Kansai International Airport
The Sinking Island: How a ¥1.45 Trillion Airport Outran the Geology It Was Built On
Kansai International Airport opened on schedule in 1994 as a triumph of Japanese engineering — the world's first major airport built entirely on an artificial island. Ground-settlement behaviour proved substantially more severe than early projections anticipated.: the soft Osaka Bay clay compressed far faster and deeper than laboratory models predicted, and the first island had settled more than eight metres before a single flight landed. Decades of jacks, seawall raising and monitoring have kept the airport operating, but parts of the islands are now projected to reach sea level by the 2050s. PIA analysis identifies geological forecasting and long-term settlement contingency planning as central lessons from the project., not of construction — and an enduring lesson in pricing geological uncertainty.
Watch: Japan Built an Airport on the Sea. Now It's Sinking.
Prefer the episode page? Watch this investigation as a documentary episode →
Source: investigation key facts.
| Measure | Value |
|---|---|
| Original budget | ≈¥1.45 trillion for Phase 1 (commonly cited, ~US$14bn at historical rates) |
| Final cost | ≈US$20bn by 2008 including Island II works and settlement countermeasures (commonly cited figure); remediation spending continues |
Contents
What Happened
Kansai was a rational response to an impossible site problem. Osaka's existing Itami airport was boxed in by suburbs, noise rules and litigation; after the land-acquisition battles of Narita, building five kilometres offshore in Osaka Bay promised 24-hour operation with no displaced residents. The price was the seabed: up to 20 metres of water over deep layers of soft alluvial and Pleistocene clay. Engineers knew the islands would settle as the clay consolidated under hundreds of millions of cubic metres of fill, and designed for it — over two million vertical sand drains to accelerate consolidation, and a terminal on 900 columns mounted on hydraulic jacks. The islands were planned to stabilise and remain above +4 metres Chart Datum Level, the threshold needed to keep waves from overtopping the seawall. The forecasts failed almost immediately. Landfill for Island I began in 1987; by 1990 the island had settled about 8.2 metres against roughly 5.7 metres predicted — a pattern that would persist. The airport opened on 4 September 1994 settling at around 50 centimetres per year, far above design assumptions. Engineers excavated beneath the terminal, inserted iron plates under the jacks and raised columns in stages; at least US$150 million went into raising and strengthening the seawall. The airport kept working — runways were repaved in forgiving asphalt rather than rigid concrete — but the margin designed into the island was being consumed decades early. The definitive audit came from outside. In 2015, University of Illinois engineers Gholamreza Mesri and Jeffrey Funk published a settlement analysis in the ASCE's geotechnical journal: by December 2012 average seabed settlement exceeded 12.9 metres on Island I and 14.2 metres on Island II — the newer island, opened in 2007, was sinking faster than the first despite everything learned. They projected parts of Island II reaching the +4 metre warning level around 2023 and sea level around January 2056, with Island I following around 2067. Operator data cited in 2025 reporting puts cumulative average settlement at about 13.7 metres on Island I and 17.5 metres on Island II, with Island II still sinking at roughly 21 centimetres a year across measured points. Cost figures vary by what is counted: Phase 1 is commonly cited at about ¥1.45 trillion (roughly US$14 billion), and cumulative spending including Island II and remediation was reported at around US$20 billion by 2008, with continuing countermeasure costs since. The operational consequence arrived in September 2018, when Typhoon Jebi flooded runways and underground plant rooms and a drifting tanker severed the only bridge, stranding thousands of passengers — a demonstration that a sinking airport's real vulnerability is the shrinking margin between its runways and the sea.
Why It Matters
Kansai is the canonical case of a megaproject whose central engineering assumption — the rate of ground consolidation — was wrong from day one and could never be fully corrected afterwards. Every subsequent decision (jacks, seawalls, asphalt runways, groundwater management) has been an exercise in buying time within a shrinking safety envelope, at cumulative costs that rival the original construction budget. The case matters beyond aviation: reclaimed-land airports, ports and coastal districts are being built worldwide on comparable geology, and Kansai shows that laboratory-derived soil models can diverge from field reality by 40 per cent or more at exactly the scale where the error is unaffordable. It also demonstrates that 'on time and on budget' at opening is not the same as delivered — the true cost profile of this project is still accruing three decades later.
-
proposal
1968
Osaka region begins planning a replacement for the congested, curfew-bound Itami airport; offshore siting chosen to avoid noise and land-acquisition conflict.
-
construction
1987
Landfill construction of Island I begins in 18-20m of water over soft Osaka Bay clay, with over two million vertical sand drains to accelerate consolidation.
-
warning
1990
Island I has settled about 8.2 metres against roughly 5.7 metres predicted; engineers begin corrective measures beneath the terminal.
-
opening
1994-09-04
Kansai International Airport opens on schedule — the first major airport on a wholly artificial island — settling at around 50cm per year.
-
construction
1999
Construction of Island II begins, applying lessons from Island I's faster-than-expected settlement.
-
delay
2000-08
At least US$150 million is committed to raising and strengthening the seawall as the +4m CDL design margin is consumed early.
-
opening
2007-08
Island II's second runway opens; the newer island soon proves to be settling faster than the first.
-
failure
2012-12
Measured average seabed settlement reaches 12.9m (Island I) and 14.2m (Island II), far beyond the design trajectory.
-
inquiry
2015
Mesri and Funk publish their ASCE settlement analysis, projecting parts of Island II at the +4m warning level by 2023 and at sea level around January 2056.
-
approval
2016-04
Kansai Airports (ORIX–VINCI consortium) takes over KIX and Itami operations under a 44-year concession, inheriting the settlement liability.
-
failure
2018-09
Typhoon Jebi floods runways and underground plant rooms; a drifting tanker severs the access bridge, stranding thousands of passengers.
-
warning
2023-09
The window Mesri and Funk projected for parts of Island II reaching the +4m CDL warning level arrives.
-
inquiry
2024-12
Operator data cited in regional reporting shows cumulative average settlement of about 13.7m on Island I and 17.5m on Island II, with Island II still sinking about 21cm per year.
Root cause through the Decision Quality Model™ lens
Is the decision being made in the right context with clear objectives?
Have multiple viable options been generated and considered?
Is the decision based on relevant, reliable data — not assumptions?
Are the trade-offs between competing priorities explicitly understood?
Is the logic connecting information to conclusions valid?
Are the stakeholders committed to implementing the decision?
Decision Quality Model™
How Good Decisions Produce Good Outcomes — explore the framework →
Future Readiness Score™
Is Your Organisation Ready for What Comes Next? — explore the framework →
Project Failure Pyramid™
The 4-Level Diagnostic for Cascading Failures — explore the framework →
Continue Investigating
The Weekly Brief — one investigation, one executive lesson and one project warning sign each week.
Subscribe freeLessons for Leaders
Point Estimates Kill Projects Resting on Untestable Assumptions
Kansai's settlement forecast came from laboratory consolidation tests that understated field behaviour by tens of per cent. Where one physical assumption carries the whole design, model the full probability distribution and design to the adverse tail, not the mean.
Build in the Option to Be Wrong
The 900 jacked columns and asphalt runways were the saving features of Kansai — deliberate adjustability that let operators buy time. Design for adjustability in the variable you trust least; it is the cheapest insurance available at design stage and priceless afterwards.
Replicating a Design Does Not Replicate Its Behaviour
Island II, built a decade later with all of Island I's lessons, settled faster (14.2m versus 12.9m by 2012). Geology is not transferable between sites a few hundred metres apart. Treat each foundation as a new engineering problem, not a copy-paste.
Opening Day Is Not Delivery
Kansai opened on schedule and was celebrated as a Civil Engineering Monument of the Millennium — while already consuming its design margin at several times the planned rate. Whole-life cost and performance monitoring must be part of the business case, not an afterthought.
Degraded Margins Convert Hazards into Disasters
Typhoon Jebi in 2018 flooded an airport whose settlement had eroded its freeboard, and severed its single bridge. Risk assessments must use the asset's current degraded state, not its as-designed state, when sizing climate and hazard exposure.
Independent Technical Audit Beats Internal Reassurance
The most consequential analysis of Kansai's trajectory came from Mesri and Funk, external academics, decades after opening — and was initially contested by project insiders. Programmes with compounding physical risk need periodic independent re-baselining against original assumptions.
Executive Recommendations
Commission probabilistic geotechnical baselines before fixing design — Require settlement and ground models to be expressed as distributions with field pilot embankments where feasible; freeze design elevation only after early instrumentation confirms the model.
Instrument and re-baseline continuously against original forecasts — Publish settlement telemetry against the design curve on a fixed cadence, with automatic triggers for design review when divergence exceeds defined bands.
Fund whole-life countermeasures in the original business case — Ring-fence a remediation and monitoring endowment at financial close so that future operators are not negotiating for maintenance capital against annual budgets.
Red-team single-point physical assumptions — For any project resting on one dominant physical forecast, mandate an adversarial external review before financial commitment, and revisit it at defined intervals.
Align concession structures with inherited physical liabilities — Where assets are privatised or concessioned, allocate long-run settlement and climate risk explicitly in the contract, with data access obligations for the operator.
PMOS Intelligence
Preview — illustrative assessment; PMOS is in development“Kansai's engineers did not act recklessly — the design was state of the art, and the airport has never closed for settlement. The failure is subtler and more common: a single point estimate for a deeply uncertain variable (consolidation rate) was embedded in a fixed design, and the contingency philosophy assumed the forecast was right. When reality ran 40 per cent faster, there was no cheap way back. The lesson for any programme resting on one untestable physical assumption: buy the option to be wrong before you pour the fill, because afterwards the only options left are jacks and prayers.”Ramesh's Insider Take — opinion
Documentary Evidence
Settlement of the Kansai International Airport Islands
Mesri and Funk; Journal of Geotechnical and Geoenvironmental Engineering 141(2) · 2015
By December 2012 average seabed settlement exceeded 12.9m on Island I and 14.2m on Island II; parts of Island II were projected to reach the +4m CDL warning level around September 2023 and sea level around January 2056, with Island I around 2067.
AeroTime analysis citing Smithsonian Air & Space reporting on Kansai settlement history
AeroTime analysis citing Smithsonian Air & Space reporting on Kansai settlement history
Island I settled about 27 feet (8.2m) by 1990 against a predicted 19 feet (5.7m); at least $150 million was spent on the seawall; total costs reached about $20 billion by 2008.
This Week in Asia / South China Morning Post syndication via The Star, June 2025, citing operator data
This Week in Asia / South China Morning Post syndication via The Star · 2025-06
Cumulative average settlement stands at about 13.66m on Island I and 17.47m on Island II; in 2024 measured annual settlement averaged 6cm (17 points, Island I) and 21cm (54 points, Island II).
Tomorrow.City feature on Kansai's subsidence and climate exposure
Tomorrow.City feature on Kansai's subsidence and climate exposure
Typhoon Jebi in September 2018 flooded the airport and a tanker collision severed the access bridge, stranding thousands of passengers; critical plant was located below the flooded level.
Mesri and Funk 2015 abstract, University of Illinois IDEALS repository
Mesri and Funk 2015 abstract · 2015
The airport islands were designed to remain above +4m Chart Datum Level to prevent wave overtopping of the seawall, using over two million vertical sand drains and roughly 400 million cubic metres of fill across both islands.
Everything from this investigation
One investigation. Many outputs. Everything derived from this case, connected in one place.
Read & download
- Executive summaryRead the executive summary
- PDF briefDownload the PDF brief
- TimelineExplore the timeline
- EvidenceReview the documentary evidence
- SourcesSee the source list
- Framework analysisRead the framework analysis
Watch & listen
- DocumentaryWatch the documentary
- Episode pageVisit the episode page
- Related documentariesBerlin Brandenburg Airport
- ShortsNot published
- Podcast episodePlanned
Share & follow
- LinkedIn articlePlanned
- Medium articlePlanned
- Substack articlePlanned
- Downloadable templatesPlanned
- NewsletterGet this investigation by email
Connected research
- CompaniesKansai Airports
- CountriesJapan
- IndustriesAviation & Airports
- Related investigationsBoeing 737 MAX · Denver Airport Baggage System · China's Ghost Cities
- Settlement of the Kansai International Airport Islands (Mesri and Funk, 2015, J. Geotech. Geoenviron. Eng. 141(2)) University of Illinois IDEALS repository (open copy of ASCE paper) · 2015 · peer-reviewed-engineering-paper
- Kansai Airports official website (operator of KIX, Itami and Kobe) Kansai Airports Co., Ltd. (ORIX–VINCI consortium) · 2026 · official-operator
- Kansai International Airport Land Company (asset owner) official site KIAC, Government of Japan-affiliated land company · 2026 · official-asset-owner
- Japan's Kansai Airport sinking due to subsidence issues This Week in Asia / South China Morning Post, syndicated by The Star · 2025-06-22 · news-analysis
- Japan's incredible sinking airport: How long does Kansai have? AeroTime · 2025-02-07 · aviation-trade-press
- Is Kansai Airport sinking? What the future holds Tomorrow.City · 2026 · urban-infrastructure-analysis
- Kansai: The Airport That Sank (technical explainer with primary-source reference list) Pager Power · 2024-01-16 · technical-consultancy-explainer
Frequently Asked Questions
By December 2012 average seabed settlement exceeded 12.9 metres on Island I and 14.2 metres on Island II. Operator data cited in 2025 reporting puts cumulative averages around 13.7m and 17.5m respectively. Island I settled about 8.2m by 1990 — against roughly 5.7m predicted — before the airport even opened.
The artificial islands sit on up to 20 metres of soft alluvial and Pleistocene clay on the floor of Osaka Bay. The immense weight of roughly 400 million cubic metres of reclamation fill compresses the clay — a process called consolidation — which proceeded far faster than laboratory tests predicted. Over two million vertical sand drains accelerated but could not contain the settlement.
Mesri and Funk's 2015 peer-reviewed analysis projects parts of Island II reaching sea level around January 2056 and Island I around 2067, with the +4m safety threshold on Island II passed around 2023. Japanese engineers have contested elements of that prognosis, noting ongoing countermeasures. What is not disputed is that the design margin has been consumed decades early.
Phase 1 is commonly cited at about ¥1.45 trillion (roughly US$14 billion at historical rates). Including Island II and settlement countermeasures, cumulative spending was reported at around US$20 billion by 2008, and remediation — jacks, seawall works, monitoring — continues to add cost today.
In September 2018, Typhoon Jebi — the strongest typhoon to hit Japan in 25 years — flooded Kansai's runways and underground plant rooms, while a drifting tanker crashed into the only road and rail bridge, stranding thousands of passengers. The episode exposed how decades of settlement had eroded the airport's freeboard margin against storm surge.
