NTSB issued the investigation report of the towing vessel Mary Lynn in which a fire broke out last May resulting to damages estimated at over $700,000, but hopefully, no injuries or pollution reported.
he investigation contributed the engine room fire to the overpressurization of the fuel day tank (which did not have an independent vent) and the main engine fuel return system which ultimately led to ignition of spraying diesel fuel from a main engine’s fuel system onto an uninsulated engine component.
About 0315 on May 18, the Mary Lynn tied up at the Gasconade Street fleet near mile 175 of the Upper Mississippi River in St. Louis after dropping off barges about 3 miles downriver. The vessel’s crew planned to take on fuel, lube oil, and potable water from a delivery tug and barge about 0600. The chief engineer woke up about 0300, in advance of his 0500 watch, to prepare for the fuel, lube oil, and potable water transfer. He went to the engine room, conducted a visual inspection of the space, and then began the weekly task of removing residual water from the four fuel storage tanks (no. 2 port and starboard and no. 3 port and starboard) using a dewatering filtration system.
About 0330, while the chief engineer was still operating the dewatering filtration system, the fuel delivery tug and barge arrived, about 2.5 hours earlier than expected. The chief engineer shut off the dewatering system pump and closed the fuel storage suction and return valves for the tank he was dewatering at the time, leaving all fuel return line valves closed. The chief engineer later told investigators that he had thought the no. 2 port and starboard fuel storage tank fuel return line valves were open, but he did not physically check their positions. The chief engineer then shifted his focus to preparing for the fueling operation, completing the bulk cargo transfer checklist (used for fuel) at 0335.
About 0500, the transfer of 25,550 gallons of ultra-low-sulfur diesel fuel to the Mary Lynn was completed. The fuel delivery barge and tug departed the Mary Lynn at 0537, and, about 0540, the captain arrived in the wheelhouse to relieve the pilot for his scheduled watch. The chief engineer changed the disposable fuel filters (not a part of the main engines) for the fuel transfer pumps and the fuel suction cartridge filters on the fuel supply line for both main engines (the chief engineer said the changeout was usually done based on engine running hours). Once the changeout of the fuel filters was complete, he informed the captain that the engines were ready for use.
On the morning of the casualty, the chief engineer awoke and began preparations for taking on fuel, lube oil, and potable water. The fuel delivery tug and barge arrived about 2.5 hours earlier than expected, while he was dewatering the fuel storage tanks. He secured the return and suction valves for the tank he was dewatering and went on to prepare for the transfer of fuel, lube oil, and potable water. Although he thought that the two no. 2 fuel storage tank return valves were open, the chief engineer said he did not physically verify their position, and thus, he inadvertently left all return valves to the fuel storage tanks from the fuel day tank overflow line closed. Almost immediately after the Mary Lynn got under way on May 18, the starboard main engine failed to meet the ordered rpm, and the captain brought the Mary Lynn back to the fleeting area to troubleshoot the issue.
The chief engineer found that the fuel suction cartridge filter housings for the starboard main engine were “sucked dry,” an outcome that he attributed to a damaged O-ring. While the chief engineer was troubleshooting the fuel pressure issue with the starboard main engine, he opened the no. 3 fuel storage tank suction valves and started one of the fuel transfer pumps, which had been turned off during fueling and had not been restarted before getting under way, as was the vessel’s standard procedure. After the chief engineer started the fuel transfer pump, the fuel day tank level began to rise.
Because the fuel day tank did not have its own independent atmospheric vent, tank venting was dependent on the four fuel storage tanks’ vents via the overflow line, return header, and opened fuel return valves. Since the chief engineer had not opened any of the four storage tanks’ return valves during or after the fueling process, the day tank essentially became unvented while the engines were running and consuming fuel. Once the day tank filled to capacity, the operating positive displacement transfer pump began overpressurizing the day tank, the supply lines from the day tank to the engine-driven fuel pump, and the return lines from the engine-driven pump to the day tank (entire main engine fuel system).
The day tank and return line pressure would have also risen based on the pressure the positive displacement engine-driven pump could produce and the rate the engine was consuming fuel—with the engine-driven fuel pump directing excess fuel not consumed (not injected into cylinders) back to the day tank. The resulting pressure increase caused the weakest part in the system to fail first, which, in this case, was the bypass sight glass bowl that tied into the return line (as witnessed by the chief engineer). The two sight glass bowls on the spin-on fuel filters were known to be sources of failure and could potentially contain hairline cracks from overtightening. Once the system reached a high enough pressure (likely exceeding 60 psi), and with no other means to relieve the pressure, the fuel supply bypass sight glass bowl on the port main engine spin-on fuel filter assembly broke. The breakage and separation of the sight glass bowl caused hot, pressurized, and atomized fuel to spray into the engine room.
For towing vessels, regulations require that, for vessels built after January 2000, “each integral fuel tank must have a vent that connects to the highest point of the tank, discharges on a weather deck through a bend of 180 degrees, and is fitted with a… flame screen.”3 The rules allow that vents from “two or more fuel tanks may combine in a system that discharges on a weather deck.” As described in the regulation, the vents would combine (tie-in) above the fuel level in the tanks, not under head pressure beneath the fuel tank level, as was the case aboard the Mary Lynn. The fuel tank venting regulations did not apply to the Mary Lynn, since the vessel was built before 2000. The design of the Mary Lynn’s system allowed for the potential pressurization of the day tank to exceed atmospheric pressure if a crewmember secured valves on other tanks. Had the Mary Lynn’s day tank been fitted with its own vent line, even with the fuel return lines inadvertently left closed, the overpressurization of the return line would have not occurred.
The chief engineer had about 32 years of experience working on towing vessels and with EMD engines. While he told investigators that he did not feel tired at the time he was awake and working in the engine room, he reported receiving less than 5 hours of sleep in the 24 hours preceding the fire, consisting of a 1-hour nap the previous afternoon and 3.5–3.75 hours of sleep before waking earlier than his scheduled 0500 watch to prepare for fueling. His longest continuous sleep—roughly 5 hours—was on May 17 (the day before the fire) from about midnight to 0500. Given the engineer’s accumulated sleep debt over the previous 24 hours and waking early during an off-duty time when he would normally be asleep, the chief engineer was likely affected by acute fatigue. Compounding his fatigue was the time at which the fuel preparation tasks were completed on the morning of the fire. The chief engineer woke up about 0300, during a circadian low period outside of his regular work/sleep schedule.
The circadian rhythm dips and rises at different times of the day, and, according to research, an individual’s strongest sleep drive generally occurs between 0300-0700.5 The performance effects of fatigue during these circadian low periods are exacerbated when a person is already sleep deprived. Effects of fatigue can include a reduction in vigilance, concentration, memory, as well as reduced performance on complex or sequential taskings requiring high levels of attention. The nature of the chief engineer’s task of preparing the boat to take on fuel required him to recall and remember sequential steps in a process while maintaining an accurate mental model of the fuel system’s configuration. When this process was interrupted by the fuel barge arriving earlier than expected, it disrupted the sequential nature of the task. The noted effects of fatigue likely impacted the chief engineer’s attention, memory, and performance of this complex task when returning from the interruption.
Lessons Learned: Tank Ventilation
Subchapter M regulations for towing vessels require vessels built after 2000 to have vents for each fuel tank. Regulations for vessels ranging from small passenger vessels to cargo ships require that fuel tanks be independently vented from the highest point of the tank to atmosphere on a weather deck. Tank ventilation is important to ensure a valve line up error does not lead to the overpressurization of or vacuum in a fuel tank. Operators should be aware of their fuel tank ventilation system arrangements. On vessels without independent fuel day tank ventilation, it is critical to ensure proper valve position during transfer and operation of the fuel system.