Dust Explosion in a Specialty Pharmaceutical Plant

(West Palm Beach, Fl) February 6, 2006 -- The explosion occurred in a blending operation handling particulate bulk solids. Multiple ingredients are added to a mixer manually from 50-pound sacks. The mixer is a tumbling type multifaceted drum that is located in a stand alone, dedicated room. The ingredients are added from a staging area located in the room immediately above the mixer through a port that is closed via a butterfly valve for mixing. The rotation is quite slow and the mixer is supported by externally mounted bearings. There is a shaker screener on the top of the fill port. Manual additions of ingredients in small quantities, no moving parts, external bearings, slow rotation speed, and excellent housekeeping all support the conclusion that in ordinary circumstances this apparatus would not be considered to be a likely locus for a dust explosion.

Nevertheless, an explosion occurred in the mixer drum during the addition of the last but one bag of ingredients. The explosion developed slowly as these things go, with the initial ignition occurring at the bottom of the drum, some 16 feet below the worker who saw the flame as he was pouring the contents of the 50-pound sack into it. Incredibly and most fortunately, although injured, this worker was able to turn away in time to avoid what could have easily been a fatal experience.

The flameball left a clear imprint on the staging room via burn marks and heat discoloration that extended throughout the room and out of the room via open doorways on two sides. One sprinkler head that was above and in close proximity to the fill port opening was knocked off the piping and this allowed water to pour into the room. Other sprinkler heads in the room were intact, undamaged, and did not deploy. Obviously, dust explosions happen too fast for sprinkler systems.

Investigators found faulty bonding and grounding for the shaker screener and for attached 4-inch diameter, wire-reinforced, plastic tubing that was used to connect the screener to a central dust collector. There was no evidence of flame spread propagation to the dust collection system. Since the faulty bonding and grounding was found above the mixer (the bonding and grounding for the mixer drum was OK), but ignition was observed to occur at the bottom, it was concluded that a static spark from this source would not be a likely root cause factor in this accident.

The ingredients were sent to a lab for testing for reactivity (Kst and Pmax), for ignition sensitivity (i.e. Minimum Ignition Energy - MIE), and for resistivity. It was found that the material being added at the time of the ignition had several peculiar properties: It had low MIE (<10 mJ), high resistivity (>1010 ohms), and a fraction of it was so fine that it behaved and looked more like smoke than like dust. The other materials all had relatively high resistivity values, too, which led to the conclusion that the source of ignition was a kind of transitional bulking-brush discharge that ignited the vapor like component of the ingredient being added. Although, it would normally be considered to be very unlikely that manual additions of material in the quantities encountered in this process could be ignited via a bulking-brush discharge, this was the most plausible explanation for this event.

Lessons learned: Care should be taken to learn of any materials being processed that have low MIE. Generally, I suggest that any particulate material with an MIE = 20 mJ should be carefully considered and handled in a manner so as to guard against static ignition. In this case, the solution included properly bonding and grounding all equipment, and the ingredient involved was replaced with a much less reactive alternative. Even though conventional wisdom would suggest that bulking-brush discharge ignition would be most likely to occur in large process vessels such as silos that are being filled via rapid pneumatic conveyors, care should be taken to recognize that even smaller quantity operations, with manual filling can lead to trouble for materials that have both high resistivity and low MIE. Although no harm came from the wire-reinforced, plastic tubing in this incident, this material was removed and a flex-metal duct was installed instead. Plastic tubing can generate a lot of static and is not recommended especially when handling low MIE dusts. Metal wire has been found to break in the tubing and it is not sufficiently reliable or effective to bond and ground it.