[thedoc (OP)]

When a plane crashes, there’s always a huge search for the elusive ‘black box’ to solve the mystery. But what is this, and how does it work? Heather Douglas put this to David Barry, senior lecturer in aviation safety, from Cranfield University…
Listen to this Show

or  

If you want to discuss this show, or ask a question, this is the place to do it.

[CliffordK]

Wilf,

Interesting question.

I think currently the black boxes are part of the fuselage, and sink with the airplane.  Your generator doesn't appear as if it would work fully submerged on the sea bed.

However, there has been a discussion on whether they should be designed to eject at, or before impact, and then potentially transmit to satellites, radios, whatever.  Perhaps they could duplicate the black boxes, one to go down with the plane, and one to float on the surface.  The black boxes also have to be extremely bomb-proof durable.  Your system would have to be designed so that it would be fully protected, then autonomously dynamically deploy.

In the future, at least some of the black box data will be streamed to a satellite, rendering the black boxes less vital.

I heard today that the black boxes only hold about 2 hours of voice data.  Normally that should be enough to recover the last few moments of what was happening onboard the plane before a crash.  However, the Malaysian jet apparently flew for several hours after the incident occurred initiating the course change.  Potentially a transatlantic or transpacific flight could also fly for several hours after an incident before finding a suitable airport to land.

Anyway, as far as the black boxes, several months worth of pings would be helpful.  Perhaps one could have a primary and secondary system, so after 60 days or so, a secondary low-power system kicks in giving say a pulse every few minutes in order to extend activity for a much longer period.  Once the wreck is located, it may still take some time to locate the black boxes.  Or, if the plane broke up, it could have a debris field of several miles to scan.

[cheryl j]

This is probably a really dumb question, but if the location of my cell phone can be tracked, why doesn't every plane have a satellite tracking system so you'd at least know when and where it went down? 

[CliffordK]

MA370 had a number of unique circumstances allowing the plane to be "lost". 

It took 2 years to recover the flight recorders from Air France 447, but apparently they had salvaged some components of the plane, as well as several bodies within the first month after the crash.  Longer lasting pinger batteries would have helped in that case too.

Consider the passengers...  They recommend they use the seat cushions as flotation devices, or if they are lucky, an inflatable life jacket.  If they are lucky enough to survive the crash, a seat cushion may only give 12 to 24 hours of survival time in the best circumstances, less in the arctic.  Rafts or inflatable ramps, perhaps a bit longer.  Beginning rescue efforts immediately, and knowing where to start are key to improving survival.  Assuming the plane did go down over the sea, anybody that hasn't been saved by now has essentially zero chance of survival.

Any life-raft, or the ramps to be used as life-rafts should also have satellite uplinks as well as radio locators.

Anyway, transponders and satellite uplinks should have been able to provide reasonably good location information, but they either failed, or were shut off at about the time of the course change of the plane.

I agree that it is very odd that the air traffic controllers don't know where all planes are all the time, and there should be more redundancy preventing just turning off communications (assuming it was a deliberate act).

"Hand-Overs" should include a 3-way call so that an air traffic controller is always watching.
Malaysia to Airplane.
Malaysia to Thailand (land line)
Thailand to Airplane.

Loosing communication with the plane, or seeing a plane without transponders should have sent out immediate alarms, but the whole incident was hampered because it happened between multiple jurisdictions. 

Had the incident occurred halfway across the English Channel, I would hope the response would have been different including scrambling military jets from a half a dozen countries to investigate.

[alancalverd]

I agree that it is very odd that the air traffic controllers don't know where all planes are all the time, and there should be more redundancy preventing just turning off communications (assuming it was a deliberate act).

"Hand-Overs" should include a 3-way call so that an air traffic controller is always watching.
Malaysia to Airplane.
Malaysia to Thailand (land line)
Thailand to Airplane.

The function of air traffic control is to regulate the flow in and around congested areas. At any time there  are thousands of aircraft in the sky, mostly near population centers. Of all types, military, private, chartered, helicopters, gliders, balloons.... the ones least likely to cause a problem or come to grief are scheduled passenger and freight liners cruising over the open ocean. Hence as far as safety is concerned, there is no point in tracking them unless the pilot requests assistance.

Whilst it would be interesting and possibly preventative to recover the flight recorders from MA370, knowing where it went down would be of little value to the passengers and crew. I had a couple of instructors who had spent part of their early careers floating in the English Channel amid the smoking wreckage of their Spitfires, but that's a tiny area, heavily patrolled at the time by several navies, and bailing out of a single seat fighter is a lot easier than exiting a pressurised liner with 300 other guys fighting for the door. The only completely successful ditching of a modern airliner was done by a gliding champion, from low altitude, in the relatively civilised Hudson river. Most others broke up on impact with 100% fatalities.   

3-way handover is standard procedure, but if it involves a long cruise phase you may get "XYZ call ABC at 20 miles, they have your details, have a nice day" and hear nothing for the next 6 hours. ABC won't be interested unless you are at least 40 minutes overdue at the reporting point: you may have diverted for weather, met an unexpected headwind, or declared an emergency and landed at an intermediate point, in which case your primary communication is with PQR, not ABC. So for about 4 hours out of 6, nobody on the ground has any reason to talk to you or question your position.

The military radar function is to protect territory against incoming threats. A scheduled plane cruising at 35000 ft is not a threat, particularly if it is flying away from your territory, and even an abrupt course change is initially a matter for the civilian emergency services unless they request an escort.

As for the original question, battery power is proven reliable, more crash-survivable than most mechanical systems, and pre-flight-testable. However long the battery lasts, there will always be scope for a longer one, but again anything that lasts longer than about 24 hours  will have little influence on passenger survival in open sea.

We often carry or wear buoyant personal locator beacons in light aircraft. These transmit a standard distress signal and GPS position to the satellite network when activated by impact or by hand, but the advantage of a light plane is that, if it is landed in trees or ditched with a little skill, there's a fair chance of the PLB ending up on the surface. 30 minutes' battery life is then entirely adequate to initiate a search although the visible strobes usually work for a lot longer. Quite different from a black box strapped down in the hull of an airliner. 

[CliffordK]

One of the books I read long ago was "30 Seconds over Tokyo", about the Doolittle Raid. 

Anyway, the one of the planes ran out of gas about 100 yards from the beach, if I remember right.  All 3 crew members survived the crash, but they were pretty battered.  They concluded that their plane would have been much better off had they retracted the landing gear.  I think another plane on the raid landed on water, gear up, without significant injuries.

Planes are a lot bigger now, but they should be capable of water landings. 

Perhaps Boeing should buy back some of the end-of-use Boeing 747's, equip them for remote navigation, and crash them into the ocean just to determine the parameters for 100% survival. 

Busses, trains, trucks, and armored vehicles all use GPS locators, and it is possible to know where they are 100% of the time.  Somebody may not be monitoring all of them all of the time, but it is possible to program in parameters which raise alarms if one deviates from the planned course.

Say a pilot decides to divert to another airport for mechanical reasons or even a rambunctious passenger.  Somebody should start actively watching its progress, and make sure it actually gets to its destination.

As mentioned, in optimal circumstances, the open water survival time might be a day or so.  The survival in the water for the Titanic passengers may have been less than an hour. 

If one waits for a plane not to land at the destination airport, then waits another 40 minutes before doing anything, then one can guarantee nearly a 100% mortality rate, even if they had a perfect water landing.

Perhaps one doesn't have to monitor every Cessna in the sky, but any plane that is carrying over 10 passengers should have an active monitoring system.

As communication and tracking systems on MA370 either were shut off, or failed, I can not think of a single reason why they didn't raise immediate system wide alarms.

Part of the problem was that there was no central tracking of information.

Thai air traffic controllers - no "report in".
Thai military - noting aircraft on radar without transponder.
Inmarsat satellite - received "pings", but no additional data.
Malaysian air traffic controllers watched the plane go out of their control area.

Put it all together, and one should have noted the problem while the plane was still in the air.

An active tracking system would be one step ahead of all of that.

[alancalverd]

One of the books I read long ago was "30 Seconds over Tokyo", about the Doolittle Raid. 

Anyway, the one of the planes ran out of gas about 100 yards from the beach, if I remember right.  All 3 crew members survived the crash, but they were pretty battered.  They concluded that their plane would have been much better off had they retracted the landing gear.  I think another plane on the raid landed on water, gear up, without significant injuries.

Returning from a bombing raid, the plane would have been several tons lighter than at takeoff, flying slowly, and falling from a fairly low altitude. Most WWII bombers would glide at less than 100 knots with the flaps down, but certainly you should retract landing gear to ditch.

Planes are a lot bigger now, but they should be capable of water landings.

As demonstrated by Chesley Sullenberger, an A320 can indeed survive ditching on smooth water in the hands of someone used to dead-stick landings at low speed in daylight. Quite different from a terminal dive at 600 knots from 30,000 ft into a dark ocean.

Perhaps Boeing should buy back some of the end-of-use Boeing 747's, equip them for remote navigation, and crash them into the ocean just to determine the parameters for 100% survival.

The optimum ditching configuration is already in the operating manual, but far from easy to achieve in real life, and damn near impossible at night or over open ocean with no horizon or distance reference. 

Busses, trains, trucks, and armored vehicles all use GPS locators, and it is possible to know where they are 100% of the time.  Somebody may not be monitoring all of them all of the time, but it is possible to program in parameters which raise alarms if one deviates from the planned course.

But deviation from the planned course is not an emergency - it's usually the means of avoiding one.

Say a pilot decides to divert to another airport for mechanical reasons or even a rambunctious passenger.  Somebody should start actively watching its progress, and make sure it actually gets to its destination.

Your progress is monitored by whoever you choose to talk to, usually either your last contact or your new destination. Thanks to the blessed European Union we now have at least 2,280 nonoverlapping radio channels so it's up to the pilot to initate contact as nobody on the ground knows which channel he has selected.  Nobody on the ground or anywhere else can "make sure it actually gets to its destination": that's entirely in the hands of the pilot and the laws of physics. With luck you can say where it last appeared on radar.

As mentioned, in optimal circumstances, the open water survival time might be a day or so.

Actual ditchings are far from optimal. Just look at the shambles when you land on time at the right place: passengers ignore instructions, clamber over each other, and faff about doing everything except evacuate in a safe and orderly manner. The only thing that went wrong in Sullenberger's case was a dumb passenger opening an aft door when told not to. Getting a fit, trained, disciplined military crew out of a fairly intact bomber (they were presumably wearing lifejackets throughout the overwater phase) and swimming 100 yards to the beach is a hell of a lot different from getting 300 screaming or partially asphyxiated civilians out of a crushed airliner, however nicely the stewardess smiles.     

If one waits for a plane not to land at the destination airport, then waits another 40 minutes before doing anything, then one can guarantee nearly a 100% mortality rate, even if they had a perfect water landing.

But nearly always a 100% survival rate if they have landed at an alternate airport, which is where nearly all diversions end up.

Perhaps one doesn't have to monitor every Cessna in the sky, but any plane that is carrying over 10 passengers should have an active monitoring system.

It's the little Cessnas and their ilk that fall out of the sky most often, get lost, or wander into the paths of jets. Thankfully, commercial operators pay to keep air traffic control working for everyone's benefit. Airliners do have "active monitoring" but what they do and where they go is determined by the pilot and the laws of physics.

As communication and tracking systems on MA370 either were shut off, or failed, I can not think of a single reason why they didn't raise immediate system wide alarms.

Perhaps they did, but by definition having done so, it wasn't clear where to look or what to do. And if a plane flies for a couple of hours down a known track, it will take at least as long for another one to reach the point at which it disappeared. 

Thai air traffic controllers - no "report in".

Not usually required if overflying on a scheduled route. No reason for panic on the ground: flight may have delayed, cancelled or diverted.

Thai military - noting aircraft on radar without transponder.

Transponders fail from time to time. No apparent threat so no interest to the military.

Inmarsat satellite - received "pings", but no additional data.

Not sure what data they collect.

Malaysian air traffic controllers watched the plane go out of their control area.

And since it didn't re-enter, no further interest.

Put it all together, and one should have noted the problem while the plane was still in the air.[

Who "one"? what "problem"? Plane was flying under control, no evidence of mechanical failure, pilot diverted.

An active tracking system would be one step ahead of all of that.

But it can't fly the plane or guess the pilot's intentions.

[Ryan Mercer]

I can readily think of two ways to better help.

First, dead man switch type deal... an instrument in the cockpit beeps every 15 minutes or 30 minutes, GPS location is transmitted and the crew has 30 seconds to push a button. If a check in is missed a computer somewhere notifies a human being "flight xyz overdue for dead-man reset", immediately someone tires to reach them on radio if no one is reached you know where they were 15 or 30 minutes ago and have a much better idea of where to start looking for the aircraft if it is in fact missing assuming a new location isn't transmitted.

Second, 2 pods in the fuselage if the same forces from a crash are detected that trigger the pinger, tiny explosive charges go off blowing 2 access panels on the body and floating GPS beacons pop out and start screaming effectively saying "we went down right here" so you know EXCATLY where the plane went down and can start your search there.

[CliffordK]

I could imagine hitting a button every few minutes would be a pain. 

Radio status updates should be expected regularly anyway, with a method to ensure the plane doesn't get lost between controller jurisdictions.

There should be 100% tracking of every "flight ready" 5+ passenger plane, 100% of the time, whether they are on the ground or in the air.  Perhaps a status update ping every minute or so.

Tracking should be either decentralized, or centralized with redundancy.  Multiple data sources should at least be able to be accessed together (so, the Inmarsat data would have been available to accessed live by flight controllers looking for MH370 while it was still in the air, although at that time, the data wasn't organized to demonstrate a flight path).

Loss of a plane in the air or a significant unexpected course deviation should trigger an immediate investigation, starting with status probes by radio, then perhaps getting some military aircraft in the air. 

Mig-25 and Mig-31 planes or similar aircraft should be able to catch up with just about any civilian aircraft, although not really designed for search and rescue.  I believe the USA retired all of the equivalent SR-71 planes.  Range would be limited.  Scrambling some AWACS jets would also be beneficial. 

In the case of the MH370, after going "dark", it presumably crossed back over Northern Malaysia, and then went south parallel to the Australian coastline.  There should have been opportunities to intercept and shadow it, even without super-fast aircraft.

People have mentioned that an electrical malfunction could knock out some communication.  Redundancy would be the key, as well as not bringing all systems through a single common area on the plane.  However, any malfunction that would knock out 100% of the communication from a jumbo jet should still be considered an emergency.

[Ryan Mercer]

I could imagine hitting a button every few minutes would be a pain. 

They are pushing buttons, monitoring instruments, filling out volumes of paperwork etc for every flight, pushing one more button every 15-30 minutes wouldn't be too much more to shoulder. Simply having it go *ding ding ding* until you press the button wouldn't be too annoying, at least not for a fast food worker that's doing the exact same thing 8 hours a day with fryers. If it's acceptable enough for the deep-fryer at McDonalds it shouldn't be considered too annoying in the off chance it saves lives by providing a much smaller search radius in the event of a crash.

I believe the USA retired all of the equivalent SR-71 planes.

Almost certainly replaced by better, classified, technology. Only reason we know about the SR-71 is because they were decommissioned and their existence declassified. However, I see no need to have fast aircraft searching large areas when you could just build in systems that give you a much smaller search area in the event of a crash, like my dead man system.

[alancalverd]

Evidence suggests that all the tracking comms on MH370 were intentionally disabled, incuding one (ACARS) that doesn't have an "off" switch but only a circuit breaker. No reason to suspect explosion or other malfunction, just pilot choice.

I'm inbound to A and get a call "runway obstructed, suggest you divert". So I call B and divert. No problem - happens from time to time and every pilot always has planned alternates. But if I forget to push the deadman switch (a diversion is a busy time, what with explaining to the passengers, talking to B C and D to find a landing slot, resetting all the NAV kit, trying to find the approach charts...) I'll invoke a major pandemonium of expensive airplanes looking for wreckage even though there's nothing wrong. Pain in the taxpayer's backside if the missed response was over land, and no hope of achieving anything useful if it was in the middle of the Pacific, 5 hours' flying from anywhere.

Radio call from The Centre? Well yes, but I have over 2000 nonoverlapping channels on the radio - which one are they going to call? And as they are all VHF, they are essentially line-of-sight only: 200 miles is pretty much the limit in good weather, so you will need to establish repeaters all over the ocean. 

[Ryan Mercer]

I'm inbound to A and get a call "runway obstructed, suggest you divert". So I call B and divert. No problem - happens from time to time and every pilot always has planned alternates. But if I forget to push the deadman switch (a diversion is a busy time, what with explaining to the passengers, talking to B C and D to find a landing slot, resetting all the NAV kit, trying to find the approach charts...) I'll invoke a major pandemonium of expensive airplanes looking for wreckage even though there's nothing wrong. Pain in the taxpayer's backside if the missed response was over land, and no hope of achieving anything useful if it was in the middle of the Pacific, 5 hours' flying from anywhere.

No. Just no. The system would have an audible alert (visual too). How is a multi-person flight crew going forget to push the button going "ding ding ding ding ding" and then completely ignore when they radio you "flight abc please respond, you have failed to push your button" and then continue to ignore "ding ding ding ding ding" and radio calls until you show up at a runway to land "oops, sorry wasn't paying attention to you screaming at me on the radio to reply and the ding ding ding".

If you can ignore ding-ding-ding-ding followed by a radio call after a minute or two of the audible alert, you probably are dead and if you do happen to be alive you have no business flying an aircraft after ignoring an alarm AND radio calls.

[CliffordK]

If one has real time monitoring, then the "dead-man" switch is redundant.  MH370 disappeared because it made an odd turn and either disabled or otherwise lost function of its communication equipment.  Real time monitoring would have either detected it "vanishing", or been able to better track its movements and alert authorities that the course didn't make any sense (no matter if someone was pushing a switch or not). 

Consolidating information from multiple groups and jurisdictions is complex, but would have helped with the early response and search for the lost aircraft.

Some wreckage of Air France Flight 447 was found shortly after it crashed, it just took quite some time to locate the sunken airframe.

As far as the expense of false alarms...  consider the cost of searching for MH370 will likely top $100 Million.  The search for Air France 447 cost around $40 Million.  It is quite possible that better use of our Military resources would have reduced the associated search costs.  The hijacking and WTC attacks would have required a very rapid response, but might have had decreased impact had the military responded.

I believe the USA retired all of the equivalent SR-71 planes.

Almost certainly replaced by better, classified, technology. Only reason we know about the SR-71 is because they were decommissioned and their existence declassified.

The military likes to parade around their very expensive toys.  And it is now all on Wikipedia long before the first flight.  NASA had a Russian Tu-144 for a while, but apparently canceled the program in 1999.  The F-18 is sub Mach 2 (still snappy, but not as fast as the Mig 25/31).  The F-22 is only Mach 2.2.  New faster planes are in development, but I don't believe they've been deployed yet.

The high altitude reconnaissance has mostly been replaced by satellites, and now drones, thus the U-2 and SR-71 became obsolete.

If the US military had been looking for the MH370, they probably could have picked up the IR signature, or perhaps the jet trail/contrail via satellite, but there was not enough information early enough to make it happen.

[alancalverd]

If you can ignore ding-ding-ding-ding followed by a radio call after a minute or two of the audible alert, you probably are dead and if you do happen to be alive you have no business flying an aircraft after ignoring an alarm AND radio calls.

I say again: radio range is 200 miles at best and a lot less in bad weather, and there are over 2000 channels to choose from. How is anyone going to call me in the middle of the Pacific?

Now imagine you are landing in weather minimums. Let's make it a multicrew ship. You've been held at 6,000 feet for 20 minutes (because it was really disgusting on the ground) and now you are cleared to make a precision approach. Your P2 is handling the radio and calling out status and course corrections, and you are using all three hands and feet to keep the ILS centered on what you hope is the runway. Your brain is in overdrive (believe me - or try it in a simulator!). Are we optimally configured? What is the best braking procedure for this runway state? Crosswind? Wheels down and locked? Downbursts? Any horn, bell, bleep or ping may initiate a go-round but do we have enough fuel, and what is the missed approach procedure? And meanwhile we are pitching up and down and side to side. Then your bloody dinger starts dinging. Do you waste a hand and several neurons to cancel it, or concentrate on making a safe landing, ignore the dinger, and thus scramble fighters and helicopters to do a pointless low-level search in a hurricane?   

Automatic GPS may help to find wreckage, but when things go wrong, or even get slightly interesting, you really don't want to waste time and effort pushing the button that you are supposed to push when things go right.

[CliffordK]

I say again: radio range is 200 miles at best and a lot less in bad weather, and there are over 2000 channels to choose from. How is anyone going to call me in the middle of the Pacific?

Are planes still using analog radio?
A digital satellite uplink is the way of the future, and it may well be more robust than the old radio technology.  One may not wish to dispense with radio communication altogether, but digital point to point communication is very nice. 

While satellites are concentrated over populated areas, I presume that almost anywhere one can acquire satellite uplinks, except perhaps when crossing the North or South poles.  Ahhh, with the satellites orbiting at 36,000 km, they are even accessible at the poles.

With the Apollo rockets, there was a communications blackout during a critical 5 minutes of re-entry where the ground station just hoped to hear from the crews again.  However, since our airplanes are using century old technology, they may enter communication blackout zones lasting hours, and the ground stations just hope they show up again?  Then if they are a half a day late, someone might go out and look for them?  I feel safer all the time!!!

[alancalverd]

Are planes still using analog radio?

Yes, and for several very good reasons. Most importantly, lack of capture by strong signals in adjacent channels, and the ability of the ground station to "stamp on" a distant aircraft in order to talk to a nearby one - it's the guy nearest the airport who needs priority service, and the 200W ground station is issuing orders which must be obeyed whilst the rest of us with 5W airborne are passing the time of day.   

  However, since our airplanes are using century old technology, they may enter communication blackout zones lasting hours, and the ground stations just hope they show up again?  Then if they are a half a day late, someone might go out and look for them?  I feel safer all the time!!!

Not a real problem. If you survive a ditching (unlikely) your position will be reported via satellite by the ELB in your liferaft. If you don't survive, the wreckage is of purely academic interest.

Time to get real, folks. There are about 100,000 scheduled flights every day, and about the same number of military, private, commercial and business flights. Say  some 70 million flights per year. About once every ten years, one disappears without apparent reason. So how much additional time, money, talent and aggravation is it worth spending to find out what went wrong once in a billion times? And in the case of MH370, there's nothing to suggest anything went wrong - all the evidence implies that the pilots decided to fly somewhere other than the scheduled destination, and either they landed at a place they don't want you to know about, or gravity took over when the fuel ran out. Knowing where won't change anything.