John Dow, icing specialist, Small Airplane Directorate, Federal Aviation Administration (Ret.)

The problem of aerodynamic stall with ice accretion has been a problem for decades and will likely remain a problem for the foreseeable future.

The airplane is certificated for flight in icing!

To comply with the regulations, 14 CFR Part 25 Appendix C describes the icing envelope or conditions. The envelopes define horizontal extent, liquid water content, droplet sizes and temperature. I have yet to see my first airplane rigorously examined for certification in temperatures warmer than approximately -5 degrees C (21 degrees F), which was the prevailing temperature in the nearly catastrophic loss of the Saab 340 at Bathurst.

Testing at warmer temperatures produces large glaze ice and run-back ice, often in shapes, locations, textures and thickness that are impossible to accurately predict with ice shape prediction codes. Accident and incident reports show this part of the envelope is very often the temperature range existing at the time of the event. There is water and ice and, yes, it is difficult to determine if the wing has ice accretion or is just wet, especially if the crewmember must look over his shoulder at a surface that may be 40 feet away. Water at below freezing temperatures strikes the wing and must eject 80 calories per gram to change phase from liquid to solid. That process requires heat flow out that can only occur with a local temperature lower than the water. As the air flows aft over the airfoil it will transit lower temperatures in the region of lower pressure (that's where the lift comes from).

So at static air temperatures near freezing where the stagnation temperature is above freezing, until the water moves aft on the upper surface it will not freeze, then the shapes can be ugly, sharp and large, or sometimes a nearly transparent coating. The supercooled liquid water can also freeze aft of the ice protection system and accordingly cannot be removed. To add insult to injury, the real world infrequently contains some icing clouds that are outside Appendix C. The effect can be even more adverse, and weather conditions conducive to the formation of droplets larger than certification standards appear to have been associated with the Bathurst upset.

Trap set - trap sprung

Usually very little ice can result in a significant increase in stall speed. Landing stall margins by regulation can be no less than 1.3 V stall, so it doesn't take much of an increase in stall speed to get close to the stall angle of attack. A roll maneuver will further erode margin to stall, with or without ice. There is no way the pilot has a clue that the margin to stall has eroded and his stall speed has increased until he gets there. As the ice builds and the drag increases it will bring the airplane closer to the new stall speed unless corrected. An additional factor is the adverse effect of ice accumulation on propeller performance. The crew of the incident airplane did not activate their propeller ice protection system prior to the upset.

Stall recovery - too little too late

Current stall recovery procedures emphasize maintaining pitch attitude while applying power at the first sign of stall (stick shaker). This technique doesn't really work well with ice. Several problems exist: 1) First sign of stall in icing may occur prior to stick shaker, so if the pilot waits for the shaker the airplane may be deeper into the stall regime. 2) Since the drag of the ice is greater than a clean airplane and aerodynamic stall has already occurred, drag is dramatically higher and there may be insufficient excess thrust for the airplane to accelerate. Uncommanded roll in this incident airplane started about 20 seconds before the official pronouncement of stall, but the autopilot was attempting to keep the airplane wings level. Maximum power took 20 seconds to apply after that point. By then the airplane had almost hit the dirt and the airplane pitch increased another 10 degrees into the stall regime.

For these kinds of conditions, disconnect the autopilot at first sign of stall, reduce the angle of attack by promptly applying nose down control input or extending flaps, while applying maximum power. Then recover the airplane.

Payload - payload - payload

Stall speed is the sacred cow of performance. Manufacturers and their customers want the lowest possible stall speed and accordingly test carefully. Takeoff and landing performance is predicated on a safety margin of 20% (1.2 V stall) for takeoff and 30% for landing (1.3 V stall). Higher stall speeds mean lighter airplane weights for a given limiting condition and some paying passengers keep their money in their wallets instead of in the airline's cash register.

Delta systems

There are some rational and encouraging practices with respect to adjusting the stick shaker firing threshold by decrementing the stall angle some amount when the airplane is in icing conditions. In the case of the Saab 340, it takes crew action to activate the system with a switch. Since this switch only functions when selected ON by the pilot, it is fair to ask if the crew of the incident airplane would have used it, since they did not activate the de-icing boots or the propeller ice protection systems. However, even with such a system, the most difficult question to answer is how much is enough?

What is the solution?

This icing event is not a surprise to me as much as it was to the crew, but in fairness to them it is a scenario that has been repeated over the years in many different kinds of airplanes in many countries. Some have been fortunate to survive. Some design changes along the lines of those recommended by the ATSB are in order. In the meantime, the recovery procedures must be ingrained into the crews through training and practice. >> Dow, e-mail jdowsr@earthlink.net<<