You are on page 1of 11

L1.

3 Deformation of a fluid element, pressure, and the stress tensor

Recommended reading: -

At the completion of this lecture you should be able to…

• Define the rate of strain and rotation tensors and describe (in words) what these
tensors represent.
• Define pressure in a viscous fluid.
• Define the total and viscous stress tensor in a Newtonian fluid

1.3.1 Deformation of a fluid element.

In this section we will develop equations which can be used to describe the deformation of a
small ‘element’ of fluid. This element contains a fixed mass of fluid which initially occupies a
volume which corresponds to an infinitesimally small cuboid. By considering how this element
deforms over an infinitesimally short time period, we will show that the rate of deformation of
the fluid element can be described by the sum of a ‘rate of strain’ tensor and a ‘rate of rotation’
(or vorticity) tensor. In the following lectures we will relate the rate of strain tensor to the stress
tensor in the fluid. Once we have defined our stress tensor, then we will be able to analyse the
forces acting on our small fluid element, which will allow us to derive a set of partial
differential equations which describe the conservation of momentum within our fluid (the
Navier-Stokes equations).

Let’s consider the small element of fluid shown below. The component of stress (the force per
unit area) acting on the top, front and right-hand sides of the element are defined as acting in a
direction outwards from these surfaces (see the right-hand side of figure A below). Now, if the
fluid is at rest then the only stress acting on the surfaces of the element is due to the pressure.
Thus, 𝜎𝜎11 = 𝜎𝜎22 = 𝜎𝜎33 = −𝑝𝑝.

𝑝𝑝 𝜎𝜎22

𝑝𝑝 𝜎𝜎11

𝑝𝑝 𝜎𝜎33

Figure 1.3.A. Normal stresses oppose static pressure in a stationary fluid.


However, in a viscous flow, the stress will not (generally) be normal to the surface. In fact, it
usually contains an additional normal and a shear component which is proportional to the
viscosity of the fluid. In these flows it is convenient to consider the fluid stress to be the sum
of a pressure stress and a viscous stress.

A viscous stress occurs when the shape of an infinitesimally small fluid element is changed in
a flow. The motion of a fluid element may be considered to be the sum of four components –
(a) translation, (b) rigid-body rotation, (c) shearing motion and (d) compression/expansion.
Such motions for a 2D element are shown in figure B below.

(a) (b) (c) (d)

Figure 1.3.B. Translation, rotation, shear deformation and compression of a fluid element.

Only shearing motion (c) and compression (d) will give rise to a viscous stress which is
proportional to the rate of these motions since there is no change in shape of the fluid element
in translation (a) or rotation (b).

The ‘deformation’ of a small fluid element can be described by considering two points which
correspond to two vertices of our element. At time 𝑡𝑡, the element is a cuboid with sides of
length 𝛿𝛿𝑥𝑥1 , 𝛿𝛿𝑥𝑥2 and 𝛿𝛿𝑥𝑥1 each of which is parallel to the 𝑥𝑥1 , 𝑥𝑥2 and 𝑥𝑥3 axes. Let’s consider the
deformation of the edge which lies parallel to the 𝑥𝑥1 -axis. Point A is located at 𝐱𝐱 and point B
is located at either 𝐱𝐱 + 𝛿𝛿𝑥𝑥1 𝐢𝐢1, as shown in the figure below. By analysing the relative motion
of points A and B we can deduce the rate at which the element deforms.

𝐱𝐱 + 𝛿𝛿𝑥𝑥3 𝐢𝐢3

𝑥𝑥3

𝑥𝑥2
𝑥𝑥1

𝐱𝐱
𝐱𝐱 + 𝛿𝛿𝑥𝑥1 𝐢𝐢1
𝐱𝐱 + 𝛿𝛿𝑥𝑥2 𝐢𝐢2
Figure 1.3.C. Definition of rate-of strain for a fluid element.

The velocity at point A is given by 𝐮𝐮(𝐱𝐱, 𝑡𝑡), while the velocity at point B is 𝐮𝐮(𝐱𝐱 + 𝛿𝛿𝑥𝑥1 𝐢𝐢1 , 𝑡𝑡).
The difference in velocity between points A and B is equal to the rate at which the edge is being
deformed (stretched or compressed). We can define a ‘rate of deformation’ in the xi direction
of the edge of our fluid element aligned with the 𝑥𝑥1 -axis as being equal to the difference in
velocity between points A and B divided by the length of the side. Thus, the rate of deformation
in the 𝑥𝑥𝑖𝑖 -direction of the edge of our element parallel to the 𝑥𝑥1 -axis is

𝑢𝑢𝑖𝑖 (𝐱𝐱 + 𝛿𝛿𝑥𝑥1 𝐢𝐢1 , 𝑡𝑡) − 𝑢𝑢𝑖𝑖 (𝐱𝐱, 𝑡𝑡) 𝜕𝜕𝜕𝜕𝑖𝑖 (𝐱𝐱, 𝑡𝑡)
lim = .
𝛿𝛿𝑥𝑥1 →0 𝛿𝛿𝑥𝑥1 𝜕𝜕𝑥𝑥1

We could repeat the analysis for the other sides of the element and deduce that the rate of
deformation in the 𝑥𝑥𝑖𝑖 -direction of the edge of element parallel to the 𝑥𝑥2 - and 𝑥𝑥3 -axes is,
respectively, given by the following two expressions.

𝜕𝜕𝜕𝜕𝑖𝑖 (𝐱𝐱, 𝑡𝑡) 𝜕𝜕𝜕𝜕𝑖𝑖 (𝐱𝐱, 𝑡𝑡)


, .
𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3

Thus, we can define a rate of deformation tensor, D, which quantifies the rate of deformation
in the 𝑥𝑥𝑖𝑖 -direction of the edge of our fluid element aligned with the 𝑥𝑥𝑗𝑗 -direction

𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕1


⎡ ⎤
⎢𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 ⎥
⎢𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕2 ⎥
𝐃𝐃 = ⎢ .
𝜕𝜕𝑥𝑥 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 ⎥
⎢ 1 ⎥
⎢𝜕𝜕𝜕𝜕3 𝜕𝜕𝜕𝜕3 𝜕𝜕𝜕𝜕3 ⎥
⎣𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 ⎦

The components of D can be written using suffix notation as

𝜕𝜕𝜕𝜕𝑖𝑖
𝐷𝐷𝑖𝑖𝑖𝑖 = .
𝜕𝜕𝑥𝑥𝑗𝑗

𝐃𝐃 can be decomposed into a symmetric and an anti-symmetric part. The symmetric part will
be denoted 𝐒𝐒 and is defined

1
𝑆𝑆𝑖𝑖𝑖𝑖 = �𝐷𝐷 +𝐷𝐷 �,
2 𝑖𝑖𝑖𝑖 𝑗𝑗𝑗𝑗
or

𝜕𝜕𝜕𝜕1 1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕2 1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕3


⎡ � + � � + �⎤
⎢ 𝜕𝜕𝑥𝑥1 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥1 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥1 ⎥
⎢1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕2 1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕3 ⎥
𝐒𝐒 = ⎢ � + � � + �.
2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥2 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥2 ⎥
⎢ ⎥
⎢1 �𝜕𝜕𝜕𝜕3 + 𝜕𝜕𝑢𝑢1 � 1 �𝜕𝜕𝜕𝜕3 + 𝜕𝜕𝜕𝜕2 � 𝜕𝜕𝜕𝜕3 ⎥
⎣2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥3 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥3 ⎦
While the anti-symmetric part is denoted 𝛀𝛀 and is defined

1
Ω𝑖𝑖𝑖𝑖 = �𝐷𝐷 −𝐷𝐷 �,
2 𝑖𝑖𝑖𝑖 𝑗𝑗𝑗𝑗
or

1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕2 1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕3


⎡ 0 � − � � − �⎤
⎢ 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥1 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥1 ⎥
⎢1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕1 1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕3 ⎥
𝛀𝛀 = ⎢ � − � 0 � − �.
2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥2 ⎥
⎢ ⎥
⎢1 �𝜕𝜕𝜕𝜕3 − 𝜕𝜕𝜕𝜕1 � 1 �𝜕𝜕𝜕𝜕3 − 𝜕𝜕𝜕𝜕2 � 0 ⎥
⎣2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥3 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 ⎦

Hopefully you can see that 𝐷𝐷𝑖𝑖𝑖𝑖 = 𝑆𝑆𝑖𝑖𝑖𝑖 + Ω𝑖𝑖𝑖𝑖 where

1 𝜕𝜕𝜕𝜕𝑖𝑖 𝜕𝜕𝜕𝜕𝑗𝑗 1 𝜕𝜕𝜕𝜕𝑖𝑖 𝜕𝜕𝜕𝜕𝑗𝑗


𝑆𝑆𝑖𝑖𝑖𝑖 = � + �, Ω𝑖𝑖𝑖𝑖 = � − �.
2 𝜕𝜕𝑥𝑥𝑗𝑗 𝜕𝜕𝑥𝑥𝑖𝑖 2 𝜕𝜕𝑥𝑥𝑗𝑗 𝜕𝜕𝑥𝑥𝑖𝑖

The symmetric part, 𝐒𝐒, turns out to be the ‘rate of strain tensor’ which describes shearing and
compression of the element, while the anti-symmetric part 𝛀𝛀 is equal to the ‘vorticity tensor’
which describes the ‘solid-body’ rotation of the element.

In a Newtonian fluid, the rate of strain tensor, 𝐒𝐒, is assumed to be linearly related to the viscous
stress tensor, 𝛕𝛕. It can be shown (using clever mathematics) that for an isotropic fluid, this
linear relationship is given by the following general expression

𝜏𝜏𝑖𝑖𝑖𝑖 = 2𝜇𝜇𝑆𝑆𝑖𝑖𝑖𝑖 + 𝜆𝜆𝑆𝑆𝑘𝑘𝑘𝑘 𝛿𝛿𝑖𝑖𝑖𝑖 ,

where 𝜇𝜇 is the fluid viscosity, 𝜆𝜆 is a viscosity coefficient which will be defined in the next
lecture and 𝛿𝛿𝑖𝑖𝑖𝑖 is the Kronecker delta which is defined 𝛿𝛿𝑖𝑖𝑖𝑖 = 1 for 𝑖𝑖 = 𝑗𝑗 and 𝛿𝛿𝑖𝑖𝑖𝑖 = 0 for 𝑖𝑖 ≠ 𝑗𝑗.

1.3.2 Compression

In this section we will derive an expression for the change in volume of an infinitesimally small
fluid element. We will then show that show that the rate of change of the fractional volume of
the element is given by the sum of the diagonal components of the rate of strain tensor 𝑆𝑆𝑖𝑖𝑖𝑖 =
𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 .
Consider an infinitesimally small element which has dimensions𝛿𝛿𝑥𝑥1 , 𝛿𝛿𝑥𝑥2 and 𝛿𝛿𝑥𝑥3 at time . A

𝜕𝜕𝑢𝑢1
𝛿𝛿𝑥𝑥 𝛿𝛿𝛿𝛿
𝛿𝛿𝑥𝑥1 𝛿𝛿𝑥𝑥1 𝜕𝜕𝑥𝑥1 1

𝑥𝑥2

𝑥𝑥1

Time 𝑡𝑡 Time 𝑡𝑡 + 𝛿𝛿𝛿𝛿

short time later, at 𝑡𝑡 + 𝛿𝛿𝛿𝛿, the element has moved to a new location and it has expanded slightly
in all directions.

Figure 1.3.D. Expansion of a small fluid element.

At time 𝑡𝑡, the difference in velocity between the two sides of the element normal to the 𝑥𝑥1
direction is equal to 𝜕𝜕𝑢𝑢1 /𝜕𝜕𝑥𝑥1 multiplied 𝛿𝛿𝑥𝑥1 . Provided that 𝛿𝛿𝛿𝛿 is a sufficiently short period of
time, the distance which the element has expanded in the 𝑥𝑥1 -direction is approximately equal
to the difference in velocity at time 𝑡𝑡 multiplied by 𝛿𝛿𝛿𝛿. Thus, the change in length of the
element in the 𝑥𝑥1 -direction over the time interval 𝛿𝛿𝛿𝛿 is given by the following equation.

𝜕𝜕𝑢𝑢1
𝛿𝛿𝑥𝑥 𝛿𝛿𝛿𝛿 = 𝑆𝑆11 𝛿𝛿𝑥𝑥1 𝛿𝛿𝛿𝛿
𝜕𝜕𝑥𝑥1 1

Similarly, the change in length of the element in the 𝑥𝑥2 - and 𝑥𝑥3 -directions are respectively
given by the following equations

𝜕𝜕𝑢𝑢2 𝜕𝜕𝑢𝑢3
𝛿𝛿𝑥𝑥 𝛿𝛿𝛿𝛿 = 𝑆𝑆22 𝛿𝛿𝑥𝑥2 𝛿𝛿𝛿𝛿, 𝛿𝛿𝑥𝑥 𝛿𝛿𝛿𝛿 = 𝑆𝑆33 𝛿𝛿𝑥𝑥3 𝛿𝛿𝛿𝛿.
𝜕𝜕𝑥𝑥2 2 𝜕𝜕𝑥𝑥3 3

The volume of the element at time 𝑡𝑡 is

𝑉𝑉(𝑡𝑡) = 𝛿𝛿𝑥𝑥1 𝛿𝛿𝑥𝑥2 𝛿𝛿𝑥𝑥3 .

The volume of the element at time 𝑡𝑡 + 𝛿𝛿𝛿𝛿 is thus

𝑉𝑉(𝑡𝑡 + 𝛿𝛿𝛿𝛿 ) = 𝛿𝛿𝑥𝑥1 (1 + 𝑆𝑆11 𝛿𝛿𝛿𝛿)𝛿𝛿𝑥𝑥2 (1 + 𝑆𝑆22 𝛿𝛿𝛿𝛿)𝛿𝛿𝑥𝑥3 (1 + 𝑆𝑆33 𝛿𝛿𝛿𝛿)

𝑉𝑉(𝑡𝑡 + 𝛿𝛿𝛿𝛿 ) = 𝛿𝛿𝑥𝑥1 𝛿𝛿𝑥𝑥2 𝛿𝛿𝑥𝑥3 [1 + (𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 )𝛿𝛿𝛿𝛿 + 𝑂𝑂(𝛿𝛿𝑡𝑡 2 )]
Note that the term 𝑂𝑂(𝛿𝛿𝑡𝑡 2 ) in the equation above indicates that the remaining terms are of the
‘order’ 𝛿𝛿𝑡𝑡 2 . The fractional rate of change of the volume, or rate of dilation, of the element 𝑉𝑉 is
defined

1 𝜕𝜕𝜕𝜕 1 𝑉𝑉(𝑡𝑡 + 𝛿𝛿𝛿𝛿) − 𝑉𝑉(𝑡𝑡)


= lim ,
𝑉𝑉 𝜕𝜕𝜕𝜕 𝑉𝑉(𝑡𝑡) 𝛿𝛿𝛿𝛿→0 𝛿𝛿𝛿𝛿

1 𝜕𝜕𝜕𝜕 [(𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 )𝛿𝛿𝛿𝛿 + 𝑂𝑂(𝛿𝛿𝑡𝑡 2 )]


= lim
𝑉𝑉 𝜕𝜕𝜕𝜕 𝛿𝛿𝛿𝛿→0 𝛿𝛿𝛿𝛿
Note that in the limit 𝛿𝛿𝛿𝛿 tends to zero, we can neglect the contribution from the 𝛿𝛿𝑡𝑡 2 terms.
Therefore we have

1 𝜕𝜕𝜕𝜕
= 𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 = 𝑆𝑆𝑖𝑖𝑖𝑖
𝑉𝑉 𝜕𝜕𝜕𝜕
Thus, the fractional rate of change of volume (or rate of dilation) of the fluid is given by the
scalar quantity 𝑆𝑆𝑖𝑖𝑖𝑖 . Note that for an incompressible fluid, we expect the net change in volume
of our fluid element (and therefore the rate of dilation in the fluid) to be equal to zero. Thus,
for an incompressible fluid, 𝑆𝑆𝑖𝑖𝑖𝑖 = 0.

1.3.3 1D shearing motion (boundary layer/Couette flow).

Consider the deformation of the small fluid element (shown in figure E below) immersed in a
flow where the velocity is described by the following equation

𝐮𝐮 = [𝑢𝑢1 (𝑥𝑥2 ), 0,0].

Recall that the rate of strain tensor is defined by

𝜕𝜕𝜕𝜕1 1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕2 1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕3


⎡ � + � � + �⎤
⎢ 𝜕𝜕𝑥𝑥1 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥1 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥1 ⎥
1 𝜕𝜕𝜕𝜕𝑖𝑖 𝜕𝜕𝑢𝑢𝑗𝑗 ⎢1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕2 1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕3 ⎥
𝑆𝑆𝑖𝑖𝑖𝑖 = � + � , 𝐒𝐒 = ⎢ � + � � + �.
2 𝜕𝜕𝑥𝑥𝑗𝑗 𝜕𝜕𝑥𝑥𝑖𝑖 2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥2 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥2 ⎥
⎢ ⎥
1 𝜕𝜕𝜕𝜕
⎢ � 3 + 1� 𝜕𝜕𝜕𝜕 1 𝜕𝜕𝜕𝜕3 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕3 ⎥
� + �
⎣2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥3 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥3 ⎦

For this particular flow, the only non-zero components of the rate of strain tensor are

1 𝜕𝜕𝜕𝜕1
𝑆𝑆12 = 𝑆𝑆21 =
2 𝜕𝜕𝑥𝑥2

Using the definition of the viscous stress tensor yields

𝜕𝜕𝜕𝜕1
𝜏𝜏12 = 𝜏𝜏21 = 𝜇𝜇 .
𝜕𝜕𝑥𝑥2
Thus, even though the element is only deforming along the 𝑥𝑥2 direction, there are shear stresses
which act on four of the six surfaces of the element.

𝜏𝜏21
𝑥𝑥2

𝜏𝜏12 𝜏𝜏12
𝑥𝑥1

𝜏𝜏21

Figure 1.3.E. 1D Shearing motion (boundary layer/Couette flow).

The 𝜏𝜏12 component acts in the 𝑥𝑥2 direction on the surfaces which are normal to the 𝑥𝑥1 direction.
The direction of these components is shown by the horizontal arrows in figure E. Note that the
direction of the shear stresses on the upper and lower faces are opposite. Complimentary
stresses 𝜏𝜏21 act in the 𝑥𝑥1 direction on the surfaces which are normal to the 𝑥𝑥2 direction. These
ensure that the element does not accelerate in rotation at an infinite angular velocity as its size
becomes infinitesimal.

1.3.4 Vorticity

The vorticity tensor is defined by

1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕2 1 𝜕𝜕𝜕𝜕1 𝜕𝜕𝜕𝜕3


⎡ 0 � − � � − �⎤
⎢ 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥1 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥1 ⎥
1 𝜕𝜕𝜕𝜕𝑖𝑖 𝜕𝜕𝜕𝜕𝑗𝑗 ⎢1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕1 1 𝜕𝜕𝜕𝜕2 𝜕𝜕𝜕𝜕3 ⎥
Ω𝑖𝑖𝑖𝑖 = � − �, 𝛀𝛀 = ⎢ � − � 0 � − �
2 𝜕𝜕𝑥𝑥𝑗𝑗 𝜕𝜕𝑥𝑥𝑖𝑖 2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 2 𝜕𝜕𝑥𝑥3 𝜕𝜕𝑥𝑥2 ⎥
⎢ ⎥
⎢1 �𝜕𝜕𝜕𝜕3 − 𝜕𝜕𝜕𝜕1 � 1 �𝜕𝜕𝜕𝜕3 − 𝜕𝜕𝜕𝜕2 � 0 ⎥
⎣2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥3 2 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥3 ⎦

The vorticity vector is related to the vorticity tensor by the following expression

𝜔𝜔𝑖𝑖 = 𝜖𝜖𝑖𝑖𝑖𝑖𝑖𝑖 Ω𝑘𝑘𝑘𝑘

1.3.5 Pressure in a static fluid.

In your Thermofluids module last year you derived the following equation which governs the
pressure in a fluid at rest (𝑢𝑢𝑖𝑖 = 0)

𝜕𝜕𝜕𝜕
= 𝜌𝜌𝑔𝑔𝑖𝑖 ,
𝜕𝜕𝑥𝑥𝑖𝑖

where 𝐠𝐠 = [0, 0, −𝑔𝑔].


Consider the pressure in a liquid of constant density 𝜌𝜌 which is at rest. Integrating the equation
above with respect to 𝑥𝑥3 yields the following well-known equation describing the variation of
pressure within the liquid.

𝑝𝑝(𝑥𝑥3 ) = 𝑝𝑝(0) − 𝜌𝜌𝜌𝜌𝑥𝑥3

If the density of the fluid does vary, then the integration is not so straightforward. Question
1.3.4 considers just such a case.

Recall that in fluid static problems, the normal stress opposes the hydrostatic pressure and the
stress due to hydrostatic pressure acts equally in all directions (Pascal’s law). Therefore, the
stress tensor in a static fluid is given by

−𝑝𝑝 0 0
𝛔𝛔 = � 0 −𝑝𝑝 0 �, 𝜎𝜎𝑖𝑖𝑖𝑖 = −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 .
0 0 −𝑝𝑝

𝑝𝑝 𝜎𝜎22

𝑝𝑝 𝜎𝜎33
𝑝𝑝 𝑝𝑝 𝜎𝜎11 𝜎𝜎11

𝑝𝑝 𝜎𝜎33

𝜎𝜎22
𝑝𝑝

Figure 1.3.F. Normal stresses oppose static pressure in a stationary fluid.

1.3.6 Pressure in a viscous flow

When there is relative motion of particles within the fluid (which occurs when there are velocity
gradients within the fluid), this ‘rate of strain’ will produce an additional stress within the fluid
which is proportional to the viscosity of the fluid, 𝜇𝜇. This additional stress is known as the
viscous stress and is denoted by the viscous stress tensor 𝛕𝛕 which we learnt in the previous
lecture is linearly related to the rate of strain tensor, 𝐒𝐒, by the following equation

𝜏𝜏𝑖𝑖𝑖𝑖 = 2𝜇𝜇𝑆𝑆𝑖𝑖𝑖𝑖 + 𝜆𝜆𝑆𝑆𝑘𝑘𝑘𝑘 𝛿𝛿𝑖𝑖𝑖𝑖 ,

where 𝜆𝜆 is a viscosity coefficient which it yet to be determined and 𝛿𝛿𝑖𝑖𝑖𝑖 = 1 for 𝑖𝑖 = 𝑗𝑗 and 𝛿𝛿𝑖𝑖𝑖𝑖 =
0 for 𝑖𝑖 ≠ 𝑗𝑗.
Thus, the total stress within our fluid acting in the 𝑥𝑥𝑗𝑗 -direction on a ‘surface’ normal to the 𝑥𝑥𝑖𝑖 -
direction, 𝜎𝜎𝑖𝑖𝑖𝑖 , is the sum of the compressive stress produced by the hydrostatic pressure, −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 ,
and the viscous stresses within the fluid, 𝜏𝜏𝑖𝑖𝑖𝑖 , i.e.

𝜎𝜎𝑖𝑖𝑖𝑖 = −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 + 𝜏𝜏𝑖𝑖𝑖𝑖 .

Normally we understand pressure to be a scalar – it is the normal force acting on a differential


surface element, independent of the orientation of the element. However, inspection of the
components of the stress tensor

𝜎𝜎𝑖𝑖𝑖𝑖 = −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 + 𝜏𝜏𝑖𝑖𝑖𝑖 = −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 + 2𝜇𝜇𝑆𝑆𝑖𝑖𝑖𝑖 + 𝜆𝜆𝑆𝑆𝑘𝑘𝑘𝑘 𝛿𝛿𝑖𝑖𝑖𝑖 ,

indicates that in general the compressive components of the stress tensor are not equal, i.e.
𝜎𝜎11 ≠ 𝜎𝜎22 ≠ 𝜎𝜎33 (see below).

𝜎𝜎11 = −𝑝𝑝 + 2𝜇𝜇𝑆𝑆11 + 𝜆𝜆(𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 ),

𝜎𝜎22 = −𝑝𝑝 + 2𝜇𝜇𝑆𝑆22 + 𝜆𝜆(𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 ),

𝜎𝜎33 = −𝑝𝑝 + 2𝜇𝜇𝑆𝑆33 + 𝜆𝜆(𝑆𝑆11 + 𝑆𝑆22 + 𝑆𝑆33 ).

Thus, Pascal’s law does not hold in such a flow. Recall that we still need to determine a value
for 𝜆𝜆. This is done using a hypothesis due to the famous physicist George Stokes who proposed
that the fluid pressure is equal to the average of the 3 components of compressive stress acting
at a point in the fluid.
𝜎𝜎𝑘𝑘𝑘𝑘
𝑝𝑝 = − ,
3
which is consistent with the definition of pressure in a static fluid (although in a static fluid
𝜎𝜎11 = 𝜎𝜎22 = 𝜎𝜎33 ). Substituting the expressions for 𝜎𝜎11 , 𝜎𝜎22 and 𝜎𝜎33 into the equation above
yields 𝜆𝜆 = −2𝜇𝜇/3. It turns out that Stokes’ hypothesis works for monatomic gases, however,
in other fluids a small correction must be applied which yields

2
𝜆𝜆 = − 𝜇𝜇 + 𝜇𝜇𝐵𝐵 ,
3
where 𝜇𝜇𝐵𝐵 is the bulk viscosity. Note that stresses proportional to bulk viscosity usually only
become significant in extreme cases of compressible flow – e.g. in a shock wave which forms
in supersonic flow over the wing of military aircraft. Thus, for most low Mach number
situations encountered in practice, the effect of bulk viscosity can be ignored. We will ignore
the effect of bulk viscosity in this course.

Thus, we have the following expression which defines the stress tensor in a Newtonian fluid
which obeys Stoke’s hypothesis

2
𝜎𝜎𝑖𝑖𝑖𝑖 = −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 + 𝜏𝜏𝑖𝑖𝑖𝑖 = −𝑝𝑝𝛿𝛿𝑖𝑖𝑖𝑖 + 2𝜇𝜇𝑆𝑆𝑖𝑖𝑖𝑖 − 𝜇𝜇𝑆𝑆𝑘𝑘𝑘𝑘 𝛿𝛿𝑖𝑖𝑖𝑖 .
3
Let’s remind ourselves of the meaning of the stress tensors. The notation 𝜎𝜎𝑖𝑖𝑖𝑖 denotes that the
stress acts in the xj direction on a surface element whose normal is in the xi direction (see figure
G below). Note that the stress tensor is a function of position and thus the components of stress
acting on opposite faces of the element in figure G do not necessarily have the same magnitude.

𝜎𝜎33
𝜎𝜎31
𝜎𝜎32
𝑥𝑥3
𝜎𝜎22 𝜎𝜎13
𝜎𝜎12 𝜎𝜎21
𝜎𝜎11 𝜎𝜎11
𝜎𝜎23 𝜎𝜎12
𝑥𝑥1 𝜎𝜎13 𝜎𝜎23
𝜎𝜎21
𝑥𝑥2 𝜎𝜎22
𝜎𝜎32
𝜎𝜎31
𝜎𝜎33

Figure 1.3.G. stresses acting on a fluid element.

Questions

1.3.1. The velocity 𝐮𝐮 of a steady airflow above a horizontal surface located at 𝑥𝑥3 = 0 is

𝐮𝐮 = [𝑎𝑎𝑥𝑥3 − 𝑏𝑏𝑥𝑥32 , 𝑐𝑐𝑥𝑥3 , 0 ],

where 𝑎𝑎 = 1𝑠𝑠 −1 , 𝑏𝑏 = 0.1𝑚𝑚−1 . 𝑠𝑠 −1 and 𝑐𝑐 = 2𝑠𝑠 −1 . Calculate the numerical values of all the
components of the viscous stress tensor at 𝑥𝑥3 = 1m if 𝜇𝜇 = 1.83 × 10−5 Pa. s.

1.3.2. Show that the vorticity tensor is made up of the components of the vorticity vector 𝜔𝜔𝑖𝑖 .

1 0 −𝜔𝜔3 𝜔𝜔2
𝛀𝛀 = � 𝜔𝜔3 0 −𝜔𝜔1 �
2 −𝜔𝜔 𝜔𝜔1 0
2

1.3.3. Consider a fluid which is rotating as a ‘solid body’ so that

𝐮𝐮 = [− Ω𝑟𝑟sin 𝜃𝜃 , Ω 𝑟𝑟cos 𝜃𝜃 , 0 ],

where 𝑥𝑥1 = 𝑟𝑟cos 𝜃𝜃 and 𝑥𝑥2 = 𝑟𝑟sin 𝜃𝜃. Calculate the components of the vorticity tensor.

1.3.4. Assuming the temperature in the atmosphere varies linearly with height according to the
following equation,

𝑇𝑇(𝑥𝑥3 ) = 𝑇𝑇(0) − 𝜆𝜆𝑥𝑥3 ,


derive an expression for the variation in pressure within the atmosphere. Assume that air can
be regarded as a perfect gas. Calculate the pressure at 𝑥𝑥3 = 5km if 𝑔𝑔 = 9.81m.s-2, 𝑝𝑝(0) =
101kPa, 𝑅𝑅 = 287J.kg-1.K-1, 𝑇𝑇(0) = 293K, 𝜆𝜆 = 0.0065K.m-1.

1.3.5. The velocity 𝐮𝐮 of a steady airflow above a horiztonal surface located at 𝑥𝑥3 = 0 is

𝐮𝐮 = [𝑎𝑎𝑥𝑥3 − 𝑏𝑏𝑥𝑥32 , 𝑐𝑐𝑥𝑥3 , 0 ]

where 𝑎𝑎 = 1𝑠𝑠 −1, 𝑏𝑏 = 0.1𝑚𝑚−1 𝑠𝑠 −1 and 𝑐𝑐 = 2𝑠𝑠 −1. Calculate all components of 𝜕𝜕𝜏𝜏𝑖𝑖𝑖𝑖 /𝜕𝜕𝑥𝑥𝑗𝑗 .

You might also like