Gas dispersion modelling estimates how a released vapour cloud may travel and dilute after loss of containment. It supports facility layout, hazardous area review, gas detector placement, emergency response and quantitative risk assessment. The usefulness of the result depends on how well the release scenario, fluid properties and environmental assumptions represent the real facility. A sophisticated model cannot compensate for a poorly defined source term.
Define the release source
The model needs release pressure, temperature, composition, hole size, inventory and duration. Isolation and blowdown can significantly change release duration, so assumptions about shutdown performance should be traceable to the design. Two-phase releases may require different treatment from dry gas. The source term should match the scenario being assessed rather than a generic leak size applied everywhere.
Weather conditions
Wind speed, atmospheric stability and temperature affect dispersion. Studies often evaluate several weather categories to represent stable and unstable conditions. Low wind stable weather can create longer flammable clouds, while higher wind can increase dilution but transport gas in a different way. The selected cases should reflect local meteorology and the purpose of the study.
Congestion and geometry
Buildings, equipment and terrain influence gas movement. Simple models may be suitable for some screening tasks, while CFD can provide more detail in congested areas. The required model fidelity depends on the decision. For detector placement around complex modules, geometry may matter more than for a broad offsite screening calculation.
Flammable and toxic criteria
Dispersion results are usually compared with concentration thresholds relevant to ignition or toxicity. For flammable gas, lower flammable limit fractions may be used to understand cloud extent. Toxic assessments use exposure criteria appropriate to duration and substance. The chosen endpoint should be clearly reported so drawings and risk results are not misinterpreted.
Gas detection
Detector placement should consider credible release locations, gas density, ventilation and cloud movement. A dispersion plot can support coverage studies, but detection philosophy also needs to consider voting, alarm response, shutdown actions and detector availability. Detection is a mitigation or escalation-control barrier only if the resulting action occurs in time.
Model uncertainty
Every dispersion model contains uncertainty in leak size, direction, weather and physical modelling. Results should not be presented as exact boundaries. Sensitivity cases can show how conclusions change. For decisions involving occupied buildings or emergency response, conservative assumptions may be appropriate, but excessive conservatism can also produce impractical designs.
Release duration and isolation
For many gas releases, duration is strongly influenced by detection and isolation. A fast-closing isolation valve can reduce total inventory released, but only if the valve is positioned effectively and the upstream and downstream inventories are understood. Blowdown can further reduce duration but may create a simultaneous flare or vent load. The modelling basis should therefore be linked to the shutdown philosophy.
Using results in design
Dispersion results can influence equipment spacing, building location, air intake placement, muster points and detector coverage. They should be interpreted together with ignition likelihood, occupancy and emergency response. A cloud reaching a location does not by itself define risk; frequency and exposure also matter.
Connection to hazard studies
HAZID and HAZOP identify the release scenarios that need assessment. Dispersion modelling then provides evidence about consequence extent and can support bowtie, LOPA or QRA decisions. Keeping the source scenario traceable across these studies prevents modelling from becoming disconnected from the actual risk basis.
How Pro Engineers can assist
Pro Engineers provides process safety engineering support that connects hazard identification with consequence assessment and practical design decisions. Visit the Pro Engineers home page for our broader process and process safety engineering services.
Release direction and elevation
The orientation and elevation of a release can materially affect dispersion. A horizontal jet from high-pressure piping behaves differently from a low-velocity vent or a release close to grade. Nearby structures can deflect the cloud and create recirculation zones. The modelling basis should therefore reflect credible leak geometry rather than assuming every release is unobstructed and horizontal.
Isolation and inventory segmentation
Inventory segmentation is often one of the most effective ways to reduce release duration. Isolation valve location determines how much material remains connected to the leak after shutdown. Blowdown can reduce pressure further, but it may introduce additional vent or flare loads. Sensitivity cases can help demonstrate whether faster detection, additional isolation or smaller segments provide meaningful consequence reduction.
Using dispersion for detector placement
Gas detector studies should not simply place detectors at fixed spacing. Credible leak locations, ventilation paths, gas density and congestion should inform the design. Detector coverage should then be linked to what happens after detection: alarm only, shutdown, isolation, ventilation response or blowdown. The effectiveness of the barrier depends on the complete detection-to-action sequence.
From dispersion to QRA
Dispersion results are often one input to a broader Quantitative Risk Assessment. QRA combines consequence extent with event frequency, ignition probability and occupancy. This helps distinguish a large but extremely infrequent cloud from a smaller event that contributes more strongly to overall risk. The same source-term assumptions should be maintained between the consequence model and the QRA.
Reviewing model results
Engineering review should focus on the decisions the model supports. Check whether the predicted cloud interacts with occupied buildings, air intakes, ignition sources, muster areas or neighbouring equipment. Sensitivity to wind direction, shutdown time and hole size can reveal whether the conclusion is robust. Where the decision is sensitive to one uncertain assumption, additional data or a more detailed model may be justified.
Documentation
A good dispersion report states the software, model version, fluid basis, weather cases, release assumptions, isolation logic and concentration endpoints. Clear plots should show the source and relevant receptors. This documentation allows future modifications to identify when a change in operating pressure, composition or isolation arrangement invalidates the original result.