The aerodynamic effects of accreted ice on the continued safe flight of an aircraft are a complex subject because of the many forms that such ice accretion can take. In certain circumstances, very little surface roughness is required to generate significant aerodynamic effects; as ice-load accumulates, there is often no aerodynamic warning of a departure from normal performance. Stall warning systems are designed to operate in relation to the angle of attack of a clean aeroplane and cannot be relied upon to activate usefully in the case of an ice-loaded airframe.
This subject is given a comprehensive treatment in an article originally published by Transport Canada and listed below under 'Further Reading'. That article examines the way in which the crucial balance between lift and drag is affected by ice and considers how ice protection systems work. This SKYbrary article provides a summary of the main points.
The Adverse Aerodynamic Effects of In-flight Icing
The aerodynamic effects of ice on an airfoil are a function of the location of the ice with respect to the airfoil’s pressure distribution, the ratio of the ice shape height over the chord length of the wing (k/C ratio), and the geometry of the ice shape itself. These parameters are very difficult, if not impossible, to quantify when viewing the wing from the cockpit.
Contrary to past thinking, there is no reason to believe that the effects of icing are cumulative in any linear or proportional sense. Larger ice shapes can have substantial effects on lift, drag and pitching moment. However, considerable work over the past sixty years has repeatedly determined that very little surface roughness is required to generate significant aerodynamic effects. In all cases, these effects are dependent on angle-of-attack which, consequently, makes control of the angle-of-attack absolutely critical. Perhaps the most insidious aspect of icing effects is that there is often no aerodynamic warning of a departure from normal performance.
A thin ice surface roughness, often characterized as hoar frost, can cause a very early and abrupt peak on the lift curve, followed by a precipitous drop in lift. The insidious aspect of this is that, up until the peak is reached and flow separation occurs, the lift curve may be quite normal. Consequently, there is no reason to expect aerodynamic warning. Further, the flow separation can occur well in advance of artificial stall warning.
Testing has indicated that the drag curve for hoar frost may yield little degradation until the angle of attack is very near that at which the early stall will develop. Even then, the increase in drag may be easily manageable.
Hoar Frost has been identified as causal in a large number of accidents. Typically, this results from the flight crew either not detecting the thin ice accretion before increasing the angle of attack, or the flight crew delaying operation of the ice protection system until a minimal ice thickness has developed.
Larger ice shapes may create similar effects. In some cases, the lift curve may peak at an angle of attack substantially lower than that for Hoar Frost. However, large drag rises are usually associated with larger ice shapes. These drag increases are also directly related to angle of attack. What is particularly dangerous is the tendency for drag to increase much more rapidly with increasing angle of attack than the pilot is accustomed to. It is very possible for the drag, at higher angles of attack, to exceed the available power.
FAA research has shown that, for aircraft with pneumatic de-ice systems, such effects can be experienced with intercycle ice shapes typical of a mixed rime/glaze condition when the de-ice cycle time was three minutes. Considerably better ice shedding was observed when the de-ice cycle time was reduced to one minute intervals, even in more severe icing conditions. Optimum ice shedding was not achieved until the pneumatic de-ice boot had completed several cycles.