Carburettor Icing: Why It Happens and How to Beat It

Carburettor icing stops engines that are in perfect mechanical order, and it is one of the most common causes of power loss in light piston aeroplanes. It is not a winter problem, it does not need cloud or rain, and it will happily form on a warm summer afternoon with the outside air temperature above +20 °C. The reason is simple: a carburettor manufactures its own cold. This article explains the physics, the conditions, the symptoms in both fixed-pitch and constant-speed aeroplanes, and the one thing pilots most often get wrong about carb heat. (American material spells it carburetor; the aeroplane behaves the same either way.)

Why a Carburettor Manufactures Its Own Ice

A carburettor is, in effect, a small refrigerator bolted to the induction system. Two entirely separate physical processes cool the charge inside it, and they add together.

1. The pressure drop. Air accelerating through the venturi — and then squeezing past a partly closed throttle butterfly — loses static pressure. The expansion happens too quickly for meaningful heat transfer, so it is effectively adiabatic and the temperature falls with the pressure — the same pressure and velocity relationship you meet in our PPL Principles of Flight guide. At normal cruise settings this accounts for a drop of the order of 5 °C. At small throttle openings, where the butterfly rather than the venturi is the real restriction, the local drop is larger.

2. Fuel vaporisation. This is the dominant term. Petrol enters the airstream as a spray of liquid droplets and must change state to a vapour before it will burn. The latent heat required for that change comes out of the surrounding air and metal, and nothing else. Depending on fuel volatility and mixture strength, vaporisation alone can remove up to about 25 °C.

Together the two mechanisms can cool the charge by as much as 30 °C below ambient — the figure quoted in the national safety literature, and larger drops are on record — and they do it within a few centimetres of induction tract. Any water vapour carried in that air now condenses onto the cold metal, and where the metal is below 0 °C it freezes there: on the venturi throat, on the butterfly and its spindle, and on every edge and corner downstream of the fuel discharge nozzle.

The deposit does not have to be thick to matter. It reshapes the venturi, chokes the throat and progressively strangles the mass of air reaching the cylinders. Power falls. The mixture goes rich as well. The ice builds in the venturi throat and around the butterfly — at and downstream of the fuel discharge nozzle — so the air is choked off after the fuel has already been added; the narrowed throat keeps the depression at the main jet up rather than letting it fall in step with the airflow, and ice on the throttle plate acts on the mixture much as closing the throttle does. Fuel flow holds up while air mass flow falls, so the fuel/air ratio richens — which is why a carb-iced engine usually runs rough well before it runs down.

Why Carburettor Icing Is Worst on Warm, Damp Days

A widespread misconception is that carburettor icing is a cold-weather phenomenon. The opposite is closer to the truth, and the reason is absolute humidity: the warmer the air, the higher the saturation vapour pressure, so warm air can carry far more water vapour before it condenses. Take a day at +25 °C with a dew point of +18 °C. That air holds a great deal of water per kilogram, and a charge starting at +25 °C and cooled by up to 30 °C arrives at about −5 °C — comfortably below freezing, with a generous supply of condensate to deposit on the metal. A dry winter day at −5 °C has almost nothing available to freeze out.

That is why serious carburettor icing has been recorded at ambient temperatures approaching +30 °C, and at humidities most pilots would describe as comfortable rather than damp. Clear blue sky is not a defence.

The standard planning tool is the carburettor icing probability chart — the one reproduced in the UK CAA's Safety Sense Leaflet 14, Piston Engine Icing, in the equivalent national publications and in most PPL textbooks. Dry-bulb temperature runs along one axis and dew point along the other; plot the intersection and read off the band it lands in. There are four bands, in increasing order of risk:

• Light icing at cruise or descent power.
• Serious icing at descent power.
• Moderate icing at cruise power, serious icing at descent power.
• Serious icing at any power setting.

The band candidates most often forget is the second — serious icing at descent power on its own — because it reaches a long way into the warm, comfortable-feeling part of the chart. Two further features are worth memorising:

• The envelope extends a long way above +25 °C. There is no ambient temperature at which you can stop thinking about it.
• A very large part of the chart carries at least a descent-power risk. Low power is the common factor almost everywhere.

In the air you rarely have a chart to hand, so use the temperature/dew point spread as the practical proxy: the smaller the gap, the higher the relative humidity and the greater the risk. The METAR gives you both figures before you start; our PPL meteorology questions cover dew point and spread from the exam side.

High-risk situations in practice: the first flight of the morning off a wet grass strip, flying after rain or in drizzle, in and just below cloud, over water, and on hazy, humid summer afternoons. Very cold, very dry air — well below about −10 °C — contains so little water vapour that refrigeration icing becomes unlikely, but the threat there simply changes character to impact ice.

Fuel Ice, Throttle Ice and Impact Ice

Examiners rarely ask about "carb ice" in isolation. They ask you to distinguish the three forms of induction icing, because each forms in a different place, in different conditions, and is treated differently.

Fuel ice is the one that breaks the assumption about temperature, because it is driven by vaporisation rather than ambient conditions. Throttle ice needs air that is already near freezing, since the pressure drop alone is a modest cooler; in practice the two form together on a carburetted engine and are cleared together. A fuel-injected engine has a throttle butterfly too and it cools in exactly the same way, but with no fuel evaporating there the temperature drop is small, so throttle ice is uncommon — and if it does form there is no carburettor heat to clear it, only the alternate air source.

Impact ice is a different animal. It is ordinary airframe-type icing, accreting on the intake and filter, and it blocks the induction system from the outside in. It affects every piston engine regardless of how fuel is delivered, which is why fuel-injected aeroplanes still carry an alternate air source.

Why the Low-Power Descent Is the Dangerous Regime

If you remember one flight regime, remember this one. A glide or low-power descent stacks four separate problems on top of each other.

1. The cooling is at its worst. With the throttle closed or nearly closed, the pressure drop across the butterfly is at its maximum, and so is the local temperature drop.
2. A small deposit is a large restriction. The gap around a nearly closed butterfly is tiny. Ice that would be trivial at cruise can block a substantial fraction of it, and in the worst case can seize the butterfly in position.
3. The engine is not making heat. An idling engine has a cool exhaust, and the exhaust is where carburettor heat comes from. The one system that would clear the ice is at its least effective precisely when the ice is forming fastest.
4. You are not watching. Descending is what you asked the aeroplane to do and the RPM is low anyway, so the symptoms are hidden. The ice is discovered when you open the throttle to level off or go around and the power does not arrive.

This is the regime of the long descent from height, the practice forced landing, the glide approach and the orbit at low power while waiting for traffic — and it is why so many carb icing events end in a power loss at the worst possible moment rather than in the cruise. A long descent in cold, damp air is doubly awkward: the induction system is at its coldest and, as our guide to cold weather altimetry explains, the altimeter is over-reading at the same time.

The defence is preventive, not reactive. Select carburettor heat to hot before you close the throttle, not after the engine has already cooled and the ice has begun. Leave it hot for the descent as your Pilot's Operating Handbook directs, and open the throttle periodically — typically every 1,000 ft in a prolonged descent, or as your POH and your school's procedure specify — both to keep the induction system warm and to confirm the engine is still delivering.

Carburettor Icing Symptoms: Fixed-Pitch versus Constant-Speed

Carburettor icing is insidious rather than dramatic. It builds over minutes, and the indications drift slowly enough that the eye adapts to them. What you look for depends entirely on the propeller.

Fixed-pitch propeller. The RPM is your flow meter. With the throttle untouched, a slow, unexplained decay in RPM is the classic first sign. In level flight it may show up instead as a gradual loss of airspeed, or as the discovery that you have unconsciously fed in throttle to hold height. As the restriction worsens the mixture goes rich and the engine begins to run rough, and vibration follows. Left alone, the sequence ends in a complete power loss.

Constant-speed propeller. The RPM will not move at first, because the governor fines off the blade pitch to hold the RPM you selected as the power falls away. The indication is therefore a slow decay in manifold pressure at a fixed throttle setting. Only when the blades reach the fine pitch stops does the governor run out of authority and the RPM start to fall as well — and by that stage you have lost a great deal of power. Rough running arrives later still. This is a genuinely examinable distinction: fixed-pitch loses RPM, constant-speed loses manifold pressure.

Two habits catch it early. First, note the RPM or manifold pressure you actually set in the cruise and compare against that figure at every FREDA-type check, rather than trusting your impression of it. Second, treat any unexplained roughness or power decay as carb ice until proved otherwise — applying heat unnecessarily costs a few seconds and a little power; not applying it can cost the engine.

Beware the masking effect too: if you correct a slow power loss by opening the throttle in small increments, the aeroplane behaves normally right up to the moment the throttle is fully open and the ice keeps building.

Carb Heat: Full Hot, and Leave It There

The carburettor heat control moves a valve that stops drawing air through the normal filtered intake and instead takes it from a shroud, or muff, wrapped around part of the exhaust. The prescriptive certification standard for these systems — FAR/CS 23.1093(a)(1), still the reference for the legacy trainer fleet — required a preheater giving a 90 °F (50 °C) heat rise at 75 % of maximum continuous power on a sea-level engine with a conventional venturi carburettor, and 120 °F (67 °C) on an altitude engine, demonstrated in air free of visible moisture at −1 °C. Types certified under CS-23 Amendment 5 meet the performance-based ice protection requirement of CS 23.2415 instead. Either way the practical point is the same: enough heat to keep the charge above freezing in most conditions, and to melt an existing deposit if there is exhaust heat available.

Select Full Hot, Not Partial

Unless your aeroplane is fitted with a carburettor air temperature gauge and the POH gives you a procedure for using it, partial heat is a trap. It can raise the induction temperature out of a range that was too cold to form ice and straight into the range that forms it most readily, and it can melt impact ice into water that refreezes further downstream. Full hot, or cold. Nothing in between.

The initial roughness is confirmation, not a warning. If ice is present, applying full heat makes things briefly worse: the melting ice passes through the engine as water, the running becomes rough and the RPM or manifold pressure may drop further. This can last anything from a few seconds to the best part of a minute with a heavy build-up. That roughness is the proof that you found ice. Hold the heat on and wait. The engine will then smooth out and settle at a value higher than before you applied it — still slightly below the original cold-air figure, because hot air is less dense, but clearly recovered. Selecting cold because "it made it worse" is the single commonest and most dangerous error on this subject.

If there was no ice, you get a small immediate drop and nothing else: on a typical trainer of the order of 50–100 RPM, steady, with no roughness and no recovery. That drop is the expected loss from the warmer, less dense and consequently richer charge — no ice involved. Check your own type's expected figure in the POH.

Why the Run-Up Carb Heat Check Exists

Selecting hot during the power checks and seeing the expected clean RPM drop confirms that the valve is moving and the hot air source works. It is also the first ice check of the flight: a rise in RPM, or a drop that recovers above where you started, means you have already picked up ice while taxiing or idling on a damp morning, and the heat should stay hot until the RPM stabilises. In icing conditions, or after a long hold, apply heat again before line-up and return it to cold before you open the throttle.

Why Carb Heat Is Cold for Take-Off

Three reasons, and you should be able to give all three:

• Hot air is less dense, so it costs you power, and take-off is the phase where you need all of it.
• The hot air is unfiltered. On the ground, and particularly on grass, gravel or dusty surfaces, that means ingesting abrasive debris straight into the engine.
• A hot charge at high power raises the risk of detonation.

There is also little to gain: at full throttle the butterfly is wide open, the pressure drop is small and refrigeration icing is at its least likely. As always, the POH is the authority — a small number of types give different guidance for severe conditions.

One further point: applying heat enriches the mixture, so at altitude the handbook may call for re-leaning with heat applied. Some modern installations also reduce the risk by design, siting the carburettors where they pick up engine heat — but never assume immunity. If there is a carb heat control in the cockpit, the aeroplane can ice.

The Carb Heat Drill

The whole of the above, in the order you would use it:

1. Suspect ice on any unexplained power decay, RPM or manifold pressure loss, or roughness.
2. Carburettor heat full hot. Not partial.
3. Expect it to get worse first. Roughness and a further drop mean you have found ice.
4. Leave it hot and wait — a few seconds, or up to about a minute with a heavy build-up.
5. Confirm recovery against the figure you noted in the cruise: above the iced figure, a little below your original cold-air setting.
6. If it does not recover, treat it as a partial power loss. Keep the heat hot, work through the POH rough-running drills, and get the aeroplane somewhere you can land while the engine is still giving you something.

Fuel-Injected Engines: Impact Ice and Vapour Lock

A fuel-injected engine has no carburettor venturi in which fuel and air mix and cool together. Fuel is metered and delivered to injector nozzles at the inlet ports, close to hot cylinder heads, so the vaporisation cooling happens downstream of any restriction and in a part of the engine that is already hot. There is no cold, wet venturi throat for ice to grow in. That is why injected aeroplanes have no carb heat control, and why refrigeration icing is not a threat to them.

They are not immune to induction icing as a whole, however.

Impact ice. In visible moisture at or below freezing, ice accretes on the intake lip and blocks the air filter exactly as it would on a carburetted aeroplane. The remedy is alternate air: a source of warmer, unfiltered air taken from inside the cowling. Many installations use a spring-loaded door that opens automatically on the pressure differential once the normal intake blocks, often with a manual selection as well. Because alternate air is warmer and unfiltered, selecting it costs a little manifold pressure — and on an automatic system, an unexplained drop in manifold pressure in icing conditions may be your first clue that the door has opened itself.

Vapour lock. The characteristic injected-engine problem is fuel vaporising in the lines or pump on the supply side, where a vapour bubble interrupts the flow the pump is trying to deliver. It is a hot-weather and hot-engine phenomenon: a heat-soaked engine bay after a short turnround, a hot restart, high ambient temperatures, or a climb where falling pressure lowers the boiling point of the fuel. Symptoms include difficult hot starts, fluctuating fuel flow, rough running and power loss. The remedy is procedural rather than mechanical — the electric boost pump and the handbook's hot-start technique — and it is entirely POH-specific.

So the exam-ready summary is: carburettor icing is a carburettor problem; impact icing is everybody's problem; vapour lock is the injected engine's own speciality.

What the PPL Exam Expects You to Know

Carburettor icing sits in Aircraft General Knowledge, but it surfaces in Meteorology and in Principles of Flight too, via the venturi. The points that come up again and again:

• The two cooling mechanisms — the pressure drop through the venturi and past the throttle butterfly, and the latent heat absorbed by fuel vaporisation — and the combined drop of as much as 30 °C.
• That icing is possible at ambient temperatures well above freezing, with high humidity and no visible moisture.
• How to read the icing probability chart from dry-bulb temperature and dew point, that it carries four bands, and that low power widens the risk band dramatically.
• The distinction between fuel ice, throttle ice and impact ice.
• Fixed-pitch: falling RPM. Constant-speed: falling manifold pressure. Rough running in both.
• Full hot, not partial; initial roughness confirms ice; carb heat cold for take-off because of power loss, unfiltered air and detonation risk.
• Injected engines do not suffer refrigeration icing but do suffer impact icing and vapour lock, and use alternate air rather than carb heat.

If you want the venturi physics properly grounded, our PPL Principles of Flight guide covers the pressure and velocity relationship that underlies both the wing and the carburettor.