Beware The Icing Hazards Masked By Average Droplet Size, Scientists Warn

Expansion of the icing envelope for aircraft certification purposes, as proposed by the Federal Aviation Administration (FAA), will not cover all the icing conditions likely to be encountered by an airplane during its service life. The envelope needs to be expanded, claim a group of distinguished atmospheric scientists.

In 2010, the FAA proposed an Appendix O to cover supercooled liquid droplet (SLD) conditions. (See Aviation Safety Journal, July 2010, “Significant Regulatory & Related Activity”) This new appendix would theoretically cover icing conditions not defined in Appendix C of the regulations.

The icing conditions in the 1994 accident at Roselawn, IN, involving a twin-turboprop ATR-72, prompted the National Transportation Safety Board (NTSB) to recommend the FAA include much larger droplets than defined in certification regulations. This recommendation is the rationale for the belated publication of the Notice of Proposed Rulemaking with Appendix O in 2010, fully 16 years after the Roselawn crash.

Supposedly, Appendix C covered only 99% of the water and droplet sizes in so-called “cloud icing” conditions. Appendix O was intended to cover the conditions of freezing drizzle and freezing rain produced by other distinctly different processes of formation that are not part of the cloud icing conditions. Thus, an airplane certificated to both appendices should be able to cope successfully with any icing encounter while airborne.

Not so, claim the scientists. After examining the data used as a basis in the proposed Appendix O, and comparing these data to other data collected by instrumented research aircraft, they conclude in their submission:

“We therefore are concerned that adoption of these rules will lead to a false sense of security that they will protect against the icing hazard of freezing drizzle and freezing rain, when we have evidence this will not be the case.”

The essence of their argument is familiar to students of Statistics 101 and those gamblers who frequent craps tables at casinos. It is similar to the way two dice can land, showing a total count of seven on the top surface. There are six combinations: 1 & 6; 2 & 5; 3 & 4; 5 & 2; and 6 & 1. The average number of spots for all six combinations is 3½. The corollary in icing is what is referred to as the mean volumetric diameter (MVD), a hypothetical diameter characterizing all the sizes of droplets in the cloud for which half the mass of water is in droplets larger, and half is in droplets smaller. A dice has no face with 3½ dots and there need not be any droplets with the exact MVD.

The scientific evidence is that MVD, similar to the 3½, bears no relation to hazard. There are icing cases similar to rolling a 6 and a 1 that are the real hazards (and the other five combinations not so much). The way the icing envelopes are defined date back to the 1940s, but evidence now shows that other metrics are warranted. Scientific evidence supporting the need for reexamination has existed from multiple studies beginning in 1984 and revisited in the late 1990s.

Yet, as “nature abhors a vacuum’, the aviation industry abhors a change – and that is the seminal message in the scientists’ letter.

Extracts of the scientists’ submission to the docket follow:

June 21, 2011


Docket Operations, M-30

U.S. Department of Transportation

1200 New Jersey Avenue SE

Room W12-140, West Building Ground Floor

Washington, DC 20590-0001


Re: Supplemental Comments to Docket Number FAA-2010-0636

Dear Sir or Madam:

The following comprise our supplemental comments to the Docket with respect to the Notice of Proposed Rulemaking (NPRM) … published in the Federal Register June 29, 2010 … We recognize that the comment period has closed. However, the following has taken substantial time and effort to thoroughly review the data that the proposed Appendix O was based upon, compare it to our results, and prepare substantive comments.

On the basis of independent measurements of the icing hazard, obtained with a research aircraft while supporting research projects that studied icing environments, we argue that the proposed rules will not provide adequate protection against some of the most serious icing hazards. [Emphasis added] We explain the reasons for this assertion below …

Our main concern is that … the draft regulations implicitly assumes that the icing hazard is represented adequately by … liquid water content (LWC) and the droplet size distribution (DSD) selected from one of two average distributions on the basis of the median volume[tric] diameter (MVD). No justification has been offered to relate the plotted parameters to performance in icing. Incorporating these figures into the regulations will imply that the icing hazard is determined by these properties, so it is only necessary to demonstrate ability to encounter conditions characterized by these values. However, we suggest that for a given LWC and MVD there actually can be great variability in the icing hazard because real size distributions vary substantially from those shown in Fig. 2 for freezing drizzle and Fig. 5 for freezing rain. Those figures result from averaging many different size distributions, all of which can have different effects on performance, and that averaging can obscure the icing hazard …

We have experience and data to support these assertions. A summary of the effects of icing on performance of our Beechcraft Super King Air 200T (operated by the University of Wyoming and henceforth called WKA), first published in 1984 … concluded that there was no observed correlation between MVD and the impact of icing on performance. This same conclusion was arrived at and published in all the subsequent articles based on a much larger data set … The fundamental reason MVD is not correlated with performance is MVD represents cloud droplets rather than drizzle drops … Indeed, the most hazardous encounters in that data set and in subsequent studies in which we were involved had the same LWC and MVD as many other encounters that led to much smaller effect on performance. (We had the benefit of a continuous measure of the effect on performance of the aircraft to accompany our measurements, something that was not developed for the data set used as the basis for Appendix O, so we can defend the preceding statement with performance data.) We therefore are concerned that adoption of these rules will lead to a false sense that they will protect against the icing hazard of freezing drizzle and freezing rain, when we have evidence that this will not be the case.

The substance of our argument is that the proposed envelopes for LWC vs. temperature and average drop size distributions mask the most adverse conditions that have been measured by combining them with conditions that pose only a minor hazard. The envelopes in the draft Appendix O focus on average properties of the supercooled drop size distribution and do not represent the important effects of variations from that average distribution, but those variations often lead to variations in ice roughness and in the locations of accretion. Certain forms of icing with very adverse distributed ice roughness from freezing drizzle can accrete in a few minutes and can quickly create significant drag and associated controllability problems for airplanes, even in cases where the visual appearance of this ice accumulation is not remarkable …

In our measurements, performance (as measured either by potential rate of climb or by increased drag on the airframe) exhibited no correlation with MVD, further leading us to question the usefulness of this measure of icing severity …

Post-accident forensic weather analyses of icing-related accidents by scientists specializing in these phenomena support the occurrence of the icing conditions that we assert are not accounted for in the draft of Appendix O, and those analyses have pointed to the likely involvement of a particular type of freezing drizzle in the accident record of various airplanes. These conditions tend to produce ice features having distributed roughness that do not have significant thickness or mass …

We suggest that additional steps to address these problems and guard against the most serious icing hazards are needed before new envelopes are inserted into the regulations. The proposed new regulations could delay efforts to address the problems raised in these comments and would lead to unnecessary effort to meet inadequate requirements.


The crux of the matter now rests with the FAA in the rulemaking process. Does the FAA proceed with the proposed Appendix C and Appendix O envelopes or revisit them? Given the pre-eminent stature of the commentators above, the FAA will have some important decisions to make. Ignoring the comments above is one option, but that course does nothing for the safety of aircrews and passengers flying in icing conditions.