Thursday, July 03, 2014

Photo Finish

I checked with maintenance regarding my recent post about the fire protection system. Our fire detector is photocell activated. The photocell is sensitive to a particular wavelength of light, supposedly characteristic of fire.

I didn't get to ask the obvious questions about why fire detection systems aren't overwhelmed with false positives from the sun, an awesome source of almost every electromagnetic wavelength, and able at some time of the day or some practical bank angle, to shine through every chink and opening in an engine cowling. An apprentice had a more pressing question about reinstallation of an exhaust manifold, and I was wearing silly girl shoes for a social event, so I teetered back out of the hangar to the office and left the maintenance to the maintainers.

Along with the answer to my question, and before the apprentice needed direction, I received the following story. A certain air carrier's night maintenance crew (they know who they are) conducted an inspection on at least one of the carrier's fleet. Their fire detection system used visible light, and the crew followed the manufacturer's instructions to test it by shining a flashlight on the detectors. The system failed the test. The crew was preparing to replace several thousand dollars worth of aviation grade electronic gear, when someone realized that that instructions were as old as the system, twenty years or so, and predated the invention of the LED flashlights the crew were using. The diodes were not emitting the necessary wavelength to trigger the detector, because when the instructions said "flashlight" an incandescent flashlight was assumed.

 The Wikipedia article on Fire Detection suggests that our photocell might be tuned to multiple wavelengths, possibly including UV and/or IR to compare the ratios. This still doesn't explain how an ordinary incandescent flashlight would trigger it, but the sun wouldn't, but does make the concept of photometric flame detection clearer. While looking for more information on flame detectors I discovered that there is a whole new generation of fire detection that uses machine intelligence to recognize the appearance of actual flames, something that is pretty fricking clear to humans. Combined with smoke detectors, something a commenter suggested on the previous post, there are enough different kinds of combustion detectors on aircraft to warrant a survey post on the topic, some time when my working day is shorter.

I was also entertained by the Wikipedia article on the Flashlight, which explains why Americans call them flashlights. The earliest zinc–carbon batteries could not provide a steady electric current so when combined with the inefficient carbon-filament bulbs, the result was a light that frequently blinked off so the battery could recover.

13 comments:

Josh said...

I came across a colleague in the sim who said the King Airs he flew had exactly this problem - spurious fire warnings when flying close to sunset at a particular sun angle.

They learned the hard way after an engine shutdown in flight.

Aviatrix said...

If I had to name one reason why I kept this blog I'm not sure whether it would be "because I have to tell someone all this stuff" or "because when I wonder things, people like Josh have stories that exactly answer my questions" would be the first reason. I am so happy to read this comment. Maybe it's because I have a science degree and this resembles the delight of having an experiment deliver data that supports a hypothesis.

Anonymous said...

The sun is much hotter than either an incandescent bulb or any reasonable engine fire, so I expect sunlight contains much more UV (relative to the amount of IR) than the light from either of the others.

Anonymous said...

Ha! All great points and I SURE DO HOPE that your systems are working. That said, a lot of 'flying' fine and smoke detection systems are as simple (or worse) than the early in-home detection systems. Not worth a dime! In some cases, "Smoke" detection is little more than the loss of a reliable signal from an emitter to a detector, or measured heat change over a time-limited period. Common folks!! The science of fine/smoke detectors HAS advanced beyond that state and far better systems ARE available. Why they are not more widely used is a COST issue, addressed only by the operators. For freighters or mission-specific aircraft, some will argue that two or three at risk is different from 400-500 at risk. I refuse to qualify the value of life in quantity, although I do understand the differences. Do the operators care about their crews?
The AVH website reports a constant string of 'cockpit or cabin smoke or odor problems that usually result in expensive diversions - and with no fault found. (The fifth force does not create those odors or smoke.) And, Of Course quick diversion is necessary, but at what expense?) Far better and far more reliable detection systems ARE available, so why not use them? Cost Cost and COST! Pilots routinely flying one, two or three-person crews, hauling freight or executing technical missions are NOT any less valuable than those with 400-500+ souls aboard, so say I. Why not mandate state-of-the-art F&S detections systems - Those that really work... for all commercial aircraft? Air operations are expensive - Duh? Are smaller, lower level commercial operations worth a far greater risk to a few pilots and a handful of crew? Ahem!! NO! The civil authorities should mandate state-of-the-art detection and alerting systems for ALL commercial aircraft, much as they did with TCAS systems a few years ago. Pilots (and non-flying crew members) in commercial, but non-passenger carrying operation such as Madam Avaitrix fly ARE entitled to the the best safety systems available. They are professionals trying to earn a living. Are they not entitles to the same basic safety standards applied to any other kind of work? Hell yes! Smaller scale commercial flying is not exempt from safety regulations - just because a few are willing to take the risk. Those pilots should not have to take those risks, just to earn a living and/or build the hours necessary hours leading to a slightly better job. Please, let's not kill them before they get there. Did I mention ALL **Commercial** Aircraft? Yesterday was not too soon.

Michael5000 said...

I'm not going to spoil the joy of rank speculation by looking into the matter, but my crude understanding of lightbulbs has always been that the electrical charge renders the filament more or less on fire. If that assessment was more or less on target, it might more or less explain why a traditional flashlight bulb would simulate better than one of them newfangled LED jobbies. QED.

Michael5000 said...

* would simulate FIRE, that is.

Aviatrix said...

Oh yeah, no problem understanding why an LED flashlight would not do the job an incandescent flashlight would. In fact while the story was being told, I understood what was going on the second I was told they were using an LED flashlight. The issue was how an incandescent flashlight simulated fire and the sun, which is also basically fire, did not. But as Josh has given an example of how the sun DOES give false positives, it is now all understood.

Rereading the post I see that I spent no effort explaining that, expecting the reader to leap to the same understanding, but perhaps I should clarify. Nah, they can read the darn comments.

Anonymous said...

Reminds me of a story I heard about troubleshooting analog electronic circuits that had a hum (mains frequency, 60 Hz) no one could explain, and only when the cover of the box was opened. Turned out that glass-packaged diodes picked up light from fluorescent tubes and altered their behavior such that the light-induced modulation could get into the original signal.

Once you know the answer, every problem is trivial...

-zb (let's see if the name shows up automatically or if I end up being officially anonymous - but I promise I'm still real...)

Aviatrix said...

Sorry, zb for the extra effort ir may take to comment. I just couldn't take the spam comments anymore.

Anonymous said...

No worries. I just think it's funny that the link above my comments says "Pay no attention to the man behind the curtain." - The effort isn't too big fo rme because I can use an existing openid accout. It works, and I'm not bothered that google doesn't make it look too pretty.

mattheww50 said...

Most likely the system compares the ratio of two wavelengths of light (IR and visible). From that information you can determine the temperature assuming it is a reasonable fascimile for a black body (most things are). The 2700K or so that you get from an incandescent lamp is a pretty good approximation for burning carbon in a flame. The sunlight has a temperature of about 5000K.
The LED flashlight has essentially no IR component, so the apparent temperature seen by the sensor will be far far above 2700K. Unlike the filament of an incadescent lamp, or the sun, an LED is NOT a black body radiator

Unknown said...

Cedar Glen, you raise an interesting point. It brings me to another interesting thought which may bolster your argument.

Fire detection systems on turbine powered airplanes are, generally speaking, far superior to those in use on piston airplanes. Also fires are far less common in these engines. I have flown variations of one type of Gerrett turbofan since 2004. This engine went into production in 1970 and over 11,000 have been built to date. I was told recently during a training session that there has never been an inflight engine fire on one of these engines. Yet, sophisticated fire detection and suppression systems get installed with every one of these engines.

When you compare the above with the higher likelyhood of fire in a gasoline powered piston engine and poorer (or no) fire detection / suppression systems, you end up with a compounded inequity in safety. It would seem almost more logical to install the better systems in the airplanes with the higher likelyhood of fire but, for manufacturers and for operators, it creates a bit of a conundrum. Every new piece you bolt onto an airframe weighs something, and weight eats into the usefulness of the airplane. The turbine powered airplanes have much higher useful load capability, and that after bolting all sorts of gizmos on to the airplane, many of which are not available to, say, a pilot in Northern Ontario in a Navajo hauling people or freight around.

Not an easy problem to solve.

Sorry Aviatrix... I didn't mean to stray so far from the original topic!

Majroj said...

Late to the party…
1. Artif intelligence with proper sensors can spot flames the human cannot (such as well-oxygenated alcohol or other smokeless liquids), and sometimes fire behind a visual obstacle.
2. Ratio of wavelengths is a key. Some systems also use "flicker detection" since most combustion is not perfectly constant.
3. LED flashlights…had to carry an incandescent light for diagnosing patients as the LED I carried for power outages would not show redness or blood properly for the human eye to detect. So most if not all LED's are nowhere near what most combustion looks like.