Hemingway once wrote that a man goes broke “slowly, then all at once.” There may be months of lead-up to an accident. Maybe there’s a weakness baked in right from the start that only crops up after years of operation or an unusual cascade of errors. Maybe there is some seemingly insignificant change whose dangerous consequences were never considered when the design was made so long ago.
But from the flammability analyzer’s point of view, the final development of the worst accidents occurs in seconds. It is part of the last line of defense against fire and explosion. There’s precious little time between the rise in concentration and the explosion. That’s a very good reason to pay close attention to the analyzer’s response time.
For many industrial operations, especially large industrial machines with ovens and dryers, a good model for an explosion is an uncontrolled release of an excess of liquid fuel, typically solvents, into a heated chamber. This produces a sudden increase in concentration limited only by the solvent’s vapor pressure and the temperature. The concentration quickly ramps up, and within seconds it exceeds the Lower Flammable Limit. Another few seconds, a doubling of concentration, and it reaches stoichiometric. It is now perfect for an explosion. It is the concentration easiest to ignite, and the one that produces the highest burning velocity and pressure rise.
“I heard the alarm go off, and when I turned to look, the ceiling came down on me.” From the loss of control that flooded the machine with solvent to the explosion took maybe twenty seconds. The concentration was climbing 10% LFL per second or more. The time to alarm was only a few seconds faster, not fast enough to do any good. Why? A long sample line was installed, and it took ten seconds for the sample just to reach the analyzer. With the concentration climbing 10% LFL per second for the ten seconds it took to reach it, the analyzer was running not just ten seconds, but 100% LFL! behind the actual event.
The analyzer’s T90 time is pretty meaningless if it isn’t designed and installed correctly. A time lag from sample tubing and sample filters can take too many seconds. So consider only the “time to alarm” as the analyzer is actually installed. Of course, a slow analyzer never gets any better even if it is installed perfectly. But more commonly, a reasonably fast analyzer is ruined by added sample tubing and filters. The acid test is to inject a high concentration at the sample probe, and measure the seconds until the alarm activates.
So a good question to ask is this: what happens when the concentration suddenly rises by 15% LFL per second and keeps rising to at least 200% LFL?