The VoePass ATR-72 crash in Brazil was terrifying to watch. Our own resident commercial pilot, 121Pilot, has an analysis of what might have led to a plane falling out of the sky.
Shocking Imagery
If you’re reading this, you’re likely already aware of the accident involving VoePass ATR-72 operating between Cascavel and Sao Paulo in Brazil.
If you haven’t seen the videos that were captured of the flight’s final moments I would suggest going here first where you can see them. They capture something I’ve never seen before and I doubt many others have. The image of an airliner in a flat spin moments before impact. I don’t want to attempt to analyze the videos in terms of nose position and bank angle too deeply, but the central feature you see in both is the aircraft descending in a fairly flat attitude with little to no horizontal movement as it spins around its axis.
Data Points
Looking at the Flight Radar 24 data which is also linked in Kyle’s article, I notice that throughout the flight there are significant variations in the reported ground speeds. Look at the picture below and you can see that the recorded ground speeds are all over the chart which means using this data does come with risks.
The data we have shows the Barometric Altitude, Ground Speed, and Track. The Barometric Altitude should reflect the altitude displayed on the aircraft’s altimeter to the pilots. Ground Speed is speed over the ground which is not airspeed. If an aircraft had a true airspeed of 300 Kts into a 100 Kt headwind its ground speed would be 200 Kts. Track is the course over the ground which is different from the aircraft’s magnetic heading. For example when you see planes landing into a strong crosswind the aircraft’s heading will not match the runway heading but the aircrafts track would as the plane is crabbed into the wind. Keep those differences in mind as you look at the data.
That being said, we know the flight was being operated in icing conditions and if the aircraft was accumulating ice we might expect to see a loss of speed which is something I do see in the data starting around 1607 UTC and the general trend of the graph starts to get lower. The reported altitude readout seems very consistent until the final minute of the flight where it drops rapidly from 17,000 feet to the last recorded value of 4,100 feet with very low ground speeds. We also see that the aircraft’s track is very consistent right up until the final minute where we see a significant change in track from 049 to 061 while the altitude is still showing 17,000’ and then 068 as we see the first indication of a descent. Here’s the data leading to 1621 UTC.
| Altitude | Ground Speed | Track |
|---|---|---|
| 17,000’ | 234 Kts | 61 |
| 16,925’ | 200 Kts | 68 |
| 16,850’ | 143 Kts | 75 |
| 16,725’ | 200 Kts | 76 |
| 16,750’ | 219 Kts | 75 |
| 17,200’ | 130 Kts | 62 |
| 15,500’ | 56 Kts | 121 |
| 14,800’ | 102 Kts | 110 |
| 14,125’ | 113 Kts | 101 |
| 12,125’ | 35 Kts | 252 |
| 11,675’ | 78 Kts | 242 |
| 11,075’ | 81 Kts | 236 |
| 10,625’ | 87 Kts | 221 |
| 10,025’ | 93 Kts | 207 |
| 9,475’ | 77 Kts | 172 |
| 7,600’ | 12 Kts | 108 |
| 7,325’ | 46 Kts | 89 |
| 6,575’ | 13 Kts | 89 |
There are a few more data points, but I think the trend is pretty clear here. Something else I noticed in the videos was a fluttering helicopter like sound that accompanies the aircraft in its descent. To me, this signals that the power on both engines was at or near idle up until impact.
The aircraft involved is an ATR-72-500 which is a high wing T-tailed turbo prop. The T-tail configuration is important because a T-tail, if stalled, is susceptible to entering a deep stall that may be unrecoverable due to a lack of airflow over the tail.
Some caveats seem in order. I’m an airline pilot not a test pilot and I don’t have a degree in aerodynamics, nor am I an expert on the certification requirements for airliners. We are also very early in this event with limited data to work on and this isn’t an airplane I’ve flown personally.
Looking at the data, I’m struck by the fact the aircraft track over the ground suddenly changed just before its descent began. The speed data isn’t reliable enough to draw any major conclusions, however the sudden drop from 200 to 143 could be an indication of a stall entry. Also, I noticed that we see the speed come back up just before a sudden jump up to 17,200’ and a drop of speed back to 130 Kts just before the data gets really ugly with the next data point being at 15,500’ a ground speed of only 56 Kts and track over the ground of 121 degrees. The continued rapid descent from that point along with the low-speed readings and significant track variations are telling me that we’ve just seen the entry into the spin at that point.
Bear in mind that airliners are neither designed or tested to be recovered from spins. It should, of course, be possible and in my research, I found one report of a 737 that was spun and recovered but that’s a part of the flight envelope we aren’t trained for and should never be in. Just getting into a spin would potentially require control inputs that you should never make and would also require ignoring the stick shaker and pusher or overriding those devices along with potentially very aggressive rudder inputs.
For those that don’t know, a stick shaker is a device that will vibrate or shake the control stick to warn a pilot of an impending stall. Responding immediately to this warning and how we are to respond is a central feature of all airline pilot training programs. The stick pusher is essentially a warning of last resort as a ram will fire shoving the stick forward and forcing the nose down to prevent a stall. That being said, as the Colgan air crash sadly showed it is possible for a pilot to still stall the aircraft despite these safety features.
Normal spin recovery procedure is for the pilot to reduce the power to idle, release the controls, and apply full opposite rudder to the direction of rotation. This procedure should work in a normal spin in an ATR-72, but you’re a test pilot at that point in the unknown. What’s even more unknown is recovering it from a flat spin, especially since it’s a T-tail and those are difficult to impossible to recover from a fully developed stall.
Also bear in mind that airline pilots don’t practice and aren’t trained in spin recovery even with the enhanced unusual attitude recoveries we practice in the simulator. To get into a spin requires two things. First, the aircraft has to be stalled and second, we need yaw or a rotation around the aircrafts center. Think of a model aircraft sitting level on the head of a pencil and you spin it around while keeping it perfectly flat. That’s yaw. Where that yaw came from right now is a complete unknown. It could have been rudder input. It could have been asymmetric thrust caused by one engine being choked from ice while the other kept making power. We just don’t know.
But what we do know is that somehow the aircraft was stalled and while it was stalled it began to rotate entering a spin. That spin likely contained aggravating factors that moved it from a normal spin (which in a general aviation aircraft would be about 45 deg nose down) to a flat spin (10-20 deg nose down). A fully developed flat spin can be difficult to impossible to recover so at that point there may have been nothing at all the crew could have done to recover the aircraft.
This is also a very different accident sequence than we saw with American Eagle flight 4184 that crashed in Roselawn Indiana after a loss of control in icing conditions. In that accident the aircraft did not enter a spin but ended up nearly inverted hitting the ground at high speed and in a nose low attitude.
Given how little we know there are no true conclusions or lessons to be learned right now. Commercial aviation remains incredibly safe. But, the accident today and the lives lost are a sobering reminder of the potential for a flight to rapidly move from boring and routine to terrifying in a matter of moments. As I wrote about recently, all of us in the cockpit need to remain vigilant even when things seem dull and routine.
Questions?
If you have questions about flying or airline operations you can email me at ask121pilot@yahoo.com. I will then respond in my Ask Your Captain column. If you have questions or comments about this specific story, please leave a comment below.
Thank you for your additional insights on this tragic accident. The video takes your breath away as one is keenly aware of the outcome while observing what looks like a flat stall. I believe there have been prior accidents related to ATRs in icing conditions. American Eagle flight 4184, (31 Oct 1994), comes to mind. That was another ATR-72 accident with the loss of life of 68 folks flying from IND to ORD. The plane encountered icing conditions while in a holding pattern for O’hare. As you said, it is too early to jump to conclusions as time and additional analysis will tell if this is a related event or not. RIP to the souls that were onboard and healing thoughts to their loved ones, who’s lives changed in an instant.
In Brazil , the I ride a dugout canoe downriver , or an outboard boston whaler upriver . The map clearly shows rivers going the traveler’s way .
Moral of the store is avoid this aircraft if possible.
This is not the first time an ATR in icing has proven to be a lethal combination as an analysis of previous crashes involving this aircraft will illustrate. Personally I go out of my way to avoid flying on this aircraft type.
So there have been several comments that suggest this aircraft is unsafe in icing and should be avoided.
The American Eagle Crash revealed a design defect that has since been corrected and there have been no repeats of the issue that led to that hull loss. I found two other accidents that happened as a result of an icing encounter but both of those had crews that did not apply and follow proper procedures.
Any aircraft has the potential to bite if you’re not flying it properly. Personally right now I see no reason to think the ATR is an unsafe aircraft that should be avoided.
Dunno, but as a private pilot with several airline pilot friends. seems like maybe the aircraft flew into known icing conditions and failed to recognize it or just do nothing about it. Looks a lot like a previous ATR-72 crash in the U.S.
Simon at AV Herald has since reported that the speed fluctuations that these sites have been showing have been corrected and they have admitted to an error in the data. It seems the plane was flying at the same speed up until the last seconds before.
Here is Simon’s statement at AV Herald:
“The oscillating ground speeds by some radar tracking website, widely discussed here and on the Internet, have been identified erroneous on this as well as previous flights of this aircraft. That website has admitted to that and published a speed readout that shows no oscillations and no problems whatsoever until the very end.”
Also, it’s now being reported that the plane suffered a tail strike in March with an experienced structural damage. It had just returned to service a month ago after undergoing repairs but issues were being reported since then.