← Back to Gallery

Dam Failures

Six Walls of Water Set Free: An Illustrated History of When Engineering Lost Its Argument with Gravity

"Every dam is built to fail eventually."
— Civil engineering proverb. The job of the engineer is to ensure failure happens at the end of design life, not in the middle of the night.
6
Catastrophes
92
Years Spanned
~175k+
Lives Lost
4
Failure Modes
5
Countries
1

St. Francis Dam — California's Worst Civil Engineering Disaster

San Francisquito Canyon, Los Angeles County, March 12, 1928 • Built in Two Years, Failed in Three

The St. Francis Dam was the keystone of the Los Angeles Aqueduct, designed by William Mulholland, the self-educated Irish immigrant who had built Los Angeles's water empire. Two years after completion the foundation rock — a friable schist on one abutment and a paleo-landslide on the other — failed. Twelve billion gallons of water tore down San Francisquito Canyon and across the Santa Clara Valley to the Pacific, killing at least 431 people in the night. The disaster destroyed Mulholland's career and reshaped American dam-safety law.

🔨

William Mulholland — Builder of Los Angeles

1855–1935 • Self-taught engineer, head of LA Bureau of Water Works

Born in Belfast, Mulholland began as a ditch tender in 1878 and rose to chief engineer of the Los Angeles Aqueduct (1913), the project that brought Owens Valley water 233 miles to the city. The St. Francis Dam was its terminal storage. Mulholland personally inspected a leak the morning of the failure and pronounced the dam safe. It collapsed at 11:57 p.m.

"Don't make a fuss about the leak. It's just normal seepage."
— William Mulholland to dam keeper Tony Harnischfeger, 11:00 a.m., March 12, 1928. Harnischfeger and his family were among the first killed when the dam collapsed at 11:57 p.m.
📝
August 1924 – May 1926
Construction
Built in 21 months by Mulholland's Bureau of Water Works without an outside engineering review. Mid-construction, Mulholland increased the height by 20 feet (and again by 10) without recomputing structural margins.
🌊
May 4, 1926
Reservoir filling begins
St. Francis Reservoir starts to fill. Cracks and seepage are observed almost immediately on both abutments but dismissed as normal "drying shrinkage" of fresh concrete.
🐋
March 7, 1928
Reservoir reaches full capacity
Reservoir reaches its maximum elevation, 1,835 feet, holding 38,000 acre-feet (12.4 billion gallons) for the first time.
👀
March 12, 1928, morning
Mulholland inspects, declares it safe
Dam keeper Tony Harnischfeger reports muddy seepage at the west abutment. Mulholland and assistant Harvey Van Norman drive up, examine the leak, and declare the dam sound. They leave around noon.
💥
11:57:30 p.m., March 12, 1928
Catastrophic failure
Within seconds, the entire dam fails. Both abutments give way; the central buttress section topples. A 140-foot wall of water (later models suggest the wave was over 100 feet at canyon mouth) sweeps down San Francisquito Canyon at ~18 mph.
🎪
March 13, 5:30 a.m.
Flood reaches the Pacific
The flood travels 54 miles to the ocean at Ventura. It destroys the towns of Castaic Junction, Fillmore, Bardsdale, and Santa Paula. Recovered bodies number 431; estimates of total dead range from 432 to over 600 (many migrant workers were never identified).
August 1928
Coroner's inquest & Mulholland resigns
A jury finds the dam's design and construction unsound but assigns no criminal liability. Mulholland: "Don't blame anybody else, you just fasten it on me. If there is an error of human judgment, I was the human." He resigns and lives the rest of his life in seclusion.
👨
Tony Harnischfeger

The dam keeper who reported the morning leak. He, his six-year-old son Coder, and his common-law wife Leona Johnson were killed within minutes of the breach. Their bodies were never recovered.

👨🏻‍🔨
Harvey Van Norman

Mulholland's chief assistant who accompanied him on the morning inspection. Took over the LA Bureau of Water Works after Mulholland's resignation.

🏠
City of Santa Paula

Hardest-hit downstream town. The flood reached its outskirts at ~3:00 a.m., killing roughly 100. The St. Francis Dam Disaster Memorial sits in the Mountain View Cemetery.

📖
Charles F. Outland

Survivor and historian whose 1963 book Man-Made Disaster remains the canonical account. He grew up in Santa Paula; his family escaped before the wave reached town.

⚖ Place in the Failure Lineage

The archetype of failure-by-foundation: a dam that satisfied its designer's intuition but failed at its rock contact. Future cases — Malpasset (1959), Vajont (1963), Teton (1976) — all involved geology that the designers either misread or ignored. St. Francis founded the modern principle that a dam's foundation matters as much as its concrete.

2

Malpasset Dam — The First Modern Arch Failure

Var, France, December 2, 1959 • The World's Thinnest Arch Dam Discovers a Hidden Fault

Designed by André Coyne — the most celebrated dam designer of his generation — Malpasset was a 1.5-meter-thick double-curvature arch dam, an elegant minimum-concrete design that exploited the geometric strength of two intersecting curves. After heavy rain in late November 1959, the reservoir filled to capacity for the first time. A wedge of foliated gneiss in the left abutment slipped along a previously unknown fault. The arch lost its bearing, snapped, and 50 million cubic meters of water reached the town of Fréjus 21 minutes later. The disaster ended Coyne's life — he died nine months later at 69 — and recalibrated the world's relationship with thin-arch dams.

🇫🇷

André Coyne — Designer of 70 Dams

1891–1960 • French civil engineer, founder of Coyne et Bellier

Coyne pioneered the thin double-curvature arch dam, building 55 in France and another 15 abroad — including Tignes, Mauvoisin, and Roselend. Malpasset was, geometrically, the most efficient of all of them. The post-failure inquiry concluded that the design itself was sound; the geology had betrayed it.

"Le barrage n'a pas cédé: c'est le rocher qui a cédé." (The dam did not give way: the rock gave way.)
— André Coyne, after the failure. He died nine months later, in July 1960, of a heart attack widely attributed to the strain of the inquiry.
🏗
April 1952 – April 1954
Construction
Built to provide drinking and irrigation water for the Var département. The arch is one of the thinnest ever attempted at this height: only 1.5 m thick at the crown, 6.78 m at the base.
📊
1954–1959
Slow filling, no problems detected
Filling proceeds gradually over five years due to insufficient watershed runoff in dry years. No alarming displacements are recorded; instrumentation is sparse by modern standards.
🌧
November 28, 1959 — December 2
Heavy rains fill reservoir to top
Five days of unusually heavy rainfall raise the reservoir to within 30 cm of the spillway crest — the highest level it has ever held. The chief engineer requests permission to open the floodgates; the reply is delayed.
💥
9:13 p.m., December 2, 1959
Catastrophic failure
A wedge of fractured gneiss in the left abutment slips. The arch loses bearing and shatters. Engineer André Ferro at the dam house is the first killed. The wave reaches 40 m at the breach.
🎧
9:34 p.m., December 2
Wave hits Fréjus
The flood reaches Fréjus 12 km downstream in 21 minutes. The Roman amphitheater is partially submerged. The total death toll is 423, including dozens of campers and a U.S. Army camp at Campécool.
📖
1960–1964
Inquiry & geological re-evaluation
The Caquot Commission identifies a previously unknown fault parallel to the foliation of the gneiss, intersecting the abutment. No core borings had crossed it. Coyne dies in 1960. The design itself is exonerated.
👨
André Ferro

Resident engineer and dam keeper killed in the breach. Found at his post in the dam house, which was swept downstream and destroyed.

🎧
Town of Fréjus

Roman-era city of 30,000. The flood killed 423, devastated the western suburbs, and damaged the 2,000-year-old aqueduct and amphitheater. President de Gaulle visited within 48 hours.

👨🏻‍🔬
Pierre Londe

French rock-mechanics pioneer who post-failure developed the "Londe wedge analysis" method, now standard for dam abutment stability assessment.

🇩🇪
Alfred Caquot

Octogenarian dean of French civil engineering who chaired the inquiry commission. His report exonerated the design and made geology, not concrete, the chief lesson.

⚖ Place in the Failure Lineage

St. Francis (1928) had warned about foundation rock; Malpasset proved the warning was generic, not local. Even a flawless arch designed by the world's leading engineer could fail if the abutment geology hid a fault. Every modern dam now requires a "failure mode and effects" review of its foundations.

3

Vajont — The Dam That Did Not Fail

Belluno, Italian Alps, October 9, 1963 • A Landslide Filled the Reservoir, Then Overtopped the Dam

The Vajont Dam itself never failed. Built by SADE (Società Adriatica di Elettricità) and at 261.6 meters the world's tallest arch dam at completion, it stood throughout the catastrophe and stands today. What failed was Mount Toc on its left flank: a 260-million-cubic-meter wedge of carbonate slope, slowly creeping along beds of clay-rich marl, slid into the full reservoir in 45 seconds. The displaced water rose 250 meters above the crest, sent a 50-million-cubic-meter wave over the dam, and obliterated the village of Longarone in the valley below. Roughly 2,000 people died. SADE engineers and government inspectors had observed the creep for years and continued to fill the reservoir.

🏔

Carlo Semenza — The Designer Who Died Before the Disaster

1893–1961 • Chief engineer of SADE

Semenza had designed Vajont as the keystone of an integrated Piave hydroelectric system. After the November 1960 sliding incident at the reservoir's western flank, he commissioned a confidential study by Austrian geologist Leopold Müller. Müller's report (March 1961) identified the entire north flank of Mount Toc as a paleo-landslide. Semenza died of a heart attack months after reading it.

"Quel di che giurato avete una colpa che non avete... Voi non avete colpa. Le colpe dello Stato sono altre."
— Tina Merlin, journalist of L'Unità, after years reporting the warning signs the authorities suppressed. She had been prosecuted for libel in 1959 for accusing SADE of negligence. The 1961 verdict acquitted her.
🏗
1957–1960
Construction
SADE builds the world's tallest arch dam, 261.6 m, in a steep limestone gorge. The reservoir is sized at 168.7 million m³. The mountain rim, including the 1,921-m Mount Toc, was already nicknamed the "walking mountain" by locals.
November 4, 1960
First major slide (700,000 m³)
During first reservoir filling, a 700,000 m³ mass slides into the lake from Mount Toc's flank. Geologist Edoardo Semenza (Carlo's son) maps a perimeter crack 2 km long around the entire north face.
📝
March 1961
Müller report identifies paleo-landslide
Austrian geologist Leopold Müller, at SADE's confidential request, identifies the entire north flank as an ancient landslide ready to reactivate. The report is filed but not acted upon. Carlo Semenza dies in October 1961.
📊
1961–1963
Reservoir filling continues
SADE continues filling the reservoir through 1962–1963. The mountain creeps; movement reaches 3.5 cm per day in late September 1963. Workers are evacuated September 26.
💥
10:39 p.m., October 9, 1963
The slide — 45 seconds
A 2-km-long wedge of Mount Toc, 260 million m³, slides into the full reservoir at ~110 km/h. Air-blast and water-wave reach 250 m above the crest. Approximately 50 million m³ of water overtops the still-intact dam.
🔥
~10:43 p.m.
Longarone obliterated
The wave reaches Longarone four minutes later. The town of 1,348 has its center scoured to bedrock. ~1,450 die in Longarone alone; total death toll across multiple villages is approximately 2,000. Many bodies are never identified.
1968–1971
L'Aquila trial
SADE engineers and ENEL officials are tried for negligent disaster. Convictions; reduced sentences on appeal. Italian Supreme Court (1971) confirms convictions for several. The reservoir is permanently drained; the dam still stands today as a memorial.
📝
Tina Merlin

Journalist for the Communist daily L'Unità who wrote alarming articles in 1959–1961 about Vajont's geology. SADE sued her for libel; she was acquitted. After the disaster, her reportage became the basis for Marco Paolini's stage monologue.

🎥
Marco Paolini

Italian playwright whose 1993 monologue Il racconto del Vajont, broadcast live on RAI in October 1997 from atop the dam, was watched by 3.5 million viewers and made the disaster a national memory.

🏠
Longarone

The valley town largely scoured to bedrock by the wave. Rebuilt in modernist style by Giovanni Michelucci. The new town's church, designed by Michelucci, is shaped to recall the wave that destroyed the old.

👨🏻‍🔬
Edoardo Semenza

Carlo Semenza's son, a geologist who mapped the slide perimeter in 1960 and recognized the paleo-landslide. His warnings preceded Müller's by months. He spent the rest of his career documenting the case.

⚖ Place in the Failure Lineage

The unique case where the dam itself was vindicated and the engineers convicted. After Malpasset (geology), Vajont escalated to a wider lesson: that the reservoir's flanks, not just the dam's foundation, must be evaluated. Every modern reservoir study now includes paleo-landslide mapping — a discipline born from Mount Toc.

4

Banqiao Dam — The Worst Dam Disaster in History

Henan, China, August 8, 1975 • Typhoon Nina Triggers a Cascade of 62 Failures

The Banqiao Dam was built in 1952–1953 by the People's Republic of China during a national push for water control. Nicknamed the "iron dam," it was designed for a 1,000-year storm. In August 1975, Typhoon Nina stalled over Henan province and dumped 1,631 millimeters of rain in three days, including 701 mm in a single 24-hour period — a 2,000-year event. Banqiao and the smaller Shimantan Dam failed within hours of each other. Their collapse triggered a cascade through the watershed: 62 dams failed in total. The official Chinese death toll, suppressed for two decades, was around 26,000 from the floods themselves and another ~145,000 from the famine and epidemic that followed. It is the deadliest dam failure in human history.

🌏

Chen Xing — The Hydrologist Who Was Ignored

1917–? • Chinese water engineer

Chen Xing had warned in the early 1950s that Henan's dam-building program lacked sufficient sluice capacity and that simultaneous reservoir flooding was inevitable. He was politically demoted as a "right-deviationist" during the Great Leap Forward. After the disaster, he was rehabilitated and headed the rebuilding commission — a vindication 22 years late.

"The dam will hold. The Party has built it well."
— Local Henan official, August 7, 1975, refusing to authorize emergency releases. Sluice gates were too few and partly blocked; communication lines were down by the time orders could come.
🏗
1951–1953
Construction during national water-control push
Banqiao is one of dozens of dams built across central China to control the Huai River basin and provide hydroelectricity. Designed for a 1-in-1,000-year storm with 12 sluices.
1955–1956
Sluice capacity halved by political directive
During design revisions under Soviet advisors, the sluice count is reduced from 12 to 5. Chen Xing protests; he is demoted. The reduced sluices are inadequate for the design storm.
🌧
August 5–7, 1975
Typhoon Nina stalls over Henan
Typhoon Nina, weakened to tropical depression by terrain, stalls over the Huai watershed. 1,631 mm of rain falls in three days. On August 7 alone, 701 mm falls in 24 hours — double the design event.
📣
August 7, 22:00
Communication failures, denied release request
Banqiao's chief engineer requests permission to open all sluices for emergency drawdown. Communications fail; permission arrives too late. Telephone lines collapse with the rain.
💥
1:00 a.m., August 8, 1975
Banqiao & Shimantan fail simultaneously
Both dams overtop and fail. The Shimantan reservoir empties in 5.5 hours, releasing 120 million m³. Banqiao releases 600 million m³ in 6 hours, generating a wave 10 km wide and 3–7 m deep moving at ~50 km/h.
🌊
August 8–9, 1975
Cascade of 62 dam failures
The flood wave overwhelms downstream reservoirs. 60 more dams across Henan fail. ~26,000 die directly in the floods. PLA helicopters drop airdropped rations to ~11 million stranded survivors.
August–September 1975
Famine & epidemic raise toll
Subsequent famine, dysentery, and typhus kill an estimated 145,000 more in the affected zone. Total death toll estimates run from 26,000 (China) to 240,000 (foreign reconstructions); the consensus figure is roughly 170,000.
📖
1995
Disaster declassified
China's Communist Party permits public discussion of Banqiao for the first time. Yi Si's reports are published. The dam is rebuilt by 1993 with a fully restored 12-sluice configuration.
🔔
Yi Si

Chinese journalist whose 1998 essay "The World's Most Catastrophic Dam Failures" first publicly described Banqiao. Reprinted internationally as appendix to Patricia Adams's The River Dragon Has Come.

👨🏻‍🔬
Qian Zhengying

China's first female Minister of Water Resources (1975–1988). Took office days after Banqiao. Pushed dam-safety reforms; later defended the Three Gorges project against critics.

🌍
People's Liberation Army

Mobilized 30,000 soldiers and dropped 12,470 tons of supplies by air to the cut-off survivor population of ~11 million. The largest emergency airdrop in PLA history.

📖
Patricia Adams

Canadian environmental researcher whose 1998 book The River Dragon Has Come documented Banqiao for English-language readers and put the death toll into international perspective.

🔴
Outcome: Worst Dam Disaster in Human History
Roughly 170,000 dead, 11 million displaced. Suppressed by China's government for two decades, the disaster is now openly discussed but rarely emphasized in domestic media. Banqiao was rebuilt by 1993 with a redesigned spillway. China later cited the experience to justify the Three Gorges Dam's massive flood-control capacity.

⚖ Place in the Failure Lineage

Vajont was a single landslide; Banqiao was a system collapse. Where St. Francis or Malpasset killed hundreds, Banqiao killed in the hundred-thousands because dozens of dams failed in cascade. It demonstrated that dam systems must be designed against simultaneous extreme events — a lesson now baked into hydrologic design standards worldwide.

5

Teton Dam — Failure Caught on Camera

Madison County, Idaho, June 5, 1976 • The Bureau of Reclamation's Worst Day

Teton Dam was built by the U.S. Bureau of Reclamation in 1972–1975 as part of the Lower Teton Project for irrigation and flood control along the Snake River. On Saturday morning, June 5, 1976 — the reservoir filling for the first time, just three weeks from completion — small wet spots appeared on the downstream face. By 7:30 a.m., they were leaks. By 11:00 they were torrents. At 11:57:30 a.m., still photographers and the Bureau's own movie cameras captured the dam giving way live as the right abutment caved into a sinkhole. The flood killed 11 people, drowned 13,000 cattle, and devastated Rexburg and Wilford, Idaho. Eleven dead is small by global standards; Teton's significance is that it was the first major U.S. dam failure to be photographed in real time, instantly debunking the Bureau's reputation for invincibility.

🏔

U.S. Bureau of Reclamation — The Builder

Founded 1902 • Hoover Dam, Grand Coulee, Glen Canyon… and Teton

The Bureau, founded under Theodore Roosevelt's 1902 Newlands Act, had built nearly every great American dam of the 20th century. Teton was its first complete failure. The General Accounting Office and an independent panel found that dispersive clays in the foundation rhyolite, combined with inadequate grouting of fissures, allowed internal erosion (piping) to undermine the embankment.

"It's going! It's going! It's gone!"
— Eyewitness on the canyon rim, captured on Bureau of Reclamation 16-mm color film, 11:57 a.m., June 5, 1976. The film of the actual collapse is one of the most-viewed engineering disaster recordings ever made.
📝
1971–1972
Design & environmental opposition
The Sierra Club and Friends of the Earth sue to stop construction, citing rhyolite foundation concerns and seismicity. The lawsuit fails. Construction begins June 1972.
🏗
1972–November 1975
Construction
Earthfill dam, 305 feet high. Foundation rhyolite is heavily fissured; the design relies on a "key trench" of impervious clay penetrating into the rock and on extensive grout curtain.
🌊
October 1975 – June 1976
Reservoir filling
First filling proceeds at 4 ft/day — faster than the Bureau's own internal guidelines. By June 4, the reservoir is at 5,301.7 ft, 64 feet below the planned crest but rising fast.
👀
7:30 a.m., June 5, 1976
First wet spots reported
Workers note small wet spots and seepage on the downstream face near the right abutment. By 9:30 a.m. the flow has eroded a visible channel; by 10:30 it is muddy — the diagnostic sign of internal erosion (piping).
🚫
11:30 a.m., June 5
Bulldozers attempt to plug
Bulldozers are dispatched to push fill into the growing breach. Two narrowly escape; their operators jump clear as the equipment is swallowed by a sinkhole.
📷
11:57:30 a.m., June 5
Failure on film
The right abutment collapses into the breach. The dam is fully overtopped within minutes. Photographers Eunice Olsen and a Bureau of Reclamation 16 mm camera capture the collapse in color — the first major dam failure ever recorded live.
🏠
June 5–7, 1976
Flooding of Rexburg, Wilford, Sugar City
The 80-billion-gallon flood reaches Rexburg in two hours, Wilford and Sugar City within four. ~25,000 are evacuated. 11 die; 13,000 cattle drown. Property damage: ~$2 billion (1976 dollars).
📚
December 1976
Independent panel report
Wallace L. Chadwick's independent panel concludes failure was due to internal erosion of the dispersive clay core through fissured rhyolite. The Bureau is excoriated for inadequate filter design and aggressive filling schedule.
📷
Eunice Olsen

Wilford schoolteacher whose color photographs of the morning seepage and the final collapse, taken from the canyon rim with her family's still camera, became the iconic visual record.

👨🏻‍🔨
Wallace L. Chadwick

Chair of the independent review panel. His report (December 1976) became a teaching standard in dam-safety curricula worldwide and triggered the National Dam Safety Act of 1979.

👨🏻🏠
Rexburg residents

~10,000 evacuated. Most made it out before the flood arrived because police, alerted by the Bureau, drove door-to-door starting at 11:30 a.m. The two-hour warning saved hundreds of lives.

🐟
13,000 cattle

The disaster's most numerous victims. Snake River pasture stock that could not be moved in time. Their carcasses contaminated the floodplain for weeks.

⚖ Place in the Failure Lineage

If Vajont killed dam-engineering credibility in Italy, Teton killed it in the United States. The disaster ended the Bureau of Reclamation's golden age and effectively ended the era of new American mega-dams. The famous color film — played in every dam-engineering course since — brought failure home to a public that had previously trusted concrete.

6

Edenville & Sanford Dams — Twin Failures in the Climate Era

Midland County, Michigan, May 19, 2020 • A 500-Year Storm in a Pandemic Spring

On May 19, 2020 — in the middle of the COVID-19 pandemic shutdown — the 96-year-old Edenville Dam in central Michigan failed during a 500-year rainfall event, sending a wall of water down the Tittabawassee River that overtopped and breached the Sanford Dam 12 miles downstream. The Federal Energy Regulatory Commission had revoked Edenville's hydroelectric license in 2018 specifically because its spillway capacity was inadequate — only half the volume required to pass the probable maximum flood. The dam's owner, Boyce Hydro, contested the order and continued to operate. No one died, thanks to a 90-minute downstream warning, but 11,000 were evacuated and downtown Midland was inundated to the second floor of the historic Dow Diamond ballpark. The case became a textbook for climate-era dam-safety failures.

🌧

Boyce Hydro — The Owner

2006–2020 • Lee W. Mueller, principal owner

Boyce Hydro Power LLC purchased the Edenville and three other dams in 2006. After FERC's 2018 license revocation for inadequate spillway capacity, the company filed bankruptcy days after the failure. A federal investigation faulted both Boyce's management and decades of regulatory inaction by Michigan EGLE.

"It is going to fail. The only question is: when, and how badly?"
— FERC engineering staff memo on Edenville Dam, dated September 10, 2018, recommending license revocation. The order was issued the same week. Boyce Hydro continued to operate the dam under Michigan state oversight until May 19, 2020.
🏗
1924–1925
Construction
Edenville Dam built 1924 to dam the Tittabawassee for hydroelectric power; Sanford Dam built 1925 downstream. Both are earth-fill dams owned originally by Wolverine Power Company.
1999–2018
Repeated FERC warnings
FERC inspections from 1999 onward identify Edenville's spillway as undersized: capable of passing only ~50% of the design probable maximum flood. Owner Boyce Hydro repeatedly fails to fund upgrades.
September 10, 2018
FERC revokes license
Federal Energy Regulatory Commission revokes Edenville's hydroelectric license, an unprecedented action. Operation transfers to Michigan EGLE under state authority. State permits continued reservoir storage.
💊
March–May 2020
Pandemic context
Michigan is in COVID-19 shutdown. Stay-at-home orders are active. Disaster planning is severely degraded; Red Cross and FEMA shelters operate at reduced capacity.
🌧
May 17–19, 2020
500-year rainfall
A frontal system stalls over central Michigan. 4–7 inches of rain falls across the Tittabawassee watershed in 48 hours. Reservoirs rise rapidly. Edenville's inadequate spillways cannot keep up.
💥
5:46 p.m., May 19, 2020
Edenville fails
Edenville Dam's east embankment fails by overtopping. Wixom Lake (Edenville reservoir) drains in roughly 90 minutes, sending a surge down the Tittabawassee toward Sanford Dam.
🌊
~7:45 p.m., May 19
Sanford overtops & breaches
Sanford Dam, just 12 miles downstream, is overtopped by the surge from Edenville. It breaches in stages overnight. Combined flood reaches downtown Midland (population 42,000) by morning.
📙
May 21, 2020
Independent forensic team
A six-person independent forensic engineering team (chaired by John W. France) is named under federal supervision. Their May 2022 report finds liquefaction, not overtopping, was the primary failure mode — surprising the engineering community and reshaping the lessons.
🎮
Dow Diamond / Midland

The historic Great Lakes Loons ballpark and downtown Midland flooded to the second-floor mezzanine. Dow Chemical's headquarters complex sustained ~$50M in damages. The town has spent $200M+ on recovery and dam-removal redesign.

📝
John W. France

Chaired the independent forensic team. His 538-page May 2022 report identified static liquefaction of the dam's upstream slope as the failure trigger — an unexpected mechanism that has reshaped earthfill dam-safety reviews nationwide.

💰
Boyce Hydro

Filed Chapter 11 bankruptcy on July 31, 2020, ten weeks after the failure. Subsequent litigation has consolidated downstream property owners' claims; settlement has not been reached as of 2026.

🛡
Michigan EGLE

State Environment, Great Lakes, and Energy department. Took over Edenville oversight after FERC's 2018 revocation. Subsequent state legislative reforms (2021) increased dam-inspection budgets sevenfold.

🟪
Outcome: Climate-Era Dam Safety Reform
No fatalities — but the disaster exposed the United States' aging dam infrastructure: about 92,000 dams, average age 60 years, an estimated $93 billion in deferred maintenance. The 2021 Bipartisan Infrastructure Law allocated $3 billion for dam rehabilitation, the largest such investment since the National Dam Safety Act. Edenville's failure mode (static liquefaction) is now standard in earthfill review checklists.

⚖ Place in the Failure Lineage

The 21st-century template: aging private infrastructure plus increasing 500-year storms (which arrive every 10 to 30 years now in some watersheds) plus regulatory inaction. Where the 20th-century failures were about geology or engineering judgment, Edenville is about ownership, climate, and political will — a template likely to repeat as the global stock of dams ages past its design life.

Comparative Analysis

DamYearTypeFailure ModeDeathsVolume ReleasedSeverity
St. Francis1928Concrete arch-gravityFoundation rock failure~431+47 million m³Major
Malpasset1959Double-curvature archAbutment fault wedge42350 million m³Major
Vajont1963Arch (intact)Reservoir landslide overtopping~2,00050 million m³ waveCatastrophic
Banqiao1975EarthfillStorm overtopping (cascade of 62)~170,000+~7 billion m³ totalCatastrophic
Teton1976EarthfillInternal erosion / piping11250 million m³Minor (deaths)
Edenville/Sanford2020Earthfill (twin)Static liquefaction0~120 million m³Minor (deaths)

Key Patterns Across Dam Failures

⛩ Geology, Not Concrete

St. Francis, Malpasset, Vajont, and Teton all had perfectly adequate dam structures: it was the rock, soil, or surrounding slopes that gave way. The 20th-century lesson: a dam is only as strong as its foundation.

🔔 Warning Signs Ignored

Each disaster was preceded by visible signs — leaks (St. Francis), creep (Vajont), seepage (Teton), inspection reports (Edenville). The common cause is institutional pressure to keep operating.

🌧 First Filling Risk

St. Francis, Malpasset, Vajont, and Teton all failed during or immediately after their first complete reservoir filling. Modern protocols require slow filling with extensive monitoring; these case studies are the reason.

🏠 Scale of Death Depends on Warning Time

Banqiao (no warning, ~170k dead) versus Teton (90-min warning, 11 dead) versus Edenville (advance evacuation, 0 dead). Modern dam safety lives or dies by emergency action plans and rapid notification.

📚 Each Disaster, One New Law

St. Francis → California Dam Safety Act 1929. Malpasset → French Decree 1968. Vajont → criminal liability for engineers. Teton → National Dam Safety Act 1979. Edenville → 2021 Bipartisan Infrastructure Law allocations.

🌡 The Aging Inventory

The world holds ~58,000 large dams. 25%+ are over 50 years old. Climate change is increasing extreme rainfall events. The combination predicts that 21st-century failures — like Edenville — will be more frequent than the 20th century's record suggests.

Interactive Mega Timeline — All Six Failures Compared

Drag to pan • Scroll to zoom • Hover for details