Ice-Contaminated Tailplane Can Bring About Deadly Consequences

However, pilots of twin-turboprop aircraft continue to find themselves victims of an ice contaminated tailplane stall, perhaps because they may not recognize the signs and understand the differences as compared to a traditional wing stall.   

A few months after the tragic accident of Colgan Air Flight 3407 near Buffalo, New York, in 2009, a video originally created by the NASA Glenn Research Center in 1998 that focused on an ice contaminated tailplane stall (ICTS) was coincidently posted to YouTube. Unless you’ve had your head in the sand or just recently became a pilot, you’ve probably viewed it once or twice. That video went viral within the pilot community and seemed to become the de facto standard for recovery when any type of fixed-wing aircraft departed controlled flight in icing conditions.

 The video emphasized that the recovery from an ICTS is the exact opposite of a normal wing stall—which is true. That is, a normal wing stall requires you to push forward on the yoke or stick to break the stall, whereas an ICTS requires the exact opposite response—you must pull back hard on the controls.

The important point here is that very few aircraft are actually susceptible to tailplane stalls, but many pilots watching this original video came away with this understanding. The video really didn’t emphasize this. In fact, many that watched it walked away somewhat biased that all aircraft are highly susceptible to a tailplane stall, which is simply not the case. 

 If you fly a twin turboprop, you should pay close attention. The Twin Otter, for example, is one of the most susceptible aircraft to an ICTS, according to Kurt Blankenship, a research pilot and deputy of aircraft operations at the Glenn Research Center, which uses the Twin Otter for most of its research flights in icing conditions.

Most tailplane stalls are usually induced when the aircraft has accreted ice on the horizontal stabilizer and the pilot extends or raises flaps and/or adds or reduces power with flaps extended. In other words, you can’t have a tailplane stall without full flap or near full flap deflection. 

If you don’t have flaps extended, then you are having a wing stall, not a tailplane stall.   

The key point is that any action that changes the angle of attack of the horizontal stabilizer with flaps extended can induce a tail stall. Moreover, it’s usually on the high end of the flap speed range when the stall occurs. Something as simple as adding power with flaps extended can trigger the event, which is exactly what you see in the NASA video. The test pilot in the video was experiencing some control difficulty, but when adding full power, the aircraft reached the top end of the flap speed and the tail stall occurred abruptly.    

In talking with Blankenship, the Colgan Air accident had little to do with icing. In fact, on March 25, 2009, National Transportation Safety Board (NTSB) investigators indicated that icing probably did not contribute greatly to the accident.

“The plane basically trimmed itself to a stall,” Blankenship said. “It had a stick pusher on it which he [the captain] fought against and pulled back and held it until the plane finally went over, and it was too late.” 

Could it be that the captain thought he was experiencing an ICTS, even though he said nothing on the tape?

“[I think] the pilot simply panicked,” Blankenship said. “...He was low to the ground and [pulling back on the controls] was a natural reaction.”

Other mistakes were made such as the copilot retracting flaps without it being called and not having a sterile cockpit during the approach to land.

“[This accident] was a low-speed awareness issue, and that should be the focus of the training...and recognizing where you are at in the [power curve] regime,” Blankenship said. “They were slow, [so] it was a wing stall, clearly.”

In response to some of the confusion that was generated by the original 2009 video, NASA produced a second video in 2016 that is essentially an “icing wing stall video that emphasizes wing stall over tail stall,” according to Blankenship, who is featured in the video.

In this update, the characteristics of both flavors of a stall in icing conditions are discussed, but Blankenship emphasized that pilots flying most single-engine-piston or turbine aircraft should be prepared to recover from a normal wing stall, not a tailplane stall. The new video is sponsored by the FAA and can be viewed on YouTube. It’s very well done and worth a look.  

Even with this new video, there are still pilots finding themselves victims to an ICTS.

On the morning of October 18, 2022, the pilot and other occupant on a Beechcraft King Air 90 (N515GK) were killed in an accident while on approach to Mid-Ohio Valley Regional Airport (KPKB). According to FlightAware, the flight departed John Glenn Columbus International Airport (KCMH) at 6:40 a.m. EDT and was headed to Parkersburg, West Virginia.

Local news reported that the crash occurred near Marietta, Ohio, around 7:10 a.m. just 4 miles from the airport. The news video shows the aircraft plummeting to earth in almost a vertical orientation, suggesting a loss-of-control accident.

When I heard about the accident later that day, I decided to do some weather forensics and see what the conditions were during the early morning hours in the Parkersburg area.

At first glance, the icing environment looked pretty nasty, but this is a King Air with a certified ice protection system. There are a lot of potential factors to cause such an accident that may or may not have included weather. However, given the robust icing conditions that were indeed present on descent into Parkersburg, I suspected that this may have been associated with an ICTS.

About a week or so after the accident, the NTSB posted a preliminary report. Sometimes the agency’s initial findings might offer the surface observation at the time of the accident or perhaps a pilot report or two. I was pleasantly surprised to read the following:  

"Preliminary weather information at the time of the accident indicated that there were pilot reports throughout the area for trace to moderate icing conditions and AIRMETs moderate icing. Weather satellite data showed supercooled liquid water clouds from 1,300 feet agl to about 8,000 feet agl."

Having already done some analysis of the weather for that morning, the fact that the NTSB even mentioned the clouds were primed for icing really caught my attention and made me even more confident that this King Air pilot fell victim to an ICTS. 

According to the FlightAware log, the pilot climbed to the cruising altitude of 11,000 feet after departing Columbus. Given the marginal VFR conditions at Parkersburg, the flight was likely vectored to an instrument approach at KPKB.

The audio captured by LiveATC clearly suggests that the pilot checked in as being established on the RNAV approach for Runway 21. In fact, the pilot was cleared to land by a Parkersburg Tower controller moments before the accident. Furthermore, the routine observation issued at 6:53 a.m., or 16 minutes prior to the accident, reported light southwest winds, 10 sm surface visibility with an overcast ceiling of 1,400 feet as shown below.

KPKB 181053Z 26003KT 10SM OVC014 03/01 A2980 RMK AO2 SLP094 T00330006

On this particular morning, the lowest freezing level in the area was forecast to be in the range of 1,000 to 3,000 feet msl based on the one-hour lowest freezing level forecast from the Rapid Refresh (RAP) model.

A surface cold front was quickly moving to the southeast and was located just to the northwest of the Parkersburg airport at 5 a.m. , or two hours prior to the accident. 

By 8 a.m. EDT, or one hour after the accident, the cold front had passed well to the southeast of the route and was located in western Virginia. This means that the entire flight was on the cold side of this cold front.

According to Ben Bernstein, who is an expert on airframe icing environments, it is common for a stratocumulus deck to form in the wake of such a cold front when cold air pours in aloft and destabilizes the air near the surface and subsidence (sinking air) above the front caps the vertical development of these clouds giving them a quilted-like appearance when viewed from above. The highest liquid water content is usually in the tops of these types of clouds. 

This is easiest to see on the color infrared satellite image as a large area of yellow and pale green located just behind the cold front. If you compare the leading edge of the clouds, it’s almost coincident with the location of the cold front at the time of the accident.

Looking more carefully at the infrared satellite image, the bright yellow color equates to a cloud top temperature (CTT) of roughly minus-10 degrees Celsius. This implies that the stratocumulus cloud deck is likely dominated by water in the liquid state giving rise to a potentially hazardous airframe icing environment within those clouds. 

Moreover, the Skew-T log (p) analysis from the Rapid Refresh (RAP) model shows a clear stratocumulus cloud signature. This includes a nearly dry adiabatic (unstable) layer just above the surface with moist instability in the saturated cloud layer between 2,500 feet and 7,000 feet msl and a capping layer above. The cloud top temperature is a rather warm minus-8 C at 6,000 feet, which is fairly consistent with the IR satellite image. 

Lastly, the freezing level is roughly 2,200 feet msl or 1,300 feet agl.

This puts the aircraft in icing conditions during a descent from 6,000 feet to as low as 2,500 feet msl. Based on the FlightAware track log, that’s an exposure of approximately seven to eight minutes. 

This weather system was producing some rain at Parkersburg as the cold front passed shortly after 5 a.m. Notice that in the remarks section in the METAR below, rain began at 20 minutes after the hour (5:20 a.m.) and ended at 30 minutes after the hour (5:30 a.m.). Given the observation of rain, the clouds had some depth. 

KPKB 180953Z AUTO 26008KT 10SM OVC016 03/00 A2979 RMK AO2 RAB20E30 SLP089 P0000 T00330000

Moreover, the cloud depth also increased and ceilings lowered in the overnight hours leading up to the time of the accident before the ceiling began around 2,900 feet and reached 1,300 feet at 7:53 a.m. From an icing perspective, this is a huge red flag. Such a deepening of the clouds provides a better environment for the growth of larger drops and possible supercooled large drop (SLD) icing in the form of freezing drizzle.

Other clues included nearby observations that morning from Ohio University Airport (KUNI), Cambridge Municipal Airport (KCDI), and Zanesville Municipal Airport (KZZV) included light drizzle (-DZ) and light rain (-RA) that again emphasize the potential for a large drop environment.

The Current Icing Product (CIP) agreed that this is a warm-topped icing scenario as can be seen above by the red area. WMPCP means precipitation other than snow is being observed at the surface with a cloud top temperature warmer than minus -12 C. This largely hints at an SLD event given the warmer cloud tops which imply the cloud was dominated by water in the liquid state plus the time of day when FZDZ aloft tends to be relatively common.

The real smoking gun is the supercooled liquid water (SLW) content in these clouds. CIP identified that the SLW content was likely over 0.5 grams per cubic meter during the descent which is quite high, but not unusual for a continental stratocumulus cloud deck especially in the tops where the air is the coldest and the SLW content is the highest. Shown above is the SLW content at 6,000 feet msl at the time of the accident.

Given the analysis above, it’s reasonable to assume that airframe icing may have played a causal factor in this accident, especially if ice accreted behind the protected surface on the tail of the aircraft. In fact, when the NTSB released the factual report on May 30, 2024, it determined that the probable cause of this accident to be “structural icing on the tailplane that resulted in a tailplane stall and subsequent loss of control.”  

The icing environment on that morning may not have looked all that threatening to many pilots, especially those flying an aircraft with a certified ice protection system. In fact, most standard briefings would not have picked up on the significant nature of the icing environment. Other than a G-AIRMET for moderate icing, there were no SIGMETs or even pilot weather reports of severe icing.

Even so, the clues were indeed there, assuming you know what to look for. 

This feature first appeared in the October Issue 951 of the FLYING print edition.