Wind energy is expanding into regions where tropical cyclones are not a rare anomaly but a defining feature of the climate — the U.S. Gulf and Atlantic coasts, the Taiwan Strait, the South China Sea, Vietnam, the Philippines, Japan, Korea. With fifteen co-authors from national labs, universities, and industry, we have just published a paper in Wind Energy Science that sets out where the science stands, where the engineering practice stands, and the gaps between them. This is the technical brief.
The under-design / over-design trap
The wind industry's design standards were built around the North Sea — an extra-tropical wind climate that looks almost nothing like a tropical cyclone (TC). When a developer takes those standards and applies them to a typhoon-exposed site, two failure modes open up at the same time.
Under-design. The default turbulence models (Mann, Kaimal) assume neutral boundary layers with stochastic gusts that average out cleanly over 10 minutes. They were calibrated on North-Sea-like conditions. Tropical cyclones don't behave that way — they have coherent structures larger than the rotor, rapid directional shifts of 90° or more, and turbulence driven by air–sea heat fluxes that standard spectra don't capture. The IEC standards' new T-Class sets a 57 m/s reference wind, but it says almost nothing about the character of that wind. A turbine designed strictly to T-Class can still be hit by loads it was never asked to survive.
Over-design. The opposite is just as costly. If a developer responds to TC uncertainty by stacking blanket conservative factors on top of T-Class assumptions, the result is a turbine that's too heavy, too expensive, and too slow to install for the project to pencil out. Markets like Vietnam or the Philippines don't have margin for that. Cost without resilience is just cost.
The way out of this trap is not to pick a side. It's to get the physics right at the specific site, so the right margin can be set — neither catastrophically thin nor financially fatal. That's a research problem the industry has not solved.
What a tropical cyclone actually does
Tropical cyclones are coupled atmospheric–oceanic systems. The schematic below shows the structure of the event: an eyewall containing the most intense winds, spiral rainbands of organised convection, a turbulent surface boundary layer interacting with breaking waves and storm surge, all advected across the ocean by the storm's translation. Almost none of these features is captured by design conditions developed for extra-tropical storms.
The modeling gap
Cyclone turbulence is fundamentally different from extra-tropical turbulence in five practical ways for turbine design:
Higher turbulence intensities. The eyewall is a turbulent factory. Standard intensity assumptions are too low.
Larger coherent structures. Convective cells and mesoscale vortices in rainbands can be larger than the rotor diameter. Isotropic-turbulence assumptions fail.
Rapid direction shifts. During storm passage the wind direction can swing 90° or more in a short window. If the turbine is yaw-locked while idling, sustained crosswind loading takes over — and the structure was not designed for that.
Wind–wave coupling. Waves are driven by the storm's wind, but they also feed back into the boundary layer through surface roughness and spray. You cannot treat them as independent inputs — yet that is effectively what many design workflows do.
Idling does not mean unloaded. During shutdown, turbines parked into the wind still see substantial loads — and "parked" is ill-defined when the wind direction is shifting that fast.
What "right" simulation looks like
The research community is starting to be able to model this properly. The figure below shows a coupled mesoscale → microscale simulation of Hurricane Laura (2020): Laura's large-scale structure develops in a 4.5 km grid; a vortex-following grid at 1.5 km tracks the storm; and turbulence-resolving microscale domains at 33 m and 11 m capture what an individual turbine actually sees. This kind of coupled simulation is research-grade today. The remaining work — and a central message of the paper — is translating it into design loads that an engineer can defend.
The standards gap
Design wave heights and IEC partial safety factors are currently calibrated for extra-tropical conditions. Two issues compound each other. First, the 50-year design wave height curve in TC basins diverges from the extra-tropical fit at long return periods — so a structure designed against the extra-tropical curve is exposed to wave heights it was not asked to absorb. Second, the IEC load partial safety factors used to convert characteristic loads into design loads were calibrated assuming an extra-tropical hazard distribution. In a TC regime, those factors need to be re-tuned to deliver equivalent structural reliability. The current values, used as-is, do not.
Physics alongside the statistics
There is a parallel gap on the risk side. Catastrophe models fit historical loss data to a portfolio of named-storm exposures — an approach that is invaluable where the record is deep. In tropical-cyclone basins it is not: the catalog is short (a few decades), the climate has moved underneath it, and the assets being built today are not the ones the historical losses were calibrated against. A 15 MW floating turbine in a typhoon basin does not behave like a 2 MW fixed-bottom did a decade ago. This is where a physics-based layer complements the statistical view rather than replacing it — and where a cat modeller or risk consultant most needs an engineering input they can stand behind.
The paper argues for a different framing: treat TC risk as a probabilistic, physics-based calculation — convolving hazard intensity, structural response, damage probability, and consequence. The hazard side needs synthetic catalogs of 10,000 to 100,000 years of TC activity, because the tail of the distribution is where the design decisions live. The fragility side needs site-specific aeroelastic simulations on the actual turbine. The consequence side needs operational and financial loss modeling. Without all three, what comes out is a number, not a defensible risk estimate.
Eight gaps we identified
The paper sets out eight specific research priorities:
1. Observational capacity. Long-range scanning lidars and Doppler radars at turbine-relevant heights (20–350 m), particularly offshore. Current measurements are sparse, temporally limited, and rarely over open water.
2. Synthetic and probabilistic storm models. Validated against offshore TCs, with future-climate scenarios baked in. Catalogs spanning 10–100k synthetic years.
3. High-fidelity simulations and reduced-order models. GPU-based LES that can feed engineering surrogates. The tiered chain: high-fidelity → ROM → standards.
4. Turbulence and load frameworks. Beyond Mann/Kaimal. Beyond 10-minute averages. Capturing shear, veer, coherence across the rotor layer and around the storm.
5. Loads, standards, mitigation. Probabilistic safety factors tuned for tropical basins. Wave–current misalignment treated explicitly. Tuned mass dampers, advanced control, novel substructures, evaluated systematically. Critically: TC fatigue treated as its own limit state, not spread over lifetime.
6. Breaking waves and ocean currents. An under-characterized loading mechanism for monopiles in shallow-to-intermediate depths. Current design relies on simplified representations.
7. A definition of TC survivability. No consensus exists across academia, developers, regulators, and the insurance industry on what "surviving" means. Structural survival? Operational readiness within X days? Allowable damage state by component? This has to be defined before risk transfer can be defined.
8. Integrated risk assessment frameworks. Hazard × fragility × consequence, tied together. TC-specific reliability targets. Downtime losses inside the probabilistic frame. A transparent basis on which developers, owners, lenders, insurers, and brokers can balance upfront investment against long-term resilience.
Why this matters now
Auction Round 7 in the UK signals a renewed push for offshore wind. Japan is pursuing offshore wind at scale in typhoon-exposed waters. The Philippines is starting from essentially zero. The U.S. Atlantic and Gulf are scaling. Developers, owners, lenders, investors, insurers, and brokers are getting more sophisticated about what they expect to see — and the question they ask is increasingly specific: which turbines on which sites at which return periods?
The under-/over-design trap is the practical answer to the practical question. Getting the physics right — site by site, component by component — is the path out.
Reference wind speeds offshore need reconsidering
The thread running through all eight priorities is a single, uncomfortable conclusion: the reference wind conditions that anchor offshore design were set for an extra-tropical world. A single reference wind speed and an extra-tropical turbulence spectrum cannot represent a basin where the design-driving event is a tropical cyclone — nor a climate in which the return periods behind those references are drifting year on year. The T-Class 57 m/s reference is a start, but a number without the character of the wind behind it is not a design basis. Reconsidering the reference wind speed offshore, basin by basin, is not an academic nicety: it is the difference between a turbine that is quietly under-designed and one that is priced out of the market.
Where Parametrica comes in
This is the early, hard part of a project we exist for — resolving the site-specific hazard with high-fidelity, coupled simulation, the kind shown for Hurricane Laura above, and turning it into design loads and risk numbers a developer, insurer or broker can act on. It is also the foundation for the parametric insurance models we are building for tropical and extratropical cyclones, or extreme weather more broadly: once the hazard is resolved from the physics up, a transparent, defensible trigger follows. Bespoke engineering today; parametric products next.
Citation
Deskos, G., Wang, J., Arwade, S., Fisher, M., Hirth, B., Guo Larsén, X., Lundquist, J. K., Myers, A., Pang, W., Pringle, W. J., Rogers, R., Sanchez-Gomez, M., Sun, C., Yamaguchi, A., Veers, P. (2026). Grand Challenges in Designing Resilient Wind Energy Systems in Areas Prone to Tropical Cyclones. Wind Energy Science Discussions, 2026, 1–44. 10.5194/wes-2026-32