Ten Disasters That Were Predicted, and Ignored

On July 31, 1985, Morton Thiokol engineer Roger Boisjoly sent his company's engineering vice president a memo about the rubber O-rings that sealed the joints of the shuttle's solid rocket boosters. After a recent flight in which a primary O-ring had failed to seal, he warned that a joint failure could cost lives, and he urged a dedicated team to fix the problem. The team was set up, but its own members complained it was constantly delayed.

Cold made the seals worse. A January 1985 launch at an O-ring temperature of 53°F, the coldest to that point, had shown erosion in both boosters. On the evening of January 27, 1986, with a freezing night forecast in Florida, Thiokol engineers told NASA on a teleconference that they should not launch below 53°F. NASA managers at Marshall Space Flight Center pushed back. During an offline caucus, a senior Thiokol executive told the engineering vice president to take off his engineering hat and put on his management hat. Thiokol then recommended launch.

Challenger lifted off the next morning in 36°F air and broke apart 73 seconds later, killing all seven crew members. The Rogers Commission found the decision to launch was flawed and that Thiokol management had reversed its engineers to accommodate a major customer.

Fourteen miles above Johnstown sat the South Fork Dam, an old earthen canal dam that had partly failed in 1862. In 1879 a group of wealthy Pittsburgh men formed the South Fork Fishing and Hunting Club, bought the site and rebuilt the dam to create a private lake. Benjamin Ruff, who supervised the work, had no engineering training. The club lowered the crest to fit a road, put fish screens across the spillway, and never replaced the drainage pipes an earlier owner had sold for scrap, so the lake could not be drained.

Johnstown's biggest employer worried. Daniel Morrell, head of the Cambria Iron Company, raised the dam's condition with the club, and in a December 1880 letter Ruff assured him that Johnstown was in no danger from the club's enterprise. Engineers who looked at the work had concerns too, according to a peer-reviewed case study by the Association of State Dam Safety Officials.

After a torrential storm on May 30 and 31, 1889, the lake rose fast. Workers tried to cut a second spillway and clear the screens. Warnings were telegraphed down the valley, but residents had heard false alarms before. The dam gave way around 3:10 p.m., and the wave hit Johnstown before 4:10. The official toll was 2,209 dead. The club was never held legally responsible.

In July 2004, more than a year before Katrina, FEMA paid for a five-day disaster simulation at Louisiana's emergency operations center in Baton Rouge. Officials from about 50 parish, state, federal and volunteer agencies faced an imaginary storm called Hurricane Pam: a slow, strong Category 3 that pushed storm surge over the levees and flooded New Orleans.

The assumptions read like a forecast. Planners expected about 300,000 people not to evacuate in advance, 500,000 to 600,000 buildings to be destroyed, nearly all communications to fail, and thousands of rescues by boat and helicopter. The scenario estimated more than 60,000 deaths and parts of southeast Louisiana left uninhabitable for over a year. A House select committee later called the exercise prescient.

Follow-through lagged. Workshops continued, but a second full Pam exercise planned for summer 2005 did not happen, apparently for lack of funding, and the committee found that implementation of Pam's lessons was incomplete. Former FEMA director Michael Brown testified that his request for money to act on those lessons was denied. Katrina came ashore in late August 2005. The levees failed, the city flooded, and at least 1,100 people died in Louisiana alone, according to the committee's final report, A Failure of Initiative.

On April 14, 1912, Titanic's wireless room was busy. According to the U.S. Senate inquiry, at least three ice warnings went directly to Captain Edward Smith that day. Around noon the White Star liner Baltic relayed a report of icebergs and field ice near Titanic's track. That afternoon the Californian reported three large bergs to the north. A message from the German liner Amerika about two large icebergs, south of the course, passed through Titanic's own radio on its way to Washington.

The last warning from Californian came roughly an hour before the collision. The Senate committee found the reports placed ice on both sides of Titanic's lane and in its immediate vicinity. Its judgment was unsparing: no conference was called, no heed was given, the lookout was not increased and the speed was not reduced. The ship was making at least 21 knots, the fastest of the voyage, on an exceptionally clear night over a calm sea.

Shortly before midnight ship's time, the lookout called out an iceberg dead ahead. It was too late to turn. Titanic sank in under three hours. Of the 2,223 people the Senate counted aboard, 1,517 died, a toll made worse by lifeboats with room for only 1,176.

Above the coal-mining village of Aberfan in South Wales, the National Coal Board piled mine waste into towering tips on the mountainside. One of them sat on top of a spring. Local officials were nervous about it for years, and in 1963 and 1964 the Merthyr Tydfil Borough Council and the Coal Board exchanged letters under a heading about the danger from coal slurry tipped behind the Pantglas schools.

At 9:15 a.m. on Friday, October 21, 1966, the last day before half-term, the tip slid. A tide of black slurry poured down the hill, destroyed a farm cottage, and engulfed Pantglas Junior School and about 20 houses. The children had just come back to class from morning assembly. The village was in fog, and nobody below saw it coming; the men on the tip could not phone a warning because the telephone cable had been stolen repeatedly. In all, 144 people died, 116 of them children.

A tribunal of inquiry sat for 76 days. It found that the Coal Board had no tipping policy and that its chairman's claim not to have known about the spring was false. Its report described a story of bungling ineptitude and a failure to heed clear warnings, and it placed the blame on the Coal Board. No one was prosecuted.

William Mulholland was the self-taught engineer who brought water to Los Angeles, and in the 1920s he built the St. Francis Dam, a 205-foot concrete wall about 35 miles northwest of downtown. A 1917 state licensing law exempted municipal engineers like him from outside review. According to a peer-reviewed case study by the Association of State Dam Safety Officials, his crews did little geologic investigation, drilled only a handful of shallow borings, and skipped foundation grouting even though workers on the nearby aqueduct had already complained about the treacherous rock.

The warnings in the final days were local and specific. With the reservoir nearly full in early 1928, leaks at the abutments worsened. On the morning of March 12, the dam keeper, Tony Harnischfeger, saw muddy water flowing from a new leak and called Mulholland. Mulholland and his deputy toured the dam for well over an hour, concluded the leak was not dangerous, and drove back to the city.

At 11:57 p.m. that night, the dam collapsed. A wall of water roared down the canyon, killing more than 400 people, the dam keeper among them. Warnings phoned ahead by switchboard operators saved many more. Mulholland told the coroner's jury that if there was an error in human judgment, he was the human.

The Vajont Dam in the Italian Alps, about 60 miles north of Venice, is one of the tallest dams in the world at 860 feet. The engineering of the arch held. The problem was the mountain beside the reservoir. Locals called Mount Toc "the mountain that walks," and during the first filling a local newspaper warned that landslides there were recurrent. Authorities responded with legal action against the paper.

The mountain kept sending signals. In November 1960 a rockslide of nearly a million cubic yards dropped into the reservoir. After that, a clear pattern emerged in the monitoring data: each time the water rose, the slope crept faster, and each time it was drawn down, the movement slowed. By early October 1963 the slope was moving almost 8 inches a day. In the final days, the mayors of nearby Erto and Casso urged people to leave the slopes, but downstream towns got little warning.

On the night of October 9, roughly 350 million cubic yards of rock slid into the reservoir in under a minute. A wave overtopped the dam and swept down the valley, reaching Longarone about four minutes later. More than 2,000 people died. The dam itself survived largely intact, which is the point: the danger had been in plain sight on the slope.

When Fukushima Daiichi was licensed, its designers planned for a tsunami of 3.1 meters, about 10 feet, based on the wave that a 1960 Chilean earthquake sent across the Pacific. In 2002, using a new method from the Japan Society of Civil Engineers, the operator, Tokyo Electric Power Company (TEPCO), raised that figure to 5.7 meters. Geologists, meanwhile, had found sand deposits showing that a huge tsunami struck this coast in the year 869, and that such events recur roughly once a millennium.

In 2008 TEPCO ran preliminary simulations based on a repeat of the 869 earthquake. They suggested the plant's tsunami risk had been badly underestimated, with possible run-up of more than 15 meters. TEPCO said it doubted the results and planned further study. According to a 2012 Carnegie Endowment analysis, that follow-up appears not to have happened, and TEPCO reported the results to the regulator only on March 7, 2011.

Four days later, a magnitude 9.0 earthquake sent a tsunami roughly 13 meters high into the plant, knocking out the plant's emergency power and leading to fuel melting in three reactors. The Carnegie authors concluded the accident was preventable.

On October 8, 2018, the engineering firm Morabito Consultants delivered a structural survey of Champlain Towers South, a 12-story condominium on the beach in Surfside, Florida. Much of it was routine: sealant, balcony tile, stucco cracks. One section was not. The waterproofing beneath the pool deck and entrance drive had been laid flat, the report said, so water sat on it instead of draining. The firm called this a major error in the original design and warned that the failed waterproofing was causing major structural damage to the concrete slab below.

The report added that failing to replace it soon would cause the deterioration to expand exponentially. It also noted abundant cracking and spalling in the garage columns, beams and walls, including exposed, deteriorating rebar. The fix was expensive and disruptive. An April 2021 letter to residents laid out a roughly $15 million repair program and said the deterioration had gotten much worse, but the concrete work had not begun.

In the early morning of June 24, 2021, part of the building collapsed, killing 98 people. In June 2026, federal investigators at the National Institute of Standards and Technology reported that the collapse began weeks earlier, when connections between garage columns and the pool deck slab failed. They traced the building's thin safety margins to its original design and construction, made worse by added loads and corrosion.

Seventeen years after Challenger, NASA lost a second shuttle to a known problem. Foam insulation had been falling off the external tank since the program's earliest flights, and the Columbia Accident Investigation Board found that debris had struck orbiters on most missions with usable launch imagery. In October 2002, just months before Columbia's last flight, a chunk of foam from the same bipod area of the tank came off during an Atlantis launch. Managers still cleared the next missions to fly.

On January 16, 2003, 81.7 seconds after Columbia lifted off, a piece of foam broke away and hit the leading edge of the left wing. Engineers spotted it in launch film the next day and grew alarmed. Over the following week, three separate requests went up the chain to have military imaging assets photograph the wing in orbit. The request was canceled. When managers asked whether anyone had a formal requirement for imagery, nobody could name one.

On February 1, during reentry, superheated gas entered the breach and destroyed the wing. Columbia broke up over Texas, and all seven astronauts died. The board concluded NASA's culture had as much to do with the accident as the foam did.