Somewhere back in the mid-1980s, I was flying a Fairchild Metroliner enroute to Richmond, Virginia, from who-knows-where. It was winter, we were IMC, and we were getting ice. We had no autopilot. I probably had a new first officer, or I may even have been conducting initial operating experience… I can’t recall… but in any event, the cockpit workload was pretty high. At some point, I decided that I didn’t have time to keep checking the wing to see when it was time to pop the boots again. I reached down and pushed the rocker switch down, to automatic, instead of up, the single-cycle position.
I had never done that before. Like most pilots of the pneumatic boot crowd, we were well schooled on the notion of ice bridging. Although the airplane was equipped with an auto cycle option, nobody ever used it. The premise was that unless the ice accretion rate was precisely aligned with the cycle timing, you would end up with ice bridging. Instead, we delayed operation of the boots until a requisite thickness of ice had accreted.
A few minutes later, I remembered the ice, and checked the wing. To my utter surprise, it was nearly clean.
Ice bridging is the idea that if you operate the boots too early, you will stretch the ice but not fracture it. When the boot deflates following the cycle, the stretched ice will remain, with more ice building on top of it. Ergo, the next cycle of the boot will be useless, because the boot will simply expand into the cavity underneath the previously stretched ice. It can be traced back to at least the 1943 edition of Civil Aeronautics Bulletin No. 25, Meteorology for Pilots, which stated that, “If aircraft is equipped with leading edge boot type de-icers, allow ice to form to inch thick, then crack off by inflating de-icers. Use de-icers periodically as ice forms. This prevents ice from forming a pocket in which the boot expands…” The identical phrase is found in Zweng’s Airline Transport Rating textbook from 1947, so it is difficult to know who had actually thought about the idea, and who was copy-and-pasting.
It was enshrined in the FAA Aircraft Icing Handbook as late as 1991, which stated that, “Bridging is the formation of an arch of ice over the boot which is not removed by boot inflation. This can occur if the system is activated too early or too frequently, especially in glaze icing conditions.” The Handbook went on to opine that, “A certain degree of pilot skill is required for safe and effective pneumatic boot operation. Actuation when accreted ice is too thin may result in ‘bridging’ where the formation of ice over the boot is not cracked by boot inflation. Thus, attention is required to judge whether the cycle time continues to be correct as icing conditions change.”
Yet there is not a single test conducted in anyone’s icing research wind tunnel that has been able to replicate ice bridging, nor are there any accidents that document ice bridging as a cause or contributory factor. There are precisely two reports in forty years of NTSB data that use the term. To be fair, there are plenty of accidents for which we do not have any specific cause aside from a generalization, or even any pilot accounts, so we don’t know what many of those pilots saw; that said, there are also precisely two reports in the entire NASA Aviation Safety Reporting System that use the term ice bridging. In the FAA Accident and Incident Data System, the term does not appear. Most interestingly, the term is never used in the Transportation Safety Board of Canada’s accident data, nor does it appear in Canada’s Civil Aviation Daily Occurrence Reporting System (CADORS).
On the other hand, when developing data for the FAA Technical Report, “The Icemaster Database and an Analysis of Aircraft Aerodynamic Icing Accidents and Incidents,” I was able to identify a subset of 177 accidents and incidents involving aircraft equipped with pneumatic boots in which something could be said about the number of times the pilot cycled the deicing boots. There were 45 events in which the boots were not operated at all. In 44 events, the boots were cycled less than four times. In 26 cases, the system was cycled an unspecified but multiple number of times. In 40 cases, the system was operated but no information was found in the reports regarding how often it was operated. In the remaining 22 cases, the system was cycled automatically.
The general trend that might be extracted from this relatively small amount of data is that, not surprisingly, a large subset of accidents and incidents involve a failure to operate the boots at all, and the smallest subset involve automatic cycling of the boots. This would be a bridge too far; there are far fewer aircraft that are even equipped with an automatic boot cycling feature, and they tend to be slightly larger scale aircraft. One would expect that their appearance in the event data would be less frequent, regardless of the boot operating mode.
Nevertheless, it seems likely that in many cases, either when the boots were not operated, or were operated only a couple of times, the pilot was delaying activation until the requisite ice thickness had been attained. The lower frequency of events in which the boots were cycled multiple times suggests that this technique may lead to a better outcome, whereas the practice of delaying boot activation has almost certainly created a significant number of catastrophes.
It turns out that there are two, entwined ideas embedded in the delayed activation practice for operating pneumatic deice boots. One is ice bridging; the other is immediate and efficient shedding of the ice. First, we need to untangle these ideas, next, we need to evaluate the value of what we find.
In 1997, I was a participant at the FAA Deicing Boot Ice Bridging Workshop, held at the Ohio Aerospace Institute in Cleveland. The consensus at the workshop was that the idea of ice bridging may have evolved from experience with early, low-pressure, slow inflation time systems, and was very unlikely with a high pressure, rapid inflation/deflation system. The current version of Advisory Circular 91-74B, “Pilot Guide: Flight in Icing Conditions,” sums the argument up this way:
“A traditional concern in the operation of pneumatic boots has been ice bridging. This is attributed to the formation of a thin layer of ice which forms to the shape of an expanding deicing boot without being fractured or shed during the ensuing tube deflation. As the deformed ice hardens and accretes additional ice, the boot may be ineffective in shedding the bridge of ice. Studies done in the late 1990s have established that there are few, if any, documented cases of ice bridging on modern boot designs. In addition, several icing tunnel tests sponsored by the FAA since 1999 showed no ice bridging on modern boot designs. Known cases are confined to boots of designs dating back a half century or more.”
I’ll admit that I have always been skeptical of the “Known cases… dating back a half century or more.” If you think about it for a minute, it is actually kind of hard to imagine how the ice accretion has enough flexibility to stretch but not fracture, and then immediately exhibits enough strength to retain its shape in the airflow following boot deflation, without any structural support beneath it. It is possible that something may have occurred with early, low-pressure, high-volume de-icing boots, perhaps because they were poorly deflated following the inflation cycle. That might have provided adequate supporting structure for further ice accretion as well as biased the subsequent expansion volume, reducing the boot’s effectiveness. Conversely, it may be that in conditions of rapid ice accretion, a boot that remained inflated for an excessive period accreted ice over the inflated boot.
Curiously, however, the pneumatic deice boots on a DC-3 were either on, meaning a continuous cycling, or off. In the meticulously thorough accident report for United 21 at Chicago in 1940, a DC-3 that stalled on short final, several pilots reported using the deice boots continuously during the descent into Chicago, shutting them off as required for the landing. None of the interviewed pilots made any mention of a concern about bridging. Indeed, a search of several decades of early accident reports reveals no reference to the problem, although this again must be mediated by the fact that most of those reports involved aircraft with no recorders, and in many the crew did not survive, so like many contemporary accidents, we don’t actually know what they saw.
Writing in Air Facts in February 1944, Captain Bob Buck described pneumatic boot operation in detail; he discussed the likelihood of considerable residual ice under certain circumstances, particularly glaze ice, and advocated his preference for delaying boot operation until at least 1/4 inch of ice had accreted in order to improve the shed efficiency, but he also emphasized not waiting too long. He described tenacious ice resulting from a late activation as ice that “may break up and blow away, but fairly large chunks will hang on to the rubber boot. A piece perhaps three by four inches will sit broadside to the air stream and be holding on to the boot by only a square inch of its surface—but it will hold on and merrily go up and down with the boot.” Although he wrote this within months of the publication of Civil Aeronautics Bulletin No. 25, Captain Buck made no mention of ice bridging.
If we dig a bit into the historical research, it is fairly clear that the driving force behind a published procedure recommending a delayed boot activation has been to obtain the most effective shed of ice in a single cycle. Before going into that, I need to explain three terms used in the discussion of pneumatic boot operation:
- “Pre-activation ice” is the ice which accretes on the wing before the ice protection system is activated. This may be the traditional quarter inch, or simply the ice that accretes before the ice detector is triggered.
- “Residual” ice is the ice remaining immediately after a boot cycle.
- “Intercycle” ice is the ice which accumulates during the period between the end of the last boot cycle up to the beginning of the next boot cycle. It is completely dependent on how often you cycle the boots, as well as the rate of ice accretion.
In 1964, Dean Bowden was the lead author of an extensive report for the FAA entitled “Engineering Summary of Airframe Icing Technical Data,” better known in the industry simply as ADS-4. Bowden had been a research engineer at the NACA, and at the time this report was issued, he was employed by Convair. There is no mention at all of ice bridging in ADS-4, but like Captain Buck, Bowden discussed delayed activation of the boots as a means to obtain a more efficient shed of ice:
“After a de-icing cycle, small particles of ice may still adhere to the surface. This is residual ice and may be used as a measure of de-icing system effectiveness. Residual ice may be minimized by permitting from one quarter to three eighths of an inch of ice to accrete before actuating the de-icers. Unless removed, the residual ice may cause an airfoil section drag increase of from five to fifteen per cent and is independent of the number of de-icing cycles.”
In other words, it is not the threat of bridging that is a concern, but rather a pernicious drag rise resulting from unshed residual ice. This conclusion curiously did not elaborate on earlier work done by Bowden while he was at the NACA. In the 1956 NACA paper entitled “Effect of Pneumatic Deicers and Ice Formations on Aerodynamic Characteristics of an Airfoil, Bowden investigated the drag rise created by both the inflation of pneumatic deicing boots and the ice that remained between boot cycles, using a NACA 0011 airfoil. (I referenced this paper in my previous article, In-Flight Icing’s Hidden Threat: The Landing Flare). Using a variety of icing rates and airspeeds, Bowden concluded from his work in the icing tunnel that “Minimum airfoil drag in icing (averaged over a de-icing cycle) was usually obtained with a short (about 1 min) de-icing cycle.” He expanded on this in his analysis (be careful here… note that he is referring to airfoil drag, not total airplane drag):
“For the lower ice accretion rates, the average drag increase reduces slightly with increasing cycle time. The difference in drag increase between 4- and 1-minute cycles, however, is only 2 to 6 percent. For higher ice-accretion rates, the average drag increase is greater for the longer cycles. For an ice accretion rate of 4.8 pounds per hour per foot span and glaze-icing conditions, the drag increase for the 4-minute cycle is double that for a 1-minute cycle. A fixed de-icing cycle is often desired to simplify deicer controls. Where this is the case, the short cycle (1 min) represents the best compromise for the spanwise deicer.”
Some of this work was revisited by Mike Bragg, Andy Broeren, and Gene Addy in 2002, and reported in the FAA Technical Report titled “Effect of Residual and Intercycle Ice Accretions on Airfoil Performance.” This investigation used a NACA 23012 airfoil section, which is very representative of many twin engine, propeller-driven aircraft ranging from the Cessna 310 up through the Embraer 120. As with Bowden’s work, this investigation concluded that a longer period between boot cycles resulted in significant drag penalties. Using a three-minute cycle time yielded at least a three-fold increase in minimum airfoil drag. A one-minute cycle yielded a much smaller intercycle ice shape and a much smaller increase in airfoil drag.
However—and this is a big however—their investigation also concluded that “decreasing the initial activation time did not substantially affect the resulting intercycle ice shape. This means that the boots were just as effective when they were activated 11 seconds after the start of the spray as when they were activated after a quarter-inch of ice was allowed to accrete (252 seconds) on the leading edge, for the one condition tested.” Bowden had not investigated any differences between early or late activation of the boots.
Of perhaps even greater importance, the investigation concluded that regardless of cycle time, the intercycle ice shapes on a NACA 23012 airfoil will reduce the maximum coefficient of lift by 50% or more. This is considerably more lift loss than that measured by Bowden in his 1956 research, and was true even for a rime icing test condition in which the boot was cycled every minute. We need to be cautious, as the magnitude of this degradation may have a lot to do with the smaller scale of the test airfoil. Nevertheless, it clearly points toward the truth of pneumatic boots: they will never yield a completely clean wing. In other words, optimum boot operation or not, the airplane is going to stall at a much lower angle of attack. That is a real foot stomper.
In this study, the researchers noted that the intercycle ice shapes were much more detrimental than the residual ice shapes. This may be significant with regard to what pilots are seeing when they believe they are witnessing ice bridging. Most importantly, they noted that “The residual and intercycle ice shapes reached a steady state after two or three boot cycles.” That may quite significant with respect to my own observations regarding how often pilots operate the boots and the subsequent outcomes.
Further research by Gene Hill, et.al, also using a NACA 23012 airfoil and reported in the FAA technical paper titled “Investigations of Performance of Pneumatic Deicing Boots, Surface Ice Detectors, and Scaling of Intercycle Ice,” reached very similar conclusions. Gene was a former Boeing aerodynamicist who served as the FAA’s National Resource Specialist for Icing between about 1996 and 2006. A couple of the statements from his report are worth considering:
“The ice shedding efficiency of the deicer’s initial cycle improved with increased ice thickness. However, the value of the cleaner shedding of the initial deicer cycle tended to deteriorate during subsequent cycling of the deicer, with the deicer shedding efficiency being equivalent after a few cycles to an initial deicer cycle with a thin layer of ice.
“Throughout the investigation… no ice bridging across the deicer tubes was observed. Residual ice that remained on the deicer following its cycling remained until sufficient thickness accumulated for the ice to be shed during a subsequent cycling of the deicer.”
We also need to be cautious with these results, as they all used wind tunnel speeds of 170 knots or greater, so operations at slower speeds may yield less efficient shedding. The ice shedding shearing force at 170 knots is almost 300 percent greater than it is at 100 knots. That said, in ADS-4, Bowden had noted the improved efficiency of the boot with increased thickness, just as Bob Buck had explained, and Hill noted this as well. All three research programs—Bowden in 1956, Bragg in 2002 and Hill in 2005—concluded that a shorter, one-minute cycle period yielded the minimal intercycle ice shape, and Bowden and Bragg associated this with minimal drag rise. None of these researchers report any indication of ice bridging, and both Bragg and Hill departed from conventional wisdom when they concluded that you would end up with the same steady-state intercycle ice shape regardless of when you activated the boots.
So if ice bridging really doesn’t occur, what are pilots seeing when they claim they have encountered it? If we go back to what Bob Buck said about chunks of ice holding on and “going merrily up and down with the boot,” and then consider the observations of both Bragg and Hill regarding the remaining residual ice and the larger intercycle ice, we may have some insight. These observations lead me to suspect that much of what pilots believe is ice bridging is actually residual ice that becomes rimed or glazed over during a somewhat lengthy intercycle period. While the pilot may be expecting a highly efficient shed of ice the first time he/she pops the boots, there will almost always be a quantity of residual, and then intercycle, ice on the boots at any given point. With caveats for visibility, lighting, and so forth, the pilot may be led to believe the boots are not working properly. The longer the dwell time between cycles, the worse the subsequent intercycle ice may appear.
The conundrum created by the ice bridging myth has driven investigators and members of the NTSB nearly apoplectic in recent years. On March 17, 2007, a Cessna Citation 500 experienced a wingtip strike during landing at Beverly, Massachusetts, that resulted in substantial damage. The pneumatic boots had not been activated at all. The Board reported that the pilot stated “that he had heard about ‘ice bridging,’ at the training provider and his company, and that that you do not want to ‘blow’ the boots too soon as you can get a ‘hollowed area.’” The copilot correctly stated the manufacturer’s recommendation, that the boots should not be operated “unless you have 1/4 to 1/2 inch of ice.”
In the Beverly report, the Board referenced three other accidents involving the Citation 560: one at Eagle River, Wisconsin, in 1995, which was attributed to an ice accretion of an inch of ice, and a second at Augsberg, Germany, attributed to an ice accretion of just over an inch of ice. The third accident was their earlier investigation of the Circuit City Citation crash at Pueblo, Colorado, in 2005. In that case, one of the 44 events in my data, the crew was aware of the ice accretion and cycled the boots one time (that could be documented). Yet, with somewhat less than the required 1/4 inch of ice accretion, they stalled and lost control.
Board Member Deborah Hersman wrote a partial dissent to the Circuit City probable cause, joined by Member Katherine Higgins, stating that “This concern about ice bridging was reinforced to this Board Member during a conversation earlier this month, with a well-respected pilot of a modern Cessna aircraft equipped with pneumatic boots, who repeatedly spoke of ice bridging and the guidance from the manual requiring a 1/4 to 1/2 inch of ice accumulation before activating the boots.” She cited FAA Advisory Circular 25.1419-1A, Certification of Transport Category Airplanes for Flight in Icing Conditions, which says:
“Many AFM’s specify a minimum ice accumulation thickness prior to activation of the deicer boot system. This practice originates from the belief that a bridge of ice could form if the boots are operated prematurely. Although the ice may not shed completely with one cycle of the boots, this residual ice will usually be removed during subsequent boot cycles and does not act as a foundation for a bridge of ice to form. The AFM procedure for boot operation should be to operate the boots at the first sign of ice and not wait for a specific amount of ice to accumulate.”
The real lesson from all of this work lies with the frequency of boot cycling. Regardless of whether you activate the boots after accreting a manufacturer-recommended thickness of ice (which your AFM or POH may still require) or activate them immediately upon accreting ice, after a few cycles you will have about the same shape and quantity of residual-to-intercycle ice, and that is probably the best you can do. The key is in the time between cycles. In general, a one to three-minute interval between cycles is best, leaning more toward the one minute end.
The challenge is that you are going to be operating in a single-pilot environment in lousy weather and probably with a manually cycled system. If you cannot select an automatic cycling option, you will have to work at cycling the boots frequently. Don’t wait until starting the approach to cycle the boots the first time; that first cycle, and the second and maybe third, will not get the wing as clean as it could be. It is probably going to require multiple cycles. You want this condition to be achieved before beginning the approach, bearing in mind that operating the boots too close to the flare may be detrimental in itself. Be sure to know what the manufacturer’s recommended procedures are for boot operation on final approach.
There are two more important points. First, AC 91-74B plainly states the following:
“An ice adhesion inhibitor should be applied to pneumatic deicing boots in accordance with the maintenance manual and is highly recommended. Testing in 2005 showed that the proper application of ice adhesion inhibitors improved ice shedding at colder temperatures and a reduced amount of residual and intercycle ice. Any product that is not recommended by the airplane or boot manufacturer should be approved by the FAA.”
Second, there will likely be a manufacturer-recommended approach speed adjustment, and you need to know what it is and use it. This may have been published through an Airworthiness Directive. Recall the conclusion from Mike Bragg’s research that, even with optimal boot cycling, the maximum coefficient of lift can be reduced by as much as 50%. While some of this is possibly due to the small scale of the test airfoil used in the icing tunnel, it still suggests a significant reduction in the maximum coefficient of lift. The speed additive is intended to keep the approach angle of attack safely away from the reduced peak of the lift curve.
And remember, as Dean Bowden showed in 1956: when you flare, the drag is likely to go right through the roof. Be prepared to land with power.
Finally, keep in mind Bob Buck’s admonition about boots from 1944: “First of all they do not completely mean that an airplane can be flown in all kinds of ice for an indefinite period. They will take care of light icing conditions and occasionally moderate ice, but their prime objective is to help you get up through and down through ice.”