[t0mas88]

The Cirrus overshot the centerline of the runway it was supposed to land on. The common way for these situations is for the plane to be on an intercept angle towards the centerline. That's a max 30 degree angle for an instrument approach, but this was a visual approach so it could have been a sharper angle. All it takes to make this mistake in a Cirrus (and other G1000 avionics type small airplanes) is to forget 1 button on the autopilot mode. If it isn't set to capture the final approach track (either GPS or ILS) it will continue straight ahead which in this case means into the side of another airplane.

One thing that makes it more likely is that US air traffic control makes heavy use of visual approaches, and then it's allowed to point two aircraft at collision courses on the same altitude because they can see each other. The European way to do this is to have them intercept at different altitudes so if one overshoots they pass over/under. But it results in lower capacity per runway than the US system.

[azalemeth]

Indeed.

Something else I'd like to point out is that it might seem easy to 'blame the Cirrus' pilot or them call out for inattention, but doing so by itself isn't helpful. Aviation is so safe partly because it has managed to turn a culture of blame into a culture of continuous improvement and shared learning: I'd be very surprised if the airport's procedures came out of this unmodified, for example.

[elliekelly]

I’ve probably mentioned this book on HN a hundred times but “Black Box Thinking” discusses the aviation culture of avoiding blame and making sure a problem isn’t repeated and how other industries (like medicine) would benefit hugely from a similar approach. It’s really interesting.

[lbriner]

I'm not sure how scalable this is in modern penny-pinching times. In the old days, airlines had to prove they were safe to attract business and this involved accepting that certain practices were harmful and they were therefore improved.

Once we got to the 1980s, we had so many airlines trying to survive that corners were cut, recommendations were not followed and various accidents were essentially negligent.

Now that lots of smaller airlines have been merged into larger ones, we now have Boeing type problems where the cost of manufacture, safety and development is so much higher than before, no-one wants to put a new plane through the whole approvals process, we just want to re-badge a 737 and get it into service.

Similar things happen on the railways in the UK where we have the RAIB to do a similar "no-blame" analysis of a crash/accident yet still time and time again, the same problems surface - lack of preparation, lack of training and lack of following procedures.

[elliekelly]

I agree and it’s somewhat addressed in the book - if you find the premise at all interesting you should definitely read it, even if you’re skeptical of the practicality of implementing the philosophy (for lack of a better word) it’s interesting and the examples are compelling.

Some of the examples are whole-cloth cultural changes of entire industries (usually commercial flight actually, IIRC), but some are small, simple, changes that can be implemented by one or two people and still have a dramatic impact. One of the smaller-scale examples from the book that really stuck out was the attitude & approach of a surgeon in an operating room. When surgeons approach mistakes from the perspective of “okay, this happened, let’s focus on how we fix it” mistakes are reported to the surgeon quickly, the surgeon gets accurate information quickly, and can respond appropriately. Result: more mistakes are reported but the surgeon has fewer complications and better outcomes.

When surgeons approach mistakes by getting angry or assigning blame to the nurse who did X or the resident who did Y those surgeons have fewer (reported) mistakes but worse outcomes. Why? People don’t fess up because they fear the consequences. And when a mistake is identified, people don’t give accurate and complete information because their primary concern is KYA rather than fixing the problem at hand.

[EricE]

Would also help computer "science" as well :p

[phkahler]

>> I'd be very surprised if the airport's procedures came out of this unmodified, for example.

I'm curious. The airport is at about 5900ft and they were at 6400. If that's AGL for them that seems like a long and high approach. If not, then they were going to do a 3 mile straight in at 500 feet AGL? Either interpretation doesn't fit my (limited, student) experience.

[azalemeth]

KAPA is indeed 5900 MSL, and those heights are also all MSL.

Looking at the FAA's charts, https://aeronav.faa.gov/d-tpp/2104/05715R17L.PDF, this combination of heights and distances isn't totally crazy -- I state without proof (and wait to be corrected!) that the approach is modified because it is a relatively high altitude airport and there may be quite steeply changing terrain underneath. I've never flown there (not even on X-plane) and I'm a low-hours UK person.

I haven't done the trig and meterological lookup to work out what their AGL altitude was on that day, but at the very least it's not crazy-wrong from the published chart...

[sokoloff]

If the 1 button on autopilot mode is the red autopilot disconnect, “time to hand-fly” button, I agree with you.

This turn to final (with the unusual additional warning to “do not fly through final”) is a visual maneuver and I’d expect most every pilot to be hand-flying at that point. (My autopilot and navigator is capable to make that intercept, but it’s way more tedious and distracting to program it than to just fly it.)

[t0mas88]

In the airline world it would at the very least be encouraged and in many cases mandatory to have the underlying approach programmed for this anyway. Even more so if told not to fly through final you would have the localizer up and monitor it.

In a Cirrus with what is probably a GFC 700 with flight director capabilities I would expect any competent instrument rated pilot to have the FD on and approach mode armed (the 1 missing button I meant) exactly to avoid this mistake.

Great to hand fly, but in a capable airplane just plain stupid not to use all the tools. And even mandatory on the professional side of things in many cases.

[sokoloff]

NB: There is no instrument approach to 17R @ KAPA.

I agree I'd have an extended centerline up (it's up by default if I zoom the MFD in close enough in my lesser-equipped A36), but this is a fully visual maneuver.

Almost no one is going to define a user waypoint near the touchdown zone for 17R and pull an OBS line off that just so they can use the FD/AP to make an entirely routine turn to final.

[t0mas88]

That makes things a lot harder for the Cirrus pilot. Easy to get the lineup slightly wrong with for example some wind correction in from far away. Crazy that ATC had them do this at the same altitude as conflicting traffic with no underlying approach as a safety net. You can ask someone not to go through final, but that's very easy to miss judge from a couple of miles away.

[sokoloff]

Slightly harder? I'd agree with that. A lot harder? 20 hour student pilots make visual turns from base to final from 3 miles out every day. This is not Top Gun material.

[t0mas88]

I have an instructor rating, and in teaching GA I've seen this hundreds of times also from experienced pilots. It's all easy in theory until in practice there is some sun in your eyes, the nose is pointing away from the runway due to wind and something inside needs attention (even more likely in a Cirrus, lots of tools = lots of distractions). Takes all of 10 seconds to be completely through the centerline and if reallt unlucky into the path of another aircraft.

Can't really blame this on the Cirrus pilot alone. They probably made the mistake here, but they were not set up for succes (no navigation available as backup) and ATC had zero safety net (both at the same altitude). So my prediction is that the final report will include recommended changes and not just "pilot error shouldn't fly through final, any student pilot could do that"

[ddoolin]

If the sight picture is the usual one for the pilot, sure. I think if you throw a 20-hour pilot at an unfamiliar airport with a different pattern configuration than they're used to (right or left, direct to base, etc), your chances of overshooting can go up a lot. Throw in tight parallels and it's not great.

I train at an airport with parallels with a tight-enough runway separation gap to necessitate a 15-degree offset in both T/O & L on the GA runway. > 100-hour me overshot into the adjacent approach when being cleared direct to base for the "commercial runway" which I've probably only been given once before. Shameful, yes, and a learning experience, but I never overshoot on the adjacent (we also have different TPAs for the parallels, probably for this reason though).

[azalemeth]

My gut feeling is that a deconfliction policy will appear on the charts and in the airport's procedures. Just because something can be done correctly, if the momentary-cockup-consequence is potentially the death of a large number of people, it's a good idea to make sure that there's some sort of defence in depth.