Icing Effects on Stall Speed
Sometimes a relatively minor accident brings to light safety issues of broad importance. After the crash of a Cessna Citation in Fargo, North Dakota, the NTSB decided to classify the accident as only a limited investigation and didn’t send its own investigators to the accident site. As the case developed, however, two important safety issues emerged. The first was how icing affects stall speed and stall warning in early Cessna Citation series airplanes, and the second was the role of non-certificated charter flight arrangements on safety.
First, let’s look at the simple facts of the case. The Citation II (CE 550), N941JM, experienced a hard landing at Hector International Airport (KFAR) in Fargo at 1353 CDT on Nov. 30, 2018. The business flight from Sloulin Field International Airport (KISN) in Williston, North Dakota, to Fargo took just more than an hour and was uneventful until the last few minutes. The airplane sustained substantial damage and nine of the 11 persons on board received minor injuries.
What made this hard landing an accident was an aerodynamic stall at a height of about 100 ft. above the runway. The pilot, who was flying the two-pilot airplane as a single pilot, forgot to lower the flaps and flew through icing conditions on approach. The airplane banked right and, according to witnesses, “fell out of the sky.”
The right wingtip struck the right edge of Runway 18. The airplane bounced and fell back on its belly. It came to a stop after sliding 635 ft. across the grass infield. The pilot shut down the engines, turned off the battery and assisted the passengers as they exited the main cabin door.
The outboard section of the right wing was pushed up and aft. The nosewheel assembly was separated from the fuselage and the left main landing gear components were pushed through the upper wing surface. The wing, tail and windshield showed evidence of ice. The angle-of-attack (AOA) probe was intact but covered with ice, and when tested, the heating element to the probe did not work. As is often the case after a hard landing or runway excursion, the flap handle was found in the LAND (down) position, but the flap indicator was in the UP position. When investigators later were able to examine the flap cable and drive chain, it showed the flaps were indeed up.
An NTSB meteorologist verified that a cloud deck existed over the approach path between 3,100 ft. MSL and about 1,300 ft. MSL. The airport elevation was 901 ft. MSL. The temperature and dew point at KFAR were almost the same, -1C, and there was an AIRMET in effect for moderate icing below 10,000 ft. The current and forecast icing potential (CIP and FIP) predicted a 60-70% chance of light icing below 3,000 ft. MSL near the accident site.
The NTSB determined that the probable cause of the accident was “the pilot's failure to lower the flaps during the approach and maintain sufficient airspeed while flying in instrument meteorological and icing conditions and the accumulation of ice on the wings' leading edges, which resulted in the exceedance of the airplane's critical angle of attack and subsequent aerodynamic stall. Contributing to the accident was the pilot's lack of proper qualification to operate the airplane under a single-pilot exemption due to his lack of total turbine time, which led to task saturation and his failure to properly configure the flaps for landing.”
We know what happened and the probable cause. Now let’s dig a little deeper into why. Why did the pilot stall the airplane and what risk factors were missed when this flight was organized?
The Effects of Structural Ice
Determining the extent to which the ice may have helped to induce the aerodynamic stall was a task performed by an NTSB performance specialist. In the absence of a flight data recorder (FDR), which was not installed on the airplane, the specialist constructed a simulation model of the airplane’s flight path, speed, AOA and other performance data using a model developed by the simulator company CAE. The specialist matched the model data to radar data from the FAA’s Airport Surveillance Radar (ASR-11) located at KFAR. He also used the fact that the pilot reported entering the clouds at 1340 CST at 3,100 ft. MSL about 10 nm from the airport.
The pilot stated that he flew at an approach speed of 120 kt., but the radar study showed his ground speed gradually bled off from 134 kt. at 2,600 ft. to 120 kt. at 2,100 ft., and then, within 24 sec., fell off to 106 kt. During the last 2 min. of flight, the airspeed was as low as 99 kt. and AOA was very close to stall AOA. When the specialist added a 5% reduction in lift coefficient to model the effect of icing on the flaps-up lift curve, the AOA exceeded stall for the last 30 sec. of flight.
The actual lift reduction might have been slightly more or slightly less than 5%. The specialist referred to figures from the 2014 Embraer Phenom 100 Gaithersburg, Maryland, accident investigation to come up with the 5% icing lift penalty. In the Phenom accident, investigators found there was a 10% lift reduction resulting from icing when that airplane stalled and crashed on approach.
On the NTSB Form 6120 the pilot filled out after the accident, he checked boxes indicating the actual conditions during the approach were light freezing rain and severe mixed icing. The term “severe” implies ice was accruing faster than the deicing equipment could remove it. While it’s understandable that the pilot had the impression the icing was severe, the photos of the aircraft after the crash look more like the ice was light to moderate, and that is what the NTSB meteorologist found was likely to be the case.
The pilot was aware that ice was accruing on the airplane, and he reported that he actuated the deicing boots several times during the approach. It is not clear when he did that. Knowing exactly when the deicing boots were actuated and for how long would have been very helpful in establishing whether the airplane’s deicing equipment was deficient or simply not used often enough. An FDR could have provided that information.
The airplane did have a cockpit voice recorder (CVR). It did not record the pilot verbalizing any checklist items or comments about activating the pneumatic deicing boots or windshield anti-ice. The CVR report also did not mention the sound of aerodynamic buffeting or pilot reaction to an impending stall.
The airplane had an optional AOA system designed to be used as an independent reference check against approach airspeed, but the AOA probe was found to be ice-covered and investigators could not verify that the heating element worked.
Two witnesses on the ground reported seeing the wings “waffling” or “fluttering,” and the passenger seated in the cockpit right seat stated the airplane began “fishtailing.” These could have been manifestations of the stall buffet. A passenger who was seated where she was able to watch the captain said he had the control wheel all the way toward him, as close to his chest as he could pull, and his arms were shaking from the effort.
The accident airplane was manufactured in 1980. One of the interesting facts about this early model Citation was that it had no stall warning system, at least as we know it. The Cessna 550 Airplane Flight Manual (AFM) stated: "Stall warning is achieved aerodynamically, aided by stall strips on the inboard section of each wing. The strips disrupt airflow over the wing, causing that area to stall first, accentuating pre-stall buffet. The pilot is alerted to impending stall by aerodynamic buffeting, which occurs at approximately VS1 + 10 in the clean configuration and VS0 + 5 in the landing configuration."
When I learned this, I was quite surprised. Then I read 14 CFR Part 25.207, which addresses certification requirements for stall warnings in transport category airplanes. It states in paragraph (b) that the required stall warning “must be furnished either through the inherent aerodynamic qualities of the airplane or by a device that will give clearly distinguishable indications under expected conditions of flight.” In other words, if the airplane has sufficient aerodynamic buffet, no other stall warning system is necessary. I don’t know of any other transport category airplanes that lack a stall warning system. The question to me is, did this stall buffet provide the pilot with sufficient warning?
The NTSB established that the pilot left the flaps up, accrued some ice on the airplane, got slow and stalled. What is not clear is if the aerodynamic buffet occurred, or if it occurred too late to alert the pilot to recover from the stall. The pilot’s action in pulling full aft yoke indicates he either didn’t recognize the stall or had not absorbed the training on stall recognition and recovery that ensued from the Colgan Air Flight 3407 accident.
If the NTSB had established that the stall warning, with or without icing, was inadequate, it would have been justified in recommending the installation of an actual stall warning system on the airplane. It did not, even though a similar scenario had happened before, on Feb. 16, 2005, near Pueblo, Colorado.
Editor's Note: A subsequent article by Roger Cox will examine the stalling warning system and use of a non-certified charter.