There are three incidents in the H2Incidents database related to improper purging, and each of them occurred in a different type of setting. Nonetheless, there is a common thread that links the incidents in terms of what went wrong. In each incident, standard operating procedures related to system purging were ignored or altered without thorough consideration of the potential consequences. Hydrogen was ignited in all three incidents, and serious explosions could have occurred if larger volumes of hydrogen had been in use.
Setting — Power Plant
Description: A large, hydrogen-cooled generator is driven by steam turbines at a power station. During maintenance shutdowns, the hydrogen cooling loop in the generator is purged with carbon dioxide. After CO2 concentrations are measured with a densitometer to verify the complete removal of hydrogen, the generator is purged with air, and the maintenance is performed. This purging procedure was used in this incident. The CO2 reading was reported to be 100% CO2 at the top of the generator. The cooling system was then purged with air and a pipe in the cooling loop was cut to install some new instrumentation. When the pipe was cut, pressurized gas was emitted at the opening. Workers assumed the gas was either carbon dioxide or air and proceeded with the new instrument installation. Unfortunately, there was still at least some hydrogen in the pipe and the rest of the cooling loop. When the welder struck an arc, a flame developed at the pipe opening and flashed back into the generator. This caused a low-level explosion within the generator shroud.]
It is very difficult to completely purge hydrogen out of a large, complex piece of equipment. Uniform mixing and dilution is unlikely in all the partially enclosed spaces and crevices. If a hazardous operation such as welding must be performed with an air atmosphere (instead of inert gas) in the equipment, reliable gas concentration measurements should be obtained at several different locations. A direct measurement of hydrogen concentration may well have been more reliable than the 100% CO2 reading on the densitometer. Also, the gas composition should have been determined at the welding site as well as the top of the generator.
Setting — Hydrogen Production Facility
Description: An instrument engineer at a hydrogen production facility was arresting the hydrogen leakage in tapping a pressure transmitter containing 131-bar (1900-psi) hydrogen gas. The isolation valve was closed and the fittings near the pressure transmitter were loosened. The pressure dropped from 131 bar to 51 bar (1900 psi to 740 psi). Then the fitting was further loosened and the instrument tube slipped out of the ferrule and got pulled out of the fitting. With the sudden release of the 51-bar (740-psi) hydrogen, there was a loud pop (like a fire cracker) and the spark-proof tool was observed to have black spot on it. The volume of the hydrogen gas released was small, since it was in the tapping line only.
Safe work procedures should be prepared and followed. Hydrogen should be purged from the system to create an inert atmosphere before working on tubing and joints.
NOTE: This incident is a good example of why fittings should not be used to depressurize a hydrogen system.
Setting — Laboratory
Description: An experiment was conducted to produce a small quantity of hydride material in an enclosed reaction chamber. During the experiment, the researcher observed a small flash and a whooshing sound coming from inside the chamber and extending into the glove bag. His hand was in the chamber, and he felt heat through his glove. In order for the flash to occur, both oxygen and an ignition source had to be present inside the glove bag. At the time the incident occurred, the normal inert purge gas (nitrogen) was being used in another project and was unavailable, so a decision was made to use argon instead. Apparently, the normal nitrogen purge gas flow rate was much higher than the argon flow rate. There may not have been a sufficient volume of argon or enough time to purge the glove bag and create enough positive pressure in the bag during the work activities. Reaction product collection activities may have caused purge gas to exit the bag (as designed), but when the glove bag needed additional purge gas to expand, it may have drawn in room air in addition to the purge gas, thus introducing oxygen. The ignition source for this event may have been spontaneous combustion with oxygen due to the extremely fine particle size of the reaction product. An electrostatic spark induced from the plastic glove bag could also have been the ignition source for the flash. However, if the inert atmosphere were of the proper quality, an ignition source would not have been a problem.
Reemphasize policies and practices on how process changes should be evaluated for direct and indirect impacts on the process. Increase the purge gas flow rate to ensure complete purging of the system. Extend the inert gas purging time of the reactor system before and after a run and use oxygen chemical indicator strips to indicate the quality of inert atmosphere in the enclosed system before opening the collection chamber.