Skip to content 
On August 16, 2024, one week after the Voepass Flight 2283 accident, Guilherme Lima da Silva, founder and CEO of ATS4i, was interviewed by Claudio Lucchesi on the Revista Asas channel. The ATR 72-500, flying from Cascavel to Guarulhos, crashed in Vinhedo, São Paulo state, on August 9, killing all 62 people on board. The interview did not discuss the causes of the accident: it covered the physics of in-flight icing and how aircraft engineering deals with it.
Note: the final word on the accident belongs to CENIPA. Both the interview and the 2024 LinkedIn post focused on the physics and engineering of icing, with no conclusions about the causes of the accident.
Cover image: frame from the interview (Revista Asas, YouTube).
Video in Portuguese, with subtitles: Revista Asas Cultura e História da Aviação, Aug 16, 2024, about 36 minutes.
Summary
- Supercooled droplets are in a metastable equilibrium and freeze when they hit the aircraft: wing, stabilizer, engine inlet, probes and antennas.
- Rime ice is white and opaque. Glaze ice is transparent, runs back before freezing and forms ice horns, which are very harmful to lift.
- Ice on the upper surface of the wing reduces lift and can lead to a stall. On the lower surface it mainly adds drag.
- Supercooled large droplets (SLD, 40 to 100 µm) characterize severe icing: water runs back aft of the protected area and forms an ice ridge.
- Aircraft certified to FAR 25 Appendix O demonstrate that they detect the condition and exit it. Earlier designs must avoid it.
- To escape, the general rule is to descend and gain speed: in icing, “speed is life”.
Icing on the route on August 9, 2024
At the time, Guilherme posted on LinkedIn the icing conditions on the day and time of the flight, with a note: “This is not a judgement or a link between the accident and the ice condition existent. There is no official information enough to take conclusions. It is a presentation of a fact: there was ice in the route.”
The forecast icing area was huge, which is uncommon in Brazil at this severity. The conditions may have been caused by the meeting of cold and warm air masses. A pilot who flew in the region that day reported severe icing to air traffic control (O Globo, Aug 9, 2024, in Portuguese).
What does “severe” icing mean? The FAA defines icing intensity by its effect on the aircraft, and this is the scale pilots use in pilot reports (PIREPs):
| Intensity | Ice accumulation | Reference rate (outer wing) | Recommended action |
|---|---|---|---|
| Trace | Ice becomes noticeable. The rate of accumulation is slightly greater than the rate of sublimation. | < ¼ in/h (6 mm/h) | Consider exiting before it gets worse |
| Light | May create a problem if flight is prolonged (over 1 h). Requires occasional cycling of manual deicing systems. | ¼ to 1 in/h (0.6 to 2.5 cm/h) | Consider exiting the condition |
| Moderate | Requires frequent cycling of manual deicing systems. Anything more than a short encounter is potentially hazardous. | 1 to 3 in/h (2.5 to 7.5 cm/h) | Exit as soon as possible |
| Severe | Ice protection systems fail to remove the accumulation, and ice accumulates in areas not normally prone to icing, such as aft of protected surfaces. | > 3 in/h (7.5 cm/h) | Immediate exit, required by regulation |
Source: FAA, AC 91-74B and Order JO 7110.10, §8-1-7.
The FAA itself stresses that severe icing is aircraft dependent, as are the other categories: severe icing may occur at any accumulation rate when the ice exceeds the tolerance of that aircraft. The current definition of severe icing includes accumulation aft of protected surfaces, which is the ice ridge explained below. The older scales, based only on cloud liquid water content, are no longer used. Forecast products such as the Aviation Weather Center’s CIP/FIP describe the expected icing intensity in the atmosphere (from trace to “heavy”), not the severity for a specific aircraft (AWC). This is also why it is unknown which criterion was used in the severe icing reports of that day.
What was said in the interview
Why ice forms
In clouds, water droplets can stay liquid below 0 °C. They are in a metastable equilibrium, like a cart resting in a small basin at the top of a roller coaster: the impact with a surface is enough to make them freeze. That is why ice appears where the air hits the aircraft: the leading edges of the wing and horizontal stabilizer, engine inlets, antennas, probes and air intakes.
Rime ice and glaze ice
Rime ice is white and opaque, like the ice in old freezers. Droplets freeze on impact and trap air, which lowers the ice density. It forms a kind of shell on the leading edge.
Glaze ice forms close to 0 °C. Kinetic heating of the droplet and the latent heat released during solidification delay freezing, and water runs over the wing as a film or rivulets before freezing downstream of the impact point. The result is ice horns, which destroy lift. Research at the University of São Paulo showed that the horn tends to form at the laminar-turbulent transition of the boundary layer, where heat transfer increases in a step. The horn then changes the flow and the droplet impingement and feeds itself.
Because it is transparent, glaze is hard to see. Clear ice on top of the wing before takeoff is one of the cases that most affects pilots, which is why the walk-around inspection is essential.
Ice on top of and under the wing
In aviation jargon, ice on top of the wing means loss of lift and ice under the wing means drag. The upper surface accelerates the air and lowers the pressure; ice there destroys the boundary layer and can lead to a stall. Underneath, the effect is similar to carrying an external store: with enough power, the aircraft compensates.
SLD, severe icing and the limit of pneumatic de-icing
Supercooled large droplets (SLD) measure 40 to 100 µm, against about 20 µm for a typical cloud droplet, and carry much more water. SLD, found in freezing drizzle and freezing rain, are among the severe icing conditions.
With so much water arriving, latent heat prevents it from freezing on the leading edge. On aircraft with pneumatic de-icing boots, the boot inflates in cycles and breaks the ice on the protected area, but the water runs back and freezes aft of it, forming a ridge the boot cannot remove. This ridge separates the boundary layer, and the separation can reach the aileron, flap or elevator, which lose effectiveness.
New aircraft are certified to FAR 25 Appendix O, created after icing accidents in the 1990s such as the ATR 72 accident at Roselawn, USA, in 1994. Certification demonstrates that the aircraft detects the condition and has time to exit it, not that it can keep flying in it. Earlier designs must avoid these clouds and, if they enter them, exit correctly, using escape procedures developed by NASA. SLD alerts come from meteorology and air traffic control, since the onboard weather radar is simpler.
How to exit: descend and gain speed
The aerodynamic force is proportional to ½ ρ V² times the lift coefficient (CL). With CL degraded by ice, you compensate with density or speed. Descending increases air density, and accelerating increases lift with the square of the speed. Hence the saying “in icing, speed is life”. There are exceptions, such as tailplane icing, where the procedure is different, but they are rarer.
Protection systems and ice shedding
Wings, stabilizers, engine inlets, windshields, pressure and angle-of-attack probes and propellers are protected. The most common systems are hot bleed-air thermal systems (piccolo tube), electric heating and pneumatic boots. The choice depends on the energy available. Besides lift and drag, there is the shedding hazard: on aircraft with rear-mounted engines, ice that breaks off an antenna or radome can be ingested by the engines. If simulation shows the shed piece is too large, the design adds heating to reduce the ice size.
Aircraft size and collection efficiency
Criticality depends more on aircraft size than on propulsion. Thinner wings have higher collection efficiency: a larger share of the water ahead of the aircraft hits the surface. On very thick wings, such as those of large airliners, many droplets are deflected by the flow, and the lift margin is also larger.
eVTOLs and ice detectors
Electric aircraft have no pneumatic source: the energy for ice protection comes from the battery used to fly. FAA and EASA have been discussing certification for escape from icing conditions, with an icing envelope adapted to the ceiling of these aircraft. An automatic ice detector earns credit in escape time, because action starts earlier; a visual cue watched by the pilot does not earn the same credit, since the time to look and decide is also exposure time.
What CENIPA’s final report found
In July 2026, CENIPA, Brazil’s aeronautical accident investigation center, published the final report of the investigation (PDF in English, in Portuguese). According to press coverage, the report concludes that the accident resulted from a combination of factors rather than a single cause. The factors highlighted include:
- weather conditions favorable to severe icing during the flight;
- faults in the de-icing system and in the ice detector, also recorded on previous flights, with the system switched on and off several times during the flight;
- repeated degraded-performance and low-speed alerts that did not lead to corrective action;
- nose-up inputs during the stall, contrary to the recovery procedure;
- informal maintenance and defect-reporting practices at the airline (“normalization of deviance”) and insufficient oversight by ANAC, Brazil’s civil aviation authority.
The report includes recommendations to EASA and ATR, among them a review of degraded-performance procedures, and to ANAC. For technical details, the reference is the report itself.
The final word on the accident belongs to CENIPA. The 2024 interview and LinkedIn post made no link between icing and the accident: they covered the physics and engineering involved. The topics discussed, such as severe icing with SLD, the limit of pneumatic de-icing, speed and ice detection, help to understand the terms that appear in the report.
Frequently asked questions
Does the interview discuss the cause of the accident?
No. Revista Asas made clear at the opening that the investigation was CENIPA’s role. The conversation covered icing physics and ice protection engineering.
Can I watch it with English subtitles?
Yes. The interview is in Portuguese and the player above opens with subtitles on. If English is not listed, open Settings, then Subtitles/CC, then Auto-translate, and choose English.
Does the ATR 72 have an ice protection system?
Yes. The leading edges of the wings and stabilizers have pneumatic de-icing boots, and the propellers, windshields and probes are electrically heated.
What is SLD?
Supercooled large droplets, 40 to 100 µm in diameter, typical of freezing drizzle and freezing rain. They run back aft of the protected area before freezing and are one of the severe icing conditions.
Where can I check icing forecasts?
From meteorological and air traffic services. The US Aviation Weather Center publishes icing forecasts with SLD and keeps historical data.
Read more
- Full interview on the Revista Asas channel (YouTube, in Portuguese)
- CENIPA final report on PS-VPB (PDF)
- Ice formation analysis: ATS4i services
- FAPESP PIPE: iceAccretionFoam, from 2D beta to 3D MVP (2026)
- ATS4i Presents iceAccretionFoam at the SAE AC‑9C Committee in Italy (2025)
- Numerical simulation for ice protection systems on an airfoil (2023)
- “The Icing Engineering Process”: SUNY Korea lecture (2022)
- Guilherme’s LinkedIn post on icing along the route (Aug 2024)
- ATS4i publications
Need to assess icing in your project? ATS4i provides ice formation analysis, ice protection system design and simulation, and certification support. Talk to an engineer.
