Exam problem category 3: Diffusion

The first part of this exam problem contains two MC questions.

Question 3.1 [3 marks]

Superposition of several solutions of sub-problems can be used to solve the diffusion equation.
Why?:

a) The diffusion equation is linear

b) The diffusion equation has homogenous boundary conditions

c) The diffusion equation is time dependent

d) The diffusion equation has only one coefficient

Question 3.2 [3 marks]

Zero flux boundary conditions can be applied using the following knowledge:

a) Recognising a fixed temperature applies a zero flux

b) Recognising in the mid-point of a symmetric analysis has a zero flux

c) Recognising zero flux is the same as a zero diffusion coefficient

d) Recognising steady-state conditions result in a zero flux

The second part of this exam problem contains open sub-questions to a single problem.

A large oil refinery is located above a sandy soil aquifer, which does not have any ground water flow.
The aquifer is horizontally oriented and has a thickness of 5m. A river is located 150m away from the
refinery, parallel to the refinery.

An environmental inspection found that there is a concentration of benzene under the refinery of 40
mass percent, and at a location directly between the river and the refinery (i.e. 75m away from
both), the concentration was found to be 20 mass percent and very close to the river 1 mass
percent. The refinery operator tells the environmental inspector that the pollution cannot have been
happening for a long time as at the river the pollution is minimal, and simply stopping the leak will
be sufficient.

To analyse whether the pollution indeed is a recent phenomenon the environmental inspector
models the pollution using the following governing equation:
𝜕𝐶
= 𝐷𝛻 2 𝐶
𝜕𝑡
where C is the concentration of benzene, t is time and D is the diffusion coefficient of benzene in
water, which has a value of 1.1 × 10−9 m2/s. The river maintains a zero benzene concentration in
the aquifer where it joins the river.

Question 3.3
Consider the diffusion problem only in the aquifer and in 1D. Sketch the problem domain,
highlighting the appropriate initial and boundary conditions. [3 marks]

Question 3.4
Using superposition, explain how the problem can be decomposed into more easily solvable parts
and how each part can be formulated such that after superposition the whole problem can be
solved. [6 marks]

Question 3.5
Explain using sketches how several individual solutions for each part of the solution can be used to
satisfy the initial conditions. [6 marks]

Question 3.6
Using the information that the environmental inspector found, advise whether the pollution is
recent, or has been happening for a long time. Defend your answer. [4 marks]

Answers
Answer to Question 3.1
a) The diffusion equation is linear
The equation is linear and therefore the change in any impulse is the same as the impulse being
applied separately and summed.

Answer to Question 3.2

b) Recognising in the mid-point of a symmetric analysis has a zero flux
In a symmetric analysis, there cannot be any flex in the centre (otherwise it is not symmetric). By
extending the domain (a ghost domain) to twice the real length a zero flux condition can be created.
This is known as mirroring.

Answer to Question 3.3

x unit
area

Boundary cond. Initial condition Boundary cond.

𝐶 = 40% 𝐶 = 0% 𝐶 = 0%

150m
Answer to Question 3.4
There are two steps. The boundaries are non-homogenous and there is also spatial/temporal
distribution.

We can split the solution into a steady state solution with non-homogenous boundary conditions (h)
and a transient (and spatial) solution with homogenous boundary condition (g):

𝐶(𝑥, 𝑡) = ℎ(𝑥) + 𝑔(𝑥, 𝑡)

We then need to adjust our initial conditions such that we are still solving the same problem. The
global initial condition 𝐶0 is modified with the steady state solution to create the initial condition for
g:

𝑔|𝑡=0 = 𝐶0 (𝑥) − ℎ(𝑥)

Then g can be solved by splitting into a temporal and spatial solution such that we can more easily
solve the heat equation.

𝑔(𝑥, 𝑡) = 𝑣(𝑥)𝑤(𝑡)
The initial condition shown for g holds for w.

Answer to Question 3.5

The modified initial condition (i.e. for g) can be determined to have the form of:

Steady state
x=150m

Global initial condition

x

g initial condition

The solution of g takes the form of:

 ∞
𝛼𝑛2 𝜋2 𝑡 𝑛𝜋𝑥
−
𝑇 = ∑ 𝐵𝑛 𝑒 𝐿2 sin ( )
𝐿
𝑛=1
The exact form of the equation is not important, but the fact that you have multiple individual
solutions which start and end at zero (the homogenous boundary condition). However they are all
different. See below:

x=150m

Ind. solution 2
x

Ind. solution 1

g initial condition

These can be summed (by careful selection of the coefficients) such that they add up to equal the
initial condition. A large number is needed to obtain a realistic result.

Answer to Question 3.6

The data points can be seen plotted below and fall on the steady state solution (a linear solution),
therefore the pollution has been occurring for a long time. The low diffusion coefficient confirms
that this would be many years.

C x=75m

Steady state x=150m

x
Grading schemes
Question 3.1 and 3.2
Right=100%, Wrong=0%

Question 3.3

Element Percentage of total

amount of points if
correct
1 Has domain dimensions 33%
2 Boundary conditions identified and at right locations 33%
3 Initial conditions identified within domain 33%

Question 3.4
Element Percentage of total
amount of points if
correct
1 Identify split to form steady state and solution with homogenous BCs 25%
2 Identify modification to BCs needed for g 25%
3 Identify split to time and space domain (multiplicative) 25%
4 Formulate the answer in correct equations 25%

Question 3.5
Element Percentage of total
amount of points if
correct
1 Sketch approximate initial conditions for both global and local 25%
problem
2 Demonstrate knowledge of equation form that satisfies homogenous 25%
BCs, i.e. zero at both edges
3 Sketch a series of individual solutions 25%
4 Describe how they can be summed to satisfy initial conditions 25%

Question 3.6
Element Percentage of total
amount of points if
correct
1 Plot data points from inspection 33%
2 Identify linear trend and match to steady state solution 33%
3 Convincing argument that diffusion coeff is low and therefore long 33%
timeframe to get to steady state