Category: Flying

This is a follow-up to Landing a CRJ in Toronto. Much has been written about Endeavor’s passion for diversity, inclusion, and equity (consistent with parent company Delta’s passion for DEI) so it seems fair to say that any landing by Endeavor is a “DEI landing”.

Viewed in isolation, a video of the DEI CRJ-900 landing that resulted in a crash/fire/flip doesn’t look that bad.

The approach angle and descent rate doesn’t seem alarming, though maybe there is some vertical acceleration at the very end. If viewed next to a video of an ordinary nothing-bent CRJ-900 landing, though, the abnormality jumps out. A normal landing has a dramatically longer flare and float, with corresponding much lower vertical speed on touching the runway.

Pilots transitioning from little pistons to airliners are admonished to “fly it on” and not try to hold the plane off the runway for as long as they did in their Cessna/Piper/Cirrus days. The ground spoilers on a jet don’t pop up until “weight on wheels” sensors on both main legs are positive. Therefore, a long float and butter-smooth landing chews up a lot more runway than an, um, “positive” landing in which the ground spoilers pop up right at the 1000′ markers. The DEI-enriched Endeavor crew apparently took the “fly it on” mantra too literally.

One other aspect of landing a jet of this size that might not be familiar to pilots with piston experience: the always-present-in-a-piston option to go around by adding power and climbing out doesn’t exist below about 50′. Once the thrust levers are pulled back, there is no procedure for adding power back in and trying to take off again. It might be doable, despite the long spool-up time for the heavy engines, but there is no training in this method. Maybe an airline crew would try this if a fire truck or another aircraft suddenly began to block the runway. Other than that, thrust levers back means a commitment to the landing and it might not be obvious how to fix a co-pilot’s mistakes (though a failure to flare, on the other hand, could be obvious and could be fixed with aft pressure on the yoke while saying “I have the controls”).

Related:

• My visual approach, and Asiana’s (in which a beginner CRJ pilot floats down the runway at La Guardia)

I wrote about my experience landing a Canadair Regional Jet in Toronto (July 2013):

On week after I completed IOE I was assigned to fly with a young recently upgraded captain to Toronto. I had about 75 hours of experience at this point during one month of flying the CRJ. The Tower cleared us to land on runway 33R. I had the plane set up perfectly. We were 3-4 miles from the runway and descending in a stable configuration. Then the Tower controller changed his mind: “Cancel landing clearance. You’re now cleared to land Runway 33L.” This is a shorter runway that starts about 2000′ farther away than 33R and also requires a horizontal sidestep of about 3500′. I would have to add some power and maneuver the airplane to line up with the other runway.

A good CRJ pilot would have added exactly the right amount of thrust so that it wouldn’t be necessary to touch the levers again until 50′ above the ground when it was time to pull them back to idle. How did I handle the situation? I added too much power. Then I took some back out. Then I had to add some back in. Then I finally got us stabilized close to the 500′ above-the-ground minimum altitude that our company rules called for (if not stable at 500′ in visual conditions, go around; if not stable at 1000′ in instrument conditions, go around). After we’d pulled off the runway and cleaned up the plane I said “That was so embarrassing. I feel like I should mail my ATP certificate back to the FAA.” The captain replied with one of the wisest and kindest things that anyone has ever said to me: “Nobody was born knowing how to fly a 53,000 lb. jet.”

Conditions were more challenging today, I’m sure, and the results were worse than even my weak effort (ABC):

Friends have been asking me how this could have happened. flightradar24 has some good info, especially the weather report (METAR):

CYYZ 171900Z 27028G35KT 6SM R24L/3000VP6000FT/U BLSN BKN034 M09/M14 A2993 RMK CU6 SLP149

The same article says that the plane was landing on runway 23 (magnetic heading 237, which is about 227 degrees true heading in that part of the world). The METAR says that the wind was from 270 (true), a 40-degree cross-wind (works out to nearly 20 knots of crosswind). The wind was blowing at 28 knots gusting 35 knots, which is a recipe for bumps. The visibility is said to be 6 statute miles, which is inconsistent with the “runway visual range” (RVR) of only 3000-6000 ft (1/2-1 mile) in blowing snow (BLSN). Maybe the explanation is that the visibility was quite good except near the surface where the strong wind was blowing accumulated snow around. Temperature was -9C. The clouds didn’t begin until 3400′ above the runway and, therefore, instrument flying didn’t play a role in this unfortunate event.

Various American media outlets have been highlighted on X for blaming Donald Trump, which seems far-fetched given that it was a Canadian-built airplane landing at a Canadian airport.

How could the plane flip over? It can’t be wake turbulence from another aircraft because a strong wind will blow the wake turbulence clear of the final approach course and runways.

Based on the detachment of both wings, my guess at this point is that the plane began to slide sideways on the runway, caught on a pavement imperfection, and flipped over as a car might. Maintaining directional control on the runway at higher speeds is done primarily with the rudder (operated by the same pedals that operate the nosewheel steering). In other words, steering is accomplished via an aerodynamic mechanism even if the wheels are rolling on the runway. Beginner pilots are prone to forgot to keep steering with rudder after the wheels touch. They think that the flight is over and now it is time to relax, even though their Cessna or Cirrus is still going more than 60 knots, possibly with a strong tendency to head for a side edge of the runway. Airline crews, of course, will be much less prone to this human frailty, but the CRJ900 lands at around 130 knots and that makes directional control more challenging. Poor visibility from blowing snow certainly wouldn’t help. The only thing that the CRJ has going for it compared to the trainer aircraft in this situation is that a 20-knot crosswind (see above) is a smaller percentage of a 130-knot forward speed than it is of a 60-knot forward speed.

[Update: video has emerged of a hard landing, maybe hard enough to snap off one of the main (under-wing) landing gears, which would certainly start the plane in a sideways direction. Why would the plane come down rapidly? Gusty winds could be a factor. If a strong headwind suddenly shifts to a tailwind, for example, the plane loses a lot of airspeed instantly and, below a certain speed, the aircraft becomes less efficient as it gets slower. (Below the stall speed, the aircraft mostly stops flying and, therefore, will sink like a rock.) Given a long runway, pilots can usually deal with this possibility by choosing to fly at a higher-than-standard approach speed (add half the gust factor is the conventional formula) and, also, the standard approach speed provides a significant margin over stall.]

The flight attendants are today’s heroes, certainly, for getting everyone out!

Our wind limitations from a smaller earlier version of the CRJ:

(It would have been a 27-knot crosswind limitation, I think, given the runway conditions being reported by the control tower. Below the chart there is a note saying that reported gusts are to be ignored in determining whether a limitation will be exceeded.)

Here are some flashcards for the CRJ900 from Endeavor:

Technically, this may have been a “dry” runway and, therefore, the reported crosswind was less than the 32-knot limit. On the other hand, reports might not have matched the reality for directional control.

(One confusing element of our life with the CRJ was that wet runways were actually considered “dry” so long as the runway was grooved.)

One aspect of the DCA Black Hawk-CRJ tragedy that is notable to a civilian pilot is the low reported number of hours of both the pilot and instructor on board, i.e., 500 and 1,000. A civilian helicopter pilot won’t get anywhere near a turbine-powered helicopter until beyond the 1,000-hour mark and that turbine-powered helicopter will be a used single-engine sightseeing machine, not a $20 million Black Hawk in more-challenging air taxi service. The pilot-in-command with 500 hours had been a military aviator for 6 years, which meant that she was flying fewer than 100 hours per year, less than a lot of hobbyists.

The U.S. military seems to start with a “cost is no object” philosophy when it comes to aircraft, e.g., training new pilots in a $6 million (pre-Biden price) twin-engine Eurocopter rather than in a $400,000 (post-Biden price) single-engine Robinson. Once the magnificent machines are delivered, however, the military then seems to decide that they’re too expensive to fly casually. Why not a fleet of Robinson R44s or, if Avgas is too complicated to keep in inventory, turbine-powered Robinson R66s, that would enable Army helicopter pilots to get significant real experience flying helicopters? (Order the Robinsons without the optional SAS/Autopilot so that the Black Hawk pilots get comfortable flying without the crutch of stability augmentation. Don’t subject our military heroes to the challenge of keeping a Robinson R22 under control, though!)

On second thought, when the government operates aircraft it usually manages to spend vastly more than what civilian operators spend. So perhaps it would make more sense to give the military pilots a stipend to use at local flight schools where the retail rental price would be much lower than the military’s cost. Reuters points out that sending migrants via military planes costs perhaps 10X what it would cost to purchase economy-class tickets (even when the military operates the exact same type as an airline, the cost is vastly higher).

A follow-up to What I learned about aircraft paint at Oshkosh…

Let’s start at the Okeechobee, Florida airport (KOBE):

This Skyhawk-with-Hawk might be the way to put a golden retriever on the side of a Cirrus (via wrap):

Back in Stuart, Florida (KSUA), a subtle design that whispers, “this Challenger 600 won’t be available for charter” (KR = Kid Rock, I think):

Shades of blue for a PC-12:

How about cars? Here’s a neighborhood mom’s ride:

Wouldn’t it look better with a light wrap of some kind on the sides?

Tough to think of a way to improve this Miami Dolphins fan’s truck in downtown Abacoa, but maybe an annual wrap with some information about the current or most recent season?

For those who want to save our planet without being associated with the Nazi Elon Musk, the Volkswagen ID.Buzz presents a broad canvas:

Ugly from any angle, if you ask me, and crying out for a wrap:

Once AI is doing all human work I think we’ll have a lot more time, energy, and money to wrap everything in custom designs.

Friends have been asking me about this evening’s crash between a U.S. military Black Hawk helicopter and a Canadair Regional Jet (CRJ) that was on final approach to DCA (Reagan National).

It’s a terrible tragedy, of course, and has led to speculation on X regarding terrorism. A review of the ATC recording shows that there was plenty of room for human error. Because the liveatc.net server is overwhelmed right now, I copied over the relevant recording of DCA Tower. Note that military aircraft communicate via UHF and, therefore, we will hear Tower talk to the Black Hawk, but not the Black Hawk talking to the Tower.

Below is the airport diagram. The Potomac River is to the right and above. Runway 33 begins at the center right of the drawing near the “Elev 10” (10′ above sea level) and “EMAS” (Engineered Materials Arresting System, designed to stop a plane overrunning opposite-direction Runway 15). The runway name of “33” indicates that an aircraft landing on it would be pointing roughly magnetic 330 (333 in this case) or northwest.

At 12:20 Bluestreak 5307, a CRJ-700, checks in and is cleared to land Runway 1 after rejecting an ATC-proposed change to 33 (“unable”). At 12:57 Bluestreak 5342, another CRJ-700, checks in and accepts a modified clearance to land Runway 33 (helps ATC get more departures out). At 13:50, the Tower says that winds are from 330 (northwest) at 15 knots, gusting 25 knots (will be bumpy in a helicopter).

At 15:05 there is some communication with PAT25 (the Black Hawk). At 15:50, Tower tells PAT25 about the CRJ’s lateral and vertical location and also where it is heading (“PAT25 traffic is south of the Woodrow Wilson Bridge a CRJ at 1200′ [landing?] Runway 33”). After an inaudible-to-us reply from the Black Hawk on UHF, the Tower says “visual separation approved” (this approval can be given in Class B airspace only if an aircraft says that it has positively identified another aircraft; we were given this approval every 5 minutes or so when operating our Robinson R44 helicopters in Boston Class B airspace for photo and sightseeing tours over the city; it was necessary because we were within a certain number of miles of the airliners even though we were never anywhere near the approach or departure paths of the jets).

At 17:25, DCA Tower asks PAT25, “Do you have the CRJ in sight?” (in hindsight, would have been much better if Tower had asked “Do you have the CRJ at your 10 o’clock in sight?”) Presumably the Black Hawk pilots answer in the affirmative, having seen or continuing to see what they believe to be the CRJ that ATC is talking about, but we can’t hear this on a recording of the VHF traffic. DCA Tower then instructs the Black Hawk to pass behind the CRJ (might require a slight turn or slowdown).

At 17:47, we hear background conversation in the Tower (a reaction to the crash, perhaps).

At 18:10, American Airlines 3130 is told to go around. The recording for the next few minutes indicates some rough times inside the Tower.

It’s too early to say definitively what caused the crash, of course. However, it seems that there were multiple jets in the air and even multiple CRJs. It is easy to see airplanes, especially airlines, at night, but not necessarily easy to tell a CRJ from an ERJ or a CRJ from an Airbus A319. A two-pilot crew in a Black Hawk would almost surely be able to avoid a crash with an airliner had they seen it more than 1.5 minutes earlier, which they say they did. Thus, the most plausible explanation is that the Black Hawk crew and DCA Tower were talking about two different airliners (i.e., talking past each other).

So… there were some excellent humans with excellent training in the airliner, in the Tower, and in the Black Hawk. Everyone was operating in the most restrictive low-altitude airspace (Class B) that we have in the U.S. and under time-tested rules that have ensured safety despite congestion. At the same time, however, we have the limitations of a natural language (English) and the human brain, which may latch onto and commit to the first plausible airliner that it sees.

A few potentially complicating factors:

• Black Hawk pilots will fly with helmets, which reduce peripheral vision.
• Black Hawk pilots may use night vision goggles (NVGs), which make it easier to see dark stuff on the ground but harder to see brightly lit objects, such as a CRJ in landing configuration. NVGs dramatically reduce peripheral vision (Update: Pete Hegseth says that they were using NVGs)
• Military aircraft sometimes use modern standard ADS-B transponders that transmit x,y,z position, airspeed, and direction, but perhaps not always, and therefore the collision warnings provided by modern avionics might not be triggered
• Airliner Traffic Collision Avoidance System (TCAS) has some inhibitions below 1000′ and below 500′ so as not to distract pilots during landing, so even if the Black Hawk had its ADS-B transponder on the airliner’s avionics might have inhibited a collision warning
• Visual clutter from all of the city lights; it’s easier to pick out airports and aircraft at night in places where there aren’t brightly lit buildings, parking lots, and towers

For those who aren’t regular readers of this blog: I’m an FAA-certificated helicopter instructor as well as a former CRJ pilot for a Delta Airlines subsidiary. Landing and taking off at DCA were part of the Delta job. I also teach an aeronautical engineering class at MIT. I have never flown a Black Hawk, but I have trained experienced Army Black Hawk pilots to fly the Robinson R44. I have spent hundreds of hours in Class B airspace, the same kind of airspace that surrounds DCA, in helicopters while airliners were landing at Boston’s Logan Airport (it was very rare for us to need to cross the final approach course, for the jets, though; we typically avoided airliners by flying over the top of the airport at 1,500′ or above or by staying as low as 300′ above the ground when underneath the final approach course to an active runway).

What could have prevented the accident?

First, let’s reflect on the fact that last night’s situation was a common one for the past 60 years or so and there weren’t any previous accidents. So the interaction among river-following helicopters and landing/departing airliners wasn’t obviously unsafe. On the other hand, safety rested on human excellence and vigilance and none of us can be vigilant 24/7.

The easiest way to have prevented the accident would have been to eliminate the Army aviation unit involved in favor of Singapore-style congestion pricing for surface transport in the D.C. area. The aviation unit exists primarily to ferry around senior military personnel who don’t want to sit in traffic like the peasants must. As D.C. traffic has intensified over the decades and helicopters have become safer (twin engines; everything precision-machined; two pilots) more VIPs have decided that they’d rather get around by helicopter than by car. But let’s assume for this post that congestion pricing can’t happen and military brass won’t use Zoom and, therefore, what is essentially an air taxi operation is required.

Winston Churchill defined a fanatic as someone who won’t change his mind and won’t change the subject. That’s certainly me when it comes to the crying shame of modern software capabilities not making it into the cockpit or onto the workstations of air traffic controllers. Our desire for FAA-certified perfection makes it prohibitively expensive to put the kind of intelligence that we expect from a $500 drone into a $30 million airliner or $20 million Black Hawk. Imagine if the Black Hawk had an onboard assistant that could have said to the pilots “There’s an airliner at your 10 o’clock that you’ll hit if you don’t slow down to 50 knots.” That would, presumably, have redirected their attention away from whatever airliner they thought they were supposed to focus on and prevented the accident. All of the data necessary for such an assistant are available in any non-antique aircraft: position, velocity, track over the ground, position and velocity of other aircraft (broadcast via ADS-B, which is its own disappointment). The only thing that was missing on the Black Hawk was a $1,000 computer wired to the audio panel and a straightforward-to-write-but-ruinously-expensive-to-certify computer program. Similarly, ATC could have benefitted from a program that spoke “It doesn’t seem as though the Black Hawk is doing anything to avoid the CRJ, despite your instruction.” The controllers will, no doubt, share some blame for not noticing an alert on their screens, but these types of alerts are too common and insufficiently specific for humans to deal with reliably hour after hour day after day.

(Check out Beacon AI for an example of a company that is trying to deliver smarter in-flight software to deal with the fact that we demand ever higher levels of safety in a world where humans aren’t getting smarter or more vigilant. Beacon AI has some military contracts and things may move faster in that domain because the military is not bound by FAA certification rules.)

What about “Trump blames DEI for weakening FAA in aftermath of Reagan National plane crash” (The Hill)? Although the FAA has invested heavily in DEI, I don’t think that was the proximate cause of this accident. There will inevitably be a distribution of ability among air traffic controllers. DEI-based hiring will sadly increase the number of those with lower ability, just as in any other field of endeavor. On the other hand, there are only 37 Class B airports in the U.S. out of roughly 500 airports with control towers. Thus, these 7 percent most-critical airports are going to draw their tower controllers mostly from the top 7 percent of all tower controllers. A mediocre or low-performing controller can be parked at an out-of-the-way airport that has just a fraction of KDCA’s roughly 800 operations per day (Westover, Massachusetts is a civilian-military airport that has a control tower and only about 50 operations per day, for example). In 1996, the FAA was trying to bend its rules to favor women (report). In 1999, the FAA was working on bending the rules to favor “African-Americans” (report). See also “Obama-era FAA hiring rules place diversity ahead of airline safety” (Fox News, 2018) and this undated recruiting video. Perhaps it would be fair to blame the FAA’s focus on DEI as a factor in slowing the agency’s ability to adopt innovation simply because time and money spent on DEI can’t be spent on improving operations.

Some previous articles that I’ve written about the negative impact on safety of the financial and calendar obstacles to certification (perfection is the goal and it becomes the enemy of near-perfect solutions that would be huge safety improvements):

• Boeing 737 MAX crash and the rejection of ridiculous data
• Certification process for the 737 MAX silent gradual pusher system
• Kobe Bryant crash: NTSB says that it was all the pilot’s fault

The classic paper on our limits as humans: “Gorillas in our midst: sustained inattentional blindness for dynamic events” (1999). The video that was part of the experiment (see if you can spot the gorilla, which roughly 50 percent of the experiment subjects (super smart undergrads?) missed):

Next steps for the NTSB and FAA

A reporter asked me what happens next. My answer:

• I don’t think that there will be anything interesting to learn from the cockpit voice recorders and flight data recorders other than, perhaps, a precise altitude for the crash (Update: the CRJ’s black box places the jet at 325′ MSL (NBC).)
• the NTSB and FAA will pull the tapes (maybe they’re still actual tapes) from DCA Tower so that they can can hear both the UHF (Black Hawk transmissions) and VFR communications
• they’ll look at whether the Black