May 9, 2007
In Preliminary Findings, CSB Investigators Say 2006 Danvers, Massachusetts, Explosion Caused by Solvent Vapor Accumulation, Lack of Ventilation inside Building; Flammable Liquid Safety Standards Were Not Implemented, CSB Investigators Find
Danvers, Massachusetts, May 9, 2007 - In preliminary findings released today, U.S. Chemical Safety Board (CSB) investigators said that the November 2006 explosion which destroyed a local ink and paint manufacturing plant and damaged scores of nearby buildings was most likely caused by the inadvertent overheating of solvents left stirring overnight in an unsealed mixing tank, releasing flammable vapor which accumulated and ignited.
The vapor built up in the facility because the solvent ventilation system was routinely turned off when workers departed the facility at the close of business each evening, an unsafe practice which investigators called the "immediate cause" of the accident. The explosion occurred at approximately 2:45 a.m. on November 22, long after all employees had departed the facility for the night. All fuel sources for the explosion other than flammable solvent vapor - including natural gas - were ruled out.
The findings are set to be presented at a community meeting this evening in Danvers, beginning at 7 p.m. at the Sheraton Ferncroft Resort, 50 Ferncroft Road, with the five CSB board members in attendance. CSB Chairman Carolyn W. Merritt will preside. Following the presentation of findings, the Board will take comments from the public.
The explosion during the overnight hours injured ten community members, damaged or destroyed more than 100 homes and businesses up to one mile away, and damaged numerous vehicles and boats. More than 50 families still have been unable to return to their homes, CSB investigators said. No workers were injured in the blast.
"The Danversport explosion caused the most serious community damage of any U.S. chemical accident since the CSB was established in 1998," said Chairman Merritt. "But for the fortuitous timing of the explosion, nearby residents could have easily been killed by flying debris or the collapse of heavy building structures. We all have a strong stake in preventing such devastating accidents that disrupt communities."
The Danvers facility housed two independent, privately held firms that split from a single company in 1985. CAI Inc. produced solvent-based commercial printing inks, while Arnel Corporation produced solvent-based stains and coatings. Only CAI was operating a solvent-based process the night of the explosion, when the facility was unattended, witnesses stated. The CSB investigation team noted that the facility did not comply with federal and state regulations concerning the storage and use of flammable liquids and solids and also lacked related permits and licenses required under the state fire code.
The building lacked floor-level ventilation systems to prevent the spread of flammable atmospheres from process equipment, as required under the OSHA flammable and combustible liquid standard. In addition, Massachusetts fire regulations require that flammable liquid storage equipment located inside buildings must be vented to the outside and must have approved automatic shutoff valves, and fire-resistant materials must be used for attached piping. These safeguards were not in place, CSB investigators said.
Investigators said that the mixing tank likely overheated because steam heating valves either malfunctioned or were left open due to an operator error. The steam valves were evidently destroyed in the massive fire that followed the explosion, making it impossible to confirm an exact mechanism for the presumed overheating. However, it was determined that during the work day on November 21, a 3,000-gallon mixing tank operated by CAI had been charged with powdered resin and flammable solvents, including heptane and propyl alcohol, and heated by opening two manual valves on a connected steam piping system. The mixer was then left stirring overnight.
"The operator told CSB investigators he believes he closed the steam valves before leaving for the evening, but because the valves were never recovered their actual position and functional condition cannot be determined," said CSB Supervisory Investigator John Vorderbrueggen, P.E. If the steam valves either leaked or remained open, 240-degree steam would have continued to heat the tank without restriction, evaporating off large quantities of highly flammable solvent vapor. During the overnight hours, vapor continued to flow out of the tank and throughout the building, eventually finding an ignition source, such as a spark from an electric device.
"By the process of elimination, that's likely to be what happened," Mr. Vorderbrueggen said. "If the ventilation system had been left on, however, the accumulation of flammable vapor would have been greatly reduced. The practice of turning off the ventilation system at night - which was unsafe in a facility that handled large volumes of flammable solvents - was the immediate cause of the accident."
"We noted that the companies lacked important safeguards for operating solvent-based processes. Such safeguards are critical to prevent a single foreseeable error or mechanical failure from leading to a catastrophic event," Mr. Vorderbrueggen said. For example, neither company used checklists or formal written procedures to help ensure the correct sequence of operator actions. There were no automatic alarms, shutdown systems, or interlocks to prevent overheating of the mixing tank. "When hazardous processes are intended to be left running unattended, it is particularly important to use multiple safeguards, sometimes called layers of protection, to prevent catastrophic accidents," Mr. Vorderbrueggen said.
Investigators ruled out all other possible fuel sources for the initial explosion, including industrial-grade nitrocellulose stored on-site; heating oil that was stored inside the building; and all potential sources of natural gas.
CSB Supervisory Investigator Robert Hall, PE, an engineer who formerly participated in Department of Transportation pipeline investigations, said, "We found no evidence to suggest that natural gas could have caused this accident. No gas line ran into the facility, and there is no evidence of any sufficiently large gas leak near the facility prior to the explosion." Mr. Hall said that the nearest the gas line was at least 150 feet away from the CAI/Arnel building, and that there was no credible scenario by which gas could travel along or inside a sewer pipe into the facility.
"Even in the remote possibility of a low-pressure natural gas leak into the sewer system, gas would have flowed toward the houses on nearby Bates Street as well as the CAI/Arnel facility, and there was no indication that any of the Bates Street properties were affected by gas. In addition, structures have roof-level plumbing vents that would be expected to safely release any natural gas to the atmosphere before it entered the interior of a building," Mr. Hall said. "Furthermore, there was no tell-tale natural gas fire of the kind that almost always follows a major natural gas leak and explosion. Gas leaks simply don't heal themselves."
Mr. Hall stated that the CSB's blast modeling confirmed the conclusion that the explosion was not fueled by natural gas. CSB investigators mapped hundreds of blast damage markers throughout the community and used these measurements to perform computer modeling of the explosion. At the community meeting tonight, investigators plan to show a diagram estimating the size and extent of the sudden blast wave that caused the widespread community damage. None of the damage to offsite buildings was caused by fire.
Mr. Hall stated, "Even under optimal conditions for an explosion - with an ideal natural gas-air mixture filling the entire available building volume - a natural gas explosion would not yield sufficient energy to account for the community damage offsite. The CSB's blast modeling indicates that a natural gas explosion could not account for the blast effects that severely damaged dozens of buildings hundreds of feet from the explosion. However, flammable solvent vapor from the mixing tank would contain enough energy to account for the observed community damage, according to our calculations."
Chairman Merritt said, "We are acutely aware of how this accident affected the Danvers community. It is our hope that these preliminary findings will help inform citizens about what happened and provide information that will help other communities assure that companies handling hazardous materials are operating safely. We are committed to an open and transparent investigation and are keenly interested in the community's comments tonight."
Mr. Vorderbrueggen said the CSB investigation will continue and that activities will include laboratory analysis of the volatility of ink mixtures produced at the facility, further explosion modeling, and analysis of opportunities to improve applicable codes, regulations, inspections, and enforcement. He said the team hopes to complete its final report within six months.
The CSB is an independent federal agency charged with investigating industrial chemical accidents. The agency's board members are appointed by the president and confirmed by the Senate. CSB investigations look into all aspects of chemical accidents, including physical causes such as equipment failure as well as inadequacies in regulations, industry standards, and safety management systems.
The Board 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. Visit our website, www.csb.gov.
For more information, contact a member of the CSB public affairs staff: Sandy Gilmour 202-251-5496, Kate Baumann 202-725-2204, Jennifer Jones 202-577-8448, Hillary Cohen 202-446-8094, or Dr. Daniel Horowitz, 202-441-6074.