EART60061 Chap.10 Diffusion, Fick's Law and Dispersion Simulation
Diffusion, Fick's Law and Dispersion Simulation
Diffusion and Dispersion links the statistics of measurement to the modelling of contaminant spread. Random molecular motion means a diffusing pollutant spreads into a normal-shaped profile, and Fick's law makes the diffusive flux scale with the concentration gradient.
Cutting distance into cells and time into steps turns the law into difference equations solved by arithmetic, and matching the result with a widening normal curve shows the two descriptions agree. The lake spill exercise starts from a single spike, conserves mass, and reveals a limiting diffusion coefficient of about one third, above which the grid produces artefacts.
A final simulation reproduces dispersion with 500 randomly moving particles, using conditional statements and iteration as a first step towards programming. It also explains why realistic free-water diffusion is far too slow to see in 100 steps, and why a real lake would mix by convection and wind as well.
What this chapter covers
- 01
Diffusion as random molecular motion
- 02
Fick's law and concentration gradients
- 03
Difference equations across a lake
- 04
Matching a diffusing profile with a normal curve
- 05
The limiting diffusion coefficient
- 06
Conservative pollutants
- 07
Random-walk simulation and iteration
- 08
Mass conservation and degradation
Worked example · free
Advance a pollutant spike two steps with a diffusion coefficient of 0.2
- 1Step 1: centre = 1 - 2 times 0.2 = 0.6; each neighbour = 0.2.
- 2Step 2: centre = 0.6 + 0.2 times (0.2 - 1.2 + 0.2) = 0.44; each neighbour = 0.2 + 0.2 times (0.6 - 0.4 + 0) = 0.24; each outer cell = 0.04.
- 1Mass: 0.44 + 2 times 0.24 + 2 times 0.04 = 1.00, so mass is conserved.
- 1Variance: 2 times 0.24 times 1 + 2 times 0.04 times 4 = 0.80 cells squared, equal to 2 Kd t.
Key terms
- Fick's law
- The rule that diffusive flux scales in proportion to the concentration gradient, with the diffusion coefficient as the constant.
- Diffusion coefficient
- The constant linking flux to concentration gradient, which rises with temperature and differs between contaminants.
- Iteration
- Repeating a calculation in the same cell or program step until a stopping condition is met.
- Conservative pollutant
- A pollutant that stays in solution and does not degrade, so its total mass remains constant as it spreads.
- Random walk
- A sequence of moves whose direction is chosen at random each step, whose spread grows with the number of steps.
Diffusion, Fick's Law and Dispersion Simulation FAQ
Why does diffusion produce a normal distribution?
Each molecule moves under countless random collisions, so the spread of many molecules behaves like random error. The normal distribution models random variation, and the exercise confirms it: a profile left to diffuse under Fick's law can be matched by a normal curve with a larger standard deviation.
Why not use a realistic diffusion coefficient in the lake model?
With distance in kilometres and time in days, free-water diffusion coefficients are around ten to the minus seven, far too slow to show change in 100 spreadsheet steps. A real lake would also mix by convection and wind, so a larger working value is used to see the process.
What is the limiting diffusion coefficient in the spill exercise?
About 0.33. Above one third the centre cell falls below its neighbours after the first step, and larger values make concentrations alternate between positive and negative. These are artefacts of discretising time and distance into steps, which finer resolution would reduce.
What does the random-walk simulation show?
Five hundred particles each move one unit left or right at random every step. Their frequency distribution spreads into a normal shape whose variance grows with the number of steps, matching a normal curve driven by an iteration counter. It reproduces dispersion without writing out every intermediate column.
Why is Excel limited for this kind of model?
A spreadsheet stores layout and formatting for every cell, so copying a model across thousands of columns is slow to set up, slow to run and hard to search. Excel is excellent for understanding a model, but high-resolution models are usually written in programming languages.
Exam move
Run the spike calculation by hand for two steps with two different diffusion coefficients, checking mass and variance each time. In the spreadsheet, try values either side of one third and describe the artefact you see. Then follow the simulation instructions once, naming the conditional statement, the iteration settings and the stopping rule, and explain why the normal curve's variance must grow with the iteration counter.
Close by summing the cells of your spreadsheet at two times to confirm conservation, and explain what change would make the pollutant non-conservative.