Rectilinear Flows

Brian G. Higgins∗
Department of Chemical Engineering and Materials Science
University of California
Davis, CA 95616

May 31, 2013

Introduction
Rectilinear flows are a special class of flows whose kinematics can be de-
scribed by v = v1 (x2 , x3 , t)ee1 . In these notes we will show how to derive the
governing equations that describe these flows, and discuss the properties
of these flows through several example flows. Rectilinear flows are a limit-
ing case of lubrication flows. Thus the analysis of such flows provides the
necessary backdrop for understanding the properties of lubrication flows.

Governing Equations
We will examine a special class of flows whose kinematics can be described
by
v = v1 (x2 , x3 , t)ee1 (1)
This motion is said to be rectilinear since there is only one component of
the velocity, and for convenience this component is taken to be in the e 1 -
direction. One can readily show that this motion is entirely consistent with
the equation of continuity for an incompressible flow. That is, the motion
satisfies
∂v1 ∂v2 ∂v3
∇·v = + + ≡0 (2)
∂x1 ∂x2 ∂x3
To put it another way, a rectilinear flow described by Eq.(1) satisfies con-
tinuity exactly. It is of interest to compute the components of the velocity
∗
email: bghiggins@ucdavis.edu

1
gradient ∇vv for such flows. Recall that ∇vv = (∂vi /∂xi )eej e i . Thus there are
at most two non-zero components of ∇vv :
0 0 0
 
∂v1
[∇vv ] =  ∂x2
0 0  (3)
∂v1
∂x3 0 0
A simple calculation shows that the non-linear terms in the equations of
motion vanish for a rectilinear flow:
∂vj ∂vj
v · ∇vv = vie i · e k e j = vi e i δik
∂xk ∂xk
∂vj
= vi ej (4)
∂xi
∂v1 ∂v1 ∂v1
= (v1 + v2 + v3 )ee1 ≡ 0
∂x1 ∂x2 ∂x3
as v2 = v3 = 0 for rectilinear flows. Similarly one finds that
 2
∂ v1 ∂ 2 v1

2
∇v= + e1 (5)
∂x22 ∂x23
Thus for a rectilinear flow the Navier-Stokes equations reduce to
∂v1
ρ e 1 = −∇p + µ∇2II v1e 1 + ρgg (6)
∂t
where ∇2II is the two-dimensional Laplace operator defined by
∂2 ∂2
∇2II = + (7)
∂x22 ∂x23
It is convenient to replace the gravitational body force ρ g with a scalar
potential φ. Let
φ = ρ g ·x (8)
such that ∇φ = ρ g . Next we define a modified pressure p̂ = p − ρ g · x .
Then the governing equation for rectilinear flow becomes
∂v1
ρ e 1 = −∇p̂ + µ∇2II v1e 1 (9)
∂t
If we examine the components of Eq. (9) we observe
∂v1 ∂ p̂
ρ =− + µ∇2II v1 (10a)
∂t ∂x1
∂ p̂
0=− (10b)
∂x2
∂ p̂
0=− (10c)
∂x3

2
It follows from these equations that the modified pressure p̂ is at most a
function of x1 and t:

p̂ 6= p̂(x2 , x3 ), p̂ = p̂(x1 , t) (11)

Moreover, it follows from Eq. (10a) that ∂ p̂/∂x1 is a constant, or at most a

function of time. For notational convenience let
∂ p̂
G(t) ≡ (12)
∂x1
so that we can write Eq.(10a) as
∂v1
ρ = −G(t) + µ∇2II v1 (13)
∂t
Figure 1 shows several examples of rectilinear flow geometries. Note for v1

Figure 1: Examples of rectilinear flow geometries

not to depend on the coordinate x1 , the flow cross-section must not vary
with x1 , and the entrance and exits effects should be negligibly small. For
the latter effects to be negligible, the flow channel must be sufficiently large
compared with the flow development length at the entrance and exit planes.
The method to quantify such effects is discussed in a separate lecture.

1 Examples of Rectilinear Flows

Steady flow in an elliptical pipe
In this example we consider steady flow on a pipe with an elliptical cross-
section defined by a major axis a and a minor axis b. The cross section of

3
the pipe is taken to be uniform in the flow direction x1 , and is described by
the following function:
x22 x23
f (x2 , x3 ) = + 2 −1=0 (14)
a2 b
The flow geometry is shown in Figure 2:

Figure 2: Flow in an elliptical pipe

The governing equation for steady rectilinear flow follows from Eq. (13):

µ∇2II v1 = G (15)

subject to the no-slip boundary condition along the pipe wall:

x22 x23
v1 = 0 on + 2 −1=0 (16)
a2 b
We look for solutions of the form
x22 x23
 
v1 = α + 2 −1 (17)
a2 b
where α is a constant to be determined. This form for the velocity field
automatically satisfies the no-slip boundary condition Eq. (16) and satisfies
Eq.(15) provided that  2 2 
G a b
α= (18)
2µ a2 + b2

4
Thus the velocity profile in the elliptical pipe is given by
 2 2  2
x2 x23

G a b
v1 = + 2 −1 (19)
2µ a2 + b2 a2 b

Clearly the velocity v1 is zero on the boundary of the ellipse. The constant
G can be determined from the flow rate Q:
Z Z
Q= v1 dx2 dx3 (20)

The above integral can be done numerically. For illustrative purposes let
us consider the simpler case when a = b, i.e., the flow channel is circular in
cross-section. Moreover, let r2 = x22 + x23 , and let R2 = a2 at the wall. Then

GR2  r 2 
v1 = ( ) −1 (21)
4µ R
The volumetric flow rate at any cross-section is the
Z R
πGR4
Z Z
Q= v1 dx2 dx3 = v1 2πrdr = − (22)
0 8µ
Recall that G defines the modified pressure gradient:
∂ p̂ dp 8µQ
G≡ = − ρ g1 /g = − (23)
∂x1 dx1 πR4
where g1 is the component of the body force acting in the x1 direction.
Solving for p we get
 
8µQ
p = ρg1 /g − x1 + constant (24)
πR4

Thus the pressure drop over a flow length L is given by

p0 − p1 8µQ
= − ρg1 /g (25)
L πR4

Rectilinear flow in a rectangular duct with free surface

In this example we consider a free surface flow in a rectangular duct of width
2b. The height of the liquid surface is taken to be h, and we assume that
the contact angle the liquid makes with the vertical walls of the duct is 90◦ .
Thus the free surface is parallel with the base of the duct. The origin of the

5
coordinate system is located along the center line of the channel floor. The
unit vector e 1 is aligned in the flow direction. The vectors e 2 , and e 3 , are
normal to the free surface and the side walls of the duct, respectively. For
rectilinear flow the governing equation for the component of velocity in the
flow direction e 1 is
∂v1
ρ = −G(t) + µ∇2II v1 (26)
∂t
As before G(t) is the forcing term that drives the flow. In this case it will
turn out to be the gravitational body force. The boundary conditions on
the velocity field are the no-slip conditions along the duct walls:

v1 (b, x2 ) = v1 (−b, x2 ) = 0, 0 < x2 < h (27)

Along the duct floor, the no-slip condition is

v1 (x3 , 0) = 0, −b < x3 < b (28)

Note that since v1 is always parallel to the channel walls, the no-penetration
condition v ·n
n = 0 is automatically satisfied. At the free surface the traction
condition must be satisfied:
 I
T − T II · n + ∇s σ + 2Hσn

n=0 (29)

Here T is the stress tensor of the liquid, n the outward directed unit normal
( in the e 2 direction) to the gas-liquid interface, σ the interfacial tension,
H the mean curvature of the interface, and ∇s the surface gradient . The
superscripts I, and II, denote the liquid and gas phases, respectively. If the
interface is flat (the case considered here), then 2H = 0. Further, we will
assume that the gas phase exerts negligible shear on the liquid, i.e., the gas
phase is sensibly stagnant so that

T I = −pI I (30)

where pI is the ambient pressure in the gas phase and I is the unit tensor.
For the liquid phase the stress tensor is given by

T II = −pII I + τ (31)

where τ = µ(∇vv + ∇vv † ) is the viscous stress tensor. With these simplifica-
tions the traction condition Eq(29) becomes

−pI I + pII I · e 2 − τ · e 2 + ∇s σ = 0


(32)

6
but
τ · e 2 = τij e i e j · e 2 = τi2 e i (33a)
∂σ ∂σ
∇s σ = e1 + e3 (33b)
∂x1 ∂x3
Equations (33) are vector equations. Clearly the first equation has a com-
ponent that is normal to the interface (i.e in the e 2 direction). Taking the
scalar product of Eq. (33a) with e 2 gives
−pI + pII + τ22 = 0 (34)
Recall that the kinematics of the flow are given by v = v1 (x3 , x1 )ee1 ; thus
the components of the viscous stress tensor are
 
∂v1 ∂v1
0 ∂x2 ∂x3
 ∂v1
[ττ ] =  ∂x 0 0  (35)

2
∂v1
∂x3 0 0

From inspection we deduce that the specified kinematics for the flow ensure
that τ22 = 0. So the normal stress balance at the gas-liquid interface reduces
to:
pI = pII (36)
Consider next the components of traction boundary condition that lie in the
interface. This vector can be decomposed into components that lie in the
e 1 and e 3 directions:
−τi2e i · e 1 + ∇s σ · e 1 = 0 (37a)
−τi2e i · e 3 + ∇s σ · e 3 = 0 (37b)
From the kinematics the components of the stress tensor that appear in Eq.
(37) are  
∂v1 ∂v2 ∂v2
τ12 = µ , τ32 = µ + =0 (38)
∂x2 ∂x2 ∂x3
So the shear stress boundary conditions at x2 = h, become in terms of the
velocity components:
∂v1 ∂σ
µ = (39a)
∂x2 ∂x1
∂σ
0= (39b)
∂x3
p = pII
I
(39c)

7
An additional boundary condition that must be satisfied at a liquid-gas
interface is the kinematic condition. For a non-evaporating fluid, the kine-
matic condition is a statement that the velocity of the interface must also
be the velocity of the fluid particle at the interface. Mathematically for a
interface described by h(x1 , x2 , t), the kinematic condition is

∂h ∂h ∂h
+ v1 + v2 = v3 (40)
∂t ∂x1 ∂x2
In our case h = constant, and v3 = 0, and thus the kinematic condition is
satisfied identically. We can now summarize the boundary value problem
that must be solved to determine the velocity v1 (x2 , x3 ) for steady flow in a
duct:

∇2II v1 = G (41a)
∂v1 ∂σ
µ = 0, = 0, at x2 = h (41b)
∂x2 ∂x3
v1 (b, x2 ) = v1 (−b, x2 ) = 0, for 0 < x2 < h (41c)
v1 (x3 , 0) = 0, for − b < x3 < b (41d)
I II
p =p , at x2 = h (41e)

Several comments are in order. First, in order to have rectilinear flow in

the duct, there cannot be a surface tension gradient in the x3 - direction.
However, it is possible to have a constant surface tension gradient in the flow
direction x1 . Recall that G(t) = ∂ P̂ /∂x1 = ∂p/∂x1 − ρ g1 . Since pI = pII
along the free surface, it follows that there cannot be any pressure gradients
in the flow direction. Hence G(t) = −ρ g1 . That is, the flow is driven by
the component of gravity acting in the x1 - direction.

Film flow down an incline plane

A simple application of the development pursued in the previous example
is steady film flow down an incline plane. If edge effects are negligible (the
width of the film is much larger than the film thickness h), it is permissible
to assume that the velocity field in the liquid film depends solely on the
coordinate perpendicular to the flow, in our case x2 . Thus the velocity field
is given by v = v1 (x2 )ee1 . The geometry of the flow is illustrated in Fig. 3.
For steady rectilinear flow without surface tension gradients, the bound-

8
ary value problem that we need to solve is:

d2 v1 G
2 = (42a)
dx2 µ
dp
0=− + ρg2 (42b)
dx2
v1 = 0, at x2 = 0 (42c)
dv1
= 0, at x2 = h (42d)
dx2
pI = pII , at x2 = h (42e)

Note in this case the body force is given by

g = g cos α e 1 − g sin α e 2 (43)

We can integrate Eq. (42b) and using the boundary condition Eq. (42e) we
obtain the hydrostatic head in the liquid film

p(x2 ) = pII − ρ g sin α(x2 − h) (44)

Integrating Eq. (42a) and applying BCs Eqs. 42c and 42d, we get

G x22
 
v1 (x2 ) = − hx2 (45)
µ 2

But recall that G = ∂ p̂/∂xi = −ρg1 = −ρ g cos α. Thus the velocity field
becomes
g cos α x22
 
v1 (x2 ) = − − hx2 (46)
µ 2
Let Q be the volumetric flow rate per unit width of film, then
h
Gh3 ρg cos α h3
Z
Q= v1 dx2 = − = (47)
0 3µ 3µ

Solving for the film thickness h we get

 3
3µ Q
h= (48)
ρ g cos α

9
Rectilinear flow in the plane
In this section we consider rectilinear flow in the plane. That is, we assume
that the flow has an arbitrary direction in the x1 − x2 plane, and further
the flow does not depend on coordinates in the the plane. Specifically we
consider a flow field with the following kinematics:

v = v1 (x3 , t)ee1 + v2 (x3 , t)ee2 (49)

The components of the velocity field for this flow are

∂v1 ∂v
∇vv = e3e1 + 2 e3e2 (50)
∂x3 ∂x3
One can easily deduce that the specified kinematics given by Eq. (49) ensure
that
∇ · v = 0, and v · ∇vv = 0 (51)
Substituting the velocity field Eq. (49) into the Navier-Stokes equations we
obtain the following equations for v1 ,v2 and p:

∂v1 1 ∂p ∂ 2 v1
=− +ν + g1 (52a)
∂t ρ ∂x1 ∂x23
∂v2 1 ∂p ∂ 2 v2
=− +ν + g2 (52b)
∂t ρ ∂x2 ∂x23
1 ∂p
0=− + g3 (52c)
ρ ∂x3
A compact notation is possible if we define the following two-dimensional
vector and gradient operators:
∂ ∂
v II = v1e 1 + v2e 2 , ∇II = e 1 + e2 (53)
∂x1 ∂x2
As before we can introduce a scalar potential for the gravitational body
force, viz., φ = ρ g · x so that the modified pressure is p̂ = p − ρ g · x . With
these definitions the governing equation of the v II and p̂ are:

∂vv II 1 ∂ 2v II
= − ∇II p̂ + ν (54a)
∂t ρ ∂x23
1 ∂ p̂
0=− (54b)
ρ ∂x3

10
We will now apply this above formulation to study steady flow between two
parallel planes separated a distance h apart. The lower plane is allowed
to move with velocity V such that V · e 3 = 0. That is, the lower plane
moves parallel to itself. We would like to determine the velocity field and
volumetric flow rate perpendicular to the plane. Gravity is not important
and can be ignored. The governing equations and BCs for this motion are:
d2v II 1
2 = ∇II p (55a)
dx3 µ
v II (0) = V (55b)
v II (h) = 0 (55c)

Integrating with respect to x3 gives

1 x2
v II = ∇II p 3 + B x3 + C (56)
µ 2
where B and C are vector constants of integration. Applying the BCs at
x3 = 0 gives:
C =V (57)
and applying the BC at x3 = h gives
1 h V
B = − ∇II p − (58)
µ 2 h
Hence the velocity field becomes
1 x3
v II = ∇II p (x23 − hx3 ) − V (1 − ) (59)
2µ h
The volumetric flow rate perpendicular to the planes is
Z h
h3 h
Q= v II dx3 = − ∇II p + V (60)
0 12µ 2
This equation shows that the volumetric flow rate need not be in the direc-
tion of the imposed pressure gradient, but depends also on the direction of
the boundary motion V .

Closing Remarks
The above development can be found in most textbooks on fluid mechanics.
The following references were helpful in preparing these notes.

11
References
[1] G. K. Batchelor, Introduction to Fluid Dynamics, Cambridge University
Press,1967

[2] B. R. Munson, D. F. Young, T. H. Okiishi & W. W. Huebsch, Sixth

Edition, Wiley & Sons, 2009

[3] F. S. Sherman, Viscous Flow, McGraw-Hill, 1990

[4] S. Whitaker, Introduction to Fluid Mechanics, Kreiger Publishing

Co.,1968

12