So, I just watched the Breaking Bad episode with the midair collision. I thought it had to be unrealistic writing, but was shocked to see this has happened in real life. From playing video games, I know it is very easy to get objects to collide in 2 dimensions, but very hard in three dimensions. So why do air traffic controllers always send planes to altitudes that are round thousands, like 10000, 20000, 30000, etc. feet, turning a 3D scenario into several different 2D scenarios? Why don’t they just fuzz the numbers a bit? If they want to send a couple planes to 20000 feet, why not just make up an arbitrary deviation and tell one plane 20356 and another 19765? They can still pretend those planes are at 20000 feet, and keep all the 2D safeguards for the 20000 feet layer. But now if an unlikely scenario occurs where those planes would collide in 2D, they will miss each other by 591 feet (or whatever the fuzzing adds up to). Since all the 2D safeguards are kept, there wouldn’t even be a be a need to remain cognizant of the fuzz amounts. Why don’t they do this?
I’m a charter pilot, not ATC. There are numerous rules governing what you’re asking about, and rather than trying to list them and their many codicils and exceptions, I’ll just say… They also separate planes by distance as well as altitude. That pretty much negates most collision risks.
As to your suggestion of (very) odd altitudes like 19765… The rules allow for some leeway in altitudes, but not a lot. This is due to inherent degrees of error in altimetry. But basically, unless you’re a test pilot, you’re not expected to hold altitude plus or minus one foot. Modern autopilots are very good, but I wouldn’t trust my jet to that tolerance, and it’s very modern.
ATC has many, many procedures and double-checks to prevent collisions. And OP, you’re correct that it’s quite difficult to put two high-speed aircraft into a collision. I seem to recall it’s hard to do even when it’s attempted intentionally.
That said, the system is slowly changing. We’re going away from a radar-based system, more toward satellites. Not sure what that will do to our rules and tolerances, but the current system works well.
Also, turbulence can knock you off your Flight Level.
Anyway, I found some Eastbound/Westbound FL rules -
Aircraft do not fly so precisely, especially in turbulent air, and this “fuzzing” would reduce the buffer between altitudes for which horizontal separation
(what you call “2D safeguards”) is not maintained.
I also introduces added complexity, that at present can only be communicated verbally by radio, increasing the potential for the kind of human error that is usually the cause of rare midair collisions - and with negligible benefit, given how well the system works already.
To my mind, the strangest element of commercial aviation (aside from the fact that we still have human pilots!) is the fact that we still do rely so much upon verbal communication by radio. Of course there’s a cross-check on altitude with the data that ATC see from the transponder, but primary instructions are given verbally over the radio. Perhaps we might stop using round-number altitudes when this changes.
enipla beat me to it, and posted with more detail that I was going to.
I’ll add that IFR traffic is assigned round-thousands of feet. VFR traffic flies at thousands+500 feet.
A mnemonic to remember the information in enipla’s post is ‘Odd fellows fly East.’
ETA: VFR and IFR Cruising Altitudes
.
Until recently, in UK FIRs, the quadrantal rule was in force instead of the semi-circular rule. This meant that you flew odd thousands, odd thousands + 500, even thousands or even thousands + 500, depending on what direction you were flying.
This meant you didn’t have a situation where planes on near-complementary tracks were flying at the same height.
A plane flying from Ireland to Britain on a track of 145M would fly at 3000’ (semicircular rule) in Irish airspace as far as the IFR boundary, then climb to 3500’ to conform with the quadrantal rule.
Unfortunately now the quadrantal rule has been abolished.
GPS-linked autopilots now permit position and altitude to be held so precisely that there’s reason to think there’s now an *increased *risk of collision between planes at the same altitude and on the centerlines of the same airways. Some pilots will deliberately offset their routes and altitudes, within IFR tolerances, as a preventive measure.
I recall on flights to or from Logan (Little Rock-Dallas-Logan and back) seeing aircraft flying the opposite direction below mine.
I do something similar when flying en-route low-level VFR, i.e. never fly at round-number altitudes (there is no obligation to do so for low-level VFR flights).
But to the extent commercial pilots do this, I think it’s also en-route, when FL separation is more substantial, and plus-or-minus a couple of hundred feet presents no risk of encroaching on the adjacent FL. Does any commercial pilot deliberately use offset altitudes on approach/departure, when vertical separation maybe much tighter?
Pretty well covered already.
The short answer to why it’s the way it is now is that’s the way it developed back then. In the Olden Dayes altimeters weren’t all that accurate. And altimeter settings weren’t all that accurate either. With the result that two airplanes both flying in almost the same spot at, say, 20,000 feet might be 200 or 300 feet different from each other in actual true altitude.
So each ply of the 3D collision avoidance solution had to be some altitude +/- the 99.9% tolerance stack up of bad altimeter setting, bad altimeter, and bad flying. That meant, and in much of the world still means, 1000 foot separation in the lower altitudes and 2000 separation in the jet altitudes.
Any effort to fuzz the altitudes would be statistically increasing separation between airplanes at the same altitude, where lateral separation is totally required, and simultaneously *reducing *statistical separation between altitude plies where lateral separation is absolutely *not *required.
IOW, statistical safety would be reduced by fuzzing altitudes, not improved. And that’s before we consider any increased budget for errors with more complex altitude clearances.
In general, if we were inventing separation rules today from scratch, and assuming all airplanes involved always have all the latest gizmos working correctly, we might choose a different mix of vertical and lateral separation standards.
But we’re not starting from scratch and most of the planet can’t get too far ahead of the oldest slowest country and the oldest least-well equipped airplanes. There are ever-expanding pockets of airspace where you need the latest toys to play in that sandbox. But it’s the work of decades to finally make such upgrades mandatory for 100% of every fleet and every bit of airspace.
Nobody operates IFR airplanes off altitude on purpose. Nobody. TCAS gets antsy if you’re 100 feet off and being 200 feet off is a fairly automatic violation.
VFR people operating above (IIRC) 1200 feet AGL (or is it 3000?) who are not operating at a round 500’ and correct thousands for hemispheric direction are committing a violation and endangering other aircraft in the airspace.
As mentioned by others above, there are provisions, mostly in oceanic airspace outside of radar or ADS-B surveillance, to operate with pilot-chosen *lateral *offset (*not *vertical offset) from the official GPS track. As much as anything, they’re designed to resolve wake turbulence concerns, not reduce collision potential. In any case, pilot-chosen *lateral *offset is not permitted in any radar-controlled airspace I’m familiar with.
Ref the earlier comments about voice radio. … the industry is driving towards more use of datalink messages for control. Certain parts of the Pacific and Atlantic are that way now. Before I retire in a few years it’ll be commonplace in at least some of the high altitude airspace in the US. For some tasks I’m sure it will be an improvement. For others it will not.
One significant thing that will be lost is that we each learn a lot listening to what’s going on with all the other aircraft. Which altitudes are smooth or not. How the guy ahead of you on the same route is handling thunderstorm avoidance, etc. How busy is the controller which correlates to how busy is the airspace and how much personal attention we can reasonably ask for. And best of all, listening in to the party line promotes “situational awareness” and error checking. If the last 4 airplanes on the arrival have all been told to descend to 12,000 and you’re told to descend to 11,000, it’s smart to ask why the difference. Sometimes it’s on purpose, sometimes it’s a goof. If, in the future, all you receive is text messages on your screen, you’ll be very hard pressed to reasonableness-check them as effectively as we do now.
Everything we do and everything we change in how we do it involves safety & cost tradeoffs. Which are pretty carefully analyzed by a lot of smart people with a lot of data to hand.
Talking about the lower end of General Aviation, it has always been interesting that in actual practice, what LSLGuy said above and the whole thread in general covers big iron stuff very well.
The little guys who live & die down in the muck with the old and not so well equipped aircraft really need to remember that the ears are the early warning devices of things going pear shaped with the aircraft, and it’s passengers.
The eyes are the outside is the world of eyes are best. Along with instrument scans that help with aircraft health.
This combination requires constant attention and that is hard to do for long periods.
As of yet, there is no tech that can replace these abilities yet in the safe operation of the flying world. Add in the fact of 50 & 60 year old aircraft still being flown …
3000’ AGL in the US, other countries differ. FAR 91.159 applies to cruising flight only, of course.
Airways are +/- 4 NM wide. If you’re in that range, you’re legal, aren’t you?
That tolerance is based on VOR sensitivity at the center of an airway, IIRC. These days, I’d guess the majority of IFR traffic is navigating the airways with GPS. Hence the need for SLOP (Strategic Later Offset Procedure). GPS can keep you on the center very accurately at any point on an airway, which makes it too accurate for its own good under a system designed originally for VORs.
Slightly off topic, but another example of GPS being a bit too accurate can be seen with older autopilots. When I was instructing I saw a lot of small GA planes with Garmin 430 / 530 units. If equipped with an autopilot that could utilize nav data, you’d often see the plane do fine with VORs but snake back and forth when flying GPS courses. This was because the older mechanical autopilots were not designed to handle tolerances that were so accurate. I usually suggested to those pilots that they fly in heading mode and just follow the purple line.
Thanks for the cite. It’d been years since I’d thought about that rule.
91.181 requires operation along the centerline of the route cleared. Unstated, but in the case law, is that means “to the best of your/your aircraft’s ability.”
The +/-4 miles is about airspace design and terrain clearance. It’s intended to absorb the fact that, especially pre-GPS, being a couple miles left or right of where you intended to be was as precise as was possible.
Allowing folks back in those days to deliberately add a couple miles offset would put them statistically outside the safe area.
You’re right that today, where we see enroute nav performance within a couple dozen feet left or right, some of that fudge *could *be used as offset and still keep the aircraft within the statistical safety envelope. But that’s not the way the regs or procedures are written today.
The key IMO is that lateral offset won’t help in the US where the most common risk situation is aircraft crossing paths co-altitude. As well, it won’t do much in a lot of the US where most airplanes are not on published routes but are instead going directly to some random place a long way away. It also doesn’t do too much in those oceanic areas with mostly one-way traffic. Which includes the North Atlantic and much of the North Pacific.
Where it would / does have value is in regions with bi-directional traffic on common routes. And doubly so in regions where climbing and descending is commonly done along the enroute structure rather than in dedicated “on- and off-ramps” = SIDs/STARs.
I never did instrument training, but other pilots I talked to who had, they mentioned that the instructor would fail you if you deviated by 25 feet (or was it 50?) either way from assigned altitude.
IIRC ATC collision errors would be one aircraft descending onto another - the visually hardest part for pilots to see, above and below (and why distance separation is crucial). Most flight paths around airports involve fixed “lanes” and circuits so it’s less likely two planes would be randomly assigned to head toward each other, particularly with 180-degree differences in direction.
There was an accident in northern Canada when GPS units first came into general use; in the middle of nowhere, two aircraft collided head-on in the middle of nowhere. they were both following the straight line from A to B (no explanation why they were not properly altitude separated). After that, the recommendation was to always fly X feet to the right (?) off the actual GPS line.