Either gunpowder or hydrocarbon vapor, the principle behind the blast is the same
Plain old gunpowder and high tech hydrocarbon vapors have more in common than one might think. Basically, there are two types of explosions — high yield “detonating explosives” like TNT and low yield “deflagrating explosives” like gunpowder and, you guessed it, flammable vapor clouds of normal hydrocarbons. Low yield explosives simply burn, whereas high explosives produce a shockwave that moves through the explosive source at extremely high velocity.8
When poured on the ground and ignited, gunpowder burns rather than explodes. It is well accepted today1.5 that a vapor cloud of normal hydrocarbons like propane, butane and others, when released in an open area and ignited, also simply burn.
What causes a flammable vapor cloud to create overpressures? The same thing that causes overpressures with gunpowder — restraint or confinement of yet unburned gunpowder under pressure from the hot, expanded combustion products. This causes the flame front to speed up until explosive overpressures are generated. Creating a confinement by rolling gunpowder in paper creates firecrackers powerful enough to blow off a finger. It is only logical that placing a large volume of flammable vapors in a confinement also creates an explosion.
Unlike gunpowder, a vapor cloud does not need complete confinement on all sides. Depending on the fuel-to-air mixture, sometimes only a two-sided confinement is needed to cause overpressures. Examples of two-sided containment include the space under objects like truck trailers, elevated temporary buildings and railroad box cars. Internal explosions in three and four-sided enclosures, including buildings, are more common to vapor cloud fire sites. Four such explosions occurred north of London in December 2005 when a vapor release spread flames through 20 fuel storage tanks at the Buncefield oil depot.
A good air-to-fuel mixture made these explosions particularly powerful, causing several massive long duration low pressure waves of one and two psi that traveled much further than would normally be expected. These waves buckled the walls of sheet metal buildings far from the vapor release. Other damage included broken glass windows and cracked brick facades, but mainly the waves crushed the sheet metal walls of large warehouses inward. The damage was the same at almost all of these type buildings. Facing sheet metal walls were pushed in, allowing the pressure wave to enter the building, flow through it, then push in the back wall. This has been seen at other vapor cloud events, such as the Southern Pacific Englewood Yard explosion in Houston, TX4. Why were these distant sheet metal buildings damaged? Sheet metal buildings tend to be big structures with large surface wall areas. Sheet metal fails at very low pressures (see Chart 1) compared to other methods of construction such as wood, brick and concrete.
Containment overpressures from gunpowder can be pretty high. Otherwise a bullet just rolls out of a gun barrel. The overpressures from a containment of flammable vapors can be shocking. A containment explosion under a large trash truck container at Buncefield caused enough of a shock to the ground beneath it that a reserve water tank some 100 feet away suffered a distortion or bulging of the wall just above the tank floor. This bulging is called Elephant Foot distortion and, until now, is only caused by a fast up and down thrust of the ground during a earthquake.
Now, move to the really mystical type of high pressure destructive force that has even sent the professional academics back to school. The high pressure, estimated at 40 to 50 psi, wrinkled and crushed cars (see Photo C) in the open areas next to the Buncefield site. This was briefly acknowledged in the Buncefield Investigation Third Progress Report. “The magnitude of the overpressures generated in the open areas of the Northgate and Fuji car parks is not consistent with current understanding of vapor cloud explosions.”
If low yield explosives need containment to achieve higher pressures, what interferd with the flame front in a vapor cloud enough to create an overpressure? Remember that a deflagration can have high pressure without real detonation. A low pressure deflagration occurs at about 10 psi, medium deflagration at about 50 psi and a high pressure deflagration at about 300 psi. Between 10 to 15 psi destroys a typical structure (see Chart 1).
When vapors ignite, the expansion pushes unburned gas ahead of the flame front. The flow ahead of the flame causes a layer of turbulence. That turbulence enhances the burning rate1, increasing the flame speed and the pressure in the reaction zone of the flame front. When other geometrical conditions and obstructions are introduced, more pressure increases in the flame reaction zone. Because this zone travels with the flame front, it applies this high pressure to any obstruction in the vapor cloud as the front hits it, passes over and then pushes from the rear. As a result, the obstruction is damaged on all sides not just the front side.
At Buncefield, the most common obstructions were cars. It is clear that the pressures were in excess of 35 psi because on some cars tires were deflated. The flame front pressures were high enough to push the side wall of the tires, with an internal pressure of about 35 psi, off the rims. Yet the flame front was so fast that only some of the cars were set on fire.
Only objects in the vapor cloud were wrecked like these cars. Other cars 100 feet outside the vapor cloud received minor damage. With a normal explosion, the pressure is reduced only by increasing the distance. It cannot go from 50 psi to one or two psi in 100 feet. The only way pressure can drop that fast is a flame front reaction zone. The high pressure stopped at the edge of the vapor cloud like a miraculous gift.
My experience includes the onsite investigation of seven large vapor cloud losses and detailed studies of nine more VCEs. The onsite investigations were to develop recommendations prevent future releases and to minimize damage from similar types of release. These site visits included large vapor cloud events in hydrocarbon processing plants, tank farm overfills such as Buncefield in 20052 and New Jersey in 19834 and transportation events outside of processing plants.
Historically, the first widely studied hydrocarbon based vapor cloud loss involved the 1974 destruction of the Nypro Chemical plant in Flixborough, UK1. A hastily formed acronym — Unconfined Vapor Cloud Explosion (UVCE) — was developed to describe the underestimated phenomenon.
About 20 years ago, after several more fire cloud explosion losses, the term was quietly changed by dropping the word “unconfined” for a new term, “Vapor Cloud Explosion (VCE).” Newer losses clearly indicated that there were almost no unconfined vapor explosions5. Therefore, damage was attributable to something else.
The Texas City VCE
On March 23, 2005, a series of explosions occurred at the BP Texas City refinery during the restart of a hydrocarbon isomerization unit (see Photo D). Fifteen workers were killed and 180 others were injured. All of the fatalities were in work trailers located downwind of the atmospheric vent stack. The explosions occurred when a distillation tower flooded with hydrocarbons and was overfilled, causing a geyser-like release from the vent stack.
The vapors drifted under the trailers, resulting in a “partial containment explosion” powerful enough to turn the wooden trailers into a pile of broken splinters, but without igniting a fire.
The Houston VCE
At the Englewood railroad yard, a 38,000 gallon Butadiene rail tanker was damaged in a collision with an empty tank car. When the flame front ignited, captured vapors under the box car (see photo E) resulted in a partial containment explosion powerful enough to blow the floor upward and out the top of the boxcar. The walls of the car collapsed on either side.
The Newark VCE
A 70,000 gallon overfill of a storage tank in Newark, NJ2, created a vapor cloud that spread across a large parking lot and into an adjacent facility before reaching an ignition point (see Photo G). The partial containment explosion under empty aluminum gasoline road transport trailers flipped one backwards and pushed another upright. The last tanker was only partially in the cloud. It was badly deformed at the front but had no damage to the thin aluminum parts at the rear. The damage went from extremely severe to zero damage in a distance of only three feet down the side of the tank.
The Buncefield VCE
After the release of about 140,000 gallons of gasoline from a tank, there were at least five enclosures that exploded at the Buncefield2 terminal vapor cloud loss. The vapors entered the adjacent terminal’s fire protection water pump building constructed from red hollow tile bricks. When the flame front reached the vapors in the building, it exploded. This produced low pressure waves that damaged sheet metal buildings nearly 500 feet away and left the pump and diesel driver in the open (see Photo H). Note that the building that once housed the pump is completely gone.
The Jaipur India VCE
A drastic example of a building-containment caused explosion is the structures damaged at the terminal in Jaipur, India, in 2009 (see Photo F). The terminal had a large release of gasoline from a pipeline next to a storage tank. Gasoline vapor cloud spread into many buildings with ignition more than one hour later. The buildings exploded when the flame front reached them, damaging other objects in the areas.
PDVSA, Amuay Refinery in August
A suspected massive leak of an LPG mixed material created a large vapor cloud that extended outside the tank farm into the local community in two different areas. The crushing effects on the flame front reaction zone were seen on the vapor space of the open top floating roof tank (see Photo B).
Caribbean Petroleum San Juan Puerto Rico
A vapor cloud from a gasoline tank overfill not only ignited 12 tanks but caused the vapor space on several tanks to be crushed in by the flame front reaction zone (see Photo A).
What Are the Two Major Destruction Mechanisms?
In the flame front reaction zone, massive pressures envelop objects inside the vapor cloud with the crushing effects from all sides. This is seen in the crushing of the vapor space of storage tanks and oil drums.
Containment of the flammable vapors from partial or complete containment can cause the vapors to explode with pressures that wreck the surrounding area and equipment. Containment can include:
• Partial confinement in a building or a shed with a roof and some open walls.
• Partial confinement beneath a building or structure that is elevated two to three feet off the ground, even if all sides are open.
• When the vapors enter a building or complete enclosure.
Something Can Be Done
Limit the damage in a processing plant by removing the buildings and enclosures or by keeping the vapor out any structure that must be within the expected vapor cloud area.
1. Remove empty buildings, sheds or enclosures that vapors can enter or get under.
2. Structures that must stay (such as control rooms and motor control center buildings) should close doors, windows and other openings and should be pressurized with air from an elevated source.
1. Lewis, D. J. Unconfined vapor cloud explosions, Prog. Energy Comb. Sci. Vol. 6, pp 151 - 165.
2. Buncefield Major Incident Investigation, and transfer depot, ‘Report to the Health and Safety Commission and the Environment Agency of the investigation into the explosions and fires at the Buncefield oil storage and transfer depot, Hemel Hempstead, on Dec. 11, 2005 Buncefield Major Incident Investigation, 13 July 2006.
3. Hazardous Materials Accident at the Southern Pacific Transportation Company’s Englewood Yard in Houston , TX, Sept. 21, 1974 NTSB Report Number: RAR-75-07.
4. Bouchard, J.K. Gasoline Storage Tank Explosion and Fire: Newark, NJ, January 7, 1983. National Fire Protection Association Summary Investigation Report.
5. AIChE Guidelines for Chemical Process Quantitative Risk Analysis (2nd Edition) 2000. 2.2.1. Vapor Cloud Explosions (VCE) 188.8.131.52.
6. Wingerden, K. Etal, Detonations in pipes and in the open, Bergen, Norway.
7. Allen, G, Limit the damage from a vapor cloud loss, HPI Safety, Security and Environment 2011.
Gene Allen joined the Oil Insurance Association after graduating from Texas A&M University as a loss prevention engineer in 1974. He has been involved with loss prevention activities at refining, manufacturing, storage and distribution facilities.