6 min read

Our blog on the Asiana crash got a lot of comments, 99 at this writing. There were a lot of different opinions on why the crew was unable to fly a successful visual approach on a clear day.

The discussion is interesting, educational and civil. Reading it gives a peek into the minds of a lot of different pilots.

UPS airline crash

Can we prevent crashes like this with more technology or more training?

On what you might call the far left are the folks who say that more technology is the answer and that the greatest potential for safer flying would come from eliminating human pilots and letting high-tech do everything. On what you might call the far right are those who think you deserve to bust your butt if you can’t fly by the seat of your pants. A lot of comments are between those two extremes and there are many constructive thoughts.

One thing that always amazes but never surprises me is how some people want to blame an accident like that on the airport. Yes, the approach was over water, and, yes the glideslope was out of service, but neither of those facts were a secret and the nature of piloting is such that things like that have to be dealt with. If you want to say an approach is dangerous because it is over water, what do you say about the approach at Telluride, Colorado? Airports are what they are, there for the taking or leaving.

To show that things do happen in threes, there have been two more high-profile accidents on visual approaches since the Asiana crash. These accidents are equally thought-provoking and offer more lessons to learn.

The NTSB probable cause on any of these accidents won’t be out for a year or more. Much of what is included in this discussion comes from sources other than the NTSB. The object is not to judge but to try to learn something while the events are fresh in mind.

A turboprop Twin Commander was cleared for an ILS approach to Runway 2 at New Haven, CT. The ceiling was about 800 feet; one report said there was some rain, and the visibility was said to be five miles. The wind was out of the south at 12 knots with higher gusts.

It was widely reported that the pilot requested a straight-in landing on Runway 2 but that the controller told him to enter on a downwind for runway 20, which does not have an instrument approach.  The NTSB did not include this in the preliminary report.

The ceiling was right at minimums for the circling approach.  That constitutes a demanding time for a single pilot flying a complex and relatively heavy turboprop twin. The demands could have consumed the pilot’s concentration because the airplane apparently departed controlled flight and crashed into a house while circling. It had every appearance of a low-speed loss of control.

Back up to the ILS approach and the pilot’s reported request for a straight in landing. It would have meant a downwind landing which is also a demanding maneuver. The amount of downwind would have been in excess of the ten knots that is considered for most Part 23 airplanes. Airliners under Part 25 are sometimes certified for more than ten knots downwind but some airlines set the company maximum at less than the certified value.

New Haven approach

Would a straight in ILS to runway 2, with a downwind, have been safer than attempting to circle?

I think a request for a downwind landing instead of a circle at minimums would be an appropriate effort to minimize risk in this case. The risk on the downwind landing might be a trip off the end of the runway though the runway was just over 5,000 feet long and that should have worked for an airplane with strong brakes and robust reverse thrust.

You have to wonder if the controller would have allowed a straight in landing if the pilot had been more aggressive in seeking approval. The pilot could have invoked the “E” word and done whatever he wanted but that hardly seems appropriate.

Two people were lost in the Commander and two lives were lost on the ground. I would bet that before all is said and done the FAA will part with a big chunk of taxpayers’ money on this if indeed the pilot requested to land straight in,  the controller wouldn’t approve it, and there was no traffic reason for the disapproval.  If the pilot did make this request, I don’t know why the NTSB did not mention it in the preliminary but it wouldn’t be the first time the NTSB covered for the FAA.

The third accident was the UPS crash at Birmingham.

The crew was making an approach to Runway 18. The longer runway that is normally used was closed for maintenance. There is only a localizer (and GPS LNAV) approach available for Runway 18. That was the most expeditious approach for the direction from which they were coming. The wind was relatively light and at the investment of a little extra flying they could have flown the ILS to Runway 36.

There is a hill north of Runway 18. Looking at it on the chart doesn’t suggest a big hill out there but I have taken off on Runway 36 on a hot day and examined that hill closely while climbing out over it.

Everything was functioning properly and the airplane was being flown through the autopilot until one second before impact. The crew had reported the runway in sight so they were likely flying visually.

They got a sink rate warning a number of seconds before impact, but it is not clear whether it was a one-time or a continuous warning.

It appears that they descended into terrain, on a downslope, while flying wings-level at the correct airspeed.

The approach was over an area without a lot of ground lights and there was a well-lighted area (including the airport) up ahead. This could have created an illusion of greater than actual height though a PAPI was available and operational and should have given a visual clue that they were too low.

As I was thinking about this, I wondered if the folks who wrote the ground prox software considered a case where an airplane is descending over descending terrain. The accidents that led to the current ground prox systems all involved airplane flying into rising terrain and while I don’t know a whole lot about software development it could be possible that descending terrain was not considered.

There is a history of premature arrivals on night visual approaches so this accident goes onto a long list.

To me, these three accidents favor the arguments that we need more emphasis on basic flying skills. Two appear related to judging actual height above the ground when operating visually and flying a safe approach without electronic guidance and one to basic flying technique. One might have related to a controller refusing a pilot’s request for a different runway and two to crews failing to request a different runway, one with an electronic glideslope.

These are complex matters that we all need to think about. The folks on the far left who want complete automation are for something that is not likely to happen any time soon. That means they will have the luxury of not being proven wrong right away. The anti-tech folks should get over it. We are not going backward in this area and the ever-present challenge will be to make the most of the pilot/technology interface. Quite obviously that is not currently being done very well.

Richard Collins
21 replies
  1. Tom Yarsley
    Tom Yarsley says:


    With great amusement, I enjoyed your characterization of me as being on the far left! Generally, I’m seated to the right of Attila the Hun at dinner.

    I agree that complete automation is not likely to happen soon – in the sense that it soon will become the predominant method, with or without regulatory compulsion. But the reason for this is not a lack of adequate technology. It is human reluctance to cede yet another domain to machines – especially one that’s been imbued with so much hero mystique.

    These landing-phase accidents are a consequence of process variability. When the process is executed correctly, the outcome is good. When it’s executed badly, the outcome is something else. Automated machines are a proven way of mitigating process variability. That may offend many or even most pilots’ egos, but it demonstrably is true.

    To the extent that “left” means “radical,” I’m going to go off and enjoy the compliment!

  2. Jim Scott
    Jim Scott says:

    Re: the UPS incident. I may have missed it but I’ve seen no discussion as to whether the pilots were CONSCIOUS at the time of impact.

    Since they had reported the field in sight they must have been awake VERY shortly before the CFIT incident. While pilots have certainly fallen asleep while flying, I doubt there is any precedent for TWO pilots falling asleep while on final approach to their intended runway.

    That leaves some sort of sabatoge or accidental discharge of some substance that could have rendered the pilots unconscious. Of course, if there was any such HAZMAT on board, the records will show that.

    But if there was sabatoge, likely via an altitude triggering switch then autopsies of the pilots may or may not show any such substances since the half life of some pretty deadly chemicals is measured in hours.

    I for one just don’t buy it (not easily anyway) that two highly skilled pilots in visual conditions and on a final approach…when the adrenaline is flowing even with experienced pilots…just didn’t NOTICE that they were about to “land without benefit of a runway.”


  3. Vinton Land
    Vinton Land says:

    Mr. Collins,

    I have been reading your articles in Flying since 1969. I followed your struggles with 100LL valve troubles in your Skylane, vaccum failures, etc. I owe you a great deal of credit for my survival since earning my private license in 1973. You have taught me to read accident reports and always keep in the back of my mind that oversights and loss of control are deadly. The Birmingham accident puzzles me be because one of the caveats I have learned from you is the use of autopilots in IMC. I have partnered with others in a Cirrus, but now that I finally own my own 1973 Cherokee 180 without an autopilot or Garmin 430, I wonder how you feel about pilots like me with Instrument ratings flying without an autopilot. Thank you

    • Dick Collins
      Dick Collins says:

      I flew without an autopilot for years. As I got older and flew more in complex airspace, though, I came to like the things. I did make do with a simple autopilot (no pitch) until the end and found that to be very adequate for me. I think you will be fine with no autopilot right up to the moment when you want one badly.

      • Stephen Phoenix
        Stephen Phoenix says:

        Of course the same could be said of de-ice equipment, radar, turbine engines, a co-pilot and so on. Bottom line, one should strive to take on flights that are within the reasonable capability of the equipment at hand and the pilot. Thinking that through before launching will reduce the stress level (of the pilot and maybe the airplane).

      • Vinton Land
        Vinton Land says:


        I flew my new (old) 1973 Cherokee Challenger home yesterday without an autopilot and it was very stable with 4 hours of flight time. It is good to know you are still there to guide us amateurs! I might get fancy now and install a STec 30 and G430, but that is later. Good, basic, old time fliying was fun! Thanks!


    • Duane
      Duane says:


      I also own and fly a Cherokee 180. It had the original Piper Century II wing leveler in it when I bought it, but the old unit was non-functional. I eventually had the Century II overhauled and it only worked a few months more then crapped out again. I flew hundreds of hours VFR cross country in my Cherokee without benefit of an autopilot, but there were times when I worried that if I inadvertantly wandered into IMC, or if I was even just partially incapacitated, the ending would not be pretty.

      So I then spent the bucks to install a new STEC single axis autopilot, and my comfort level increased a great deal thereafter on cross-country flights. I also installed a Garmin 400W GPS to link up with the STEC to do GPS approaches in IMC. It’s not cheap to put anything new in your panel, especially when the airframe isn’t worth all that much … but if one flies a lot of cross-country and/or carries non-pilot passengers, one is bound sooner or later to need the backup available with a functioning autopilot.

      Think of it as your trusty co-pilot who never gets disoriented and always keeps the sunny side up.

      • Vinton Land
        Vinton Land says:


        Thank you for the note.. I want to go in the same direction as you sometime in the coming months. I was wondering if you might share what avionics shop you used and were you satisfied? Thanks again.


  4. Jim S
    Jim S says:

    Thre ya go….you won’t need it until you need it. (-:

    For those of us who carry passengers, a simple wing leveler that Mr. Collins used could very well be a life saver.

    In IMC with passengers, it seems to me that without AT LEAST a wing leveler, we are imposing an automatic death sentence on our pax if something should happen to incapacitate us.

    I for one, would not carry non-pilot passengers on any flight for any reason without at least a wing leveler…and aboslutely CERTAINLY not in IMC or the reasonable chance of IMC developing along the route of flight.

    We have all read and even watched those wonderful cases where a non-pilot has managed to pull off a survivable landing when the pilot became incapacitated. But we can’t know how many pax perished with their pilots simply because they had no clue how to even send a radio message…let alone keep the airplane right side up for long enough to do so. And in IMC…such pax, to put it in the vernacular…would be screwed even IF they could use the radio and even IF they had VFR flight training. (I believe I’ve read that the average survival time of VFR pilots entering IMC is about a minute and a half).



  5. Rod Smith
    Rod Smith says:

    Dick- you are 100% right on and correct as usual.I agree on the 690 crash,and landing downwind(even wet runway)would be no problem in that aircraft(all systems normal of course)it was the best choice for him.In the UPS accident the weather was good vfr but went down just before their arrival for a couple of hours,if I remember it was like 600 and five in light rain,then loc only and who knows the actual wx north of the airport pushed them to ease down just a little more after mda,maybe with runway in sight but not papi(this approach not authorized @night without papi,and I not saying papi was inop just not in sight thru rain/clouds)and like you I got a good look at that hill taking off north in a Navajo many years ago on a hot summer day.As for the 777 crash I wont waste my time on that one.

  6. Eric-Jan Oud
    Eric-Jan Oud says:

    Approaching low over a hill top has 1 often overlooked side effect that may “confuse” pilots and/or automated systems alike: the venturi effect of that hill.
    Wind velocity will be highest at the hill top. Compensating for that, slowing back down to approach speed, just might drop the airspeed below critical values after passing the hill top. The lower relative airspeed after passin gthe hill top will increase descend rate. Especially if the “loss” of airspeed is compensated by dropping the nose, that will initially result in an even higher descend rate, before actually picking up speed …
    That might explain the downhill CFIT of the UPS crash @ Birmingham (not having looked at any other documentation on that crash, I must admit; just a brainwave)

  7. Cary Alburn
    Cary Alburn says:

    I guess I lean pretty hard toward the “basic flying skills” end of the argument, although I’m not against reasonable help from technology. For instance, I had an Alpha Systems AOA gauge installed in my 63 P172D a few years ago, although I have way more than enough hours (2400+) and years (40+) to be able to “feel” impending stalls–and I love the thing. I now have in-flight weather via my very new iPad Mini, Foreflight Pro, and Stratus II, and I am elated to see how well that combination works.

    But I have a hard time understanding why so-called experienced pilots can’t land an airplane, even a big one, in VFR conditions and instead rely on the doo-dads to do it for them. Or don’t they get any practice, even in sims, using the yoke/stick and rudder instead of flipping switches and twisting knobs?

    Regarding the Commander accident, I’m at a loss to understand why the pilot didn’t simply tell the controller that he was going to do the straight in–he was the PIC, not the controller, and as such, he should have been more assertive. I’m sure there were more, but a couple of incidents in my past come to mind, where PIC assertiveness was appropriate.

    *One was IFR to Jackson, WY, in a Cutlass RG about 15 years ago when I started picking up light ice near the Dunoir VOR. Any ice in a Cutlass is too much, so I immediately advised Salt Lake Center and was told I was #4 for the approach. I knew what types of aircraft were ahead of me, all FIKI-protected twins, so I replied, “No sir, I’m #1–I’m picking up ice, and I need to be cleared for the approach immediately.” The other 3 were held, and I was immediately cleared for the approach. If I hadn’t been, I would have resorted to the “E” word.

    *The other was IFR into Grand Island, NE, in my P172D about 5 years ago, before I had the 430W installed. This one wasn’t quite as critical as the Jackson event. I was cleared for the LOC/DME BC 17 approach, but without DME, I couldn’t take it. I told the controller that I wanted the ILS 35, that I had no DME. He said, “The winds are 20 knots from the south–I’ll have to give you the VOR 17 approach.” I replied, “No sir. Clear me for the ILS–I’ll deal with the winds when I break out.” He did, I flew the ILS, and broke out plenty early to make a nice circling approach to 17. Sure, I could have shot the VOR 17 approach, but the ILS was a whole lot easier–and I was the PIC, not the controller.

    Anyhow, I see the major issue of all 3 of these accidents as the pilots all abdicating their PIC authority from themselves to either technology or the controllers. Fundamentally, as pilots we need to fly the airplanes–that’s why PIC means Pilot In Command.

  8. Daniel
    Daniel says:

    The UPS pilots did not have an ILS available that night. The only ILSs available at BHM are on 6 and 24, which was closed.

    The only approach available to 36 was a non-precision RNAV. While the current NOTAM was issued after the accident, that approach hasn’t been allowed at night for a long time.

    • Jim S
      Jim S says:

      I still find it VERY difficult to believe the 2 supposedly conscious and at least reasonably sober UPS pilots could drive it into the ground on any kind of night…let alone a VFR flight.

      From what I’ve read, there was ZERO pilot communications with ATC..re: emergency conditions or anything else…while the pilots just proceeded on their merry way to a CFIT accident.

      The most LOGICAL explanation is that they were not…in fact…conscious.

      However, a full analysis of the CVR transcript would weigh heavily on that speculation. BUT…the most recent NTSB comments that I have found refer ONLY to the fact that the CVR was operational because it recorded the ground proximity warning sounds including the PULL UP voice.

      NO mention (that I have found)refers to any PILOT voices having been recorded. There is ZERO chance that 2 conscious pilots would say NOTHING in the last minutes of the flight. The “sterile cockpit” rules do NOT mean “silent cockpit” and all manner of rountine dialog between the Captain and FO takes place during those last minutes.

      I’m NOT a conspiracy theorist and generally laugh off that sort of thing. But LOGIC is dear to me and there is nothing logical about the scenario (including mute pilots on final approach) as reported so far.



  9. Roger Rowlett
    Roger Rowlett says:

    Two comments:

    It is very easy to have a CFIT incident at night with hills off the end of the runway if you are not careful:

    1. If transitioned to a visual approach, it is difficult to see dark hills between you and the runway. Usually the first clue is the disappearance of the VASI or PAPI. In a large plane, the cockpit could be high enough that the bottom of the plane won’t clear the terrain, even though the pilots can see the visaids (although they should be telling you you are low.)

    2. If following a GPS VNAV profile on a non-precision approach, which my GNS430W will happily do right to the runway, there is no guarantee of terrain clearance if you descend below the MDA. It just plots a straight line from the FAF to the runway. A few pilots have fallen into that trap. You can experience this first hand at Sidney, NY, on the GPS 25 approach and pilots have complained about it, not knowing how their GPS units actually work. If you follow the horizontal needle below the MDA you will fly into the hill.

    UPS planes and pilots probably don’t have problem #2, but #1 is possible.

    • Jim S
      Jim S says:

      Good points…but the NTSB reports re: the CVR state that the GPW did, in fact, function. It commanded pull up but there was no pull up. I haven’t read any comments re: any FDR data except that it detected no pre-impact engine failure or fire so this really was a CFIT incident. But it remains unclear whether there was any ATTEMPT to pull up.

      I seem to recall that significant engine spool ups can be heard on the CVR but I’m not sure of that and haven’t read any comments along those lines.

      Finally, re: intervening hills…certainly they have caused pilots to come to grief in non-precision approaches in IMC and/or at night. But if you review the post-accident photos, you will see that the front third of the fuselage is almost completely in tact.

      That is not a signature for a CFIT into the face of hilly terrain. The pilots apparently perished from fire/smoke not impact.

      In due time, the FDR data will become available as will “enhanced” analysis of the CVR. If neither of those devices prove an attempt to pull up was made in spite of the known GPW warnings, then the nearly inescapable conclusion would be that both pilots were not conscious at the time of the incident…with the only remaining question being “Why?”

      1. Passed out in drunken stupors? Possible but extremely unlikely.

      2. They BOTH fell asleep from fatigue on the final approach course? Ditto the above.

      3. ??????????????



  10. Tilly M
    Tilly M says:

    No one who writes on this accident points out that the longer runway was only closed until 6:00am daily, until about 10 minutes AFTER the crash. No one has pointed out that if the crew had requested a 10 min. hold, they could have had the longer runway. Having flown this approach many times, I can tell you there is not many lights at all short of the Guard ramp and the airport on this runway. Your rendition of this accident is very disappointing. Was the crew even qualified for this approach? Possibly would explain their silence. The airport is in a valley PERIOD. The rise they hit has already been lowered once due to another accident but the airport will always be in a valley. The Notams are so long for this airport and construction you need extra fuel to listen to all. So, there it is, choose the longer runway when flying a heavy.

    • Jim S
      Jim S says:

      @Tilly M….I don’t know whose “rendition of the accident was disappointing.” But first, I don’t believe that there are 2 UPS pilots in their system who are not “qualified” for that approach…or any other.

      They may not have shot the particular approach but that doesn’t mean they aren’t “qualified” to do so.

      As for intervening rising terrain, I think the laws of physics prohibit running into rising terrain as long as the pilots can see the approach guidance lights…and there is no evidence of which I am aware that they did not.

      Of course, utter incompetence or negligence can be factors explaining why they saw the lights disappear and did nothing about it. But there are other equally plausible explanations as well…which is what I was probing. Ruling out sabotoge in this day and age would not be an optimal investigative technique in my view.

      Finally, I don’t buy the thesis that a 10 minute hold would have provided a longer runway and avoided the accident. The pilots obviosly thought that the available runway was long enough and I know of no SOPs which require the use of the longest available runway even if holding to get access to it might be required.



  11. Paul Miller
    Paul Miller says:

    Re: Roger R’s #2;
    I soloed in a Piper Vagabond at Sidney,NY in 1957 off the old cinder strip.Instructor(still there) Mr. Ford’s words about the hills surrounding were stay well above them!Never had a problem.Later a private rating in a Pacer at Cooperstown-Wellsville with hills again and it was innate to stay above them. Now as a CAP SAR pilot with G1000 in a C182 we deal with Mt LeCompte near TYS,but down where we work it’s still “Stay above the hills”. Old lessons die hard-it’s time to get back to basics and Fly The Plane-with assistance from the added gadgetry-but with all eyes outside.
    Your words have always been Gospel,too,Dick.

Comments are closed.