Thanks, yes, the NTSB told me all the CAB files were turned over to the National Archives. The National Archives said the only CAB files they got on this incident were the final reports.

I've uploaded a new version to http://we.tl/TvTrucG3Ak

This incorporates the material I found at the Chicago Tribune. It has a couple graphics to discuss the trajectory at the end of the flight.

Based on input here and from some emails I think my understanding of the situation has improved, and I'm characterizing the crash not as a failed attempt to make a belly landing, but more as an attempt to manage a descent that they couldn't stop.

This draft also has an explanation of the mechanism that can explain the leveling of the bank. The CAB discounted this, because they thought the bank was always increasing. Once the calculations showed that it couldn't always increase, suddenly this return-to-neutral makes much more sense.

Interesting.

I see you have incorporated some of the suggestions from the forum.

The only point of contention is that you seem to equate bank angle with a given rate of turn. While this in broadly true, it only applies in a balanced turn with ailerons and rudder acting together.

Given that there was no effective aileron control, the turn became progressively more unbalanced as the roll increased. So the aircraft was effectively skidding sideways through the air (in a horizontal axis).

That kind of makes a nonsense of any attempt to calculate a turn rate based on the angle of bank of the aircraft. That is to say, an aircraft in banked or even knife edge flight, may have no turning moment whatsoever. Or alternatively there may be a considerable initial rate of turn that then progressively reduces as bank angle increases, due to the ailerons and rudder becoming increasingly out of balance.

That would account for the impact site being where it was, despite calculations suggesting that this would not have been possible with progressively increasing bank angle and rate or turn. The rate of turn didn't increase linearly with increasing bank angle. In fact it may well have reduced somewhat.

Of course. That's what I came here for. I've also been in email discussions with pilots.

Yes, I understand the turn and bank calculations can be off, and I don't yet know how much off. I need to play with a simulator. So for the time being this is an approximation.

New version posted at https://we.tl/beXUsgYPvX

Two significant additions to this release.

The first is discussion of an Argentine Navy Electra crash in '89. Video is available at The video shows what an Electra actually does in a high bank right wing impact, compared to what the CAB said of Flight 706. I use a bunch of frame grabs to show how this contrasts with Flight 706 regarding trajectory, crash dynamics, and survivability. Very useful lessons.

The second is a series of posed images to show what N137US would have looked like in different bank angles. My most encouraging supporter (a retired NWA pilot) gave me a big metal Electra model, and I posed it on a stand to replicate perspectives. The most important is the view from the control tower, presented as a storyboard in correct scale.

The Argentinian accident invites a lot of interesting comparisons with the original crash. The scale representations of the aircraft as it would appear in flight are a great idea. It is possible to appreciate how difficult it is to estimate bank angle from any significant distance and without a clearly defined reference. Well done.

I am still uneasy at your criticism of the original accident reports. The investigators at the time identified a disconnected aileron cable as the component commanding full right wing down. The aircraft was effectively uncontrollable from this point onwards, so no matter what action the flight crew performed, the crash was inevitable.

It is possible to run thousands of different scenarios through a modern computer and come up with several possible solutions to explain how the aircraft ended up at the final crash position and with the wreckage arranged as found. At the time of the accident, such tools simply didn't exist and imvestigators were forced to rely on experience and rules of thumb to arrive at a conclusion.

So while your scenario may indeed be more accurate in terms of describing the flight path and sequence of events during impact, it doesn't actually serve to improve flight safety, which is the primary purpose of any accident investgation.

I continue to have no doubts whatsoever that the flight crew on that day did everything they could to save the aircraft, but the physical evidence, photographs and contemporary reports do not indicate a nearly successful attempt at belly landing.

That is my interpretation of the evidence, yours is different and backed up by computer modelling and comparison to other incidents. I am happy to accept that what you propose is a possibility but not the most likely probability as to the exact sequence of events.

You have collected a mass of data from various sources and you are now probably the best informed person on the planet with regards to this accident. You are entitled to express your opinion as an expert after all that research and to be taken seriously. The experts who investigated the accident at the time were also doing their best but with the added constraint of needing to find the cause as soon as possible and without the benefit of modern technology. The truth probably lies between both sets of conclusions.

I shall be studying your latest report in more depth over the coming weeks.

I'm impressed that they all got out alive from that crash. What is clear is that there was a lot of disturbance of the wreckage during the search and freeing of casualties. The same is probably true for most crashes where there might be the possibility of survivors. That makes any reconstruction of the incident more difficult for investigators.

Where I saw a possible similarity is in initial contact by a wing with the ground. Also the increasing angle as the aircraft climbed out from the runway and control was lost.

While it might not be possible to make direct comparisons between the two crashes, the video gives an indication of the sorts of time scales involved and what a witness to the original crash might have seen.

Other than that, I agree with you megan.

G0ULI, you're correct that my fault-finding has nothing to do with the cause of the crash. Megan had made the same point earlier. But my whole subject is the impact sequence, and I explicitly say that I don't question the cause of the crash. With regard to that part of the investigation I have nothing to criticize.

My subject is the impact sequence and plane attitude, which was investigated so badly as to qualify for my term "slapdash". If I seem harsh, it's because I lived through that and watched others live through it as information trickled in. There were many people who cared very much what happened. After over 50 years we can see that the news we waited for was misreported badly. It makes me indignant, I confess.

I've adjusted my phrasing and my understanding, based on quite a few conversations with pilots. In the document I now characterize it as not so much an attempt to make a belly landing as an attempt to manage a descent. With more altitude they might have had much better results. They were headed for the ground and were hoping not to hit too hard, rather than having the goal of putting it on the ground.

I'd be delighted if someone had the computational tools to re-enact the possibilities and see how it plays out. I don't have the equipment for it. As I say in the document, I can show that the CAB/ALPA reports are wrong, but I can't prove that I'm right. I'm confident of being pretty close, but that's about it.

A couple pilots have discussed the flight with me. They agree that this maneuver would never be tried at all (at a safer height) and there will be no data on it or first person reports. They suggest I try a flight simulator and see what results I get, and I've started on that project.

I presented the Argentine crash not because it was like the Chicago crash, but because it's like the CAB said the Chicago impact started but the Argentine crash played out differently from the first impact. No two crashes are going to be alike, but finding this one gives us the chance to draw a couple lessons from it.

I agree that the crew did all they could, and we knew that since '61. I get a lot of drive-by psychoanalysis out of this. People think I'm trying to salvage reputations that don't need salvaging, or trying to deal with trauma that I was over at least 30 years ago.

Being the best-informed person on the planet on this particular crash is setting a fairly low bar, considering that everyone who studied it in person is probably dead. I'd rather be known for my book The Passionate Ape. :)

Hi Megan, glad you're still here. I have valued your input on this thread. Thanks for the video link.

As I said, the Argentine crash is not a match for the Chicago crash. The similarity starts and ends with the right wing impact.

But three things can be taken away from the Argentine crash: The trajectory of a plane with no lift, the rotational effect on an Electra when the right wing hits, and the survivability of a moderately low-angle Electra impact. These are all highly relevant to understanding the dynamics of the Chicago crash and how close they came to a survivable incident.

ETA: Your video has much better quality. The clip I found was passed through some steadying software and was electronically zoomed, and lost a lot of detail. I'll be able to redo my graphics with better quality next release.

Oh yeah, that's a major reason why the Argentine crash is so interesting. N137US ended up tail-first and upside-down, and the Argentine plane ended tail-first and right side up.

It seems highly likely that the CAB investigators had in mind something like the Argentine impact sequence when they viewed the wreckage. It sort of fits the description of cartwheel, in a sloppy way. And the tail section being upside-down apparently didn't faze them. But being upside-down is the kiss of death to their model, because it means N137US arrived at the final site with high energy and nose-first, and then flipped over.

I presented the Argentine crash for contrast, not to show the similarity. The Argentine plane crashed just how the CAB said N137US crashed, but the Argentine plane behaved in a completely different manner. That's why (as I said in my document) the Argentine crash refutes the CAB model. If N137US crashed as the CAB said, a backward slide would be normal. But being upside-down, N137US obviously didn't crash that way.

Do you see the implication?

ETA: One of my pilot friends read the CAB report and busted out laughing at the description. He said "so it hit the right wing, and then defied the laws of physics for 800 feet."

Several pilots have said that any information about the flight dynamics will have to come from a simulator. I've loaded X-Plane, which is the best one for my purposes, and am using a L188C model written by David Starling. The manual says the Electra at the conditions for Flight 706 should stall about 110 knots, and the model does. The maintenance manual says the rudder can deflect 30 degrees right and 25 left, and the model has it exactly.

The model appears to be an accurate representation, and we can probably take its behavior to be correct within a smallish percentage of error. I'd be interested in hearing from pilots about the feel of the model.

One of the questions is whether an Electra can knife-edge fly. Turns out it can at any speed over about 250 knots. At lower speeds there isn't enough airflow over the rudder to get the nose up. I've run a number of tests, all starting at 1k feet over Puget Sound, since Flight 706 got to 300 feet tops and ORD elevation is 680 feet. My tests start at 160, 200, 250 and 300 knots, and I throttle up as needed as soon as the bank goes vertical. At 160 knots the plane has no chance, and is essentially in free-fall for the first 500 feet. At a starting 200 knots after a few tries I could get to 250 knots about 50 to 150 feet over the water. Starting at 250 or faster the plane had no trouble at all.

There will be some discussion of this in the next version. The answer for now is that knife-edge flying was not possible for Flight 706 because it was about 100 knots too slow.

My next step is to write a program harness for X-Plane so I can run turn and bank trials and collect more rigorous numbers. The questions to address are (1) what bank can the rudder induce, or what unwanted bank can it counter, and (2) what do rudder inputs do to turn radius calculations.

Hand-flying suggests that the rudder alone can turn the plane completely over, which impressed me. I'll be interested in seeing how the data turns out.

Excellent idea to get a feel for how the aircraft handled and the effectiveness of the rudder. Even though sustained knife edge flight is not possible at speeds below 250 knots, the rate of descent is presumably reduced somewhat, even at the slower airspeeds? Inappropriate use of rudder can certainly turn an aircraft inverted and/or induce a spin.

I'm certain that there's at least some lift from the side of the fuselage, but I haven't measured the exact rate of descent to determine how much. I hope to do that when I develop some firmer numbers. My first goal is to program the harness for running the simulator, and that is unfortunately a little ways away due to competing demands for my time.

Other commitments have taken me away from this project for the last month. I don't yet have the program harness working. Programming is my field, and I understand the problem, but don't have a solution yet.

Meanwhile I've been trying to get some useful information out of the simulator. Bear in mind that the simulator results are all that's available, but they can only be considered an approximation. I covered the fact that knife-edge flying is not possible at 160 knots. Two other questions are answerable by hand-flying: What are the behavior characteristics of an Electra banking right under full left rudder, and what attitude would result in a five degree descent slope.

Several pilots have agreed that the rudder must have been hard left. This would be an instinctive response, it would have an immediate effect, and they would have maintained it until they regained level flight. So we have an airspeed of about 160 knots and hard left rudder in a right bank. The first issue is to determine what this does to the turn radius. All my illustrations are based on balanced turns, because no one could estimate what the rudder inputs would do to the radius.

We have evidence that the bank angle at impact was near 34 degrees, so I used that angle when collecting a representative radius value, but I'm not saying this radius calc is itself evidence for a 34 degree bank. A balanced turn at 160 knots and 34 degrees gives a turn radius of 3370 feet. A 35 degree bank at 160 knots and hard left rudder in the simulator gives a turn radius of 5275 feet. I'll be running more trials to get an average, but it's in this ballpark. The left-rudder turn radius is around 1.5 times that of a balanced 35 degree turn, give or take.

In my illustration on page 20 of the 160602 document, I said that the red-dotted path was the most extreme possible, but that was based on a balanced turn. That path would have resulted in an impact bank angle of about 22 degrees, which made it unrealistic. Given the new unbalanced turn radius estimate, that path becomes much more reasonable. The plane could have reached the 271 heading with considerably more bank than 22 degrees. So the initial emergency probably started with a higher bank farther southeast than I had thought.

The second question is: What attitude would result in a 5 degree descent slope. At 160 knots a 90 degree bank results in free-fall, as discussed earlier. When this project showed me that they weren't in a 90 degree bank as reported, I wondered why they didn't climb. The simulator says at a 34 degree bank and 160 knots with hard left rudder it's easy to climb. So I'm looking for a bank angle extreme enough that the plane has insufficient lift to stay up, but not yet stalling.

As the bank increases, the elevator angle has less and less effect vertically, and starts to serve only to tighten the turn. Meanwhile the rudder has more responsibility for keeping the nose up, but it's going too slow to do that. N137US weighed 93,000 lbs and would have stalled at about 63 degrees in a balanced turn at 160 knots. My numbers are rough, but it appears that the simulator is giving stall warnings just above 55 degrees. Pulling back on the yoke increases the stall likelihood by lifting the nose, while not pulling back keeps the plane descending.

The exact rate of descent is highly sensitive to airspeed and bank angle, but a descent of 1400 to 2000 feet per minute seems within reach. The 5 degree slope Lockheed estimated would be 1400 feet per minute. The simulator run suggests they were at the stall boundary and unable to stop descending. My latest report characterizes their effort as just trying to manage an unstoppable descent, but seeing the stall warning makes it clearer.

This view, rough as the numbers are, is the best evidence I've seen that my 34 degree impact bank angle estimate is too low. This gives about a 20 degree gap between the simulator's result of a maximum 55 degree bank in flight and physical evidence of a mid-30 bank at impact. I can see a couple possibilities. One is that the flight bank and impact bank are both somewhere in the middle, and computation errors in the simulator are showing a steeper bank than they actually could have had.

The other is that in the last two or three seconds the bank angle was changing, perhaps as the aileron boost unit was returning to neutral. If the second one is the case, then the turn radius was decreasing and the elevators would have been able to pull the nose up. In other words, it's possible the impact was actually a flare that they were unable to initiate until too late.

The idea that the bank was changing rapidly in the last couple seconds feels conceptually awkward, a bit too close to magical thinking. I'm inclined to think the first possibility is the better one.

The rate-of-descent data is very rough because I have a really tight window to get into position and collect data, and I'm no pilot. I wasn't able to collect turn radius data at this extreme bank. If I can get the harness working it will give better numbers.

Meanwhile, I'm always interested in feedback.

I've uploaded a proof-of-concept video to my weTransfer page. This is intended to show the kind of data I'll capture using the X-Plane simulator. This is a 15 mb MP4 video. The frame rate is low, but the goal here is to capture data, not to make a smooth video experience. https://we.tl/GU7wQaiDOZ

I envisioned three levels of testing with the simulator. The first was to test basic handling characteristics of an Electra. One of the objections raised to my analysis has been the proposal that Flight 706 not in a shallow bank but was knife-edge flying when it hit, and this explains the slow descent. Hand-flying the simulator shows that this is impossible because Flight 706 was going about 100 knots too slow for knife-edge flying. Hand-flying is sufficient for this proof, but that's about my limit as a pilot.

The second is to determine the characteristics of an Electra in a banked turn. Turn radius is critical to understanding how Flight 706 got to where it impacted on the airport grounds, and the angle of attack it could have had. So the next set of tests involves putting the simulated plane into turns of varying angles and different rudder deflections and measuring the turn radius of each. A pre-requisite was to write a simulator-controlling harness, which now is showing basic functionality. The first job in this test series is to calibrate the simulator and ensure that it accurately reflects what we could expect of an Electra. The link shows part of such a calibration step.

Non-Techies skip this part: The internal control is via a Lua plugin running inside Gizmo. Gizmo is a Lua/X-Plane interface plugin. This Lua script is a dumb plugin; it passes simulator data (eztracted by Gizmo) to the external program and the external program passes data values for the plugin to stuff into the sim through Gizmo. The plugin makes no decisions. The external program is a Delphi 10 (Pascal) program that has rudimentary flying abilities. Delphi 5 was my last installed version, and it's the best IDE I ever used, but it's unable to deal with W10. Embarcadero is giving away an intro-level Delphi 10 for free for the next four or five days, so it you ever want it this is a great opportunity. The intro level has plenty horsepower for this job.

The uploaded video shows the level of control needed to establish Electra flight characteristics. The harness can hold the sim plane within about a half a degree of bank angle, and within very few meters of height. There is a ringing effect, so the longer it runs the smoother it gets. With this harness I can put the plane into a stable state and then mess with rudder inputs and the like. The large readout shows the time it takes to make a circle, control surface deflections, headings, etc. It's not all wired up yet. The goal at this point is to calibrate the simulator and refine the Delphi flight controller so it smoothes more quickly. The video shows the Electra (in 1960's livery) circling over Elliot Bay and Bainbridge Island.

Not shown in the readout are the longitude and latitude values that determine exact coordinates. These are captured in a log file with other data, and I use those to compare to the speed and timing, which determines circumference, to check on the circle dimensions. There is a distortion in these measurements due to the way the simulator handles the 'flat earth problem', so I'll have to do the real measurement flights in Jakarta, not Seattle. Circles flown near Seattle turn out to be north-south-elongated ellipses, buit circles flown near the equator are circles.

The third set of tests will be to replicate possible flight paths for Flight 706. I won't be able to work on this until I have more data, a feel for how the sim handles, and a better Delphi flight controller. If the sim cannot be shown to be a good representation of Electra handling, then more work would be wasted. Even if the sim is only a marginal flight model, there should be some useful information about how rudder deflections alter the turn radius. If the sim turns out to have a dependable flight model, then there may be more useful data forthcoming. At this point I can't be sure the sim will generate anything useful beyond disproving the knife-edge flying conjecture.

All good stuff. Obviously you need to concentrate on one set of data at a time to establish the correct control responses. Don't forget to consider the effect of various engine power and propeller pitch settings once you have your basic data established. For initial analysis, assuming the engines were developing full power at lift off would be reasonable, but changes to engine power settings may have been made as the bank developed. This would change the airflow and therefore the effectiveness of the flight control surfaces on the wings and tail. If you don't get reasonable results from your initial simulations, this may prove to be the cause of the problem.

One thing that occurs to me while watching your videos is that the ailerons are functional throughout the simulation run. You can vary the aileron inputs to attempt to duplicate the actual flight path taken by the original accident aircraft, but does that actually represent the true aileron positions during the accident flight?

Once the aileron drive cable broke, the ailerons may have been held in a fixed position for a time by powered assistance until the pressure bled away or the power boost was turned off. Alternatively, the ailerons may have moved due to airflow over the wing and prop wash from the engines.

Because of this, it is difficult to conclude that the turning circle was continuous or at a constantly and steadily increasing bank angle. Clearly the aircraft banked to the right after take off and that bank increased to a point where the aircraft could no longer remain in the air. There is the possibility that the bank angle increased in a series of 'jumps' as the crew struggled to regain control and as the aileron deflection altered due to crew inputs such as turning off powered assistance and changing airflow patterns over the wing.

Rather than assuming a steadily and smoothly incresing bank angle, you might want to try increasing the bank in a series of steps that might correspond to attempts to regain directional control.

The initial cable break would require a left rudder deflection to try and correct the bank. The ailerons would progressively become more susceptible to aerodynamic forces, and propwash effects as engine power levels were adjusted and the bank angle increased.

It seems clear from your tests so far that this wasn't a nice neat curving climb and descent with a steadily increasing bank. You have already concluded that this simply wasn't possible if the aircraft was to finish up at the actual crash site.

Perhaps a take off with a gentle 5° bank which then increases steadily to perhaps 30° bank at the highest point in the flight, followed by a jump to 45° increasing to 60° over perhaps 5-10 seconds, then a further roll to 70+ degrees to the impact point. I feel this would more accurately reflect the possible flight profile and allow the turning radius to agree with the results recorded in the accident investigation as to impact point in relation to the runway and airfield.

The problem is that once the aileron cable broke, it is very difficult to establish the position of the ailerons between that point and the moment of impact, when witness marks would reveal the deflections with reference to the wings. Did the ailerons remain in one fixed position, did they flutter in the slipstream, were they influenced by the engine power settings, or did they move in response to control power boost devices being cycled off and on, perhaps repeatedly?

Considering the flight as a series of step changes which temporarily mitigated the situation, but then failed is probably the way to go in understanding this accident.

Well, I CAN alter the aileron settings, but I don't. Once the turn is initiated I leave the ailerons in that position for the remainder. I use a variety of settings in different runs, but it's stable within each run. There's a little jiggling out past the second decimal, but that's generated by the sim and I can't control it. For this series of tests and examples, I'm presuming the ailerons aren't moving after the cable parted. I've been doing some other runs using a scenario where the ailerons return to neutral, but no useful results yet.

It's possible the ailerons were moving significantly after the break. In order to test possibilities, I need a backstory to justify the exploration. I can't presume they were moving at random and find all the outcomes. Statistically such research is impossible because the permutations are endless. Logically it becomes meaningless because I could concoct any movements I like in order to bolster my case.

If there's a solution (and there's no guarantee I'll find one) it needs a rational scenario for a framework. The system returning to neutral is one. The idea that they were doing something that resulted in some sudden movements is interesting, but I'm not sure how to approach it. The most accessible premise is that there was no movement, since the deflection noted in witness marks is a reasonable deflection to start a moderate turn.

Take a look at https://www.youtube.com/watch?v=MDc4fHJFchY

That's a demo I did to illustrate ideas in an email thread I've got going with a couple pilots. I've been exploring the reaction when the left rudder is initiated at different points after the cable failure using different rates of change and different degrees of aileron deflection. I'm not getting satisfactory results, but it's getting closer.

I've been doing some experiments with hand-flying the simulator in high-angle banks with hard left rudder, to see how it behaves and to see at what point it can't maintain altitude. It's possible that the real answer lies in a higher bank than I'd previously considered, maybe in the mid-50 range, with a rapid left rotation on first impact.

The damage geometry says the plane was banked no more than the mid-30's when the prop left scars on the embankment, but the embankment actually had several strikes. I'm impressed with how fast the plane rotates counter clockwise about its longitudinal axis and right about its vertical axis when it hits the right wing a glancing blow, and then straightens again under rudder influence. Perhaps the answer is that it hit the embankment at a mid-50 bank, and left prop marks while in a mid-30 posture within a matter of a few dozen feet.

I don't know how much weight we can put on impact behavior in the sim, since it's a flight sim and not a crash sim. It's very doubtful that we can trust the plane model to accurately reflect the effect on the plane from tearing off half a wing.

The videos certainly demonstrate that flight is considerably prolonged with the application of left rudder to counteract aileron deflection.

The question of the aileron positions during flight can perhaps be simplified somewhat by considering that the aircraft maintained a constant turn to the right. This suggests that the aileron position at the point the cable broke was at the very least maintained.

Application of left rudder failed to rectify the situation, so on the balance of probability, the ailerons appear to have moved somehow to a position where they became even more effective at rolling the aircraft despite the application of full left rudder. The aircraft would have been flying in an unbalanced sideslip. Would the disrupted airflow over the wings have been enough to move the ailerons, sucking them further from a neutral position?

It is probable that the engines on the right wing would have had the power pulled back at some point in order to try and get the wing to drop. The left engines would be left at full power to try and prevent an incipient spin developing which would have led to an immediate crash.

As you say, the closer you look at all the variables and interactions that could have been involved, the more difficult the problem becomes in arriving at a credible solution, rather than one that just happens to fit.

This is why I suggest that the flight needs to be considered as a series of events with the aircraft reacting to the failure and the crew's attempts to corect the problem.

Aircraft takes off a bit long and low on the runway, not necessarily a problem.
Aircraft begins a standard banked departure turn.
Aircraft climbs in a banked turn.
Aileron cable breaks.
Crew are unable to adjust the angle of bank using ailerons.
Bank increases.
Up elevator commanded to maintain height?
Progressive application of left rudder to the limits.
Further movement of aileron due to disrupted airflow over the wings in sideslip?
Aircraft starts to descend.
Crew reduce engine power to right wing engines, maintain full power on the left.
Does the application of differential engine power move the ailerons still further out of the neutral position and increase the bank angle?
Aircraft continues to roll and descend.

I have a hunch that doing the right thing in reducing engine power on the starboard engines while leaving the port engines in full power may have resulted in the ailerons being moved out into a more effective, rather than reducing to a neutral, position in the slipstream. Flight training would have taught that applying more power to the port (left) wing should raise the wing and turn the aircraft to the right. So while the flight crew reacted correctly and in accordance with their training and experience, it is possible that the "correct" course of action would have been to completely and counter intuitively reduce engine power, particularly on the port wing first. This would inevitably have resulted in a crash landing, but just possibly in a more wings level attitude.

As I say, this is just purely supposition and it is possible that the crew did finally take a counter intuitive approach to control of rhe aircraft when it became obvious that a crash was inevitable. That could certainly account for the apparent reduction in bank angle before the initial impact, which is what you have been trying to explain all along if I have my facts straight.

I doubt that these possible elevator control inputs, aileron positions, engine power settings and rudder inputs can be duplicated using personal computer flight simulation software, but approximations can be made to test the ideas. As you have already demonstrated, just applying a hard rudder against increasing bank angle dramatically prolongs the flight.

I think you are making genuine progress in explaining what actually may have happened. You are certainly building a fairly strong case that tends to support and explain your version of events.

I don't think the accident investigators at the time were at fault in any way. They had to work with the resouces available to them at the time and no doubt with considerable commercial and government pressure to produce a result quickly.

You have the benefit of being able to review and model potential scenarios over and over without any constraints. I would be surprised if you don't eventually arrive at a better solution to this tragedy, one that demonstrates that your theory is correct and also agrees with the results recorded at the time, if not the conclusions.

This is going to take a couple posts to reply to. Pardon while I use this one to vent.

With regard to faulting the investigators, my fault-finding is limited to their description of the flight and impact. I give them credit for finding the root cause and tracking down the people and practices that led to the failure.

I was raised in a flying family, and we all understood the vital need for accurate and objective analysis of accidents. It was something we discussed from time to time, and we were aware of my dad's views on the subject. Accident investigation is pretty close to sacred. We've all seen the lengths investigators now go to in retrieving recorders and salvaging wreckage from inaccessible locations.

When I realized that the investigators in this case had misreported as badly as they did, I was stunned. I didn't know they would do that; it was inconceivable. The actual statement is relatively innocuous - they said the plane slide tail-first and right side up to a stop. The pictures show it tail-first and upside down. By itself this is not vital data. Quite a few have told me, some with as much patience as they could muster, that this is irrelevant to the root cause, which of course I knew already.

What this factoid does do is tell us that they didn't much care about this aspect. I agree they were probably under pressure to close the case, but even so their work in this area (as reported officially) does not meet a minimal standard of care. I say 'as reported' because I cannot believe there weren't people focused on every aspect of the flight. The report implies several teams working on different areas, just as they do today. They did site surveys and debris mapping. There had to be people who understood the impact scene much better than did the person who wrote the report. But for some reason, what made it into the official report was wrong in this respect.

We can now say with confidence that the flight sequence in the report was flat out impossible. I'm reasonably sure that there was dissension at the time. Heck, they had 54 witnesses and it happened on airport grounds. There is a hint of witness disagreement, and the CAB clearly disagreed with Lockheed over the flight path. I'd love to see Lockheed materials, but they have not responded. I think any pilot, given a description of a vertical bank extended over a couple thousand horizontal feet at a height of 300 or so, would dismiss it out of hand. Correct me if I'm wrong.

The CAB report could have addressed this without negating the value of the conclusions. They could have said there was confusion about flight details, but not about what broke. Instead they drafted the cartwheel-and-backward-upright-slide scenario with perfect confidence on the afternoon of the crash and went public with it. I find this unconscionable. The ALPA report writer appears to have done even less work on it and written more purple prose.

Their handling of this phase left us with a blank space that we didn't know about. If I seem to take all this personally, it is because I was personally injured by their carelessness. They left us with a misunderstanding for 55 years. As I peel apart the layers, it's clear that there are things that friends and family would have liked to know, and it's too late.

Now that the lapse has become clear, I'm simply trying to fill in what they failed to address. My materials are very sparse, and the chances of more information turning up are virtually nil. So this project is now closer to its conclusion than its inception, but there's still something we can extract.

Knowing that the plane arrived at its final site nose-first and right side up tells us that the entire impact sequence was misrepresented. By back-calculating along the debris path, it tells us that the first impact was described wrong, and this says the flight itself was misrepresented. The flight simulator can not give precise numbers, but it's adequate to show patterns and trends. And reverse-engineering the flight seems to open up a window on what the crew was doing and thinking.

There are moments of this, I confess, which are extremely creepy for me.

So even though the investigators found the root cause, I fault them for their misreporting.

On the technical side, let me explain what I think may have happened.

We know that at 165 knots knife-edge flying an Electra is impossible. That is, a 90-degree bank is impossible, and an 80-degree is impossible, because the rudder can't lift the nose. But flying level, the rudder hard left can flip the plane over by swinging the nose left. In a shallow bank the rudder can easily overcome even extreme aileron deflection.

In a shallow bank there is some small portion of the rudder force which is actually lifting the nose, not swinging it left. The more extreme the bank, the larger is the nose vector going up and the smaller is the vector going left. At 90 degrees the entire force is trying to lift up, and it can't manage that at 165 knots.

Think of rudder deflection as a curve of decreasing effectiveness correlated to an increasing bank angle. At any given speed up to about 250 knots, there's going to be a point where the left rudder cannot lift the nose. As long as the nose can't lift (i.e. move left) the plane can't roll left by using the right wing dihedral and differential leading edge exposure. I ran some experiments with the simulator, and in an Electra at about 165 knots this point is about 55 degrees. A bank more extreme can't be countered by the hard left rudder, and less extreme can be.

A couple pilots have told me that no one trains for what happened to my dad and his crew. I've had two suggestions that they may have not called for enough left rudder or done it early enough. At the same time, I was looking at the flight as a purely mechanical problem. A hard left rudder in a right bank is an extremely unstable configuration. Either it responds quickly or not at all, and if not at all the plane crashes very fast. My goal was to figure out how the plane could get 4,000 feet from the runway, and it occurred to me that increasing the left rudder slowly rather than quickly might give an appropriate result.

Now I had a narrative suggested that would explain a slow increase, and a computer problem that seemed to favor a slow increase. So I started running trials and got a huge bunch of very close matches to the observed flight. I put one on Youtube with comments, and might try to tweak it a little and then put some more up.

See https://youtu.be/cVUta2x3pPc

Try this idea on for size: When they realized that the plane was not responding to control column left input, they first thought they had an uncommanded autopilot activation. The autopilot was carded inoperative. They spent a few seconds on this, and then started calling for left rudder.

But the left rudder input they used was appropriate to shallow banks, and by the time they started they were already at about 40 degrees. They increased the left rudder to something that would have countered a steeper bank, such as 30, but by then they were at 50. In effect they were chasing the bank down the curve of rudder effectiveness.

I'm told that pilots never train for anything like this, which suggests they are not trained that the rudder will become increasingly useless at high bank angles. (If someone can give me feedback on this, please do.) Finally in desperation they did what they should have at first but couldn't have known - they went to hard left rudder. The plane was by this time near the bank angle of no return, if I can call it that, so recovery might have been slow.

As the plane started to respond the bank lessened, but at this point they were too low and there was too much downward inertia to recover.

To put this in a larger picture, if Dad had an aileron loss of control, or if an uncommanded right bank started while they were straight and level, they would have had no trouble at all controlling it. They could have come around and landed, maybe after a big sloppy ponderous turn. Might have been a messy landing, but all doable.

Their problem came from the fact that they were already in an intentional right turn when the cable parted. This meant they didn't know they had a problem until they tried to ease the bank. There would have been some delay in figuring out what was going on and what to do. Further, this meant they were already in an attitude where the rudder would have decreasing effect. They were some way down the curve to the point where it would be useless, but they didn't know that.

The CAB report said their problem was that the cable parted at too low an altitude to recover. I think maybe that should instead read that the cable parted when they were already in a banking turn. If they had an uncommanded bank while flying level, they would have had no great trouble at any altitude. At the point they first knew there was a problem, they were already moving fast into rudder-degrade territory. I think the intentional bank was a decisive factor.

Does this make sense?

Without a doubt this is the closest simulation run yet to approximate the flight path. That really is quite an impressive run. I think you may have cracked it.

Any pilot will tell you that control inputs, particularly on large aircraft, need to be made in a considered manner otherwise you will tear the wings or tail off. The aerodynamic forces really are large enough to cause instant failure of the tail if full rudder is commanded suddenly. Smooth, measured, control inputs are generally one of the marks of a good aviator. The Airbus crash in New York shortly after 9/11 was caused by a relatively inexperienced pilot violently applying full rudder inputs when they hit wake turbulence from a preceding aircraft. The tail broke away and the aircraft crashed.

Something must have convinced the investigators that the aircraft hit the ground the right way up and facing in the direction of travel. Most likely witness marks on the ground from equipment in the lower part of the aircraft or detatched parts from landing gear, etc.

At some point after that initial impact the fuselage rolled and rotated, ending up facing the wrong way and upside down. My best guess would be a wing snagging the ground and the rotation taking place about that point. Indeed there may have been a couple of rotation points with perhaps the nose digging in flipping the fuselage upwards and then a wing snagging and dragging the fuselage around before the wing broke off and the fuselage rotated upside down and backwards. It would all have happened extremely rapidly, one or two seconds at most, so plenty of room for confusion about the exact sequence.

I am begining to come to the conclusion that the description of the impact sequence was hastily written and not revised because it was not strictly relevant to the root cause of the accident or its' survivability. The investigators found the root cause of the mishap, with the broken aileron cable and were then able to account for the flight path and impact point. After that, the investigators were not too concerned with the break up sequence other than points that affected their investigation as to the cause. There was a need to be able to demonstrate the aileron cable broke in flight, for instance. Having effectively backed themselves into a corner with an early publicity release about the likely impact sequence, perhaps the investigators decided to just leave that as the official record, rather than go through the trouble and embarrassment of issuing a retraction. Such a correction might have cast doubt on the investigator's abilities or accuracy, certainly in the public's eye. I don't think there was any malicious purpose behind the inaccuracies.

To summarise, I think that you have arrived at a reasonable simulation of the flight path, one that reasonably matches the witness reports and investigators findings. You have demonstrated that the impact sequence, break up and final disposition of the aircraft wreckage were not accurately described by investigators after the initial impact point of the main fuselage. The aircraft may have slid for some distance relatively upright, but clearly it rotated upside down and end for end. The investigators did not accurately describe that sequence, but may have been constrained by a too hastily released press report. They left their report written to agree with the early press release so as not to cast any doubts on the investigation. The root cause of the accident was established and amending the accident report to more accurately account for the final position of the aircraft wreckage would not materially affect any of the findings.

I think that is possibly the most accurate account that can now be written after all these years and without new evidence coming to light.

Megan, you are correct. The point I was trying to make is that pilots are generally predisposed not to make large and sudden control inputs, particularly in commercial aircraft. It makes the passengers feel uncomfortable.

Two interesting wrinkles turned up. Been going through all my Electra manuals looking for anything on control surfaces and problems.

An Eastern and a Qantas manual both mention that in the event of a loss of authority over any control surface, the best course is to engage the autopilot. The autopilot communicates directly with the boost units, bypassing all the cabling, which of course is where their problem was. But in this incident the autopilot was tagged inoperative. I don't know if that means it was completely nonfunctional with parts yanked, or was in need of adjustment and was considered not trustworthy, so I don't know what would have happened if they turned it on. Investigators suspected that the crew thought the autopilot had engaged and were trying to turn it off, which suggests it was in fact operational but not trustworthy. In any case, they weren't able to take the one recommended course, perhaps because they thought it was the problem.

The second factoid is that in the event of a control surface problem (and only when the autopilot is not available) the default action should be to disconnect the boost for that surface. The Qantas manual says that they can safely cut all boost units in such a case, and recommends it. In a normal flight attitude this would be no problem, but they were in a bank that was getting steeper by the second. Cutting the boost increases the force needed at the flight station, and also reduces considerably the range of movement of the surface, leaving enough movement for what the manuals refer to as normal needs. But in this instance they needed to do full left rudder. So if they followed the manual instructions (presuming NWA specified something along these lines) they were inadvertently crippling their ability to counter the bank using the rudder.

ETA: Just found it in another manual. The movement range with boost off is approximately half for all control surfaces.

Without any artificial boost, the control surfaces could only be moved as far as the physical strength of the pilots would allow against the aerodynamic forces. About 50% of the full range assisted movement seems reasonable.

It would be interesting to know what the problem with the autopilot was and the process for disabling it when it was inoperative. Just a circuit breaker pulled, a fuse removed from a panel, or switched off with a placard placed over it.

If the autopilot was capable of operating the ailerons independently of the manual flight control cables, then this aircraft should never have been allowed to fly. With the auto pilot inoperative, the cable presents a single point failure mode where maintenance of normal flight becomes impossible.

A bit like taking off in a modern jet with only one generator operative and the other generator and APU flagged inoperative. It shouldn't ever happen, but it does.

One question to consider is whether when the cable broke, did the controls lock in position because the broken ends of the cable jammed and snagged preventing movement? That would certainly give the impression that the autopilot had somehow engaged and was fighting the crew for control. Under the circumstances it is probably reasonable to first assume that the autopilot has engaged somehow. Something as vital and simple as a mechanical control cable breaking would probably not be the first item to spring to mind, especially in what was a complex aircraft for those times.

So perhaps the root cause was that the aircraft was despatched with an inoperative autopilot, removing the only possibility of maintaining control in event of a mechanical failure of the flying control cables.

I went back and looked at the report. The CAB said the autopilot was tagged inoperative pending a required modification. This suggests that the parts may all have been there and functional, but there's no indication what exactly the crew would have known. The report did mention that the autopilot switch was accessible.

My comment about wondering if the crew thought they were fighting an autopilot that had activated uncommanded comes from recollections of what we discussed at the time. We had a lot of conversations with NWA pilots and some contact with investigators, which is why the ALPA report is a carbon we were given. The CAB report doesn't directly address that.

The CAB report said spark damage indicates there was power in the autopilot, but they couldn't rule out something happening in the impact sequence.

The CAB also said the boost units were in the engaged position for the elevator and rudder, and disengaged for the ailerons. This is the ideal configuration, but they also said they didn't rule out that the impact forces acted to engage or disengage something that was set the other way during the flight.

This whole incident has really started to flesh out for me now. My next document version is going to have a section covering these systems and how they're intended to (and perhaps did) interact.

I've uploaded another video to Youtube. This is my best sketch so far of the plane's flight.

https://www.youtube.com/watch?v=D4_b1ekxKkY

The video includes several camera angles, simulating views by witnesses in different locations. It becomes abundantly clear why witnesses disagreed, and why the CAB and ALPA got the bank angle wrong.

Earlier I thought the shallow bank meant they should have been able to recover, but eventually I acquiesced to the consensus and agreed they couldn't. Now that I look at a simulation that ends in the shallow bank evidenced by the impact damage, it looks more than ever like it was close to recoverable. The video makes that clear as well.

I can't swear the video portrays the real flight, but it matches the site survey and the impact damage, and clarifies witness statements. I'm putting this out as an approximation of what the real flight must have been, with a certain degree of slop in all directions. My video matches the site survey within 1.5%, which feels pretty good.

I'm grateful to all who both agreed and disagreed with me. You've helped me clarify my thinking about this.

Craig,

Excellent video and simulator work. You clearly demonstrate that it is virtually impossible to estimate bank angle accurately unless the aircraft is flying directly towards or away from the observer. From a side on viewpoint, as soon as both wings are visible, the brain sees the aircraft in a more severe bank than it actually is.

This psychological effect doubtless contributes to popularity of airshows. The spectators see spectacular manouevers from the ground which are relatively benign from the pilot's seat.

Bank angles of up to 30 degrees are part of flying a normal instrument holding pattern, so are not normally of much concern to pilots, but it would look like a much steeper bank from many points on the ground.

Your reconstructed flight path and bank angles do indeed seem to match the accident report very closely. I doubt that anyone will ever be able to do a better job of recreating the flight path than this. Impressive work. Well done.

A brand new puzzle has turned up. It's almost charming.

I've been scanning newspaper archives online. The best source has been the Chicago Trib. They followed the story closely, and they carried the graphic (morning after the crash) showing the rotation and erroneous backward slide.

On Tuesday they reported that one engine was found buried eight feet deep in mud and one other engine had been located, with two still missing. On Wednesday they reported that all had been found, with three of them (2, 3 and 4) 'close together' on open ground. It seems very odd that they could find one of those three on the day after the crash, and it took another day to see two more that were close to it. Since their frame of reference was that these were about 30 feet from the forward fuselage, we may take 'close' to mean something on that order of magnitude, as opposed to (say) a hundred feet apart.

This was open ground with no other wreckage in the area except the forward fuselage. In my document I pointed out two objects in photographs which were candidates to be engines, in the area the Chicago Trib described. An Electra engine is about 12 feet long and 29 inches high lying on its side. If you think of a Jaguar F-type, you're not far off. There were (per the Trib) three of these within about three car-lengths of each other, but they couldn't find two of those for a day.

Further, the one buried in mud was said to be number one. There is a distinct line in the aerial views where something went zooming off to the southwest from near the tracks, which I strongly suspect but can't prove was an engine. But if this was engine one, then it separated before the left wing or nose touched the ground. This seems as unlikely as not seeing two engines lying on the ground within a dozen yards or so.

This was not the CAB report; this was the Trib reporter relating what he thought the CAB rep said. I'm familiar with terrible reporting, and this sounds like something was garbled in the communication. These two Trib stories, taken together, are invalid on their face. I suspect two engines on open ground near the forward fuselage were just as I described in my document - engines one and two, which separated just east of their final location.

The irreducible core of this report is that the CAB said they could not account for the full complement of engines until Tuesday, and on Monday they were puzzled by this.

This is not something a reporter is just going to make up; it had to have come from the CAB. So why was there difficulty accounting for all of them? As in several other pieces of this documentation, the question is how much we have to discard before the rest makes sense.

Nothing was moved from the crash scene under official auspices until Wednesday at the earliest, when mapping was done. The Bridgeport Post reported on Monday that the rail line was out of service 'for a time', so the track was reopened to traffic on the day of the crash. The question is then to ponder what might have put it out of service.

The track was clearly shifted about a foot, but the distortion is over about 40 feet, and is within the range of normal distortion for poorly maintained track. The railroad could have simply stopped rail traffic while the site was surveyed for damage. There may have been some broken ties that either needed replacing or at least needed review before they allowed rail traffic to resume. There may have been some sheet metal and other wing material on the track, which could have been moved by several people lifting it off. Or it might have been engine four.

I have said that it looks like engine four was left lying on the railroad tracks. Two pictures show an object on the rails that looks to be the right size, with a couple people standing near it.

The railroad would have been the ones to move engine four off the tracks, if that really was engine four in the pictures. They had the rail-mounted equipment and know-how to pick it up and hand it off to the CAB at a convenient intersection perhaps.

I'm intrigued by the possibility that they did this without coordinating with the CAB, thus resulting in a missing engine until it was located sitting on a flatcar. It seems more likely that they did get some kind of ok from a CAB investigator at the tracks, and then this wasn't communicated to the CAB site coordinator. Either way, the CAB investigators were left thinking an engine was missing, and this impression lasted long enough for the spokesperson to brief the Trib reporter, who then got it a little garbled in retelling.

This small puzzle feels almost parenthetical. The odd story of the missing engine does not by itself prove that engine four was left at the tracks. But if engine four did indeed land on the rails, my confusion scenario explains how an engine could be reported as missing for a day, and therefore constitutes circumstantial evidence which is admittedly thin but entertaining.

There doesn't appear to be a plan showing the position of the major components in the official reports, although it would seem essential for a thorough investigation. Interesting that no such document has shown up in the records.

The four engines are probably the most concentrated blocks of mass in the entire airframe, but they are not that large when cowlings and accessories are stripped away from them.

The ground was relatively soft, so all the engines were probably partially buried in ground that was pretty chewed up by the crash and response vehicles. I don't think it is too much of a suprise that it took a couple of days to get everything found and mapped out.

I suspect only the wing tip broke away in contact with the embankment and perhaps one engine on first contact with the ground beyond. The remaining engines separated more or less together at a later stage of the impact sequence. That would account for them all being found close together, with one engine a distance away.

I'm pretty sure there would have been a lot more damage to the railroad track and embankment if an engine had made contact at that point.

If an engine had been left on the railroad tracks or the embankment, that would have made for a terrific news photograph. Someone would have documented something as dramatic as that for sure. So the debris on the railroad track is most likely bits of wing and perhaps a propellor tip.

The engines possibly breaking up is certainly a valid issue. The CAB report said that in only one case did they fail to get torquemeter readings from an engine, which suggests that only one engine broke apart at the drive shaft. One newspaper article mentioned that the investigators were surprised at how well the engines held together.

They mapped the site, of course, and all that documentation existed once. But when the CAB moved their records to the FAA all that was discarded. There's nothing in the National Archives and I haven't been able to turn up anything else. If I had a debris map all this would be easily settled. That's what makes this a detective story.

The energy of an engine hitting the embankment is one of the things I've wrestled with. I've tried to find experienced crash investigators to give me some feedback, but haven't found any yet. I go into that at some length in my document, but it's all guesswork.

You're right about the photographer getting that shot if he could. That plus the Bridgeport Post article suggests that whatever was on the track was moved really quickly. Something was there in the aerial view, but not when he was at the tracks. I have a faint hope of hearing more from the Chicago & Northwestern Historical Society, which maintains a lot of unsorted old journals and records, but their first response was not encouraging.

The ALPA report says that engine power indications came from the engine torquemeter itself, not from the cockpit instruments. The CAB report said that the only cockpit instrument with anything on it was the captain's artificial horizon, and they didn't trust the reading.

The ALPA said the engine power indications were read off of engines one through three, but no reading could be obtained from number four. It also said that engine four separated at the embankment. As I've said, engine four hitting the embankment is an excellent reason for not being able to recover the torquemeter value.

I agree, the torque settings would have been inferred from the position of components within the engines, rather than cockpit instruments. The position of fuel feed and oil feed valves, or witness marks, would provide an approximate indication of the power settings on the engines.

If engine four separated on or shortly after contact with the top of the embankment and then tumbled some distance to a final resting position, the disruption and damage to the engine might cause several sets of witness marks which would be difficult or impossible to accurately decode.

Thanks Megan.

I hadn't found documentation on how that works and was curious about it. See, that's what I come here for!

Ok, the missing torquemeter value doesn't tell us anything useful. Good to know.

Megan

An impressive amount of detail in your post. I had no idea that engine monitoring technology had advanced to that stage when the Electra was produced. I was thinking of the mechanisms within the engine that controlled the power output rather than the instrumentation side of things. I think your explanation is the correct one. Thank you.

I tracked it down in a pile of manuals dated '56 and '58 sent by my biggest supporter. The pickup circuitry was vacuum tube technology. It's all transistors now, I presume. This is very cool, and just as Megan described.

I'm way behind schedule on this project. Taking off a few months was intentional; this thing really wears on me for personal reasons and I needed a break. I had hoped to finish the write-up on the flight simulator, but it's at least a couple months away.

In addition, I'm exploring the prospects of animating the breakup, as contrasted to simulating the flight. The flight simulator was intended to play out real physics in a virtual world, to give my best approximation of the flight. But I have no access to the kind of tools it would take to simulate the impact and breakup. This would require vast amounts of structural data and probably a supercomputer to run the sim. But I should at least be able to use an animation tool to show what I think happened and make it more accessible to the reader/viewer. A side-effect of importing from the flight sim into an animation tool is that I can redo all my fairly crude line drawings with much better images. So I may delay the next version in order to at least do the import and redo images, before I get into the animation itself.

Meanwhile I've posted two interim files, which I've been circulating among my contacts since February. The first of these is a write-up of the missing engine puzzle discussed here in the thread. It will become a chapter in the next version. https://we.tl/2iOuz04If7

The second is an attempt to understand what was going on in the cockpit. This was pretty creepy, and is why I took a break from the project. This discusses the resources they had to work with, and I try to fit what they might have been doing into the time they had. A pilot friend said he'd forward this to an Electra instructor, but I don't have results yet. I would be very grateful to pilots for their feedback on this write-up and on whether I've come anywhere near the pacing of what would be going on under stress conditions like this. https://we.tl/x0L59Bpfmj

My impression from the sparse documentation at the time is that the initial efforts were devoted to recovering the casualties from the crash. Mapping of the wreckage wasn't initially seen as a priority, hence the confusion about the engine location(s). Nobody bothered to map it all out until the third day, when reports indicated that all the major parts of the airframe and engines were accounted for.

Cockpit resource management was in its relative infancy in those days. The crew would most likely be looking to the Captain to direct attempts to recover the aircraft, so probably not a lot of discussion, rather orders to perform certain actions.

With regards to the initial response and air accident investigation, there would have been very little organised communication initially between the first responders and agencies attending the scene. Undoubtably a scene of utter confusion for a day or two until proper lines of communication and assignment of responsibilities had been organised.

I wonder whether any of the current generation of P3 Orion pilots can give any insights considering that the two aircraft types are closely related.

Have you tried 'Failure Analysis' in Menlo Park, California? They might be of some help in (forensic) animation? PM me for contact info on a P-3 Navy Pilot.

Was whirl mode (investigation or AD) in play at the time of the crash?