N72EX (Kobe Bryant) Crash Reconstruction with new ATC Audio
 

Hi. This is my first post to PPRuNe. I created an amateur reconstruction of the crash of N72EX and it includes new (previously unheard) ATC audio that I obtained through a Freedom of Information Act request to the FAA. I am not a pilot, and I am hoping that some experienced pilots can take a look and give me feedback. Because I am a new member I do not have privileges to embed URLs yet, so if someone could repost with a live link I would really appreciate it. "https ://youtu.be/M_Dpm144KXo" . I am aware of two audio transcription typos where I misspelled VFR.

That’s a pretty good effort on your part stringing what you could together.

Yes, very good job, perhaps you should be working for the FAA:ok:

It certainly seems to confirm that disorientation was a key factor - if he entered cloud in a turn, that is likely to produce the leans very quickly and the uncorrected left bank (apart from one brief attempt at wings level) is a good indicator of that. I have seen a number of students enter IMC like that and perform a similar manoeuvre - a turn with the climb dropping off eventually turning into a descent with them getting more and more confused by the indications because they are not focused on the AI.

Just one small point - the graphics say VRF while the audio correctly says VFR.

Another minor point: I've never heard any aviation professional pronounce ADS-B as "addsbee". It is always spelled out: "A D S B".

Thanks for you comment. BTW, I mentioned the VFR typos in the original post I wish YouTube would let me fix it, but I decided to let it go once someone first pointed it out because I didn't want a "new" video with just that fix to show up in my subscriber feeds. I still wonder if I made the right decision :)

I have been saying "addsbee" so long in my brain it never occurred to me that I was mispronouncing it. Thanks for pointing that out!

Fantastic job airplanecrazy. The last two transmissions from 2EX I hadn't heard before, quite chilling.

I wonder if he had issues engaging the autopilot, any S-76 guys care to comment on how easy it is to engage PIT/ROL mode on the fly? It's almost as though he engaged the AP (or at least thought he did) and then got distracted in the cockpit by a dropped iPad, passenger interruption etc. I just can't fathom that an ATPL certificated pilot, albeit not current, sitting in an IFR certified helicopter, could not conduct a simple climb on instruments, in my opinion there had to be an external influence.

Thanks for the kind words. For anyone answering Sir HC's first question, I believe that the helicopter had an SPZ-7000 Flight Control System installed (at least it did when sold by the State of Illinois, and I find no record of it being removed in the Airworthiness documentation.).

I think you underestimate the potential for over-saturation when things suddenly go all white outside if you haven't been in those conditions and practiced IMC flying very recently. It's certainly a perishable skill. Here's a good example of an EC135 IFR-rated pilot who wanted to practice an ILS approach to keep proficiency (responsible decision in my opinion), however, unfortunately he was sans instructor and very quickly became over-saturated and lost it while on the way in.

Yes, that report is classic saturation trying to deal with real IMC as a single pilot.

What can make it worse is engaging an autopilot that has its own limitations and tends to meander about whatever datum was set - this is especially true of 3 axis APs that have to use cyclic pitch to do things like hold ALT or VSI - they are slow, lagged to reduce control inputs and don't deal with turbulence well.

Unless you have experienced it to the point you were scared, it is difficult to explain how powerful the spatial disorientation illusions can be, convincing you that you are in a turn to the right (for example) when you are actually close to wings level and causing you to push left bank because it 'feels right'.

Constant practice in actual IMC conditions is the only way to prepare you for IIMC - or perhaps try teaching students to fly IMC, that sorts you out when they screw it up:ok:

Please forgive my naïve question (as a non-pilot). Should the pilot have declared an emergency to ATC when he decided to "climb above the layers" at 17:44:34? I can't imagine that it would have significantly changed the outcome, but I am curious. Thanks.

I sure appreciate task saturation, having done my fair share of both two crew and SPIFR. The NTSB report you referenced looks like he forgot to activate the approach and follow the glide slope and then failed to properly sequence to GA mode when he realized that he was full scale deflection.

In my opinion, the two accidents are vastly different, with the Delaware one, the pilot having to fulfill a clearance, a busy cockpit with a FD not giving useful information, reading back clearances etc etc.

Kobe's accident, all the guy had to do was pull pitch and climb, no airspace to contend with, no clearance to get/maintain, no icing to deal with, he didn't even need to talk to SoCal. I think the decision to continue IMC in a VFR only helicopter would be weighing on him, but he sounded pretty calm on that radio call. As I first said, I just find it unfathomable that an ATPL with even the bare instrument rating couldn't maintain a heading and airspeed, I believe there had to be something else that added to his workload.

At the very least he should have told ATC he was no longer VMC and was climbing to regain visual reference.

Nope. Despite holding an ATPL and being a CFII, ending up in cloud for the first time, of itself, could easily have been enough for the whole show to turn to worms.

I'm not familiar with the SPZ-7000 installation in the S76, but in any event the pilot didn't need to engage any pitch mode to save the day. All it needed was to center the heading bug, push the HDG button on George, increase power and confirm climbing. Single cue mode, that's all. It takes five seconds to do that. George takes care of keeping the wings level, your left hand controls the rate of climb, your eyes confirm climbing wings level, and you tell ATC what happened.

The safest thing to do if committing to the climb in that type is to press the GA button and then figure out the next few steps (in this case recognizing that he needed ATCs help). If selecting one mode in a S76B, I’d made it IAS then add power. Pressing HDG and adding power will give an initial climb, but the PBA will quietly convert that to airspeed while you are not looking. Best fly 3 cue IMC and at night if you want to avoid nasty surprises.

I have the pilot's manual for the SPZ-7000 and I see this "The go-around mode may be engaged by pressing the button on the mode selector or by depressing the go-around switch on the collective. When engaged, this mode will cancel all other modes that are engaged. Upon engagement the collective will be commanded to obtain 75 knots or more while the roll axis will return the aircraft to a level roll attitude and will maintain the current aircraft magnetic heading. The pitch axis will be commanded to a positive rate-of-climb of 750 fpm " Does pressing the GA button activate the autopilot if the autopilot is currently off, or would the autopilot also have to be turned on to actually command the aircraft?

The problem comes if you are not used to using the AP upper modes and if you spend all your time VFR without needing them. What might take an experienced user 2 button presses to achieve a predictable outcome (ie wings level climb) can take far longer and if you press the wrong button or have the HDG bug in the wrong place you can easily end up disorientated.

I don't entirely agree. In an aircraft that hasn't been maintained to the IFR standards, probably for years, there is too much stuff that needs to be working properly for 3-cue to work as advertised. And if it doesn't work as advertised you end up with utter confusion. Trust me on that. The safest thing to do, and this is what we teach in the simulator, is to boil it down to the minimum which is HDG hold (single cue), increase power manually, check wings level and climbing, and then adjust pitch trim to get 70 knots-ish. If a pilot who is already confused about his situation is then loaded up with deciphering what 3-cue is or isn't doing it 'aint gonna be pretty. But in the off-chance everything does work, then yeah, pushing the GA button should save the day. But I wouldn't trust it in a critical situation in an aircraft only maintained to VFR.

And having refreshed my memory, this is what GA mode does in the SPZ-7000 ...

https://cimg6.ibsrv.net/gimg/pprune....8dbbdfbd96.png

Gullibell - I presume there is a minimum speed for engagement?

How accurate is the airspeed readout in the video?
If he was coupled up in the climb and allowed the airspeed to bleed below 60kias, the autopilot heading reference would switch from roll to yaw. Could explain the sudden left turn and disorientation.

Yep, for SPZ-7000 it's 60 KIAS.

I don't believe that theory. If HDG mode was engaged the bank angle won't exceed 20 degrees by system design. I think he had quite a bit more than 20 degrees left bank towards the end. I would be very surprised if the pilot had made any use of the autopilot. Regrettably.

The speeds shown in the video are what I would call kinematic speeds (combination of ground speed and vertical velocity from GPS/ADS-B) The speeds shown do not reflect any wind (which I assumed were calm the entire time based upon the METAR at VNY). They also do not reflect effects of air density on indicated air speed. Given those caveats, I think the speeds are pretty close to both TAS and IAS. As a sanity check, the lowest ground speed reported in the flightradar24 ADS-B data for the duration of the video is 109 knots (I would give a link to the data, but I don't yet have that privilege "www.flightradar24.com/blog/wp-content/uploads/2020/01/N72EX-Granular-Data.csv"). I think the aircraft was well above 60 knots the entire time.

I also note from that ADS-B data the altitude, ground speed, rate of climb, bearing, or turn rate are never constant for the duration of the video. Would one of those be mostly constant, for at least a little while, if the autopilot was engaged?

ATP and CFII aside, from the information I’ve seen this pilot flew VFR in a predominately VMC environment. It appears recurrent training/checking was accomplished in the actual aircraft using a view-limiting device and not in a simulator. I suspect he maintained basic instrument currency by performing the minimum requirements every 6 months during his 135 checks. Given these supposed circumstances, an IIMC recovery would be at best, an accident waiting to happen.

Also, consider how he achieved the initial rate of climb.

The S76 is always flown with both AP's on. The AP's can be in stability augmentation mode or attitude retention mode. For the flight director to control the aircraft the AP's must be in attitude retention mode with the force trim ON. The S76 should be flown in attitude retention mode, and I would be very surprised if this aircraft was being flown in stability augmentation mode. Therefore, one button press on the flight director would have given safe control of the aircraft to George (assuming FD2 was the active flight director).

Having SPIFR experience in the 135, and many a hour in the 135 simulator (flying and observing), I surmise the pilot lost control of the aircraft during his reconfiguration for the missed approach.

First, it appears the controller violated ATC procedures by vectoring the aircraft in violation of the recommended distance/angle from the approach gate. This is what precipitated the accident. At his point, I guess the pilot failed to arm the GS, or armed the GS after passing through it. This would explain why the aircraft never descended.

Second, when the pilot realized the AP didn’t capture the GS, he requested a missed approach. However, I believe he may have been in a slightly confused state of mind (unusual position and circumstances). He was then instructed to perform the published missed, which suddenly increased his workload. I would guess he hadn’t familiarized himself with the missed approach instructions, and therefore needed to quickly find/read them. This also helps to explain the unwarranted climb at this time. Most pilots would (eventually) place the FMS into NMS/GPS mode in order to fly this missed procedure, which requires a reconfiguration. I suspect the pilot may have inadvertently disconnected the AP at this point. Disconnecting is an easy mistake, especially if one doesn’t perform the reconfiguration in an orderly fashion. With the sudden, unexpected AP decoupling, and a slightly confused state, I expect the aircraft quickly lost airspeed, stability and the pilot lost control.

An ATC procedural error combined with pilot task saturation and FMS/AP misconfiguration followed by the pilot’s inability to perform an unusual attitude recovery resulted in this crash.

As always, I could be completely wrong.

Yes you are. The pilot was not on a PROC so there was no GS to capture. He was not being vectored by ATC, he was flying VFR OCTA following a major highway.

Nope, the aircraft was at high speed immediately prior to the accident sequence. The pilot was in control, he just lost situational awareness.

he was talking about the EC135 accident report quoted earlier

I wonder what position the NTSB found the force trim release switch?

Take your finger off the trigger, you're scaring everybody.

I have started a g-force analysis of the flight using a program called "Tacview" and what I currently see in my model is that from 17:45:12 (after the sharp turn is established) to 17:45:22 (about when the pilot finishes saying he intends to climb to 4000), the combination of turn and acceleration downward appear to give about 0.9 - 1.1g down in the seat with very little lateral (side to side) or longitudinal (backward and forwards) g's. This is very preliminary, but I would like like to know if that even sounds possible to you all? I can reproduce a similar effect flying the X-Plane S76 simulator, but 1) It is a not a full fidelity simulator, 2) It doesn't really tell me how it would feel in the cockpit; and 3) I don't know if a pilot would notice himself/herself making the control inputs necessary to preserve a 1G in the seat as they are descending/turning (assuming the pilot is not getting attitude information from the AI and is flying by feel). I still have a lot of work to do to be confident in these preliminary results because I am still working on my pitch/roll model, working on my filtering software to deal with GPS position gaussian error, and working on my algorithm to deal with the 25 foot quantization of altitude in the ADS-B data, but it looked interesting...

A fundamental concept of successful IMC flying is understanding the fact that you cannot fly by feel. If there's no AH, then flying off the remaining instruments is possible, however if a pilot's just looking out the window trying to regain references (or ignoring the instruments in some other way), they won't be flying for much longer.

What would you like to know whether it is possible?

To fly wth 0.9g to 1.1g, and little lateral or longitudinal acceleration? I would say that is the aim. The lateral and logitudinal g's normal only come in at the moment of contacting terra firma (i.e., hard landing or crash).

I see that I did not phrase my question well, so I will try again :) The hypothesis I am exploring is that the pilot's attention could have been diverted away from the attitude indicator from 17:45:12 - 22 and that he "subconsciously" applied control inputs while distracted to maintain a 1g trajectory for those 10 seconds, even though the helicopter was banking and descending at the end of that period. My question is does the hypothesis seems possible to actual pilots? I have no experience to judge whether it would have required more than "subconscious" concentration from the pilot to achieve that trajectory, or to know if the pilot would have noticed making those control inputs (e.g. it would require the cyclic to be in an unexpected position).

Edit:- Ha! Our posts crossed:-)

Airplanecrazy, who did the AMAZING video reconstruction says:-

Since no one competent appears to have answered I will give it my best!?

I have done zero instrument flying, some other flying and I even had a helicopter lesson:-)

Your statements above seem reasonable to me. As the g tended to increase due to the turn the pilot could well have compensated automatically with a bit of forward cyclic.

In fact, thinking further, without some aft cyclic the aircraft may well have maintained 1g all by itself. To get more than 1g, up collective (+power) would be required and also aft cyclic to maintain the flight path as a level turn.

Perhaps all that was required was a little bit of roll input or perhaps the helicopter's natural instability would have been sufficient?

Maybe someone who understands this properly can respond to airplanecrazy? I have no knowledge of any autopilot or stability systems.

The reconstruction is very powerful and the scenario presented suggests that the pilot became distracted or maybe had a medical issue.

In addition to the amateur ADS-B collections there are I believe official ones. They may add to the available data and are likely to have proper timestamps. There are new satellite ADS-B data gathering systems too. I believe at least one is in pre-production testing and that they provided the data that persuaded Boeing to ground the 737 MAX. Boeing seemed to feel it safe (for them) to discard as irrelevant the amateur-collected material that showed a flight path eerily similar to the LyonAir crash.

https://www.flightradar24.com/blog/s...radar24-users/

In straight and level constant speed flight you experience 1G, just as if you were standing on the ground not moving - all forces are in balance.

If you roll into and maintain a steady state turn while maintaining height you will experience an increase in steady state G - 1.4 G at 45 deg AoB and 2 G at 60 deg AoB because the you are accelerating towards the centre of the turn.

As Apolloheli pointed out, you can't fly IMC by feel because of the lack of the visual cues your brain needs to orientate itself and the fact that acceleration in all axes affects the vestibular organs of your inner ear causing powerful sensations of pitching, rolling or yawing.

Whilst it is possible to fly instruments without an AH/AI it takes practice and is far more likely to end in tears than when using an AH/AI.

In the thread accident, all we know is that what he wanted to do with the helicopter isn't what the helicopter was actually doing - typical symptoms of disorientation.

If you end up IMC when you weren't expecting to and don't lock straight into an instrument scan, things are going to go wrong quite quickly - if you have the aircraft trimmed straight and level and don't move the controls apart from raising the lever a little to initiate the climb then you might survive to VMC on top.

If you go into cloud already in a turn you are halfway to the crash if you don't lock onto instruments and if this happens while you are moving your head around looking for external references you have a bullet in the chamber and your finger on the trigger.

For those that haven't experienced it, I cannot explain how powerful the illusions of bank can be - the leans - and how easy it is to end up in a spiral descent. If you have the height and awareness to recognise the situation you can recover it with good instrument flying skills and I have had to take control from disorientated students in this configuration many times in cloud. If you don't have the height and awareness you are doomed.

Thanks for the kind words!

I sent a FOIA request to the FAA for the ADS-B data they have. In discussions with the FOIA officers, I think there is a good chance I will get it. :)

i agree - like a lot of things it’s hard to understand without experiencing it. I’ve had full blown leans three times - once in IMC and twice in VMC. The IMC encounter was after a 270 degree turn from ATC vectors, and when I rolled level it felt like I was banking the opposite direction. It required complete fixation on my scan until the felling began to dissipate. The VMC encounters were once in clear visibility above a smooth full cloud layer that had a slope to it, giving a false horizon (I assume a temperature gradient?). Secondly in hazy conditions with no visible horizon, approaching the coast, that was straight but not perpendicular to our approach, giving a false horizon. I had to revert to instruments to overcome the sensation of banking which was almost overwhelming in both cases.