What was wrong with the 737 MAX?

The software solution to a hardware problem.

Mar 31, 2021

In the past few years, it was one of the most infamous aviation news headlines: Two 737-800 MAX aircraft crash due to flight software errors. Lion Air and Ethiopian Airways had planes crash within 5 months of each other, killing 346 people. Both of these crashes happened for the same reason. Why wasn’t the issue identified after the first crash? And more importantly why wasn’t it identified before the first crash? These nuances, as well as the software and physics principles that caused them, are the topic of today’s discussion. Thank you for reading and don’t forget to subscribe if you haven’t already!

Let’s begin with why. The 737 Max is an iteration of a Boeing staple: the 737. This is extremely common for airlines to do. As technology advances, passenger comfort can be improved, or they’re looking for something to do, aircraft companies will iterate on their past designs. Sometimes this will result in a slightly larger plane, a different seating configuration, or in the 737 MAX’s case, larger engines. These engines are larger than their predecessors because they have a higher bypass ratio and are resultingly more efficient. Boeing wanted the MAX to be able to serve longer distance routes and could achieve this with brand new, more fuel-efficient engines. However, putting larger engines on an already in-production aircraft has a variety of impacts, many of which are not necessarily obvious.

None of these implications went unnoticed, including the one that posed a danger to the aircraft and its passengers. To grasp this, we have to take a crash course on torque. Torque is simply what results from a force being applied at a distance. The larger the force, or the larger the distance, the greater the torque. Let’s take a peek at the figure below.

Now let’s apply this to the 737: [1]

As the size of the engine increased, the center of it got lower to the ground, increasing the distance between the thrust force and the centerline of the aircraft. This increase in distance increased the torque, and this increase in torque led to a tendency for the nose of the plane to rotate (pitch) upwards. At this point in production, they couldn’t change anything about the plane physically to account for this, so they took an alternate path: software.

To mitigate this issue, and theoretically make less work for the pilots, they integrated software that would automatically point the nose of the plane back down if it noticed its pitch was increasing. This way, the pilots wouldn’t have to constantly correct the pitch of the plane—a nice thought. The only issue here is that it assumes that the plane actually knows its orientation.

Planes have a whole bunch of sensors. These can measure airspeed, altitude, temperature, as well as the plane’s pitch. Usually, there are two of each of these in case one fails. Even so, if both fail, usually the pilots can take the plane off autopilot and fly based on what they see. This is much harder, but they’re trained to do it. However, in this case, the software that pointed the noise down was not just part of autopilot, because it would need to be constantly correcting throughout the entire flight. As such, if both of the pitch sensors failed, it wasn’t as simple as “just switching them off”. This is exactly what happened in the case of Lion Air and Ethiopian Airways. Both pitch sensors failed.

When the sensors failed, the plane thought its nose was pointed almost straight up. So, to account for this, it automatically began to point the nose downward. Unfortunately, the plane was flying level at the time, so this action caused the plane to enter a nosedive. There was, however, a way to fight this. The pilots could pull the nose back up themselves and there actually was a way to disable the system altogether, but due to insufficient training, the pilots were not aware. As a result, the plane continued to pitch down, and the pilots continued to pull up. Over time, sadly, the software won.

After the first crash, Boeing released a bulletin outlining how to disable the system should it behave like it did for Lion Air. Unfortunately, it wasn’t fully processed by some airlines until too late. And 5 months later, the same thing happened again.

Since these incidents, Boeing has added further redundancy in the design of the system, as well as including an easy and widespread way to simply turn it off if the pilot believes it’s behaving incorrectly. Unfortunately, for many people, this solution came too late. Boeing was charged with fraud and a $2.5 billion dollar fine for withholding information from the FAA that could have prompted the earlier discovery of this as a potential risk. [2]

I don’t mean for this to scare you, but I know that it might. While this was an issue with the airplane, no other planes are known to have it, and all the ones that did have since been grounded and fixed. The one good thing from these incidents is that it has been thoroughly investigated so it won’t happen again. For these two crashes in the past 5 years, there have been over 180 million flights. That places the odds at 90 million to 1 of being on a fatal crash. [3] Or, you’d have to be in the air all day, every day, for 40,000 years straight to be statistically likely to be in one. All this goes to show that, statistically speaking, planes work, and if they don’t, it’s at least nice to be able to know why. Thanks for reading!

It's Not Rocket Science

Discussion about this post