Lecture 12

Weston Barger

August 3, 2016

1 Conservation Laws
We begin to discuss a class of mathematical models (in the form of PDE’s) which are
derived from conservation laws. First, what is a conservation law?

Definition 1.1. A conservation law is an equation which accounts for all the ways that
the amount of a particular quantity can change.

This accounting is one of the basic principals behind mathematical modeling and can
be applied to many, many quantities. A few examples include mass, momentum, energy,
population, and vehicles on a road. This language is a little vague. So, let’s give an
example.

1.1 Derivation of a scalar conservation law

Suppose we are in a one-dimensional setting. Let us call it the x -axis. Suppose that the
quantity we are interested in is Q.

Example 1.2. The quantity Q could represent

 the number of cars in a traffic flow problem

 the number of particles in a system

 the energy of a physical system

 the number of people in a social dynamics problem

1
 Let u(x , t ) denote the density of Q, and let S = [a, b] denote a segment on the x -axis.
The quantity in S is given by
Z b
Q= u(x , t ) dx .
a

The quantity Q can change in two ways:

 Q could enter or leave through a or b, or

 Q could be created or destroyed in S .

It follows that the rate of change of Q in S , dQ

dt , is given by the rate at which Q enters or
leaves at x = a, plus the rate at which Q enters or leaves x = b, plus the rate at which Q
is created or destroyed within S . We make the following definitions.

 Let φ(x , t ) be the flux function. The function φ(x , t ) is the rate at which Q is
flowing past a position x at a time t . This function has units
unit of Q
φ(x , t ) ↔ .
time
We follow the convention that if φ(x , t ) < 0, Q is flowing to the left, and φ(x , t ) > 0
denotes rightward flow. Note that the total rate at which Q is flowing into S through
the endpoints is given by

φ(a, t ) – φ(b, t ).

 Let f (x , t ) be the source function. The source function describes the rate at which
Q is begin added or removed from a system at a point x at time t . The units of
f (x , t ) are
unit of Q
unit of time unit of Q · unit of length
f (x , t ) ↔ = .
unit of length unit of time
Note that the total rate at which Q is created or destroyed in S is
Z b
f (x , t ) dx .
a

2
Therefore, we set

dQ d Zb Z b
= u(x , t ) dx = φ(a, t ) – φ(b, t ) + f (x , t ) dx . (1.1)
dt dt a a

We call (1.1) a conservation law in integral form.

An alternative form of the integral conservation law (1.1) can be written if we make
the assumptions that ut , φx and f are all continuous. We can rewrite (1.1) as
Z b Z b Z b
ut (x , t ) dx = – φx (x , t ) dx + f (x , t ) dx ,
a a a

so that
Z b
[ut (x , t ) + φx (x , t ) – f (x , t )] dx = 0. (1.2)
a

If we assume that (1.2) holds for all a and b,then since we have assumed that the integrand
in (1.2) is continuous, the fact that the integral in (1.2) equals zero implies the integrand
must me zero i.e.

ut + φx = f . (1.3)

We call (1.3) a conservation law in differential equation form.

1.2 Constitutive relations

Even if we consider f (x , t ) as give, (1.3) is one partial differential equation for two quan-
tities u and φ. A constitutive relation relates u and φ. In many cases, such a relation
gives φ as a function of u. Then φ = φ(u), and we have

ut + φx (u) = ut + ux φ0 (u) = f .

Example 1.3. The inviscid Burgers equation

u + uux = 0

is a conservation law with f = 0 and φ = u 2 /2.

3
1.3 Examples of conservation laws
1.3.1 Plug flow chemical reactor

A plug flow chemical reactor consists of a long tube in which a chemical A is fed in at one
end, a reaction occurs, and the resulting chemical B flows out the other end. Let us use
a conservation law approach to derive an equation for the amount and distribution of A
throughout the tube.
Let a(x , t ) denote the concentration (e.g. kg/m) of A at position x and time t in the
tube. In this setting, our flux φ(x , t ) gives the rate (e.g. kg/s) at which a chemical A
passes through our pipe at a position x and time t . The source term f (x , t ) describes the
rate (e.g. kg · m/s) at which the chemical A is removed from the medium by the reaction.
Let us make the assumption that A is moving a constant speed c through the tube.
Then
mass
φ(x , t ) =
time
length mass
= ·
time length
= c · a(x , t ).

This is a constitutive equation. This type of flux function describes an advection process.
Here, changes in a are due to movements in the medium.
Within the tube, assume that the chemical reaction removes A at a rate proportional
to concentration i.e.

f (x , t ) = –ka(x , t ),

where k > 0 is a constant. The negative sign is to indicate that the chemical A is begin
removed.
With the flux φ(x , t ) and the forcing term f (x , t ) derived above, we get that the
conservation law in differential equation form is

at (x , t ) + cax (x , t ) = –ka(x , t ).

1.3.2 Diffusion

Suppose we have pollution spreading in stagnate water in a pipe. Let u(x , t ) denote the
concentration of the pollutant (in mass / length). Assume further that no pollution is

4
being created or destroyed in our pipe i.e. f = 0.

Figure 1: This figure shows the initial concentration of pollution in our pipe.

We expect that pollution will flow from areas with higher concentration to areas with
lower concentration. If ux (x , t ) > 0 then an area of higher concentration lies to the right,
so we expect the flow to be to the left i.e. φ(x , t ) < 0. Similarly, if ux (x , t ) < 0 then an
area of higher concentration lies to the left, so we expect the flow to to be to the right i.e.
φ(x , t ) > 0. This is visualized in Figure 1. The simplest way to capture the phenomenon
we just described is Fick’s law:

φ(x , t ) = –σux (x , t ),

where σ > 0 is the diffusion constant i.e. the willingness of the pollution to spread. So,
construction our differential law gives

ut + φx = f
⇒ ut + (–σux )x = 0
⇒ ut = σuxx .

We have arrived at the diffusion equation a.k.a the heat equation.

1.3.3 Traffic Flow

Let us turn our attention to the study of traffic flow on an interstate highway. We approx-
imate the number of cars per unit length by a continuous function u(x , t ). Let us make
the simplifying assumption that there are no exists or entrances i.e. f = 0. As before, we

5
need to determine the relationship between φ and u. Let v be the average rate at which
cars are traveling on the highway. Then

φ = Rate at which cars are passing x at t

number of cars
=
unit of time
number of cars unit of length
= ·
unit of length unit of time
= u · v.

Now, we need to determine how the density of cars on our stretch of highway relates to
the average velocity the cars are traveling i.e. we need to relate u and v . From your
experience on the road, you should be able to vouch for the fact that when u is large (
high car density) that v is low, and when u is low, v is at or near the speed limit. The
simplest way to do this is

v = v` – au,

where v` is the speed limit and a is some positive constant. We set v` /a = u` to be the
maximum density possible. We can rewrite the velocity as
!
u
v = v` 1 – ,
u`

for u ∈ [0, u` ]. Thus, the constitutive relation becomes

u2
!
φx = u · v = v` u – .
u`

Thus, our conservation law becomes

ut + φx = f
!
2uux
⇒ ut + v ` ux – =0
u`
!
2u
⇒ ut + v` 1 – ux = 0.
u`

6
2 Method of haracteristics
Recall that we used the method of characteristics to solve the wave equation utt = c 2 uxx
with initial data u(x , 0) = u0 (x ) and ut (x , 0) = 0. We used characteristic lines through the
xt -plane to track the transport of the initial data u0 (x ), resulting in a complete solution.
We will now use the method of characteristics to solve several conservation law initial value
problems of the form

ut + φx = f , u(x , 0) = u0 (x ). (2.1)

2.1 Advection equation

Suppose that our constitutive relation is

φ(x , t ) = cu(x , t ),

for some constant c. Then the intial value problem (2.1) becomes

 x ∈ (–∞, ∞), t ∈ (0, ∞),
ut + cux = 0, (2.2)
 u(x , 0) = u0 (x ).

The method of characteristics uses special curves in the xt -plane along which the PDE
(2.1) becomes an ODE. Choose a point x0 on the x -axis, and consider a curve

(x (t ), t ), where x (0) = x0 ,

i.e. (x (t ), t ) is a curve in the xt -plane which starts at x0 . Let û(t ) be the restriction of
u to the curve (x (t ), t ). That is, let û(t ) = u(x (t ), t ). As x (t ) moves along the curve
(x (t ), t ), the value û(t ) = u(x (t ), t ) changeds at a rate of û 0 (t ). By the chain rule
d d
û 0 (t ) = û(t ) = u(x (t ), t )
dt dt
∂u ∂u dx (t )
= + ·
∂t ∂x dt
!
dx (t )
= ut + ux .
dt
Now, if we were to select the curve (x (t ), t ) such that
dx (t )
= c, (2.3)
dt
7
then we would have
d û d
= u(x (t ), t )
dt dt !
dx (t )
= ut + ux
dt
= ut + cux .

Now, since we have assumed that u(x , t ) satisfies (2.2), we have that
d
u(x (t ), t ) = ut + cux = 0.
dt
Therefore, selecting x (t ) such that (2.3) holds, then u(x (t ), t ) is constant. In other words,
the function u(x (t ), t ) is constant when (2.3) is satisfied. Since we have assumed that
x (0) = x0 , we have the ode

dx (t )
= c, x (0) = x0 ,
dt
whose solution is

x (t ) = x0 + ct .

This curve is called the characteristic curve of the advection equation. Using the
fact that u is constant along characteristic curves, we can find the solution to (2.2) for all
points (x , t ) in the plane. Since u(x0 , 0) = u0 (x0 ), we have

u(x , t ) = u(x0 , 0) = u0 (x0 ) = u0 (x – ct ).

Example 2.1. Let us solve the problem


 x ∈ (–∞, ∞), t ∈ (0, ∞),
ut + 4ux = 0,
 u(x , 0) = arctan(x ).

The characteristic curve at the starting point (x0 , 0) is found by solving

dx (t )
= 4, x (0) = x0 .
dt
We get

x (t ) = x0 + 4t .

8
Since u is constant along (x (t ), t ), we have

u(x (t ), t ) = u(x (0), 0) = u0 (x0 ) = arctan(x0 ).

Thus,

u(x , t ) = arctan(x – 4t ).