You are on page 1of 6

BioMEMS & Microfluidics

Solutions to sheet 3, SS 23

M.Sc. Fulvia Del Duca, M.Sc. Sebastian Freko


Fachgebiet Neuroelektronik
Fakultät für Elektrotechnik und Informationstechnik

BioMEMS & Microfluidics SS 23


Exercise 1

A Newtonian fluid is confined between two parallel infinite planes coplanar to the x/z-plane (compare Figure 1).
The plates are located at y = ±h. Assume that all pressure differences are negligible and the plates move with
velocities of uL and uH (lower and upper plate, respectively) in the x-direction.

a) Starting from the Navier-Stokes equation, simplify and solve for the steady-state flow between the plates
as a function of y.
The Navier-Stokes equation
∂u
ρ + ρu · ∇u = −∇ p + ∇ · η ∇u .
∂t

∂t = 0), no convection (u · ∇ u = 0, since u ⊥ ∇ u


can be simplified by assuming a steady flow ( ∂u
throughout the system), only negligible pressure differences (∇ p = 0), and a Newtonian fluid (η =
const), this simplifies to:
0 = η ∇2 u .

Taking into account all symmetries in the system, we can further simplify this to

∂2
0=η u,
∂y2

which integrated with respect to y gives:



C1 = η u.
∂y

Since η ∂y u = τvisc,xy , this shows that the viscous stress in a Couette flow is constant. We can continue
by integrating with respect to y to get
C1
u= y + C2 .
η
Inserting the boundary conditions u(y = −h) = uL and u(y = h) = uH yields
 
1 y
u= ( uH − uL ) + uH + uL ,
2 h

a) b)

uH

2h uL
y 1 cm
0.5 cm

x
z

Figure 1: Schematic depictions of the systems discussed in Ex. 1 and 2. a) As described in Ex. 1, a fluid is confined
between two plates moving with uL and uH . b) In Ex. 2, two fluid reservoirs filled to 1 cm and 0.5 cm are connected via a
small cylindrical channel.

BioMEMS & Microfluidics 1 SS 23


y

−h

Figure 2: Flow in the system described in exercise 1.

or, in three dimensions,  h i


1 y
2 ( u H − u L ) h + u H + u L 
u= .
 
 0 
0

Fig. 2 shows a plot of the flow field in an exemplary x/y-plane.

b) Evaluate the strain rate ε and the stress τ in the system. How does the magnitude of the viscosity affect
u, ε, and τ?
The strain and stress in the system are given by
   
uH − uL − uL
0 4h 0 −p η uH2h 0
 u −u   u −u 
ε= H
 4h
L
0 0
 τ= H
η 2h
L
−p .
0 
0 0 0 0 0 −p

As can be seen, neither u nor ε are affected by the viscosity. This is because both u and ε are measures
of fluid movement. τ, however, is a measure of force per unit area and thus depends on the viscosity.
From τ = 2ηε − pδ, we see that—as expected intuitively—deforming a viscous fluid requires more
force.

Exercise 2

Consider a circular microchannel (radius R = 20 µm, length l = 5 cm). On both ends, the channel is connected
to cylindrical fluid reservoirs with a diameter of d = 4 mm. The reservoirs are filled with water to heights of
1 cm and 0.5 cm (compare Figure 1).

a) Simplify the Navier-Stokes equation for the microchannel. Restrict your discussion to the parts of the
channel that are far from the connection to the reservoirs.
Starting with the Navier-Stokes equation

∂u
ρ + ρu · ∇u = −∇ p + ∇ · η ∇u,
∂t

BioMEMS & Microfluidics 2 SS 23


we can simplify by assuming u ⊥ ∇u (i.e. u · ∇u = 0) and a Newtonian fluid. This leaves

∂u
ρ = −∇ p + η ∇2 u
∂t

to be solved (see below).

b) Assume quasi-steadiness and derive the velocity u(r, t).


If we assume quasi-steadiness, we can assume that the flow is in equilibrium with the respective pres-
∂u
sure at all times. Therefore, we can set ∂t = 0 to obtain

∇ p = η ∇2 u.

Using cylindrical coordinates and the symmetry of the system, we can write this to
 
∂p η ∂ ∂u
= r ,
∂z r ∂r ∂r

which—integrated—yields
r2 ∂p
u= + C1 ln r + C2 .
4η ∂z
Since u is finite for r → 0, C1 = 0. Further, we use the no-slip boundary condition to argue that
u(r = R) = 0, to obtain
R2 − r2 ∂p
u=− .
4η ∂z
dp ∆p
Next, we assume a linear pressure drop along the channel ( dz = l ) and yield

R2 − r 2
u (r ) = − ∆p . (1)
4ηl

At a given time, the pressure difference is given by the difference in hydrostatic pressure in both reser-
voirs:
∆p = ρg∆h . (2)

In order to evaluate the time dependence of the height difference, we can evaluate the flow between the
two reservoirs. Dividing the flow by the area of the reservoir yields the change of height with respect to
time in the corresponding reservoir:

dh1 Q 4Q
=− =− 2 for reservoir 1 and
dt A πd
dh2 Q 4Q
= = for reservoir 2,
dt A πd2

where A is the area of the reservoir’s base. The derivative of the height difference with respect to time
is thus given as
d∆h 8Q
=− 2. (3)
dt πd
As we are assuming Hagen-Poiseuille flow, the flow rate is given by

∆p
Q= ,
Rh

BioMEMS & Microfluidics 3 SS 23


b) 30
a) 8
20
i

6
| Q| 10−5 µL s−1

10

r [µm]
0
4
-10
h

2 -20
-30
0
0 0.5 1 1.5
h i2 2.5 3
time 106 s

Figure 3: Q-t-plot (a) and flow profile (b) far from the reservoirs for the Hagen-Poiseuille flow described in exercise 2.

8ηl
which using Eq. 2 and Rh = πR4
can be rewritten to

ρgπR4
Q= ∆h . (4)
8ηl

Combining Eqs. 3 and 4 gives the following differential equation.

d∆h ρgR4
= − 2 ∆h
dt d ηl

This classical differential equation can be easily shown to be solved by the following function.

ρgR4
∆h(t) = ∆h0 e−kt with k = (5)
d2 ηl

Ultimately, combining Eqs. 1, 2, and 5 yields


 
R2 − r2 ρg∆h0 ρgR4
u(r, t) = − e−kt with k = . (6)
4ηl d2 ηl

Figure 3 shows a plot of the mean flow through the channel as a function of time as well as the velocity
profile in the channel at a given time. As we see, the time dependence is exponential and the velocity
profile is of a parabolic nature.
ρR2
c) The characteristic time for pressure driven flows can be described by η . Comparing this to the solution
obtained above, is the assumption of quasi-steadiness justified?
d2 ηl
As we see, the above-given result is an exponential decay with a time constant of ρgR4
. Using the
given geometry and water as a fluid (ηH2 O ≈ 1 mPa s), this time constant evaluates to 5 × 105 s. The
ρR2
above-given characteristic time of the flow, however, is η = 4 × 10−4 s. We thus see that the fluid

BioMEMS & Microfluidics 4 SS 23


reacts to changes in the boundary conditions on the scale of 10−4 s, while the pressure changes in this
system happen on the scale of 105 s. Hence, the assumption of quasi-steadiness is justified.

BioMEMS & Microfluidics 5 SS 23

You might also like