Delta Flight 191
August 2, 2015 7:49 AM   Subscribe

30 years ago today (August 2nd, 1985), Delta Flight 191, a Lockheed L-1011 TriStar from Fort Lauderdale, bound for Los Angeles, crashed over a mile short of the runway at the Dallas/Fort Worth International Airport. The plane was caught in an unexpected weather event called a microburst, which causes a sudden downdraft of air. During the struggle to regain control, the plane careened across a highway north of the airport, struck a car, and skidded into two large water tanks, which broke it apart. (Flightpath/Crash diagram) The crash killed 136 of the 152 passengers and 11 crew on board, and the driver of the car. (Cabin seat diagram showing injuries and survivors)

Amazingly, 27 people survived, including flight attendant Vicki Chavis, who described the feeling of the crash: "I was facing aft and I felt us going forwards. The grass and sky and things were going by in the opposite direction. My arms were completely flailing, my legs were flailing. I had no control over my body,'' the flight attendant recalled. ...the out-of control jetliner struck a water tower. As she briefly turned her head, a flame rushing from the front of the cabin singed her hair. ...By then the aircraft had broken in two, leaving the rear cabin section largely intact and separate from the rest of the plane. And there was momentary quiet, she recalled. "We stopped. I was hanging upside down from my seat belt ... The whole fuselage on the left side of the airplane was gone."

The National Transportation Safety Board investigation determined that the crash resulted from the flight crew's decision to fly through a thunderstorm, the lack of procedures and training to avoid or escape microbursts, and the lack of hazard information on wind shear. Many improvements in air safety were made as a result of this accident.

Crash animation and Cockpit Voice Recorder audio with transcript

Similiar crash animation and CVR from courtroom proceedings (includes airspeed, wind vector, etc. info)

Live coverage from the time with Ted Koppel

Additional stories of survivors

1989 movie "Fire and Rain"

Memorial dedication
posted by agregoli (32 comments total) 29 users marked this as a favorite

This was nonstop news in the Metroplex for a long time after the crash, and that was before the 24-hour news cycle. The charred and broken fuselage sat in place for weeks (?) while they conducted the investigation. It was on the north side of the airport, and we lived an hour away from the south side, but we took a trip up there one weekend to drive by it. That's an amazing thing when you have a dad who hates to drive more than 10 miles away from home, and especially into the bigger cities, but he was an airplane and aerospace engineer, so there you go. I don't know whether that was a tourist outing or a pilgrimage in his mind. Could have been either.

The nonstop news coverage and seeing that wreckage in person was one of two things in my life that have made me an incredibly nervous flier. (The other was a scary emergency landing, also at DFW.)

Airplanes are amazing and all, but their actual fragility becomes evident once you see a huge piece of a broken one just sitting there.
posted by mudpuppie at 8:41 AM on August 2, 2015 [8 favorites]

We're also in DFW and I mostly remember that chunk of fuselage (it became a de facto news graphic for the crash) and my mom rushing out to donate blood because she was AB- and wanted to make sure any survivors who needed that had enough. Also the phrase "wind shear" was everywhere, and I think about that sometimes when I fly. For whatever reason, it hasn't made me more afraid, though.
posted by emjaybee at 8:51 AM on August 2, 2015

The rules of air safety are written in blood, but conversely once that blood is spilled, a fair bit of effort goes into writing those rules and making sure that whatever happened won't happen again. It's a lesson that many other parts of society are going to have to learn.
posted by wotsac at 8:55 AM on August 2, 2015 [16 favorites]

I commuted on that stretch of 114 for about 7 years and frequently thought of that accident as planes landed over me. There's a spot near there where they (at least up until 5-6 years ago) did disaster drills every year or two.
posted by Lyn Never at 8:55 AM on August 2, 2015

The nonstop news coverage and seeing that wreckage in person was one of two things in my life that have made me an incredibly nervous flier. (The other was a scary emergency landing, also at DFW.)

Airplanes are amazing and all, but their actual fragility becomes evident once you see a huge piece of a broken one just sitting there.

It always boggles my mind when you land and you see just how little of a wing is solid.

But we've been flying as a species for over 100 years. We've learned so much about the mechanics of flight, how to respect nature, how to survive when things do go wrong. There hasn't been a fatal US commercial flight since 2009. We're up to 1 in 45 million on the odds of dying in a place crash. We're orders of magnitude better than we ever have been at keeping people safe in the air thanks to learning from accidents like these.
posted by Talez at 8:57 AM on August 2, 2015 [11 favorites]

As plane crash porn goes, Mayday is pretty mild, although they ratcheted back the pornographic aspect of it after season 5. I love the stories of trying to understand a big, awful, multifaceted problem.
posted by wotsac at 9:00 AM on August 2, 2015 [2 favorites]

Also the phrase "wind shear" was everywhere

Absolutely! It seemed like a brand new term at the time, and suddenly it was ubiquitous.
posted by mudpuppie at 9:01 AM on August 2, 2015

I had to pick someone up from DFW the day after the crash. This was back in the days when you could go in and meet people at the gate. I remember standing at the window and watching their plane fly in over the wreckage area. It was very quiet in the airport.
posted by tamitang at 9:06 AM on August 2, 2015 [1 favorite]

I was a kid when this happened, and when I finally moved away from DFW, I was shocked at how few people were worried about wind shear.
posted by unknowncommand at 9:07 AM on August 2, 2015

From the diagram showing injuries and survivors, there was only two people who suffered no injuries at all. I could not imagine what that would be like in an environment that had so much death and destruction while threading a needle with my own life. And then just a handful of minor injuries, with the rest fatal or serious.
posted by SpacemanStix at 9:16 AM on August 2, 2015 [1 favorite]

Wind shear wrecked a *lot* of planes and killed a *lot* of people. DL191 was the one that we finally had all the parts in place to see what exactly was happening and why it was dropping planes out of the sky.

And a big part of the reason?

Because we'd trained pilots that way!

The pilots are flying an approach. They're flying a glideslope. If they're on an ILS approach, they're flying along a set of radio beams, defining a glideslope down. If they're on a visual approach, they're usually using a landing aid, like a PAPI to make sure that they're on that glideslope.

Commercial and private pilots use power to control that glide slope. If they're flying too high above it, they'll reduce power slightly, the plane slows, flies a steeper descent, and settles back onto the glideslope. If they dip below the glideslope, they add power, the plane flies a shallower path, or with enough power, climbs, and the climb back to the glideslope. Now you know why the engine note changes as they approach.

Now, those downbursts. First, they encounter a big headwind. That, in effect, increases the airspeed over the wings. That makes the plane climb -- climb over the glideslope. The pilots, by training, reduce power to get back on. Then the wind shear happens.

We should define that term. Wind shear is a sudden change in wind direction, that's all. Normally, you see this changing altitude. But in a microburst, you see this moving horizontally. So, they fly through the microburst, and instead of that headwind, they see a tail wind -- which the wings see as a big reduction in airspeed. They lose lift.

But what did the pilots just do? They just pulled back the power!

Look where the plane was just put -- big tail wind and a reduction of power. This causes a massive loss in lift. The plane falls from the sky. The pilots see this, and push the throttles forward, but it takes time for an engine to spool up. If it was a big headwind, causing a *big* lift, they pulled the throttles way back. It will take a lot of time to get that thrust up.

They're landing. They don't have time. The ground hits them before the engines wind up.

When this happens at altitude, it's a nothing burger. The throttles aren't being moved, you're not really following a glideslope. The ground is really far away at FL300. But when you're on a landing approach, the ground is right there.

DL191 happens, and we finally see the entire chain of events that leads the pilots to do what their training tells them to do, and what their training is telling them to do is crash the plane.

Now. We have better sensors that detect wind shear. A sudden climb on approach like that? You don't back the throttles. You *advance* the throttles and go around. Indeed, many modern planes can detect windshear and they'll literally call it out as a voice alert. "WIND SHEAR WIND SHEAR" -- the pilots will either just fly through it if they're far enough from the airport, or abort the landing and go around if they're too close.

And, since DL191? Landing accidents due to windshear have plummeted. Because we know now. Pilots are taught what that sudden increasing in apparent airspeed means on landing. They know -- now -- don't pull back the throttles. Better to try again later.

That flight number -- 191 -- is not the first time that the pilots, doing what they were taught, lost the plane. AA191, also a trijet, a DC-10-10, went down on May 25, 1979 after takeoff at Chicago O'Hare. It lost the left engine on takeoff, and by lost, we mean "it fell off, flipped over the left wing, and landed on the runway." Maintenance workers were using a shortcut method to remove and replace engines, one that damaged the engine mounting pins and the failed.

Per their training, the pilots reduced their speed to just over stall and tried to fly back to ORD to land. Instead, the aircraft snap rolled over and crashed. It remains the deadliest single aircraft accident in the United States.

What happened? When the engine left, it damaged the hydraulics controlling the leading edge slats on the left wing. When the pilots reconfigured for landing, the flaps on both wings went down, but only the slats on the right wing went down. This mean that the left wing had a much higher stall speed (190kts vs. 160kts) than the right wing. As they slowed down -- per their own training, drilled into them through countless simulator hours -- the left wing stalled and lost lift, and they fell. The ground was far too close.

Afterwards, it became clear -- if they kept the speed up? They get back safely. Pilots easily fly that plane back to a landing -- at 195kts.

Now? The rule is *don't* just drop your speed. Keep it up until you understand the state up the plane, reduce it *slowly*, and watch for asymmetrical airfoil conditions. You can catch them if you're ready for them. They weren't, and they paid for it.

Two planes. Over 350 gone. And both times, if the pilots had know what we know now, they would have landed safely.

We've learned a great deal from 191. We paid a great deal learning it.
posted by eriko at 9:18 AM on August 2, 2015 [144 favorites]

Wasn't there another terrible crash at that same airport a few years later?
posted by thelonius at 9:31 AM on August 2, 2015

Per Wikipedia, the late 80s were pretty hard on DFW, with four commercial flight accidents at the airport, two of them resulting in fatalities.
posted by ardgedee at 9:35 AM on August 2, 2015 [1 favorite]

I remember this very well -- I was visiting friends that week (and had lived in Dallas in 83-84) and my uncle worked at American at DFW. When I lived there, I drove over to DFW to visit him periodically and, also, for some reason, often just drove over there on LBJ from where I lived in north Dallas. I was quite familiar with the area and my friends I drove over there the next morning. I can quite vividly recall even today exactly what it looked like -- you could see it right there from 114. (Unless I'm misremembering something and 114 wasn't opened the next morning. But I think it was.)

I was only twenty years old and this was the first airline crash that really made an impression on me -- seeing the debris in person is much different than seeing it on television. And I recall learning about the crash in a lot of detail -- I'm not certain how. Newspapers and magazine reporting, I guess. What I recall is that at least initially there was a lot of discussion about wind shear in terms of this being exceptional and unanticipated.
posted by Ivan Fyodorovich at 9:35 AM on August 2, 2015

AA191, also a trijet, a DC-10-10, went down on May 25, 1979 after takeoff at Chicago O'Hare.

That crash was what I thought this FPP was about at first; I was living in Chicago at the time, having just taken my first commercial flight a couple of years before (from O'Hare to DFW, in fact), and having lost my own parents in a plane crash when I was a kid. Needless to say, AA191 kind of freaked my shit out, and it would be more than a decade until I flew again.
posted by Halloween Jack at 9:37 AM on August 2, 2015 [2 favorites]

That crash was a big deal in Chicago for years. I didn't move there until '87 and it was still coming up on a regular basis in the news.
posted by wotsac at 9:51 AM on August 2, 2015

This is the one I was thinking of. It horrified me because I was flying a day or two later, and because of the shocking and inexcusable incompetence that caused it.
posted by thelonius at 10:01 AM on August 2, 2015

That crash was a big deal in Chicago for years.

It was a miracle that it only killed who it killed. Look at a map. It missed a mobile home park by a few hundred feet, it missed the Des Plaines Oasis on the (then) Northwest Tollway by not that much more. Hit either of those and the toll would have been much higher.

As it was, it was in a very visible spot, between the tollway and Touhy Avenue. I remember that plume of smoke, actually, being about 13 at the time. The A/C was pretty much fully loaded with fuel.
posted by eriko at 10:01 AM on August 2, 2015 [2 favorites]

eriko, what do you mean by this statement (re: Chicago accident):

Afterwards, it became clear -- if they kept the speed up? They get back safely. Pilots easily fly that plane back to a landing -- at 195kts.

Do you mean that in simulations, pilots have been able to fly that plane back to safety? Because if so, wow. Without an engine? Amazing.

The smoke from the Chicago accident was visible from my high school fifteen miles to the south. To this day, when I hear "Touhy", I think AA 191.
posted by SuperSquirrel at 10:19 AM on August 2, 2015

This is the one I was thinking of.

Yeah -- another one where "Pilot error" means "don't stop digging." Yes, not a sterile cockpit (pilot error, cause #1) Yes, failure to run the checklist properly (pilot error, cause #2.) Yes, failure to configure the plane properly for takeoff by lowering flaps/slates (pilot error #3, proximate cause off accident.)

A lot of agencies would have just signed the report right there. The NTSB has a lot of rules of thumbs, and one of them is "never stop at Pilot Error." They kept digging, because on a 727, there's a system that, if you advance the throttles to take off power and the plane isn't configured for takeoff, which includes the flaps/slats being extended, an alarm goes off. That should have completely mitigated all three of the pilot errors above -- abort takeoff, and this changes from a multiple fatality and hull loss accident to a paperwork incident.

Except: that system wasn't working. (Mechanical issue, cause #4.) Redundancies and safety systems are important. They are also useless when they are broken.

Which is why we now have things like the MEL (Minimum Equipment List) which details what *has* to be working for a flight to be made, and what steps have to be taken if something isn't working. There are multiple MELs, depending on the length of the flight, how far you are from a divert airport that can handle you, and if you're over water or not. So, if it breaks, you look at the MEL, and it tells you if you can fly, what you do if you can, and if you can't, you either fix it or you get another plane or cancel the flight.

And DL1141 was one of the accidents that taught us the importance of this.

We covered something similar in the SL-1 reactor thread. In engineering (and this is really just another engineering problem in a lot of ways) we, unfortunately, often really only learn after we screw up horribly. We learned not to make rails and boilers of iron only after those boilers blew up and those rails failed. We learned the importance of not running BWRs in edge case conditions after SL-1 -- they're very stable otherwise. We learned that assuming a takeoff configuration alarm would catch an out-of-configuration error is a bad assumption if we don't check if the alarm is working.

And so on. Time and time again, we think we have it, and physics proves that nope, we missed something. Physics is fair -- the laws are set, it doesn't cheat, it doesn't change them on the fly. They are there, they are defined.

But physics is harsh. If you assume X, and Y is true, physics will follow Y -- and if that means the plane crashes? The plane crashes.
posted by eriko at 10:24 AM on August 2, 2015 [30 favorites]

Do you mean that in simulations, pilots have been able to fly that plane back to safety? Because if so, wow. Without an engine? Amazing.

Yes. All commercial airliners are built to fly the one engine *gone* case from takeoff to landing. The DC-10 had three engines, it was built to fly on two engines. If the pilots had known the left wing would stall at about 190kts, and kept their airspeed above that, they could have landed the plane, and pilots in simulators knowing that did land the plane. They would have needed to go dump fuel, they were massively over landing weight having just took off, and they would have had to have landed at a higher than normal speed. But ORD at the time had a 13000 foot runway*, so they had plenty of room to roll.

Airplane engines used to be vastly less reliable than they are now** -- one reason that trijets like the DC-10, 727 and L-1011 were built. Nowadays, the only time you see more than two engines is when the plane is simply so large than the engines would be improbably large if you built it as a twinjet with the "one engine out rule" -- so the 747, A340 and the A380 still have four. Pretty much everything else, except a couple of oddballs, uses two engines, and is perfectly capable of flying on one of them.

Of course, with an engine out, you *fly to the nearest airport and land.* Because you lose that engine, and things get much harder. Note -- not impossible. There have been cases where both engines died and everybody lived, the most famous being The Gimli Glider.

* Well, it still does. At the time, it 13R/32L, that's now 9886' long. What was then 9R/27L and is now 10L/28R has been extended and is now 13,000 feet long. This is all part of the ORD modernization project to change ORD from three pairs of crossing runways to six parallel runways. ORD is now fundamentally a 4 parallel runway airport, though two pairs of crossing runways still exist, that will change this October when 10R/28L opens and the 14/32 pair close. Then ORD will be 5 parallel, with a crosswinds pair for unusually strong SW/NE winds, which crop up every so often.

The final stage of the runway project will be to extend 27L/9R to about 12,000 feet, and build 27/9C, and finish out the interconnecting taxiways which will be the 6th runway and complete the project.

Unknown is if the West Terminal project, which was considered at the start, will happen. The current domestic terminal space is adequate. The International terminal needs more space, but that can be expanded.

** Thanks to its four Wright R-3350 Turbo-Compound radials, the Lockheed Super Constellation was often called the world's prettiest three engined aircraft. The PRTs were officially called the "Power Recovery Turbines", the mechanics called them the "Parts Recovery Turbines", because when the engines lurched themselves and spat out valves, that's usually where you found them.

Nowadays? In flight engine failures on commercial aircraft are rare enough that they'll make the local news if they land at the local airport, or national if they happen on intercontinental aircraft and they have to divert.
posted by eriko at 10:46 AM on August 2, 2015 [35 favorites]

eriko that's all so fascinating, I could listen to you talking about this for hours.
posted by Hazelsmrf at 1:58 PM on August 2, 2015 [5 favorites]

It's always a treat when eriko weighs in on a thread, and I can usually identify his responses even before I get to his byline.
posted by bshort at 2:31 PM on August 2, 2015 [4 favorites]

I hadn't thought hard about the relative fortune of it crashing where it did. These days I drive by the spot regularly when I'm in town, since my office is nearby on Touhy, and I'm either staying in an Elk Grove motel, or headed for the Western burbs.
posted by wotsac at 3:58 PM on August 2, 2015

Wow. All these years I thought the cause of the AA191 crash was completely mechanical; that when the engine fell off it locked some vital mechanism, resulting in a roll the crew couldn't maneuver out of. Thanks for the knowledge drop, eriko!
posted by offalark at 11:44 PM on August 2, 2015 [1 favorite]

Airplanes are amazing and all, but their actual fragility becomes evident once you see a huge piece of a broken one just sitting there.

cars are amazing too. they kill a fuckton more people.
posted by 7segment at 2:09 AM on August 3, 2015

Yes. All commercial airliners are built to fly the one engine *gone* case from takeoff to landing.

AFAIK, most (all?) multi-engine aircraft are designed to survive an engine failure even before takeoff. It's the most common time for an engine to fail, and you very specifically do not want to hit whatever is at the end of the runway.
posted by schmod at 5:42 AM on August 3, 2015

AFAIK, most (all?) multi-engine aircraft are designed to survive an engine failure even before takeoff.

During the roll, the Pilot Not Flying will be calling out various airspeeds that you've calculated. The important one here is V1. V1 is the speed that you'll be at when cannot stop safely with the amount of runway you have left.

When you hit V1, no matter what goes wrong, you takeoff and deal with it in the air. Before V1, if you get any number of issues, you stop.

V2 is Takeoff Safety Speed. That's the airspeed you can takeoff at having lost an engine. On a commercial airliner, V2 has to be lower than V1 -- that is, you have to be able to takeoff before you absolutely must take off. If V1 is slower than V2, you can get into a nasty place where you hit V1, then an engine fails and you're not going to reach V2 before you run out of runway, but you can't stop in time because you don't have enough runway and FLAMING DEATH AND DESTRUCTION IS YOURS. Much paperwork. Lots death. Wow.

V2min is absolute minimum takeoff speed. That's "crawl into the air speed." You hate this because this is "inches a second" climb speed, but sometimes, that's what you have to do. Below that, you simply cannot get off the ground.

Ideally, you don't lose an engine and you wait until VLOF, Lift off speed, which is well above all these minimums and the plane basically leaps into the air and climbs rapidly, which gets it out of the way of anything that might be at the runway end.

There are a lot of V speeds. Wikipedia has a sample of them but the exact ones you have depend on the plane you fly. There's no VFE if you don't have flaps, because if you don't have flaps, you can't extend them, thus there's no maximum flap extended speed.

But all planes have a few. They all have VS, Stall Speed, Vref, reference landing speed, and the one you really want to pay attention to, VNE, Never Exceed Airspeed. Go faster than this, and bad things will happen, usually involving controls not working, occasionally because things fell off. Bad. Very bad.

Part of the preflight is making sure you know the fixed ones and calculating the ones that change. V1 changes because it's dependent on several things -- how much lift you have, and of course, how long the runway is. So, you figure that out, figure out where you're starting from, add a fudge factor, and that gives you the number. The Pilot Flying flies the takeoff, the Pilot Not Flying watches the checklists and various instruments, and calls out V1, and the Pilot Flying knows at that point *not* to reject the takeoff, no matter what.

And because they've made sure that V2 is lower than V1, if the thing that just happened was "Fire, Right Engine", they know they have the power to take off, even if they're past V1. They will take off, make sure it's a real fire, then activate the fire systems on that engine and go back.

They do it that way, because one time, a pilot called out "Fire right engine" and the other pilot said "Fire, right engine" and pulled the fire bottle on the right engine. This put the right engine out.

Problem: The fire was in the left engine. They now had one very dead engine and one very dying engine. They didn't make it. Now, the rule is *take off, get stable*, then confirm what the problem is, agree that that's the problem, then take action that shuts down an engine. Because on a twinjet? Shutting down your one good engine is a bad thing.

(Yes, by the way, these are all written with V for velocity, but spoken as "speed". They're pilots, not physicists, and it's an Airspeed Indicator.)
posted by eriko at 7:34 AM on August 3, 2015 [11 favorites]

cars are amazing too. they kill a fuckton more people.

Yes, but to be fair, they get a fuckton more chances.

And I suspect if we used a tenth of the rigor in driving that we use in flying, driving would be vastly safer than flying. There are lots of failure modes in cars that kill you in flying that are nothing burgers in cars. Out of fuel? Pull over in a car, probably die in a plane. Break a window? Annoying in a car, deadly in a plane at FL300 unless you can get below FL120 before your oxygen mask runs out. And so forth.

But we don't use a thousandth of the rigor. We just hand out cars and licenses, we barely inspect them to make sure they work, we barely test drivers for competence, and we send them out on the roads without any central coordination, and when there is a crash, a cop writes a report and that's it.

Really, the only reason that cars have become safer is that the insurance industry -- which has to pay out claims -- basically is demanding that they try to kill less people. And, well, it is working. Cars today are vastly safer than before. But still, the reason that flying is so safe is there is a very rigorous process in making sure that the aircraft and the operators are safe, and when something does go wrong, there's a very deep and intensive investigation to find out the detailed causes, then procedures and equipment is changed to prevent those causes from happening again.

And really, a lot of those inspections are overkill on cars. You do not need to do a 4 hour long inspection on a car every week. A car is not going to fall out of the sky on someone. However, I think a four hour long inspection on a car every year may not be a bad idea. However, that's going to be expensive, and people are going to be pissed at the cost, doubly so when you tell them that they will fix the problem or they can't drive the car.

And that's what they do with planes. The reason that there are a number of stored 777s right now is that they've come up for D checks, which can take up to two *months* for a wide body airliner and basically involve taking the entire airplane apart and checking every bit of it for cracks. If you're going to use the A/C for another twenty years, sure, you do it, but if you're not? You park the thing -- why spend a million bucks on the D check if you're not going to fly it?
posted by eriko at 7:45 AM on August 3, 2015 [5 favorites]

If you haven't seen it, there is an excellent set of training videos from American Airlines on YouTube. Windshear & Microburst Review is especially relevant to the above discussion.

I also highly recommend Children of Magenta for an interesting discussion of automation and human factors.
posted by scolbath at 10:11 AM on August 3, 2015 [1 favorite]

Thanks for this post.

I could not imagine what that would be like in an environment that had so much death and destruction while threading a needle with my own life.

I had a friend in high school whose father was one of the survivors seated very close to the break. He was relatively okay physically, but the PTSD led to alcoholism and the end of his marriage.
posted by lost_cause at 12:46 PM on August 3, 2015 [2 favorites]

« Older Meet Jane Marie Kroc   |   You can't fuck osmosis jones Newer »

This thread has been archived and is closed to new comments