ANALYSIS OF A VAPOR CLOUD EXPLOSION
2006 Explosion at CAI/Arnel in Danvers, Massachusetts
On November 22, 2006, at approximately 2:45 AM, in Danvers,
Massachusetts, an explosion completely destroyed a local ink and paint
manufacturing facility. The blast from
the explosion injured ten community members, damaged or destroyed more than 100
homes and businesses, and broke widows up to one mile away. About 300 residents (77 families) were
initially evacuated. As of May 2007, 50
families were still unable to return home, and 16 homes and two businesses were
razed. No workers were injured, as the
facility was vacant at the time of the explosion. The accident was investigated by the U.S. Chemical Safety Board
(CSB), which presented their preliminary findings in a Public Meeting in
Danvers on May 9, 2007. A copy of their
power point presentation is available at
CSB Chairman Carolyn Merritt commented on the accident: “The
Danversport explosion caused the most serious community damage of any U.S.
chemical accident since the CSB was established in 1998. But for the fortuitous time of the
explosion, nearby residents could have easily been killed by flying debris or
collapse of heavy building structures.
We all have a strong stake in preventing such devastating accidents that
The manufacturing facility itself was completely destroyed
by the explosion. Fire was confined to
the facility and did not spread to the community. The blast from the explosion did do considerable community
The U.S. Chemical Safety and Hazard Investigation Board,
Washington, D.C., (also called Chemical Safety Board, or CSB) is an independent
federal agency that investigates chemical accidents with the objective of
protecting workers, the public, and the environment. The CSB does not issue citations or fines but does make safety
recommendations to plants, industry organizations, labor groups, and regulatory
agencies such as OSHA and EPA. While
Congress has written into law that CSB findings cannot be admitted in court for
litigation purposes, a visit to their website, www.csb.gov
shows that the organization is very blunt and specific when pointing the finger
at the root causes of specific chemical accidents and making their findings
available to the public.
Organizations involved in emergency response included:
fire and police departments
Peabody, Beverly, and other fire and police departments
Bureau of Alcohol, Tobacco, Firearms, and Explosives
State Fire Marshal
Department of Environmental Protection
Remember, the blast occurred just before 3 AM. The facility was closed. There was no one around who witnessed the
blast. When the CSB began their
investigation, there was the visible evidence of blast damage, the accounts of
emergency response personnel, off-site floor plans of the facility and some
information on facility operations, and facility workers to interview. Some photos used in the CSB public
presentation are illustrated below.
The blast site, obliterated facility at right; note
proximity of houses.
Severe structural damage to nearby houses occurred because
of the blast
Businesses were damaged because of the blast
The CSB examined what was left of equipment at the
facility. Note proximity of nearby
house at right.
A key part of the CSB investigation was to map damage
assessment as a function of distance from “ground zero” and relate to peak
blast overpressure. The above image is
a projection of the findings on an aerial photo supplied by Google taken before
The Facility floor plan was examined. Attention was focused on room “C”. Rooms E and F are also used in production,
but on the night of November 21-22, there was no overnight solvent processing
and the rooms were empty of chemicals except for small batch quantities
Overnight (unattended) solvent processing occurred in room C
during the night of November 21-22.
Workers reported that it was customary to turn off the exhaust fan
during unattended overnight processing to save building heat.
The CSB investigation eliminated (1) natural gas, (2) fuel
oil, and/or (3) nitrocellulose as the possible source of the explosion. None of these sources were capable of
producing the blast damage seen. The
nitrocellulose at the facility did not explode but burned rapidly inside the
storage trailers. The CAI/Arnel
facility did not have natural gas service.
The closest natural gas pipeline to the CAI/Arnel building was 215 feet
south, and there was no post-explosion natural gas fire or leak. The fuel oil tanks were intact after the
investigation centered around a 3000 gallon mix tank which was charged with
2000 gallons of printer ink ingredients and flammable solvents n-heptane and
n-propanol, and mixed overnight unattended.
This mix tank was one of four mix tanks in room C, but only one was in
operation that night. Also in room C
were several 500-gallon capacity tote or storage tanks containing flammable
solvents. While the possibility exists
that one of the 500-gallon storage tanks failed, resulting in solvent leaking
onto the production floor which evaporated, the more likely cause was that mix
tank overheated and the solvent evaporated.
The mix tank had no automated temperature controls nor high-temperature
steam shutoff. The tank was not
sealed. The room exhaust fan was turned
off. The vapors accumulated in the
building reaching the lower explosive limit.
An unknown ignition source explosively ignited the vapor cloud.
The lower explosive for n-heptane is 1.05%. The lower explosive limit for n-propanol is
2.1 %. The CSB estimated a facility air
volume of approximately 110,000 cubic feet.
If the facility were uniformly filled with 1.05% vapor volume of
n-heptane, the minimum mass of
n-heptane vapor can be calculated:
= 300 lbs
Here, 100.2 is the molecular weight of heptane and 386.78 is
the number of cubic feet in a lb-mole at 70o
F. Similarly, for n-propanol, a minimum mass of
358 lbs is calculated.
The actual amount of chemical in the vapor cloud would need
to be greater than the lower explosive limit.
The upper explosive limit for n-heptane is 6.7%. The upper explosive limit for n-propanol is
13.7%. Some of the flammable chemical
vapors may have leaked outside and participated in the explosion. We don’t know exactly what happened.
The maximum amount of vapor in the facility is limited by
the amount of solvent in the mixing tank.
The total amount of chemicals in the mixing tank was 2000 gallons. For example, if 1000 gallons of n-heptane
evaporated, the amount of n-heptane in the vapor cloud would be 5250 lbs.
Perhaps the best way of estimating the amount of chemical is
to back calculate from the map CSB used for overpressure damage estimates. Using the PEAC tool, we calculate a minimum
of 1630 lbs of n-heptane (or 2400 lbs of n-propanol) for the blast
pressures. More likely, there would be
some mix of the two chemicals, and the amount could be as closer to 5000 lbs of
n-heptane (or a mix the two chemicals).
Both numbers are within range of the amount of chemicals charged in the
mix tank and would produce vapor concentrations between the upper and lower
explosive limits within the building.
A big unknown when doing vapor cloud explosions is the
fraction of vapor that participates in the explosion. This is the yield factor, which is generally around 0.02 to 0.05
(on a mass basis) for hydrocarbons and closer to 0.2 for more energetic
chemicals with oxygen or nitrate in their structure. The PEAC tool when doing vapor cloud explosion analysis informs
the user what yield factor was used in the calculation. Another issue is the shape of the vapor
cloud that explodes. In the example
discussed here, the shape of the vapor cloud is defined by the volume of the
facility. Outdoors, the shape is
defined by air movement, and because of unknowns, the PEAC tool applies a
safety factor of two on the distance displayed. The PEAC user is informed of this.