Odour dispersion modelling in an odour assessment

Odour dispersion modelling is the step where an odour assessment either becomes evidence or becomes an opinion with contours attached. The chain runs from a sampled emission rate, through a description of how and where the odour leaves the site, into a dispersion model driven by several years of hourly meteorological data, and out as a 98th percentile hourly concentration at each receptor. Every link in that chain has an uncertainty attached, and the first one usually dominates the answer.

Regulators read the result backwards. They look at the contour, then ask what emission rate produced it, then ask where that rate came from. And that last question is the one most assessments answer worst.

Deriving the emission rate by olfactometry

Odour concentration is measured by dynamic dilution olfactometry to BS EN 13725, which dilutes a sample until a trained panel can just detect it, and reports the result in European odour units per cubic metre. Multiply that concentration by the volumetric flow rate and you have an odour emission rate in odour units per second, which is the number the model needs.

Report the sampling in full: the date and time of each sample, the process conditions at the time, the sampling method, the bag material, the transport arrangements and the interval between sampling and analysis. Samples degrade, and the standard sets a limit on how long they may be held, so a sample analysed outside that window is worth challenging.

Say how many samples were taken and how the representative rate was derived from them. A single grab sample from a process with a diurnal or seasonal cycle is thin evidence, and where only one sample exists the assessment should say so plainly rather than presenting the figure as settled.

Area sources need a different method. Lagoons, windrow pads and open tanks are usually sampled with a flux chamber or a wind tunnel, and the conversion from measured flux to a whole-area emission rate should be set out with its assumptions.

Source characterisation and release conditions

The model needs a physical description of each release, and errors here are easy to spot and easy to make.

For a point source, give the stack height above ground, the internal diameter at the exit, the efflux velocity, the temperature, the orientation, and whether the outlet is capped or fitted with a rain cowl. A capped stack behaves very differently from an open vertical one, and modelling a capped discharge as unimpeded vertical release flatters the result.

For area and volume sources, give the dimensions, the release height and the vertical extent. For fugitive emissions from doors, roof vents and yard activity, describe the assumptions about opening frequency and duration.

Time variation matters as much as magnitude. A process that runs eight hours a day five days a week should be modelled with that pattern rather than as a continuous release, and the assessment should state whether emissions were varied by hour, day or season.

Model selection, meteorological data and receptor grids

Use a dispersion model of the type accepted for regulatory work, state which model and version, and explain the choice in a sentence.

Meteorological data should be several years of sequential hourly data from a station that represents the site, conventionally five years, with each year modelled separately and the results reported year by year as well as for the worst case. State the station, the years used, the data capture rate, and how any missing data was treated. Set out the surface roughness at the site and at the meteorological station, the minimum Monin-Obukhov length used for urban areas, and whether terrain effects were modelled.

Receptors come in two forms. A nested grid gives the contours, and discrete receptors at named locations give the numbers that go in the table. Put a discrete receptor at every sensitive location within the study area, name it, give its coordinates and say what it is: a dwelling, a school, a hospital, a care home, a workplace.

Grid resolution needs stating. A coarse grid will miss a peak close to the boundary, and an assessment that reports no exceedance on a 100 metre grid while the nearest dwelling sits 40 metres from the source has not answered the question.

Benchmark exposure levels by odour offensiveness

Results are compared against benchmark levels that vary with how offensive the odour is. Environment Agency odour guidance sets indicative benchmarks expressed as a 98th percentile of hourly mean concentrations: around 1.5 odour units per cubic metre for the most offensive odours, around 3 for moderately offensive, and around 6 for less offensive.

Offensiveness is assigned by process type, and the assignment has to be justified rather than assumed. Sewage sludge treatment, rendering and intensive livestock sit at the offensive end. Breweries, bakeries and some food processes sit at the less offensive end. Where a site has several processes, the assessment should say which benchmark applies to which source, and whether a combined assessment uses the most restrictive one.

Benchmarks are indicative, not standards. An exceedance is an indication that further work is needed rather than an automatic refusal, and a result below benchmark does not settle the matter where there is a history of complaint. A good Odour Assessment says which it is dealing with.

Percentile reporting and why the figure matters

The convention is the 98th percentile of hourly mean concentrations across the modelled year, which means the level exceeded for 2 per cent of hours, roughly 175 hours a year.

Report that figure, and report the maximum hourly mean alongside it. The two together tell a regulator whether the exposure is a steady low level background or a small number of severe episodes, and the two situations generate very different complaint patterns.

Where the assessment reports more than one percentile, say why. Shifting from the 98th to the 99.5th percentile without explanation looks like an attempt to soften the result, and reviewers treat it that way.

Sensitivity of the result to emission assumptions

The emission rate is usually the largest single uncertainty, and it propagates directly into the predicted concentration.

Run the model at the central estimate and at plausible upper and lower rates, and report all three. Where the upper case pushes a receptor over the benchmark, say so, and set out what would be done about it. An assessment that reports only the central case and then relies on the mitigation being effective is asking the regulator to take the emission rate on trust.

Do the same for the abatement assumption. An odour control unit with a stated removal efficiency is a design intent, not a measurement, and the model should be run at a lower efficiency to show what happens when the media is near the end of its life.

This is the honest limitation of the technique. Dispersion modelling predicts dilution well and predicts emission rates not at all, so the quality of the answer is set before the model is opened.

Presenting contours a regulator can interrogate

Give the contour plot on an ordnance base at a readable scale, with the site boundary, the source positions, the named discrete receptors and the benchmark contour clearly labelled with its value and its percentile.

Then give the table. One row per receptor, with the coordinates, the receptor type, the modelled concentration for each meteorological year, the maximum across years and the benchmark applied. A reader should be able to find their own house in that table.

Include the model input file or a full parameter listing as an appendix. Where the scheme involves extraction from a commercial kitchen or a food process, keep the Ventilation and Extraction Statement consistent with the modelled stack parameters, since the two documents are read together and disagree more often than you would expect.

Councils expect a suitably qualified author for odour work and professional review is strongly advised. The professional directory is the route to a reviewer, and the report index shows the related assessments a process site usually needs.

PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. For specialist technical reports, qualified professional input may be required. This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it.