“Air France 447 and Air Asia Q8501 are clear cases where the crash was caused either by computer problems, or because of the interface between the onboard computers and the pilots”, says Tom Dieusaert in the second part of his interview with AeroTime. What should we do to avoid these accidents in the future?

In your book you have one chapter dedicated to Air France Flight 447 and one to Air Asia Q8501, although there are more incidents you cover. Is there a particular reason you chose these two specific examples to cover in more detail?

Both are clear cases where the crash was caused either by computer problems, or because of the interface between the onboard computers and the pilots.   According to reports in Air Asia’s case, there was a maintenance problem at the root of the accident: that plane had a malfunctioning RTLU, which is the rudder travel limiter unit, one of the fly-by-wire “protections” that limit the movement of the rudder to prevent excessive rudder inputs. In itself, a malfunctioning RTLU is not critical for a flight, so it very interesting how this rather small problem amounted to a fatal crash.

One of the reasons is that the more computers you stuff in a plane, the more potential problems you can have with these computers. I quote a pilot I interviewed for my book:  “Computers solve certain problems, but they create new ones”.

Tom Dieusaert is the author of “Computer crashes: When airplane systems fail”

Most troubleshooting on the modern automated planes basically boils down to resetting of the onboard computers but this procedure has its risks. You can have “spurious faults” as well as real mechanical problems that are masked as if they were a software issue. In case of the Air France 447 of AirAsia 8501, the pilots had restarted the RTLU a dozen times during the previous months before the accident, to no avail. Two days before the accident, the maintenance crew then decided to pullout the circuit breakers of the Flight Augmentation Computer (FAC), also with no result. Finally, maintenance decided to substitute the FAC by a new one, which seemed to solve the problem temporarily. But the RTLU fault messages appeared again on the fatal flight and the pilots (probably) pulled out the CB’s again in flight, after which they lost control.

Resuming, I think, if you have a system, which is over-automated, the risk exists that you’ll think “Oh yeah, I need to restart the computer” and you don't get to solve the root problem that is behind it.

Christine Negroni, aviation journalist and author believes that the biggest error done today is putting too much tech between the pilot and the plane. Do you agree with this position or do you think there are bigger problems causing plane crashes (at least among the ones you investigated)?

I do agree. It's a bit of a philosophical question, but it’s like driving a car. Thirty years ago, we would do it completely manually and with no computer controlling us. Then maybe in 20 years from now we will have totally automated cars. Now we find ourselves somewhere in between. With aviation it’s a similar situation.  There is still manual control, but automation is taking over and telling the pilot “Do this. Don't do that…”

Aviation is one of the safest ways of transportation, but tragedies, many of which are surrounded with riddles, do happen. Hundreds of professionals dig for clues and answers during investigations that can span many years. Despite their efforts, sometimes the truth is left unearthed and the public is given an explanation that only sounds plausible.

I guess that's an awkward situation for many pilots, because they don't feel like they are directly flying the plane anymore. They have so much technology in there and maybe a lot of technology that they didn't ask for. Maybe there are “smarts” and gadgets on the plane which are unnecessary. But that is the logic of automation and the industry. Sometimes the industry puts new smarts, because they want it to sell the plane. I can imagine that pilots sometimes get overwhelmed by the quantity of technical stuff. Another important risk of automation is that pilots lose their basic flying skills.

Will the pilotless planes replace the ones controlled by humans in the near future?

I think that we are going towards the unmanned aircraft. According to people in the industry, at first the two pilots will be reduced to one, with one pilot sitting in a container on the ground, controlling the plane by satellite connection. In a next phase there won’t be a pilot anymore in the cockpit and just on the ground. We talk about 15 or 20 years from now.

But according to safety specialists, the combination of a pilot together with the computer, which is helping him, is still the best combination. Because the computer can perform tasks that are boring and predictable. When there is a situation which requires skills or even improvisation, humans are much better. Otherwise, when the problem occurs and you are up there in the air, the only one who’s going to be in charge is the machine and sometimes the machines do crazy things, like the dozen examples in my book. Personally, I don't think pilotless planes are a good idea safety-wise, but the industry is moving in that direction nevertheless.  The biggest reason is because it is going to be a lot cheaper when you won't have to pay the pilots, deal with strikes and so on. Moreover, the prevailing idea in the industry seems to be that machines are more trustworthy than humans.

What change is needed in the industry to avoid the disasters like the Air France or Air Asia as covered in your book?

First of all, it’s very important that airplane manufacturers admit the malfunctions they have on their planes. That does nearly never happen. Or that the regulatory agencies take their responsibility and put the finger on these errors. There seems to be too much at stake.

Secondly, what a lot of the people in the industry also say, younger pilots need to go back to manual flying. As was the case with the Air France accident, the pilots lacked skills of flying without the protections of the fly-by-wire envelope. Probably all pilots that flying regularly on commercial airlines, should fly a little Cessna or a Piper to feel the sensation of real flying with thrust and pitch attitude. Today, young pilots are trained on simulators, which is a virtual thing and they lose the real feel of flying. When they’re up there in a cockpit, it’s like they’re playing a computer game, but at 35,000 feet.   

Tom Dieusaert, a Belgian journalist and writer, does not believe it is always pilot’s mistake at fault for plane crashes. He argues that pilots are commonly blamed for tragedies like the Air France F447, but something more subtle – the plane’s computer software – is often overlooked. Tom Dieusaert's book "Computer Crashes: When airplane systems fail" is available to order at amazon.com

The first part of the interview titled ““Blame the pilot!” The case of Air France 447” was published on November 1, 2017 and can be read here.