Under pressure to get production going, the plant had to resume production without adequate preparation. One key omission leads to the explosion.
Company procedures called for a PSSR (pre-startup safety review) to be run for the methomyl unit before it was restarted. This was a complex process that was supposed to prove that all the equipment and safety devices were in order, allowing the unit to run correctly and safely.
This was no simple restart. Bayer had replaced the DCS during the outage, installing a completely new system to replace a platform that had been running for more than 20 years. The plant was under pressure to get the unit running again because the demand for Larvin (the specific pesticide product) was seasonal and orders had to be filled within a critical time frame. Methomyl, a key ingredient of Larvin, was being pulled from inventory, but there was not enough to fulfill demand so production needed to get started without delay. Consequently, Bayer had operators, engineering staff and contractors working around the clock to finalize all the installations, calibration and checking.
Who’s checking what?
The individuals performing many of the specific PSSR functions were not operators or people that were familiar with the process. Working through the eight pages of PSSR instructions, the team incorrectly identified some items as being complete or “not applicable” that in fact were applicable and had not been completed. Other things that were correctly identified as incomplete had no dates for follow-up. The PSSR related specifically to the new DCS was also incomplete and had critical errors.
While this was going on, the control room operators were trying to understand the new control system, but there was no formal training to prep them for this huge change. They were trying to learn, as best they could in the time available, how the system worked by observing operators at one of the other units in the plant. This attempt at OJT by watching a much different process did not prepare the operators for what lay ahead.
The startup procedure for the methomyl unit called for a solvent flush and nitrogen purge before the startup. This eliminated water from the system, and also helped verify instrument calibrations, proper equipment operating sequences and that the new DCS was performing as expected. Given the amount of piping changed to accommodate the new residue treater vessel, along with the new control system and its associated instrumentation, this was a hugely critical step.
Most of the job is not enough
Plant technicians did most of the job, but not all. They performed the solvent run through most of the unit, but not as far as the residue treater. In many respects this was the most direct cause, not because it would have identified some mechanical problem hiding in the system, but for the simple fact that the solvent used needed to remain in the residue treater to be in place and heated to the correct temperature when the “flasher bottoms” (methomyl-contaminated solvent) began to reach the treater. Instead, operators began pumping the flasher bottoms with an excessively high level of methomyl into an empty vessel where the decomposition process quickly got out of hand. A runaway exothermic reaction ensued and the tank exploded.
This should not have been possible because the control system was designed to prevent adding methomyl into the vessel unless there was enough MIBK (methyl isobutyl ketone) solvent in the vessel, at the correct temperature. Operators deliberately bypassed this safety measure, with the knowledge of supervisors, which led to the explosion.