8 min read

The video of the 747 crashing after takeoff from Bagram Air Base in Afghanistan is hard to watch. The airplane had climbed only a few hundred feet and was flying quite slowly when it rolled a little to the left and then rolled off to the right, obviously out of control, as in the beginning of a spin. The only thing similar I can remember was video of a USAF B-52 starting to spin out of the bottom of a really steep turn years ago. In both cases, the airplane was quite low to begin and hit the ground shortly after control was lost.

747 crash in Afghanistan

The doomed 747 pitched up dramatically before stalling and crashing.

The B-52 was to be flown in some sort of demonstration and the pilots were really throwing the big airplane around. In that steep turn, the angle-of-attack was apparently not minded and the airplane did what any airplane, large or small, will do when AOA (or alpha) is allowed to reach the stalling point.

The 747, on the other hand, was on a routine mission to fly military equipment to another air base. Certainly nothing unusual was planned but something unusual certainly happened.

As pilots will do, after watching the video I came up with an idea on what I thought might have happened. If the center of gravity moves aft, stability deteriorates. Far enough aft, there is no pitch stability and it can finally reach an aft point where the airplane will stall regardless of what the pilots do with the controls.

The 747 had a loadmaster as a crewmember and with a lot of eyes on the weight and balance, I doubt seriously if the airplane was loaded with the center of gravity beyond the aft limit. This would mean, in this scenario, that the heavy load of vehicles could have shifted aft. To do so, it would have had to be improperly secured and that has certainly been known to happen.

Apparently this subject had come up because in a statement, National Air Cargo, operator of the 747, said that the airplane was loaded at another base and stopped at Bagram only to refuel. After the original loading, the statement said the cargo was inspected and was found to be properly loaded and secured and had passed all the necessary inspections. The cargo was again inspected prior to departure from Bagram.

The NTSB is investigating and someday we will learn the probable cause. For now, for me, it raises an important question. Most pilots know that overloading an airplane is bad but a lot of pilots do it anyway. They know that some performance and a little structural integrity is sacrificed when an airplane is flown overweight and they are willing to risk that. What I don’t think many pilots fully understand is that a transgression in the balance of an airplane can be lethal, especially if the aft limit of the CG range is exceeded.

This got a lot of attention a number of years ago, in 1977. The Commonwealth of Pennsylvania leased and then crashed one of the first turboprop Piper Cheyennes. Control of the airplane was lost shortly after a departure in instrument meteorological conditions. The investigation revealed that the CG was as much as 3.2 inches aft of the limit. There were eight adults on board which could explain this. Two would have had to be seated aft of the standard club arrangement.

Pitch stability had been a big question in the certification of the Cheyenne. In fact, to meet the regulations, Piper had to include an active stability augmentation system (SAS). Manufacturers had long used downsprings and bobweights in pitch systems to buttress longitudinal stability but the Cheyenne’s SAS was a first in light general aviation airplanes. Some pointed to it as a weakness in the design.

Piper Cheyenne

Piper’s turboprop Cheyenne required a Stability Augmentation System (SAS) for certification.

The Cheyenne was basically a piston-powered Navajo converted to a turboprop with a lot more horsepower. Horsepower is destabilizing in pitch, thus the problem.

The Cheyenne SAS kicks in when the airspeed drops below 125 knots. Then it starts applying forward pressure on the control wheel through a spring and at 100 knots it reaches the maximum push. The effect of this is to provide artificial control feel.

The stability requirements call for an airplane to always seek a trim speed and to return to that speed when disturbed. To go progressively slower or faster than the trim speed requires more pull or push.

At some point an aft CG condition can cause an airplane to reach what is called the stick-free neutral point where there is no feel in pitch. You don’t have to exert force to go slower or faster, you just have to move the elevator control. As you would expect, the aft CG limit is set ahead of the stick-free neutral point. It would be possible to control an airplane with the CG aft of the stick-free neutral point but it is difficult.

I have flown both a simulator and a variable-stability Navion operated by Princeton University and looked at both possible and impossible aft loadings and at best it will make you sweat and at worse control will be lost. It is a bad feeling, one you surely would not want to replicate without a way out.

After some original controversy about the SAS, the Cheyenne flew on. The pilots flying the airplanes apparently understood there was no margin in the aft CG limit and most kept it forward of that limit. Many disabled the SAS system because they didn’t like the way it messed with the controls when the airspeed dropped below 125 knots. I don’t think that ever contributed to an accident.

All stayed quiet until December 15, 1983, when The Wall Street Journal published a sensationalist, paper-peddling article that implied that the Cheyenne was unsafe because it lacked longitudinal stability. The report cast doubt on the FAA certification process for all airplanes.

There was so much wrong with the article that we at FLYING worked to set the record straight. Mac McClellan and I traveled to Florida, where the Cheyenne was produced at the time, and flew the FAA-mandated stability tests in each model of the airplane. It was an interesting exercise and I learned a lot about the stability characteristics of the airplane.

The hardest test to pass comes in a climb with the airplane trimmed for the best-rate-of-climb speed, with takeoff flaps, and with full power. Here the airplane must have a stable stick force curve, meaning pull for slower and push for faster, it must seek the trim speed when away from it and the controls are slowly released, and any change is speed must result in a stick force that is clear to the pilot.

At the time we flew, the Cheyenne IA was the basic airplane. It differed from the original, with 500 hp per side as opposed to 620 and with an aft CG limit two inches farther forward. In the full power climb test, this airplane had light stick forces but it had acceptable control feel and was easily controllable.

Next up was the Cheyenne II, which was the current name for the original with 620 hp per side. With the SAS operative it met the requirements though not quite as well as the Cheyenne I. With the SAS inoperative, it was at the stick-free neutral point in that full power climb and was flyable only if you knew what to expect and how to deal with it.

At the time, Piper was also building a Cheyenne IIXL with a longer fuselage than the basic Cheyennes and to keep from using an SAS system the company decided instead to limit the 620 hp engines to 500 hp for climb. Even at that, in the 500 hp per side climb the airplane was close to the stick-free neutral point and the return to a trim speed was not strong. It was the weakest of the Cheyennes that we flew.

Piper was also building the even longer-body Cheyennes III and IV and the WSJ article suggested they also had longitudinal stability problems. Nothing could have been farther from the truth. The original prototype did have problems and Piper addressed them with a new design that included a huge tail. When one of these airplanes is viewed from behind the tail almost looks bigger than the airplane.

Weight and balance chart

Pilots usually watch the weight–but what about the balance?

What this means to pilots in every day flying is that any airplane will be less stable with the CG aft. A pilot who learns in a Skyhawk might find it a much different airplane after he gets a certificate and starts taking friends and relatives for an airplane ride.

The wider the CG range, the more pronounced this can become. The V-tail Bonanza had a narrow range and had to be loaded with care but the handling qualities didn’t change a lot within the CG range. The P210 that I flew for 28 years had a wide CG range and went from an airplane with strong longitudinal stability to one that was honestly hard to fly accurately when the CG moved aft. It was so bad that I arbitrarily limited loading to two inches ahead of the aft limit, or, 50 inches aft of datum. In flight testing for supplemental type certificate mods to the airplane, the last time I looked nobody had been able to certify to the 52 inches aft that Cessna had certified. I know that one, and possibly two, P210s were lost in testing at that 52 inch aft limit.

There is one other thing to consider on the 747 in Afghanistan. There are bad guys with guns there so most operators climb to gain altitude as quickly as possible. The result is less airspeed margin above a stall than would be found in a normal climb. One report said there were thunderstorms in the area and the possibility that wind shear caused that loss of control has to be considered.

Richard Collins
25 replies
  1. Meredith
    Meredith says:

    One thing that could have been a factor in the Bagram crash is the steep max performance departure. This is usually standard in hostile airspace to avoid small arms fire. It seems possible that the straps and tiedowns just broke as all the weight of the cargo strained against them in the high angle climb. I’m sure that they will be re-evaluating the threat and the need for such a departure.

  2. Seth
    Seth says:

    The National Airlines crew is reported to have said over the radio that the load had shifted aft. How much of the load (multiple wheeled vehicles) and how far aft they shifted, I doubt we will ever determine. I DO know that the FAA has certified civilian air carriers to carry military cargo on military pallets in excess of the design limits (10,000 lbs. maximum per pallet). This is important because typically, a sub-floor of pallets are used to load large, heavy military vehicles since civilian cargo aircraft are not equipped with a dedicated floor stressed for wheeled vehicles to roll over like their counterpart military cargo aircraft (C-130, C-5, C-17). I believe there is a great potential that either a pallet (or pallets) gave way structurally and/or the cargo floor in the 747, letting the cargo shift aft during the initial climb out phase. It could also be a simple failure of the cargo restraints themselves, causing a domino effect of restraint failure. Either way, after the hard impact and large post-crash fire, there will be no way to determine what happened in the cargo box. At best, the surface position indicator’s feed to the flight data recorder will indicate if the pilot was too aggressive on climb out, or if he was trying to push the nose down the whole time.

  3. Tom Yarsley
    Tom Yarsley says:

    Back in the 1980s, we operated a fleet of Piper Tomahawks, as trainers at a mid-size flight school. The Tomahawk is an excellent trainer, but one potential “gotcha” that the vehicle does not really offer is an opportunity to load it out of its CG range. You’d need a lot of bricks in the back, along with gymnast-size pilots to do that. It makes it safer, but potentially can engender complacency in that regard.

    It was a cold late-winter morning. The previous day had brought a deluge of spring-is-coming rainfall. You get where this is going. The bottom surface of the Tomahawk’s horizontal stabilizer includes a couple of 1/4-inch water drain holes, courtesy of Piper. But not one for each “bay” (the areas forward of the spar and between adjacent ribs).

    A student with just a smidge of solo time in his logbook went out for the day’s first flight – some dual in the pattern with me. If he did well, I’d allow him to complete the hour alone in the vehicle. We weren’t more than 20 feet in the air when he said “something’s wrong,” and ceded control to me. As soon as I touched the yoke, I knew two things: he was right, and we had a problem.

    I assessed the situation. We had elevator authority, but completely unstable, knife-edge flying qualities in pitch. The pre-flight inspection had been uneventful; the control checks at runup were normal; the trim springs felt normal. There was no excess friction in the yoke tube, in either axis. But the vehicle was eager to depart from controlled flight. It appeared to be a CG issue, and since it wasn’t apparent that we had shed any parts in flight, I suspected undetected tail-area ice.

    Emergency declared, I flew what to all looked like a normal traffic pattern track, but I adjusted the power – cautiously – so as to maintain the same airspeed throughout. The no-flaps approach led to a smooth landing.

    Back at the barn, a couple of mechanics met us, followed soon by a couple of FAA types (we had a FSDO/MIDO on the field at that time).

    The fix was to drill additional drain holes in the underside of the horizontal stabilizers on all of our Tomahawks – so that there was a drain hole for each bay between the ribs. An SDR went to Piper and the FAA. Occasionally, I see a foreign-field Tomahawk, and I look for the extra holes.

    The moral of the story is that water can accumulate in lots of places, and occasionally it can take up durable residence as ice. Most of us have encountered this when encountering a rough-running engine at startup (ice in the bottom part of the spinner), or when attempting to open a wing locker on a twin Cessna. But tail-area ice is especially hazardous – even in small amounts – because of its long-arm location. It doesn’t take much extra weight at the tail to throw even an airplane with idiot-resistant loading characteristics well out of the allowable range.

    What had we done wrong during the preflight? Nothing. That’s the scariest part. The engineer in me yearns for pressure transducers at each wheel location, with active computation of weight & balance. It actually would work; it wouldn’t be as expensive as you might suspect (well, except for certification costs); and it would give a method of recording g-loading of “firm” landing events.

    And that student? He did well; got his PPASEL certificate; always a pleasant guy, but I haven’t seen him in years. I wonder if he remembers the morning when something was wrong.

    • John Townsley
      John Townsley says:

      Thanks for the annecdote. I’m ALWAYS looking for issues to think about during pre-flight. I wasn’t aware of this one.

  4. Robert Morris
    Robert Morris says:

    If this aircraft was so far out of cg to the aft why didnt it come in on its tail as it nosed in I think stab trim mechanical or setting was to blame here

    • Seth
      Seth says:

      First of all, as I said before, the ill-fated crew is reported to have radioed they were having a load shift problem just moments before they crashed. They’re dead now, so we can’t ask them why they thought it was a load shift problem, and not a mechanical problem (although they may discover that information when the cockpit voice recorder and flight data recorders are analyzed, assuming they weren’t destroyed in the post-crash fire).
      Second, why do you think the aircraft would fall backwards after stalling?? A load shift only has to get to a point where the pilot’s available flight controls are unable to counteract the nose high attitude that induced the stall. If ALL of the cargo suddenly shifted to the back of the airplane at once, MAYBE it would have fallen tail first. But aircraft designed with good longitudinal stability are designed to go nose low after a stall. Not to mention I’m fairly certain the pilot was not pulling back exacerbating the stall (as you would when intentionally stalling an aircraft for training or demonstration purposes), but fighting to get the nose down by commanding down elevator.

      • Eric
        Eric says:

        I had the same impression as Robert Morris and your explanation still does not answer it for me.
        If the load shift causes CofG to go aft of the balance limit and causes the nose to pitch up beyond the pilot’s ability to counteract it with the elevator before the stall, then how does the nose pitch down due to “good longitudinal stability” after the stall? It seems you are saying the nose will go up due to an aft CG then when it stalls the nose will come down due to design. It seems to me that the nose will drop on its own only when the CG balance is correct according to the design’s longitudinal limits. You can’t correct for an excessively aft CG with a better design. I’m not an aero engineer so please explain this.

        • Seth
          Seth says:

          The excessive nose up pitch that initiated the stall was caused by a load shift if the “witness” reports of what the crew said over the radio are correct. That means the wing exceeded the critical angle of attack for that particular wing design. As with all stalls in aircraft where the wing design demonstrates positive static and dynamic longitudinal stability, it will oscillate in the opposite direction of the initial disturbance (i.e., the nose will drop after the stall). The aft CG (and forward CG for that matter) limits are developed during engineering and confirmed with a combination of computer modeling, wind tunnel and flight testing. If the aircraft is flown within CG limits, aircraft control response and flight characteristics remain normal. If pushed past those CG limits, it makes the aircraft far more pitch sensitive and difficult, if not impossible to control (especially under the conditions of a stalled wing). But typically, aircraft that are too far past the aft cg limit and stalled don’t drop straight backwards, tail first. They enter a flat spin. In this case the CG didn’t shift so far aft as to enter a flat spin, but was far enough aft to render the aircraft extremely pitch sensitive and enter a deep stall, requiring a massive amount of altitude to recover. Unfortunately, they didn’t have thousands of feet to recover from such a deep disturbance to the longitudinal axis. So to recap, the load shifted aft far enough to make the aircraft extremely pitch sensitive, beyond control of the pilot’s flight control inputs. This caused the aircraft to pitch up abnormally and the wing stalled. Since the aft CG was not so far aft as to cause a flat spin, the longitudinal stability characteristics of that wing design made the aircraft pitch down after the stall. However, there was not sufficient altitude to regain flying speed and avoid impacting the ground.

        • K. Murphy
          K. Murphy says:

          Draw out the lift & gravity vectors for the wing & tail, and then think about how they change as the aircraft stalls. In my estimate (not an aero engineer either), the CG could shift aft enough to render the plane uncontrollable; but still be forward enough that when the wing stalls and its lift is lost the nose would drop.

          As long as the wing is flying and producing lift the nose would rise since the center of lift would be in front of the CG. Once the lift is removed the plane becomes a lawn dart, stabilized by the tail, nosing downward because gravity works on the CG unopposed by the (previous) C-lift.

    • John Townsley
      John Townsley says:

      The B-52 accident, like the C-17 crash in Alaska that followed it by about 10 years, was because the pilot rolled his own procedures, and ignored the material in the aircraft manuals. In the case of the C-17, the pilot was “well respected”…

  5. Jimmy Leaver
    Jimmy Leaver says:

    All pilots realize the importance of weight and balance, and therefore flying within the established envelope limitations. Nevertheless securing the cargo is also a vital issue, very often overlooked.

    Several accidents in the past come to my mind where the pilots starts the take off within CG limits only to enter into an uncontrollable situation which many times ends is an unfortunate crash. The cause, the cargo slid back putting the aircraft outside its established CG limits

  6. Cary Alburn
    Cary Alburn says:

    Dick’s comment that a “pilot who learns in a Skyhawk might find it a much different airplane after he gets a certificate and starts taking friends and relatives for an airplane ride” needs to be taken to heart as a lesson about aft CG. That’s why when I was instructing, I required that checking out in a bigger airplane meant one that is fully loaded, so that its aft CG characteristics can be learned. Any airplane loaded to the aft end of its CG limits flies much different from when it is flown with a forward CG. Whether its a Traumahawk with water in the tail feathers or 2 big guys in the 5th and 6th seats of a P210 or a shifted load in a 747, the difference is phenomenal if the pilot isn’t ready for it.

  7. Bob M.
    Bob M. says:

    As a 747-400 captain with about 15,000 hours flying the airplane, it was extremely difficult for me to even watch this video. When a friend said it was on line, I waited 3 days before I finally watched it. I did so, because there is always something to be learned from seeing what is rarely ever recorded. It was a heart wrenching thing to see. As always, it is not a good idea to say to anyone what you think caused the crash as the NTSB will eventually come out with its report and we will probably know all the details surrounding the event. Having said that, I think there are some things here that are fairly obvious to any pilot who has seen the video. Again, i will be careful not to say anything as fact. The airplane appears to have stalled. What might have caused it may or may not be obvious to most aviators and non aviators alike, but again, we should never say for certain, or in my opinion even speculate, as our expertise might be misconstrued as more than just conjecture. When the final report comes out, those who guessed correctly can pat themselves on the back for being right.

    Last but not least, we have lost some of our family here and we should never forget that these aviators didn’t get in those seats because of their looks. They were highly skilled at their jobs and as such, should never be criticized in any way. My heart goes out to all the families affected by this tragic event.

  8. Bill Scott
    Bill Scott says:

    Dick, I too appreciated the C-172 example you gave and intend to pass the entire article and the comments by others to my C-172 students and former students who are doing just as you describe with family and friends.
    One aspect you did not mention is the progressively aft shit of the CG as the fuel burns off. Using the C-172 example, if the four hour flight starts out with full fuel tanks and due to loading has a somewhat aft CG to begin with. The effect of that ‘slightly aft’ CG will become more evident as the fuel burns off causing the CG to move progressively aft.
    Presumably the designers took this condition into account and allowed for it in the layout of the Weight and Balance ‘envelope’. But nevertheless it is real. I flew many hundreds of hours in a C-210. As the fuel burned off and the CG shifted aft, the nose would rise and with manual trim an adjustment was made to maintain the desired altitude. As a direct result of the further nose down trim (and the reduced weight) the airplane TAS would increase. In a five hour flight the speed increase was usually about 12 or 14 kts IAS. The effect of the further aft CG was definitely noticeable during the approach to landing, particularly to a short bush strip where an extra five kts of approach speed was not desirable to maintain adequate stopping distance.

  9. Bruce Johnson
    Bruce Johnson says:

    Loss of pitch effectiveness cause by a load shift might be improved if prior to loss of energy a bank is entered and bottom rudder applied. This would require an altitude buffer to be available so the nose could go low and a forward shift to occur regaining pitch control.

  10. Bob Brewer
    Bob Brewer says:

    Dick:

    We (Piper) had Flight Safety put the assumed CG of the Pennsylvania
    aircraft into the Cheyenne II simulator when it came available…I don’t think anybody controled it after takeoff if not warned ahead of time…

    As I recall, the B-52 fiasco was 100 percent pilot error.. Apparently the guy had a reputation and the poor co-pilot had tried to get out of the flight because of his concerns….

    The Cheyenne I and III were beautiful handling airplanes.. Regarding the tail, I was on a demo with the prototype at one of the major New York airports, when a voice came over ground asking, “do you guys see that tail chasing the Cheyenne down the ramp?” Had to laugh because it was quite a sight at first…

  11. phil scott
    phil scott says:

    Looks like you nailed it, Dick. A report on the news last night said that the cargo of vehicles broke loose and shifted aft. I’ve flown to Afghanistan twice, both times in a C-17 packed with cargo, and both times it was strapped down to the max. Good thing, too, because the second time I had to sleep on the floor between a stack of pipe and a Humvee.

  12. Warren Smith
    Warren Smith says:

    Since the A/C had stopped for fuel only, the load tiedowns were obviously good otherwise the previous takeoff may have revealed a problem. The question of structural floor failure is probably what happened. Also the load weight is sometimes misread by the loadmaster, as sometimes hidden weight of an item of cargo can slip by. I was subject to such an event on a Navy A/C once in Italy where they loaded a large power converter mounted on a steel skid, and the manifest listed just the skid weight. The crew loadmaster didn’t know anything about the item other than it was as written on the manifest. Having worked with such equipment in an earlier career I immediately recognized it was being loaded too far aft and made an issue of it. My being a civilian created a bruhaha…but the pilot asked why I was concerned. [Primarily, I was a passenger,] and I pointed out the weight on the power converter’s nomenclature plate, and then that on the skid mounting. Everyone’s jaw dropped, and the unit was moved forward over the front spar.

  13. Seth
    Seth says:

    The tie downs weren’t obviously good. The accident investigation determined the vehicles broke loose and took out the hydraulics going to the tail. So not only did the pilot have a severe aft CG, he also had no hydraulics to any of the control surfaces in the empennage.

Trackbacks & Pingbacks

  1. […] melhor a questão no geral. E, para isso, eu recomendo este artigo do Dick Collins – “The weight – and the balance” – que, explorando um caso emblemático de P&B (do Piper Cheyenne) traz muita […]

Comments are closed.