Boeing 737 crash in China

Friends have been asking about China Eastern Airlines Flight 5735, the Boeing 737 (non-MAX) that departed Kunming and crashed nose-down near Wuzhou yesterday.

Without the flight data recorder it will be tough to determine what happened, but what I’ve been telling friends is that there are a variety of ways that an airplane can end up in an uncontrollable nose-down attitude.

In a conventional airplane, the wings lift up from just behind the center of gravity (CG) and the tail pushes down. If the horizontal stabilizer, which looks like a small wing near the tail, were to break off in flight, for example, thus resulting in a “no tail” situation, the airplane would nose-dive because the wings are lifting from behind the CG. See the following force diagram (source):

There is a substantial amount of overdesign in an aircraft and thus extreme maneuvers may result in a component getting stressed or cracked, but it is almost impossible for the horizontal stabilizer to come off. In the comments section below, a reader highlighted Japan Air Lines Flight 123, a Boeing 747 whose tail, and, more seriously, hydraulic systems, were damaged by the failure of a 7-year-old patch to the pressure vessel.

Is it possible to lose the downforce from the tail without parts of the tail becoming detached? Yes. This can happen due to ice accumulation (see NASA videos below). It seems unlikely that the accident Boeing got into severe icing at 29,000′ (where the steep descent began), however, because the air at that altitude is extremely cold and simply cannot hold much moisture. For the tail to stall while the wings were still lifting powerfully would likely require an unusual failure of the pneumatics, which take hot compressed air from the engines to melt ice off the wing and tail surfaces leading edges.

The horizontal stabilizer’s angle relative to the fuselage can be adjusted via the airplane’s trim mechanism. The runaway-trim-by-design is what brought down the Boeing 737 Max airplanes, but runaway trim can also occur in the non-Max 737, as in other planes. There are a variety of safeguards intended to prevent runaway trim (except in the Max where the computer actually held its finger on the “trim down” button in response to absurd data from a failed sensor), but if those safeguards fail somehow and the airplane is trimmed full nose down it might not be possible to recover.

An easy-to-understand cause of a nosedive is movement of the standard flight control surfaces, in particular, the elevator (just behind the horizontal stabilizer). This can be seen at airshows, e.g., in this video of Mike Goulian at Sun ‘n Fun (I’ll be there this year on Saturday and Sunday if you want to meet). Of course, Goulian pulls out of the dive by pulling back on the stick as he gets closer to the ground. If the elevator was stuck in the “stick forward” position does that mean that the pilots of the accident Boeing had the stick full forward? (i.e., the pilot suicide theory) No. Unlike in a lightweight family airplane, the flight control surfaces of a heavy jet are not directly connected to the pilots’ yokes/control columns. No human is strong enough to overcome the air loads of the wind rushing over the control surfaces. What drives the flight controls is 3,000 psi of hydraulic pressure generated by engine-driven and electric pumps (source):

(See also this thorough video explanation.)

How do the pilots of a heavy jet (or “pilot” if one is in the restroom) move a flight control surface then? Ignoring the modern fly-by-wire systems of the Airbuses, the standard technique is a cable that goes from the control column to a power control unit (PCU) next to the aileron, elevator, or rudder. The PCU uses the position of the cable to modulate the application of hydraulic pressure and it is the hydraulic pump that actually moves the surface. (more) Like everything else in aviation, these PCUs are almost perfectly reliable, but if one were to fail/stick it could lead to an impossible-to-control airplane. Here’s an NTSB report regarding an elevator PCU that got stuck in 2009:

On June 14, 2009, a Boeing 737-400, registration number TC-TLA, operated as Tailwind Airlines flight OHY036, experienced an uncommanded pitch-up event at 20 feet above the ground during approach to Diyarbakir Airport (DIY), Turkey. The flight crew performed a go-around maneuver and controlled the airplane’s pitch with significant column force, full nose-down stabilizer trim, and thrust. During the second approach, the flight crew controlled the airplane and landed by inputting very forceful control column inputs to maintain pitch control. Both crewmembers sustained injuries during the go-around maneuver; none of the 159 passengers or cabin crewmembers reported injuries. The airplane was undamaged during the scheduled commercial passenger flight.

An investigation found that the incident was caused by an uncommanded elevator deflection as a result of a left elevator power control unit (PCU) jam due to foreign object debris (FOD). The FOD was a metal roller element (about 0.2 inches long and 0.14 inches in diameter) from an elevator bearing. During its investigation of this incident, the NTSB identified safety issues relating to the protection of the elevator PCU input arm assembly, design of the 737 elevator control system, guidance and training for 737 flight crews on a jammed elevator control system, and upset recovery training.

See also this Wikipedia page on problems with B737 rudder and B747 elevator control due to PCU malfunctions.

So that’s everything that I know, which is to say… almost nothing relevant or helpful, unfortunately, just like everyone else on Planet Earth until and unless the flight data recorder and, perhaps, cockpit voice recorder, are recovered.

More on tailplane icing can be found in these NASA videos…

an older version…


7 thoughts on “Boeing 737 crash in China

  1. Thank you for the background knowledge. That Goulian video was a workout just to watch! He’s only pushing hard for about 4.5 minutes but the G’s he’s pulling/pushing in some places. What a Stress Test.

  2. Blancolirio explained how a stuck elevator would make it bob up & down. Someone had to be constantly adjusting the elevator to get the straight line on flightradar24. Seems to be another EgyptAir Flight 990.

  3. … the plane had three pilots on board – one more than ordinarily required. The third pilot was described as an observer seeking to increase their 556 hours of flight time.
    “From what we know, the performance of the three pilots had been good and their family life relatively harmonious,” said a China Eastern official, listing 6,709 hours of flight experience for the captain and 31,769 for the first officer. … Reuters quoted a Chinese state media source that said China Eastern is now requiring “two senior captains and a senior co-pilot on a three-person crew on some aircraft types” after the MU5735 crash.

  4. “There is a substantial amount of overdesign in an aircraft and thus extreme maneuvers may result in a component getting stressed or cracked, but it is almost impossible for the horizontal stabilizer to come off.”

    Coincidentally PBS America in the UK (yes, there is such a thing) is running a series ‘Plane Crash Recreated’ at the moment, an episode of which covered JAL123 (1985), where a 747’s tail fell off because the rear pressure bulkhead had been poorly repaired after a landing tailstrike 7 years before.

    The format has a set of several expert pilots who try to recreate the flight in a simulator, first handling it as an emergency (not so much as for the actual crew, as even lay viewers would recognise some of the scenarios) and then trying to recover from the known situation.

    Even freed from the problem of depressurisation, no pilot was able to recover this situation better than a 2000ft/min “landing” into the sea in a simulator. But the aircraft didn’t plunge nose down from altitude either in real life or in the simulations.

    Apparently 5735’s CVR was found.

  5. Yes, the picture does show the horizontal stabilisers in place, which is surprising considering that the fin extended some way in front: the whole tail was missing in images of the crash scene. As you say, the failure rendered the stabiliser controls unusable but the pre-failure trim might have been enough to keep the plane from nose-diving. So without loss of the horizontal stabiliser, a full-on dive seems to need accidental or intentional assistance from the cockpit.

    Today’s news is the finding of the FDR as well as further wreckage that might indicate a JAL123 scenario:

Comments are closed.