Fatality and Injury Rates for Two Types of Rotorcraft Accidents

DOT/FAA/AM-05/17
Offi ce of Aerospace Medicine
Washington, DC 20591
Fatality and Injury Rates
for Two Types of Rotorcraft
Accidents
David Palmerton
Civil Aerospace Medical Institute
Federal Aviation Administration
Oklahoma City, OK 73125
October 2005
Final Report

---

NOTICE
This document is disseminated under the sponsorship of
the U.S. Department of Transportation in the interest of
information exchange. The United States Government
assumes no liability for the contents thereof.

---

Technical Report Documentation Page
1. Report No.
2. Government Accession No.
3. Recipient's Catalog No.



DOT/FAA/AM-05/17




4. Title and Subtitle
5. Report Date



Fatality and Injury Rates for Two Types of Rotorcraft Accidents
October 2005
6. Performing Organization Code





7. Author(s)
8. Performing Organization Report No.


Palmerton D

9. Performing Organization Name and Address
10. Work Unit No. (TRAIS)


FAA Civil Aerospace Medical Institute



P.O. Box 25082
11. Contract or Grant No.


Oklahoma City, OK 73125


12. Sponsoring Agency name and Address
13. Type of Report and Period Covered


Office of Aerospace Medicine



Federal Aviation Administration



800 Independence Ave., S.W.



Washington, DC 20591
14. Sponsoring Agency Code


15. Supplemental Notes




16. Abstract


An analysis of the frequency of four different types of rotorcraft accidents was conducted to determine if the


number of fatalities and injuries between accident conditions was different. Accidents involving rollover, no
rollover, fire, and no fire were studied to determine if accidents with a rollover or fire might be creating
evacuation delays that contribute to the fatality and injury rates.
A search of the FAA Accident Incident Data System from January1986 to March 1997 produced 2704 accident
records for this analysis. A Chi-Square test for independence was used to determine the difference between the
rollover and no rollover and fire and no fire accident categories. Further analysis were performed on
combinations of the two main categories to determine if an event such as a rollover and fire produced more
fatalities or injuries than a rollover without a fire.
There were more fatalities in the no rollover category (P=.0001) and more injuries in the rollover group
(P=.001). Accidents with a fire produced more fatalities than accidents without a fire, (P=.0001). Rollover
accidents without a fire produced more fatalities (P=.0001) than no rollovers without a fire, and more injuries
were produced in the rollover no fire group (P=.0001) than the no rollover no fire category. The group of
accidents where the rotorcraft rolled and caught fire lead to more fatalities (P=.0001), and the no rollover group
with fire generated more fatalities (P=.0001).
Rollover accidents injure more pople, and accidents with no rollover kill more occupants. It appears as if the no
rollover condition produces greater impact forces, preventing the rotorcraft from bouncing and rolling;
consequently, the higher fatality rate. Fires produce more fatalities but not more injuries. Autopsy data might
explain this, but smoke inhallation during the evacuation and the speed of the evacuation warrent more
attention, since it may be a contributing factor in the number of fatalities.
17. Key Words
18. Distribution Statement



Accident Incident Data System, Rollover, Fire, Rotorcraft,
Document is available to the public through the


Defense Technical Information Center, Ft. Belvior, VA

Evacuation, Chi Square, CFR Title 14 Part 29.807, Injury, 22060; and the National Technical Information
Fatality


Service, Springfield, VA 22161
19. Security Classif. (of this report)
20. Security Classif. (of this page)
21. No. of Pages
22. Price

Unclassified Unclassified 9
Form DOT F 1700.7 (8-72)
Reproduction of completed page authorized
i

---

---

FATALITY AND INJURY RATES FOR TWO TYPES OF ROTORCRAFT ACCIDENTS
INTRODUCTION
Rotorcraft Operations are not routinely associated with
fi res and toxic smoke on evacuating passengers. Although
high passenger loads, but it is an area that accounts for a
commercial aviation accidents are capable of produc-
signifi cant part of general aviation activity. Statistics show
ing the highest number of fatalities per accident in the
rotorcraft generally transport fewer than four passengers
aviation industry and warrant strict regulatory oversight,
(2), thus, rotorcraft accidents do not generally get the
general aviation accidents also produce post-crash fi res
kind of attention afforded large commercial aircraft where
that may exacerbate the injuries and increase the number
the potential for a large number of injuries and fatalities
of fatalities.
is greater. Ensuring that all rotorcraft occupants have a
The large volume of highly combustible fuel carried
rapid means of escape is, however, of importance.
by commercial and general aviation aircraft increases the
Airworthiness standards for normal and transport
threat of an in-fl ight or post-crash fi re in a survivable ac-
category rotorcraft operations are governed by the Code
cident. The National Transportation Safety Board (NTSB)
of Federal Regulations (CFR), Title 14, Aeronautics and
defi nes a survivable accident as one in which the fuselage
Space, Parts 27 and 29, respectively. Normal category
remains basically intact, the impact forces are within hu-
rotorcraft have a maximum weight of 7,000 lbs with 9
man tolerance limits, and the seat belts restrain the pas-
or less passenger seats and must have at least 1 emergency
sengers during impact. Using this defi nition, fatalities in
exit on the opposite side of the cabin from the main
survivable accidents are caused by events that occur after
door. This emergency exit must consist of a movable
the initial impact. That is, the crash impact forces do
window, panel, or additional external door, to provide an
not kill the occupants, but rather the post-crash fi re and
unobstructed opening that will accept a 19-by-26 inch
toxic smoke routinely cited in the autopsy reports are the
ellipse, be readily accessible, require no exceptional agil-
primary cause of death. For this reason, government and
ity of the person using it, and be located so as to allow
industry have continually sought ways to not only prevent
ready use without crowding in any probable attitude that
the threat of aircraft fi res but to expedite the evacuation
may result from a crash (3). Transport category rotorcraft
of the occupants once a fi re has developed.
have a maximum weight greater than 20,000 lbs with 10
This study analyzes the frequency of rotorcraft ac-
or more passenger seats. The number of emergency exits
cidents involving fatalities and injuries to determine if
located on the opposite side of the main door is based on
certain types of accidents are inherently more dangerous
seating capacity; however, CFR Title 14 Part 29.807(c)
in relation to rapid evacuation capability. Four categories
specifi es a requirement for exits on the top, bottom, or
of accidents were analyzed: those involving a fi re, those
ends of the rotorcraft. Specifi cally, there must be enough
without a fi re, those in which the rotorcraft rolled over,
exit openings in the top, bottom, or ends of the fuselage
and those without a rollover. It was hypothesized that
to allow evacuation with the rotorcraft on its side, or the
rollover accidents create evacuation delays that produce
probability of the rotorcraft coming to rest on its side in
more fatalities, particularly in situations involving a
a crash landing must be extremely remote (4). The intent
rollover and post-crash fi re, where evacuation delays
of these regulations is to provide occupants with a viable
may expose occupants to toxic fumes longer than they
means of escape in any type of rotorcraft accident.
would be if the rotorcraft remained upright and the
With the large number of passengers traveling on com-
evacuation only required occupants to quickly step out
mercial fi xed wing aircraft, it is imperative that evacuations
of the rotorcraft.
are effi cient, expeditious, and can be executed without
injury. CFR Title 14 Part 29.803 requires manufacturers
DATA SOURCE
of transport category rotorcraft to demonstrate that a
full load of passengers can exit the aircraft in 90 seconds
The Federal Aviation Administration (FAA) maintains
or less, in darkness, using half the emergency exits, and
a database of every accident investigated over the last 25
with only emergency lights illuminating the cabin (1).
years. This database (approximately 140,000 accidents
Requiring these evacuations to be completed within 90
dating back to 1973), the Accident Incident Data Sys-
seconds has signifi cantly reduced the effects of post-crash
tem (AIDS), has become the comprehensive source of
1

---

i nformation for all accidents investigated by the FAA.
Aircraft make and model, phase of fl ight, weather
Rollover Accidents 365 (14%)
(Includes Fire & Non Fire
conditions, accident location, date of the accident, Total Number (2704)
Accidents)
primary and secondary causes, a narrative describing of Helicopter
Non Rollover Accidents 2339 (86%)
Accidents (1986 – 97)
the accident, and the number of fatalities and injuries
(Includes Fire and Non Fire
Accidents)
comprise a small portion of the information collected
from each event. In this study, the narratives for each
Fire Accidents 148 (5%)
accident were analyzed to identify accidents where a
(Includes Rollovers and Non
Rollovers)
rotorcraft rolled over and/or caught fi re. Statistical Total Number (2704)
of Helicopter
analysis of the data was then conducted to show if one
Non Fire Accidents 2556 (95%)
Accidents (1986 – 97)
(Includes Rollovers and Non
type of accident consistently produced more fatalities
Rollovers)
or injuries than another.
Figure 1. A Comparison of Accidents by Rollover and Fire
Status.
METHODS
searches for rollover, a total of 148 accidents involving
There are 181 fi elds used to describe an accident or
fi re, and 365 accidents involving rollover, were identifi ed.
incident in the AIDS database. Most of these fi elds are
A comparison of the accidents by rollover and fi re status
coded with letters or numbers specifying a particular
is shown in Figure 1.
condition, action, or consequence associated with the
In addition to the descriptive accident information in
event. Even though the database contains a large amount
the database, a count of the number of injuries and fatali-
of information, unique questions concerning the cause of
ties associated with each event is available. Two statistical
injuries and fatalities associated with fi res and rollovers
tests were used to compare the number of fatalities and
are not easily answered.
injuries for each accident type to determine if one type
The two factors in rotorcraft accidents thought to be
of accident signifi cantly produced more fatalities or in-
the most likely reasons occupants are not able to rapidly
juries than another. The fi rst test analyzed the frequency
egress a rotorcraft are rollover and fi re. However, no fi elds
of fatality and injury accidents for the different groups,
in the database specifi cally allow accidents to be identi-
while the second test compared the fatality and injury
fi ed by rollover and fi re. Descriptions (e.g. in-fl ight fi re,
rates of the subsets of accidents that had at least one
engine fi re, post-crash fi re, and cockpit fi re) are routinely
fatality or injury.
used to explain the primary or contributing cause(s) of an
The Chi Square (χ²) Test of Independence for Fre-
accident. Consequently, multiple searches must be used
quencies is used when data are arranged in categories.
to identify those situations where a fi re has occurred. In
Comparisons are made between the observed and expected
addition to the coded fi elds in the database, a narrative
frequency to determine dependence or independence
text fi eld is available that briefl y describes the accident.
(5). Accidents involving rollovers and fi res make up the
By conducting word searches on this narrative fi eld, it is
different categories of information, while the number
possible to categorize accidents by rollover and fi re.
of accidents with fatalities and injuries represents the
A search of the AIDS database for accidents occurring
frequency of occurrence in each category. The analysis
between January 1, 1986, and March 4, 1997, showed
focused on the difference between rollovers/non-roll-
there were 2870 rotorcraft accidents or incidents reported.
overs and fi re/no-fi re accidents. Further analysis was
Screening of these accidents revealed 166 that did not
also performed on combinations of these two categories
accurately report the number of injuries or fatalities
to determine if an event such as a rollover with fi re pro-
and were excluded from the population sample. The
duced more fatalities or injuries than a rollover without
remaining 2704 records became the dataset on which
fi re. Table 1 shows the four main categories, as well as
all other queries were based. Using the fi re codes: in-
the combinations of these categories that were analyzed
fl ight fi re, post crash fi re, and engine fi re, as well as word
using the χ2 statistic.
Table 1. Analytical Design for Each Accident Category
Fire No
Fire
Rollover
Rollover & Fire
Rollover & No Fire
No Rollover No Rollover & Fire No Rollover & No Fire
2

---

The second statistical test performed on the data
Table 2 shows the number of fatality and injury
compared the fatality and injury rates for the different

accidents for all categories.
accident groups. The mean number of fatalities and
The χ²results for the rollover and fi re categories are
injuries were calculated and a comparison of means was
listed in Tables 3 and 5, respectively.
performed using the t statistic (6).
Comparisons of accidents involving rollovers versus
those without rollover showed a signifi cant difference
RESULTS
between the two groups; fatal accidents occurred more
often without rollover while injury accidents occurred
Figure 2 shows the total number of accidents,
more often with rollover (Table 3).
accidents with fatalities, and accidents with injuries
The χ² values for the rollover accidents with fi re vs
for each category.
rollover accidents with no fi re showed there was a greater
The χ² analysis for accidents with fatalities and ac-
number of fatal accidents in the rollover with fi re category.
cidents with injuries was initially run on the two main
In contrast, no difference was found in the number of
categories: rollover vs. non-rollover and fi re vs. no-fi re.
injury accidents in these groups. A greater number of
These were then divided into the subcategories shown
fatal accidents with fi re but not rollover was found rela-
in Figure 3. Using the data in Table 2, separate χ² values
tive to those with neither fi re nor rollover; again, there
were also calculated on the number of fatal and injury
was no difference in the number of injury accidents for
accidents in each of the remaining four categories.
the two (Table 4).
Total Number of
Helicopter Accidents
2704 Accidents
662 Injury Accidents
320 Fatal Accidents
No Rollover
Rollover
2339 Accidents
365 Accidents
546 Injury Accidents
116 Injury Accidents
299 Fatal Accidents
21 Fatal Accidents
No Rollover
No Rollover
Rollover
Rollover
No Fire
Fire
No Fire
Fire
2214 Accidents
125 Accidents
342 Accidents
23 Accidents
509 Injury Accidents
37 Injury Accidents
108 Injury Accidents
8 Injury Accidents
247 Fatal Accidents
52 Fatal Accidents
13 Fatal Accidents
8 Fatal Accidents
Figure 2. Total Number of Accidents, Accidents With Fatalities, and Accidents With
Injuries for Each Category.
(1)
(4)
Rollover
Fire
vs
vs
No Rollover
No Fire
(2)
(3)
(5)
(6)
Rollover & Fire
No Rollover & Fire
Fire & Rollover
No Fire & Rollover
vs
vs
vs
vs
Rollover & No Fire
No Rollover & No Fire
Fire & No Rollover
No Fire & No Rollover
Figure 3. Subcategories of Accidents.
3

---

Table 2. The Number of Fatality and Injury Rates for All Categories.
Accidents Accidents Accidents Accidents
with no
with at
with no
with at
Fatalities
Least 1
Injuries
Least 1
Fatality
Injury
Rollover (365)
344 (94%)
21 (6%)
249 (68%)
116 (32%)
No Rollover (2339) 2038(87%) 299
(13%) 1791(77%) 546
(23%)
Fire (148)
88 (59%)
60 (41%)
103 (70%)
45 (30%)
No Fire (2556)
2294(90%) 260
(10%) 1939(76%) 617
(24%)
Rollover Fire (23)
15 (65%)
8 (35%)
15 (65%)
8 (35%)
Rollover No Fire
329 (96%)
13 (4%)
234 (68%)
108 (32%)
(342)
No Rollover Fire
73 (58%)
52 (42%)
88 (70%)
37 (30%)
(125)
No Rollover No
1965(89%) 247
(11%) 1703(77%) 509
(23%)
Fire (2214)
Table 3. χ2 Results for the Rollover Category.
Accident Type
Fatal Accidents
Injury Accidents
Rollover (1)
χ2 (1, N =2702)=14.98, p = .0001* χ2 (1, N =2702)=12.09, p = .001*
No Rollover

Table 4. Results of the Rollover and Fire Categories.
Accident Type
Fatal Accidents
Injury Accidents
Rollover Fire (2)
χ2 (1, N=365)=38.15, p < .0001*
χ2 (1, N=365)=0.10, p=.749
Rollover No Fire
No Rollover Fire (3)
χ2 (1, N=2337)=98.22, p < .0001* χ2 (1, N=2337)=2.87, p = .09
No Rollover No Fire
Table 5.
Results for the Fire Category.
Accident Type
Fatal Accidents
Injury Accidents
Fire (4)
(1, N =2702)=123.5, p < .0001*
(1, N =2702)=2.95, p = .086
No Fire
* Signifi cance greater than or equal to .001
4

---

The χ²results for the fi re category are listed in Table 5.
or injury. A comparison of the mean number of fatali-
Comparison of the accidents with fi re versus those
ties and injuries for the subsets of accidents with at least
without fi re showed there were more fatality accidents
1 fatality or injury made it possible to determine if one
involving fi re than without. There was no difference in
accident type, on average, produced more fatalities or
the number of injury accidents based on fi re condition
injuries than another. (See Table 7 for a breakdown of
(Table 5).
accidents, fatalities, and injuries.)
Comparing accidents with fi re and rollover with accidents
Toward this end, a comparison of the means was run
with fi re but not rollover revealed no difference for either
using a t test (5). The results are listed in Table 8.
the fatality or injury groups. Accidents with rollover but
In most cases, the τ statistic showed little difference
not fi re, and accidents with neither, produced χ²values
in the average number of fatalities or injuries for the
for fatalities and injuries that were consistent with the
different accident categories. The results in Table 8 did
main rollover categories (Table 6).
reveal, however, that accidents without rollover produced
Determining a difference among various accident
more fatalities than those with rollover. This indicates
types does not mean there is an analogous difference in
that not only is the frequency of fatal accidents in the
the total number of fatalities or injuries. For example,
non-rollover group higher, but more people on average
an equal number of fi re versus non-fi re fatal accidents
are killed in accidents when the rotorcraft does not roll
does not mean the total number of fatalities for the two
over. There also were more injuries in the rollover no-fi re
categories will be the same. Since the objective was to
group than the rollover fi re category. Among accidents
look at the difference between the total number of fa-
with a fi re, it was the non-rollover category that produced
talities and injuries for the different accident types, we
the most injuries.
decided to analyze accidents involving at least 1 fatality
Table 6. Comparison of Accidents With Fire Versus Those Without Fire.
Accident Type
Fatality Accidents
Injury Accidents
Fire Rollover(5)
χ2 (1, N=148)=.38, p = .54
χ2 (1, N=148)=.25, p = .62
Fire No Rollover
No Fire Rollover
χ2 (1, N=2554)=17.57, p < .0001* χ2 (1, N=2554)=11.87 p = .001*
(6) No Fire No Rollover
Table 7. Breakdown of Accidents, Fatalities, and Injuries.
Accidents
Total
Accidents
Total
1 or More
Number
1 or More
Number
Fatalities
Fatal Mean Injury
Injured Mean
Rollover 21
26
1.238
116
178
1.534
No Rollover
299
527
1.763
546
927
1.698
Fire 60
109
1.817
45
72
1.600
No Fire
260
444
1.708
617
1033
1.674
Rollover Fire
8
12
1.500
8
9
1.125
Rollover No Fire
13
14
1.077
108
169
1.565
No Rollover Fire
52
97
1.865
37
63
1.703
No Rollover No Fire
247
430
1.741
509
864
1.697
5

---

Table 8. Results of t-test of Means Comparison.
Fatal Injured
(1) Roll vs No Roll
t (34) = 3.36 p = 0.002*
t (187) = 1.5 p = 0.135
(2) Fire vs No Fire
t (66) = 0.41 p = 0.687
t (50) = 0.40 p = 0.689
(3) Roll Fire vs Roll No Fire
t (8) = 1.26 p = 0.245
t (19) = 2.73 p = 0.013*
(4) No Roll Fire vs No Roll No Fire
t (56) = 0.41 p = 0.684
t (41) = 0.02 p = 0.981
(5) Roll Fire vs No Roll Fire
t (21) = 0.83 p = 0.418
t (40) = 2.35 p = 0.024*
(6) Roll No Fire vs No Roll No Fire
t (37) = 6.47 p < 0.001*
t (170) = 1.15 p = 0.251
* Signifi cance greater than .05
DISCUSSION
effects would disable occupants more than would the
crash alone, exacerbating diffi culties with evacuation and
The effects on passenger evacuation related to rollover
increasing the likelihood that passengers would succumb
and fi re generally met expectations. For example, there
to the worsening environment. Thus, the effects of ro-
were more injury accidents with rollover than without, as
torcraft fi re on passenger evacuation warrants particular
would be expected when considering the crash dynamics
attention, especially in situations in which the rotorcraft
related to occupants being thrown around the cabin during
has rolled over.
rollover. In contrast, there were unexpected results, such
as the greater number of fatal accidents attributable to
CONCLUSION
the non-rollover condition. A likely explanation for this
circumstance could be that the impact forces that cause
Of the 2,704 crashed rotorcraft included in the analy-
a rotorcraft to roll are less severe than those encountered
sis, approximately 14% rolled over and 5% caught fi re.
in a non-rollover situation, where the rotorcraft hits the
These percentages, while small, are not insignifi cant; that
ground with such direct force that it embeds fi rmly. These
is to say, there is more than a remote possibility that a
extreme forces would also explain the relatively greater
rotorcraft will roll over and/or catch fi re in an accident.
number of fatalities in accidents that do not involve roll-
The demonstrated results of these crash factors makes
over. Comparison of these two situations suggests that
the provision of emergency exits in appropriate numbers
the impact, and not the rollover, is the more important
and locations a necessity.
consideration for passenger evacuation.
Evaluation of the effects of fi re in rotorcraft accidents
also produced mixed results. There were more fatal ac-
REFERENCES
cidents with fi re than without; however, there was no
difference in the number of injury accidents for the two
1. Title 14, United States Code of Federal Regulations,
conditions. The reason fi re produced more fatalities,
Part 25, Section 803, 1997.
but not injuries, is not immediately obvious, although
2. Federal Aviation Administration Accident Incident
a likely explanation would be that the impact/rollover
Data System, 1998.
forces primarily injured occupants, while accidents
involving post-crash fi res included the effects of both
3. Title 14, United States Code of Federal Regulations,
crash dynamics and the heat/toxic byproducts produced
Part 27, Section 807, 1997.
by the fi re. This would also explain the larger proportion
4. Title 14, United States Code of Federal Regulations,
of injuries in rollover accidents without fi re, as compared
Part 29, Section 807c, 1997.
with those having fi re; i.e., the presence of fi re makes
the relative number of injuries appear lower, since many
5. Havlicek, L.L., Crain, R.D., “Practical Statistics for
initially injured passengers may be killed by smoke or
the Physical Sciences,” 1988, pp. 194.
fi re. The inaccessibility of autopsy data precludes the
6. Havlicek, L.L., Crain, R.D., “Practical Statistics for
ability to draw appropriate conclusions regarding such a
the Physical Sciences,” 1988, pp. 169.
speculation, although the combination of crash and fi re
6

---