MH370 Sim Experiments


In Mondays blog, I dissected the impact of the under-dressed sim instructor CNN used for its round-the-clock MH370 coverage, but what of the simulator itself? Like many, I assumed the simulator, which looked impressive on camera, was the sort of machine used to train and type rate pilots looking to find a 777 seat. Not quite. As described on its web site, UFlys B-777 sim is a highly detailed reproduction of the cockpit and systems, but since the 777 is fly-by-wire, how accurate does it replicate the real airplanes control laws?

Maybe not so much. I mentioned that one of the interesting things CNN did toward the end of its intense coverage was to replicate what the airplane would do if both engines ran to fuel exhaustion. They even flew the airplane briefly on asymmetric thrust, since one engine would inevitably quit before the other. This was an attempt to illuminate what would happen if the airplane ran out of gas somewhere over the Indian Ocean, with pilots dead or incapacitated.

While Boeing has, understandably, tried to suppress speculation about the MH370 disappearance, sim operators-real, level D sims-all over the world have been experimenting with what-ifs in their multi-million dollar motion boxes. Ive been corresponding via e-mail with a couple of 777 professionals, one of whom is a training captain for a major airline. This week, he sent me a summary of experiments done by one operator and reproduced by others. Bottom line: You can throw any straight ahead flight path/impact calculations out the window.

What this is, really, is not so much a test of the airplane, but of the software that runs it. No one was quite certain how the 777s control laws would degrade and adjust to the loss of engine thrust and, briefly, electric power, with no human intervention. The results are eye opening.

The basic setup involved programming the sim with MH370s fuel, weight and CG conditions and letting it run out of fuel in track hold and altitude capture. Predictably, one engine flamed out before the other and a feature called TAC for thrust asymmetry compensation automatically applied rudder.

The speed decayed from 325 knots indicated to 245 knots. When the second engine failed, TAC returned the rudder trim to zero. Then the fun started. The autopilot dropped out and the flight controls reverted to direct mode. In the 777, Boeing designed three modes, normal, secondary and direct. Direct can be thought of as the modern equivalent of manual reversion; it gives the pilots direct control authority and strips away any envelope protection.

The sim experiment revealed that after autopilot drop out, the speed came back to 230 knots, but the nose slowly pitched down, eventually reaching 340 knots indicated and a descent rate of 7500 FPM. The bank angle got to 25 degrees.

At that point, the airplanes ram air turbine, an emergency backup generator, automatically deployed and the copilots PFD came back, plus other displays. But the aircraft remained in direct control mode. The 777s EICAS was peppered with alerts, including one that the APU had failed to start, which it would automatically do after the engine failure. But with no fuel, no APU start.

The airplane then essentially entered a stable, evidently non-damping phugoid with a maxium descent rate of 8000 feet and a pitch excursion range of 9 degrees down to 6 degree up and bank angles between 5 and 25 degrees. The speed fluctuated between 220 and 340 knots indicated.

The exercise was terminated at 10,000 feet, but there was no reason to believe the phugoid wouldnt have continued until surface impact. Although these findings are largely academic, I found them interesting nonetheless, especially the autopilot drop out. In the CNN sim clip, the airplane simply picked up a wings-level glide, suggesting to me that either the flight dynamics arent well modeled or they didnt allow the experiment to continue long enough. Either way, its inconceivable that the impact would have been survivable, if indeed anyone was still alive to survive it. Whether the type and violence of the impact would have had an effect on surface debris field distribution and thus probability of detection is similarly academic. One theory held that a survivable impact might leave the airplane largely intact and thus less or no surface debris. All we know for sure is that not a trace of the airplane has been found.

And that leads to this disturbing thought. My correspondent has mentioned to me a couple of times of having nightmares about what transpired in that cabin, never mind the cockpit. Im quite certain hes not alone and Im equally certain your imagination is vivid enough to construct a palette of scenarios, so I wont catalog my own. Theres not much comfort in knowing that we can at least ponder how the airplane might have behaved, but at the moment, with precious little else known, its at least something.

Wednesday a.m. addition: CNN shot this background video that offers more information on the simulator.

Join the conversation.
Read others’ comments and add your own.