HAZEL – Hazardous Atmosphere Zoning for European Locations
Built on verified scientific sources
Free. Auditable. Fully open source.
Beta Version 1.0.0 · first public release, October 2026
HAZEL (Hazardous Atmosphere Zoning for European Locations) is a free, open-source tool for estimating how far, and for how long, hazard zones extend after an uncontrolled release of a dangerous substance.
It was created for the detailed analysis of (hypothetical) incidents at industrial facilities and in the wider field of dangerous goods logistics, above all road and rail transport. Its architecture follows the official technical documentation of ALOHA, the program developed by NOAA and the U.S. EPA, the DEGADIS 2.1 dense-gas model, and published literature that the scientific community treats as reference material — all of it listed below.
The words “European Locations” describe where the project comes from, not where it works. The physics engine rests on mathematical relationships that behave identically at every point on the globe.
Please note: HAZEL is an independent tool for training and planning and holds no official government certification. ALOHA is a registered trademark of NOAA. The HAZEL project is not affiliated with NOAA, the U.S. EPA or any other government agency.
Why does HAZEL exist?
Atmospheric dispersion modelling has long been squeezed between two kinds of barrier: financial and technological.
On one side, the market is dominated by commercial, closed-source (proprietary) software. These tools can be very accurate, but they need expensive licences that often have to be renewed. We believe that tools which support engineering decisions and directly protect human life should not sit behind a paywall.
On the other side there are respected, free reference tools, with ALOHA at the head. Yet here too a different barrier has grown over the years: technological progress. ALOHA is a classic 32-bit desktop program whose Macintosh version is supported only up to macOS 10.11 El Capitan, and it was last updated in September 2016. Apple has since dropped 32-bit applications altogether (macOS 10.15 Catalina, 2019) and moved to its own Apple Silicon (ARM) processors, which cut users of current Macs off from this kind of classic software. On Windows, 32-bit programs still run thanks to the WOW64 subsystem; the Apple ecosystem was left without native support.
With the first public beta of HAZEL (October 2026) we set out to remove both barriers at once: a tool that is free, fully open source, and resilient to the architectural changes of tomorrow.
What can HAZEL do?
HAZEL covers the chain from “something is leaking” to “where does the danger reach”: first the rate at which a substance enters the air, then how the cloud spreads, then — if it ignites — fire and explosion effects.
Release rate
A known rate, an evaporating puddle, or a damaged tank (horizontal or vertical cylinder, sphere) holding a liquid, a gas liquefied under pressure or a compressed gas. Liquid from a tank forms a spreading puddle.
Toxic gas dispersion
Gaussian plume for gases that mix like the surrounding air, including low-level inversions, elevated sources and puddles treated as area sources.
Heavy gas dispersion
A separate model for clouds heavy enough to slump and creep along the ground, such as chlorine or a gas that leaves the source very cold.
Automatic model choice
HAZEL picks Gaussian or heavy-gas dispersion from the cloud's density, the release rate and the wind (Richardson number) — and lets you override it.
Fire and explosion
Flammable area (flash fire), vapour cloud explosion overpressure, BLEVE fireball, pool fire and jet fire. The explosive mass is taken from the dispersing cloud, not typed in by hand.
Indoor concentration
How much of the outdoor cloud reaches the inside of a nearby building — input for a shelter-in-place decision.
Map and reports
Zones on OpenStreetMap, the concentration at any clicked point, KML export for GIS software and a printable PDF report.
Chemical library
3,148 substances with their exposure thresholds, searchable by name or CAS number, plus your own local entries for missing properties or new substances.
Works offline
Install it like an app; after the first load no connection is needed to run a simulation. Saved scenarios stay on your own device.
Step-by-step instructions are in How to use; what HAZEL deliberately simplifies is listed under Known simplifications.
How accurate is HAZEL?
We did not take the formulas on trust. Nearly 100 scenarios were run side by side in ALOHA 5.4.7 and in HAZEL, and more than 250 resulting values were compared. The table gives the picture in six groups; the full 18-row breakdown with test identifiers is on the About page.
| What was compared | Values | Typical deviation | Worth knowing |
|---|---|---|---|
| Release rates: puddles, tanks, two-phase and compressed-gas releases | 70 | within ±5% | Compressed-gas tank contents within ±2%. Two-phase releases last about 1.4–1.5× longer, while peak rates and zones are unaffected. |
| Gaussian plume: point and elevated sources, inversions, zone width | 35 | within ±1–3% | Zone width is 4–7% narrower than the reference. |
| Gaussian plume from a puddle (area source) | 22 | within ±10% | Larger gaps only very close to big puddles, where both tools call the result unreliable. |
| Heavy gas dispersion | ~60 | 5–20% shorter | A systematic bias towards shorter zones, strongest for small sources within ~100 m. Keep a safety margin. |
| Fire and explosion: BLEVE, jet fire, pool fire, flammable area, vapour cloud explosion | 49 | ±1% to ±10% | BLEVE ±1%; pool-fire distances about 8% shorter; the others within about 10%. |
| Indoor concentration | 6 | within ±4% | Building assumed on the cloud's centreline, the most exposed position. |
The choice between the Gaussian and heavy-gas model matched ALOHA's in 25 of 26 comparisons; the one difference (flashing ammonia) is deliberate and explained on the About page.
What these numbers do — and do not — say
Agreement with ALOHA measures how faithfully HAZEL reproduces a respected reference method, not how closely any model matches a real cloud. Even the best dispersion models are uncertain: in his review of field experiments with dense gases, Havens (1992) notes that predicting dispersion within a factor of two or three is already difficult. HAZEL is still a beta release and holds no official certification — treat it as a training and planning aid that supports professional judgement, not as the sole basis for action during a real incident.
What is HAZEL based on?
A guiding rule of the project: no formula enters the code until it has been checked against its source. What could not be verified is either left out or openly listed as a limitation, but never guessed. These are the sources behind the engine:
- ALOHA Technical Documentation, version 5.4.4 (NOAA Technical Memorandum NOS OR&R 43, 2013) The backbone: release rates of puddles and tanks, Gaussian and heavy-gas dispersion, thermal radiation, explosion overpressure and indoor infiltration.
- Spicer, T.O. and Havens, J. (1989), User's Guide for the DEGADIS 2.1 Dense Gas Dispersion Model, U.S. EPA, EPA-450/4-89-019 The reference edition of the heavy-gas model, including the equations for the dense “blanket” that forms at the source.
- Havens, J. and Spicer, T.O. (1985), Development of an Atmospheric Dispersion Model for Heavier-Than-Air Gas Mixtures, Vol. I, U.S. Coast Guard (DTIC ADA171522) Confirmation of heavy-gas constants that the NOAA summary names but does not give.
- Havens, J. (1992), Review of dense gas dispersion field experiments, Journal of Loss Prevention in the Process Industries 5(1) Background on how dense-gas models, DEGADIS among them, compare with large field trials (Maplin Sands, Thorney Island, Goldfish and others).
- Chamberlain, G.A. (1987), Developments in Design Methods for Predicting Thermal Radiation from Flares, Chem. Eng. Res. Des. 65, 299–309 The original source of the jet-flame geometry.
- Wilson, D.J. (1987) and Sherman, M.H. (1980), as used in the ALOHA documentation Indoor concentration: the building as a slow filter, and the estimate of its air exchange rate.
- AIChE (1994) guidance on vapour cloud explosions and the Baker–Strehlow–Tang method, as applied in ALOHA Explosive mass of a flammable cloud and the resulting overpressure.
- Smith, J.M., Van Ness, H.C. and Abbott, M.M., Introduction to Chemical Engineering Thermodynamics The Pitzer correlation behind the real-gas correction for compressed gases; the acentric factor is estimated with Edmister's method when it is not entered.
- U.S. Department of Energy, Protective Action Criteria (PAC) The AEGL, ERPG and TEEL exposure thresholds that make up HAZEL's database of 3,148 substances.
- Side-by-side comparison with ALOHA 5.4.7 Nearly 100 scenarios, used to confirm each model and, where the documentation leaves room for interpretation, to choose the reading that reproduces the reference.
What we noticed along the way
Reading several sources side by side turned up a few things worth sharing:
A sign in a formula
Chamberlain's original paper gives the slant length of the jet-flame frustum with a minus under the square root; the ALOHA documentation prints a plus. Re-deriving the geometry confirmed the original, so HAZEL uses the minus.
One formula, two models
A dense cloud is described as a uniform core with Gaussian-shaped edges. In HAZEL the width of those edges comes from the same Briggs lateral-spread relationships as the Gaussian plume, so the two models join smoothly instead of living in separate worlds.
A constant that is not 0.4
The DEGADIS derivation of the source blanket uses a von Kármán constant of 0.35, while 0.4 is the usual value elsewhere. HAZEL follows each source where it applies, and says so in the code.
One cloud, everywhere
The explosive mass for a vapour cloud explosion is not a new formula: it is the same concentration field that draws the zones on the map, integrated between 90% of the lower and the upper explosive limit. The cloud you see is the cloud that explodes.
When a formula is ambiguous
For the indoor model, a key equation was unclear in the copy of the source we worked from. Instead of guessing, we checked its units and compared it with the original Lawrence Berkeley infiltration model; both agreed on the implemented form.
A threshold that had to be calibrated
The ALOHA documentation gives a Richardson-number threshold of 1 for switching to heavy-gas dispersion. With HAZEL's way of computing the number, 1.95 reproduces ALOHA's own choice of model — a difference we state openly in the guide.
What is HAZEL written in?
To make the engine immune to unexpected changes in hardware and operating systems, HAZEL is built on open web technologies — JavaScript and HTML5. Moving all of the calculation into the browser removes the dependence on operating-system vendors and closed runtime environments (vendor lock-in). The same code runs natively on almost any modern device: desktops and laptops with Windows, macOS, Linux or ChromeOS, and phones and tablets with Android, iOS, iPadOS or HarmonyOS.
As a Progressive Web App (PWA) with a service worker, HAZEL combines the freedom of a website with the advantages of traditional desktop software: one click adds it to your desktop or phone's home screen.
On first launch the service worker stores the complete calculation engine and the database of more than 3,000 chemical substances on your device. From then on HAZEL works fully offline and needs no connection to run a simulation. Whenever you start it while online, it checks for a newer version and refreshes the files that have changed. The entire program weighs less than 10 MiB.
Can I host and modify HAZEL?
Yes. HAZEL is open source: you may host your own copy, modify it, redistribute it, sell it on and build it into your own systems. The one thing we ask — and the licence requires — is that you credit the authors. The complete source code and the licence text are in the GitHub repository.
HAZEL consists of static files only, so any ordinary web server
will do — no database or back end is needed. The one file to edit
for your own installation is config.js; the optional
live-weather connection is switched off by default and described
in the repository.
The version we host is labelled beta, so modifications that improve the accuracy of the calculations may still appear. If you run your own copy, it is worth checking back from time to time.
How can I help HAZEL?
If you have sources — in particular older handbooks that were never digitised — that could replace one of the simplifications we had to make, we would be glad to hear from you. To see where HAZEL stands today, read Known simplifications on the About page and the notes in How to use, and see whether you can help find a better approach.
You can send what you have to damian@kocie.mba or open an issue on GitHub. With your consent, we credit everyone who helps as a co-author on the About page.