SOUTH CHARLESTON, W.Va. — Union Carbide Corp. today blamed equipment failure and human error for an Aug. 11 poison gas leak at its Institute plant that sent 135 people to hospitals.
The company also said the poison cloud was composed of 650 pounds of methylene chloride, a suspected cancer-causing agent, and 2,800 pounds of aldicarb oxime "decomposition products." By that, the company meant the aldicarb broke down in a storage tank before being released.
"The emission was caused by a self-sustaining reaction, which was initiated by the accidental heating of a reactor vessel used for storage," the company statement said.
The company acknowledged that the workers thought the temporary storage vessel was empty, as stated previously by Air Pollution Control Commission Director Carl Beard. A Carbide investigation showed about 4,000 pounds of the mixture remained in the tank.
Carbide said the chemical mixture had been transferred to the tank between Aug. 1 and Aug. 7 after excess aldicarb oxime was introduced into the reactor where the pesticide ingredient aldicarb is made.
During that period, steam accidentally entered a jacket surrounding the vessel. How the steam entered was not explained.
"On Aug. 7, an effort was made to move the remaining mixture back to the aldicarb reactor," the company report said. "Pumping was started and continued for a period of approximately 2 1/2 hours when a check indicated the pump was no longer pumping.
Pump Shut Down
"At this point, the pump was shut down and it was believed the (storage vessel) was empty," the report said.
"It is now known that approximately 4,000 pounds of the mixture remained."
At a news conference, assistant plant manager Larry Dupuy said that the temporary vessel had been designed for processes that required steam and that the chemical solution was heated for 10 days without operators knowing about it.
"It should not have been operating," Dupuy said, adding that "the steam could have been isolated from the vessel."
Carbide said the gas leak occurred after a rupture disk and a safety valve opened, and three gaskets blew out when steam continued to heat the reactor.
Plant Manager Hank Karawan said there were numerous violations of plant procedures. "We have a standard operating procedure . . . when starting up units or doing something different."
The first thing a review would have detected would be that the chemicals were sensitive to heat, Karawan said. He said the workers failed to monitor the vessel for a 10-day period. A computer monitor was not programmed to watch it either.