Fatal Accident due to Bacterial Hydrogen Production in Atmospheric Storage Tank
A fatal accident took place at an onshore processing facility for slop water from the offshore petroleum industry.
Drilling fluids, or mud, are typically oil-water emulsions consisting of base oil (continuous phase), water (dispersed phase), and emulsifying agents. Used drilling mud, or slop, is mud enriched with water and rock cuttings from drilling --- typically 60-80% water, 10-20% emulated base oil, and 10-20% rock cuttings. The used drilling fluids are collected in slop tanks on oil platforms and later shipped to onshore facilities for further processing.
On the day of the accident, two operators were trying to remove the lid from a manhole on top of a 1600-cubic meter storage tank. However, they were not able to unscrew the rusted bolts holding the lid in place, and decided to use an angle grinder equipped with a cutting disc to cut the bolts. Shortly after the cutting began, an explosion occurred inside the tank. The explosion caused sections of the weld between the tank wall and the tank roof to rupture. The most serious damage to the tank roof was in the vicinity of the manhole. One of the operators was thrown over the railing and was killed. No other tanks in the tank farm were damaged. T
he accident investigation focused on finding the direct causes of the accident, and in particular on identifying the fuel component in the explosive mixture. A sample of base oil collected from the liquid surface inside the tank was tested by an external laboratory, and the closed cup flash point according to ASTM D6450 and ISO-2719 was determined to be 92° C. Hence, there was no reason to suspect that oil vapors played a significant role in this accident.
The liquid surface contained small bubbles trapped in a thin oil film, and gas chromatography (GC) revealed that the primary constituent in these bubbles was hydrogen gas. Microbiological analysis by direct counting with an epifluorescence microscope and quantification of DNA sequences by quantitative polymerase chain reaction (Q-PCR) showed extremely high biological activity: more than 100 million bacteria per milliliter. Most of the organisms identified in the solution belonged to groups of fermenting bacteria: Clostridia (Acetobacterium woodii), Bacilli (Alkaliebacterium, Desemzia), Fusobacteria (Propionigenium/Ilyobacter), and Spirochaeta. These are all capable of producing hydrogen under anaerobic conditions as part of the energy metabolism. The solution contained only about 0.1% metanogene bacteria: Methanosaeta (Metanofollis limninatans), Methanogenium and Methanoplanus. This is consistent with the negligible content of methane observed in the GC analysis. Hence, the fuel was hydrogen gas produced by bacterial fermentation of hydrocarbons inside the tank. Unfortunately, the growth substrate that sustained the high bacteriological activity could not be unambiguously identified from analysis with gas chromatography-mass spectroscopy (GC-MS). The ignition source was most likely mechanical sparks produced by the angle grinder, which could have entered the tank through an open 6-inch inspection hatch on the roof of the tank. This is consistent with the very low minimum ignition energy of hydrogen-air mixtures.
