1 An illustrative calculation: a block sliding

down a liquid covered plane

Consider an inclined plane along which a thin lm liquid of continuously ow-
ing. Release a block of mass m, density
b
, and dimensions H, L and W, from
rest and ask how its center of mass translates as a function of time. Let us
denote by x the position of the block along the plane.
Although the block will both slide along the plane, and sink into the
uid lm, lets assume that the latter process is suciently slow (or perhaps
inhibited by surface tension or a mechanism attached to the block to maintain
its height above the plane) that we can focus on determining x(t). Then
we just have a one-dimensional problem. Clearly we need to understand the
frictional resistance exerted by the liquid lm. After that, this is essentially
an elementary physics problem with sliding friction.
First, the mass of the block is m =
b
HLW. We know apply Newtons
second law F = ma to the sliding ow. In the x-direction the block is acted Vectors are bold
faced characters.
on by gravity, which exerts a force mg sin directed along the positive x-
direction, and there is also a frictional force from the uid below the block.
The ow under the block basically resembles the shearing motion introduced
when providing a rst denition of viscosity, and so the force that resists the
motion is directed opposite the sliding direction with magnitude

h
dx
dt
.(LW),
where LW is the area of contact between the block and the liquid lm. Thus,
by Newtons law we have
m
d
2
x
dt
2
= mg sin
LW
h
dx
dt
. .
friction
. (1)
We can rearrange this equation as
d
2
x
dt
2
+

hH
b
dx
dt
= g sin , (2)
which is to be solved with the initial conditions x(0) = 0 and
dx
dt
(0) = 0.
You should recognize this equation as being of a simple enough form that
you can solve for x(t) by two integrations. However, rather than rushing away
to do that we make two observations, which both add much more physical
insight into the system.
(i) Let us consider long times, when you might think that the system can
establish a steady-state sliding speed,
dx
dt
= constant, or
d
2
x
dt
2
= 0. Then,
1
equation (2) reduces to
dx
dt
=
hH
b
g sin

. (3)
This result provides a typical speed in the sliding block problem and
emphasizes that the block slides more slowly for higher uid viscosities.
(ii) How long does it take the block to establish this steady-state speed? Let
us now let v =
dx
dt
and write the governing equation as
dv
dt
+

hH
b
v = g sin , (4)
Next, we nondimensionalize this equation. By performing this relatively
simple step we will identify the representative or characteristic speed of
the block and the typical time over which the velocity changes from its
initial state of rest to its nal steady speed. In this case we will learn
something even before we solve the detailed dierential equation.
Let V = v/v
c
where v
c
is a constant you are to determine. It should
have dimensions of speed. Also, let T = t/t
c
where t
c
is a constant with
dimensions of time. Substituting into equation (4) and rearranging gives Recall vc is a
constant, so by
the chain rule:
dv
dt
=
d(V v
c
)
dt
=
v
c
dV
dt
; also
d
dt
=
1
t
c
d
dT
dV
dT
+
t
c

hH
b
. .
denes t
c
V =
t
c
g sin
v
c
. .
denes v
c
. (5)
The form of the equation suggests that if we seek simple multiplicative
coecients in the nal equation then we identify
t
c
=
hH
b

and v
c
= t
c
g sin =
hH
b
g sin

(6)
Hence it remains to solve the simplied equation
dV
dT
+ V = 1 , with V (0) = 1, (7)
which we note has no parameters, and so clearly indicates that the form
of the solution is independent of the uid viscosity, the block density
or dimensions, or the plane inclination. This equation has the solution
V (T) = 1 e
T
.
Sketch solution; emphasize dimensionless format, scale of speed and time
Final remark: Of course, the block also will sink into the lm and so the
separation distance between the block and the plane h(t) decreases in time.
Perhaps later in the course we will see how to analyze this aspect of the problem
and so solve the coupled problem of sliding with decreasing separation distance.
2
2 An introduction to dimensional analysis
Here we introduce the idea of dimensional analysis and nondimensional pa-
rameters. We will use these ideas throughout the course and later will try to
be more systematic since there is a formal mathematical description that can
be useful. For now we will content ourselves with illustrating the spirit and
style of the ideas and give a few examples.
In your previous science and mathematics courses, you have made mathe-
matical models, developed equations, obtained analytical solutions, etc. There
was probably one idea that was overlooked when you had your rst few courses
and this concerns some structural features of any problem statement that gen-
erally constrain the form of the answer. Here we illustrate how the dimensions
used to express a given problem or question provide valuable information about
possible forms of the answer.
2.1 An example from introductory mechanics
A projectile of mass m has initial height h above the ground. It is projected
horizontally (x) with velocity v
0
. Find the horizontal displacement when the
projectile hits the ground. We neglect resistance from the air, but do note
that later in this course we will learn how to calculate such resistance.
Here is how you would proceed in an introductory course. Recall that in
a uniform gravitational eld (g is the gravitational acceleration) the vertical
motion z(t) of the projectile follows from
z(t) =
1
2
gt
2
+ h , (8)
so that the object hits the ground (z = 0) when t =
_
2h/g. The corresponding
nal horizontal displacement follows from x(t) = v
0
t = v
0
_
2h/g = x
nal
. This
algebraic formula, in one form or another, is typically how the answer is left
in most books and courses. However, notice that given the phrasing of the
question, the basic distance involved in the problem statement is h. So, how
large is the nal horizontal displacement x
nal
relative to the initial vertical
displacement? From the exact answer we can write Recognize that h
is the basic unit
of length given in
the problem
statement!
x
nal
h
. .
dimensionless distance
=

_
v
2
0
gh
. .
dimensionless
parameter
. (9)
3
This equation completely describes the solution in a compact and useful form.
Notice that this result says that the projected distance relative to the initial
height h is the same if the ratio v
2
0
/gh is the same. We refer to v
2
0
/gh as a
dimensionless parameter. Further, we can immediately observe that the ratio
x
nal
/h is doubled if h is reduced by a factor of four or if v
0
is increased by a
factor of 2.
We will return to this problem shortly and explain how the basic func-
tional form of equation (9), though not the multiplicative constant

2, could
have been predicted knowing only the dimensions of the quantities given in
the problem statement.
2.2 Dimensions and units
It is useful to distinguish the dimensions of a given quantity from the units
in which it is measured. In mechanics, which is concerned with force and
the corresponding motion, it is common to take as fundamental dimensions
mass (M), length (L) and time (T). Then, for example, by Newtons second
law, F = ma, we speak of force (F) as a derived quantity with dimensions
ML/T
2
. It is convenient to introduce the bracket notation [.] to represent the
dimensions of a quantity. For example, we would write [F] = ML/T
2
.
It is helpful to be familiar with the many variables that arise in common
science and engineering problems. For example, Table 1 indicates in the rst
column dierent variables, in the second column the dimensions of the variable,
and in the third column the dierent expressions for the variable using dierent
systems of units.
Remark: It is also useful on occasion to view the fundamental dimensions as
force (F), length and time. In this case, mass is a derived quantity with (do
this as an exercise) dimensions FT
2
/L.
2.3 Mathematical description of nature: dimensional
consistency
Consider your entire experience studying quantitative subjects. When making
comparisons, it is necessary to compare like with like. Thus, we take it
for granted that
(i) Equations must be dimensionally consistent. For example, The equation
could not be
x(t) =
1
2
gt
3
.
4
variable dimensions
1
systems of units
mass M kg, g, lb
m
time T sec, min, hr, years
length L m, cm, microns, ft, yard
velocity L/T m/s, cm/s, ft/min
acceleration L/T
2
m/s
2
, ft/s
2
force ML/T
2
kg m /s
2
=N, lb
f
= 32.2 lb
m
ft/s
2
torque ML
2
/T
2
kg m
2
/s
2
energy ML
2
/T
2
J = Nm; erg = g cm
2
/s
2
pressure M/L/T
2
Pascal = N/m
2
, psi = lb
f
/in
2
Table 1:
1
Here we are considering mass (M), length (L) and time (T) as the
fundamental dimensions. N = Newtons. You are probably familiar with SI
and cgs units, but not so familiar with the English system, which is basically
only used today in the US and United Kingdom (and one or two other places).
We will occasionally use the English system just to give you some exposure to
it.
x(t)
..
L
=
1
2
gt
2
. .
L
T
2
T
2
dimensions are: L = L (10)
(ii) A complete statement of a physical law must be independent of the
units used for representation and measurement. If this were not true,
then two observers using two dierent measurement systems could arrive
at dierent conclusions.
2.4 Dimensional analysis for arriving at the structure
of answers
(i) A ball of mass m is dropped from a height h and falls due to gravity g.
Determine the time to contact the ground.
Answer: We expect that the time can at most depend on m, h and g.
It is common to express the potential functional dependence as
= f(m, h, g) (11a)
dimensions: T M L L/T
2
(11b)
5
But the time does not have dimensions involving mass, so m cant ap-
pear in the answer since it is the only variable with dimensions involving
mass. Because the only variables remaining are h and g, we note that
the only way to obtain a result with dimensions of time is
_
h/g. Hence,
we see that the answer must have the form
= c
1
_
h/g or

_
h/g
= c
1
(12)
where c
1
is a constant. The ratio /
_
h/g is dimensionless and represent
the ratio of the actual fall time to the natural time,
_
h/g in the
problem statement.
(ii) A ball of mass m is shot vertically in the air with initial velocity v
0
.
The gravitational acceleration is g. Determine the maximum height h
to which the ball can rise.
Answer: The solution form is potentially
h = f(m, v
0
, g) (13)
Again, we conclude there can be no dependence on m, and as the only
way to obtain dimensions of length is v
2
0
/g (note that g has dimensions
L/T
2
, so we need the v
2
0
in the numerator to eliminate dependence on
time), we conclude
h = c
2
v
2
0
/g or
h
v
2
0
/g
= c
2
(14)
where c
2
is a constant.
(iii) Drive at speed v along a circle of radius R. Determine the acceleration
a.
Answer: Since a = f(v, R), then elementary dimensional inspection
yields a = c
3
v
2
/R. Also, we should reiterate that this form of argu-
ment does not determine the constant c
3
, nor the sign of the acceleration
(whether it is directed inwards or outwards).
(iv) Reconsider the example from section 2.1. We are interested in the nal
horizontal distance x
nal
achieved by the object projected horizontally
from height h. Read the problem statement. Clearly, if the only variables
6
involved in the solution are those mentioned in the statement, then at
most we can expect
x
nal
= f( m
..
M
, h
..
L
, v
0
..
L
T
, g
..
L
T
2
) (15)
In this case, we can rst conclude that the nal result must be inde-
pendent of m, since x
nal
has dimensions of length (L) and there are no
other variables with dimensions of mass (M). So, we can at most expect
the dependence of x
nal
on the other variables to be
x
nal
= f(h, v
0
, g) (16)
Now if we measure x
nal
relative to h then the only nondimensional
description is
x
nal
h
= f
1
_
v
2
0
gh
_
, (17)
which is precisely the functional form of equation (9) that was obtained
by obtaining the exact solution of the trajectory equations. Nevertheless,
even without knowing what is the precise functional form of the solution
(i.e what is f
1
in equation 17), we note that simply arriving at the
dimensionless form (17) is a signicant step forward. For example, we
can conclude that doubling v
0
has the same eect as decreasing g by a
factor of 4.
3 One more view of elementary dimensional
analysis: Motion of a mass on a spring
Let us examine in dierent ways the solution of the simple harmonic oscillator
of a mass m connected to a spring (spring constant k).
What are the units of k? Again, using M,
L, T as indicating
dimensions.
Answer: Force = kx [k] =
F
L
=
M
T
2
.
Now we know that this system undergoes (possibly damped) oscillations.
What is the typical period of the oscillation? In the case of damped vibrations,
what is the typical time over which the damping occurs? We refer to both
questions as asking about time scales.
7
The dierential equation for x(t): Let x(t) denote the displacement of
the spring from its equilibrium length. x(t) satises the dierential equation dots denote time
derivatives
m x =

forces = kx m x + kx = 0 . (18)
We typically wish to solve this equation subject to initial conditions
x(0) = x
0
and x(0) = v
0
. (19)
For this discussion, well choose v
0
= 0, so the spring begins at rest with
displacement (from equilibrium) x
0
.
Four ways to look at the idea of a characteristic time:
(i) What is the approximate period of the oscillation?
Answer by dimensional arguments alone: Clearly the period must
depend on the parameters in the problem. Therefore we expect that
= f(m, k, x
0
) as these are the only three parameters that appear in
the introduction above. Below we note the dimensions of each of the
quantities:
= f(m, k, x
0
) (20)
M
M
T
2
L
In order to obtain a quantity with units of time, we see that we must
eliminate the mass and so
_
m/k has units of time. Thus we write
(m/k)
1/2
(21)
and note that this type of inspectional (dimensional) analysis can not
determine the multiplicative constant (in this case 2 if we refer to the
period of the motion) in the equation.
(ii) Of course, for the initial conditions x(0) = x
0
and x(0) = 0, we can solve
equation (18) exactly. You should verify that the solution is
x(t) = x
0
cos
_
(k/m)
1/2
t
_
. (22)
From this exact solution we deduce that the period of the motion is
2 (m/k)
1/2
. It is thus natural to speak of the (undamped) spring mass
system as having a characteristic time scale
_
m/k since this is the typ-
ical order of magnitude over which x(t) changes (say from its maximum
to its minimum).
8
(iii) Another look at (18): Lets only think about the order of magnitude of
the terms in the equations, and indicate their dimensions. We observe
that when typical displacements, say x, occur on a time scale , then
the dimensional estimates of each of the terms in (18) are
mx

2
, kx = 0
2
m/k again (m/k)
1/2
(23)
Since equation (18) is linear in the displacement x(t), then x does not
enter the nal estimate in (23).
(iv) Nondimensionalize the equation: It is possible to write (18) so that all
the dimensional variables (and even the multiplicative constants) are
eliminated. Let is dene new, dimensionless variables according to
X = x/x
0
, T = t/
_
m/k (24)
Then, substituting into (18) and (19) we obtain an equation for X(T):

X + X = 0 , X(0) = 1 ,

X(0) = 0 . (25)
If you are to solve or plot the solution to this equation you will see that
the magnitude of (changes in) X is less than or equal to 1 (we typically
write O(1), and pronounce the big Oh as order), when changes of T
are O(1). Please do not be
alarmed by this
language. It is
simply useful in
more complicated
problems for
expressing ideas
related to
approximate
answers of
complex
problems.
3.1 A mass-spring system with damping
We have emphasized above that it is useful to express solutions in dimension-
less form. We will give one additional example of this style of thinking applied
to a dierential equation familiar from your previous mathematics and physics
courses.
Many basic models (mechanical, electrical, physiological) involve second-
order, constant coecient ordinary dierential equations (ODEs). For exam-
ple, a mass-spring oscillator with linear damping (coecient ) has a displace-
ment from equilibrium x(t) that is described by the equation
m
d
2
x
dt
2
+
dx
dt
+ kx = 0 . (26)
Let us assume that this equation is to be solved with the two initial conditions
x(0) = x
0
,
dx
dt
(0) = 0.
9
Upon a rst inspection of this problem you might think that understand-
ing the possible solutions requires independent specication of four parameters:
m, , k, and x
0
. In fact, this is not true: the structure of the solution only
involves one (dimensionless) parameter.
Making a dierential equation dimensionless: Let X = x/x
0
and T = t/t
c
,
where t
c
is a constant, with dimensions of time, which is to be determined.
Substituting into equation (26) and rearranging slightly gives
m
kt
2
c
d
2
X
dT
2
+

kt
c
dX
dT
+ X = 0 , with X(0) = 1,
dX
dT
(0) = 0. (27)
To make the coecients of the equation simple, we see that it is convenient to
choose t
c
=
_
m/k in which case the equation is
d
2
X
dT
2
+

mk
. .
dimensionless
parameter
dX
dT
+ X = 0 , with X(0) = 1,
dX
dT
(0) = 0. (28)
Therefore, the basic physical problem only involves ONE DIMENSIONLESS
parameter. The form of the parameter
2
/mk makes clear that increasing
by a factor of two is equivalent to increasing m or k by a factor of four.
Exercise: Consider instead equation (26) subject to the initial conditions
x(0) = 0,
dx
dt
(0) = v
0
. Put the problem statement in dimensionless form.
10