Abstract: At 1:55 a.m. on October 8, 1998, Employee #1, a refinery operator, and his coworker assistant, were trying to stop a leak from the bottom head of a coke drum. Instead, the bottom head opened, releasing super-heated petroleum liquid that was above the auto-ignition temperature. A fire broke out and Employee #1, who had been standing by the bottom head, ran past an exit ladder, which was 16 ft from the bottom head flange periphery, to the west end of the platform. He jumped 40 ft to the ground and sustained internal injuries and multiple fractures on the right side of his body. The bottom head had a Hahn and Clay FACT (Force Actuated Closure Technology) system instead of the usual flanges. The FACT system worked somewhat like an hydraulically-driven Mason jar lid and consisted of the bottom flange of the coke drum, a sealing plate under that, and a force plate under that. There was a gasket between the bottom flange and the sealing plate. The force plate was suspended from the bottom flange by long retaining bolts. Between the sealing plate and the force plate were two sets of opposing wedges (which acted like the threads of a Mason jar), and an annular metallic balloon. The FACT could be closed by inflating the balloon with nitrogen, which pushed the sealing plate up against the gasket. The lower set of wedges were then rotated about 1-1/4 in. to tighten the wedges up against the sealing plate. The balloon was then deflated. Retightening was essentially a repeat of this procedure. A locking safety cam, which was supposed to fall into place when the system was in the closed position, had not been functioning for two weeks. The hydraulic controls, where the operator was working, were 26 ft from the bottom flange periphery. In order to help the hydraulics of the system, Employee #1 stood by the bottom head, encouraging the closing action by pounding on the hydraulic cylinder force tabs with a sledge hammer. The hydraulic valves which survived the fire (the cylinders did not) were tested for leaks and malfunctions, and none were found. After the fire, the FACT system was found in the open position. No explanation was evident, other than that Employee #1 had pushed the hydraulic lever in the 'open' direction when he was supposed to be retightening it. The FACT system manual stated that during retightening, "when the balloon is vented, the possibility of process stream leakage exists. Observe site safety precautions." The refinery's operating procedures did not discuss how the FACT was to be retightened or what the "site safety precautions" were.