦ 8μ L dt 8μ L
H πρ gR 4

= e −τ τ ≡ 8μ L A t
H o
T

Requires infinite time for complete drainage


for 90% drainage for 95% drainage for 99% drainage
τ = 2.3 when H H o = 0.10 24 μ L AT τ = 4.6 when H H o = 0.01
t∞ ≡
πρ gR 4
Squeezing flow
Time dependence comes from a time
dependent change in the geometry
Laminar creeping flow

Order of magnitude
Axial Q = 2π R 2 H& UR R R
= >> 1
Radial Q = 2π R × 2 H × U R H& 2H H

Radial velocity is much greater than axial velocity


(except very close to the plate)
∂u z ∂p ⎡ 1 ∂ ⎛ ∂u z ⎞ ∂ 2u z ⎤ ∂p
ρ =− +μ⎢ ⎜r ⎟+ 2 ⎥ =0 p = p(r , t )
∂t ∂z ⎣ r ∂r ⎝ ∂r ⎠ ∂z ⎦ ∂z
Since the disks are rigid bodies, uz along the surfaces is independent of
radial position -> assume uz is everywhere independent of r

1 ∂ ∂u 1 ∂ du
(rur ) + z = 0 (rur ) = − z = f ( z , t ) ur = 12 rf ( z , t )
r ∂r ∂z r ∂r dz

∂ur ∂p ⎡ ∂ ⎛1 ∂ ⎞ ∂ ur ⎤ Assume viscous effect dominates


2
ρ =− +μ⎢ ⎜ (rur ) ⎟ + 2 ⎥
inertial (accelerative) effect
∂t ∂r ∂
⎣ ⎝r r ∂r ⎠ ∂z ⎦

dp ∂ 2u r μ r ∂ 2 f rf H 2 ⎛ dp ⎞ ⎡ ⎛ z ⎞ ⎤
2
=μ 2 = ur = = ⎜ − ⎟ ⎢1 − ⎜ ⎟ ⎥
dr ∂z 2 ∂z 2 2 2 μ ⎝ dr ⎠ ⎢⎣ ⎝ H ⎠ ⎥⎦
BC : f = 0 on z = ±H
1 dp 3H& 3H& μ R 2 ⎡ ⎛ r ⎞2 ⎤
− =
H
∫−H
2π rur dz = 2π R 2 H& μr dr 2 H 3
p=
4H 3
⎢1 − ⎜ ⎟ ⎥
⎢⎣ ⎝ R ⎠ ⎥⎦
This pressure resists the movement of the disks toward each other,
so an external force is required to drive them together
R
F = 2π ∫ Tzz z =H r dr
0
∂u z ∂u z 1 ∂ 3μ H& ⎡ ⎛ z ⎞2 ⎤
Tzz = p − 2 μ 2μ = −2μ (rur ) = ⎢1 − ⎜ ⎟ ⎥
∂z ∂z r ∂r H ⎣⎢ ⎝ H ⎠ ⎦⎥
R 3π R 4 μ H&
F = 2π ∫ p (r )r dr =
0 8H 3 Stefan equation

If the disks are driven under constant force


3
dH 8 F H
H& = − = o
dt 3π R 4 μ
1 1 16 Fot
− =
H 2 H o2 3π R 4 μ

good for high viscosity, slow squeezing


Squeezing flow of inviscid fluid
Slip boundary conditoin
Radial velocity is independent
of the axial position

1 ∂ du H& H&
(rur ) = − z = C (t ) uz = − z ur = − r
r ∂r dz H 2H

⎡ ∂u ∂u ∂u ⎤ ∂p && H& 2 ⎞
∂p ρr ⎛ H
ρ ⎢ r + ur r + u z r ⎥ = − − = ⎜ − ⎟
⎣ ∂t ∂r ∂z ⎦ ∂r ∂r H ⎜⎝ 2 4 H ⎟⎠

For constant speed squeezing


ρr 2 H& 2 R ρπ R 4 H& 2
p=
8H 2
0 ∫
F1 = 2π rp (r ) dr =
16 H 2
Strictly from the unsteady
nature of the flow
Draining of a liquid film from a vertical plate
A uniform film of initial thickness H suddenly begins to
drain, and the initial amount of liquid ultimately drains
completely off the plate
Assume:
film thickness varies gradually in the x-direction
(nearly parallel flow)
Viscous thin film with low Reynolds number
(neglect inertia, surface tension)

∂u ∂ 2u ∂ 2u g⎛ z2 ⎞
=ν ( 2 + 2 ) + g u = ⎜⎜ hz − ⎟⎟
∂t ∂x ∂z ν⎝ 2⎠
Lubrication approximation Quasi-steady approximation:
almost parallel flow the flow field for any film thickness h(x,t) is the
same as the steady flow for uniform film thickness
h ( x ,t ) gh3
The flow rate per unit width in the y-direction q=∫ u ( z , t ) dz =
0 3ν
Any difference between the flow in and out of film must appear as a
change in film thickness
∂h ∂h ∂q ∂ gh3 gh 2 ∂h
qx − q x + dx = dx =− =− =−
∂t ∂t ∂x ∂x 3ν ν ∂x
h = Axα t β Nonlinear partial differential equation

g
β Axα t β −1 = − ( A2 x 2α t 2 β )α Axα −1t β
ν
1/ 2 1/ 2
⎛ν ⎞ ⎛ νx ⎞
A = ⎜⎜ ⎟⎟ h = ⎜⎜ ⎟⎟
⎝g⎠ ⎝ gt ⎠

1.Thickness is infinite at t=0


2.Thickness is zero at the top

Excellent at least after an initial period of time


as long as we are not concerned with the film thickness near the top plate
Leveling of a surface disturbance
Magnetic recording system
Slider velocity ~ 10m/s
Air gap ~ a few hundred nanometers or less
Lubricant to prevent contact ~ 30-50Å
In the order of molecular size, certain
phenomena are not accounted for by NS

Contact of the slider with the lubricant results in a furrow


-> how quickly the lubricant flows back into the furrow to restore the protection

Assume the disturbance to the film


thickness is simusoidal
H ( x) = H + h sin kx

k≡ h << H
λ
Assumption: lubrication approximation, quasi-steady state
∂ 2u x ∂ 2u x ∂p ∂p
0 = μ( 2 + 2 ) − =0
∂x ∂y ∂x ∂y
Neglect any effect of gravity because the film is so thin
Surface tension provides the dominant force for restoration of the uniform film
σ 1 d 2H
p ( x, t ) = =− 2 p = −σH ′′ = σhk 2 sin kx p′′ = −σhk 4 sin kx
R ( x, t ) R dx

p′ 2 a
ux = y + y+b p ' = ∂p / ∂x
2μ μ p′ 2 p′H
ux = y − y
u x = 0 at y = 0 2μ μ
∂u
μ x = 0 at y = H ( x, t ) ≈ H
∂y
H ( x,t ) H p′H 3 dQ p′′H 3 ∂H
Q=∫ u x dy ≈ ∫ u x dy = − =− =−
0 0 3μ dx 3μ ∂t
∂H ∂h dQ p′′H 3 σhk 4 sin kxH 3
= sin kx = − = =−
∂t ∂t dx 3μ 3μ
1 ∂h σk 4 H 3
=− ≡ −β h = ho exp(− β t )
h ∂t 3μ
Decay rate for the disturbance
A disturbance to an extremely thin film will decay very slowly
If one uses a lubricant of low viscosity,
1.Low viscosity lubricants will have a smaller load-bearing capacity,
and it will be easier for the head to crash through the lubricant
2.Since the high centrigugal force tends to produce a radial flow of
lubricant off the disk
-> design of new lubricants is important (i.e. that bond chemically to the
topmost solid layer of the disk

You might also like