A BLEVE, drawn through the whole sequence
bleve_physics.py from propane properties, not quoted from a headline:
the video and this page read the same module.
A propane sphere is a very ordinary object. It holds a liquid that is not a liquid at room temperature and atmospheric pressure, and the only reason it is one is that the steel is holding it at about 8.4 bar. Nothing is being refrigerated. Nothing is being held back except the wall.
A boiling liquid expanding vapour explosion is what that arrangement does when the wall stops holding. It is not a chemical event and it does not need a flammable liquid: the explosion is the expansion, and the fireball afterwards is a separate consequence of the fuel happening to be fuel. This page walks the sequence, gives you the overpressure table, and ends on the three things a design report has to size before a vessel is allowed anywhere near a fence.
Out now
Try Reflux Student free on Aspen Plus
Reflux Student drives your own Aspen Plus V14 from plain English. You type what you want, it opens the case, makes the change in Aspen, runs it, reads the result back and tells you what it verified. Aspen does the math. You keep the judgment. Windows, your own licence, and free to start is a trial with a limit on it, not a free product.
Enter your email on the next page and the download link lands in your inbox, so you can open it on the Windows machine Aspen lives on. Windows will say it does not recognise the app the first time: click More info, then Run anyway. The certificate is new and Windows trusts it by reputation, which takes downloads to build. Nothing is wrong with the file.
Not on Aspen Plus? Tell me which simulator you use → and you hear the day your build exists.
Why it is a liquid at all
Propane boils at -42.1 °C at atmospheric pressure. A sphere in a yard on a warm day is at about 20 °C. Put those two numbers together and the liquid in that vessel is 62.1 °C above its own boiling point, and has been for years.
It stays liquid because it is sitting under its own vapour pressure. Antoine for propane at 20 °C gives 8.38 bar absolute, which is 106.8 psi on a gauge. That is the whole arrangement: a liquid that wants to be a gas, and a steel wall that will not let it.
This is why a BLEVE has nothing to do with combustion. Any liquid held above its atmospheric boiling point by pressure alone can do it. Water in a fire tube boiler is the classic case, and boiler explosions killed people by the thousand before anyone worked out why.
The part that fails is the part with nothing behind it
Put a pool fire under the sphere and the whole shell starts taking heat. Below the liquid line, that heat has somewhere to go: the liquid boils against the inside of the wall, carries the heat away by convection, and the steel stays near the liquid temperature. Wetted wall is not the problem.
Above the liquid line there is only vapour, and vapour has neither the density nor the heat capacity to do that job. The heat arrives and stays in the plate. That band of steel climbs through a few hundred degrees, and structural steel loses roughly half its yield strength by 550 °C and most of the rest by 700 °C. The wall is still holding 106.8 psi the entire time.
It also gets worse on its own. The relief valve lifts, vapour leaves, the level drops, and the dry band gets taller. A relief valve sized for fire does not prevent a BLEVE. It buys time.
What is released, and how fast
When the dry band tears, the pressure in the vessel goes to atmospheric in milliseconds. Every kilogram of liquid in the tank is now 62.1 °C above its boiling point with nothing holding it there.
The liquid cools itself the only way it can, by boiling part of itself away. The adiabatic flash fraction is the superheat divided by the latent heat:
- x = cp × ΔT / hvap = 2.45 × 62.1 / 425.6 = 35.7 %
- That fraction flashes essentially instantaneously. The rest does not stay in the tank either: it is thrown out as an aerosol of fine droplets, and that aerosol is what burns.
The expansion is the explosion. One volume of liquid propane at 20 °C becomes about 260 volumes of gas at atmospheric pressure, from 500.5 kg/m³ of liquid against 1.8933 kg/m³ of vapour. A sphere that was one-third full of liquid has to make room for a few hundred times its own volume, and it does that by becoming fragments travelling at a few hundred metres per second.
Only after that does the flammability matter. The cloud finds the fire that caused all of this, and the fireball is a second event with its own radiation footprint. A vessel of hot water does the first half and stops.
The overpressure table, and what it does to a building and a body
These are overpressures: the pressure of the blast wave above atmospheric, not the pressure the vessel was holding. The two are different quantities and comparing them directly is a mistake.
| Overpressure | What it does |
|---|---|
| 0.5 psi | thin glass fails |
| 1 psi | glass gone, doors blown in |
| 3 psi | light structures collapse; eardrums rupture |
| 5 psi | roofs and walls of a house collapse |
| 10 psi | unreinforced buildings destroyed |
| 20 psi | near-total destruction |
For calibration: a car tyre runs at about 30 psi, and total structural collapse is 20. Blast injury to people mostly comes from the building rather than the wave. Eardrum rupture has a threshold somewhere around 3 to 7 psi depending on which dataset you use, and lung damage begins near 15, but at 5 psi the walls are already coming down on whoever is inside.
Two that are in every syllabus
Feyzin, France, 4 January 1966. A propane sphere at a refinery south of Lyon was being drained of water through a bottom valve. The valve froze open, the escaping propane found an ignition source on a nearby road, and the pool fire burned under the sphere until it failed. Eighteen people were killed, most of them firefighters who were cooling neighbouring vessels, and eighty nine were injured. Two adjacent spheres went afterwards.
San Juanico, Mexico, 19 November 1984. A pipeline rupture at a PEMEX LPG terminal outside Mexico City released a cloud that found an ignition source and burned back to the plant. Fifty four vessels went in series over about ninety minutes. Around five hundred people were killed and thousands were burned. It remains the most severe LPG accident on record.
The thing both have in common is that the first event was small and the second was not. Feyzin started with a drain valve. San Juanico started with a pipe.
The three things a design report has to size
This is the part that turns the story into engineering. Before a pressure vessel holding a liquefied gas is allowed to exist next to anything, a design report owes three specific numbers, and none of them is optional.
- A relief valve sized for fire. Not for a blocked outlet, not for thermal expansion: for the heat input of a pool fire against the wetted area, which is usually the governing case and usually the biggest orifice. API 521 gives the correlation. It does not stop a BLEVE, it delays one.
- A fireproofing specification. Insulation or a water deluge on the shell and, just as importantly, on the supporting legs. Unprotected steel legs can fail before the shell does, and a sphere that falls over tears its own piping out.
- A separation distance to the fence. Worked backwards from the overpressure table above and from the fireball radiation: you choose the overpressure you are willing to expose the boundary to, and that choice sets the distance. This is what the table is actually for.
Next time you drive past a white sphere, the question worth asking is not whether it can do this. It is whether someone did those three calculations, and how far away the fence ended up.