Monday, November 16, 2026

Local, National, International: Three Engineering Accidents and Why They Made the News

Last week brought three widely reported occasions in which the plans and intentions of engineers went awry, but in different ways and with different consequences.  Together, they say something about why such incidents are newsworthy and how the size of a mishap or accident, measured by the number of people affected, influences the way the incident is covered.

 

The local incident happened on Wednesday, Nov. 11.  A building site at the intersection of Bedford Avenue and Flatbush Avenue required the excavation of a square area to a depth exceeding a full story, right next to a three-story apartment building with a store on the ground floor.  Normally, the weight of a building is carried by the ground beneath and beside it.   Removing underlying layers of soil immediately next to a building requires shoring the exposed vertical surface with a partition that can withstand the resulting horizontal pressure formerly opposed by the removed soil. 

 

Around 4 P. M. that day, the shoring apparently failed, and the soil underneath the apartment building came crashing into the excavation, removing its support from underneath that side of the building.  A portion of the building then collapsed, killing eight people and injuring eighteen more. 

 

Somewhere in the world, a building or part of a building collapses nearly every day.  But when more than one person is killed, and when the manner of death is peculiarly horrible, it can make national headlines.  Imagine that you're sitting on your couch watching a soap opera one second, and the next second everything around you vanishes into a heap of rubble that ends your life.  The power of such an image makes it fearful all out of proportion to its unlikelihood, which is another reason it was so widely covered.

 

The national incident happened on Friday (yes, it was the 13th) of November.  The U. S. Federal Aviation Administration (FAA) was installing some new telecommunications equipment as part of a long-delayed upgrade.  When a flight controller at an FAA-operated control center speaks with a pilot, his or her voice is carried over a complex system resembling a specialized telephone network, which automatically routes the conversation between the control center and radio transceivers sited in locations that will reach the plane in question.  Evidently, due to conflicts between the new parts of the system and the legacy parts, it began to break down to such an extent that the FAA managers declared a "ground stop," meaning no flights were allowed to take off anywhere in the U. S.  Many flights in progress were ordered to turn around or land at a different airport than planned.  After five hours, the stop was partially lifted, but the disruption to air traffic was so extensive that it will be many days before flight schedules return to normal. 

 

Unlike many countries, where air-traffic-control organizations are private and paid for with operating fees from the airlines, the U. S. government directly operates and pays for the FAA, which makes it subject to the political whims of Congress.  Although funds have been allocated for upgrades, even the very process of improvement can cause problems, as we saw last week.  No fatalities or injuries were directly attributed to the shutdown, but thousands of travelers and tons of cargo were affected, and the monetary cost of the disruption both to airlines and their customers is probably in the many millions of dollars. 

 

The international accident occurred Thursday afternoon, Nov. 12.  Unlike the other two, this one affected me personally, as most of my email comes through my university, which uses Microsoft Outlook as its email service.  I began to notice Thursday afternoon that the last email I'd received came around noon.  I can usually count on a steady flow of (mostly junk) email twenty-four hours a day, so I checked around on the internet and discovered that not only Microsoft email, but other Microsoft services such as Teams were also affected.  Updates from Microsoft online stated that a failed "health check" on many servers had caused them to go offline, leaving the vast bulk of Microsoft traffic to be handled by a few remaining servers, which quickly got overwhelmed. 

 

As it happened, I wasn't urgently needing to email anything, so the problem amounted for me to an almost welcome respite from having to check my email.  Eventually, old emails began to show up close to a day later.  But governments and health-care organizations around the world, which were part of the estimated 400 million users of Microsoft internet products globally, had more serious problems, although no fatalities or injuries were directly attributed to the outage.  However, an estimate by the Anderson Economic Group put the outage's price tag at around $7 billion from lost sales and other causes.

 

There's an old saying in the news business:  "if it bleeds, it leads."  If an engineering-related incident kills or seriously injures anyone, especially if the numbers are high, it's going to make the news somewhere.  And if the circumstances are particularly gruesome or horrifying, which well describes the Brooklyn building collapse, it's even more likely to receive at least nationwide coverage.  Thousands of people die in more mundane ways every day without the world taking notice, but a morbid part of human nature is attracted to sudden, violent, and unexpected death caused by a failure of our built environment that we normally trust so well. 

 

If death and injury are not present, large numbers will often make the news cut.  One flight delay goes unnoticed, but thousands of delays and cancellations across the entire U. S. with knock-on effects on international flights made the FAA ground stop highly newsworthy.  And when the cause isn't a natural one, but a failure of what should be one of the most reliable government telecommunications networks around, the news value increases further because politics gets in the mix. 

 

Finally, although Microsoft is a publicly-traded private company, its size in terms of revenue exceeds that of some small nations, and its reach is truly global.  So even something as comparatively trivial as having one's email hung up for a day or so made the news, because it affected so many people. 

 

Lessons were learned from all these mishaps, and let's hope that if anything like them happens in the future, at least they won't happen again all in one week.

 

Sources:  For the Brooklyn building collapse, I referred to Wikipedia's "2026 New York City building collapse" and https://www.cnn.com/2026/11/11/us/live-news/brooklyn-building-collapse-nyc.  For the FAA ground stop, I referred to https://apnews.com/article/flights-halted-airports-faa-f81b0904331dafad6133f291e5f531a8.  And for the Microsoft outage, I referred to Wikipedia's "2026 Microsoft outage" and https://www.securityweek.com/microsoft-365-outage-disrupts-outlook-exchange-online-and-teams/. 

 

Monday, November 09, 2026

Learjet 60 Crash: What Failed?

Around 10 AM on Sunday, Nov. 1, tech entrepreneur Adam Robinson, his wife Helen, their two small children, their nanny, and two friends of the family boarded a Learjet 60 aircraft at Austin's Bergstrom International Airport for a vacation trip to the Bahamas.  The charter jet's captain and pilot was 55-year-old Jeff Gregory, and his co-pilot was 20-year-old Nathalia Gadelha Filgueiras.  The plane took off from Bergstrom at 10:37 AM with a route that took it over southern Florida.

 

Severe thunderstorms were in and around Palm Beach when the plane crossed the west coast of Florida.  At 12:40 PM, Nathalia, who was evidently piloting the plane at the time, texted her pilot boyfriend that she was encountering turbulence and ice.  This was about five minutes after she had decreased the plane's altitude from 41,000 feet, where it had been for most of the flight, to 35,000 feet.  At 12:47 PM, she requested authorization from air traffic control to climb to 37,000 feet, after reporting extreme turbulence that changed the plane's altitude up to 700 feet, icing, precipitation, and an outside temperature of -47 F. 

 

According to records recovered later, the plane then entered a steep dive, which was reported by the co-pilot in the next minute or so as the plane headed at an average speed of 300 MPH (480 km/hr) almost straight down.  It crashed nose-first into a mobile home park in Greenacres, a subdivision of Palm Beach, at 12:49 PM.  The crash was caught on a dashcam and shows a huge fireball erupting from a crater that quickly filled with water from the local rainfall.  All nine people aboard the jet were killed and one person on the ground was injured.  The number of fatalities was the largest of any accident involving the Learjet 60.

 

Questions were raised about the qualifications of Co-pilot Filgueiras, who had less than 2,000 hours of flying time but was fully certified to fly the Learjet.  Her father publicly defended her record and her professional behavior in making the MayDay call when the plane began its nosedive, and not losing her composure even after she and her passengers were facing certain death.

 

It took more than a day to drain the large crater the plane dug itself into, but over the following week efforts were under way to recover the bodies and the black-box voice recorder of the plane to aid the National Transportation Safety Board (NTSB) investigation into the crash.  Typically, the NTSB will issue an initial report of facts discovered about 30 days after an accident, but the full investigation is likely to take a year or more.

 

While the final word will come from the NTSB, anyone may speculate about the reasons why a plane crash happened.  This one is remarkable not so much for the number of fatalities, although it is a record for the type of plane involved.  Factors making it newsworthy include the striking dashcam video capturing the exact moment and aftermath of the crash, the poignancy of an entire family being wiped out in one instant, and the relative youth and consequent inexperience of the co-pilot. 

 

Without prejudging the investigation outcome in any way, there are some general facts and policies that are relevant to this accident.

 

In general, pilots are well advised to avoid penetrating thunderstorms whenever possible.  A well-known pilot's manual, Modern Airmanship, says the following about them:  ". . . a pilot can expect to encounter thunderstorm areas frequently.  A knowledge of how to avoid them in planning and flying, or how to penetrate them as a last resort, is basic airmanship."

 

For reasons that may emerge later in the investigation, the co-pilot of the Learjet 60 decreased the plane's altitude from 41,000 feet to 35,000 feet around 12:35 PM.  As the Bahamas are not that far from the Florida coast, this may have been part of the initial approach to the Bahama airport. 

 

For whatever reason, the move to a lower altitude appears to have placed the aircraft in a region of severe turbulence and icing.  Temperatures below -40 C (-40 F) are not unusual at high altitudes above 30,000 feet.  But when supercooled water is present, as in the upper reaches of a thunderstorm, a solid object such as a wing surface can rapidly accumulate a crippling amount of ice.  The rapid changes of altitude experienced by the Learjet show that it had encountered either updrafts or downdrafts within the storm, further complicating the problem of flying the plane. 

 

After the crash, the plane's elevator, a tail surface used to control its pitch, was found in pieces a mile or more from the crash site, indicating that it separated from the plane in the air. 

 

The Learjet 60 is not a fly-by-wire aircraft, which means that the force necessary to move the control surfaces is provided either directly by the pressure on the controls, or through a power-assist hydraulic system that still allows haptic feedback (the "feel" of a control), rather like power steering on an automobile.  One can speculate that an ice buildup on the control surfaces either distorted the airflow around them or prevented full movements necessary to control the plane.  In any event, once the pilots lost control of the aircraft and it went into a dive, the airfoil movements necessary to end the dive and gain control might not have been possible.  And the separation of the tail surfaces from the plane sealed its fate, as there was nothing more they could do to avoid the crash.

 

Further investigation will reveal a more detailed sequence of events that led to this tragedy.  Even pilots with decades of experience are sometimes taken by surprise when flying in or near thunderstorms.  And it's likely that a combination of situations, each one of which could have been dealt with successfully by itself, may have led to the crash when they all happened at once. 

 

Our sympathy goes out to those who lost loved ones in the crash.  And I hope that the full story will serve as a lesson on what pilots can do to avoid similar dangerous conditions in the future.

 

Sources:  I relied on the Wikipedia article "2026 Palm Beach County Learjet 60 crash," as well as the following websites:  https://www.tmz.com/2026/11/02/florida-private-plane-crash-victims-identified/, https://abc7news.com/amp/post/florida-learjet-crash-grieving-father-60-pilot-nathalia-gadelha-filgueiras-defends-daughters-reputation/20022148/, and https://www.aerotime.aero/articles/learjet-elevator-pieces-florida-crash.  The quote about thunderstorms is from p. 288 of Van Sickle's Modern Airmanship, 6th Ed. edited by John F. Welch (Tab Books, 1990).

 

Monday, November 02, 2026

The Toledo Train Crash: What's Wrong with Spain's Trains?

The Spanish high-speed rail network is second only to China's in its size and number of passengers carried per year.  From its opening in 1992 until this year, there were no fatal accidents on the high-speed trains, which travel at speeds up to 160 MPH (250 KPH) across the countryside.  But the Madrid-Seville line, which goes south of the capital through the province of Toledo, was the site of not one, but two major accidents this year that caused dozens of fatalities. 

 

Both private and publicly-owned train systems share the line.  The private system, Iryo, and the public system, Renfe, must share maintenance responsibilities on some basis.  The January accident happened when a poorly welded rail joint separated.  The stresses of supporting trains going faster than some planes can fly must have contributed to the failure.  But even a slight unevenness in the track is intolerable at such speeds.  When the northbound Iryo train hit the faulty joint, three of its carriages derailed and landed on the adjacent southbound track.  Twenty seconds later, a southbound Renfe train collided with the wreckage.  Forty-six people were killed and over two hundred were injured.  Systems designed to detect a breaking rail failed to activate in time to warn either train.

 

Last Wednesday, October 28, all trains on a 39-mile (63-km) stretch of the Madrid-Seville line near Toledo were stopped automatically when a computer failure disabled the LZB (short for Linienzugbeeinflussung) automatic train-control system.  This included both the Iryo train, which left Madrid first, and the Renfe train, which followed only six minutes later.  When it works, LZB resembles what in the U. S. is called "positive train control" in that it monitors speeds and can automatically apply brakes if the engineer neglects to do so in an emergency. 

 

But without LZB, rail traffic controllers had to get into radio communications with all the affected trains in order to bypass the automatic directions from the now-nonfunctioning LZB system. 

 

At 8:36, a controller told the Renfe train it could start running at normal speed, as the Iryo train was 15 miles (25 km) ahead of it, and presumably would get going at about the same time.  What the Renfe driver didn't know was that the Iryo engineer couldn't restart the train.  Although details of the communications between the two trains and controllers have not been fully investigated, it is clear that the Renfe driver didn't get the message that the Iryo train was still stopped, and that he shouldn't proceed. 

           

The line has an older backup train-control system, possibly using signal lights, but unfortunately both trains were on the same control "block" (section of track).  So there were no signals from the older backup system to inform the Renfe driver of the stopped Iryo train ahead.

 

When the Iryo train came into view, it was too late to stop the Renfe train, as it takes at least a kilometer (0.6 miles) to come to an emergency stop.  The train was going over 100 MPH when it collided with the rear of the Iryo train, killing at least 69 people and injuring hundreds more.

 

Although dull and boring compared to the excitement accompanying the opening of a new system such as Spain celebrated with the inauguration of its first high-speed rail line in 1993, maintenance of equipment is crucial to the safe and reliable operation of any system as complex as a high-speed rail line.  Without having detailed information on the conditions that prevail among controllers, maintenance workers, and management of the Spanish high-speed rail system, anything I say at this point must be speculative.  But based simply on the causes of these two accidents, which spoiled the system's safety record in a spectacular way, I would say that those responsible for system maintenance have come up short, for whatever reason.

 

The track failure in the January crash was due to a faulty weld.  I don't know how the Spanish system inspects its tracks, but it's at least possible that the fault would have shown up in a routine track inspection, which often uses sophisticated equipment to detect flaws before they become as obvious as a broken welded joint.  And the system designed electrically to detect a track breakage also failed, whether due to faulty maintenance on it or for some other reason. 

 

Turning to the October crash, failures on the Spanish tracks of the German LZB system are so common as to have become unremarkable.  It relies on physical cables laid along the tracks, and is thus vulnerable to vandalism and other types of damage.  But this particular outage was due to a computer problem, which might have been avoided by running identical backup systems as NASA and other agencies do when human lives are involved. 

 

Instead, controllers and engineers became accustomed to LZB failures, and relied instead on overworked train controllers, overworked at least according to the train-controller's union.  And somehow, the critical information that the Iryo train wasn't moving failed to get from the Iryo train's engineer, to a controller, to the Renfe train.  If LZB had been operational, it would have probably detected the Iryo train's presence and stopped the Renfe train, or prevented it from starting in the first place.

 

While the controllers played a role in this tragedy, it would be unfair to accuse them of being the sole cause.  Poor maintenance and the lack of fail-safe systems for the LZB control devices are at least as much to blame.  The world's greatest safety system is no use if it's out of commission.

 

Spain's Minister of Transportation, Óscar Puente, has attributed the accident mainly to controller error, but is under pressure to resign in the face of the second major fatal accident on high-speed rails this year.  The first accident should have occasioned a major housecleaning in which accidents waiting to happen, such as poorly inspected rails and LZB systems that fail several times a month, would be cleaned up.  But that didn't get done, and now there will be 69 fewer rail passengers looking forward to Christmas—or to anything else.

 

Sources:  I referred to an Associated Press article "Spain's latest deadly crash puts human error and systemic failure under scrutiny" at https://apnews.com/article/spain-train-crash-toledo-causes-unions-669ef8c0c12c8c9b720e721e7532ccbb, the Wikipedia articles "2026 Toledo train collision" and "2026 Adamuz rail disaster," and incidental questions to Google's AI function, which is hard to avoid but convenient when a specific question comes up.