In an interview to The Indian Express, he draws on early video footage of the Thursday crash and his decades of experience to explain what may have gone wrong with AI 717, what questions investigators will now be asking, and why patience — not speculation — is essential in the hours and days after a crash.

From your expert perspective, what is your initial impression of the air crash given that it occurred shortly after takeoff and below 1,000 feet?

One of the first questions, when looking at the video, is whether the airplane was properly configured for takeoff. During takeoff —when the aircraft is heavy, low, and slow—it needs extra lift, which is achieved by deploying the trailing-edge flaps (flaps are adjustable, hinge-like surfaces on the trailing edge of the wing that can be lowered to change wing shape in order to increase wing area and curvature, allowing the aircraft to operate efficiently at lower speeds during takeoff and landing). If the correct flap setting wasn’t used, the wings may not have produced enough lift to support the aircraft’s weight at low speed. This can lead to what we call getting on the “backside of the power curve” (an aviation expression for when an aircraft is flying so slowly that more power is required to maintain altitude than to maintain speed). So even with fully functioning engines, the aircraft can “mush” or settle into the ground if the wings aren’t properly shaped to generate sufficient lift. The thing that’s curious about the video is that the flaps appear to be up or at a minimal setting that wouldn’t be typical of a normal takeoff.

Also, at 600 feet, the landing gear (the system of wheels, struts, and other components that allow the aircraft to safely land and take off) should have been up (or retracted). So the question is: why wasn’t it? Was it a hydraulic issue or an electrical problem? Landing gear is hydraulically held in place and retracted, while flaps are electrically operated. So why were these systems in an improper position at that low altitude?

Former senior air safety investigator with the US National Transportation Safety Board Greg Feith.

When you watch the video, the aircraft’s pitch attitude — the nose-up angle — appears normal for a climb, yet the airplane is settling into the ground. That could indicate inadequate thrust from both engines. There have been similar cases — such as a Northwest Airlines crash in the U.S. in August 1987 — where improper flap settings during takeoff led the aircraft to lift off briefly, settle into the ground, crash on a highway and it killed the majority of the people. Aircraft performance is going to be a critical aspect of this investigation.

Another key question for investigators will be whether the engines were producing proper thrust. This is a highly computerised and technically advanced aircraft. The crew has to account for the aircraft’s weight. It was also very hot that day which significantly affects both engine and overall aircraft performance. If the crew didn’t factor in the high temperature or didn’t set the correct thrust, the engines might not have generated enough power to keep the plane airborne after liftoff. Investigators will examine the crew’s procedures: how the aircraft was set up for takeoff, especially considering they were likely operating at a heavy weight with 242 people on board and possibly a substantial fuel load. In addition to pilot actions, they’ll be looking closely at mechanical aspects —specifically, whether the engines were generating adequate thrust.

So many different aspects in this investigation based on early information from the video, but there are a lot of things that won’t be known until we get better data.

You mentioned the importance of the appropriate flap setting for takeoff. When is that typically done — before takeoff or during the takeoff roll?

The trailing-edge flaps are typically set by the flight crew before initiating takeoff. These flaps change the shape of the wing to produce more lift at low speeds when deployed. As the aircraft climbs and gains altitude, the crew gradually retracts the flaps so that, by the time the plane reaches cruise altitude, the wings are in their optimal shape for high-speed flight.

You mentioned that the nose of the aircraft appeared to be in a climb attitude even as it was descending. Could that suggest the engines were not producing enough thrust?

Based on the pitch attitude and the fact that the flaps don’t appear to be in the down position, it’s possible the pilots were maintaining a climb attitude but didn’t have enough thrust for the aircraft to actually gain altitude. Given how low they were, they may have been trying to keep the airplane flying by holding that attitude — essentially gliding it forward as far as possible. If there was insufficient thrust, it could have been due to engine rollback or even flameout (read: completely shut down) caused by a fuel issue.

Remains of the Air India plane that crashed moments after taking off from the Ahmedabad airport on Thursday. (Photo: PTI)

What does the altitude of 600 feet tell you about the phase of flight and the pilot’s workload at that moment? 

Typically, in that particular phase of flight, the airplane would have been on the takeoff roll, then rotated, and started its initial climb. Once you establish a positive rate of climb — or rate of ascent — the callout is “positive rate, gear up.” The pilot monitoring calls “positive rate, gear up,” and the flying pilot gives the command and executes it. That’s why it’s curious that at 600 feet, the gear was still down. So is that because of an engine problem — or problems with both engines? Is it a hydraulic problem? Or did the pilot become consumed with handling another issue and simply didn’t get the gear up? There are a number of possible scenarios based on what the pilots may have been dealing with. Takeoff and climb, under normal conditions, aren’t considered a high-workload phase, but they are a phase where both pilots are actively monitoring instruments, engine performance, and aircraft behaviour. If a problem developed that prevented them from retracting the landing gear — or if there was a hydraulic issue and the flaps retracted on their own — that would cause performance problems. Now the pilot flying has to expend significant mental effort to understand what’s happening and figure out the appropriate corrective action.

How much room does a pilot typically have to recover from an emergency at such a low altitude — just seconds after takeoff?

That’s a hard question to answer because it really depends on the nature of the problem. For example, if during takeoff and the initial climb only one engine failed, the airplane is certified to fly on a single engine. So the pilots would just continue what they were doing—climbing and monitoring the aircraft’s performance to reach a safe altitude. You can have an engine failure and still keep flying. Now, if they had a total electrical failure, the airplane would still fly, but it would require a different kind of corrective action. So being at 600 feet is just an altitude—what the pilots should do depends entirely on the specific situation they’re facing.

With more than 8,000 hours of flying time, the captain was highly experienced. Does that make this crash surprising to you?

At this point, there’s no reason to be surprised — because we don’t yet know what the crew was dealing with. Was it an issue with the aircraft that couldn’t be corrected at such a low altitude? Take the example of Jeju Air and the 737: that crew flew through a flock of birds, and both engines rolled back. The pilots, relying on their experience, managed to get the plane back to the airport. But they didn’t complete all the necessary procedures — like lowering the landing gear or properly slowing the aircraft—so it went off the end of the runway. In this case, we simply don’t know what these pilots were facing. Until we have a better understanding, it’s hard to be critical or supportive of the crew’s actions because we don’t know what they were dealing with.

How common are air crashes during the takeoff phase?

For a major air carrier worldwide operating as an airline, takeoff accidents are rare. They do happen and have happened because I’ve investigated them over my career. But they are rare events.

Could you explain why crashes during takeoff are less common than, say, those during landing?

Typically during takeoff, the crew has had time to prepare the airplane while they’re sitting at the gate. They’re configuring it, making sure the flight management computer has all the relevant information necessary to determine the proper engine thrust for takeoff. As they’re heading to the runway, they’re setting the airplane up — putting the flaps at the appropriate setting for the conditions they’re taking off in.

Then, on the takeoff roll, one pilot is flying the airplane, and the other is monitoring what’s going on — airspeed, engine thrust, and everything else. Unless the aircraft encounters something like a flock of birds — like what happened with the “Miracle on the Hudson,” where Sully Sullenberger flew through a flock of geese and lost thrust — that’s a very rare event. Or, if the airplane is taking off in bad weather that causes wind shear or some other issue, that can also affect the aircraft’s ability to fly. But again, those are rare events. Pilots typically don’t take off in dangerous weather. They’re very aware of conditions during takeoff.

Remains of the Air India airplane after it crashed. (Express Photo: Bhupendra Rana)

Now, we tend to see more accidents during landing, because the aircraft is committed to its destination. Weather might have more of an adverse effect on landing performance than on takeoff. Unless something happens that severely impacts the airplane in a way that prevents the pilot from taking corrective action, accidents during takeoff—like this one—are very rare.

We’ve had them before, like Northwest 255 in Detroit, years ago. The pilots didn’t put the flaps down into the proper configuration, so when they tried to take off, the wing didn’t produce enough lift. The airplane, similar to what we see in this video, lifted off briefly and then settled back into the ground—crashing on a highway and, unfortunately, killing everyone onboard except for one person. That happened in 1987 and killed 154 people. But again, that kind of accident with a major commercial airline is extremely rare today.

If you were leading this investigation, what would your top priorities be?

If I were running this investigation, there would be multiple priorities. First, of course, would be recovering the flight data recorder and the cockpit voice recorder. That will give the investigative team the most reliable information to determine the direction the investigation should take — whether it’s pointing to an aircraft issue, a pilot issue, or a combination of both. So that would be a top priority.

Another priority would be collecting as much video evidence as possible — including what’s already circulating online, but also checking for any additional footage from airport security systems. Can we see the airplane at the gate? Can we observe whether it was properly configured while taxiing or during the takeoff roll? That kind of information could be captured by security cameras and would help investigators assess whether they need to focus more on mechanical systems or on flight crew actions and procedures.

And then, of course, there’s the physical wreckage. Examining the accident site for any visible signs of damage or failure can help determine whether there was a mechanical malfunction or or failure of the aircraft.

Given that much of the physical wreckage appears to be completely destroyed, what kind of clues can the accident site still hold to indicate a mechanical failure?

Correct — but that’s the forensic part of the investigative process. You’re looking for components or parts of the airplane that are not in a normal state. For example, if investigators find the trailing-edge wing flaps, they’ll examine the mechanical devices that move them — like the jack screws — to measure how much the flaps were deployed. If one flap is at a five-degree position and the other is at fifteen degrees, that’s an anomaly. The next question is: why did that happen?

In the cockpit, investigators will look at the physical positions of the thrust levers, the flap handle, and the landing gear handle. Then they’ll compare those positions to the data from the flight data recorder. If the recorder shows the landing gear or flaps were in the “up” position, but physically we know the gear was still down, that’s another anomaly.

That’s why the forensic, on-scene work is so important. Even in a crash like this, where the aircraft appears completely destroyed, there’s still a lot to learn. It’s like working a crime scene — sometimes it’s the smallest pieces of evidence that matter most. As the saying goes, the devil is in the details. It’s not always the big things — it’s often the little things that provide the biggest clues.

How long does it typically take for an investigation like this to conclude — or at least to arrive at some conclusive findings?

In this case, the investigative process is really twofold. The first part involves gathering as much information as possible to determine whether there was some kind of deficiency, inadequacy, or failure within the broader system. And when we say “system,” we’re talking not just about the aircraft, but also about policies, procedures, regulations — anything tied to the airline, including training and operational practices.

We also need to identify whether there were any mechanical failures or malfunctions with the aircraft itself. All of these factors can have an adverse impact on flight safety. The goal is to find out what went wrong as early as possible — so if there’s something that needs to be fixed, it can be addressed immediately, while other aircraft are still flying. That’s why investigators push to get actionable information quickly. But the full investigation typically takes between 18 and 24 months.

How quickly do you think investigators can gather actionable information — so that corrective measures can be identified and implemented?

I think if the cockpit voice recorder (CVR) and the flight data recorder (FDR) provide useful information because these boxes, while they are built to be somewhat indestructible, they have failed in the past. I’ve had to deal with that in the past. Assuming we get good data from both the CVR and FDR, that’s a good start. In addition, aircraft constantly transmit data to the ground, so we have access to other repositories as well. I would expect that if there is any kind of issue that has an adverse effect on flight safety,  I would expect the investigators to identify it within a week.

Any final thoughts you’d like to share?

I think right now, with all the media coverage and the chatter on the internet by so-called experts — and all the hypotheticals and theories floating around — it can mislead people, especially passengers who are about to get on an airplane or are making decisions based on what they’ve read online.

That kind of information has to be taken with a high level of caution, because a lot of people are speculating without having any real facts — aside from a video that’s circulating. I’ve read a bunch of stuff online that I know is completely wrong. But to someone who’s untrained or not familiar with aviation, it may sound like fact when it’s actually fiction.

So, people really need to take a step back and allow the investigative process to begin. There’s a team coming from the AAIB in the UK and the NTSB from the US as well. As more information comes out over the next 48 to 72 hours, we’ll hopefully get a better understanding of what took place — whether it was an issue with the airplane, the pilot, or a combination of both.