How does ice form on wind turbine blades?

Atmospheric icing reaches a blade by two routes: in-cloud icing, where supercooled droplets in cloud or fog freeze on contact and build rime, and precipitation icing, where freezing rain or drizzle produces glaze, and wet snow its own heavy, wet accretion.

Rime ice and glaze ice

Rime is the usual form of ice on wind turbine blades at hilltop and ridge sites, where moist air is lifted into cloud. IEA Wind Task 19's 2011 expert group recommendations for cold climate projects, drawing on ISO 12494, give soft rime a density of 200 to 600 kg/m³ and hard rime 600 to 900 kg/m³, forming from 0 °C down to about minus 20 °C. It grows into the wind, on the leading edge and thickest towards the tip, where relative velocity and droplet collection are highest; a field study of 50 m blades after a 30 hour event, published in PNAS, photographed ice up to 0.3 m thick near the tips. That is the classic rime ice blade accretion: a rough, opaque horn that destroys lift long before it adds much mass. Glaze, from freezing precipitation or wet in-cloud icing between 0 °C and about minus 6 °C, is the densest ice a blade carries at around 900 kg/m³; smooth and clear, it is harder to see and sheds in heavier pieces.

Meteorological icing, instrumental icing and the IEA ice classes

In the IEA definitions, meteorological icing is the period during which ice is actively forming; instrumental icing is the period during which ice remains on the structure or disturbs the turbine. The second can be far longer than the first, so weather-model icing hours understate what the rotor experiences. The IEA recommendations classify sites on both and attach an expected loss to each class:

  • Class 1: meteorological icing up to 0.5% of the year, instrumental icing below 1.5%, production loss 0 to 0.5% of annual output.
  • Class 2: 0.5 to 3% and 1 to 9%, loss 0.5 to 5%.
  • Class 3: 3 to 5% and 6 to 15%, loss 3 to 12%.
  • Class 4: 5 to 10% and 10 to 30%, loss 10 to 25%.
  • Class 5: above 10% and above 20%, loss above 20%.

Where a site lands in two classes the recommendations say to use the highest, and at least one year of on-site ice measurement before a loss estimate is relied on.

How much production does wind turbine blade icing cost?

It depends on the site, the turbine and the year. The classes above put the icing production loss at a few percent of annual output for moderate sites, and at 10 to 25% or more for severe ones. Operational data support the ranges: a DNV analysis of more than 30 operating wind farms in the Nordic region reports losses above 50% in winter months and above 10% on an annual basis, and within a single event the PNAS study measured power loss of up to 80%. The icing losses wind turbine owners record also swing from one winter to the next, so two winters say little about the twenty-year mean. For an operating asset the useful number comes from SCADA; IEA Task 19 publishes an open-source method that identifies icing periods from the power curve and temperature and sums the deficit.

What does ice do to blade loads, vibration and safety?

Ice on wind turbine blades changes the aerodynamics, the mass distribution and the dynamics of the blade at once. The IEA recommendations list aerodynamic imbalance and, with long exposure, increased component loading among the direct effects, and note that a once-per-revolution torque variation is typical even of lightly iced rotors (source). The blades never ice identically, so the rotor runs out of balance, which the drivetrain and tower carry as fatigue load; German turbine guidance, as summarised in the IEA Available Technologies report, assumes an ice mass imbalance for seven days a year in fatigue assessment. A 2026 model-scale study in Scientific Reports summarises the dynamic side: ice alters stiffness and damping, shifts natural frequencies, and shedding introduces an impulsive excitation that can affect fatigue life. An iced aerofoil also stalls earlier, one reason iced turbines trip on vibration alarms.

Ice throw, ice fall and safety stops

The IEA distinguishes ice fall, from a turbine at standstill or idling, from ice throw, from an operating rotor (IEA Task 19 ice risk recommendations, 2022). The Canadian Renewable Energy Association's guide notes that thrown pieces can weigh several kilograms and gives the widely used, conservative throw distance of 1.5 times the sum of rotor diameter and hub height. The IEA report uses a localised individual risk of one in a million per year as the threshold for negligible risk. For one turbine type at 78 m hub height it reports measured counts of ice pieces of 50 g or more from about 80 a year at class 1 to more than 3,200 at class 5 for an idling, unheated turbine, and an extrapolated three times that when operating. Regulators commonly require a stop on ice detection near roads or buildings, and in some jurisdictions a turbine stopped on an ice alarm may not restart until the blades have been checked visually, in person or by camera. In our experience the stops, rather than the aerodynamic penalty, can be the larger part of the icing production loss.

How is ice detected on a wind turbine?

Detection is where most practical weaknesses lie. The IEA Task 19 ice detection guidelines classify methods by where they measure, nacelle or rotor, and what they measure.

Ice detection wind turbine methods compared

The power curve method compares actual with expected output, gated by a nacelle temperature below a threshold near freezing. It needs no extra sensor and is, in the guidelines' words, widely adopted by wind turbine OEMs, but its sensitivity is limited by the natural scatter of the power curve and it can overstate icing, stopping turbines in conditions that were not severe. Nacelle-mounted sensors correlate with rotor icing but are not the same thing, since a large rotor dips into cloud the nacelle never reaches. Blade-mounted sensors, including those tracking a shift in blade natural frequency, measure the rotor itself and are preferred where structural protection is the purpose. Cameras serve detection and the visual check before restart. Owners should ask which of these the controller uses, since it decides how much lost production is false alarm.

Blade heating, anti-icing blades and what a retrofit has to earn

The IEA distinguishes anti-icing, which keeps ice from forming while the turbine runs, from de-icing, which lets ice build, stops the turbine and removes it (Available Technologies, 2018). What owners call anti-icing blades are, in practice, blades with an active heating system and a controller deciding when to run it.

Heating, coatings and operational measures

Two blade heating technologies dominate. Hot air systems circulate heated air inside the blade; the IEA report rates them simple and robust but mostly limited to de-icing and inefficient, since heat must pass through the laminate to reach the surface. Electro-thermal systems place heating elements at the surface along the leading edge, with better power consumption and spanwise control but increased lightning risk and expensive repairs if damaged. Consumption must be netted off the recovered production, and the PNAS author warns that conventional heating risks damaging composite material through overheating. Passive coatings are attractive on paper, but the IEA report calls them unproven on turbines and leaves them off its retrofit list, and the PNAS author reports that coatings alone have not eliminated ice at the leading edge. Operational measures complete the set: preventive stops at the onset of an event, curtailment to reduce loads or throw distance, and heated wind sensors.

A worked example with stated assumptions

The following uses assumed values, not measured or market data. Take a 4 MW turbine at an assumed 35% net capacity factor, about 12,260 MWh a year. Assume a 5% icing loss, inside the class 2 to 3 range, or about 610 MWh; assume the heating system recovers 60% of that after its own consumption and the stops it needs, about 370 MWh; and assume electricity worth EUR 60 per MWh. The recovered value is about EUR 22,000 per turbine per year, around EUR 220,000 over an assumed ten remaining years before discounting, and the retrofit, including downtime, maintenance, lightning protection, warranty and certification, must cost less than that. At a class 1 loss of 0.5% the same sums give about EUR 2,200 a year, at which a retrofit will rarely pay; at a class 4 loss of 15%, about EUR 66,000, at which it may well.

How icing damage is distinguished from other causes in a failure investigation

Ice rarely breaks a blade by itself. When a blade fails in winter, the investigation has to establish whether icing was the initiating cause, a contributing load, or simply the season in which an existing defect finally propagated, because each answer places the cost with a different party. We work through a short list:

  • What do SCADA and alarm logs show before failure: temperature, power curve deviation, vibration alarms, stop codes, and any ice detector or heating record?
  • Does the damage match an ice mechanism? Impact from shed ice is local; leading edge erosion is progressive and rain-driven; lightning leaves attachment marks at receptors.
  • If the blade is heated, is there overheating, a burnt heating element, or a lightning path through the heating conductors?
  • Where did the fracture start, and was there a laminate wrinkle, a bond line defect or a prior repair there? An ice-induced load that finds a manufacturing defect points to the defect, not the ice.
  • What did the last pre-season inspection record at that location?

The sequence, not the season, is the finding. A wrinkle that fails under an ice imbalance load in February is a manufacturing question, and if the wrinkle is serial it is a fleet question, whatever the weather.

Frequently asked questions

What is wind turbine blade icing?

Wind turbine blade icing is the accretion of atmospheric ice on the rotor blades, as rime from supercooled cloud droplets or as glaze from freezing rain and drizzle. It reduces lift, adds uneven mass and can trigger vibration alarms and safety stops.

How much production do icing losses cost on a wind turbine?

Published ranges run from under 0.5% of annual output at lightly iced sites to more than 20% at severe ones. Nordic operational data show losses above 10% in some years and above 50% in single winter months. The reliable number for a turbine comes from its own SCADA record over several winters.

Can ice on wind turbine blades damage the blade?

Directly, seldom. Indirectly, yes: uneven ice creates imbalance and stall-induced vibration that add fatigue load, shed ice can strike the tower or a following blade, and heating systems bring thermal and lightning risks. Whether icing caused a failure or revealed an existing defect is a question for a root cause investigation.

Does blade heating pay for itself?

In our experience it can at severe sites and seldom does at lightly iced ones; the IEA recommendations mark anti- or de-icing as recommended from class 3 upwards, worth considering at class 2 and not needed for energy at class 1. The decision rests on the measured icing loss, the tariff, the remaining life and the full cost of the retrofit.

How far can ice throw reach?

The conservative rule of thumb used in Canada and in German building regulation is 1.5 times the sum of hub height and rotor diameter, and thrown pieces can weigh several kilograms. Site-specific trajectory modelling usually gives a smaller area, and regulators often require a stop on ice detection where roads or buildings lie within it.

How Apex Wind can help

We are an independent blade engineering consultancy in Denmark. We sell no heating systems, coatings, sensors or repairs and take no fee from anyone who does, so our view of a retrofit or a loss estimate is an engineering view only. In a technical due diligence we check that the icing loss in the model comes from the site's own data and that the detection and protection strategy is what the controller actually does. When a blade has failed in an icing season, our blade failure investigation establishes the sequence and states whether ice was the cause, a contributor, or only the weather. If either question is on your desk, contact us.