I ~~LEVEL'I5

SSAFETY STUDY -

PRIME AUCLEAR AIRLIFT FORCE Ip.HAF)

MISSIONS
INVOLVING THE USE OF C-13 AND .141 AIRCRAFT \

Major John 6. Dean

DTIC
ELET

""- - - '.,- ,.- . ._ ,. , ... .

lj.t

DIRECTORATE OF PUCLEAR SURETY

Air Force InspectioN I-Nd Safety Collor
.A.irthand Air Force Base, New Mexico $11117

i C pb paove

81 8 24 002 1t
:.,. ', .- *. . . . ... ~-. -
. . ...- ---- - -- - - - . - - - _ . . .--. . .. ' -
 I \w

THIS" D0 UMENT IS BESTl

QUALITY AVAILABLE. TIE COPY
FIURNISHED TO DTIC CONTAINED
A SIGNIF!CANT NUMBER OF
F2:ICLES -HETH DO NOT

THIS DOCUM4ENT CONTAINED

REPRODUCED FROM BLANK PAGES THAT HAVE
BEST AVAILABLE COPY BEEN DELETED
AFISC-TR-81-001
IiI
This final report was prepared by the Directorate of Nuclear Surety,

Kirtland Air Force Base, New Mexico, Major John G. Dean was the Project
Officer.

This report has been authored by an employee of the United States

Government. The United States Government retains a nonexclusive, royalty-

free license to publish'br'reproduce the material contained herein, or allow

others do so, for the United States Government purposes.

This report has been reviewed by the Public Affairs Office and Is

releasable to the National Technical Information Service (NTIS). At NTIS,

it will be available to the general public, including foreign nations.

This technical report has been reviewed and is approved for publication.

WIL'1A1 J ENDR S
Director of Nuclear Surety 4
I. iI T 41

'S,'
N,-VV

-- -
 UNCLASSIFIED
SECURITY CLASSIFICATION Oft Y41S PAGE (Sh.ord~eittd

REPORT DOCUMENTATION PAGE .. RE COMSTPcLTINGS - -

1. RMPORfT
,,,,,
s.,,-001.oo
NUMU.R
n.;o 3
GOVT ACCESSOIN NO, -- RECIPIEN-"S CAYALOG NUMB "

4. TITLE (anid SUSiI) ,S TyYPE OP 111POW10 & P9RIOO COVEMEI1O

Safety Study - Prime Nuclear Airlift Force
(PNAFr Missions Involving the Use of C-130 Final Report
and C-141 Aircraft *, 0FORMIN
4...
.iota. RePORYT
_- MUMVIE

7 AUTHOR(I) S. CONTRACT O9 GRANTr NUMU0,R*)

Major John G. Dean

S. PERFORMING ORGANIZATION NAME AND AOORESS 10, PROGRAM EILMENT, PROJECT. TASK
AREA & WORK UNIT NUMiMERS
Directorate of Nuclear Surety/SNA
Kirtland AFB NM 87117
* ,|ti, CONTROLLING OFFICE NAME AHO AOORKSS 1'. REPORT OATE
January 1981
Directorate of Nuclear Surety/SNA PAGES
Kirtland AFB NM 87117
14. MONITrORING AGENCY NZMi AODR111$(l1 dlfflel~tn lmm Co~nt.lllne 0e1i09)
t3. NUMSERO 32
IS. iCU,RITrY CLASS. (of this *',fIN.)

iUnclassified
; |S=, OE{ FICAIFATIONi
OOWNGRABING

':1 'IS.~~
OIS'TRI11UT"ION STATEMlENT Wte his RetoottlSG [O L

!Approved for public releasel distribution

i" :iunl imqt ed.

* T17.Utai m ,oUu m at ^i .n t o itse suagrilLc nMered II 01oCM JU, It gtjf st I 1g.

~ R"Oft)

Approved for public Yelease; distribution unlimited.

"is,SUPPLEMENTARY NOTES

19. KEY wORDS (Cowmuoiuon rov.,.. suitio moioayad Iduutity by block numbor)
n~iti

Broken Arrow C-130, C-141

Accident Rate Air Transportation
Prime Nuclear Airlift Force (PNAF) Nuclear Cargo

10. AISTRACT (Contidiw a" ,.vorue. side It nooaieiae? and Idenltiy by block dumb~t)

This report estimates the accident rate for C-130 and C-141 aircraft while
involved in the air transportation of nuclear weapons as cargo. Causes and
factors involved in accidents are shown, and beneficial .practices are
identified.

N UNCLASSIFIED
DD , 1473 EDIotI oN
O ovGi,OUsoaEtE
SECURITY CLASSIFICATION OF THI4S PAGE (WhImenDito EnW6

P.r~l ,r" IIr.rt :

 AFS-R8-001I
CONTENTS

INTRODUCTION3

OBJECTIVES4

4
DISCUSSION
9I
* DATA BASE

KEYS TO ACCIDENT TABLES

25
ACCIDENT RATES

2
CONCLUSIONS

3
REFERENCES

3
DISTRIBUTION LIST

APPENDIX

INTISG~~
DTIC TOB
BYa

1oti

trbtin.

Code

4*xl~bli'
II AFISC-TR-81-001

ILLUSTRATION
Figure Page
1 Classification and Type of Service, U.S. Air Carriers 13

TABLES
Table Pae
1 C-141 Total Flying Exposure by Year 12

2 C-130 Total Flying Exposure by Year 12

3 Accidents, Rates by Aircraft Make and Model, U.S. 14

Certificated Route Air Carriers, All Operations,
*i 1968-1978 (1978 Preliminary)
4 Accident Rates and Exposure for Selected Aircraft Types, 14
All Operations, All U.S. Air Carriers

C-141 Accidents 18
5

6 C-130 Accidents Involving Aircraft Destruction 20

7 Accidents Involving Selected Aircraft Types, 22

All U.S. Carriers, All Operations, 1967-1976
8 Aircraft Destroyed (C-130, C-141, Commercial) 27
9 Accident Rates for Accidents Resulting in Destruction 27
of the Aircraft (Per 1,000,000 Departures)
10 Aircraft Accidents by Phase of Flight 27

11 Accidents by Cause/Contributing Factors 28

12 Causes/Factors-Certificated Route Air Carriers, 28

1968 through 1977, from NTSB Report

2
 AFISC-TR-81-001

INTRODUCTION
This is a study of the safety aspects of transporting nuclear weapons by

military cargo aircraft. The safety history of nuclear cargo airlifts and other
related operations was used to predict the expected frequency of accidents.

Several kinds of accidents could have been chosen as a basis for the study.
The basis chosen was that of a "Broken Arrow" accident.

A Broken Arrow is defined as an accident or unexpected event involving a

nuclear weapon that results in any of the following consequences: nuclear

detonation; nonnuclear detonation or burning; loss, theft, seizure, or destruc-

tion; radioactive contamination- actual or perceived public hazard. Some

"elements of this definition describe events that are much more likely to happen
than others. The most likely is "perceived public hazard." This is Judged to
correspond to any accident in which an aircraft carrying a nuclear weapon is
* destroyed or irreparably damaged.
Occurrences of "actual public hazard" are much less likely to occur.

Studies by Sandia National Laboratories 1,2

' attempt to statistically describe

P, accident environments and the response of classes of nuclear weapons that could
be involved in those accidents. This approach gives probability numbers for
7.1 the occurrence of accidents defined in terms of what actually happens to the
3
weapons. Other studies incorporate damage models that attempt the next step
of finding the probability of occurrence of accidents defined in terms of what

ultimately happens to people and property.

_ /There is considerable uncertainty in each of these steps, especially when

the accident is defined in terms of consequences to people and property. These

uncertainties arise because the data consist of small or ambiguous samples.
The justification for "one more study" must ultimately rest on the reduction
of some of this uncertainty and therefore on data. This study is bosed on a
large body of accident reports that were carefully screened to obtain a

"u ... ,, .. 1
 AFISC-TR-81-001

consistent set applicable to nuclear airlift operations. The data is presented -

in tabular form in the report, and individual summaries of the accidents are in
a separate appendix.

OBJECTIVES
Determine the expected frequency of occurrence of accidents to cargo air-
craft transporting nuclear weapons that would result in destruction of or
irreparable damage to the aircraft.
Identify factors contributing to the accident rate that can be changed by

improvements to the system.

DISCUSSION
The study objectives require determination of an accident rate for C-130

and C-141 aircraft. The rate needed is destroyed aircraft per amount of flying
]
*exposure. We will primarily use a "departure" as a unit of flying exposure,
* Iwhere a departure is one takeoff (followed ultimately by landing and including

* all between). The preference for departures, instead of miles or hours of

flight, is because the accident data show a very low incidence of accidents
in cruise flight. Also, to keep the magnitude of the numbers near one, the

rate will usually be expressed as destroyed aircraft per million departures.

If nuclear weapons were carried as routine cargo on a representative
sample of all kinds of C-141 and C-130 missions, a very direct analytical

approach would suffice. Assuming a similarity between the operational condi-

tions of the recent past and of the near future, one could use the observed

accident rate to predict the future accident rate by statistical means.

PNAF operations are not strictly typical of all C-141 operations or of

all C-130 operations. But, despite the differences that exist, the direct

approach could still be used if the historical rate used was PNAF destroyed
aircraft per million PNAF departures. This historical rate for both the C-130

and C-141 is zero; however, we will show that this fact permits little precision

4
 "AFISC-TR-81-001

in the analysis since it would almost always be observed. That is, we will

later show that the C-141 has an accident rate of about three and one-half
destroyed aircraft per million departures and the PNAF C-141 rate is of the
order of one per million departures. The whole history of C-141 PNAF flying

is of the order of 10,000 departures. Therefore, assuming the Poisson dis-

tribution applies, out of a large number of samples each of 10,000 C-141

departures one would expect to find zero destroyed aircraft accidents in any
given sample about 96% of the time. Even though we have accurate data on PNAF
accidents (zero of them) and on PNAF departures for both C-141 and C-130 opera-
tions, we cannot precisely predict accident rates by direct methods because the
historical sample is too small.

An indirect method of predicting the PNAF accident rates is to use the

larger sample of historical data, representing all C-141 operations and all
C-130 operations. This data could be used directly if there were no differences

between PNAF flights and typical flights. However, differences are known to
exist and their influence must be allowed for.

A significant area of difference is that the overall history will include

many different types of operations, and some of these may be of a class having

a very different accident rate from PNAF operations. An example would be

* combat airlift operations. This atypical class must be excluded from the data

base by deleting the accidents and the departures attributable to the excluded
operations. The remaining data would be a large historical sample of all

operations having approximately the same intrinsic hazards as PNAF operations.

Another source of differences is that factors influencing accident rates

may be present in actual PNAF operations to a different degree than they are
present in the larger "all operations similar to PNAF" sample. These factors

are grouped in this study as factors involving crew selection and training,

factors involving maintenance, and factors involving conditions of flight.

5
/:
 AFISC-TR-81-001

To make the best possible prediction of accident rate from the "all operations

similar to PNAF" sample, the effect of each of these differences must be

estimated, and a correction for the effect included. It is worth noting that

any exclusions made in going from "all operations" to "all operations similar

to PNAF" are aimed at excluding operations having hazards not found in PNAF
operations; while corrections made for effects of crew selection, maintenance,
and conditions of flight are to account for hazards that are present in PNAF
operations but are possibly present to a different degree.
The step In the analysis of excluding from the data base those types of

operations having, as a class, a very different accident rate would best be

done by examining historical accident rates for all of the various types of
operations. Unfortunately, the data base will not permit this. The accident

' reports are very complete, and one can easily assign an accident occurrence
to a given type of operation and then accumulate totals. However, there is

'1 no detailed breakdown available on flying exposure by type of operation. Thus,

the rates cannot be obtained. The only alternative is to make judgments that
certain operations involve hazards not found in PNAF operations and then to

exclude accidents occurring during those operations. Having done this, one
must then also exclude all of the flying exposure related to those operations.
However, we have already said that the data to make that exclusion is not

available. The unhappy result is that a poorly supportable estimate is

required. In the C-141 data, no accidents that destroyed aircraft are
excluded, and we assume all C-141 operations to be "similar to PNAF." In
the C-130 data, exclusions are needed for actual combat operations, combat

airlift proficiency training, initial crew training including maneuvering

related to combat aircraft, low-level search and rescue, and weather recon-
naissance typhoon penetrations. The excluded accidents and flying exposure

are discussed in the "Data Base" section.

(-.
1I*1 . [ 'I"T I[lI~
 AFISC-TR-81-001

Estimates of the effects of crew selection, maintenance, and conditions of

flight to allow adjustments to the accident rate predicted from "all operations

similar to PNAF" are obtained by examining a body of accident data concerning

commercial aircraft. The comparison involves commercial aircraft generally

similar to the C-141. Four important assumptions are made. The first is that

the correction, used as a multiplier, that is estimated for obtaining the C-141

PNAF rate from the C-141 "all operations" rate is also applicable to the C-130.

Only the C-141 and similar commercial aircraft are actually compared, The
comparison is not repeated for the C-130 and large commercial
turboprop air-

craft. The next two assumptions are that PNAF crew selection results in crew

proficiency equal to that found in the commercial flying used for comparison

and that, likewise, the PNAF maintenance practices result in equipment reli-

ability equal to that in the comparison commercial flying. The last assumption
is that PNAF conditions of flight are less frequently as hazardous as those

found in the comparison commercial flying.

Since the comparison commercial flying has a historical accident rate

that is over three times better than the corresponding C-141 accident rate,

all of these last assumptions tend to project a safer picture of PNAF operations.

S The effect of the crew selection assumption and the aircraft maintenance
assumption is to say that the PNAF accident rate is better than the "all

operations similar to PNAF" rate and, for the C-141, is equal to the compari-

son commercial flying accident rate. If the reader disagrees with the assump-

tions, they at least allow rapid mental adjustments to the conclusions. For

example, the commercial rate is roughly three times better (lower) than the

C-141 "all operations" rate. If one believes that PNAF crew selection and

maintenance practices are ineffective, use the C-141 "all operations" rate.

If one believes that PNAF crew selection and maintenance practices are very

much better than commercial practice, one could estimate a commensurate

S. . , , - -u ~ n- .n - u n l I l l I l l I l I
 AFISC-TR-81-001

further improvement. The assumption made in this study, that of equality, is

based primarily on the author's personal perceptions. A check of the reason-

ability of this assumption was made by providing a draft copy of this study to
the Headquarters, Military Airlift Command office in charge of nuclear airlift
operations and to some Air Force Reserve C-141 pilots who are also commercial m
airline pilots. They concurred that the assumption was reasonable. The special
PNAF procedures for crew selection and maintenance are established by Military
Airlift Command Regulation 55-18, Volume I (Cl). 14 The part applicable to
crew snlection is Chapter 2, paragraphs 2-7 and 2-8. Maintenance is covered

in Chapter 8, especially paragraph 8-2, "Aircraft Selection and Preparation."

The assumption that PNAF flying is less frequently as hazardous as the
comparison commercial flying has to do with the character of the accident
histories for the C-141 and the comparison commercial flying. By the method

used to select and tabulate iccident data in this report, 40% of the accidents
that destroyed commercial aircraft involved weather as a cause or contributing
factor. Only 11% of the destroyed C-141 aircraft similarly involved weather.
Because of the small number (nine) of destroyed C-141 aircraft, one of which
was caused by weather, this apparent difference is not conclusive. However,

I it is supported by the perception that commercial aircrews are under pressure

to adhere to schedules and routinely fly into weather conditions that C-141
aircrews avoid. A National Transportation Safety Board special study11 reports
that 47% of air carrier accidents occur during instrument landing system (ILS)

precision approach, indicating a significantly increased hazard during adverse

weather landings. PNAF missions especially avoid those conditions since the
extra restrictions in their mission planning result In a substantial weather
margin built in. The restrictions that are most effective in this respect are
over-flight restrictions and selection of alternate/emergency airfields with
nuclear airlift support capability. By the time all of the restrictions have

,8- . - .
AFISC-TR-81-001 ,

been met, the flight plan is so constrained that, if the destination area
weather is marginal, you usually just don't go. Avoiding adverse terminal
area weather may further improve the PNAF accident rate by roughly 30%.

DATA BASE
Tabular summaries of all the data used in this study are presented in

this section. Most of the source data is organized in a separate appendix

because the accident reports are privileged and distribution is limited by
AFR 127-4, "Investigating and Reporting U.S. Air Force Mishaps."

While data on several different classifications of aircraft accidents are

summarized, the accident class used as a basis of comparison and for conclusions
in this study is an accident in which an aircraft is destroyed or irreparably
damaged. There have been no such accidents on PNAF flights of either C-141
or C-130 aircraft,

Data on all C-141 flights over the whole history of the aircraft through
1979 are used, in part, to estimate the accident rate for C-141 PNAF flights.

Data on all C-130 flights through 1978 are used, in part, to estimate the

accident rate for C-130 PNAF flights. Data on certain U.S. air carrier
operations are also used. All data on the C-141 and C-130 aircraft were
obtained from the Air Force Inspection and Safety Center at Norton AFB,
California. ,7,8,9 The civil aviation data were obtained from the National
4 10
"Transportation Safety Board (NTSB), Washington D.C. '
Table 1 summarizes total flying for the C-141 aircraft. None of this
total is excluded since no significant amount of C-141 flying differs suffi-
ciently from PNAF flying.
Table 2 summarizes total flying for the C-130 aircraft. Excluded flying
is shown and deducted from the totals. Exclusions were made for flights
conducted under conditions which differ significantly from PNAF flights.
The large number of excluded accidents in the C-130 history of 60 destroyed

9
 AFISC-TR-81-001

aircraft requires a substantial correction to the amount of flying. However,

how much flying to exclude is not known and has to be estimated. This Is

because the flying history data for a type of aircraft is reported in a

separate system from accident reports and is used primarily for different
purposes. Thus, we cannot determine how much flying Is associated, for
instance, with low-level flight operations or with combat-zone operations

where actual combat was taking place. So, we do not have a good basis for
setting the correction.
An estimate is made by noting that the years 1966 through 1973 had the
most departures per year, exceeding other years by about 70,000 departures
. each year. These years span the peak Vietnam war period, so the total correc-
tion for combat-related operations is estimated at 500,000 departures. The
other excluded activities are estimated to account for 200,000 departures
* over the 18-year history of C-130 operations.
Figure 1 shows the categories used by the NTSB in tabulating data on U.S.
air carriers. All of the tables of commercial aircraft accident data use
these categories. The NTSB data are from References 4, 10, 11, and 12. Tables

1 directly extracted from these references are so labeled.

are directly included or condensed in the separate appendix.
References 10 and 12

The data on U.S. air carriers, used to compare to C-141 data, include
all operations of certificated route carriers, supplemental carriers, and
commercial operators of large aircraft that involved aircraft types similar
to the C-141. The aircraft types included are shown in Table 3, along with
their accident rates and total flying hours for the years 1968 through 1977.
Table 3 only applies to certificated route carriers, but their operations

account for 94% of the total flying hours by U.S. air carriers during 1977.
The selected aircraft types shown account for 84.25% of the flying hours for
certificated route carriers during the time period 1968 through 1977.

10
 AFISC-TR-81 -001

The accident rates in Table 4 come from detailed tabulation of commercial

aircraft accidents shown in Table 7. The "All Accidents" category is defined
more restrictively than the NTSB definition which counts accidents in which

passenger injuries occur but the aircraft is undamaged.

Tables 5, 6, and 7 are summaries of the accidents considered in this study,

Table 5 shows C-141 accidents; Table 6 shows C-130 accidents; and Table 7 shows
the commercial aircraft accidents used in this study for comparison purposes.

These tables summarize the circumstances of the accidents in four broad areas:
SIaccident class; cause of the accident; phase of flight in which the accident

occurred; and categorization of the type of accident. The commercial accidents

in Table 7 have a reduced list of causes and factors and are not categorized

by accident type. A full list of definitions is provided in the "Keys to

Accident Tables."

&.

'1*'

"i ~11

F~ ~ * - *
 AFISC-TR-81-001

TABLE 1. C-141 TOTAL FLYING EXPOSURE BY YEAR

YEAR HOURS FLOWN NUMBER SORTIES NUMBER DEPARTURES
65 35,367 37,450
66 189,240 39,794 122,007.
67 461,772 96,082 194,333
68 672,627 163,439 244,166
69 642,291 208,654 253,917
70 612,518 147,266 251,790
71 487,929 125,318 235,288
72 471,440 121,151 213,996
73 362,532 97,014 181,814
74 286,377 78,500 177,351
75 314,771 85,134 169,149
76 281,622 77,981 155.365
77 299,191 83,461 171,598
78 - 282,594 81,205 170,983
TOTAL 5,400,277 1,404,998 2,577,256
* (2.08 Hr/Departure)

1
i TABLE 2. C-130 TOTAL FLYING EXPOSURE BY YEAR

YEAR HOURS FLOWN NUMBER SORTIES NUMBER DEPARTURES

65 554,237 313,325
66 730.887 242,761 469,245
67 659,861 283,436 448,183
"68 594.058 334,372 446,338
69 537,126 350,559 436,509
70 504,113 241,335 422,852
71 487,137 185,962 430,005 .4
72 480,989 156,418 413,695
73 399,605 131,720 374,987
74 360,549 117,736 371,934
75 365,181 151,764 383,740
76 336,592 124,444 323,726

1
77 334,624 126,973 335,040
78 348,168 144,420 364,841
TOTAL 6,693,047 2,590,900 5,533,420
Excluded (Combat-Related) - 500,000 Departures
Excluded (Other) - 200,000 Departures
PNAF rotal - 4,800,000 Departures
(1.21 Hr/Departure)

12
 AFISC-TR-B1-001
= o--
4 4.J4.)- r. 1 c S.
0E41 w 0 Q
,.0 -U C
"" ) CDr-"0$
W0c4)
'001
LA 4J, (D.U %
- r004)
1- 4--) w4
0.} W >:3 C' (A; U'l-
00 -cc
.IV).J d ,--- " .1- ,, - *I-;
00 01-----
LLJTm
CD0 41-4
r- .,.--4-.
0,,0:
woD -j
*~0 0.00) -'
(A 0.-
.3
. o.L ,': . 1-.
-4c. .9-0
0 CA
LLJ 4-
_!m ,
'0 t I-
4.,' .o L.
U06.
4A-V
131
AFISC-TR-81-001

TABLE 3. ACCIDENTS, RATES BY AIRCRAFT MAKE AND MODEL

U.S. CERTIFICATED ROUTE AIR CARRIERS, ALL OPERATIONS
1968 - 1978 (1978 PRELIMINARY)*
ACCIDENT RATES PER
AIRCRAFT ACCIDENTS AIRCRAFT 100,000 AIRCRAFT HOURS FLOWN
MAKE I MODEL TOTAL FATAL HOURS FLOWN TOTAL FATAL
B-747 28 2 2,851,904 0.98 0.07
B-707 1/ 67 14 2/ 10,906,499 0,61 0.10
B-720 10 1 1,947,518 0.51 0.05
8-727 93 10 20,299.441 0.46 0.05
8-737 12 1 2,952,316 0.41 0.03
DC-8 56 5 6,296,514 0.89 0.08
0C-9 43 11 2/ 9,409,311 0.46 0.10
DC-10 12 2 1,975,911 0.61 0.10
L-I011 12 2 1,052,458 1.14 0.19
CV-880 5 1 687.067 0.73 0.15
BAC-i-1l 8 0 1-040,980 0.17 0.00
TOTAL 346 49 59,419,919 0.58 0.08
1/ A sabotage accident which occurred 8 September 1974 Is included in
all computations except rates.
2/ Includes midair collision accidents nonfatal to air carrier occupants,
excluded in fatal accident rates.
Note: These makes and models
certificated route air of aircraft
carriers, butarethis
the list
most does
widely used by
the entire accident experience for this category of not contain
operations
during the indicated years. The types shown flew a total of
53,585,612 hours from 1968 through 1977, while all types and
models flew 63,597,427 hours in the same time period.
Reference 4

TABLE 4. ACCIDENT RATES AND EXPOSURE FOR SELECTED AIRCRAFT TYPES,

ALL OPERATIONS, ALL U.S. AIR CARRIERS

HOURS FLOWN RATE PER 100,000

YEAR DEPARTURES DEPARTURES ACCIDENTS DEPARTURES
(TOANDS HOUR (100,000) ALL* DESTROYED ALL* DESTROYED

1967 4945 1.0 49.5 12 5 .242 .101

1968 5395 .g6 51.8 20 5 ,386 .097
1969 5678 .91 51.7 27 4 .522 .077
1970 5451 .88 48.0 20 7 .417 .146
1971 5381 .88 47.4 19 4 .401 .084
1972 5309 .88 46.7 24 5 .514 .107
1973 5480 ,87 47.7 19 5 .398 .105
1974 5036 .86 43,3 16 6 .370 .139
1975 5090 .87 44.3 16 2 .361 .045
1976 5247 .87 45.6 13 3 .286 .066
TOTAL 53,013 475.9 186 46 .391
.097
Accidents having damage classified as "substantial" or more by the NTS8.
This differs from the NTSB "All Accidents" rates which include injury-only
type accidents that result in no damage to the aircraft.
Note: Accident occurrences taken from NTSB accident briefs7 which are con-
densed in the appendix.

"14
 AFISC-TR-81-001

KEYS TO ACCIDENT TABLES

USAF -Reports
Injury Classes
F - Fatal
Mj - Major (required hospitalization)
Mn - Minor
N - None
Damage Classes
0 - Destroyed/Irreparably Damaged
MJ - Major
Mn - Minor

N - None
NTSB Reports

Injury Classes
F - Fatal
S - Serious
I.iN - None/Minor
1 X/Y - For collisions with other aircraft, "X" is injuries aboard accident
Io aircraft and "Y" is injuries aboard other aircraft.
Damaqe Classes
D - Destroyed
S - Substantial
M - Minor

N - None

All Reports
Causes/Factors. This includes the following categories of causes and contrib-
uting factors as discernible from the accident report:
Weather

15

. A' -
 AFISC-TR-81-001

Aircrew
Judgment: Aircrew used poor judgment and endangered the aircraft.
Wron Action: Aircrew procedures were improper (misapplied controls,
etc.j.I

Communication: Aircrew communication procedures were improper

(failed to make a communication, used wrong communication procedure,
missed hearing a communication, or misunderstood a communication).
Crew Rest: Aircrew violated crew rest rules.
Training: Aircrew was inadequately trained in an area significant
to the accident.
Maintenance
Personnel Error: Poor maintenance.
Procedures/Data: Maintenance personnel followed stnding rules, but
the procedures or technical data were wrong or faulty.
Equipment, Test Gear: Faulty maintenance equipment contributed to
the accident.
Airport, Airways, Facilities
Controller Error: Controller (including all ground personnel who
issue instructions, clearances, and other information to the aircrew)
made an error.
Communication: Same as for alrcrew communication but applies to
controllers.
Radar, Radio, etc: Ground electronic equipment failure contributed
to the accident.
I, Ground Operations:
the accident. Nonmaintenance ground activities contributed to

Aircraft Materiel Failure

Engine: Includes foreign object damage (FOD).
Instruments, Flight Controls: Self-explanatory.
Navigation, Communication, Radar: Electronic equipment failure.
Landing Gear, Brakes, Tires: Self-explanatory.
Power, Hydraulics: Electric or hydraulic power generation and
distribution system failure.
Airframe and Control Surfaces: Includes spoiler, flaps, And cargo
door failures.
16
 AFISC-TR-81-00i

Other (Self-explanatory)
Phase of Flight (Aircraft status when accident occurred)
Static, Ground Operations: Aircraft was parked or being towed.
This includes parked and undergoing maintenance. Engines and/or
power systems can be running.
Taxi: This includes taxiing on the ramp, taxiway, and crossing
runways. It does not include extension of takeoff or landing roll.
Takeoff, Initial Climb: From start of takeoff roll until departure
of airport vicinity with aircraft stabilized on departure heading,
speed, and climb rate.

Prolonged Climb: From initial climb until cruise altitude,

In-flight Normal: Cruise flight, including altitude changes not
associated with departure or arrival at destination.
Let-Down, Approach: Descent associated with arrival at destination
through start of final approach.

'1
I
Landing: Final approach through turn off of active runway.
Unknown: Damage was detected during postflight inspection, and time
of occurrence cannot be determined.

First Type of Accident (If included, this section describes the initial
occurrence of the accident.)

17

* i
AFISC-TR-81 -001
2, f
PL Ask
II. t. u 6W MUW I
c W 2 W1 4Sd#t 3.'
Cc'
I'Z
6-S .-F
'. LL v.
IC2
tQ Z 1 LL. 45
210 .. ' 4C U. I.u 4i' .. a
2 -
- - - c mc
r: NJ ! ; ? -- eA a l S
- z1
wC
'8 Va. CL 0) goa
SU ~~ ~ a ~ a K c a s wI
-b. -60 L- C. ,a
- -- cC t t
USD 0-
LIOU a- M s
t4 a 3, 4
4'i V f 7
0
-- ena A3 5 V 4 '4
CA* C Q9 0 0
v-
U- 1%"L&CLL
i
WN
ta,
no cc'CULC-si18
 AFISC"TR-81 -001
0 12 X
%L - I 1
U.a cm %T2
i
sA tS a 3
WWI 1 A
O1 m-S
0. t4E*. . -c mn wjJ..
'in .6 U
&-. L t -i.'. ft c . 4 6 i~
w aw a4 asa
. S - = 5. . UY rM w a
In -0L:
A r.. C W A-
4 2%* Ile
-L, Z2 iW i
tt me
w0 0mini
* . r do ad 0' 4iC
C do C
yuu
01 a4 ai M- 190. Zoe
-a cai .
-Y- t4I
'a
ipi ; aI I Cn
C
4. Cc
.6 1.
= 4 e--C 0 4
- -.
'a- Aa~ r 0
-C.
Z .""
'Iii
'-l
co 2q V2 M
19
 AFISC-TR-81 -001I
%A Id
- I., Wa a L, --
C6cmt
uC - -. 6L
- ia 6:; - *
m
44.0 -t -6 ''l1 .6
1:: a 4 1
Lu Eli a.,
04044. ..
C0W b4 P4.4
Ed az-"
Al 44 A 44 --
a.1 41 E
1.41g 4 .:2- V44
44UI t.4t C5 z-3C ,' ,t 3f -O
a -- i .,! 14
4-C S4Ze CL Ea.0 1-5
.jat
-- 'A to 1 'A .n 4 6a
i
- F4
, Z'0I -
C
it cC C j C C Moe
t4a
U~~~~t CC C C
1C c
A3IU C U " 0 0 U'.
o .0 -L2 -O20
:1UC aC. I
4 ~AFISC-TR-81 -001
uJ~x -
I %0 u
CI1C
mm
n e5
La
A35-
.40 'aLmC(
CcC
CM. 0-4 OL - L r I
Li C
14U.0
Ca c
-C -J
410
LA21
- - Sjji
AFISC-TR-81 -001
444
L~. 1 h4 4
1.e ii6n
A - 'A C14 C
- IA W ICV -I0V , A
ae
Ina0' LW a w0w b 9L
a--
Sujtflzu.~~~ u n s u r00~~ncnu gmrt
0
m . CIZ c NWtt
- ~ ~ ~ -.
-- - - - - - - - - -
9, LIV..
W L4 S. I-U t
i b0 A A10 .1V
A-. 0 Z VI nt L -A IlC
NI0 CNNIC mON
A 10, V0
22
*1 AFISC-TR-81 -001
tt Ut U 5
-r c m c %
S-S
c4 Z~2 'A -- ZcI ~ -
1. -. 1! a 1
c~ J~ U
C3 u #AAVf 'A A'A a ~ XQA M tn 0 'Au CA 39 ~ SCwAu0UAiL
u
w~
aVIz ac a: a !a -Att ul Wu LA a
w w
. ~w A adx a
2 o itaa w
CL 0% on cc m mS - C 0% ii-
C3 0'' 8 0 -0 go do
N
in amn doo C O a~n XU do 5 0 0
mago
N NN Nm N NNjcmN Em Nm
U14*
NO~.W NN' ~, N Ny Cm O
ORN N N N
c4c a
*~ uJ
0
'A 0C W IAu0lAOCL n i~ Mta 'A 0 C
A 0Ai 010 'A uViSE
In 0 -da
U U.
IA 'A rt.l. .U yi AM k U. z S.w
caaa oa a as 0 t a n i
;zz0
0d
LO IL.
t0 1 .. N ' I .C.C.NN C N C 000 N NIU
C ~ N I C N N N N C N I S 0 N23
AFISC-TR-81 -001
I 4
1Z I
I ~a m ~~I
x cu u~ as~ xAS 0~~LL0 kuZm
U0 0% 1LAn IA A
39z
mu .6fl . li w MU w IA0 ~ wA a UWtlA nn IA VWtXf Wj *
as cc 3 w j oa' m m ' ~-ago
r. In N N N 04 MO w8
1'' uA IA I
-i - i - - - - - - --
4- 4
.41
-i -- Y -- t 4 - -9 a-
aA a s- a La a I 4A &Aa as a- x sa Vnm
I- ur4A
amw -0t a to 0 0 ami 00a to w wgn.-
CA 2
m M; Za-10
CL
.ja M"a
sn Iii t t L L i ~ o ft O ~ i C d U A E ~ i z n
24a
 AFISC-TR-81 -001

.1 ACCIDENT RATES
Total accidents and exposure for C-130, C-141, and commercial aircraft
similar to the C-141 are shown in Table 8.
The resulting rates are shown in Table 9 and apply fleet-wide to the

I, aircraft types shown. Exclusions have been made only for accidents occurring
during missions completely unlike PNAF missions. No corrections have been

made for pilot selection, maintenance controls, or restrictive conditions of

flight. The 90% and 98% confidence intervals are taken from Molina's tables 1 3
by Interpolation. The 90% interval is found by taking the interval between the
values: "what (high) value of frequency of occurrence would cause the observed
number of accidents or fewer to occur in this number of departures only 5%of
* .the time," and "what (low) value of frequency of occurrence would cause the
1 observed number of accidents or more to occur in this number of departures only
5% of the time," The 98% interval Is similarly defined, except that 1% is used

rather than 5%. The assumption made in determining these intervals is that the
Poisson distribution function is applicable-in this case, a very good assumption.

No further approximations are made as the intervals come from tables of the
actual integral distribution function.
Tables 10 and 11 show accident rates by cause or contributing factor and
by phase of flight. They are taken directly from Tables 5, 6, and 7. Table 12
shows causes and factors from the NTSB annual report (Reference 4) and is shown
for comparison.

.' The phase-of-flight tabulations in Table 10 show that negligibly few

accidents that destroy aircraft (of the type considered in this study) occur
during the "inflight-normal cruise" phase of flight. Virtually all such

accidents occur during takeoff or during letdown and landing, with about
twice as many occurring in the landing phase as in the takeoff phase.

For this reason, it is inappropriate for this study to give accident rates

25

m , - .
 AFISC-TR-81 -001

for destroyed aircraft in terms of hours or miles of flight. The preferred

method is to use departures.

Table 11 shows which causes and contributing factors are associated with

accidents that resulted in destroyed aircraft. Although the meaning of this

table is somewhat obscure, it does contain useful information. First of all,
"note that the table does not apply to normal day-to-day flight conditions.
It applies to literally one in a million accidents. Also, it does not indicate

SI how frequently the accidents occur, only what events were associated with the
accidents when they did occur. An example may help. It is interesting that
in accidents that destroyed commercial aircraft, aircrew errors in judgment or

actions were involved in a little over half of the accidents. This is also
true for destroyed C-141 aircraft, but the accident rates show that the C-141

'I accidents occur nearly four times more frequently. Therefore, Table 11 shows
that, when an accident situation occurred, the military pilots and the civilian
pilots had made the same kind of lapses and errors that led to that accident
situation. When we also look 3t the frequency of accidents (if we simplisti-

cally place all accident blame on pilots), we would conclude that the military
pilots made these same kind of errors four times more frequently. Therefore,

,A the purpose of Table 11 is to characterize the accident, not to describe

"accident rates. Differences appearing in this table show differences in the
circumstances of the accident.
Table 12 shows the percentage distributions of causes or related factors

for commercial aircraft accidents. For the 10-year period (1968 through 1977),

weather was the most frequently cited cause/factor in U.S. certificated route
air carrier accidents, followed by personnel and the pilot. The pilot,

followed by weather and personnel, was the most frequently cited cause/factor

in fatal accidents.

26
 AFISC-TR-81-O01

TABLE 8. AIRCRAFT DESTROYED (C-130, C-141, COMMERCIAL)

AIRCRAFT DEPARTURES AIRCRAFT DESTROYED

C-130 4,833,000 27 (1965-1978)
C-141 2,577,000 9
Commerclal 47,590,000 45
Note: The rates and confidence limits are shown in Table 9.

TABLE 9. ACCIDENT RATES FOR ACCIDENTS RESULTING IN DESTRUCTION

OF THE AIRCRAFT (PER 1,000,000 DEPARTURES)
AIRCRAFT 98% (LOW) 90 LOW) MEAN 90% (H!) 98% (HO
C-130 4.23 4.65 5.59 8.62 9.62
C-141 1.39 1.82 3.49 6.09 7,30
Commercial .67 .74 .97 1.24 1.35
Note: These rates are "overall rates" not "PNAF rates"-see text for
explanation of confidence Intervals.

t.j

iTABLE
r 10, AIRCRAFT ACCIDENTS BY PHASE OF FLIGHT

C.-141 C-130 COMMERCIAL* C-141 COMMERCIAL*

ACFT DES'T ACFT DEST ACFT DEST ALL ALL**
NO. % NO. % NO. % NO. % NO. %
Static 1 11.1 2 6.5 0 0.0 5 10.9 12 6.5
Taxi 1 11.1 1 3.2 0 0.0 6 8.7 24 12.9
Takeoff 1 11.1 9 20.0 11 23.9 6 13.0 36 19.4
Prolonged 0 00 0 0.0 5 10.9 5 10.9 15 8.1
Climb
Inflight 0 0,0 4 12.9 0 0.0 3 6.5 17 9.1
(Cruise)
Let Down 4 44.4 5 16.1 5 10.9 6 13.0 12 6.5
Landing 2 22.2 10 32.3 25 54.3 11 21.7 67 36.0
Unknown - .- 3 10.9 - -

* Selected Aircraft Types, All U.S. Air Carriers, All Operations.

a ** Does not include accidents resulting in passenger injuries without aircraft
damage.

27

... . .. . . ..... . ,. . , . ,-i . . . . . i .....

 AFISC-TR-81-001

TABLE 11. ACCIDENTS BY CAUSE/CONTRIBUTING FACTORS

C-141 C-130 COMMERCIAL C-141 COM4MERCIAL

CAUSE/ ACFT DEST ACFT DEST ACFT DEST ALL ALL
CONTRIBUTING FACTOR NO. % NO. % NO. % NO. % NO. I
Weather 1 11.1 8 25,8 19 41,3 5 10.9 56 30.1
Aircrew: Judgment 4 44.4 7 22.6 7 15,2 8 17.4 21 11.3
Wrong Action 3 33,3 17 54.8 26 56,5 5 10.9 76 40.9
Communication 2 22.2 3 9.7 - - 2 4.3 - -
* Crew Rest 3 33.3 3 9.7 - - 3 6.5 - -
Training 2 22.2 3 9.7 - - 3 6.5 - -
Maintenance or 3 33.3 15 48.4 9 19,6 32 69.6 68 36.6
Aircraft Failure

Maintenance 2 22.2 5 16,1 - 18 39.1 - -

Acft Failure 2 22.2 12 38.7 - - 28 60.9 - -
Airport/Airways/Ground 3 33,3 2 6,5 12 26.1 14 30.4 36 19.4
Operations/Other
. (Total No. of Accidents) (9) (31) (46) (45) (186)

Note: The percentage totals exceed 100% because multiple causes/factors can
be cited in any accident.

TABLE 12. CAUSES/FACTORS-CERTIFICATED ROUTE AIR CARRIERS,

1968 THROUGH 1977, FROM NTSB REPORT*

PERCENTAGE OF PERCENTAGE OF
CAUSES/FACTORS TOTAL ACCIDENTS FATAL ACCIDENTS
Weather 48.3 46.3
Personnel 46.6 42.2
Pilot 39.5 62.5
Airport/Airways/Facillties 9.0 4.7
Landing Gear 8.8 3.1
Power Plant 7.3 4.7
Systems 6.6 9.4
Miscellaneous 6.3 12.5
Instruments/Equipment 2.7 3.1
Airframe 2.4 6.2
Terrain 1.7 0.0
Undetermined 1.2 6.2
Rotorcraft 0.7 3.1
Note: The percentage totals exceed 100% because multiple causes/factors
can be cited in any accident,
* Reference 4

28

LI
 AFISC-TR-81-001

CONCLUSIONS
PNAF C-141 Accident Rate
The accident rate for all C-141 accidents is three and one-half per million

departures.
The commercial accident rate for aircraft types similar to the C-141 is

one per million departures.

* !PNAF differs from all C-141 aircraft in areas of crew selection and main-

tenance, but it is not greatly different in conditions of flight. PNAF differs

from commercial flights in conditions of flight, but it is assumed to be similar
in terms of crew selection and maintenance.
The character of accidents that destroyed C-141 aircraft (non-PNAF) and
similar commercial aircraft shows that commercial accidents involved weather

about four times more frequently and involved maintenance or materiel failure

about half as frequently.

These differences are interrelated and are not separable because they are
not due to independent causes.

If PNAF crew selection and maintenance were equal to the commercial popula-
tion, the difference in conditions of flight would make the PNAF accident rate
lower than one per million departures.

As a conservative high estimate, the PNAF accident rate is Judged to be

* one destroyed aircraft per million departures.
This estimated rate could easily be in error by as much as a factor of two;
however, for this type of problem, a factor of two uncertainty Is not especially
significant.
C-130 Accident Rate

The C-130 accident rate for the whole fleet, considering all flying and
accidents that are not completely unlike PNAF flying, is about five and one-

half destroyed aircraft per million departures.

29
AFISC-TR-81-001

Materiel failure seems to be a more significant factor in C-130 crashes

than for the other aircraft considered in this study. Thus, the special main-
tenance practices for PNAF have a potentially greater effect.
The accident rate for PNAF C-130 missions is estimated to be less than
two destroyed aircraft per million departures.
Use of C-130 Aircraft
Whenever short runways and other adverse field conditions exist, it is
safer to use C-130 aircraft to carry nuclear weapons to and from such fields
than it is to use C-141 aircraft because of the C-130's ability to operate from

smaller airfields.
C-130 and C-141 accident rates are not greatly different. In fact, it is
not possible to state with high statistical confidence that they are different
at all.
PNAF Practlces/Im ortant Factors

Crew selection for skill and maturity is important.

Special maintenance practices and controls are probably valuable. They
are probably most important as applied to the C-130.
Avoidance of adverse weather is important, especially on landings.
All these practices, taken together, probably cause the PNAF accident rate

to be half an order of magnitude lower than the fleet average. They may have
as great an effect as a full order of magnitude reduction,
Accident Reporting
The USAF accident reporting system does an excellent job of reporting the
circumstances of accidents. The use of this accident data is severely limited
by the extremely poor reporting of flying data from which exposure can be
determined.

30
 AFISC-TR-81-001

REFERENCES
1. Accident Environments Expected in Air Force C-5- C-141, and C-130 Aircraft
Accidents, SAND 75-0231, Clarke, Foley, Hartman, and Larson, August 1979.

2. Severities of Transuortation Accidents, SLA-74-0001, Clarke, Foley, Hartman,

and Larson, July 1976.
3. Forward Look, Final Report, Safety and Security Risk Analysis of DOD Peace-
time Nuclear Weapon Transportation, Vol 39, SAND 79-1441, Ling, W. H.,
July 1979. (Secret-FRD)
4. Annual Review of Aircraft Accident Data, U.S. Air Carrier Operations, 1977,
NTSB-ARC-78-2,-National Transportation Safety Board, 6 September 1978.
"5. Accidents, Rates by Aircraft Make and Model, U.S. Certificated Route Air
Carriers, All Operations, 1968-1978, National Transportation Safety Board,
20 June 1979.
'' 6. Major Accidents, 1962 to Date, in C-130 Aircraft When Aircraft Was Destroyed,
Air Force Inspection and Safety Center, 1976. (Privileged Report, AFR 127-4,
Not Releasable Outside the Air Force)

7. C-141 Accidents, 1967 to Date, When Aircraft Was Destroyed, Air Force
Inspection and Safety Center, 15 January 1977. (Privileged Report, AFR 127-4,
Not Releasable Outside the'Air Force)
. 8. C-141 Class A and B Accidents in 1966 and 1967, Air Force Inspection and
Safety Center, 12 October 1979. (Privileged Report, AFR 127-4, Not Releas-
able Outside the Air Force)
9. C-141 Class A and B Accidents, 1968 to Date, Air Force Inspection and Safety
Center, 12 October 1979. (Privileged Report, AFR 127-4, Not Releasable
Outside the Air Force)
10. Briefs of Accidents Involving Turbojet/Turbofan Aircraft, U.S. Air Carriers,
"i967-1978, National Transportation Safety Board, October 1979.
11. Special Study: Flightcrew Coordination Procedures in Air Carrier Instrument
Landing System Approach Accidents, NTSB-AAS-76-5, National Transportation
Safety Board, 18 August 1976.
12. Briefs of U.S. Air Carriers, All Operations, 1967 through 1974, National
Transportation Safety Board, July 1980.
13. Poisson's Exponential Binomial Limit, Molina, E. C., VanNostrand, D., 1942.
14. MACR 55-18, Vol. I (C-l), "Nuclear Airlift Operations,' 19 June 1979.

31
 AFISC-TR-81-001

DISTRIBUTION LIST
HQ AFISC/CC/SER, Norton AFB CA 92409
HQ USAF/IGF/XOOTA, Wash DC 20330
HQ MA.C/DOOM/IGFN, Scott AFB IL 62225
HQ AFLC NSO/LOWQ, Kirtland AFB NM 87117
HQ AFSC/IGF, Andrews AFB MD 20334
AFWL/NTS, Kirtland AFB NM 87117
Field Command Defense Nuclear Agency/FCPF, Kirtland AFB NM 87117
USDOE Office of Military Applications/Department of Safety and Emergency Actions
(DP-29), Germantown MD 20767
USDOE Albuquerque Operations Office/WSSB, PO Box 5400, Albuquerque NM 87115
Sandia National Laboratories/1230/5610/3100/8328, PO Box 5800, Albuquerque NM 87185
Lawrence Livermore National Laboratory, PO Box 808, Livermore CA 94550
Los Alamos National Scientific Laboratory/NSP-SS/ISD-4, Los Alamos NM 87545
National Transportation Safety Board, Info. Sys. Div., Bureau of Technology,
I 800 Independence Ave., SW, Wash DC 20594

'11
'-j

32

. . .