Skip to content

E. I. DuPont De Nemours Co. Fatal Hotwork Explosion

Overview

On November 9, 2010, two contractors were performing welding repairs atop slurry tank 1 at DuPont's Yerkes facility in Buffalo, New York when flammable vinyl fluoride vapor that had flowed undetected into the tank ignited, causing an explosion and fire. One contractor was killed and the other was seriously injured. The CSB final report identified failures in tank isolation, hot work permitting, and flammable vapor monitoring.

Incident Snapshot

Field Value
Facility / Company E.I. DuPont de Nemours & Co. Inc.
Location Buffalo, NY
Incident Date 11/09/2010
Investigation Status The CSB Board Members unanimously approved the final report 3-0 at a public meeting in Buffalo, NY, on April 19, 2012.
Accident Type Hot Work - Explosion and Fire
Final Report Release Date 04/19/2012

What Happened

During a planned shutdown from October 22 to November 6, 2010, asbestos insulation on slurry tanks 1 and 2 was to be removed, new non-asbestos insulation installed, and the slurry tanks externally inspected for corrosion.

On October 22, the slurry tanks were locked out by DuPont maintenance personnel so the tanks could be cleaned and entered.

On October 29, DuPont discovered damaged agitator support after the insulation was removed from slurry tanks 1 and 2.

DuPont hired Mollenberg-Betz to repair the agitator support based on the fact that they were already engaged in hot work on site and could be quickly made available.

During the tank 2 internal inspection on November 3, DuPont engineers discovered that the U-leg seal loop on the flash tank overflow line had a fishmouth split in the pipe.

DuPont engineers concluded that the slurry tank could be returned to service without repairing the split and scheduled a repair on this seal loop during the next planned unit outage.

A DuPont crew reconnected the tank 2 and 3 process piping and removed the valve locks to prepare for unit restart.

On November 6, DuPont restarted the Tedlar process with the valves aligned so that the PVF slurry flowed into slurry tank 3.

The equalizer line remained connected to all three tanks, and was not isolated or disconnected from tank 1 before the hot work was authorized.

On November 8, a compressor within the Tedlar unit malfunctioned; the unit was restarted without the compressor, more than doubling the VF vapor present in the PVF slurry flowing into the slurry flash tank.

On November 9, 2010, the contractors completed a hot work permit before starting work on slurry tank 1.

The morning of November 9, a DuPont lab technician tested the area around the top of the slurry tanks for flammable vapor concentration, including the region above tanks 2 and 3 and the immediate area above tank 1 where the repair work would be performed.

The testing indicated there were no flammables present around the top of slurry tanks.

Continuous air monitoring was also present near the top of tank 1, but the lab technician never tested the atmosphere inside the tank.

Shortly after 9 am, the contractor crew went to the catwalk above tank 1 and started work.

The contractor attached his safety harness lanyard to the agitator on tank 1, stepped out onto the tank top, and began the grinding and welding repair work.

The foreman remained on the catwalk adjacent to the closed manway cover to supervise and act as fire watch.

At the time of the incident the contractor was using an electric arc welder to weld metal support on to a C-channel beam on the top of slurry tank 1.

At 11:04 am the tank 1 level transmitter in the control room recorded a sudden increase as a fire erupted inside the tank.

The overpressure blew the steel manway cover off the tank, hurling it more than 100 feet.

Nearly the entire 25-feet-long tank top-to-shell circumferential weld tore apart; only a segment about 24 inches long remained connected.

The welder died instantly in the explosion.

The flash fire burned the foreman’s arms and head, burst an eardrum, and scratched his eye.

The fire quickly consumed the flammable vapor and self extinguished.

Facility and Process Context

DuPont has owned and operated the 100 acre facility in Buffalo, New York since 1921.

The facility employs over 600 workers and manufactures Tedlar, a polymer used as a film in photovoltaic panels and Corian, used to make countertops and other durable surface products.

The Tedlar process converts VF into polyvinyl fluoride (PVF).

VF is generated at another DuPont facility, shipped in on tanker trucks and stored at the Buffalo facility in high pressure storage tanks.

The non-combustible PVF slurry then passes to one of three insulated slurry holding tanks numbered 1, 2, or 3 in the tank farm adjacent to the production building.

Under normal operating conditions, slurry is pumped to slurry tank 3.

Slurry tank 3 is newer than slurry tanks 2 and 1; it had been replaced in 2009 due to corrosion.

Slurry tanks 2 and 1 were used as overflow storage space in the event slurry tank 3 was filled and slurry tank 1 was generally kept empty of slurry.

The flash tank and three slurry tanks are located in a tank farm adjacent to the Tedlar manufacturing building.

The slurry tanks are 11 feet in diameter, 19 feet tall, and have a capacity of 10,800 gallons.

The tank shell and top are one-quarter inch thick stainless steel.

Each tank top has a hinged, unsealed steel cover on a 24-inch diameter manway and a large agitator motor drive and gearbox.

The agitator assembly is bolted to structural steel beams that are welded directly to the tank top.

The slurry tank feed and drain lines are equipped with isolation valves.

The slurry tanks have a common overflow line attached near the top of the tanks.

Blind flanges are installed on the overflow line and isolate one tank from another if one or two tanks are operating while the third is undergoing maintenance.

The 3-foot diameter flash tank vents small concentrations of flammable vapor directly into the ambient atmosphere, so the area is equipped with flammable vapor air monitoring devices to notify personnel when explosion hazards are present.

The continuous air monitors are located on the catwalk near slurry tank 3 and on the flash tank vent pipe.

An alarm in the Tedlar control room activates if either device detects flammable vapor above the instrument set point.

Consequences

  • Fatalities: 1
  • Injuries: 1 injured; the foreman received first-degree burns and minor injuries, and the flash fire burned the foreman’s arms and head, burst an eardrum, and scratched his eye.
  • Environmental Release: Not reported in the source material.
  • Facility Damage: The explosion blew most of the top off the tank. The top and agitator assembly hung over the side of the tank supported only by a 2-foot section of the top. Minor overpressure damage occurred in the tank farm area and the adjacent production building. The steel manway cover was blown off and hurled more than 100 feet. Nearly the entire 25-feet-long tank top-to-shell circumferential weld tore apart.
  • Operational Impact: The Tedlar process was restarted on November 6 after the shutdown; the incident occurred on November 9. The fire quickly consumed the flammable vapor and self extinguished.

Key Findings

Immediate Causes

  • Flammable vinyl fluoride (VF) vapor from interconnected, in-service process tanks flowed undetected into the tank and ignited when the welder was repairing the agitator support assembly.
  • The ignition source came from the repair work activity of welding the agitator support on the tank top.
  • Welding and grinding generated hot sparks, which likely ignited the flammable vapor.

Contributing Factors

  • The defective seal loop provided a direct path for the flammable vapor and uncondensed steam to flow from the flash tank into tank 2.
  • The VF present in the slurry in tank 3 and from the defective seal loop in tank 2 had an unimpeded path to slurry tank 1 through the unblinded overflow line.
  • VF, which is heavier than air accumulated to a flammable concentration in the bottom of the tank.
  • As the steam and VF entered slurry tank 1, the steam condensed, due in part to the earlier removal of insulation from the slurry tank.
  • The slurry tank insulation had been removed during the shutdown and not replaced and the condensing steam exacerbated the VF flow into slurry tank 1 at increased concentrations.
  • The agitator shaft passed through an unsealed hole, approximately one-half inch, in the tank top, which provided a path for sparks to enter the tank or for flammable vapor to escape into the work area.
  • Welding the C-channel on the tank top using the arc welding technique increased the metal temperature above the steel melting point of 2800 °F (1538 °C).
  • The surface of the metal would quickly get red hot as welding progressed, a temperature in excess of 1200 °F (650 °C), which was significantly above the VF vapor autoignition temperature of 725 °F (385 °C).
  • The accumulated VF vapor inside the tank would ignite as soon as it contacted the extremely hot steel or if welding sparks entered the tank through the agitator shaft opening.
  • The contractors were unfamiliar with the Tedlar process and the process equipment involved.
  • The contractors did not know what the flash tank was or which chemicals were present inside it.
  • The hot work permit requirement section was not completed, even though the hot work was carried out within 35 feet of the slurry flash tank that was designed to vent VF to the atmosphere.
  • The safety precautions section of the permit included an item for Lock, Tag, Try, Think that was not checked as necessary.
  • The contractor supervisor did not answer the questions Are chemicals present? What chemicals? and Flammables/Combustibles present? Identify.
  • The DuPont construction engineer for the slurry tank work had no working knowledge of the Tedlar process.
  • He would not have known the effect of the loss of the compressor on increased VF in the system.
  • The service department worker had no knowledge of the Tedlar area or any associated dangers and had been signing permits for the Mollenberg-Betz contractors for months while they did hot work in other parts of the Buffalo facility.
  • No individual with plant-specific knowledge of the Tedlar process reviewed the hot work permit.
  • The lockout card for slurry tank 1 did not include a requirement for blinding the overflow line; that is, tank 1 was considered completely locked out with the overflow open to the tank.
  • OSHA standards do not explicitly require testing the atmosphere inside a tank before performing hot work on the outside of a tank.
  • The OSHA hot work standard does not specifically require air testing prior to the initiation of hot work.

Organizational and Systemic Factors

  • DuPont did not consider that VF could accumulate in the large vapor space in the slurry tanks and reach flammable levels.
  • DuPont did not see this as a hazard because this 0.02 pounds per hour is below the LEL for VF.
  • No management of change (Section 4.7.2) was carried out for the process to continue with this compromised seal loop.
  • DuPont personnel determined that the slurry tank 1 job could be safely completed after the process restarted and scheduled the repairs for November 9.
  • DuPont did not require a management of change if adverse safety impacts were not anticipated.
  • DuPont engineers determined that any residual VF vapor would vent safely to the atmosphere from the flash tank above the slurry tanks.
  • DuPont did not consider the possibility that flammable vapor could flow into, and accumulate in, the slurry tanks.
  • DuPont included the VF storage vessels, reactors, separators, and flash tank in the PSM program but excluded the outdoor PVF slurry tank farm from the requirements of the PSM standard because DuPont concluded VF carryover into the flash tank and product storage tanks was unlikely and insignificant.
  • DuPont further concluded that VF emissions from the flash tank would not accumulate in the tank farm area.
  • Prior to the incident, DuPont classified the slurry tank as a lower hazard operations (LHO) area.
  • Post-incident, DuPont increased the slurry tank area rating to a higher hazard process (HHP).
  • DuPont did not conclude that OSHA PSM coverage extended to the slurry tanks.
  • DuPont’s practices had fallen short of its policies.
  • The latest audits carried out by DuPont at the DuPont Buffalo facility missed many deficiencies that became apparent as a result of the November 9, 2010 incident.

Failed Safeguards or Barrier Breakdowns

  • The overflow line to all three tanks was never blinded.
  • The equalizer line remained connected to all three tanks, and was not isolated or disconnected from tank 1 before the hot work was authorized.
  • The lab technician never tested the atmosphere inside the tank.
  • The DuPont construction field engineer or contractor crew did not ask the lab technician to test the air inside the tank.
  • The DuPont hot work permit procedure did not require testing the atmosphere inside tank 1 for flammable vapor even though the work required welding directly to the tank top.
  • The individuals who signed off on the hot work permit were not knowledgeable in the operations and hazards of the Tedlar process.
  • The hot work permit requirement section was not completed.
  • The safety precautions section of the permit included an item for Lock, Tag, Try, Think that was not checked as necessary.
  • The contractor supervisor did not answer the chemicals and flammables questions on the safety task assignment.
  • The area proprietor was not knowledgeable about the area and did not walk down the area where the hot work was to be done.
  • No individual with plant-specific knowledge of the Tedlar process reviewed the hot work permit.
  • The lockout card for slurry tank 1 did not include a requirement for blinding the overflow line.
  • The contractors did not check the lock out valves on slurry tank 1 with any DuPont employees prior to starting the hot work.
  • DuPont did not properly isolate and lockout tank 1 from in-service tanks 2 and 3 prior to authorizing hot work on tank 1.
  • DuPont’s hot work permit procedure did not require testing the atmosphere inside tank 1 for flammable vapor even though the work required welding directly to the tank top.
  • OSHA’s hot work standard does not specifically require gas monitoring inside containers intended for hot work even though it is recommended by industry safe practice guidelines.

Recommendations

  1. 2011-01-I-NY R1 | Recipient: E.I. DUPONT DE NEMOURS AND CO., INC. | Status: Closed- Acceptable Action | Summary: Develop and enforce corporate-directed policies and procedures which will require all DuPont facilities to audit their hot work permitting systems prior to initiating hot work to ensure that: All potential explosion hazards associated with hot work activities are identified and mitigated; All relevant forms required for permits are completed in accordance with corporate policies and industry standards (including NFPA 326 and NFPA 51B); Appropriate DuPont personnel officially approve hot work permits, by signature or equivalent, consistent with DuPont policies.
  2. 2011-01-I-NY R2 | Recipient: E.I. DUPONT DE NEMOURS AND CO., INC. | Status: Closed- Acceptable Action | Summary: Revise corporate policies and procedures to require that all process piping, or similar connections to tanks or vessels be positively isolated, (using closed valves, blind flanges or pancake blanks) and the equipment appropriately vented before authorizing any hot work.
  3. 2011-01-I-NY R3 | Recipient: E.I. DUPONT DE NEMOURS AND CO., INC. | Status: Closed- Acceptable Action | Summary: Revise corporate policies and procedures to require that the atmosphere inside the container be monitored for flammable vapor prior to performing any welding, cutting, or grinding on the container surface.
  4. 2011-01-I-NY R4 | Recipient: E.I. DUPONT DE NEMOURS AND CO., INC. | Status: Closed- Acceptable Action | Summary: Revise corporate policies and procedures to require air monitoring for flammable vapor inside the container for the duration of the hot work consistent with industry standards (NFPA 326, NFPA 51B). Create a policy for determining criteria for requiring continuous or periodic testing for the duration of hot work.

Key Engineering Lessons

  • Hot work on a tank must not rely only on external area monitoring when the tank itself can contain or accumulate flammable vapor.
  • Positive isolation of connected piping and overflow/equalizer lines is necessary before authorizing hot work on a vessel that may be connected to in-service process equipment.
  • A compromised seal loop or similar defect can create an unexpected vapor pathway that must be addressed through management of change and mechanical integrity review before continued operation.
  • Hot work permit systems must ensure that the permit is completed, reviewed by personnel with process-specific knowledge, and includes the required atmospheric testing and monitoring steps.
  • When welding directly on a container surface, the atmosphere inside the container should be monitored for flammable vapor before and during the work.

Source Notes

  • Priority 1 final report used as the primary authority for incident facts, causes, and recommendations.
  • Priority 3 recommendation status summary used to confirm recommendation closure status.
  • Priority 4 supporting documents were used only where consistent with the final report and to supplement wording on process hazard analysis, permit practices, and public meeting context.
  • Official terminology such as Tedlar, vinyl fluoride (VF), polyvinyl fluoride (PVF), slurry tank 1/2/3, flash tank, and U-leg seal loop was preserved from the source documents.

Similar Incidents

Incidents sharing the same equipment, root causes, or hazard types.

Same Equipment

Same Root Cause

Same Hazard

← View in Knowledge Graph