A loud jet engine-like sound drew startled attention from businesses neighboring the T2 Laboratories chemical plant one afternoon in December 2007. Eyewitnesses reported high pressure venting from the top of a 2,450-gallon batch reactor designed to produce a specialized gasoline additive.
Within moments, the reactor violently ruptured with a force equivalent to 1,420 pound of TNT.
"The incident at T2 Laboratories included one of the most powerful explosions that the CSB (U.S. Chemical Safety and Hazard Investigation Board) has ever investigated, a blast that was felt 15 miles away in downtown Jacksonville (FL)," said CSB chairman John Bresland.
Four T2 workers, including a company co-owner were killed in the blast. Four other T2 employees and 28 workers at nearby businesses were also injured.
The blast damaged other businesses within one quarter mile of the facility. Four damaged buildings were subsequently condemned. Debris landed up to one mile away.
A report issued by the CSB in September 2009 blamed the disaster on a runaway exothermic reaction following a breakdown in the reactor's cooling system. No emergency backup to the primary cooling system was available.
Ordinarily, the CSB made specific safety recommendations regarding the process under investigation. Instead, this report made wide ranging recommendations to better educate chemical engineers with regard to chemical reactive hazard awareness, Bresland said.
"We would have made different recommendations had the company still been in operation," he said. "But the company was basically destroyed by the explosion. There wouldn't be any point in making recommendations to the company."
The company might have survived if management recog-nized the runaway reaction hazard associated with their product, said CSB lead investigator Robert Hall.
"Had they understood this they might have chosen to design and operate their process differently," he said. ???????
Prior to 2004, T2 Laboratories, Inc., a small privately-owned corporation, concentrated primarily on blending solvents used in the printing industry. Other than issues of inherent flammability, these solvents remained non-reactive when blended into solution as needed.
Manufacturing methylcyclopentadienyl manganese tricarbonyl (MCMT), an organomanganese compound used as an octane-increasing gasoline additive, was T2's sole reactive chemical process, Hall said. The additive was marketed under the brand name Ecotane.
"It was much more hazardous than the other processes they were performing at this site," he said.
Before 2001, T2 blended pre-manufactured MCMT to specified concentrations for a third-party distributor. That year, T2 moved from a downtown warehouse to a site in a north Jacksonville industrial area where construction began on an MCMT process line.
"The seeds were planted the first day they operated," Hall said. "That ultimately led to the explosion."
Although both of T2's owners - a chemical engineer and a chemist - had prior chemical industry experience, neither had previously worked with reactive chemical processes.
"The owners who developed the process did not fundamentally understand the hazard of the material they were dealing with," Hall said. "They did not understand the runaway reaction potential that existed."
Utilizing a patent dating back to the 1950s, the owners developed a three-step process for making MCMT within a single process reactor. After running more than 100 test batches in a one-liter glass reactor, T2 moved to the north Jacksonville site.
"My guess is that the laboratory testing was more involved with improving the process, making sure they were getting a good quality material and good yields," Hall said.
T2 hired consulting engineers to assist in the process design, control system engineering and project management. Due to limited funding, T2 purchased and refurbished used equipment, including a 12-foot high, seven-foot diameter reactor originally built in 1962.
Modifications to the reactor reduced the maximum allowable working pressure from 1,200 psig (pounds per square inch) to 600 psig. A four-inch vent pipe connecting to a rupture disk provided overpressure protection for the reactor. T2 employees told the CSB that the rupture disk was set at 400 psig.
"Had they done thermal hazard evaluation testing, which is very specific to the task, they could have properly designed the reactor relief system," Hall said.
Three of the first 10 full scale MCMT batches using the revamped reactor resulted in unexpected exothermic reactions, all during metalation, the first step in the process. In each instance the batch recipe was slightly different.
T2 did not repeat batch recipes to isolate the problem. Instead, they changed recipes in each of the first 10 batches, the CSB report stated.
"Records we reviewed show that in some early production batches there were excursions that took the temperature higher than the normal operating temperature," Hall said. "In each case the plant was able to get the reaction back under control, primarily through cooling. In the first 10 batches there were three of these events."
T2 announced successful commercial operation to its stakeholders in 2004 after Batch 11.
"They had these near misses but the response was not to investigate the hazard further, just begin making another batch," Hall said.
With Batch 42 in July 2005, T2 increased the batch size from four to six tons. No records exist of additional chemical or process analysis conducted as part of the recipe change. A greater volume of reactants increased the energy that the reaction could produce, likely altering the cooling and pressure relief requirements, the CSB report stated.
Hall compared the management situation to the loss of the Space Shuttle Challenger in January 1986.
"It's a reoccurring theme in accident investigation," he said. "In the Challenger accident you had the failure of the O-ring that led to the explosion of the shuttle. They had prior failures of the O-ring but normalized it. There wasn't an explosion, so they did the same thing again."
At T2, because operators were able to cope with these unexpected temperatures with applied cooling, the company normalized the situation, continuing to do the same thing rather than mounting an investigation into what was happening, Hall said.
At 7:30 a.m. on December 19, 2007, production of T2's 175th batch of MCMT began. The process required both heating and cooling. Hot oil circulated through three-inch piping installed around the inside of the reactor.
For cooling, water injected into a jacket covering the lower three quarters of the reactors was allowed to boil, the steam venting through an open pipe connected to the top of the jacket. The capacity of the cooling system exceeded that of the hot oil system by a margin of 10-to-1.
Each MCMT production step required the process operator to add raw materials and adjust heating, cooling and pressure using a computerized process control system. In the metalation reaction, the process operator fed a mixture of methylcyclopentadiene (MCPD) dimmer and diethylene glycol dimethyl ether (diglyme) into the reactor. An outside operator then hand-loaded blocks of sodium metal through a six-inch gate valve on top of the reactor.
At about 11 a.m., the process operator began heating the mixture, setting the reactor pressure control at 50 psig (3.45 bar) and hot oil temperature control at 360 degrees Fahrenheit. The melted sodium reacted with the MCPD to form sodium methylcyclopentadine, hydrogen gas and heat. The hydrogen gas vented to the atmosphere through a pressure control valve and one-inch vent line.
Once the mixture temperature reached 210 degrees, the process operator started the agitator. The mixing and higher temperature both increased the metalation reaction rate. At 300 degrees, the operator turned off the hot oil system as specified by established procedure. Heat generated by the metalation reaction continued to raise the mixture temperature.
Once the process reached 360 degrees, the operator initiated the control system cooling program, which intermittently injected water into the jacket. Investigators suspect that a single point failure in the cooling system left the operator powerless to control the increasing heat.
No emergency source of cooling existed.
"Had they recognized the hazard up front and designed a cooling system that was more robust and had more redundancy, that could have prevented the accident," Hall said.
At 1:23 p.m., the process operator asked an outside operator to contact the owners, who were off site. Within minutes, the owners returned. The owner/chemical engineer reported to the control room to assist while the owner/chemist searched for the plant mechanic.
An outside operator en route to the control room to investigate multiple process alarms was met halfway by the owner/chemical engineer who said he thought there would be a fire. The owner/chemical engineer motioned employees away from the reactor before returning to the control room.
At 1:33 p.m., the reactor's relief system could no longer control the rapidly increasing temperature and pressure. The three-inch thick reactor ruptured, its contents exploding. Flames also spread through the other flammable processes and storage at the facility.
"It was just a matter of time with the various batches that one got far enough out on the parameter to cause an explosion," Hall said.
CSB investigators determined that insufficient cooling was the only credible cause for this incident, ruling out cross contamination, wrong concentration of raw materials, local concentration of chemical within the reactor and application of excessive heat.
"We likely had a cooling system failure," Hall said. "They were unable to apply sufficient cooling, leading to the result we had."
Witness statements confirm that the process operator reported a cooling problem shortly before the explosion. The cooling system lacked design redundancy, making it susceptible to single point failures.
Employees indicated that T2 did not perform preventive maintenance on the cooling system, replacing components only after failure. On at least one prior occasion, the reactor cooling drain valve failed during operations and required repair. Formation of mineral scale inside the jacket could have interfered with heat removal. Also, loose scale could have blocked the inlet/drain pipe, causing it to stick open.
CSB conducted laboratory tests based on the T2 chemical recipe using a small sample size to minimize potential hazards. Two exothermic reactions were observed. The first reaction occurred at about 350 degrees as desired. A second more energetic reaction occurred when the temperature exceeded 390 degrees.
Pressure from that second reaction overwhelmed the reactor's pressure relief system designed for normal operating conditions. The CSB report stated that it was unlikely the T2 owners were aware that reaction would occur.
?"In a perfect world, T2 would have done sophisticated testing that we did after the accident," Bresland said. "Then, based on that, T2 would have designed the process in such a way that, for example, there would have been a backup cooling system," Bresland said.
The pressure and temperature rise during the second exothermic reactions was about 32,000 psig per minute (2,206 bar per minute) and 2340 degrees per minute. This was sufficient to burst the test cells.
It is unlikely that an overpressure relief device of any size set at 400 psig could have prevented the reactor failure during the second exothermic reaction, the CSB report stated. Had T2 set its four-inch reactor rupture disk at 75 psig, rather than the 400 psig used, the runaway reaction likely would have been relieved during the first exothermic reaction, precluding the second.
"Had they done thermal hazard evaluation testing, which is very specific testing to the task, they could have properly designed the relief system to better protect the reactor," Hall said.
All three steps of the MCMT process involved toxicity, flammability or reactivity hazards. A literature search by CSB found little published information on the production of MCMT other than the patents and no published information specific to its reactivity hazards.
That lack of process hazard information made laboratory testing especially important, the report stated. Initial testing done by T2 did not observe extreme exothermic behavior. Test temperatures never exceeded 380 degrees. However, the one-liter laboratory reactor did not accurately indicate the amount of cooling needed in the full-scale T2 reactor.
"The thermal hazard evaluation testing that CSB performed is a specialized group of tests that is done by a very small number of companies across the U.S.," Hall said. "These tests look at the heat produced and the reaction rate. We have used these tests in a number of cases to replicate chemistry that occurred in similar types of runaway reaction accidents."
The owner/chemical engineer held a bachelor's degree in chemical engineering and was active in his university's engineering curriculum advisory board. However, most baccalaureate chemical engineering curricula in the U.S. do not specifically address reactive hazard recognition or management, a survey by Texas A&M University's Mary Kay O'Connor Process Safety Center revealed.
Of the universities surveyed by the center, only 11 percent required process safety education in the core baccalaureate curriculum. An additional 13 percent offered an elective process safety course.
The CSB recommended that the American Institute of Chemical Engineers and the Accreditation Board for Engineering and Technology, Inc., work together to add reactive hazard awareness to baccalaureate chemical engineering curricula requirements.
It also recommended that the AIChE inform all student members about its Process Safety Certificate Program, encouraging participation.
A report on reactive hazard management issued by the CSB in 2002 documented 167 serious reactive incidents in the U.S. between January 1980 and June 2001 that resulted in 108 deaths and hundreds of injuries. At least 35 percent of those incidents were due to runaway reactions similar to the one at T2. Many of those incidents were at small manufacturing sites similar to T2, the report stated.
Four previous runaway reaction incidents investigated by the CSB since 1998 resulted in 10 deaths and more than 200 injuries.
Only one other company in the U.S. manufactures MCMT, Bresland said. That company, a market leader in fuel and lubricant additive research, development and manufacturing, was the originator of the MCMT patent.
"We don't know what their process is like but they have not had an incident so we assume they are doing things the right way," he said.