I’ve had people ask me why I’m so bent out of shape about the 737Max and the reactions of Boeing and the FAA in the last few months. For reference I am not a flight controls engineer, but I did spend about 7 years of my life testing flight controls at NASA in the 1980’s and early 1990’s. The program I worked on was the X-29 Technology Demonstrator program at NASA Dryden (now NASA Armstrong) Flight Research Center.
It’s also important that many old friends from NASA and the X-29 program will probably make their way to this blog post at one point or another so I welcome their comments regarding anything Im getting anything wrong or jumping to conclusions that I should not be.
It’s all about airframe stability, or the lack thereof…
The mission of the X-29 program was to flight test several technologies and aircraft configurations. The obvious difference in the X-29 was that the wings looked like they were on backwards. The wings had a forward sweep rather than a rear sweep like most all other aircraft. This was done because aircraft with conventual rear swept wings at high angles of attack tend to have their wing tips stall first before the wing root. The control surfaces on a wing surface at stall don’t work very well because not enough air is flowing over them. The control surfaces at the wing tip have much more command authority of the aircraft than those at the wing root because they are farther from the center line of the aircraft (kind of like fulcrum). So simply put, an airplane with forward swept wings retain better control at lower speeds than that of a conventional aircraft making them more agile, a desirable feature in fighter aircraft.
The implication of this design however is that the airframe is inherently unstable. This means that if airframe were placed in a wind tunnel it would simply behave like a bad paper airplane and immediately tumble head over heals in nearly 100% of the flight envelope (ie all parts of the sky where the aircraft flights). While it would seem that this would be undesirable, it does have a desirable feature. If you are building a fighter aircraft a statically neutral or statically unstable aircraft can turn very quickly because you are not fighting its natural tendency to want to fly straight and level.
In order for the X-29 (and since then other fighters) to be able to fly at all they need computers to augment the stability. This means the computers move all the control surfaces many times each second to keep the airplane flying stable, straight and level. The pilot inputs are added to these computations in order to actually move the aircraft. In other words, to have the aircraft diverge from a straight and level flight the flight control system is actually relaxing its authority a bit and letting the aircraft do what it wants to a bit.
Airframe instability has massive implications on technology. If the flight computers don’t work or are acting on bad data, the results are catastrophic and for the X-29, uncontrollability would happen almost instantly because of the lack of static airframe stability. When designing a system like this the critical sensors are typically triple redundant (three of them). In addition, the flight control computers typically share all the sensor data between each other voting out a sensor that disagrees with the other two sensors, removing it from further computations used to stabilize the aircraft. When the degraded system is reduced to two sensors the ability to continue to vote is removed because if the two remaining sensors disagree its generally impossible for the computers to know which one of the two remaining sensors is bad one. In this case many systems are designed to average the two remaining sensors. This way if bad data makes it into the system, then the computed (over averaged) “less bad” version is used. This is not ideal but better than very bad data (like the 737Max used for Angle of Attack after a single sensor failure).
Enter the 737Max aircraft…
The Max was an extension of the very successful 737 line of aircraft dating back to the 1960’s. The 737 like all other commercial passenger aircraft is an inherently stable aircraft without augmentation from a control system. This is very different than the X-29 which was the opposite. However, when the Max was designed, larger engines were added to compete with A320 Neo aircraft that Airbus sprung on the world. This was all fine except to fit the engines under the wings the engines were added more forward of the wings changing ratio of Center of lift to Center of Gravity.
This change didn’t make the aircraft unstable in the entire flight envelope, but it did make it neutral to slightly unstable in the take-off regime of the aircrafts flight envelope. Specifically, if the pilot used excessive thrust the increased lift produced by the engines would tend to make the nose continue to rise at an ever-increasing rate, in essence requiring augmentation to keep the pilots from over commanding pitch attitude. MCAS was an ill-conceived band aid to this problem.
Boeing made a stable airframe unstable in a critical part of the flight envelope…
With that background, my problem with this situation is Boeing “broke the rules” they and other manufactures had used for decades designing commercial passenger aircraft. They now had an aircraft that needed augmentation to be safe. This is a fundamental design change and they did it to get to market faster. It wasn’t a mistake it was intentional. In doing so they did not treat the problem with the respect it deserved. You simply can’t add a required augmentation system that is driven by a single sensor. If that sensor produces bad data, the augmentation control system believes that data and you end up the 737Max disaster. In fact, you really can’t run this system with two AOA sensors (all that the Max has) because if one fails the system is incapable of figuring out which is bad. Boeings response in their “fix” is simply disable the MCAS system when the two disagree. That’s all fine but that leaves the pilots with a neutral to slightly unstable aircraft if this occurs during take-off.
Boeing engineers obviously knew the aircraft needed computer augmentation to be safe (or they would not have added it) but failed to tell the FAA or their customers about this. The Max problem is described now as a “sensor” problem in the general press and a software change can fix it. The debate going forward is if pilots should be trained in simulators or not.
I think this glosses-over the real issue here. The aircraft is fundamentally different than previous versions. It now requires a system to augment it and that brings an entire host of problems, not being addressed), that simply makes it a different machine.
The Max needs redundant sensors to collect air data (Angle of Attack, Dynamic and Static airspeed, etc) than an aircraft that is inherently stable in all parts of its flight envelope. These systems now must work, rather than it being desirable that they work. That’s a big difference and not many people are talking about it.
FAA’s Part in this mess…
The FAA for a long time has been in the pocket of the aircraft industry. The philosophy has been that it’s in the best interest of the aircraft companies to protect their reputation. This resulted in the FAA being reduced to a check box organization. The problem is this assumption has failed. Like the Challenger accident, engineers likely knew this fundamental shift in design philosophy was occurring and the problem was likely brought it up inside Boeing. Like the Space Shuttle Challenger my guess is it was ignored by Boeing Management in order to make the schedule and the FAA not really knowing anything that was going on just rubber stamped it.
I watch with interest to see where this goes with regard to the Max series returning to the sky. Im not optimistic that the right things are being done to correct the problems.