d . levev~ - dtic · final report for january 1977 - september 1977 s[approved for public release;...
TRANSCRIPT
AFFDL. TR-78-8 D . LEVEV~
FACTOR OF SAFETY - USAF DESIGN PRACTICE
Geogre E. Muller and Clement J. SchmidStructural Integrity Branch D D CStructural Mechanics Division
April 1978Bt
TECHNICAL REPORT AFFDL-TR-78-8'
Final Report for January 1977 - September 1977
S[Approved for public release; distribution unlimited.
Best Available Copy
AIR FORCE FLIGHT DYNAMICS LABORATORYAIR FORCE WRIGHT AERONAUTICAL LABORATORIESAIR FORCE SYSTEMS COMMANDWRIGHT-PATTERSON AIR FORCE BASE, OHIO 45433
7 ~
NOTICE
When Government drawings, specifications, or other data are used for any pur-pose other than in connection with a definitely related Government procurementoperation, the United States Government thereby incurs no responsibility nor anyobligation whatsoever; and the fact that the government may have formulated,furnished, or in any way supplied the said drawings, specifications, or otherdata, is not to be regarded by implication or otherwise as in any manner licen-sing the holder or any other person or corporation, or conveying any rights orpermission to manufaqture, use, or sell any patented invention that may in anyway be related thereto,
This technical report has been reviewed and is approved for publication.
,,,< V. ' /
Y!
GEORGE E. MULLER CLIENT J.WSCHID JAMES/JOLSEN, Actg ChiefAerospace Engineer Technical Manager Struc ral Integrity Branch
Structures and Dynamics Division
FOR THE COMMANDER:
RALPH L. KUSTER, Jr., Col, USAFChief, Structures and Dynamics Division
"If your address has changed, if you wish to be removed froi, our mailing list,or if the addressee is no longer emploted by your organizati(on please notift,,W.-P AFB, 0,ý5433 to help us maintain a current mailing list".
Copies of this report should not be returned unless return is required by se-curity considerations, contractual obligations, or notice on a specific document.AIR FORC1E5678o/ 2 1 May 1979 - 270
REPORT DOCUMENTATION PAGE ~ lt. M I ' ;I iI WN~I
M I A',l I""ldON Nil I W I ll N IIN' AtAld NI'M11'I
AlI ill.-1R-/ M ii
- .- -. . ....... N441A
' All I H,164 . INI A, 1 I'M ,M ANI I NumAit "Il.
(Geoillti I .Mu 1I Itr S i 1C Ielivilt. J. Sc hilid
II ~~~~~~~ ..Il.ll .. 'l ... IliA t A'INd-
Alir Ioc HWt iiitht Dirniawic', Ihborat orv IN 4W'i iijh t -!',I t vI,, llSO Alir orco Ilicit, Ohio 4Y1,43W I O n 100
Aiprve l or Flitihl Dyicr t, e~s L 'Ir i htio u iilhll
Airr forc At llawr Ohiown 041, 104tii ' IUN~1
mote v I4 Nd - .t SI ois ll Am 4 i'l-- 0.- 1 -. cbs .If. tini I I0 Ip 1 ,1 , l
I~~~~~~~~~ dl toi'. 1,t '. etvS rt udIRl ,(1 i
Alit iimi' \,if 1t M 1 N~c oI ' -Il -- h li Ise esIoCncp
dApptroe f$ h Ieill, p til t irveml dSt ruc t ii ud1 ih's 1111 Cr1t ed
George Kt' /tl ilhe h tri Clement Jpli'n of Ih'1j w tS u ii~ .1tN o97
Ind let4 v I.1101111 dill''001,1, W J hi 1E ~c r 1W Iiin~ 1 tli ei ls,,il I '
I, dt t tC tv I l a il,- I I Ditto. reet ' trlu till A C oil c tl~te t N y Svaitt d elt it i f t y 1111 dl itl I-1 I -l Ot I u I r I he I i t a01 of s I i' t viS111 o~p
Aircm 'I attv fco- e ib lt M e e'tl oie t
"sIet AN',rly t lctla DAl-N1 'I*Il A,4 Wlt, i,.p.,,.,
A ovo o il itoitldee0117eito tt I! (Y i)rtlvlfail i
UNI.MASS Ill H11UECUPITY CLASI FICA7ION OP IHIS PAOCIfl.', IWO Kwdl
Al though not. wit hodt c rrt i. I s;i, the fac'ttor of sof'ety des iql roncept hasheCo m O ia"amost tniversally accepted lan',ure of fl•ight safety, lhere is,however, a tendency anion(q ng n ltne•n' N to hoth chadllenqe the conltinrued applicationof factors of' safety for effitient aitrframe desiqn, and yet. to avoid anychan•ve that would challnqe the 'oonfidence of future desig;ns. The use ofreliability based concepts will probablvy increase but their appl ication toairframe design may he limited. The factor of safety still covers manycoottinqencies and it appears at this timue there will be a continuing need forsoImxe fact or.
i'•t
; CU:;.; i LiII "Of " . ..
:wTA83
,:> I if__ '
INlI ASSITI Im
I ' t AI PIt( A ThtW 0T 1 O P I h,,-x.a
rI ~ AFFDL-TR-78-8•
FOREWORD
The authors, Mr. George E. Muller, Aerospace Engineer, and Mr. Clement
J. Schmid, Technical Manager, are located in the Critetia and Applications
Group, Structural Integrity Branch, Structural Mechanics Division, of theAir Force Flight Dynamics Laboratory (FBE), Wright-Patterson AFB, Ohio
45433. The work was accomplished under Project Number 2401 "Flight Vehicle
Structures and Dynamics Technology", Task Number 240101 "Structural
Integrity for Military Aerospace Vehicles", Work Unit 24010104 "Structural
Design Criteria Specifications and Design Methods." The effort was
accomplished during the time period January 1977 through September 1977.
The basic text was originally presented at the 45th meeting of the
Structures and Materials Panel of the Advisory Group for Aeronautical
Research and Development (AGARD) in Voss, Norway on September 26, 1977.
The text in this report remains essentially unchanged. Some figures
have been added and minor clarifications incorporated. Corrections of a
historical nature were made to the discussion of the V-G diagrams near
the end of Section I1.
Acknowledgement is made of the assistance provided by Mr. Robert L.
Cavanagh, Dayton, Ohio and the Air Force Museum, Dayton, Ohio in providing
the photoqraphs:
Figure 3. R. L. Cavanagh Collection
Figure 4. A. F. Museum Collection
Figure 5. A. F. Museum Collection
Figure 6. A. F. Museum Collection
Fiqure 8. R. L. Cavanagh Collection
Acknowledqement is made also for the valuable assistance lent the
authors by Or. John W. Lincoln and Messrs. John C. Grogan, Donald B. Paul,
and iohn C. Spa,'ks for their critical review of the report.
ii
AFFDL-TP-78-9
TABLE OF CONTENTS
SECTION PAGE
I INTRODUCTION 1
11 DEVELOPMENT OF THE 1.5 FACTOR OF SAFETY 4
III DEVELOPMENT OF OTHER FACTORS OF SAFETY 26
IV FACTOR OF SAFETY DESIGN CONCEPTS 31
V RELIABILITY BASED DESIGN CONCEPTS 53
VI CONCEPT INTERACTIONS 79
VII CONCLUSIONS 89
REFERENCES 92
- . PAM •i.b
ffi... i..... PA.., ..-. ,IK
V/
AN M -TR-7,tt-8
MUMMARY
The 1.1) fa tor of sdtlet y is dt hiqjhly Visible difiui'tram design para-lilt, trv The, factor is emlp iOU'i ia1 der'ived and prey ides, all a~lmost
UliriVel-i11lv accep1ted IIIIt~'dsu'eof t'lithit s~ift'ty, Al thoughi the measure is
quot h cal lonti the ctv Io a enty pinued by tle ic .io fact~ a eore
ott safepty o stffdardi ea t-tl havfra es i o idvetlpe aI tedency camnyetd
would cha,1vllenge the Collf idenlce, of future, designs. The unsettled Position
Onl tho factor' of, safet v 111,ynee cmpe110telY stabil1i:v but `it can he
clan fie'd by revOiewing it'; hi StDOric sikjimificanIce'
Inl U.S. de'Signl pratt he the( Sign~iititnce of tile 1 .5 factor of
SoIfi t Yct I tdii 1e 1 Itced int perspe;)OLti Ve bytIT reiewing I itsý deveVI ormenV~t for1
both 116 1 it a rv avid L iv iIue Ihe fac tor evolved ac, a Compromi se opinlion
based oil ftl i iht 0rrera t kils . I he .Ippro.~imlAt~e 1 .!) rdti 0 of il t ilihite stress
to Yieold Nres. tfor cort dili mater id Is cone noifii ito use durilnq tihe Same
t em periodl supported thet dlckisonl bUt did 11ot inf~luence the se ec'VLt ion Of
th 1i, . 'fat t o r Iot *aret v. S i rice 111" t ie lit'(f i t Sse Iect ionll, va r i'AtOtios
and aldaplt a t tioll to o t her a ircra ftI t vpes have beenl proposed and somletimles
used . Nrvera I va I i a t i ons anld 0\pert'ilent'll ap 1111i cat iOnsI aT no V reiewed.
I het I AC tir ot f a ftot v (it's i tri concep11)t halts recen t Iy 1 os t sýomet of i tsk
llpeA arid reliail at it v lw sod conce~ptsý have beven emphlils i :ed . As- part of
itss nu turx Idts iII gn'itc1t Oi' dovelt opmeritt proonkirim the, Air O'L- Focehssponsorei-t inves gat loris to doeitI'op re I iabi Iiit~v hww'd cri tri'l . * r Iee
oh t heset inIvest i klat iorsc 'rilt s fimi 1 11r i t it,, betweenl t he factor- of sa fevty
IndIrelidat'1ii it V 01 T Ort ' h Ir VVt soV ar rvee. AlI t huIh1 1k th us0keý otf )-'el i ab1iit y
taIed t mirlt wp v IIpill 'r Itil v i rIlOYea-W , th lit, ai v 1111 i cat ionl to 0 xi rtrralrre
dr'sý i lnr may ltir I In it old . heit fa cIt or of so et v still co Vers" 11, ian Collrt in11-
(Wlonie t" and it a p IIarI" a t this iit' it, thre w i1 ll h a toilt inuirlig 11 eied fo(Ir
srrl'Ir' I Ir t
A!! Ir I.,Gr 1 A
r I G~ll~LPP r;
1Itt''. i I Load"t o! '0 ý I! It A , ":rb l Iit'
L~~~ rmy , Na vy v utrte ( ANQ Comnitittee P roq rants and.Publ icat.ion'; 8
3 Cut J%'.1-411 Airplane 10
Cu(ni. W ;;In-4 Ai rol nn, 12
tlovi~eini PWA Ai rplane 13
MY Gav U-~llacIram.t Loo-Loq (oordindtes and RoundedCorners 17B oeinni P- P Airt'int 19
9 ýSpeci fication X-180i3 Cla~sif ications 22
10 Stdt-utural WOW't anmd Mkrint al Safety Variationsjith rhinqo in I actot', of Siety 40
11 Lxample ui Kw'tAution Env~nve Construction 44
12 .S,,iurtural M:mi tations tarqe Cruise Vehicle 4
13 A Compr ison of Wei uit & Performance Channes withDi ffve'n~tt Ki':tot of Wa.YL and Re1liability Conceptsfor a 1-.1 1 .i r'(e [do I, e Velt ic 1 47StiAtc WHO l~rt'csnwnni c of the Factor of SafetyOn e i n Comm!e 57
15 T1wo l 'tit.'Puotu n'lity P'i ii' lt ion and
1 L o ta o! I i~ w..'l[i'?.6
17 Ovet'1''au & l tndur&Y.tr~tth heq rvertues 58
S18 J;iJ : ') o i III~ lt Q r iol Safety for a Civen'~triru! 1'tll i; Wi~ lit; Wril I Vr'as Strontth Scatter 70
19 LHIIM 001&~;, W'it ii A ni I a~r In! Sofit~y for aAtlle ai A, ut.1 1w i A ; lit y VIAl Vc'r~us Strenqth
"* 'ca! ter 71
20 . 'dltt'e' ¾t . tj 1of Mulli ' t tut' I Fevott'or flsrawttut rS
fiti
I.
I-
1.
AFFDL-TR- 78-8
SECTION I
"INTRODUCTION
From the beginning of fliqht and even before power was used the
concept of safety was considered. Wilbur Wright, in a letter to his
father, Bishop Milton Wright, on September 23, 1900, wrote the following:
"I am constructing my machine to sustain about five times my weight
and am testing every piece. I think there is no possible chance of its
breaking while in the air."
Early designers, researchers, and pilots were interested in safety
and were anxious to establish facts and information identifying maximum
loads or, various parts of the airplane. Wind tunnel measurements madebefore and after 1900 were used principally to predict airplane performance
rather than structural strength, but in-flight loads measurements to
assess strength were also made during those early days. To this day,
occupant safety is a primary concern in designing manned vehicles and
the "factor of safety" has become a prominent design concept.
The historical development of the 1.5 factor of safety in U.S.
practice is larqely unknown, even among the engineers who use the factor
frequently. Although not without criticism, little if any thought or
concern is given to usinq the 1.5 factor in day to day design applications.
This fact would seem to reflect its basic acceptance. This is not true,
however, with other structural design requirements which are frequently
challenged and modified. Desiqn specifications and practices are con-
tinuously reviewed and revised.
A concerted effo t to rationalize airplane design requirements took
place durinq the 1930's as a joint effort between the Army, Navy, and
Civil Aeronautics Administration. The d6velopment of the 1.5 factor of
safety is closely related to, and interacts with, this rationalization
effort. The term rational in this case refers to a derived rather than
an arbitrary requiremenc.
I t I 'eI I I I Ir i l I I ' t io I IN I v , I JI I I I I I I II I , 10
fi hI , t I I I t tIi ai ~ I N Iv 1 1 i t1 d i Oll I tit' I !I 10dd I I I Odh
' .1 r fo rt 'l I o ' r io' ,1 1 t W t ,. e r et.l ;I ,I 1o e 1'ei'' llt''
"a, tlt Iw o V . 1 1(i uI ,,, ! ) F • ,It'd11 dl t "l,l Oitl ', 1, I-, t•, O licel''l ,t l' llkd ,* h o lv .'1
,ll)•. ,I ,+)t' • ,t!' :• ' :til',:t•l"re I , lr tit, '1••t k, o, % 1h' Ok',! ) 't, h o' H d • •h
,I'm t t'" d !00 I t \, i - I I i' It'l ttl h llt' ht ý1,•, ',1 trm I, dui t1 o h'i d I, i -l ,' , I i,t
"t Y, \ 0 O I•'l,! ~' [J , loh f ! 'W r ll %% IV lol, 't ,,tu sit, ,. ',P l i \ ý'm , it,
oi. t'rlt ,t . ift1 II t, ' �'I -! tt t" m1• t, , t I tti \ i ~ ht I ' jl t N t t t CI )
p on det i n ,n o I i'l', rj'flit k~ l IY O ',ll k h, -I, tI ,•ld o 1 1 L"it. I'M •OIN ' t tI, N ,0it0) ,• 1
"K n w ,I 'dot , ký ",,l' , k I k ,'In Pit,I r t,',l o "klla ' IftI k ll l 1) 1' 1 ,I'ttI~ t j I 1 u0'1:101
?,1oUl(" o( r ii, UI I• II r ,,l •''' 1 do" *t •!}l • " ,I'l t "ti l t A \ I " ,t it." • t'
l llt+I, I f 0 A I o t-\ IY" "l , ] r+,l,.l "[, d1 O . 1Y • o: llI t'",,1• io 1• I' , 1 • "'1 J
1 hI I lt CII I i, I I tI I Y i i a I -iI , Ii! t I N k 011 t'0 1 1 t I II '.1 til vt ifi I I t I
a I.,", ll l 'l I! l • ' kI+I' I' a" ,', , 0. 1 1 1' 0 l ,t 0 tit' -l Cl 1|1 A i"~ 'l 0 11 0 ,I1 IV ýl I I,
,roilk I t" r, l lt !ll 1'.l''I +I t' 11.t1 %Nh' A~ ~ 'l+ , tin t k,,[ ,'.l' 'Y •i I I + , ? lt
Itil Idh t " , nrit' 1,i 1;f it' IY'', t,•' , l r't ; It I• 'h I q , if I' +l i op I I + I • 1 11 .' 01'1f
0 l ll ' +., , 1I t O 1A C PO 1*1 11 t i, ,l ! h ',It I It~ I ' fN , ýl•. ') t, I Ni tt N!t ':,
110I'I ti t I'~ Il+ II'.I ! t Ii I~.i , . Ifp %1it 1'[ i++, it Ii ! , I,| I i I '
",' t Il hO t'l %,.. At l,l 00.11' " 0 11" 110t't ,ll l jtI h 110 lt l 0 I .hj " t "t '., I ,I ' l ', t ,,
"t n Ii ov 1 . . .. .. . o ... . .. .. on.....t t
AIbi - TP '1i
'ho',te .Isoc i ated rweoriL e', thalt were found ore ,el ated to the "new"mntert;t in rational:zim i strructu;'d1 ,isfety on a reliabilitý basis. This
int r:'et heo;an in the late 105)O' and early 1906'., It is concernad
almost entirely 'qith replacing the current 1.5 ftactor of safety with
protabilistic interprotations of structural safety. (.,rtainly, today's
technol oqy can better han J e the matherat i cal and comI.u tat i ona 1 aspects
Of -, !:,ore 01o111p1 eX ', ,fety eva1 Iat ionl d d ",ay 0 Iavt( pro(mp0ted the current
i nterest.
Variations to the convent i o n factor of safety a nd probabilistic
techniques that have beer, considered aný1 used by the LISAM, as they relate
to statiL structural strength, will oI,• hf reviewed to show how they
evolved and are related to the 1.5 factor of safety.
The history of the 1.5 fa(:tor of safety has already been documented
tby two of the people actuallv involved with the formulation of design
retnuiremen ts durinq th;e 1920's anid 1030'5. Mr. A. tpsteirn worked for
the tlnited ýtates> Armv Air corps Materiel Center from 1 )?q to 1940 and
preparied the ori(ilnal Air C01','. Struc tures, SpeN ifica1tion X-l11()3 in 1036.
tie coninuoed hi,, career fn the U.S. aircraft indulst.ry workinq in the
str't nra i and criteria area until his retirement'. Mr. I . R. Shanley
worker for trhe Civil Aeronauti,'s Admiinistration in the 1030's and was
knowledtge.,able of the dev, lopmnent of civil airworthinwss, r-,ýuiretents
Another source of t ivil ,iirwort hie t eqr i reeiints, ais they relate to the
fa totr ut O ety, is a hit.ktory prepared bv the LO; Aiieeles Req ional Offi(ce
of the i iv iI Aeronaut i(., Admin, traLion. lhese hiktories are given as
Rvi'erente¾, (Militorv) and .3 (Civil), lTie hitory of the 1.5 factor of
safety (liven inl thist paper is derived almost ent i'rely from these refer-
ence"v wh i Ii are the only sýpr( if i( r ource' known t-• the antholrs.
h• i - 2 : :2 •-2 -.-- . - • I i "I
At iM T
S~t J IN '11
TI ho I fj4 k 'i i \i 1' 1 tI kitN~l I N)t %i' funkiddlenflaId adf
rtepreIt-ýit " .I Itvel of de i'm 5,1 f tt wh il ha b t'-c~iiit, ill oc~poled Qj ~ndard.
I~kii~vr ~ flp~1 ~ ti~W h~i 111'iile ht,' Ce~lt i iue~ aivil ic t tcil of the 1.
* htk~l t 'O' ti it'll! ~e l dio!P!t lov tilld ket ttend to nioi d ally majocr
L li~ sit' SW *il S¶I it'~ ik di Of hill tUif I tistC i q'nif i de c'e
kdC dill PtIrhIJ 0 P (I 1iced ill PINrs'ect i we bi 'ev iewi lit bothl t 's lii Iit an
11)ndI %I I'. d5 ý1¶\¶S -Ielll ýjcll ji it'll ~ iit,'h~itI in iudolink3 !Ile. 1
hit' !, l.)1 at 1%] P'icard ole; 1 iiiIi ti~)V lit, '~ 5emp5
1 SI- i k I
It rl 5 rar * T~ II 11) f~1 it.ltr I. I vo 1.t, d
-a I oII dIp oI 't Mv1 I. Al'l i Nl Re 1"(1tk A' k 11 .ad I.
#wSI I~ s i or '1 I 11.' I I o I I , *l (t ",\* *\ ..I~ A
ossi V.1 tt: ds '"t osJ t'I ro1Ii , t~ t1 i I oil", t* Sit
II c" t.
II ts ' t 4 A I.. 11 I.I A* s I. s t o aut 'e Ij I
I.I
,--I
A 1 '1)" 1
0 0
0 N&.0
I. ', Lr
"-0 C N(
-H -4
IT
N cn
21
0 04
if) it I
i-;
" "L- .. .• -I -1;,' 4 ,," ,, ,,
, ,• * . .C),-
Uin
fif)"-4
-'I- L)'-
r
fl. ~ *;! *
AFFbL-TR-78-8
periodic revisions and was replaced in 1938 by Part 04 of the Civil Air
Regulations, in keeping with adoption of the Civil Aeronautics Act of
1938 (Reference 3a).
The Army and Navy specified design load factors for three flight
attitudes, as shown in Fi.ure 1. Two of these, low and high incidence
(angle of attack) attitudes, were associated with dive recovery initiation,
and final recovery from a pull-up to maximum load factor. The low inci-
dence dive attitude with its aft center of pressure usually designed
the rear wiag spar. The associated design load factor, which was two-
thirds of the high incident value, was based on a lift coefficient of
one-fourth the maximum lift coefficient. This was a realistic design
concept at the time, considering the relatively limited speed range of
the airplanes and the reduced lift coeffici nt at the low incidence
design point. The range of speed from stall to maximum was sufficiently
restricted that when a high load factor maneuver was performed the
airplane would generally come close to the maximum lift coefficient.
To achieve the same maximum load factor at low incide.nce, where the lift
coefficient was one-fourth the maximum, twice the speed would be required.
Such a speed could not be achieved and thus, the reduced factr was
realistic. The third flight attitude for which design load factors were
specified was that of inverted flight.
Civil airplane design load factors were originally based upon actual
acceleration measurements during Air Corps tests in the early 1920's.
To avoid establishing categories or weight classifications for various
airplane types, the lo.id factors were made dependent on airplane gross
weight and power loading. Until 1932 load factors were given in chart
form using these two variables. These load factors were modified
slightly in 1932 for airplanes having low power loadings. The require-
ments in Civi' ,.erooautics Bulletin 7-A were revised in 1934 to include
certain basic performance and design characteristics by using empirical
equa+ions hased on previous operational practice. Although' the load
factor charts were known to neglect important airplane characteristics,
such as winq loading and drag, no substantial chan.es were made in the
maneuvering load factors themFelves. The load fac-tor charts were replaced
it "' i i• ', " " .... . ... . ..
4.ý
Q0~
4-1
V, 0 14
H A I- V)
;C) 0 In)C
H H 14 H 4.
F4 4 i H0
0 H 1-4J H ) U . L
H C) L'.1 ~ Z4.U
04 -4 1
ty) U) H L3A1 ~ IH ~ 4 V) H --~ 4 *~ 4 H 11) k1)
0~ ~~~~~ 111 i ~ C ~ ~ H ~ C14IX (A rl* L).i H *J
;! I P, n W I
.' ft. Cl4 tp 44
C4 t l C l 4 1~ ~
U~ u H U U u biU ) Ll
Al II'M -I Rt 1 -.••8
ini 1)34 by anl •'mj•irical eqluation and the minimum d";ign load iactor wa',
cldu('d frtom 4.0 to . 1h l•b 'ae , of ,,at ifact"ory experience with I1arle
1 yinq boat, ( ele renc'0 3d).
Vhe tPrm "factor of safety" was rvont'ci' ted in a general sense hal
various inUterpretat. ion were applied. Durinq nthe 1• ?('s and early 1930's,
all loads wre ult imate and airplanes were ,h,,v igned to load factors which
varied for ea,'h type Rere'eiw I definos tfactor of s.afety as the rat.io
between ult. imate loadl and ma xi mum probable load. It states', that. the
least factor of safetky used for ai rplanes is usually 1'.0 and that the
termi "factor of safety" is often used incorrectly in place of the tenn
"load factor.." A nearly identical devinit.ion for factor of safety was
in use in BIulletin 7-A in 1'129, and its use may have been the result of
Mr, Niles' influence (Referencv 1). I•n a minor sens;e , terminology was
a problem and Reference 1 also gave definit.ions for design load, normal
load, ultimate load. load factor. and margi'n of safety. Simi lar terms
were also defined in the 19134 edition of IBulletin 7-A.
While writing Air Corps Specification X-1 803 in lQ3.6, Mr. Lpstein
rnoted the aiiibiuitasnp-s of the ter';i "applied" and "design" and proposed
"limit" and "ultimate." The new terms were later adopted in a joint
meetinq of Army, Navy, and Commerce Iepartment representat ives, the
forvrunnvr of the group t.hat later oriqinatitd the Army, Navy, Commnerce
(ANC) prooramr:, whicdh are shown in 11 gutr 2. 1he accopted terms first
appeared in the lq40l chang,5 to Speciicatiion X-1,801.
The first edition of the Civil Air Reulations, issued in 193h,
incltded as imi lar terminology change. The terms 'dt•sion" and "appplied''
were replaced by "'ulti|mate" and "vield B '' oth of the new te'Wrms ref'vrrped
to load, required to he withstood by the st tructuire. The term "li'mit"
was also i utroducd to spectify the "actual"' or "expecte'd" load facttor.
The 1 ili t load fact"or represented a fl itght l im itat ion for which the
ai rplane was expected to be completely airworthy (Itefereoc" 3b).
As defined in Reference 1, Mr. Niles s',,m%' to have iqoured themaximunim maneuver load factor capabilities of the airplane in his a';ess,-
,mnt of the factor of safety of '.0. lie app. ,'ently used only the 'more
7
4,t
1'
I-I
of 11---rm
AFFDL -lIR- 18-8
typi cally achiev ed maneuver load factors when assessingn the difference
between actual and des iqn loads. There were no fligiht restrictions to
limit maneuvers, for example, to half the ultimate design load factor
(to insure a factor of safety of 2.0) and it was characteristic at the
timNe to use the more typical (or average) maneuver load factors as a
basis for des.iqn requlations. The Civil Aeronautics Administrdtion
during the s ,w period specified only a load factor value (ultimate) in
the ,rder of o.0 for a typical airplane. The implication was that the
airp<iane structure should not fail before reaching1 6.0G in flight. There
were no maneuver load limitations specified but there was the assumed
factor of safety of 2.0.
Durin~j the 1920' s, operational flight load factors began to increase.
In l1•11, Reference 4 stated that a load factor of 4.5 was sufficient for
stunt.ing based on flight tests using a JN-4H airplane (Figure 3). The
Air Corps, in fliqlht tests conducteo in 1924, recorded a load factor of
7.A in a PW-1 airplane (Figure 4) flown by Jame, Dooldittle. This factor
was the hitghest reached and occurred during a sharp pull-up at 162.5 mph
(Reference 5). the 7.•8G compared to the theoretical maximum of 8.15G
at max and a design factor of 8.5G. The 7.8 load factor certainly
could not be considered in the W Waximum probable load" category when
considering the factor of satety definition in Reference 1, but rather
as an improbably high loa'd. Similarly, the thought that airplanes had
an approximate factor of safety of 2.0 was more of an opinion which was
based on limited operational data and not on aerodynamic capability,
Pursuit airplane deviin load factors were increased to 12 when it was
reali,:,d I•hat Ihr 8.5 design load factor then in effect could be readily
In IMq',' (Re'terence 6), a Navy W6C-4 airplane (Fligure 5) develoled a
load factor of 10.06 during a pull up and in 1930 (Reference 7) a PW-9
pursuit airplane (Fiqure 6) reached acceleration%; up to 9G during flight
load test proqramsi. Both of these airplanes here desi!gned t.o ultimate
Iload factors of 11G. lowever, their were no further increases in pursuit
airpN lante dtes i gn loaOl facIt or'; as a resulLt of t Ies' •es!eper i('enes.
N'
3'
'1'
AFFDL-TR- 78-8
CL
L.
AFFhL)-T.- ,-!
The Army, Navy and CAA began during the early 1930's to rationalize
their d-osin requirements. The objective was to relate the air loads more
closely to ,r t.na1 fliqht condi tions. This effort eventually resulted in
the introductiotn of glust loading conditions, airplane speed, the velocity-
acceleration (V-G) diagram, and the use of aerodynamic derivatives. A
number of the more significant milestones in the evolution toward rational
criteriai are described in Reference ?c. The 1.5 factor of safety evolved
from this rat; oali.zation process as an outgrowth of the flight test
programs conduc ted,
The formal introduction of a factor of safety of 1.5 into Air Corps
requirements occurred in 1930 but it only applied to establishing design
tail loads. The HIAD khandbook) design loads for the hoHzontal tails
at that time were admittedly arbitrary and insufficient for the expected
service of many airplanes. Reference 8 established a new method which
consisted of determining the steady-state flight path and speed of the
airplane with zero power, assuming a complete range of angles of attack
from maximum positive to maximum negative. The balancing tail load was
then computed For each of these points. The balancing tail loads so
defined were further adjusted by a velocity factor and increased 50 percentfor desion. The 50 percent factor over the computed load was termed the
factor- of safety for material. The adoption of this design technique
also introduced the use of airplane speed and aerodynamic derivatives
(wing moment coefficients) as requirements.
The flight loads proqram reported in Reference 7 was the most
comprehensive undertaken up to that time (1930). it was conducted at
the reque(t of '.he Air Corps to determine the magnitude and distribution
of loa)ds over the winq and tail surfaces of a PW-9 pursuit airplane
during manervrrNs most likely to impose critical loads. The maneuvers
included pull-ups, roils, dives, ond inverted flight. Pressure distri-
butions and load time histories were recorded. The distributed loads
measured over the tail surfaces were two-thirds of the desigii loads and
assuming that the factor of safety of 2.0 applied, the report concluded
that the design load criteria should be increasved. The same data in
Reference 7 were'again evaluated in Reference 9. In Reference q, the
14,
-�.----,- _______ - -
¾
I �
C
.4
1"L.
0
�0
A |IIM -1 }* .': 8
Mr, Shanloev e \tuot't'd sii ,1 1t" IhouItttht in Retfet•encat't' He favored
t in t a . f "actor" of %,tt't% to keep the pvttinis ibt, limit loadi rt'tla-
tivt' lv hith , on comip red to a fa, .r Of . O. Al'houqh a requitremll',elnt
ri o1lti h lim i O.tld% to tihe atn 'e of ptriihent sort haId not vyt been
writ ten, Mr, ShalN -'ikit erpretd limit load at thi tim' as got 'exceedinq
vil d trmit tl.
A•"•mUi'iriG i'i t'd 0, Mr. Ipstt'in in Ri'•eence .'1c, the factor ot safety
val1tiue that ine considered to apply won. a variiable which depended upon the
% ell of f l !, dat a used and pee',onl,1 j udqvttnt a's to thew base to use
in asws,.'; il n the tactor. In t, io, of acttual st reingth , ther'e was no way
of knowiiwt A t'rue factor of ,dfet)'t in "vjtw of tthe limited knowledge
of load, and st t , r ,1n,1 a saly is.
:At tti% po0¶int int t a ftactt" or Mte.t)tt phll'lx1>,oph' had eSsentialtly
evolved but had not beon forinolU.li:ed. Airuplhanes were flV'pq at two-
thii rd%" of Wiit ti0 load factor, permanent sot was not des.iratle,, pe•imtis-
nibl e limit ,load, should be at hioh as ponvivle, alnd a 10. ' factor was
ali ',ads in use to est,• lih de, i n tail lo.ad%. Wt. there was no formalylV
tabli s.hedl r , t ionih ip between des i load I oaIc ct or. ma\iintill aet'rodyntamii c
Imneuver capabil it, *and operat ional maneuver i Iiimith•. Thse reot ion-
shjips would e,\,ol , with the doVelopment o f e V..e G diaqram.
lho' ctoinet of a \ -A ,tiaIram Saht I dtefineso, the det'1iq" bounda r it'es of
ain airplat, i-, ,lienoi'lli attributed to Richard Rhode. Prior to thet
adopt ,ion o' ti', oncop!, 1"to fIa.ctor of -a. tt had limited Nioniticalice.
no di aora' " *, It w I '. -.%,t't P1o ,i j '~ 1 lo i i',l' t'but it repi't eenkt a
Ioio'or ', !Itt i to .ii to i at iona; rif'tit'l'i ,. Sho No%\ l iut'eal ot
u , ! wa. I the I i•i' t" s1't'r Ist ti'. .t Iir% t i i , t .hthown in Fiqute ? , a
i 0 3.
in t he nit cll o ' f v• t at', , h ,i i t ". tdari li ' " t', , iuii d'ri o ti i remelit
with tihe' N,'t\, the' U1lr "oir, had in itiated a ,tuldv of the NaV' Neries o"
%'.1 i.'. fobund. in Refer enc, 11. •u•, ot 0? It , *'lI, ltlo ' developlme t etfort
t'OW \o dt tlor n.. thele ii' 1 t'0 i i' ui'oth 'te u i'i ! and Ii oht I nd, co'rnie ni'
ofl Wt,•' oi 1t\'l•'l'=' ýh10 "i ýl \ had 1i ,10J 11•!,! •,•cl'l~lt'< 'lt',(•1
A Vi It .R-" "
authhors asumed Chat the taxillum opyera1tional loadd% we, rie t he specified
ultimate dt, i on load% divided bv a ., factor of stafety, 1he tmetaured
tall loads act'uall\ were about twO-third% i 'tt' the des• in values and
hel •ed to s•i t anti ate Cthe 10. as',umptlOl, .t1 thoaah it was a weak
assumption in a statistical sense. It was their belief that the require-
ments were based on "an ant ci pated load fact or p1u, a meargi n of SO per-
cent to allow for poInilo impaiect ion'. of materiall, alpprO\ximations of
analysis and teneral Wlck of kMOWlMdqe of loads.' In References 7 and Q,
then, we s.ee tw• concllu1ion, baved on the same data., fach conc.l usion
occurred by as•uminlll a different tacttor of safty.
Althouilh some disao,' r't,mentk t ll pan sib l a.'nfusion seemed to exist,
it CoaWd have been wore. proumablv, if the 2.0 factor had been con.-
sidered the norm, the IN2 PW,- WPM ' 1 .tlle shoul .lOt even have been
permitted to exceed A. Itf the hi vhat W1t load factor recorded by the
PW'..) airplane hid been considered and comrpatred to the I,2G design load
factor, an en', loweer 1.33 factor' of safet , could have been assumed when
evaluatinq the fliotht data in Referen•co Now consider a'gain the
facter of safety of .'.0. it" it wal sWill the accepted norm as assu.mled
in Reference 7, then the '"new" conclusion in Reference 0 , that a 10
factor of safetv pre•,aHVtJ] during MiOMht operations•, was not adequately
supported. lhe ,Assiumed need tfr a factor of safety vof 1.0 in 19?5 was
,juSt a5 valid a', a,,uml lU the l. fa,'tor exs 1ted in fliiht operat ions) inl
l31 . Neither value coold be ,ktw,,. tl •,ldta ,ti st icallV. Clearlv, a
chan.lae in thi nfk ing• had 1 occurred as a roAit ot operat ioanal p•Factice andt
available fl iglht test dtat.
Mr. I 'sto•il nlw ite 11 '" , !'ttt' .. that his thinlkifltl tol towed thik.
pattetrn. In thFCe eat, 14 '04' ', a t acor otf 0 wa'. considet red nlecessary
an implied in Re,,renc'.i I b%, N I'.. in the late 1.'0"0, actual opera-
tiOonall flx Il'o 01 thIe n at' irpla a l'n, ht''", %t'l' cami njq c loser to the ultimate
load tactor than earlier Folds.' Airplane% were flying up to two-thirds
and lmloe otf the a1ltlmate l oadh faC•tor ,Fnd nhothinlo was happen i ng to the
tastructuire; theretore, the evaoluti of at thi uk inli towadt 'a. , owt'r taw fctor of
safety W Wa, natural onv1'.
i i 1 '
AllL
A ir' C o r'I rtiqu i rellen t oI t J I ord t ho ;:II I r wut'l [)IId t*ic t o r t) oc cr at 1
ma i\ i luitl I i i t ctit,"li Ic 1 ellt arIld it WI' ý' re t ort. re otrciiendtd t fillt t he Iia II -,
ard ol il " e lkýit ,I1 '1 itidh L d t on' i .It t iic k. t he I OW rI. left
cornerl~' W,' 0400d t heit Filitt "I'MCt i 011 tt Ot h' tlt'(ii~t' tVli \iIii ft coeffi cienit
l ine 'rInd th lit, i'lat i vt c" ie ttil oi od tiac t oIii'. ho t A i I Corp I al I t reconutild-ld
tha t t ho ulppe, -r i'i till tidid ai. 1*11'l lit, I I' i q It anli to prey ,.z I do fin jt i ve
lo 011V e (IIk '0t' Itt t ic k k r Li t",Io l i 111
10 C 0 Uiit I,' t tit, it roulj'Irit I ht dMI. Ip I dI' 101,IW Ot'rItol' id it t i c f 11 Ct 0 1' Y in1
soerv i ce w it h reduced~ 101i od at, I-, f I, tr I hi I low i nic i dem ce( aIrniekl ot' at tackA
IorrIoe , Mi.. p "t v II io t od t hat 'I t Tv . . aria 1ý . is wasý i ' vilot IlY corisrva t i ye
A ri l th ti' 1t r-ýIwt Urc wasý ICt Ui I1 ' t 'enter' t han aria I v,,' ir.,i d i (,ted .Fu rt he I'
tIV kid iice it o j ui' " i a 1, i of:t aitii I ed liviler r' 'i ih li k~'011'11' WaN PIOVi dtod
ovill !a a in Rt~feli encet I . whorr1 a P' 1 'I' i, I kiinrI i IT] rplirie WIS Ohowl t L
l Ill~e ll Id ed 'I Ii II : uiP ftlttT ; 3 Vi00t icd I ki \ive ýtIart il rio lat .'50 'Ili 1 o'ý 1`10'
hou r. A r'jI rw i jijt..p d 0 1 ."1! "Ii e\11 P1'r hItou' lv~ts reachedi~ and 1 111,1011i1iu1ii
13 k I t At or' tit' wa . Joy'Vt Ioped a It l ow arii: 1 Lit' at t ack Ia t .'8 Ill Os
pt 'r hour'. Wh 1' %,,1 an i a d I ih rtml l. ! I till from I lit, 1''Id\ "li! \peltlld
Ka i vliii "ete, wclrw 'I rt "Ire'kdtl\ ItI ned I)1 re Jt- i tl ;,! 111'Wo' a1 1; i t her,
tl~ tit, t'I 'i I. ro, 1t in a Vl't" I CA ,l d .i' wh ti k 1w, . a p:1 I i d'ca to r Pit l'-
Wt'd 'li I Il 0a y1; l'e L'iJ I &1 1I,011 he l'vr 11) tit 0Wt liI'Woht to r dc i t.o ther.
lqlO.i IT t'I , I 't I o it d a Iq ý p 1 1 rut' ta I t t 1 1 . , ti I wI t I ,I it't .II \ic' I s O~i N
I I. lit .. . . . .1, IIL, ' It ItTA 1 1 Ii , W i 1 1
J ArFDL-TR-78-8
+1.
lot .10 lt( oo104114 4H1 I 1!-MilNavv~~~~ ~ ~ ~ ~ *-,Daiai (qtokodnt, n one onrRet t-row
.........
Al F1)1 -1 R- -IS1 -
ult imate design load factor would have to be raised to 15.75." This high
faCltor ws felt to be unwarranted by virtue of the accompanyinq weight
inc cease and loss ill performance. Therefore, the princillle of not
de, IgOinq to the maxi mum recorded load factors was recognized and the
maximum load factor" boundary for" the V-G'diagram was established by the
load factors then in use.
Having agreed on the boundary of the V-G diagram, the question of
how to use the diagram in conjunction with the factor of safety as a
design and operational boundary remained to be resolved. The Army and
Navy differed in opinion as to whether the V-G diagram should be an
ultimate or a limit diagram. The Navy V-G diagram represented an elastic
1imi t requirement. Yet the limit was difficult to define because the
Navy did not have a fixed factor of safety. Appendix II of Navy Specifi-
cation SS-1 (Reference 10) states that the ratios of ultimate strength
to elastic limit strength should be equal to or greater than 1.35 for
all conditions except the (live, in which case th:e factor is to be a
mini mum of 1 .5. Wino cells were to be . designed to any point within the
V-G diagram without exceeding the elastic limit of any structural
member and the horizoltal tail had to sustain the: maximum balancing load
mu 1 tip1 ied by 1.5 without permanent, set., and by 2.0 without failure.
The Navy also had ai flgliht V-G diagIram and a reqioirement that the flight
loads should not. exceed tle elastic limit. of the structure. Appendix II
suIggested a factor of 1.05 or 1.10 as the flight elastic true (yield)
fact or of safety.
lhe Air Corps study in Reference 11 recommended the adoption of a
sinllIe tfIctor of safe!v as a preferable alternative to the variety
SleL'iit'ild by the Navy. Since an Air" Corps precedent. for' usi01g a 1.5
fact or of safety had already been establ ished, it was recommended that
the 1L.5 fdctor he adopted for the V-Gi diag.iram. However, the Air Corps
also recommended that. the VWG diagram he an ultimate rather than a yi'eld
di agram as the Navy had been utsinrg it The Air Corps proposed flight
limits were to he two-thirds of the ultimate factors shown by the
di ag0ram,
,-0
AFFDL.-TR-78-8
C6
L.L
AF.FI- .R-l78-
STRESS ANALYSIS CRIT|ERIA
BAS!C; FLIGHT CRITERIA
WINGS AN) WING BItRACING
A . Ill'l' N ( (FAR
CONTROL, SURHACFES, INCLUDI.I)1NG lXI) FXDIRFACES, AND AUXI1LIARY DEVICES
CONTROl. SYSIE',S
ENEINE MOMNTS AND NACELLES
FIISELAG;E ANI) HUL.L
F I TT' I N(GS
Figure 9. Specification X-1803 Classifications(Reference 2c).
The Aeronautics Bulletin No. 7-A retained the pi'eviously described
factor of safety philosophy and the implied value of 2.0 until 1933. The
1934 revision involved a correlhtion of loading conditions with actual
fl ight, conditions and it was necessary to redefine the design or "ultimate"
load factor. An "expected" or "actual" load factor 'as defined in con-
junction with the ultimate, load factor and a factor of safety. The
ultimate load flcter was divided by the factor of safety to obtain an"HI'lied" load fa-tot which was not. allowed to catuse any permanent struc-
tu.'al deformation. Th(, actual strengtth renLuiremeT1t in the 1934 issue of
Bulletin 7-A slat( that. "The mlinIium factor of safety for any aircraft
Structr tirleo CoHnptIO1'nt tlheiefore shall he 1 .0 muless otherwi*.e spe i fied.
This rtguiires that fhe nltit muo st| t'nqtjh of any member shall be at least
l,!hO t ies, as grealt ',t its .c. it ical applied load" (R,,ference 3h).
Ovelr the, year,,, s;omke writer", have at tributt'd the orinjin of the
1.5 fattor of salety to the charactt'ni stics of the newer ;W04 alui.iintm
(2'4S] at that time. !t had a ratio of ultimote to yield stress of
approximlately 1 . 5. Act, ial 1 y, the i'eredeont of a 1 . ! fart or of safety
(for des igol tail loads) had ailready been established when the Air Corps
forliv adopted it tor overamll •1-,rctlural dt-"ii . Mr 1p100ei1 has
St ateti i! 1ete lcel Nt th'it i:i,itvrI il prop'rtit's were riot ant Air" Corps,
AFFDL-TR-78-8
This design philosophy was incorporated into Revision 6 of the HIAD,
dated March 1934 and the 1.5 factor of safety became a formal Air Corps
requirement. The use of an ultimate V-G diagram was later changed by
the Air Corps to a design load factor diagram in August, 1936, when
Specification X-1803 was issued. The original recommendation of an
ultimate diagram was based on potential structural problems due to
secondary wing bending effects in biplanes. By 1936 the rapid trend
toward unbraced monoplanes replacing both biplanes and braced monoplanes
in Air, Corps procurement caused the original concern to disappear.
The original concern that led to the ultimate V-G diagram recom-
mendation was related to the secondary nonlinear bending load effects
with load factors found in the wing spars of biplanes and braced mono-
planes. An explanation of this concern is given in Reference 1 and
relates to a DH-4 wing test as reported in McCook Field Serial Report
2391. The report concluded that wing spars should have sufficient
lateral bending strength to prevent a tendency to twist under some wing-loading conditions. To prevent lateral spar failure, the strength
requirement for the internal wing drag truss was increased 33 percent.
The Navy engineers disagreed with those of the Army regarding the need
for the additional factors and did not adopt them.
When Specification X-1803 was issued, this and additional drag truss
p. design requirements were given as part of the Wings and Wing Bracing
classification (Figure 9). To insure tors;onal rigidity, the design
requirements for internal wing truss designs were increased as d function
of the wing type and ranged from a factor of 1.33 to 3.0. These factors
were used in addition to the 1.5 factor of safety. However. as previously
noted, the 1936 change from an ultimate to a limit V-G diagram wassolely the result of new airplane design trends. The change did facili-
tate the use of one diagra for both design and operation, but the
desirabiliLy of a single diagram was already apparent when Reference 11was prepared. Consequently, the 1934 Ai4 Corps recommendation to use an
ultimate V-G diagram was only reluctantly proposed because of the
potential structural bending problems in biplane wings,
21
AHI)L - TR- 78-8.
coincidental association with material properties) in that its adoption
in conjunction with the use of the V-G diagram recognized "the principle
of all airplaane being l imited operationally to a fli iht envelope within
which it would not. experience any significant permanent set."
He further notes in Reference 2c that, "the factor of safety of 1.5
has withstood many moves to alter it, but. there was a period in 1939
when the Chief of the Structures'Branch of the Engineering Division at
Wright Field thought seriously of reducing the value of the Factor.
Newer aluminum alloys were becoming available with higher ratios of yield
to ultimate strength and he interpreted the factor as the ratio of
ultimate to yield. However, no action was taken when the following
explanation was offered: 'The factor of safety is not a ratio of ultimate
to yield strength, but i% tied in with the many uncertainties in airplane
design, such as fatique, inaccuracies in stress analysis, and variations
of material gagles from nominal values. It might also be considered to
provide an additional margin of strength for an airplane subjected to
shel 1 fi re.
Finally, Mr. Epstein notes in Reference la that, "In subsequent
years there have been various assessments made as to the significance of
the 1.5 factor of safety but actuall.V its ori'gil (in USAF requirements)
was an opinion of what was represntative of service flig1ht operations"
(relative to design load fac'totr;).
The 1.5 factor of safety remains today in an intermittent state of
assessment. Its use in U.S. airplane design practice has never been
formally desigtnated as a fixed design en t.i t.y or as a single design
entity, althouqh it. is often viewed in t.hat. sense. Other important
structural desiq•n factorrs affect safety but. they arre nornally viewed in
a less rigid fashion. Each factor that has evolved is applied in- a
specific way. The 1.5 factor applies to the basic external ground and
flight loads while other supplemental factors apply, for example, to
pressurized cabins, castintis, and fittink1s. The size of the safety
factors selected usually depend on the design application, manu fact.uring
24
A FFA 'r- - 78 - H
c on s i dlera tLiont. I f they had be'n , 1 7sf, which Was the. al1umi nunl allony
usedl at the( t imv, would h(mav dic~tatd a factor of safety of 1 .7, since
it had an ultimate value of' 55,0 ~ps i and a yield value of 32,000 ps~i.
There were also opin ions; its noted by Mr. Shaniley in Reference 3a,
which attempted ton r'elate airplane operation and permanent set of thle
structuire, or it lack of it, to the selection of the 1.5 factor of safety.
This is often cited, as the basic reason for, thle choice of 1.5 rather than
some other number. Thle approximate 1.5 ultimate to yield stress rat~tos
of cormionly ursed materials and the( apparent lack of permanent defonnation
appeared to mean, that airplanes had niot been developinq more than about
two-thirds of their design load factor in flight. Mr. Shanley points
out that fie was not convinced that the permanent set philosophy was a
"sound argument, for at least two reasons: (1) It did not apply to
compression membersý that failed by buckl ing , and (2) tension members
were almost always critical at. jon- for which the efficiency was
generally below 80 percent." As preoviously cited, Mr. Shanley' s main
reason, for Favorinq a 1.5 factor was to allow limit loads to remain
relativel1y high (as opposed to the as soued factor of safety of 2.0).
Sinrce he also i nt erpreted 1 imi F. load ais o requinirement to preclude
*per'manvO set do rii n no rmea operationl; lihe concluded that thie only
si gni ficance to bt, p1 aced onl the two- Fbi ed: rat in wai that. it imposed
no penial ty onl existing a irplaries when worki-ntl backward from exist inrg
load factors, us inq at factor of safety of 1 .S.
From the poi nt. of view of the Air For-e, thle history of the 1 .5
fa~ctor nf safety can best he suniniirized by several of Mr. Epstein's
observations;. lit Iefererct, 'it, Mr. fpste in noted that the dec.is ion by
thle AirtTorp', to stipul a t a I . ! fart or for' subsequen01t dei go in conjunc-
tion with the( V-G( diagram wal supporteid by, and niot, thle result of, thle
fact that the% ?4ST alumiruuim alloy materia-l then conning into rise had at
1.5 ratio ti'tween ul tim&tv tensi lt and Yield st rength. lie felt that the
adopt ion of the 1 .5 tac tor of safety waFs much moen sitginifi (ant (than a~ny
;,3
AFFDL -TR-- 78-8
SECfION III
DEVELOPMENT OF OTHER FACTORS OF SAFETY
As seen in the previous section, the 1.5 factor of safety did notevolve as, the result of a concerted effort to derive a useful factor.
It evolved toqjether with other design requirements as part of an overalldesire to rationalize struictural design criteria. Other commonly used
factors 'of safety also evolved in a similar but more direct fashion than
did the 1.5 factor of safety for airplanes.
The history of the 1.25 factor of safety for missiles as a require-Plent is relatively complete. The factor evolved as part of an overall
effort by the Aoity, Navy. and Air force t~o develop strength and rigidityrequirements, for missiles. The derivat ion of the actual valute of 1 .25
which finially evolved is not as well defined, however, at least withregard to Writi ht Field record,-. The value seems to have evolved through
a phi losophical1 trial and error process.- Missile strength and rigidity
req icemnt ,inc Iudi nq the ]..'IN fact or of safety, were forinal ly publishedby the Air forice andl the. Navyý in Spec if icat ion M1 I_- M-8856 (ASG) , which
was dated 2V dune 14959,
Missilt, des iqn req[)i vemenlt s were act ively pursued by both the AirCorps and the Navyv but theyv were prece~ded byv those of the pi lotle~ss
a i r, 1~ n.e - in 1945, t he Air Corps comp iled requiremnents fov such vehicles.11ue (loutiment , "St ress Analysi s Criiteria for Winioed Missiles," did, ineltecl appl.\ 'o p ilet less ai rplant-1 and was, derived from Speci ficat ion
C 1, 1.(13 . lh titec i H fictai toin retaineod the airiplant, fctor of safetyot I .Itit' Nav v wroto I a wiert'ra1 spec ifi cation for the De 'I o nd
Colltruct ionl lit' i lc levis Aircraft , in April, 1940. This documient was
referred to " he Aerofmllti cal St andamrd,, Group (ASG) bY the Navy inl
Doemberh~, 101140. A let ter from the( ASG to the Chief of Ordinance
Ventako)11. * reo1111endhed corintinof the specifikcat ionl withP all
brankclet- ot t Phe ,try ice-s.
--- -- ' - ,. -
AFFDL-TR-78-11
standards, and the intended operational use of the airplane which existed
at the time they were adopted. Since circumstances change, a review of
the origin of each factor is always of interest and worthwhilc. Perhaps
this review will help place the significance and future applicability of
the 1.5 factor of safety in proper perspective.
The remaining sections will discuss the history of other well
known factors of safety and variations to the factor of safety design
concept from an Air Force perspective.
At i ill, I Rý
COMt' hd MA t~ aIot f il I I Oil) and ai I I aI,-tI v I V- WOT'' tt'rmiltt'd i 1
Kikh iItI* I it I lift' i'm r I n i .-CtiI Lc'vit, kc rd i;mt' ki ll vi C i t it's Wet'lt
JCo C ht, Ndl 0 f *\ctanau I c t ti t' Ot M thi'll'd i :A t it'll)
C011' " i fu? I ritcid tO ITi Pmjdt t "11 "t' 'i nal " draf m ad sowt'dx
it11 ;'' I IOntl s c 0 1 td I ll no , unt a Ptnd OMNI I' I +)} dii tit , fat:.' Cd' to"
o l .t c• It, ' re I I', •t a I',ld 'o I ln t I J 0 ?.T I)'- t't 0 I'm I 11 ,ha id b ll ' kit •it'd
I lt',1,% 1 0 (Ih I ti , Vl INJ Itd k oi, ci t C t' U V . I l i. ' " fl 0 Uan 1 • 41' i " , c t C t,' i Ct r'I
It I t,, I I, l'.f 1•j . t3 1i I ttc o .r . e I ' J 111d fndt, uO l t i ,In t 0 t kitt
cit ho I C t, c 0$1 Cdi k t0t" itj I. i l' c~t1 ae td j ld v C P I li, 3 t~i % S 01t, k h't %I
s notlld t, a I, tio f1 a % I I o I'f0, o 'I o0 .lPclit , I . I A' 0V. t W1 It S .i l
ft'lua Nk vi) Iv to ld dI e ,si t ac ' " C i't 01, I! ik l lt .I v 11.1"t ii0t' 1, Ca " ilti S -i It'
JTpp Iit Ion.ll tu he lNi r ,N. as p ,t'v Ialu I s I tio' lit I . vaI ut• Was a
,l "i -,oO ,' ta i" c,1,,,ed !ti, prts," ,v kdite'1I.tltis .' l't,( l'i" 11,0 01
o'spt'k. It tor ;jny t st '11 P ' t'litI i M t i0 ' t ato l t 111to rea So., ' I'
" t' iill Cr•.l'd • , INt w :'ii S I ' C t o, li. I tll , ' i. il k' vi.' dk- i~ tcd 'l I jltc, i -
pa' I). i's rc~sW Oute c it dcralIil- an l 't' h los' f ha C work, flot.
Al i 'ti ',tn.! h d,, ; e, ::t'ev'. 'ITý' I , 1011at t'd !W 110k e iiler 1 ' iIhlo
Ftl i s o J I cl I, I f Ct oI,' o toi 'la ,cC l C a .t' Ia t•,'1 0 '" l'CtISt 'S ? T iil I t, j .'C' .
tilt ittltt -t r ,",\\c\ I I V , s 1% ,,ITN t' i ll appaklt tiit'cn , ,h 11"nn i tn' s Nd 11- i
on I 1woI.,l t!i1 tC jilt1 a, orI Ca l itre' I Iit'd I tait 1 i' dI lild
II I r. Pll# aw In,*' fl I I t n Cll! n't ll ht, 10 -:1 dixra t x .1 l it I hId
putil I Oled " I ft'I' ,I t 1a C a ti j 'ta 110d CMeO .... " t, I I,, t ot 1r %hIll k, u'- .I 1) h
Al CPtoi'' f lM11 -M A "ll i pn .", i tit, k n C11 f I f tIf rm a '\ an , 11r 1 oi' t', NaIl,I t'll 1a Arwn. In Iiittl•, • a i, Ct , di d i kep t
St.'~~~~~~~inClatTlcd ',li ;lcv il~ lt Ot t. dl'i.' ii't ail' fll~•il i \ •;ll l I1icOl;~ t''•| '
11'rta1 ( n 0T t od o 4o it'l T N 011 I t ' :' 11MO O Ca.f *a C C ~ ' i'.11
bu . tit' ! tiC t' h , I 'll 0 I" t I no 01 t' I o 0 1l II k, t 0n ti•Cci. imoa.l C It , t lit,,c
tol'l t•I g' kl ! I I n "Odi:, O ll \ 0 M ,l • o• a~'l•t' l ( tf c' tlt•J~''t,
N J I I lit,t t 111 t ' i t'O Wk I o 1a111) N I d I f t• l . O - t, Vt!' ,i 01-11 'J ittI tt. 11, s;Y " S iltIk, A -
at0- , klV" iq ro'm, nt)I s st I 'u l'•Ctl C ltid t ''"a11 ''•l:ti",I Cl oll 0 1a t1 I o i, l ajlnd 0l&d i tiN
c oind i t i v.1)S IMVI ) h . 111n 1n tC tp t a 11 ) prlxlaxt I it f ma I I 't'v colis I dt'r'd . .
I k' , )'atst In %I I, lI t at't, r i t\ a 'i 1'.." * o 'ra 2I lit, 1 1t r: I .:I t t' l Vi ,.l#aI qu\
tI i poflita t O ti f ov it, t , A,,t' kit .1 11 di t o- aj ' I tii. ua A S u Jkar1,rI ai~ i kur dt
ta- a k, ill'." iI'. C'i~. It Coa ol0'. . k,',i' !,t, doi C, e ode vd
AVI Pt - I R- 7,1-8
In I952', a t ri -service Gu i ded Mi ss il e lask Group was formed by the
Office of Standardizat ion, Defense Supply Mana;ement Aqency. The task
uroup was composed of tivye subccomikiittees, one of which was to prepare
structural criteoria and data requirements. This'suhconnittee first met
in January, 1Y53. The Navy wrote the first draft of a guided missile
strength and riqidity specification for the Criteria Subconmiittee in
February, 1453. This draft specificatkion req'ired a 1.15 factor on
yield strenqth, a 1.5 ultimate factor for loading conditions hazardous
to personnel or to the launch airplane, and a 1.0 ultimate factor for
all other loading conditions.
The Air Force, prior to this time. had been using a variety of
ultimate safety factors including 1 X, 1.15, 1.25, 1.30 for its winged
and balistic missiles. Occasionallv, more than one factor was used on
the same missile as occurred on the Matador. The ultimate factor changed
from 1.15 to 1.25 between early ,lesiqns and the "B" model. At the time
the above Navy draft ;pecification was written (14531 the Air Force had
already informallv established the 1.25 ultimrrate factaor of qafety as a
standard value for missileq. This difference in Air Force and Navy factor
of safety philosophy became very evidnt by the third meeting of the
Criteria Subcommittee anti thu factor of Wafetv became the most contro-
vWsi',1l issue to be resolved.
ilhe miat ter was said to be "resolved." as roted in the next draft
specification, by deletino the use of any specific factor of safety and
all owinq each user of the spec ification to insert their own valte. This
approach to the fact'or of safety di sagreemrient appeared in what was trnmed
the "final draft" of the specification in Olle, 1o5e . However. the
actual finali:ation of the specificattion took considerabl y longer.
The An,,v !had initial ly Ic ipated with the Air Force and Navy
during the firstt few meetinrig of the cormrritiee b•tt did not attend after
October. 1453; 1 he Army Ordinanc' Corpst felt that t, hetate-of-the-art
did not warrant the i ssutanoe of a peocitication at that time. They did,
however , subm it commenermts on later draft% when they were citrctulated fr
coordinati on. l ho oilrcomt tee's final dliaft did not circtulate for formal
AFFDL -TR-78-8
stress. Because of the uncertainty related to the design of large,
integral pressure vessels in combination with flight loads and tempera-
* tures, a 1.1 factor on yield was selected in combination with a 1.4
r.factor on ultimate. The study conclusions emphasized that high safety
factors and high reliability are not necessarily equivalent, nor do
they negate the problems of inadequate design practice or analysis,
ineffective quality control, or prevent brittle material failures. The
yield and ultimate factors were to apply to all combined aerodynamic,
inertia, pressure, and thrust loads for both the solid and liquid pro-
pellent boosters then being considered. The liquid booster propellent
tanks, however, when subject only to pressure loads, were to be designed
to a 1.25 ultimate factor because the internal pressures were considered
more predictable than those in a solid propellent booster.
The 1.4 ultimate factor of safety, as initially developed, was
intended for a specific vehicle desiqn, the X-NO booster. It was not
intended for broader application: or to he used without the 1.1 factor on
yield stress. Its use establlished a precedent, however, and it has
since been quoted in many publ icat ions. Presumedly, the two factors are
still considered appli~able to current designs, since they have appeared
in both Air Fo•rce and NASA desiton requirenlllents, for mani t'd SpdiL' vebV ik ICo.
Currentl,. both m•lssile and space vehicle ttructural design phi-
1oaophi, re factor of safetv orient,,d. However, the overall desion
reluirenlents for these velhicles, are more clomely related to I reliability
based criterina than are Current airplatne dellt- N,, In tho next secti m,
other ba.,it coniepts which relate to hoth airpllante ,t'1d missiles will he
reviewed, Tht.,e ,onkhopts will inc lode modifi c•tion,, to the conventijonal
f0t tors of 1, ot v. and erta in rt 1 iatl i lit v ha,,od c riter ia i nteracc t iots.
3 U
A FFPL -T-W- 718.-
from the miet hod found in NA(A TN' 43Y32 ( Reference 13) and I definition of
the atmocpher e in power spect ral dens ity form was taken from NACAReportý 1272 ( Roference 14). The prrohab ili stic gus;t requl rentents replaced
the origl I n discrette JIuS t requ i r~en tens found in early dra fts of
MI L--M-8t3$6. The earl icr discrete qust requ-irettents were pdatterned after
aIrplane, requ irelivnt-s and were vi qorously objected to by the aircraft
indimu ry. The, industry also emphasized the. need for a comillion atmospheric
description for- the design of the structure and the control -system.
Probabili sticall1y defi ned wind and gust descriptions were i ocl1uded in the
flinl speci fication.
In all, kiln extended period of time, was required to deloelop, co-
ord-inate, and i ssuto the 0n ji na M11 M4N (ASG) missile spec ifi cat ion,
but manlY of the roklu i remoits ISWire nlew and niever. be~fore formial 1
coord iniate'd betweenl the SNyvicets.
A related s ide 1light is the, development of tho 1.A fator of safetyv
which is usomd for manned NOac "e cls i act or oniklinated wi thi ii
the Aircraft lalorit orv. Wrikilt Air' Deveopoment Ceterlol ItWAPC Theit
facltor wa s tt df ined bY * he samit o f f i ce respons; I'bie for allI other Air
Forceo a irp Il .te and Ini ; s iI t, c r t e i a (RefteronlCe I t I lihe I .4 fac tor 'I rew
out o f a I aho ra tor'. studY t o evalIuate t theit a~pp i cabi I it o of th fi,1 . 4
fclCt or o1' sa S0t'( '' v fI ,I Io eart I I, b o o It I es fo1' Imane I spIaMCe. v ehIi cles.
Ie I)' Ir Ie t 1\-1.1' I Itr~' trnsIrover, 1b 1,e ,'eent 1'.ý 1) 1t IM. w s underi1
(itvleo'llopnt a1 t t fil t i mek and t it bsosN t or sN stt ow., wore,' an int eor1i 1 IPArIt of
lio deve I opmen t . IThe manniled oli tid, neonl t rl''. l k.'o Ic t, Wa' be Iiti 11eits i 9orred
bl. Iho I . " k~a It oi- o f ,,Ifet, t bu1(t the 0'.t dIt ol p it abi l it', Ito t tit boo st Ir
was onellts i ioned . I h tab11orat " or'S Util %~ OW, i toned Itheit usual ttS i on
L tillI 1.11ct ionl, andllII' IlrutfI,rt trino iFit I I-i t I t i on rII IIId inIlterat, ins ti" b)ut Ithe,
mnate1, 1l-11 i er't 'Pt ,I*tI; prollvidJed Ithe ma'lý or'It w i kllt tio ( fa t or'. Ik he u I ti Ia t
Ito v ie I d I tl, nes raIt io f0 th1w' tliI otIlld I la t' Kna soa bIdý, ,Noost1 1 lat o I' maeiaIs
WOe-Vlt' 0r~r v i den ia L' I 1 o mi0 IIJ III v rut ur I ,I a ftot'I and fi er
hlosoi 01 o hr rit) I I hr I. ( it '1n work i nor o I tt ''. 1' r to f dt -.' kion ' , 1 d
AFFDL-TR-78-8
their influence on current design practice. These variations to current
factor of safety design concepts tend to blend with reliability-based
concepts. One effectively leads to the other because some of the
variations are an attempt to rationalize criteria and are intended to be
a step toward a reliability-based criteria.
The operational environment has become increasingly hostile and more
and more demands are made of the airframe. To improve airplane performance
and accurately appraise structural design requirements, established
criteria must be ,antinuously updated and supported by adequate technology.
The technical support must include analytical techniques for determining
aerodynamic derivatives, vehicle dynimics. heat transfer, stress-
temperature distributions, material properties after prior random
exposures, a suitable means of qualifying a structure to actual or
reasonably representative environments, rind a satisfactory means of
flight test demonwtration. Regardless of the design conceit, the
parameters that ai ect the structue and its response must be further
explored. A realistic appraisal of our current ability to guarantee
the design of a relMabWe structure and t(: define the steps required to
r obt~ain a reasonable assurance of structural reliability is also required
(Reference 16).
Statisti.s have formed a basic part of airplane criteria from the
time that sufficient data were available to .judge the reasonableness of
values usud for maneuver load factor and diqn gust (Reference 16):
Material properties. sink sneeds, and other parameters used for esti-
mat irit fdtique life are also derived ,.tatistiially. However, all current
requirements foi .tatic strength call for a specific factor of safety ta
he apnl ied i, maximum expected loads even thoJqih ,ome of th' design
paraumeters and rlsaltinq loads were statistically derived.
For about two decades, whith spun the 1,150's and 1960's, two
additional design concepts, s"afe--life" and fail-safe," have been used
in combination with the factur of safety to desiqn Militarv airpldnes
for the fatiquc or repeated load envirinnimebt. C'hief emphasis has been
placed on the facto r of safety and safe-life a pproac.hes. 1he fail-safe
;AFFDL-TR-78-8
SECTION IV
FACTOR OF SAFETY DESIGN CONCEPTS
The desire to rationalize design criteria has never ceased. Structural
design requirements (military specifications, for example) are in a
constant process of review and revision. Because of its encompassing
influence on structural design, the factor of safety has received
considerable attention in recent years. This attention, though, relates
primarily to the factor of safety as a design concept. The specific
value of the factor of safety is usually a secondary consideration. The
actual value does not readily equate to a specific level of safety and it
is difficult to judge the difference in safety as related by a change in
factor from 1.5 to 1.4, for ixample. Similarly, the factor of safety
concept does not readily equate to an identifiable design objective that
provides or defines structural integrity. Integrity is achieved through
many interacting design facets, some of which are obvious and some
abstract. The concept primarily provides a safe operating margin between
an operational and design level of strength, Just how "safe' this margin
makes the airplane is always open to question because of numerous
unknowns and parameter variations which affect structural loads, design,
analysis, materials, operation, and the natural environment. Because of
these unknowns and variatinns, the a.tual degree of structural opti-
mization achieved is also quetion,•bt.. The apparent high degree of
structural integrity athieved by the factor of safety concept is often
the result of indirect, intuitive considerations, and reactions to
previous probles . Design and operational experience has essentially
provided the basis for the acceptability of current requirements and
the safety provided by the factor of safety concept. To overcome this
apparent lack u! precision , definition and ohi e'.tivity, and improved
design flxitnility, the use of probabilistic techniques and reliability-
* based desiirO criteria are often proposed.
"Havving• reviewed the hstory of several well known factors of safety
for airplane and mis,,ile de,;QP in ec tion 1I and III, we can now review
a number of Air Forrte ,tudie% of variat ions to these factor, and note
i 'i i -- '-•-- ~
AFFDL-TR-78-8
concept is added to a design when high reliability is required (as in
transport airplanes) or when performance penalties are not incurred (as
in fighter ai:rplanes). The safe-life concept attempts to identify,
through analysis and tV-st, the fatigue ;:ritical areas of the airframe
and offset problems which might occur w'ithin the specified lifetime of
the airplane. This presents a conflict between the nonprobabilistic
factor of safety concept and the probabilistic concept of safe service
life. Faced with an increasingly complex operating environment and a
demand for more reliable (Economical) airframes, the designer has
attempted to make the most of each concept. The factor of safety, a
static strength parameter, will not provide for time varying] effects,
and the •afe-life concept suffers from a lack of appropriate operationa!
and structural component test data. Therefore, the task of designing a
reliable structure has been to incorporate analysis methods which
combine the useful functions of each concept, (Reference 17).
More recently the term of "safe-life" has become obsnlcte and the
terms "damage tolerance" and "durability" have been introduced The
design intent to provide2 structures that are safe and economical to
maintain has not changed but the approach is different. The currett
Air Force design philosophy emphasizes both I le damage resistance 'r
tolerance to manufacturing or service induced flaws for some specified
period of service usage and the economica ma ;Lenance of the airframe.
The term damage tolerance is not new, but the emphasis on assumed
initial or service induced flaws in the airframe is relatively new. The
damage tolerance concept is intende,& to minimnze catastrophic structural
failures due to LJie propagation of undetected flaws in critical ln,.'tions.To contain the damage, fail-safe and slow crack growth design concepts
are used. The fail-safe concept contains local damage by use of multiple
load paths and tear stoppers. The slow crack growth concept protects
safety by not permitting flaws to grow through unstable rapid propagation,
This is done through inspections, or life I ihiiting in the case uf
noninspei table structure.
33
h
AFFDL-TR-78-8
The du,'ability requirements emphasize low maintenance costs during
the life of the airframe. The durability concept is intended to minimize
airframe maintenance due to cracks and related structural degradation.
The safe-life concept used a factor of 4.0 in predicting fatigue life
and was a combined deterministic/probabilistic concept. In essence, the
damage tolerance concept (which does not use a directly applied factor)
can be considered as deterministic as the safe-life concept because the
stipulated initial flaws are in fact, factors of safety on time.
The most argued "advantage" for reducing the factor of safety is
the reduction in airplane weight and the accompanyi1,g increase in
performance. An impressive discussion in favor of reducing the factor of
safety to save weight is provided in Reference 18, which was written in
1954. lhe discussion coný.iders permanent set, allowance ,or defects in
material and workmar.ship, stiffne,;w, ano meneuver load exceedances.
Proper accounting of these points during design is shown to support.a
decrease in airplane weight. The arguments seem factual and are still
current. Some facets can be updated to trdays design philosophies and
technology to further support the contentions given. The advantages to
reducing the factor of safety are shown by decreases in gross weight as
a function of factor size and proportiorate increases in performance for
representative military airplanes. Of special interest is the note that
airplanes frequently exceed design limit load factors and that suchfactors nay require an incre-se, rather than continuing to count on the
1.5 factor of safety to cover such occurrences. The projecled control
of limit load factor exceedances by the '-e of entirely automatic flight
control systems has not materializec foir piloted airplanes but is quite
comnon for missiles and space craft.
Reference 18 also p)oin.ts out the erroneous idea that the 1.5 factor
of safety always provides an actual operational level of strength
50 percent above limit load factor. Structural design procedures assume
a I inoa," loid increase betwepn ' imi t and ultimate load when the 1 .5
factor- of safety is tised. Due to aero(dynamic nonlinearities, some
parts of the dirplane reach loading conditions g,'eater or less than
34
A
Al t *I~ III I 'R
I. , If lt It,,1 it, I liltu l itI tI h eioitton Ih ' a i tp I, iut, It i tý' It I V iPM lf l ,1 .",0 '.
illt CI', I * ," o,! h i'dttl i ' i hi ' l .I I ,t (it", i.1i'lt ll 1d I 't lll I r i ',IIv t o ii - i' lll'ot"11•l ' tY t I -,, I ! 'I'f•t r' till r'villolt ý ll -pc c'.1 't(• I •' ' p111 11'11t J, 1 i, i 11'erom'I'l
hill t co th , , ,," IhokllI w;, , I.~ ) f ilt I , n' I f I c !it o lp ropr ; I v .1 oe' l•,ll i ' l h , t t, tit
I t'll I vi itu it I,,, it l of tell It h le 'ldor I It.It1in 0 li lt) il!'• ' dlfit, i 'llf. I "I
Im II , l oll 'd .] ,1 iI, f1",1111f, h( l. Jq il. '1h| nt'q i 'I ,o a T11" 1 M I N,l V (Inlknown, - r', I'• .l i tlll 1h11
l it r I 'Itw ,i. , oI t ' I fI t,' wi' ht I i kill , W iJitill l', t v I I iir't r
I II i i• t 'l a v.; ,l (i' j f li•, ' h " ll ) t ;, I ) i Ii I Y tlit''.i. V 1
Itill ti i I11 i k 1 d Ilk0 01 . ro lukl I ' t h 'ih ,I i 1'1 1',.1111v lyt, tllh t r',1l' t1 kll', tillI v I 'll c "rr 11t
d ivtrl~ l it oll v ki"11• ] I pll" orlll],: 1h '1, 1 I ll t"i11 o•po -,!!~ ,1 i•:|: ,n 0 i I it
I, I i '•t l" l I !tI' ll, Ii l {,11 , I. Ilkf fuI I o k•., ' w v Il. d,\',10011W•lll t'l .\', Ih(,
InI i~ll I Iev tI uc , f '~ f r11 %1 ,i tll 1 1 ,11 \ I I ,,111.i ho iI 11 "I t'~ , illq "V il, veli
tit ilpll ', Im I i ollai I ollk, i 1 11ll 1 ('l11t , d ilk,111 41 t, ll ~t t 0 1 tl . l i~'l t , till,, i kin t'iqu i lrtillvlt 1 ,
tll!'t hiT C0111 I ] c,l I v If• lit , l il," I.1 i till. W qf It l 0ý 1 dl, i 1 t ill ? I ',"t I fi OWl, <, h I i b
1% t, t I Ii.hod Nx,l whlvil lit",, iq (1 11 o\i biI i ( i,. I hiolhi, f (, I I, I V il 1 d0%', tll
YM'l I i II~la do.,i n ,litaln IIlt , I -a',I i olla I -t',,iqt, da t I a I'l, IIII'mllo I v lit] wo,
AM ' Il no,! v 1' , l v il ,i t o , kin i. Wo ii klllt r t II,ki li. oll tit•]'ll1, tl ql ('11 k011011 111
-llt1 ' l lit , I" t I I,0 1Clh t l 11'. tI oo hol vih , i \ ,,I t l ' •l r d lh'iill tl • lv ,I ý a rf ,llild
'IVV h ,ll1 Id m i ~it a ] iv I llak I\" I l ihIt; mat~, ki • hoI e,, prow'im i le l 1, 11 i Illpor'I a It
1,111.1 M' %,, 0 A' tlll til ,i nll',hitl , •i ow .,. A I l tn h I• , ho ii,,l'ik ini, twIoi tilt ( l
i O 'I 'C , I ' I •, r d % Il ,a ir I m' .vai 'llo o h i l . f | I ', ,,tl" I , 1 \ 1 i ,k f ual I \ I ' i I i , i til
h ie I liv(l't I -v ,I ] ',I i Il ' I, lw " a•I vh I •' 1,% ý11 ll~qt I tit- Ill i li j j ill I Ih it I i ,-, v .1 V 11
oI I !,I, [I fl r l Illl, Ii l II I f(1 ,!l i , V ,I ldt I"t, Ii I t1 i, [, li \ II•t do, l lt ! .1111 1 It (li, . Iii.1rl\
Ihim n ti till I li I ftt ed whollI ",d, d h'I. o I r I Ih. .t 1" t, IIll(, wt',lhf '1 11 .,•\' l klol-,
I I~ I tl ".c ,,t'1 1\ l ,t~it, ,d .l l 't,'I I ,11hY. I 1 I •,'a, t I I I a a ,|• I., a l lt, kill1 I it
jlo','.i 1l1v Ilk' pu t tt tit, f f or u-01,• 11\" %l~l•~ lh 1h 11 u' i ilA l \ ,i Ih ,
,a r r~n, I r l it llt t r I I |mlle \ v'w I I I lh lU, I a,!I, o ,. 1\ai~ l o e a
,0% t l, I e)FA1 v iV( it~ill Ibm I ( '.•i 1.11t 1 )10" Pit',l • 111,.
AFFDL.-TR- 78-8
Any change in structural weight is fdrther reflected as an even
larger change in gross weight. This relationship is described by a
weight growth factor which is the ratio of the delta change in gross
weight for each one mound change in structural weight. A certain range
or value of weight growth factor can be used to describe an airplane
category (transport, fighter, or bomber) but a specific factor must be
calculated for each airplane to be accurate. Although airframe weight
trends have not varied significantly in recent years, weight growth
factors have been decreasing and the overall sensitivity of airplane
performance and operating costs versus structural weight have also been
decreasing. The reasons for the change in sensiLivity are related to
technological improvements. These improvements include the use of more
efficient materials and construction techniques, greater aerodynamic andpropulsion efficiencies, and higher internal packaging densities.
Conversely, the structural weight trends (as described in Reference 19)
show an insensitivity to higher strength to weight materials and related
structural improvements. Apparently, increases in structue'al efficiency
are offset by the imposition of more severe design and operational
riqui rements.
rhe influence of current structural design conceots and packaging
density can be illustrated by reviewIng the wing content and structure
of d current fighter airpldne. The installed wing structure weighs
about 1800 pounds. The primary wing bending strength is derived from
the upper and lower wing box skins which weight about 735 pounds. A
factor of safety change would have the largest impact on the wing skins
which comprise about 40 percent of the total installed wing structural
weight. A large percentage of the total wing weight is composed of the
flaps, actuators, and seals. These and other miscellaneous components
would not be greatly affected by a change in the structural factor of
safety.
Recent examples of the factor of safety's influence on airframe
weight ha'e resulted from an unofficial Air Force design philosophy for
experimental or prototype vehicles. The unofficial philosophy modifies
36
Arrol. -Ta- mwilK ~ thlt nortra 1,1 a irp Iarml development cycle Which incl 1udes a S-en es of groundand fl ight tests to Vai idate ai rframet i nt~egr ity. linitial flight tests
MTe oorii1t 11 V res tri cted to 80 perceirt of ecvrt ain dosign mraneuver load
levels, 'ujnti qw gounld sta tic tests are completed (o ul tima te load levels.
Flight tet. a irpl anes are normall1y instrumented and critical load pointsare cl osely monlitored.* Flight mnd ground test. loads are. then correlat~ed
and (round telst~s are repea ted, if' meces'sary. to further validate the
Structure for the actual fltiiht loads before the ai rplane is rcleased tofl 1i t 100 percent of designi 1li1m1t load. Such testi-ng is complicated,
expo'nsiVt', amnd timeit conls wmi mg. Alt hougli j ust ifiable for an airplanesysteml, AhN) S tructO.I1 efficie~nc'y ints t be opt imi zedJ, experimental and
pro to type Vehiciles canrwlt beV as ri jornus ly tes ted because of cost anrdtime cons txains. To insure etulm fliigh t stifety at 100 percent of designlimit Iload, without, the vxtt'ns ivt test.rigq and associated delays, the
fol lowing) procedure has beenl established:
a1 rTh exp iiaI/ r otyearfrm~ne or rmnd ifi cat.ions to existing
a irframeits are desitoned usJi rn a I.,805 fact-or of safety on loads, which is
equialvil et tlo a theore t.ical mla rg in of, saIft~y of 10. ?5. Tile initial810 pe rcen'rt t Ii ght res t riction rino ri ly impoed (in an ai rpl are sy stemn is
aIlso t'guiul va 1en to a *0.;! .l5 rlttlil or 01 a0tety.
b. Tht pe - iiina1/poovrai-rtcame k' s tressý irstrumolnte( at
cr'it.i cal des i go po in t.s anld proot tested "w ;1 the rounld to 110 percent
ot, deLIIS im1111t lord. to itmrr'ro deN ig(11/11411111 tIW 1ing tit rintegri tv The
installed iris t rumenoltat iton is turtiler man i orod ir;t- fi ght. anld compared to
the protttt load roltsiIsa; as,. furlther ;,I1*ttV check.
It' tfIii ý desO~ i oIWOC(dure ik utr'd, theaw c'01,11,rmt is a Iiwoed to fly art
lulO per't ert Of dest'OIn limlit 10oa1 d c)Aphil1il vWithout IrIIu -1t:*1rrute, load
groruld les:t arid wit hout, reduc ing overa 11 ;artety. ActunalIly , the 1 .875
tadte ld 110 percen1t lWOrtto 101 otl t'st prov id(I"! 1 argr' il] tilurratell firrit
latit, (I . / ) t holr the' (onvent inor~ rat~t it I ( . !)' , h Vr-ef0rI, teVSt ing to
110 percet o r I imIIIi t l oadI, isýI re likely j to (Mu~e detrim(il'ltl yielding)
of thlt a i i-I amle 1h,1n corvenlt loin 1 tt' t i rig to 1100 ITVlterIt'n o~lt 10iri0 od.
AFFD)L-TR-78-8
For the experimental/prototype vehicle, the philosophy imposes no real
penalty because of its one-of-a-kind nature and flexible mission status.
Two examples of this philosophy will be discussed in the following
paragraphs.
The first example is a prototype Mach 2 fighter airplane designI which
emphasized exceptional maneuverability and typical mission objectives.
This prototype design was to be built and used as a technologyI ad-ancement demonstrator; the design was completed but manufacturing
plans were cancelled. The technology objectives have been rechanneled
to an existing airplane which will be modified instead. During the
design of the proposE demonstrator, a dual airframe comparison was made
using the 1 .5 and 1.875 factors. The comparison evolved as follows:
Tsa. The design requirelent specified the use of a 1.875 factor for
flight loads. Computerized design techniques and a highly detailed
finitexelement structurae iodel were used. The available design/analysis
flexibility allowed the weight of the airplane and airframe to be
established sepiarately fo both the 1.5 and 1.875 factors.'
b. NT o weigent comparisons were then established: (1) The airplane
gross takeoff' weight using the 1.5 factor was 26,465 pounds. Using the
1.875 factor it weighed 27,056 pjands, orun increase of 2.2 percent.
(2) The structural weight using the 1.5 factor was 5,095 pounds. For the
1.875 factor, it was 5,433 pounds, or an increase of 6.2 percent. These
weights reflect a design service life of 12,000 hours (a service life of
3,000 hours and a scatter factor of 4.0).
The weight growth factor was clculated to be 1.75 (poundF gross
weight iqurease for each pound of !.ructural weight), which is a
reasonable value for a fighter airplane.
The second example is the YF-16 prototype airplane. !he 1 .875
factor was api lied to the Fliqht 1(; ds and incrreased the structural
weiqht by 6.6 pet ent wheri -ompared to the 1.5 factor. The weight
_3
. -....
AFFOL-TR-78-8
increase has not been detrimental to its overall performance and would
be further minimized if the 1,000 hour design service life of the
prototype airplane were increased. The weight increment caused by the
larger factor to safety would be partly absorbed by the weight increase
required to meet the service life requirements.
Using the same design aalugy, a comparison of the weight increase
required for different service lives can be seen in another design study
of a fighter technology demonstrator of the same weight class. For a
1.875 factor of safety and a scatter factor. of 4.0, the airframe delta
weight increased about 25 pounds per 1,000 hours of service life.Because of the 1.875 factor, no additional weight was required to achieve
the first 1,500 hours. Damage tolerance requirements were not applied
during this study but they would have further influenced the airframe
weight, increasing it to some degree. Similarly, the 1.875 factor would
have lessened the weight sensitivity of the airframe to these requirements.
If, instead of increasing the margin of safety by 25 percent, it
were decreased by the same amount, a similar airframe delta weight could
be expected for the tecinolog) demonstrator. As noted in Figure 10,
which is based on the first example, the factor of safety equivalent to
the 25 percent margin of safety reduction is 1.125. The use of this"small" factor, when compared to the 1.50 nominal factor, would probably
not be considered by a designer even if a large reduction in airframe
weight were desired. The airframe weight reduction shown (6.2 percent)may be optimistic because the normal damage tolerance/fatigue liferequirements are not incorporated. A 1.25 factor of safety is perhaps
a more reas,)nable value to choose and is shown for comparison. It wouldprovide an approximate 4 percent weight saving and a 16.7 percen-
reduction in margin of safety. These percentage weight changes and
margins of safety are reasorable and reflect current jet fighter design
technology trends. Similar data can be found for other airplane typesin References 183 and 20.
39
AFrDL -TR- 78-8
1-44,
1-44
(4
t-4
".44
o 0
41
.1--4
t40
AFFDL-TR-78-8
Reference 18 reflects technology of the 1950's but the trends are
still applicable. Data are given for fighter, bomber, and transport
airplanes and factors of safety between 1.0 and 1.5. The average
structural weight saving shown for the three airplane types using a
factor of safety of 1.4 would be 2.5 (±0.25) percent as compared to a
1.5 factor. If a 1.25 factor of safety were compared to a 1.5 factor,
the average structural weight saving would be S.0 (+0.75) percent.
Although these structural weight trends are still current, the range and
gross weight trends in Reference 18 do not represent today's airplanes
as well. lhe range increases shown for the lower structural weights
assumed a weight growth factor of seven for all three airplane types and
additional fuel was substituted for the gross take-off weight saved.Using these assumptions, jet fighters are shown, for example, to yield a
15 percent range increase and jet bombers a 5 percent range increase or
an average of 10 percent for both airplanes. These range increases are
based on a gross take-off weight and fuel adjustment that averages
15 (±3.5) percent when a 1.25 factor of safety is used. These gross
weight and fuel adjustments would average 6 (±1.25) percent when a
1.40 factor of safety is used.
The weight growth factor of seven and the substitution of fuel for
gross weight saved will give optimistic gross weight decreases and range
increases today since weight growth factors are less. Factors in past
years for airplanes have ranged from about five to ten. The weight
growth factor for a fighter today would be about two instead of seven
and for long range airplanes like bombers and transports a value of five
would be appropriate. To illustrate, the trend in Reference 18 for a
fighter and a weight growth factor of 7 gives a gross weight savings of
17 percent for a factor of safety of 1.25. The fighter technology
demonstrator study gives a gross weight savings of 2.4 percent for a
factor of safety of 1.25 and a growth factor of 1.75. For the same
factors, Reference 18 gives a gross weight savings of 13 percent for a
bomber as compared to 6 percent in Reference 20, which used a growth
factor of 5.50. These are only trends, however, and they are debatable,
since weight estimating and the establishment of weight growth factors
41
m m •............ . ....
AFFDL-TR-78-8
is a sensitive art. To establish accurate values, the factors must be
based or. a detailed evaluation of a specific airplane and its basic
mi" SS ilolS.
Reference 20 presents a simplified but constructive parametric study
of the factor of safety. Three different vehicle types were selected
and designed using a range of ultimate factors of.safety. For comparison,
three other structural design concepts were included: the modified factor
of safety concept defined 'in Reference 21, Part I; a yield factor of
safety concept; and a reliability based concept.
The conventional factcr of safety was applied to the limit design
loads of each vehicle in increments between 1.0 and 2.0. Ihe modified
factor of safety concept applied a factor of 1.05 to speed, 1.15 tomaneuvorabilit", and 1.10 to design loads. A yield philosophy applied
factors of 1.0 and 1.10 to design limit loads.
The demonstration incluaed cruise, ballistic, and glide reentry type
vehicles. Although hypothetical mission profiles and performance figures
were used they were patterned after real vehicles and designed to
realistic structural requirements. The structural concepts used were
monocoque, semimonocoque, truss, pressure stabilized, and sandwich
honeycomb. Two structural concepts were applied to each vehicle, as
appropriate to the vehicle type, and radiating and ablative thermalprotection systems were applied separately to the reentry vehicle.
The conventional factor of safety I)hilosophy normally considers
only one factor to establish ultimate design 1'.;ads, without regarding
the variables contributing to the limit loads. The yield designphilosophy uses a single factor of safety to prevent the design limit
stresses fim exceeding the material yield strLss; an ultimlate load
factor is not used. The modified factor of safety philosophy also
applies a fdctor to the design limit loads bi;t these leads are first
established by factoring two performance parameters, as previously
noted
42
AFFDL-TR-78-8
A reliability based design philosophy normally considers the
statistical distribution and the probability of the combined occurrence
of a finite number of structural design factors in order to establish
design loads. In Reference 20, the various factor of safety design
philosophies were also compared to a structural reliability value. Thereliability value was established by considering the statistical dis-
tributions and the simultaneous occurrence of two statistically variant
design factors.
The parametric study considered both rigid and flexible structure
and aerodynamic heating effects when applicable. Weight, weight dis-
tribution, and stiffness characteristics were determined for each major
component. These parameter variations were correlated to each vehicle's
performance and structural reliability.
The scope of the investigation can be considered limited in that
the structural concepts used were simplified, the analysis methods were
not elaborate, and only one critical design point was selected for each
vehicle. However, structural weights were optimized, the effects of
plasticity were accounted for, and interaction equations were used to
account for local and general instabilities caused by combined loading.
The vehicle design is similar to a B-52 bomber in performance, size,
weight, and structural flexibility. Gust and maneuver were the criticaldesign conditions. These two parameters formed the loading interaction
curves used to establish an equivalent reliability based design, as
shown in Figure 11. By using available gust and maneuver statistics.
the riost probable combination of the two design parameters that could
cause structurai failure at each strength level were located on the
interaction envelopes, as shown in Figure 12. Although not an optimum
design, considering the complexity of other reliability bdsed concepts,
the most probable failure point is used to illustrate that a lighterstructure can be achieved by a reliability based concept as compared to
a factor of safety concept, even though both provide the same
(theoretical) reliability.
43
AFFDL-TR- 78-8
4114N/.
- _ - . . -..... -;... .. ..
CO) 4J
/ 0
4 w J
LO 0
'4"4
I-I I
_ _ _ or-,
- .... I L --
Cn
/ 0'
C)C
0 0 0~C 00 0•0r,:
44
L Il i I III .. "• . " . .. . . ........ -0
AFFDL-TR- 78-8
0 0
0. 0.
1 -4 W~
.c *
r-4 -40L
G~eG)
- - 4 0.'- cILI III.
L) 9-
aoL
-------------. -
AFFDL -TR- 78-8
The weight reduction for the equally reliable ýesign is achieved by
bringing the strength of the individual components closer togethe'.
This is similar to using different factors of safety to contour the
shape of the interaction curves of over-strength comp'~nents to fit the
shape of the composite failure boundary of the vehicle. This takes
advantage of the fact emphasized in Reference 1F, that all components do
not reach 150 percent of limit load simultaneouIsly. Part of the study
results f~r the cruise vehicle are shown in Figure 13.
The range increases shown in Figure 13 are the result of decreases
in the inert (empty) weight of the vehicle. The ratio of inert weight
change to structural weight change is about 2.0 and the growth factor
between structural weight change and gross take-offr weight change is
about 5.5. 'the decrease in inert weight duo to the structural weight
change, for example, when using a 1.25 factor of safety is about 11percent, which increases the range 1.6 percent. The dramatic perfor-mance increases, oftenl seen in such coinpari sons is not seen here because
of the nfore realistic value of the weight growth facto'r. The substi-
tution of fuel for either the inert or, gross weight saved, 3lthough afuel substitution for inert weight saved is shown, is, not considered a
realistic desiqn trade.' The performance of a specific airplane design
is nriorilly Opt imnized using available ful volume; any addi tional
Structural wrighlt saived as a result of a concerted weight reducti'onpreg ram would normoi 11y enhanrce the performance based on the or igiinal1
fuel vfl umve, Oi dependently (if additionml toe I
rati(Ilit lift, 111, flutter wti e not de~ iq goonsiderations in
Refterten A) mO ao Iheit' hi nera ti Oiln al'e 0nknown. SimilaIrly, the proba -
tli*Jit" iS hOWn in F iguroI 13 aMe based 01n1Y On the interaction of guSt.aid lmaneuver, I oad', and IO no0t cimis i tTe OVtO hem t,*Ot' tssLhh as iiia o't~~i a
proport it", and workman11li p t h'it would ASO ~O OL'tt the St rl tura I1
l~t Imms t t t dteh t V) t i k~ An et C It imi V0 1h,1 rit ht, t) go if
AFFDL.-TR- 78-8
w3Ii 0 a, rn 4 -d 0 co N'20 C C 0 0 0 C': 0
w 0 0 H 4 -4 ý4 H- -4 4 Q)
r- -4 0- Ht r4 C, C%& 0 4 HN HONH -i
w3 0
~d 0 -4
iI e
-H4
A- r
In 0 0' 0' A' In M' 0' 0
0 ..
44
0000.
'44
4. C 1N H O C 0 V' a ' a' C o C
C3~@ ON-4 c-41
011
~~i~ (A '
4.. * -4f
-44
I'4 ( _4N
H ce 0 0 * 0 * 0 9-m -4 401 *-1 4)j &.JUI., Cr (0 %'- r .Q * t
(4JA - '. 4 U. 44U. ~ a. -0 Or. 00
l4-. U I . 'L T * 0' * 0
Afim IIll:-1ll1..8
i 1. iL fi ti'.'( nut t urd 111o (iodels Asý a tje'r~i I t rend, t he relI i ai i i tybased desin I q incopt prov ided ll l ower Oilrt. t nra 1 weitihI anld hettevrporlformanllce t han di~d hel fac'tor of ;,Ifvt v conicept iav Il an0 do qua 1rel -A, 1i ty .The mod ified fatl.wIL or "I Vet y conicept kqave reso 1Ist, closerto the reel i a 1 it I basyhllýed coocep(11t 11s i il~l t wo pa rame t vi' hIn. 1.1 d id theS1 nq 11e1 ~t, ,or of sdle1,t y w; ed t o obt t ýii nI he w 111 ma t Ie des i tin I oil ds
I'heonet, i c() It l . a h1i qho'r coil tf I ;ent e c.I1 he p)1 a ced In 11' s truc turedes i tjnetd by a re' 1 bi Ib ii .vC(1Wcept t hI by a1 tIactor .1f Safety conceilpt.bo~caus Oe the * atua 11 el V I Iit olteIacteo a eh i n is tever
k nown The re Ia i1) i I i t ha bw~d concep t by coollt ra .; t ,cois 1 don.; theS t~at s .i S i 1lni I uire t)f Ohe des . (I no pa cavil ec is a nd t bus acIO l esa known
re ,b 1t n h isclated coifideuc", level of the( stat.i t~ica1 datause~ed Ill a prac t ic,11;ns hk1,.!-ver t his is; not true beausew thedes1, i til pa rame ters; tc'( lo t 4o 1, delfi nevd a 11 thevc di' Id I eli ab iIi ty o f thelsIructuhm e cann11ot beilo, hI(It I)nt a teod
T he mod i f 1 ed tf. It t o ot . a o telfvty c oip as I c l i-o I -po ra tevd inI the1c onpa ra1.i ye f c to r (If' s a Ite 1 shindy iln Re fe re n' e 0 , i s doc ulilO .et ill
Re ference il , Pal' I . Thi, cont ept wa ý deve 1oped as anl initerim des itlnme1t hod to lie 11m.4d nlit i I a I'm ,qm nomb1111er of pmlaran'tvers cool d bel detinledand a formal rel labil1I y haile conlcept e'; tabl I set. Inl Roeoterece 21.forty-onev dos Ill p llodtdmet.~ MTe described a,; s iqnli icanit to a eel labilIi tyNased S t ru t ura I des iqo11 c( "celt .IFi Itteen are relate Ud t o t he des i (1n
ef~y I t-onhlitlett anld 6wratI i ml colid i t i ow; 1 Ie hel rIIted .itni 1s's is parallel ersIareV: .Me'Odyllianii. I olces pro oll 5 itO A 111 ti'h ol*es Mid 1111s10;MT *
material. Illopwiif ; 'Id111. lI)Ivrtv Vl rd 1,a t 11ow;5I t henna 1 st ress's 1" cr. C)(Olltt
molatm-'cls,. weitiqht anidt wekllqh di't iblb ion . coimi;oilnt Inkial i(Illments"
tempera )lit're tl, arjll ka 11111 Syrs (1)thion. o IdaI .I io k*iili tn e rosio I C' it t
and avr (rmvllt ot "k ithe htif illledttb of atmt vi can11e alpi jed~a to any
vellic IV s tivqni bot it wall oritlil na I V dovelope'ui toe missle desivOIl
app1l ikt tion. 1 he w51' tit three pwam"11ofer., '.e''iieu fesible Within thlit
AFFDL-TR- 78-8
current state-of-the-ait and the concept is onsidered to be more
rational than the ulti'tvt factor of saf2ty concept because it rec-
ognizes parameters other than load as significantly affecting structural
integrity and reliability. Although the structural reliability of a
mifssile design using the 1.25 factor of safety could not be given a
numerical value, the reliability seemed highi, and the three factors
selected (1.05 on speed, 1.10 on loads or "quality," and 1.15 on
maneuverability) were chosen to give i combined strength effect similar
to that provided by the 1.25 factor. The 1.05 factor on -meed is
significant because, for some missiles, small increases iv speed resultio rapid degradation in structur.l capacity due to material degrao:tion
with rising temperature. Thu,, with the initial ,esign ba:-,d on a
higher speed (and hence higher temperature) the du:signer is i'orced to
avoid the use of materials which are unduly sensitive to temoeatures.
The 1.10 "quality" factor is applied to the structual loads incurred at
specified maximum design conditions rather than .3t conventionaly definedlimit conditions. This concept more rationally accounts for the
noril in'a'-i ties in aeroelastic and aerodynamic data that frequently occur
betwceen limit and ultimate (design) conditions. The 1.15 factor is
applied to "he naneuver load facWor to protect the vehicle from
inadvertent maneuver exceedances.
The modified factor of .afety concept was used tG design the ASSFT
glide reentry vehicles., ASSWT is an acronym for Aerotnern'odynamic/
elastic Structural Systems E~nvironmental Tests. The ASSIl program
coisistLed of designing and flying a series oi winged gIlide reentry
veh~ic:les whnich were boosted into suborbita! fliqht pa t.hs. la,,h wehiicl,
was designed to explore an area of glide reeitry technolo(ly that. could
not be defi ned in existin g ,;round test facilities. Structural flight
test. results and dat. correlations which substantiate the adequacy of
the design concept, at ieast for vehiclIs having a progranll'jd ttijectory,
is given in Reference 22.
For an ai'rframe th,,t , "ot ierudvnam.a ] hwa el1, the oad. factor
parameter is utsIll Suff-r t: I, to ,wnvvy its over,,l stOrelgith cipabi1ity.
Withthe ntoducticui of tras jent. str~esses and r~educed mat terial
pr~oper ties due to th010,1 Vrdd jents andi hig~h tmpervatures, there is no
Simuple method for' Convey i ng strenyith capability (Re fer~ence 23) . Al thoughI therm11al factor' of Stafety is inot flcC(ssari ly it true factor, of safety in
the conventionalI sense it r~epresents a des ig~n concept. that is intended topr-ovidec an equivalent measur~e of conlsevdi t ism and conf idence in theheated a ;rfirame as that pr~ovided by the conventional factor of safety
i n thle cold air-frame.
To date, unofficial Air For-ce pol icy has been to avoid using anyfac tot* on temperatture, heat transfer', or (in Unimo at, tempei-a turo, for-a-plIanes.Ti a been a revasonlable policy sin-:(, ser-vice experience
ha s dmons mrti that, thereo is little problem in keepim niWithin a%; eed-alti tude des iqn envelope. Thus', in a sense, the. temper'atoresnorm 11 I l used fork des itn rert n th av Iu 1 aiu ob ecte
for' thle S tru tore. This philosophy of not. ftc tor-i nq temper~ature i s anla ppa erent contraddiction when compar~ed to) tho phi losophy of factor~i ng
l imi t l oad, which is a lso consider~ed thei maximum to be expected(Refterence 16).
The fact or.. of saflet~y used in convent ional air-frame desigin toocorpor t e ose rvM .i sr also apply t~o Icvrod1vtMllikca11ly hea ted a ilrfr-aies
but the, convent. ona 1 fact ors dto vtint account. I'or the ,ddi t i otml uncertainties
associated with & evated tempetratur-es. Th. adc it i ena conserva t. isin
't rel1ate direoctly to, the uncer~tainlties assoiciated With) the preýdi ctioi.,Of ý011tlll-ura tempera I't LTirS and thra retua1des itin anal ys is.However, t he tact that an a i trtame v'iy hawe been built. to res ist amwodynamic heat ni noeffects does not necessarily imply an incr~ease in conl-vent ional desiqll and manl factur'i nldeit) ec is Initially, the, use (.freplatively un Ini"ili'l am mlal'teials a110 f'ol.w oit' koit. I'lt: ct ion ill the !wa ted
*a ilrft-mli and the need folr cOmpjUt inn( stt';cturl etipe t ure ISWitht onlya limfited a1OU11t, Of f Ii lh t e! ufrm il t 0,1ion * does tenld to' itt'Cicraso theli keihj ood of' delit i 1:1i it!:; In t ilno hrnwevel.. *home onts idera t Iotis
improve' and thit mud it ica t i o of touvelit iona 1s Ornc tut rIsat ety tat.ofursis not wI ert e1td (Ret~v)(. erene ?'
AFFDL-TR-78-8
As part of an overall Air Force effort to establish a rational
design criterion for aerodynamin'ally heated airframes, the velocity
factor has evolved as a simple .nd expedient way of providing andcontrolling thermal -structural design conservatism throughout the
design cycle. The details of tis concept are developed in Reference 24.
The techniques investigated included factoring structural loads, temper-
ature or heat flux, angle of attack, velocity, and atmospheric density.
It concluded that only factorc in velocity or structural temperature
(or heat flux) are likely to pr.)vide an adequate margin when considering
the exceedances of performance variables. Of the two choices, the factor
on velocity is considered the mire logical. The velocity factor provides
conservatism in a uniform way at each point in the design mission as a
function of Mach number and the selected size of the velocity factor
controls the imposed conservatism.
A thermal factor, of safety on loads would introduce an arbitrary
(unknown) conservatism; however, a direct, rather than a presumed,
margin of safety would exist at the operational level if margins were
placed on performance variables instead. Factoring a performance
parameter (speed) provides conservatism in a way parallel to that
achieved by factoring loads and is preferred to factoring temperature
or heat flux since perfornrance nargiios introduced over operational levels
are more evident and controllable.
By requiring a design speed beyond limit, the velocity factor, is
in a sense, a factor on heat transfer. However, to specify a direct
and specific factor on heat. transfer would be a design weakness, in that
a number of analytical techniques may exist for a particular area and
fligqht regime, with a large spread inl the values thvy provide. From a
str'uctural point of view, the factor of safety should be associated with
a particular analytical method. For ablation, varied factors Of safety
have been'used (generally to faCtor the thickness of the ablator), with
different considerations bei no given to each pa'ticulatr flight application
and for applications of the material as an insulator. A factor may
also I)e used it repre ,ent a conihined factor on heat transfer and on
........ ...... 5iR WWW W W WI
AI FD.-I R- 7 B-6
the scatter of the ablative material cl.aracteristics. Factors of safety
relating to tnermal-structural applications can only have meaning when
rel aI d to the trajectori :,s or flight paths for which they are needed.
A factor of safety on a nominal trajectory may be appreciably higher
than those corresponding to design trajectories based on a broad
parametric investigation or one which places a margin on altitude which
represents a factor of safety on temperature (Reference 23).
No single technique can be expected to provide conservatism in arational way for all contingencies. At best, one type of factor will
come closest to providing the desired conservatism and this has been
true of the velocity factor. Its use seems reasonable in view of
existing factor of safety precedents. A specific design requirement
for aerodynamically heated airframes is beiny formulated and tested by
design application. The basic cri teria is conventional but it is
modified to incorporate the velocity factor concept and attempts to
account for maty of the design and analysis variables which affectthermal-structural design. The criteria have not been finalized and are
based primarily on References 16, 23, and 24, which provide insight to
time related load and temperature interactions and the selection of
critical thermal-structural design points.
Most of the studies and des i tn ccncepts reviewed in tbis section
have evorvod as an attempt to further rationalize structural desigqn
criteria. As previously noted, these variations to the current factor
of safety design concept tend to blend with reliability based concepts.
The next section will discuss certain relianiiity based concepts
investigated by thie Air rorce. related des i t, parameters:, and data
collection programs.
A?
AFFDL-TR- 78-8
SECTION V
RELIABILITY BASED DESIGN CONCEPTS
The number of publications treating reliability based design increased
appreciably in the early 1950's. Initially, reliability design seemed
to imply, at ledst within the Air, Force, the eliwtination of tile factor
of safety as a design concept. Statistically defined design parameters
and parameter interactions were to be substituted for the discrete
parameters and the factor of safety. Too often, however, the magnitude
of the problem was overlooked and the acquisition of essential elements
were overly simplified.
The Air Force initiated the development of a reliability based
structural design criteria for missiles in the mid 1950's. Missiles
were one shot devices and unmanned. There was little to lose. Airplanes
in turn, were to become mere missile launching platforms that would not
require strength for rigorous design maneuvers. Combat was to be con-ducted remotely. Durin( this .sirl of revised coimnitment to systems
development, the Air Force established a program to conduct a series of
investigations that were to encompass and define the total life cycle of
a missile. Considerable prior;ity was given to this effort. The
rationalized criteria were to be based on information obtained through
data gathering programs and operational experiences. The broad goal
was to establish a reliability based structural design criteria to
replace the factor of safety for missiles. This rationalized criteria
was to be placed in the new structural strength and rigidity specifi-
cation MIL-M-8k856, initially dated 22 June 1959. However, the data base
never material ized, the 1.25 factor of safety is still used, and thr
elmiha' is to develop a reliability based criteria for missiles has
d i i n i shed.
When the need for a reliability based criteria is 'onsidered, a
comparison is usually made to exi stinq des itn requirements and tileneed is often questioned: "Why are reliability hased critera reIqu i red
whell current iequi relllentlls And industry practices have produced airplanes
5 3
AFFDL-TR-78-8
having a (seemingly) high reliability?" The answer that evolves must
in some way relate to realism in design. A reliability based design
concept, as stressed by some in the literature, is inevitable and is
the only means of providing greater realism; the only way of rationalizing
the factor of safety concept.
The Air Force first deviated in a significant way from a deterministic
design concept by defining probabilistic fatigue design requirements.
Although they were applied deterministically it) the final analysis, the
realism that operational, environmental, design and manufacturing
variations preclude the development of no-failure airframes was implicitly
emrphasized. This is further emphasized in the later changes in Air Force
philosophy by changing from a safe life concept to a damage tolerant
concept, as described in Seccion IV.
A related change in design philosophy should also be noted. In 1960,
in conjunction with the Navy, the MIL-A-8862 (ASG) Specification,
"Airplane Strength and Rigidity, Landing and Ground Handling Loads".
was established and incorporated an unfactored design load concept.
This concept applied only to the landing loads analysis. Limit and
ultimate loads were not specified for these conditions. The incentive
to deviate iroum the 1.5 factor of safety concept was the realism
provided by available operational statistics of the type describeo in
Reference 17. A general dissatisfaction with the design load concept
resulted in a change back to the use of a 1.5 factor of safety inl the 1971
Air Force revision of the specification which became MIL-A-008862 (USAF).
However, the desi•.im load concept for 1,•iding loads is still viewed
favorably and is being used by the Navv.
The Air ur'e on v Ya pplied thre desi qn I ntdi. ln 'd concept to two
airplanes. On one airplane the Concept was applied to tihe first two
models; later models of the airplnte were changled rnd redesigined usi n(
the 1.5 fctor of safety. Unfortunately, the limited appli cation of the
desigqn landing load concept a,1so limit', the ex.perie oce base available
for evallutiiol. Apparenitly, there was vt'ry little, if' any, penalty
5,1
AFFDL-TR-78•-
involved between the models which employed the two concepts. The in-
frequent application of the design load concept and its eventual
elimination from MIL-A-8862 can be traced to: (1) on Air Force requirement
that transport airplanes be compatible with and certified to FAArequirements; (2) a lack of clarity in the specifications and differences
of opinion regarding which loads are affected by the design load concept
and which ones are not (carry through structure, nacelle attachments,
external tank and store attachments, etc.); (3) difficulties in t:he
interpretation of interactions with other requirements relating tomaterial yielding and aeroelastic effects; and (4) the added difficulty
of applying stotic test loads to the airframe through the lower strength
landing gear. The overriding reason for these imolementation problems,
however, appears to be poor planning. Th, concept was conceived andimplemented too quickly and without appropriate trial applications. No
loss in design efficiency is expected, with respect to static strength
requirements, by using the 1,5 factor of safety, Current durability,
damage tolerance, and dynamic taxi requirements will add more weiqht
than could be saved by using either the design load concept or the
factor of safety concept.
Design philosophies, such as those noted above, evolve over a period
of time and revisil.ins normally follow a series of trial and error
applications. Basic hiilosophies are formally adopted and maintained
in specifications and handbooks. The design specifications and handbooks
preserve past experiences, correct previous mistakes and prevent design
oversights. This effort attempts to maximize structural integrity and
reliability but the concept is not foolproof. New mistakes are always
possible with the rapidly changing state of the art and the increasing
severity of the operational environment. Because of these complexities,
the (locUMeIts are aifficult to keep) currLnt and new or revised design
requirements are normally written into the statement-of-work for any new
system development, if they occur between the revision intervals of a
specification. Criticism, then, that certain specificatioh,; are not
current or that certain requirements are not rational, may be correct
but they do not hinder new system devwlopments. When similar criticisms
55
AFFD. -TR- 7,1-13
are leveled at the 1.5 factor, of safety, however, the evaluation is not
so simple: should it he replaced with a reliability-based concept?
What is generally not appreciated is that the basic airplane design loads
(static strength) are essentially statistical in nature. The design
values are derived from i .5road S.pectrt'1 of operational experiences.
If a distribution were assumed, then the design lifiit and design ultimate
loads would represent a certain probability of occurrence and exceedance
(Figure 14). A justifiable criticism of the factor of safety, tnoigh, is
that a fixed factor does not recognize the variation of load or strength
aid does not provide a uniformly efficient structure (Reference 25).
This effect was illustrai:ed in References 18 and 20, as discussed in
Section IV.
Unofficial Air Force recognition of a variable factor of safety
concept has been established by allowing certain structural design
deviations. The static test failure of an engine inlet duct at 1.3 times
the 11.mit pressure, for example, was accepted when it was established
that the internal dynamic pressure in the duct would not exceed the
design pressure in a dive. The resultant delay, rfdesign, and cost to
bring the duct structure up to the normal strength level of 1.5 times
limit pressure was therefore avoided (Reference 25).
The Air Force has not formally adopted a reliability-based st.,t'ctural
design criteria, although some requirements have an associated
probability of occurrence. Available procedures that could be adopted
vary in concept and detail but their philosophical crinciples are the
same. References 26 and 27 summarize some of the philosophical aspects
that appear in the open literature and also note the complexity of the
reliability based design problem, as paraphrased in the followi,,q two
paragraphs.
The many proposals for a more v,"tional criteria are related to the
appearance of new structural mat rials which exhibit improved strength
and stiffness or weight characteristics. Other considerations are the
extreme increases in the structural loading environment and concern with
56
AFF[)-TR- 78-8
SDesvign Ultimate
z
-Des.Ign Limit
N C)> I.
n24 4
moo'
0
Figure 14. Statistic:al Representation of the Factor of Safety Pe-1iqnConcept
AFFDL-TR-78-8
economic costs. Designing to an acceptdble risk while keeping all design
factors in proper economic perspective would seem to be effective but
che concept of risk must be quantified before an acceptable risk levelcan be determined. The new materials also tend to exhibit variations
which could, in a deterministic design, require such large factors of
safety as to nullify the improvements. The increase in extremes of the
structural load environment are primarily new to the civil engineeringfield while economic costs dre perhaps new to the aeronautical field.
[Reference 28 notes that the aeronautical engineer has for many years
consider,,d new failure criteria (fatigue and creep), new materials and
construction (brittle materials and fiberous weaves), and more complex
loading conditions (temperature-load histories). This has resulted in
greater variability in the applied and failing loads than has been
encountered in the past.] Picking the worst possible load conditions
for design is no longer considered economically feasible under a broad
spectrum of load conditions and the statistics of extremes must be
considered for rational design. Reliability based analysis permits amore consistent approach to structural safety by including the statistical
variability of load and strenoth in the factor of safety evaluation
(Reference 26).
Most of the early studies in probabilistic design considered only
the fundamental problem in which al of the strength variables and the
load variables were lumped into two random variables. These studies
concentrated on the effects of different safely factors, coefficients of
variation, and frequency distributions. Later studies included multi-
Illember and multi load structures, different levels of failure, and the
dpplication of decision theory. Several problems mnws t be considered in
the context of a reliability based desion. First. is the reliability
analysis (if structures with derived or aFssrikt,! probabiiliLy distributions
for random variables, inc1 udil.0 load and strenltth di stri)utions;
developing and construc tinq the necessarv computational models whic'h
0A)ounlt frot in(IketerrkWinlany, the tylltps of failuore .odes (incl lding.
elastic, brittle, and l Lllapse modes) , the nurimber of lord conditions
and failure modes, and their" statistircal correlation. Another problem
is the des ion of ,i ,tructure ill thte ,•notot of a ,'andoni vari able of
'A
AFFDL -TR-78,-8
safety for a given probability of Failure; the random variable could be
cost or weight. An additional problem is that of parameter sensitivityand determining their effects in the load and strength descriptions.Most reliability analyses assume 'that strength anid load distributions are
known. The high cost of obtaining load and -strength data will probably
necessitate the acceptance of lower confidence levels in structural
design than professional statisticians asually recommend. Sajbjective
statistical analy~si is needed togJether with studies to determine the
effect on) optimum wei jt and cost due1 to changes in choice of frequency
dis"Xibution, coefficients of variation, and otner parameters (Reference 27).
As part of its program to maintain 'md develop structural design
criteria, ttie Air Force has sponsored various invostigations intended to;ead to 1reliabilil-y based criteria for both airpl anes and missiles.
One investication (Reference Nl) was parti llly presented in Section IV.
This investigation will he fur11ther discusse-d with two others (References
29 anid 30). These three investigiations eirphasized static strength
reliabilitY, al thoU(h fatikue Or- dur'dibi: i Y requirements, can be in-
corporated wi thin these concepts,. 0 her, Air- Force investigations havemore thoroughlY emphasized the f,.t igue aupects of a reliability based
CH1 ter ia. Re ference 31 . foy` exatitp1 trevats fatigue design cons idera tionswh ile Reference L.? emoha si.zes fa -i ue but ilso prov'ides 1limnited treatment
of s ta tic des ign cons i der t ions Re ference 31 is ao extens ion ofReference 32 anid Reference 33 is an evaui ot ion of the concepts; in
Reference 31 . lhese ta:i que related references, 'Ir note here or ly forcompleteness and will not he e-mpha sized fuirt ho- T he thbree inlves t iti t i OtS
reolating. primarily to Ntalt i 'tr1entith desiqli cons id'r( t ions, however,
will bevuir o in more, det a il ec aluse of their di r'ŽCt correlation
with thet factor oif safl t v ckw-'ep !I) e.\p,)nrd i lit. these, t bret' reT iabiIi tY
based efferts, thet philonsophy ot the techi; quges will be, omphas ized andriot the technical aspects of their' development or' application.
A stat istica] II v basd (.oncept dýeveloped for' Miwi sIlt deS' 1r is found
in Reference ?1I, Part 'T1. The concept is brnoad, howevet-, and can bte
applied to almost any relitiaM 1 i thibsedi des i'in Probl em. ýhte desi gn
AFFDL-TR-7,l-8
concept is basically a framework dev~loped to be consistent with the
premise that there is no definite demarcation between safe and unsafe
design, but a gradual change in reliability between safe and unsafe.
ihe idea that a sharp change exists between safe and unsafe can cause
unnecessary redesign for every slight change in design load or reduction
in allowables. Redesign should be required only when the overall
reliability of the system decreases appreciably. The term "framework"
applies because the wide scope and complexity of the problem did not
allow final refinements to be made. This concept, as discussed in
Reference 21, is described in the followini: paragraphs.
This concept uses any number of design variables that are essentially
independent of each other to determine interaction envelopes that separate
failure from nonfailure regions. Design parameters are presented so that
criticality continuously increases as parameter values increase or
decrease. By superposition, a single interaction envelope is formed
and the probability of not generating points in the failure region is
the quantitative reliability of the system for the time period considered.
To simplify the computation of reliability (which could be obtained by
integrating the content), an equivalent value of each parameter is
defined such that the envelope content is approxilat ed by simple
multiplication of the probabilities associated with the probability
value of each parameter. Anly number of statistical parameters can be
used. Limit,; are imposed only by the analysis time and data available.
As data is defined, the number of parameters can be optitized and should
include five to eight that are random varying and fifteen to thirty that
are systemaIti call y Vw rvi ng ; any (listlibution can 1be accom1odated without
the necessity of having to find a speo(al fundction of the variable.
The intersections of any two illter.- ttionl curves are defined as"nodal points," which are used as design puints. The analysis is
relative y insensitive to the exact interact ion envelope shape bet.ause
any reasona,le env, lope shape havingo the same coont et will pasI ne, tbhe
same desiign points (Figure I)). The Ls6 ignll .onditioll"n or not.J: 1'oiln Lt
are defined by two pa rairete,' , one h1vin01( a imitinlg valIue (,X . ) and
60l h
�rrnt -TR- 78-8
C,.'x
C,
4.1
1�
W i.�
�'U
10 C)� U . 0�-*4 aI.a.. 4 44
¼
44
,/1/ I)-�
4.4
.11.n �aI
�1.. (14
I,
F' I' j 'I'
(NI,,� 'F1 41'-4.' -t IL?.�jI// 414
**"I't.14j
4''
U,
I,'
14
\ 10 Nti I *I.iW. I X.I.S Itt
A.!. I *lI tIVilONI
I, I
eak t~ I pio In If i '~~ % to~ I t'1it I ot I Incka u m nl, an' .t o rettIion I at I r
K, wh i k h i I'tO ilte IIIII I Ioilat b aakn' lit roreentalit ie p'tobkihtily t
t'oiwi'i l,pnk I a I I t hn hitu arw" A O t'd to ev I nui I I I lii depe'nden t ott I~ Pt s
and tit hill- ', r illil it I I I' I ilk hn1 11111W'. WOI't detio Ioped I or hind I i noi %%~ % I 'til I i
wadlt! I I oi¾ le hod Ior' dlet vml ni no It he r't'i i ted c5on! 1 dvnt ii levvelI of
lilxt 1,01loIclv 1". 51 aIIlQ VI it~ t A". 1 to ttt~ otNI % o'tnt Ite i Ai I Ii I v a naI v% I % "itt
Ak(I1111d h I' I't kil I Il Mat tit,'IO ta .t I Inne 'Ild I Vt 0i 11 m hti tt Ititl 1. G e ter a t iki
I I tit, I Il' I o kil olt' ti hod t I i A t Itm I A h i lto tI Iloii d 101ti a I 1 t' . 1 o1
A'ii It V~ I u I o Ait I I .I II vI I it'! It I. t It l' t, tt I fill lvkIt toill ' ' ,t ' I (l i :
and lldI I V% wo o , j1"I%% 1 tha hlMe jditm %evt io
14 u
1~(14
IS
tL
A V F. -TR 7-18
Thi des ip met~hod is, charok t ori :ed by the, t~how'ht that1 t he oh i it.v
to .:O cu latep o probah Ii i tv otf a i Iorek en 10o0d and t i ;J h ,Nc tra are
aS(Olled to bev knlown, Is (i ite di ttwon'fi th1101 thei atd' 1i ý. f.0 eo i inthe true s trilktura I re~ ai Iib tY oI an o perat ionial struct~ur.a1 sstm
The method also inocorpora tes certain cons idora tions t hat appewi t.. be
overlIooked 'in Mil e-- a pproaches. Thes- woers ith ts are: ( 1) or rors or
di scrolnmc los which occur betwooen actual anId COl iul ated s~e a()thr
influential1 ef'fec t 0ý testi n(q(1 (I 1111Mea of des 1i.,n error l. lsr
(3) the liece"Ssit y for demlonlstra t i !i proof (if c omplhalore Wit h uiri e.s
and (4) the wecessi1ty for assiqninq) rvspmls ibilflt a G ctionls whichl
affect S truc turalI reliability. Other inoterac otios wi ti 00 *! r
operotba 1 o tuI, onatieria 1 , atid contracturai . area1MS 'rV .I1so inc uded.
4The overallI invest~i(,t ion and proplosed doslkt 10coliept evolved ill
three. steps. The first. 01ep PVva1 noted the variows func-tion's which
CoO tribu.te to s tý,(wt~urc. desim I qo ihe qeconid step OVO ii 'teod an 00coflored
the cut'rent. factorw of safe~ty comcept and a (hyplothetical ) purelystatistical structural r",1 iabi Ii ty CIccpt. to0 the structuralpet ~acand desi I n fo t I ga stablishled inl the, fi rs t st ep . The t~hird step
evaluated exist inq( and proposed1 reoliabili ty baseod concepts, to the Salle
standards of eval1ua tion used ill Step[ tWo. These eVAL ut~ielS concluded
tha t: (1 ) the currenlt fact ol of safetyv des i l1 t echo jilco i , a 5s t i *a('ctovy
systemi but. tOat miore s'trillqtiot flotutre reipui rellen'lls will mliltirln -v the
etfectaveness (it the system; (2) a pure ly statistical stnictur'll
reliability based system is nlot prac ti cal s tince I here is no0 way t
aiccura tel y 11ieasure structulral reliab~ility anrd it is fltot poss ibl e to
Wr itt' doi' init.i vi' reiv eent odmowstrao Ieth111 re 01ii it'i tv (11roof %If'
Cmliace), and (3) that. fbow, of the~r Known Orlwturod iit1l~b ili t~ybasedConcept', il the li terot ore todaly p'ov ides a sa a oyfounda t.ion i'or
ao euabtitatie strut tl oawl prvdes nc mli o ci a sod imn the dis it ill methods.
Which evolved, the phlosophie'i; Which %qoverned hiOue. ~ ln witl beexpanded ill pao rlqrawilh
AFFOL-TR-78-8
Tho~ fundamiental purpose of any structural design) effort is to developa.-. operational structural system that satisfactorily performs its mission.
This de.'eolopmenit is the result, of many manacement and engineering decisionswhi~ch are a key olement iii the success or failure of t he design effort.One aspect of the 'factor of safety concept is that the design effort canl
be practical and easily administered because it hqs anl inherent proof ofcompliance (test) provision. Its fundamiental problem is that it has noclearly ideritifiable quantitative design objective to satisfy. Theavailable ogic cannot resolve the comparative adequacy of different factorof safety values. lIh somie design areas, such es fatigue or high temper-Atures, the factor of safety is not even directly applicable to thedefinition of the design requirements. Thle concept only, defines areldtionship between limit and ultimate load, which normally controlsthe designi strength leve-l and does not allow for an assessment of its"correctness," other than failure. Positivo margins of safcty do not
prevent failure. Gross error's indesign loads, analysis, and largestrength scatt~ers c-ontribute to failure at limit load Or' less. Structuraltests, on thle other' hand, are a nnirly perfect disclosure of gross errors
if the strength scatter is small, as is customary. The trend towardgreater scatter and a lessoning or inability to disclose analyticalerrors req~i ires that the pos si bili ty of failures bel ow limit be con-s idered mcire seriously~ ill the future. Further, test conditimrs arenormasly selected on tile balsis of thle strength analysis and the actualdesign conldit 1olns are bvcom~i r( 11ore difficult to Simla~fite When testing.Successful ground tests, therefore, do riot. guaraintee successful
o pera t ioný-1 performance and flight testi ng will reria in anl importantdes i~il devel nplient. conlsideratiol regiard I s' of the devs in concept used.
~*~f rou I einn t~,hve evolIved primail 1y as a reaction to past
probhill a'nd;J tho assunm I im timfht ful ore s truc tural systems will have theSamle cha rac tel stir-S 'i vs t s' m iý m ot. nocessa iily valid. Whens trucIW Nualflil-e'; (it o ur,kt o i dt~ermiiii SI tic foica~:'e of Hthe f10~01, ofS fetv concept a 1 lOWN tilt doterriii at.l (ion ofthe cause, IL' ri's t)ns i lit yond the korrec 1. e .rcti on to the takenl [blufr tiaat elY, becaulso of Itlieman11Y il neraCt OIVý thettwei i'01 st1W i ýtl*R and Ot her dPS(il -n.--(ms whic.h
611~
AFFDL-TR-78..8
contribute to structural integrity, responsibility is not always
recognized until after a failure has occurred, as the result of a design
oversight. These and other considerations previously noted relate
primarily to the factor of safety concept but they also interact withreliability based concepts. Especially important are the considerations
that establish design compliance and responsibility when failure occurs.
When cause and responsibility are not determinable, neither is the
corrective 'action.
The structural design concept that evolved in Reference 29 utilized
the desirable features of the factor of safety concept and improved or
replaced those not desirable. The following bas'c characteristics are
included in the concept:
1. The deterministic type of requirements that give the factor of
safety concept its practicality ind administrability are retained.
2. A clearly identifiable objective that serves as a basis for
judging any proposed modification to the factor of safety concept is
established.
3. A structural reliability goal is part of the objective. The
goal is not a requirement since structural reliability, per se, cannot
be determined accurately enough to serve as a contractual requirement.
4. The techniques to convert the structural reliability goal into
deterministic reO(uirements based on statistical considerations are
developed.
5. The capability to deal with structural systems having large
strength scatters is incorporated.
6. Specific problems such as fatique and hiklh tomperature design
can be inteqrated into the structural design to at t,lin the defined
ob0iec t i vr.
APFFIL-TR-78-8
7. Thle crucial interfaces with nons truc turalI des ign requi remnents
are identi fied. Prov is i .ons tire made for assigning responsibility for
every function that affects structural integrity.
8. The concept of testing as a disclosure of error in formulat-ing
design requirements is utilized.
Bas ical ly, the structure should have a capabiliity to survive
desigqnated overload and understromith situations caused by undetected
errors or oversi(Ihts. The factor of' safeoty concept provides this
capability but thle provision is indirectly arid inconsistently applied.Structures with larqle strenoth ,cIttvr-; are basically more prone to fail
from understrenqth cons idera t.ions ra ther Ulan from overloading. The
concept in Reference 29 establishes separate and distinct requirements
for unders trength and overloaoe si tuati~on,-, The requ *,mients are based on
probabilities and statistics and are selected to be consistent with a
level of structural rel iabhility a ppropr iate to the airplanes miiss ion.
The desiqo and mi;ssion reVldtionShips ar'e ilustrated in Figjure 17. The
central bar indicates that thle limii t des ign 1load incl1udes a provi sion to
handle an understrenqth s tructore to avoid fail ure it limit 1loadi. The
rigqht or left bar i nd icatvtesthe overload provision. The left ha r
ill us trates a 1 argo over oi Od prov is i n and overrides the noeors trengt h
provision. Thi s could represent d ) TIl lvely low reliability. high load
factor fi ght 'or .t j rplaoe. IPl rieqht tho ill us tra tes a desigin si tuat~ion
wi thI a sila' 11 or overl -Ioa d regis i TWOer II. 11W 1111derstrong11th prOVi Sion is now
mor'e t-Ti tkd 1111d kgOverll, theW des ion. lThis could( represent a design
requ iremeont f'or a highly ~ re Ii a hi , 1O ow1oa d fac tonr t:rans port a irpl1ane.
Once t.he appropriaite des i in voltuws ire chosevn, they bev ome deterministic
and are 'Is eait) adiumnis ter a', the -onverit ionno factor of' safety
Concept.
The first i mplement ing Atop i'" to sl t a structural rel iabiIi ty
goal,1 conistenk~t With tile. Th.e'. noi iot a requiiremett and
suggJested vIHOl 11Ty ( given in Retermines .'") a iid .30. Th litIi i t a rid1
ul i. ima tied', i go1 Olnd itI iOnls Ire two I%) I rte cond it i ols ha so'd (In thle
COel1i 1i t.V 10a a1d Ie s tdtabi ShedI by ItIti t i aI or gl i tatIi v t
AFrDL -TR- 78-8
H
H .�
�Yflui ji�HH-J
- V
c�J
U
Lw
U)4-Q.
U,
I C
II, U)
__________________________________ C,
a
I C
HH . I -0? A Co [
� L
fl L
______ ________IL......W IH
1
H
IAJ� I� C'
0-
i�J� H H0�Lxi �
U)iI C,,
rIH �
C,k-�
I',
A~FFDL-TR-78-8
considerations. which reflect available experience. The design limit
condi tion is tire upper bound of a nocrni io or- expected (permissible)
oper, ti ng condi t on whil1e the? Ulitima te condition is anl abnormal opera-
tional coitdi Lion reached only as the result of an operational error orfailure (if a nonlstruIctural system. Fail ures within thfe understrength
design provision are at structural responsibility and require correction,
while failuhres froml overloads require, Operational corrections. There is
nc fixed factor- of safetyv separatinq the limit and ultimate condition,
the ultimate condition in Referenico 29 does niot represent the conventional
meaning of ultimate load, The condi tion is, separate and unrelated to the
limit condition, whose meaning does inot change. The iii timate condition
is a desiqn condition based onl a rare or- abnormal situation and may be
significantly di fferent. from the conventi~onal ulitimnate load.
The biasic structural design is, qualified anld approved by conventional
gr-ound aind fligiht test~s. Grou~nd test loads areo (efinled by a limit or all
ultimate test. factor of safety. The fitor select. dl . aries according to
the established structural rel jab i tyv tjoa I and by the number, of tests
conducted. E:Xamp11l e values are shown inl r iýJ l g 1re 18 1d 19. Fl i ght testsare conduc 'ted it) a conventionadl ma uner anýMd op~era t i0ona1 flight 10ocd
anloni tori ng are reqni red to) verify oper-ati onal consistency With) design.
If niot consis tenit. t~he s trluc tore Woul d reguial-1 mod if icat ion whenl
opera t ions aIre de torlimmi men t 1oir the oplera t i ono proce-dures mlay ~'quir
Some' changeo
fihereo art, !1,11. nyrmI*Ii j'ja in,;, %uoia if i cations. advaintales, .nd di s-
ailvaltitoges roto 1dtd to t-eli ab ility v Isod dos 'itill collcopts asexore-ssed ill
Reference 1"9. 'Most of the i;socioited idvnao rpr-oblemls are niot
flow but have a Iw~sbeen pr,0oW11 em . Unch IS the stati Stical def inIi tion) Ofdesi on data. No new irtob em,; are cre~tvtd hv i ltt Odu i nt reliability
based concepts but old problems become mot'e iomeal v defined. The ba,;ic
advanitag~e of the proposeod concwept ini Retwferen e '.is hillit v to
Cs hi shstrctualpoaoratoc infoa';Of ai gutafI1i; y1t; deli nab ILeM
gooal. lh penl,1 t'~'i I d'.t ioni I f til he l ai 11i miuiIH o Itaau >:a tolulreaatnt S
t o mot t he (j.1 I . .1."o l( I. t li'jus 1 ficka I i on of a 1;1i ,ss se ere
ru".klv ~ i~lý 1` ov ilo w. t' .~ ne. taa ed. btk - tl i i ot hv' doI'. a g
AFFDL-TR-78-8
LMI t IDes I gn Woad = ln It T"'lt Lojad Limit. Load x limit TFS
3.0,9 S~ructural Reliability Coal
.9999
~2.5
2.0-
0 1P Special ,Justt4 i-uu4 cation and Proof
um 1.5 This Region.
"1 .0l T- T
0 .02 .04 .06 .08 .10 .12 .1.4 .lb .18
STRENUTII SCATTER
Structural Rellambiitv' coal = .9999
1/
3.0- Number of Tests
S2.5"
0 10
o 2.0-
•1.o- T--1 I I -T--T-- I 1 I-[ -- T _ T
o .0' .04 .O0 .us .10 .12 .14 .It .18
SITRENtt SCATTiER
igure 118. L imit Design mnd Test. Factor of %i fety for a Given
I trLItura Reli.ihilit Goal Versm.s Stoenlgth Scatter
(Fference :9)
?0
AFFDL-TR- 78-8
Ultimate Design Load -254 Ultimate Test Load
Ultimate Load x Ultimate TFSS0
C 2Structural Reliability Gosl
Speci1 Just9fication/ .e99999999and proof of ys nee.e 9999in this regiun.
I . .99
1.0-iI
O .04 .08 .12 .16 .20 .24 .28
STRENCTH SCATTER
Strueturial Reltabiltty Goal. ,9999
b,w
0Number oi Tests
0
1 0 T
ua 1.5
1.0 . 1 1 T200 .04 .o. .,2 .,,, .20 .2., .28 .•2
"SIRENGrTi SCATTER
Figure 19. Ultiate Desiqn and 7est lctor of Saietv for' a GivenSt ructurai Reiiabilitv Go;M1 Verous strt ngtI" ".,atter(}Y:t renct.e 2q
,; }. .
AFFDL-TR-7111-
aspects iý quatititized. trade-offs can be made betweeco; cri teria reductions
and the difficulty (technical, cost, time) of providing more efficient
structural characteri stic,,.
The proposed reliability based conce'pt in Reference 29 is reviewed
in Reference 30 to identify data requirements, necessary changes to
design specifications and handbooks, interactions with nonstructural
design areas and the steps required to implement the concept. To
illustrate eachi step of the concept, Reference 30 uses several design
exampl es. First, simplified duoiruy data is employed and then realistic
data. The categories of required data are defined fur-ther by a study of
'data pertinent to the C-141 cargo airplane and then by a trial applica-
tion of the concept to its wing. Thie revision of data to reflect an
improved state of knowledge at eachi desiqn. stage and during the life of
the vehicle, and the form in which the required data might be standardi zed,
is also discussed. Although the limited study did not allow an extensive
treatment. of data requirements. Reference 30 provides insight to the
comp iexi ties of the data problem as it relates, to rel ilbil ity based
des i qn and siimilatr concepts. The data requi remen'ts , limi tat ionis, and
des ign i nterac-t ions are presen ted in the foll1owingi pa ragraphs.
To be, effective, data must be establ ished and updated continuiously.
Fundametital dat. are operational load spectra, error functions, and
s treonth dis~t-r i ht i ons. These data will cha nge periodically during the
toto l Ii fetlime of a spec ific airplanev. The particular periods5 that
piermiit prokiretssi ye upiat inq are the iniitial , detail . and final desio(n
p)ha'sev, htore, and a ftov te~;t s, aind teofore and duri noý airplane opet'at iotls
It is throuoh the airplanes opera t on that quant it ies of new data canl
beI obtai'od ; the daIta '1re alIso pertintent to ot her' des iol conicepts.
In dune, 1(M a4. speIc i a1 panel r'epo rt (Reference, 34) of tht h NACA
~' o 01W'! t on Oil A 'f t I oadý ITCroe onme nded a proworam toI obtain S ta -
i~t hl ~noiiat on 0!!lltMUVI*tOV' 01d rt't'1tt'd intlioqht loads, whetherVCaLl';ed by l'i Iot i ndi etuen t' orat.1110o;phentiC t urbUlV)O 7t110 .1he ) panl aSo
reominddt hImt ttue Alt rmoe amtd N1avy Alta inl time hi storie (if t hree~
AFFDL-TR- 78-8
linear and three angular accelerations about mutually perpendicular
axes, airspeed, and altitude to establish a statistical design base. The
reconmmended statistical maneuver load program was initiated as a joint
effort by the Air Force, Navy and the NASA in 1956.
In 1958 the Air Force outlined a long term program to collect and
utilize flight measured data. The program, initiated in 1959, was to
develop techniques for integrating the statistical data into existing
design criteria, review aad improve the data recording and reduction,.
and tc. establish fundamental requirements for structural criteria based
on statistical methods. The resulting effort identified certain
problems which were grouped into three categories: the definition of
desigr; condiL'iens, the definition of component strength distributions,
and mathematical procedures relating the first two to structural
reliability. The program also led to the sizing and establishment of a
data reduction facility by the Navy and an 8-channel recorder development
program by the Air Force.: State of the art limitations eventually
terminated the recorder development program and in turn closed the data
reduction facility in 1969. References 35 and 36 are documents relating
to this effort. Other investigations which have defined data requirements
and collection progirams for missiles are described in References 16, 24,
37, 38, 39, 'k aand 41.
More recently, the operational data recording program for airplanes
has continued as the Air Force's Aircraft Structural Integrity Program
(ASIP), as defined in Military Standard 1530A, by the same title, dated
11 December 1915. As part of the ,SIP, each airplane system will be
mc.nitored to obtainl tim, hisctorY records. The parameters necessary to
Plolilor t operatioinll t1sate and dervet? st)t e s Sp~ectra ft Critical
,suctura1 areas wil1 be measured in approximately 20 percent of the
operational force.
To accomp, sh the ASIP, the Air Force irt 'iated plan!; in 1968 to
develop a new and more universal multichannel recordino sFystem to less
s t ri n.ent standards tha n the previous 8-ch,tiane recorder. New retluirements
7 3
AFFDL-TR-7 8-8
were prepared and in June 1970 the development of a 24-channel digital
recording system and a ground playback unit were started. The system
has been developed and is now in limited use. The data tapes for each
system willI be colIlected and compiled at a single location when the
program is, fully implemented. The parameters measured and sample rates
can be varied to the specific needs of each systenm. Figure 20 is
typical of the data to be collected. Plans to establish theý necessaryparameter cor relations and design load spectra fron, the ASIP data are
being formulated.
The major objective and justification for the multichannel program,
is to provide a better tool to accomplish fatigue tracking. However,
because of the high coinnonalty hetween data needed for fatigue and
statistically based strength design,, the reliability based dcsign
c .oncepts will also benefit. When a recorder proglram on a certain system
matures to the point that statistical stability is obtained and no new
operational fatigue related information is produced, or when certain
parameters attain statistical stability and need not be recorded fulltime, ~ ~ ~ ~ ~ ~ -teruligsrlsorcrder capacity can be used to establ ish
statsticl srengh citeria or to fill knowledge tas for exomple,
the phasing of power spectral density (PSD) loads. In qust. analysis,
current rsD methods; al11ow a fairly precis,,-e but. sepairatc . determi nat ion
of shear, hendint), and tors-ion at a tliven location; the phasinq of the
three vec tot's is 1largely a tluess . The addi ti on of strain qaqo c 1 o& ters
or rose ttes at seleccted locations cool hiprovide actAual eompitft' of the
ampl i tu~de dni frequenicy reb t ionlsh i ps Such da tr will be essential1 if]
foI toede Ilstoepes plij d 10N 111ds anti st l'oti ra1StIrentit inl 0
c CililllO Hset 0 ft e ntvis.
(Id("eaIl ,v derive ddt 0 a k ht t J ltidi:d I ii -1 ateqri1 d Aconivenient. A1IM-'O)c wYOUld bet IWOVidi dd hK' Clhi i'. r 1 I(Idtiii thet ro'Llo i red
des it.1 'Ind te'.t N(L tor'; to thet rel iahili lv level., inllri' of 1iiniimoh'i'
~ ribI nu th 1 it st)IJ ret IT i h d i l-iihotti to' , tilt error t twi i tinll 01ar0 the'
nlumber. andl' Iypr, of tve'.s I leire t it ill , it wqool~ to etrps 1h i
deve lop a' i iOuII et load s pe Irum tbr 0,10 h I ,1dijon, WvHOi. IdUU Lonta in
A~r rm.-TR-/78-8
NO, IrlM NAME
I. r ChICo"k TIime
". It PrI'vte ze Altitude
1. quI valent Airspeed
.. N., Nor�n Aorm.al Acclerat' Ion at ,..(.", g.'a
"N "I I, •a Ic Actevitexlt lon at C.tC., g'
h. ~i'tct•h Rate
1. Yaw R•t•vi
ClevaLtor Pots t itt1'
Rudde'r Pos it i LlI"
I. FHap Pol It Ion
Ii . V (iouniid S'-i.od
12. ,N Nost,, (4air St eterlng Aug h,
I I. ,Strtaln ;:t I.o t Ion I
14. St ra, I1| l L -.o i t ion 2
'I . S•I-.1 iit aI t h.ica' IOil
IS, St ra I' i ;i t Lova ( ion 4
18. ,l all I.II Vl I.I's I t oIit) 1 '
lotali W i ght ot Fitel
1.0. Squaltt Switclh Makv-o'--I,ruvak S.ignal[
.I .I 1 I),Pt 11
Sol, i.0 Numi-iun
.2,' Ilt I tl�ll i.1 , Ci o W'Iguht ot ,I Catgo t1id:att'
iD'I, 1'C. 1 " W ii g Nc i ht
I i ljtut ,' ,'0. 1'1o1)0',t'd i C,, of Mill tit haunn'l RI t'•tot'dt'Vl IP,tf t',|ltl r.%lot the C- 1.11 Airphlni, ti'fIoi r,' t' (01
7 b
AF L FBI- X7M-8
the totail load occurrences for an airplane lifetime, However. th,
simultaneous consideration of both limit (nriiM&a) and ul timate
(abnorial) 1oading conditios will seldom bI I bsS ile because the
permissible Is trength level will generally be different. Each structural
location will require separate mnalysis, since both load and strength
distribution will differ from noint. to poi,nI
The statistics available, even oil ,airplanes which have e:tensive
operational l experienceC, are not. adequa te. For exalpe , additional
informAtion is required on the probahbilities of: (1) weight and weight
distribution; (?) speed and altitude; (A) tyu, s of load conditions
(Yust, pull-uK, rudder kick, etc.); (4) level of loading (in terms of abasic parimete,"); [h) tirme his tory of loaddi n (to describe local loading);
(6) associated load systems ( presure, thermtl qiradlirent, etc.). Theseprol-abilities are not independent and the reultant probatbility of each
combinrtion is also needed. In addhitiom to the avvralq oWr Ivpit<t
conditions atfilningt each seglimlent of the it on pi t•l ile,, it is ,e.ces,,,v
to, drivtye or as'ullt' tihe Tiap, and di;,tribution ,oout the mean. Without
this deti i led level of data, no reali,,lii estimitte of the risk of
ftailore c'an be made.
Alt h.iutiqh t' vitt'ii'iv iiiM terial 1 -.t'rvn th da.ta ePO'W , the allowabl e,,
ret"re'erit, on ly onet dis r rot ' e 'in I ii t he di I r • tlut i on. Ihle for nr of dat a
requ ired 0on,,i,% oA the mean and stIandard dvi at in and the shape of
the distr'ibut ion to be used. The' slo turiil .,tietsth of the final
('0illo iel1t wll l so rot let I the var , it W i',i ,,ed hv all ot theinlht'rent opprltiot;s in libilWat on arlhl .. ld 11to the ,,t trv lh of
vdrl4U%, ',ti' tl al pniI elt lk',L t tI 'i,,t,. o ill, in a inidol ' alminier alnd
tsu.ll Iv in i, in auliQ toi'o',di' ,dinitt adq til ,t i. ,al
It n f ioii lvi " i e ar ',,y to i,'IIV W ! Ia O•l e 'ViI o111 ' ar' of a
nne fiw'~' vo i potlal ion wiio'. v t WNWtiot Iiolls j . 'andw'd form.,htth l',',Ull~t t tl, t'll illXt 0 0, ti it,, fleal tilt, P10t•,0 t r l' t ollit ll y ] O, kutrl i' no+
vi ltnes (,t IOU thie A tr, tura I rl iat i tv prolv Iem .tohei ,i I, or" requi re
empha% i b !hi' oiioi di Attv'nri v hetvlwe" i•,'l t I=, r' n ind emil l " t'r iliitv
i 7~ t
AFFDL -TR- 781-.:
analysis is that the mean time to fa ilure is not d desirable measure of
rel iabi 1i ty. It is the risk of failure that is required for a structure.
There is no accep table failore rite and much more emphasis is therefore
pl aced on hitill (ahnorma I) lIoad,, Ind unusuaI Iv low strengths. This
empha,.is, then, requiires t hat the statistical representations ma tch the
appropriate tail-. of the distributions rather than the reqion near the
IIX)rP frequentVl y ocCurr nq va1 lueS.
The formal re,'ooni ton of possible errors is probably more important
than the specific definition of an error function. The error function
may desc ribe any number of discrepancies,, howevr causedi, in terms of
the distribution of the probable mean strenk1th of the structure. A
number of suitable functions ait available for inii.ifl design use. The
dekiree of dispersion (coefficient of variation) hxas relktively little
inflneltce once tile tte' t. resullIts have been incorporated, when tests are
used IS at, error disclosure. A relatively low sk would probably be
introduceo by tle adoption of a standard error function.
The implementation oi an '• iabilitYv baed tethnique will present
ctertain problems and Re ference 30 ctuqoesmts a t1,o ;taIqe process.
lInitially. there will be insuftit lent da b, available to implement a
total relliat'i litv ha,,eci detsi qn concept and emphasis hould be placed
on the comparar vt, .imi aritiv' with the eqist;rit factor of safety
de, iqn kol1ept rat tier than tihe d I tferelies. .However, even restrictini ntile rel iabil !y tont ept to d-.iqo ctmnditiort' for which data is, available
wi I heI Ip etSiabI Nih a c'orrec t undeiltard14, inq14 1Of the probabi listic
po'lls ,' and el11ouralqe the a,(qu1sit ion of the data required for further
ip ml emeri I , I i nI
I t, fi rI- t 1lIlh' woold apply the ret lahi Ii tv t oncept to selected
dc 1tIn c ondi t Iio, and pIi mlli I v m'phahN I-, fal ti I iar t v wi hI termi inol oly
anli ltlat hei1a t al rel I at I sIho1 11 t it, reid Ilv' itinpo --tanf'ie o(f pa rai'll, Ivrs
eva twI t i nq t he Im lq, I id rel ia)i lit ie t- , i -,t inq airplanlest;, anld Insuilnlr i nq
that • onlt Intl Iv wi tl e\t' Nt it] de'i&tll tin iIi pt N e\ , iNt N 1 t lhiit :Io abruipt
OhM qes, III s r'lkt turld1 illtt q) it , will O e ist. Ihe ilt of rl t iorl betweell
ArFDL-TR-78-18
static strength, fail-safe strength, fatigue strength and damragetolerance strength also require idenitification to permit the wholespectrum c *ructural reliibility to be expressed in a consistent
ila nner.
The secotid or final staqeI that of achieving a meaningful, completelyprobabil1istic concept. will not bie possible until quantities of additionalstatisti'cal data are obtained, especially for- asymmtetric flight cases andcombinations of parameters which are not independent. Not only mustevery possible cause of loading he established in probabilistic terms but
every factor affecting, the strength must be established. Unless a total
picture is assembled, nothing will :be known ahout the relative importanceof the various design conditionsý and interac~tions, about ways ofchanging the rel itbi lity results' by modi fyinq thle operating conditions,Or by r-edesig n Of the str'ucture. Whf-n rel idbiIi ty results areP fiurther
specified as a sintgle numnerical1 value, even when specified as a goal.the rel'It ivt' merit of dIi fieret value~s regarding safety and possible
redesigln must be cons idero~d. The conicept of a si nille numerical Valueofor the reliabili~ty (if in airframe or even a spe'ci fic location on the
a irfrallit is superficially att ract ive, but any real advantage is
Completel y Offset by problems of interprevtation of the number. AnylitidtiementL a1 to acceptabi lity of onle rel i bi 1i tv nlumber over another
that is Slikiiht ly different. will rema in atrbi trary. It is probable thaitI rE1 at ive risk as-Sessment tecnimiue will prove to he worthI while even
when all of the necessory statist i a 1 dat a are available and a completelyprbailiist ic de'siqgn conceplt Can1 b( 10hie'VI'd. r i na 1 imtip) 011n10 t i on Will1
tie qo'~ertod by V1el-pt'r i eck' kai ned duringq the first paeand the
ava ilab ill? v ot (d001 ion dta.
T ho e~ l, se~t iotn will at tempit to p1ak a lte ý t ne LLh kklt ., and ideals, an1d
thobo o1 prov'iou'N sect'ions, into perspo~vt i ye by fiurt her relating t hea
to i urren I des ioni pra"ctice'.
AFFOL-TR-78*-8
SECI ION VI
CONCEPT INTERACTIONS
The factor, of safety has lost some of its appeal in recent years and
probability analysis has been emphasized as a mnor'e rational ccncept. For-
merly, the complexity of reliability based concepts centered around the
*anaytial spects of the solutions hut this difficulty hds been offset by
curre: :omputer technology. Today the prime restraint is available design
data in the proper statistical form (Reference 21).
Dissatisfaction with tile factor of safety concept became more apparent
during the ear~ly 1960's when sur~veys of airplane and missile manufacturers
werev conducted in conjunction with various Air, For~ce structural design
criteria development progr~ans. The generadl industry feeling that the fac'-
tor of safety is gIrowing more and mnore( inadequate has apr-arently not changed.
The initial, dissatisfaction appl ied pr~imar~ily to missiles, but air~planes
wereP riot excluded. The sur-veys also found that the degree of availability
of -flight measur~ed data varies greatly between systems. Its quality and
quantity ar~e both deficient and the par-ameters most needed for- reliability
based design concept s arev often niot. meiasured. Cost arid the inability to
access a sys t em for- the, pluiposes of insttr~'urrenta tion ind data ica sur~emen t
ofte beomeinsur-mounitable pr-oblems.
WhetPIerIl using~ j t`,C tot' of safe~ty or' ).iel jab ili ty ba sed concept, the
dl rrameprobali 1 i ty of f~t ilure will the Sensitive to thle numlber. of
sign li ti cant des itin paramoll~t er's (assumitt rigll1 s itni ficant pa rameter-s ar-e
accounted filr) anld their' Stait i st ica 1 (i stri but ion. The design data must
01W~w ~r s'11a1 or tit I he atl1 u a11en0v-i0o11en100%, it did ued tiIi V 1Pro time nts*opera-d
tio I m I vat', iI i onII,, n11at or'I i alIp -p o tert es';. an1)d b J i I t - upl t', c S t ur'MAIJ 1 pt'ope't i VS
tilt to talI nuanber'l of specific partllAmet Wt- ego i rin statistica 1d(efin it ion
hvroines "ignilf 41icant Iv lrqg. 1ilte state of thet ait' and pr-act icel limintat ions
in s tN i'.hinna ir'~r' t~tt ist ~ aI dt t fo' ech uni ficarit par-amleterl is
%l t!)tha I t hited tr l (W st 110 11 t " mayd be 1110:1' ad ernItw t barn related to ac tualI
rrrd~ . rn1 O stilr' av,1ii rlr 1 eil da'taar' ar il'0 1 SV elotted an1d reOduced.
crls~ ior'-ah Ir of toll (mil 1 0be e 11it'rdd With 1tn Io or ou' T r rSUltS.
Qa
AFFDL-TR-78.8
Although the validIty of extrapolating data for .ýsign use is taken
for granted certain precautions must be exercised. Too often design data
have been compiled from inappropriate or limited samp'es. Data must re-
flect operational conditiovs from all segments of the Air Force; pilots
must be qualified or typical, not highly experienced flight test pilots;
weather conditions must be proportioned between good and bad flying condi-
tions, and daylight :.,nd night operations; and weather cycles occurring
during the year and over a period of years must be considered. Some para-
meters have physicil limits or practical ipper limits and any assuitied distri-
hution must consider a reasonable cut-off value. Extreme values become less
accurate as they progress away from the mean and influence design confidence.
Values selected closer to the mean could affect flight safety. As a logical
extension of the realization that all airplanes of a certain type cannot
(statistically) meet the design economic or fatigue life expectancy, there
is a trend to develop exceedance curves for design that represent average
rattier than extreme environments Forierly, airplanes were always designed
to the maximum expected or extreme environments for both static and fatigue
strength and the static loads induced viere incre,,sed by the factor of safety.
The effect of this design trend on flight safety cannot be assessed but the
iiportance of selecting proper design parameters and haing a fictual data
base increases.
The 1 mitations whicn inhibit induced load measurements have led some
to believe that the factor of safety should be retained on loads but that
all other desib, considerations (which are assumed to be well-defined)
should he evaluated 1'y a rational statistical analysis. These concepts
might lead to a refinement of curren, design procedures but do not chanlqe
the procedures since statistit-,l cinsideiations. have lonn9 been a part of
air'plane and missi le design criteria. Althotugh probability factors for
structural design are seldom exprsed in current criteria, the choice
of limit load factors for static and fatique strenWqth Mnd various environ-
mental desigIn parameters are fundamentally based on flilqht and environ-
mental statistics (Reference '13).
AFFDL-TR- 18-8
Perhaps the miost. imtportan t. conltri hbut i 0 to aireframe safety is that
of testingl. As so d ptl 1no00ted in Re'fer'ence C2, "safety regul1at ions' how-
ever' good anid sophist.i cated, shoiuld al ways provi de for, approval1 based up~on
rel1evaniit experiment , such as-to iiit' ma surinen1t 01 Or cotro 1o01 ac tua 11oads , the
measurement of the actual stren(Ith of' components, arid the demronstration of
the performiancv, of the complete. structure by proof or' stiffness tests." Thi&.
prac t.ice has J1ong beer) the custom of' tilt, aeronau tical1 eng~ineevr. Both ground
and fl ight tests are used to demonstrate design inte(ri ty and optimize the
confl i cting requirements of miniimum weighlt. anid maximumn structural relilability.
Optimization anid economic airframe 1life requ irtemen ts have in rec~ent year's pl aced
considerable emphasis onl devel1opmen tal1 testing. This emphasis will probably
incr-ease in future years. Al though testing is, a very cost effective des;ign and
substantiation tool, the expense of testing -is a major obstacle to obtaining
more appropriate stat itstical data for, relilability based design concepts anid
stati stical substantiation of structural reliability. Some of' the interact ing
roles of structural analyps is and test inq are discusse'd in Refertenice 43.
A natural extenision to curi'en t prrc t ice i s the use of add it.iona 1 fact ors
and design paramet er's. Thiis concept is le-ss complcx and easier to manage in arapidly changling designr environment111 than1 us 01jg a var1iableI 0(701o' Of Safety.
The modified fact or of sa fet y concept (Rofe'renice 21l) factors thbree perfollimancet
var iabl1es anid it. c ould be e\1pauded to incl1udte o ther's. Going forit her, Refevrence,
38 suggested that. the factor of' safe t v on 1toads for U i ssil1 s could he reduced in
certa i n instances whenl a plart. icu 1I'ar' load sourceo i s higIhl y p red ict~ahl V. t hrusv;t
for example. Hlowe~ver, Referelnce lo espresse',d the view that. althoughl Some die-
sign load sourceos are hig chly predictable, theit combi ned des il go I od mlay be, e\-
ceeded for' some designi conditionsW10 weI componkenlts Of the combined load are,
reduced. Ili ef fec t. known Con serva t i it;N C01ompnsate, for' thet 111nkn1own
The additional de~signl comipl ex ity imposedk by a new or rev i sed conicept
on structural analysis most al1so be c on'~itdered. As sOc ia tinq li specit ic ft -
tor with a specific vat' iable, rart1esOf* the nubrof tact ored variables,ý
will pi-ov ide a certa in level (it addi tional1 colmpl e' Itv to a lo. j11s tress, anla ly is
using(. d var'iable fat',CL Or f sa0tety and .,poc ifi 11,11ra11t0(w. will add d i fferenit
level of coinpl exitv,.tmotr e an,l 1v,ýk is.te' hoi q;it' can relie~ve sOi r tilt,
8l1
AFFOL-TR-78-8
bookkeeping, but considerable additional judgement in design is required as
compared to using a single factor of safety on load. Although less complex,
the single factor of safety on loads will not satisfy the objective to control
structural reliability; the single factor is a function of numerous variables
which can vary structural reliability appreciably and not impose a change on
the factor itself, Also, if new factors of safety are applied to additional
design parameters or if revised or variable factors are applied to loads, the
years.of experience and backlog of compensating design limitations and related
requirements which we have for the conventional factor of safety will be lack-
ing and design confidence will decrease until :i new base can be established.
Major changes in airframe design technology are usually iccompanied by a
comparison with existing techniques prior to adoption. Reference 30 illus-
trates a limited comparison of this type as discussed in Section V. Other
studies have made comparisons relating equivalent factors of safety to a com-
patible reliability. Such comparisons ci be found in References 28, 44, 45,
and 46. There is a similarity between reliability and factor of safety con-
ceptF which becomes more obvious when the factor of safety is viewed as a con-
cept based on the statistical definitions of many basic design parameters.
Each desiqn parameter, however, is nonvally reduced to a specific value and
the concept becomes detmrministic rather than probabilistic. To further the
analogy, the design (ultimate) and operational (limit) stresses can each be
assumed linear and represented by a frequency distribution. The ratio of the
mean stresses ,f ech assumed distribution can then he defined as a factor o0
safety. Refer, nce 4 uses this analnqy to show that the level of reliability
can , ry W idel for the same factor of1 saifety value. As proportiona 1 ch anqus
are Med to the stress rotia or to the shape of the distributions, the over-
lap ( the tail, of the distributions varies and, in turn, the reladhbilii,
varies. A fix ed tactor of safety cannot. therefure, ensure a constant level
of reliability without consideri n the tattiStics of tho desigrn st.revqth anid
operational streses. Roference l tfurther expkin us that the fact"or ot safety
can bh placed on a hhre rationa1 b,,is .od. fact. only has meaning when
related to the concept of reliability.
AFFOL-lR- 78-8
The degree of safety'desired or required also varips according to
the oarticular point in the V-G diagram which is involved. Th. left hand
corners of the Miagrain represent maximum lift coefficient and the maximum
load possible (not considering the smnall increase maximum lift co'ffi-
cientq that can occur under dynamic conditions and wnich may be used in
constructing the diagram). Thus, if a static test has been conducted
satisfactorily and validated by flight ,measured loads. it ,ould be physi-
cally impossiblP to have a sNatic structural failure at these points when
operating normaliv. A similar stat.emnent may be made of the flight controls,
which are limited i" loadintj to specific boost capacities, and in a more
gener I sense, to inlets which are designed to specified pressures corres-
ponding to maximu speeds it which tnte airplane will fly. The airplane may
physically exceed Lhe des uin speed and maneuver limits, but it has turned
out in practice that p ilot.s have kel t the aircraft within speed and maneuver
limits without difficulty. In these cases, it might be logical to consider
a lower factor of safety than that considered for tie right hand corners of
the V..6 diagram (Reference 25).
FxceptiMrs to nu-'mdaly contrelled f. liqht conditions are instabilities
which cause design load factors to be exced•d. The factor of safety does
not r,(og ii :v .i•rod\iami" in stabilities in des,•i•. Even with an extensive
operat ional bcl I'0•,oud. past experiences show that all qlt tic flight failures
cannot. hr In'e P . Several late I 's airplanes have been lst because ufunacro•lo ted "pr aei or.'lastic effects, Some los'ses were tIhp result of aero-
dy.rhii int-rdk .ons and one resultd from imprope'rly predicted spanwise wing
Ia ! .0 . SCUP *,qctura1l tailures could have been prevented within the state
-Io art ' ther. have resul ted tfro11 new phenomena not anticipated or a.4 . l iph lt:u " I to a prior ujrb'iu el Iit , up .t [Ref erenlce ').
M' m , or- ,,.ip• of, t o r , or , 1 failiure, i, c i tc•• e i d for both fighter
, , , ', art'. Ir'wq: amp I e'.nw' l amp'ld °P i tin W outro I gk r to•tma WWuIn,-
01 ti. 1 qh IWt tract. rIaI fa i Iw' ';ty% hNooavnrt 1 % cc u orvta tc thr w i nt
I I:e .u'' '.st'V l.i " ,'t and ,o' I tAi I. • -Ni kl n W, I ii'' p l anks tot am',,• ~ ~ ~ nt b 1 i;!. " I,, ,]'"]• io", wv e 0.Vt't t n al']l'l 11• O.Vt' , oVX: . ]• {•ikt t e SN U l ure'I l '(
he . 0 OW01 ki t•t It 1000 Or 1 a MOW Odd
AFFOL-TR-7.13
concept havinct al effective factor of safety lower than 1 .5, theSEý
airplanes would, pý,obably have failed catastrophically in flight. Thi s
failure projection is hypothetical, hut these real eNamples of structura'ýoverload reflect the, inherent conservatism in today's airframe and the need
for overl1oad ptrotection. It is unlikely that the dollar value associatedwith the cost of the airpla,-e and crew training, for any onie of the severalairplanes affected, could be off'set by the savings in weight, the performance
gained. or operational costs saved by using a design concept that might pro-
vide ý lower le el Of stru'ctural safety.
There iý another measure of safety to be accounIted for beyond the normaloverl oad/undt'rstrength probabitity limits of structural failure. Extraneous
causes of structural failuro may arise which are not part of an) original de-sign evaluation. Instances oC pk ir maintenance, improper assembily or reassemn-bly, substitution of improperly heat treated components, etc. , are well known.
Other phonomena suo as hydrogen eiimbri ttlemient and stress corrosion may not
be adapt,.hWe to statistical design procedures and the statistical limitationsOf S114111 coupon or' structural component tests are also well known. Even i fthese eveýnts can be statistically accounted for, their significance could
overshadow the probability of normal structural failure when consideringreliability based criteria (Reference '2ý).
The~re is still another safety aspect to consider. The factor of safet.y
does iotoract with other desiqn and anal vs s requirements, al1though it, is,
often v i-wed as an indepondent measure of st ructura' safetyv. Reference 4~7is a s nay of compari sons between existingl and proposed civil jqi enineeri norequiremen ts that emtphasize a similar interaction. The study first notes themaniy uncerta int ies in des inni and construkct ion that are covered by p)rov iii il~
overlIoad protvý i on. The facets noted are, identical in Context to t hoý-considered Witihin the factor of safety Concept for A r framv des itin . imil1ar-
ly, the tictor of safety concept is nioted to be. a crude method of cover-ino~
analytical and cotist mc ti on errors, but the refer-ence also notes that iH ha,
the mer; t of simpl icity. The study evalunated ivariety' of stt-ucturw ý 'oe
,to three 10 M. o Cdinqcnditions. It cbimpaired the quant iti to I t x eua e eIrequired for vec h des inn case and assessed t he t ucoret 1i1 oveirload I lIt'
AFF DL- R-78-8,
of the structures. The steel requirements using both the old and new
requirements were found to be very similar for each loading condition but
the presumed overload protection was found to vary. It was initially pre-
sumied and implied in the specifications that the theoretical overload pro-
tection would be proportional to the factor of safety associated with each
load system. However, the strength was not proportional to the factors of
safety used. This occurred because the design of the adjacent spans in the
structure used different factors depending on how they were loaded; in
some cases, the loaded spans were counterbalanced by an exaggerated (factor-
ed) dead load on the adjacent unluaded spans. The overload capacity of the
loaded spans was not directly proportional to the factor of safety.
Even though the study showed a variation in over!'ad capacity, exilirng
structures designed to these requirements were still considered safe for
several reasons: Occurrences of actual overload were negligible, the proba-
bility of understrength was low, and inevitable detailing excesses (design
conservatism) existed. One additional reason, however, is most important and
relates to the elastic analysis. Until recently, the complexity of the elas-
tic analysis encouraged the use of simplified assumptions which required up
to 70 percent more steel than would have been required by a more rigorous
analysis. The additional material, in turn, greatly increasedi the overload
capacity of the structure and led Reference 47 to conclude with this question:
"With the increasing use of computers, which make more rigorous analysis,
will structuref, designed according t6'--- (existing requirements) --- or simi-
lar types of load system(s) still be adequately safe?" This question ,.
equally applicabie to the 1.5 factor of safety design concept for airl'anes.
The following are similar points emphasized in Reference '9. The main
point emphasized is that the conventional factor of safety provides for
(unknown) situations that might not be recognized in a more sophisticated
(rational) desiqn procedure. Any attempt to be too sophisticated can also
lead to design procedures that are impractical. To avoid these possible pro-
blems, new and old concepts should be closely compared. Any large differenc,,s
85
h .. . . . . . . .
AFFDL-TR-78-8
in the results should be viewed with caution and not be accepted uncriti-
cally. It is further noted that different results, either more or less
critical, are not necessarily adverse and can often be justified uder appro-
pridte circumstances.
The concluding question in Reference 47 should also be expanded to
reliability based concepts: If more complex reliability based design con-
cepts are eventually implemented to rationalize existing requirements, save
airframe weight and improve performance, will the new structures still be
adequately safe? There is no immediate answer available. Hopefully, any
change in design concept will bring with it an adequate and equivalent level
of airframe safety; however, any design concept is a balance of requireatents
involving numerous parameters and design interactions and any new concept for
which experience is limited must be thoroughly evaluated and closely ;,loni-
tored to ascertain the true affect of the change on structural safety.
Generally, reliability based concepts are not proposed to improve
flight safety. Flight safety is always a concern and new design concepts
are generally not adopted until an equivalent or better level of safety is
assured. The most significant reason normally given for adopting a new
concept is to achieve a reduction in weight when compared to the accepted
norm. This reason is well intended but could also be misleading. The
accepted norm can be elusive and difficult to define and the projected ýav-
ings in structural weight can be easily overstated. Every design concept
attempts to maximize efficiency and avoid either an unconservative or an
overweight structure. In this respect, current desiqin practice has been
very effective. Refere.-ce 19 provides some insight to the history of struc-
tural efficiency for bomber and transport airplaiies up to 1964. The obser-
vation is that airplane weight trends and stru,:tural weight tractions are
very consistent. It makes no difference whether an airplane is jet powered
or propeller driven, whether it was built ,?!• years ago or is of recent (1964)
vintage. There remains a balance between ai efficient structure and/or
material and the design requiremients imposed on them. Studies of future
airpkanes (beyond 1964) also supported this trend (which still seems valid
today). It is apparcnt that if more efficieit mater ials and types of
AFFDL-TR-78-8
construction are found, that more stringent operational requirements will
be imposed on them. Time and state of the art advances seemingly have little
effect on basic structural weight trends. Reference 19 describes the weighttrend curves as basic, and as such, a technological breakthrough would be
required to significantly change them.
The trends, of course, are based on the conventional factor of safety
design and metallic materials; any improvement in structural weight trends
that might result from the use of a reliability based design concept or from
composite materials when used on an actual ai'plane is unknown. These con-
cepts may provide the necessary breakthrough. Howler, an airplane design
is the result of many compromises and interactions AThich have a neutralizing
influence on overall design with respect to any one parameter. Too often
weight and performance improvements are estimated superficially and the opti-
mistic conclusions are not achievable. A thorough study that incorporates
the major performance and airframe parameter interactions in a design eval-
uation is required to obtain a confident weight impact assessment and many
design variations would be required to establish a new trend. ihr pacing
influence on performance improvements to date have come from advances in
propulsion concepts, not structural concepts. Structural designs have
successfully kept pace, however, and structural efficiency has improved.
But as pointed out in Reference 19, when greater structural efficiency is
achieved, greater demands are made and the structural weic;ht fraction has
not changed appreciably.
The actual impact of design data, analysis, weight, and test interac-
tions on the development and application of future reliability based con-
cepts cannot be clearly defined at this time. The siwilarities between
the factor of safety and probabilistic techniques indicate that ary state
of the art imprnvements intended to benefit the implementation of a proba-
bilistic concept would also benefit and improve the factor of safety concept.
This is especially evident when considerinq a design data base and may be
an additional point to consider when evaluatinq changes in current design
"concepts.
87
A
AFFDL-TR-78-8
There appears to be a slightly subdued, but continuing interest with-in the aerospace industry to develop a design concept that corrects tha
deficiencies associated with the factor of safety concept. It has beenassumed that any lack of clarity in the merit, goals, or direction thatmight be associated with a reliability based concept will be resolved
satisfactorily as the concept is implemented.
The remaining section will briefly summarize the salient traits ofthe concepts reviewed and project a possible balance in their future
application.
t.i
88
AFFDL-TR-78-8
SECTION VII
CONCLUSIONS
To have emphasized only structural design concepts with respect to
flight safet5 is in keeping with the intended scope of this review, but
emphasiziag too limited a view could be misleading and detract from other
important safety aspects. Safety considerations in a broader view are
discussed in References 48 and 49 and serve as a reminder of the overall
scope of the safety problem. Safety is a total operational system and all
aspects must be considered in unison.
The static strength safety aspects of the airframe have been controlledprimarily by the 1.5 factor of safety. To be more precise, the factor of
safety has been the most visible design aspect of airframe safety and it
serves as a unit of measure in that regard. It provides protection to
occupants from both understrength airframes and inadvertent overloads.
But the overall concept of sdfety must again be emphasized and not just
the factor of safety; the factor does not function alone but in concert
with many other structural design and operational requirements.
The 1.5 factor of safety in U.S. design practice is fundamental and
represents a level of design safety which has become an accepted standard.
Although the concept is accepted and used without reservation, it has
remained in an intermittent state of review. Its efficiency as a design
concept has been challenged and the objectives of its design application
cannot be clearly identified. There are proponents who have encouraged
change and proponents for the status quo. To define the arguments and
differences between them is sometimes difficult.
Perhaps the 1.5 factor of safety is rational and does not require
revision. Or perhaps the 1.5 factor of safety is arbitrary and its
basic function cannot be defined sufficiently to establish a revised
value. Its history seems to say that the 1.5 factor of safety is a
mixture of h'th elements. The 1.5 factor is rational because it is
89
btaed ntll what wer cnsidered to -be r epresntattve ratis jof desi• to
olperattnq mlanleuver' loa,•dfac.to1s exp~~'erienced• durYinqt the 1420.ts andt IQ.30s
(which have not appreciably changed today) and It is arbittrary because
we still do not know the exac t design, manutiat •or i-nq, and operat in•
Intricacies and variations it protects aqatnst, or how to quantify theu.
Nelther can the do r'eM of in-fl •Iht safety provided by thi' 1.5 factor he
quantified but its successful history cannot he lightly dismissed.
Reliability and retal lsi seem to qo toqether. Probabilist ic design
pconcepts are cons ider'ed more real ist •c and have beten proposed as belnq
more rational than the fa•ctor of safety. Reliability based concepts
have, therefore, *been i;oposed to replace the fac'tor of safety concept,
Because of ant icipated implementation delays, interim des bqn techniques
have been proposed and consist of imuultiiple factor, of safety which are
related to spedif tic desbin parameters and variatble, factor% of safety
which art' related to ,,peclftic dyson need%.
INh primary just iticat ion and final objective oif the probahlist ic
conceopt I s to improve airplane performance anitd reduce operat Ing cost s,
] h I' improvement! are to be qa i ned throuqh retduced a I rframe weig qht.
ftowever, recelnt airplalne woioht MtOudli' and pas;t we bht trend, have'
stown that the actual airplane weiq!ht saved i% ofteen less than antMIX-
pated Durabilityv fatiqtti life, and damaqeI toleranc'e requirement'nt s t Iso
inluence a nirlam woiqht N and tend to supersede sayviIns q; Ifined through
improved deolqn techniques The•,' requiraetn s add weviqht beyond that
needed for stati c qtrvrnqth.
lhero, art, many d•,'•sion, operational and pons•lb loeqal rami1ficatl ion
to be defined eftore the factor of safetyv concept can be formnall y chanoqd
with assured justif'icatlon. ithe lack ot appropriate statist'ical data.
the need for procedures to establi ,h the tru"e structur"al reliability of
a diestin and to demonitrate co't rattual requireen'ts will also hinder
implm•entat ion ot a reliabi 1 ty hawed desiqn concept.
!41
I' AFFDL.- TR- 18-8
Reliability based concepts are difficult to define and understand
In summary form because of their comiplexity. but reqardless of the
advantaqes or disadvantaqes alluded to herein the concepts cannot be
liqhtlv discarded. They must he examined critlcally and objectively,
simultaneously defininq the ne:ded dessiqn parameters. Yet, a completely
.rigorous reliability based concept way he so impractical for structural
.desig n that it may be less desirail e than the easily administered and
less riqorous factor of safety concept. In time, the application of
reliability based concepts to airframe design will increase but the
deqree of their application may have a definite limit, future concepts
will Prcbably evolve tu incotrpoVaPLe hoLh a bhmpl i fication of the purely
statistical reliability based concept and the qross simplicit.y of the
factor of safety concept. The factor of safety still covers many
contiinlenies and at this time it appears there will be a need for some
factor, and to a qreater dcqree than is sometimes implied (References 21
and ,.).
The objective has not been to support or minimiz;e a particular
structural 1 K sin conc!rpt but to underscore certain points seldom
emphasized. All of the disadvantages noted apply to both the factor of
safety and rellability based concepts. the point is, that the reliability
based concept will not eliminatie all of the problems of the factor of
Safety concept or n:ceisarily offer a safer deviqni. In fact, the problems
may tend to incrase because of limited depiqn e'xperience with statisti-
cal concepts and the us, of statistically defined parameters that may be
of questionable validity. lurthermort. there may he an unjustified confi-
Oence in computed re, iabi lit v entimate,; a ithouqh reliabil i ty based desiqn
concepts have been assumed to he more ratlonal than factor of safety
concept,,. 1he phv;ical results of a reliability based design and the
related ,tat it•ical data must be more thad a matheuialical Whet&; thestatistical confidence expressed in the desiqin must be realizable, in
fact. Fina11y, reqard1 'es •f the design tonrept adopted for future use,
the safety of the airframe will depend not only on that concept but on
the ade'quarv of the total itrawtural ,eiqon criteria and the ability
o f the conrept to met't tihe proof ofi comppliane' re uirtemen ts of the
dhsiqn spec if cat ion%, (Refprtr A").
.....................................l..
AFl I O - T R- 18- 8
4111I R1 Nfl
Ni lo I A.", .*Itr. *Al rplane [vtlliq I nloimn'erino Di vidsoni, U.S.~ Arm~yAir' Set'v i tv Doytoni, 0hio0 in yIjv '11l)
'. Mi itary History:
a. [pstein * A. , "Atiothert HisIor i ca1 Note onl t~he I . !I lVct'if (fsa let y", Ridt.Forinji, oiori'li of' heit Aerospace SciencesVol .No. (' , Juie 19~6."
1) 1lpske in , A. , 'IHoilzontaol aind Vert i cal 1.1ii I nods onl Iarg pI ransport Ai t'c ri "t "1 Svecond St rut unral 1L oads Work shop forPirtoIl iII i narY Av t'o spa c: Dms i (in Projvec t s , Dpupotyv for Dove)I opron tPlanint mo. Aeronaut ical Nys tuiw Divki'. nn WPAI'l1 Ohio,:44-1 2!ept ember 1969,
c . 1).oi, A. , "SoNIP 1i kI t 00Cl Not Ws on COi t 0 ina l' FifthStrlct tla tiid1Iomki Works hop for Pro 1 in na rv A ro space DeOsImoPt'oJects, AJ. U. Iii ot Dynamics labiora tory and llepnt v forDevelopni'nit P1 annli nk, Aeronriiit ici al tu Division,18-19) Septvembt' l 13
3. Civil IsIltory
a. Sha nley, F. R, , "Historical Note nni thle 1 .5 Facto ot' o Safetyfor Al tcra iI t i Itiresw-,; Roalivr'ý F ot-1in. ,Fnnrn1 1 nt the
Aerspae S I ocu *Vol . ',No. .2. F eht'itvir 14t6.,
h . lloqert G . . * and Al ct'on A. M., *Reqin ited Fu iIor' of Sastelyfor rDe,Jn of(i'lCvil Aii'crat'l." Appendix A Io "1'lithePie-sent Aiict'aft, Ntrit md nip I mi lot' (it "A het v Rea Ii Is it
byG.M. Mantlur Ian, Aort'ospa( I I o'~ ineovin I')l ltv jwSelttiitt'l'ibe 11 ')I.
4I. Norton,~f l. and All io, I l., Adeletat ion-, in 11 itht , NACA H\~~0
Dool1 i tt Ie Ill. A 1overa t i oIvnsI in light., N A CA I R N o 2'03 . 19: "1
6 Rhode , . V. Hi, h Prussorer Ilktii Iibut ionl lyer thte Hot'i :oltd 'A)ndVer't ical 1 ail 1 Sorfavies of a PW-1 0 Pnrsni I Aiinplomi' in rli ilt,NACA UR No 307 , Jlntio 1 '),"I
1. Rhode, R. V. , I ho PressutreD -,t rillit ion ýI'or t he Wing' anrd ll 1Snjrf'aces. of a PW- Pnrqu i t Al rp-l ano itn light . NACA I K No '364, , 10.10,
It. Roche . ),.A. . Proposed Met hod of De tn oDv, ij go1ail 1I oatN fotrAirplane'., Air Covpsý Cirvuliiar No 650,* May 10 . 11110.)
P1 Ii1W NtS ( CoN 1 I N01.))
PacL * MIX. and Boqert , H .. , Compa r ~on tuf t he SIt. r 'uitiural De-.igjnRequ i revitint for A i rplI ant's wit h I oa d OH a i ned i n FuLl - Sc alIe
IT", sUIT D ist rihut ion ITests A i r Corps I nforinat ioli Crc~ular,No. 07,', June, M, t3
10. B ureaui of Atronaut icsý Sec if icat ions:
Sii Snechticat i"" fe the Stand F Ii~ht I imi t a t ions fm_ Ai Iplane s.
5-2 - pe f i t"at i on f~l the DO.fltti of Appl ied Lo~adsonl WinyCll col)(f
SS -3 - SpeC if i Cat i Onl for the St0ru(7tiral Design of- the W~inyGroup,
SS-5 ~Spec it i ca t i ,onl -fo*r Pt' te0rilm1 na*t i onl ofI th 11,[qu 1w.len-tSt'Iidi- Wing antd the, Struct ural Mean Chord from thePropt'rt lt'5 of t he Actual Wiliy.
SS- tiA -Spec i ficatfi on fo r the Deteorminat inn of Air Loads o~nWinyj Ribhs.
SS- I.' Spieci f ica t i on for Calculation o'f lerminal Veloc~ity forShdt ri i xl Pes~j i yii urpo(Sesý
SS-271 -Aerodynamics Report of Othe fi st r ihutj oi-on of- Air, L.oad-sBe tween 10111*a ne Wing". __
11. [1s t ein,. A. . inve'.Fi yat ion of the Compa ra~ti vt' Des ign Reqjuiremen~tsrfor Al rp lanes, of' the Artiy Air.Co rj~s and of the Bureau of Aeroinbt -tcs,
Alr ops lechn i cal Peport No. *11), 4 , rehruAarly 1 U4 (Uiull bid{.
1,' . Rhode. R. V. .A,)p lied Winri I oad Fact ors. NACA Sp i cal Report tothe Se'rvices, 1,113 ( tnlpubl i Shed)
131. Press, S .11 . nd S~teiner, R . , Ani Approach to the Pr-oblem of Est imat i nsevere and Repea~ted OiuSt I oads for MiSSile Op)erat otiS*, NACA YIR 433K
Set emh. er 1L !'8
14. Press, 1I., Meadows;, M . I .and I1adl ock , I , A Reevalu at ion of Dataon Atm.rropheri c. I urbu It enct- and Alirpl ane Gu' 1 Loads 0o AP101 9,1i1in) Slec tma 1 Ca cul at Von, NACA Report 27,'O6
1.Schmid, C.Jl. , Plyna Soar 11ooster- Systemn Structural Criteria,WUL "SC Sect ion Mt,moratfdlim, 'Wri ilit*Nat tversoii Ai r' Irol-ce. NsevOhio 45i433, 7 April 1010).
I t. I pste in , A. . arnd Himi ! toil, A. I ., Fxi1t riSpce and Reent ry St ruct uralDesign Cri teria, ASTD- R- t.2-M1, WI qlt-atterson Air- Tro~ce Base,O~h io 4t4 33, Dect'mher 1406.2
93
All I of I R I
KI It I \1 NCI S (r'ON I IN[I if )
1H. Baum. C.P. I.''RNeliarbi lity o mf Air frame Nt nc tutres G (eneral Aspenctsof t he 'rimtm 1 erm arid out Iitit iteo' Met hod' for At l a i ni "V,' Proceed i rnosof the Avrooparce ReIliab ilily anrd Ma inotainabrili ty ninereionce,was~h i rI Iotoi P. 1). Oi ltn -lu 1 v 1I, I1~) 64.
18. ýMonqui ian, G3. N. , "Is the Present Ai rcraft Structural Factor oifS., lefty Real1 is tic.',Arnu c lIrýinrr~R iw Sept emiber' 1954.
I') Marr , N. ii, , "Bagsic Wrilght I r'mndw for Roinher' and Transport Ai rcraft,"1 went y-1h ii id Annual Nart ionial (out erene of the Society of Aviv-nauid tIcal Weight I ny i rers , May 1 ')o4
;11. kurschviier, H. , and P riwn , I . *lt~~ii~ i c ~t~Concerilit) the(Stru~ctui'al Id tety~ fo'A)ojae e fce RTi1-TDR-63- 4os,,
'1, 1atw VAh~P. Wacnrre',N.. ,RX Wn au.r .A_ Ain I ves tiodtiaton ofthe P'fiiHit ion oif MI.i le Str'uctural QIominr Criteria Re~qulenrentson a Rel itt' I it, Sab'.i, - Par't I , ''The Inte~wntionu of furrent Datahit o Rot'cwomrv'errd Reokli rerrn t S' - Pairt I I , Rvadey. , .BI. , ''TheDevelIopmient orf a Met hod Iramrework for' Ovtr' inii og the Q~uant itatijvteSt rucu ralr Rewl i rowniert NtNeLe;r'v to Achieve ar Given Level ofStrumturmrl Rellailm'int,' WAPP.11n-oO-Yom, Wr'ight-P'atterson Air' ror'ceSOSO,' Ohio *H'-133, Au'ru't NO
2.. atirr, 1- , kroit shaprur, W. A. I,. Or Wil1son , I .W ., St ructuoralCri ter'i, arid , I. Iligmht D)atar (or'rei i aon,* ArIrl'0,1-TR-112 . Vol V,A ir- I oiwce ýi !AniamirIc' I aborat ory', Wrioi nrt*-I'tt erson Air' Eor'etPi;,l '. Ohio Noveiirt ený I o,
3.1p.~ir, A * Hrmi t1oni, A, I . . Sifnd; of' it- igr t'arrinretor's f'or Structur'eSub ject to *\enodpamuntc teat inrg. ASP 1 N-oi 45. Aeronaut ical System,;Pivki'.onr, Wrioift 'it ter''orr At' force Paw, Ohio 01~433. August 01%..
.1. GoI a11i"'nh. Nit.. PmkIrv , Wi.H., Bart t, . IR., ý Jan '%, M. , A Study toDevelopment'r Dot'. 'iii Crinfoia~ for' Aerodynrimi call H~ eat ed Swept and P 'el t a
IF Winrg St-ruc t ure% Vol 1, . Surrmitav Report." AFbiP-TIT-65-l 7, AiNFo'rte fljinht ovnramics labor-taor', Wrightf-Pt'a terson Air fni'ce Fane,kil 10I 4!1-t 33. Mai i-h 1 "100'
LI"IeiA._ Rol i ao'jiit ' Piow dures iii Ai rcraft PD'sigmni equ irypenuit'Ind l I'-i':ni Prac~tice, Unrptibli ihed Notew, Aeronarut Vcal Systerrus
Pivki'.i'n, Wrili-rttiatterwor Air' orce Bas-e. Ohio 45433.1 October' 1%-i.
X S.~ tow,'eiK 0i . , adir Mo'.el 1% " Relirabi lity~ Airrrialyis of I i-dneS1tFUAirct ".' JO 1~ of' lre St rut uir 1 i vi' oin, AS('I , Vol '16No i ll I , Nov'iibt'r- 10'0.
p"-jion,' ,Iounrr.tl of' the \''nt u-i iyji,turlPv ii AS('I, Vol Ob., No S:It 'll I'11
AV F IN - TRi 7a 8,
RI IFI RI NCI S 1 C(IN 1 N~IIF 0)
.'8 Swit. ky~, H., Iie~il''14fl for Strucitural Ri'1 ihilIityv J',ou~rnal o1fAircraftt, Vol :, No P ., Nov-Pec 140,
B.Iouton , , kent , 0._ 1 II k. M. , Mc~uh,~ A. H. * Quant itadtive'Structural fe\i in Cri teriai by Sta dis t icd 1 Met holds - Voj I.
F ''~."A Crit ique of Pr'event aind Prolposed'Approllches to St muct ura 1flt'slqn CritIeriai" - Vol 11,~ "The Phi 1o'oph\ and lImplemientation1of' t he Now Procedure,' A! I I1 -1 R - 10'. Air lorce I 1 itht DynamicsI laorat ori, Wrii nt XarIv i' Ai*\r lorce Raw., Ohio, 01433 , June 10h8.
30. Camp'ion, M.C., Hanson. I.Q. Morcook , P. S., *Implemnentat ion Studiesfor a *i e1ahil1it v-Iawd St atic'St remithI Crit erias ~'TWA, - Vol IAvlv ati u!ion,' - Vol 11 , twvil etat ion.) AmFi -TRITI 78, Ai r [or~ce
311. Witkr .CR~nr .. Reliabilitv nlssApraht
Air Imye Ilioht Pt;7amf lorv. W rni-' Wih-atrwrl i orce
Pater'on Ai For'c' Faw-', Ohio 01'133, Ocobr 00
3.Sarhpiv., Cit .Jr.. Watson. R.S.. vawtoofaReibltAnayw ppoat h to Frat ine L.i fe Vraityof Ai rcraft St ruscturesUsn C 8 10 l Serv ice Operatiional1 Pata AFMI- 0-7'.'.' AirFoc
MaeikI abtoratori . Wrih -Pat l' rwl Ar orce Bias;e, Ohio 0P433.k etruari 1'll' I.
34, vpr on Klt iti'11cal Min'sitor lod I wih Specja 1 Panel oftuNAL'A Subcor'ivit te on Ai rcra ft las ue14
hit n .Boulon, i ftrf, K,. ., Development of Struc tural PtViVin cri ter iafrom "tat kikal iIIight Vata. W.\DP R0;O-40% Wrigiht -PattersonAir F orce Saw'i', Oh io' 4433. J1anuar') 1%.&
11i riot ,8 . Cowen , A. i !;outJ on , I .<A*pp! i cat ion of Miii I liparimef i'1F10 load% Va~kta to Structural le~itin Criteria," 4411ll-10R60-131,Vol 1 -11\, Air Iorrc' Ili ht Iinaw4 *. bora I er', Wri'i h I att i r'en
h nn'~n."8 'ir.38 81-on.. S, *. 11;ir~ St suF\ oi t0,ieiindI n-i roniwnont I oeid' C'ni t er iai for 6uiijed Mi %,' i Iev% , WADC 115 67 *. , Wi'i ioht -
'alt! ir'on Air Flowi haw Oiio 44'3. , A-\u%'. 10K.
I18 F Cir . . ýxi~n'iIl.a~i'"r. A. \ . I mdnd' , . '1 .An lnwe\ i gat ion ofI aunth ,truk¾ '1 ncI turi llt-i j CrI it eria for !light Vvhti cli''.'04- !R Yi ti . Wrioht 'it t'n--. Air For Pawi'i Ohio 4,4'33,
Septelib!'r )t,..
AFFDL-TR- 7-,
RI FER[NCLS (CONCLUDED)
39. Shepherd. J.[. . WIttiq, d .W., KadImer, W.L., An Inve:*;tiqL-aijn.ofliUjjght Phase Structural Design triteria for Fl iht Vehicles,
ASP-- TR-61 62WrhtPterson Air E'c sOho45433,June 1961'.
40. Hahn, P.G. , Structural Dsi]n Criteria for StageSeparation,AS)-TR-61-671, Wright-Patterson Air Frorce Baso, Ohio 45433,Januaty 1963.
41. BOuton, I., Uijihara, B.H., Ilight Vehicle Structural DesignCriLeria, ecoveqysPhase, A TT--6.T4.- Wright-Patterson Air Force
Base, Ohio 45433, September 1963.
42. Pugsley, A.G., "The Safety of Structures," Edward Arnold, London,England 1966.
43. Bigham, FR., "The Complementary Roles of Analysis and Testing inAirc aft Structural Substantiation," National Business AircraftMe;ting, Society of Automotive Engineers, March 1970.
44. nones , B. H., Probabilisti" D[esign 1and Re lability, CompositeMaterial s, Structural Design and Analysis, Part II. Academic Pres s,~ev; York, N.Y., 1975.
45. Bording, W.C., Diederich, FW. , Parker, P.S. , "Structural Optimizationand Design Based on a Reliability Design Criterion," AlAA Launchand Space Vehicle Shell Structures Conference, April 1963.
46. Hilton, iI.I. Feigen, M., "Minimum Weight Analysis Based onStructural Reliability,' Third Symposium on High-Speed Aerodynamicsand Str'uctures, UISAF, University of California, Ryan AeronauticalCo., Convair, March 1958.
47. Beeby, A.W., Cranston, W.B., "Influence of Load Systems on Safety,"";Cvi.l _j -enijj: .and ýPLbl i C Works Review, December 1973.
48•. Christensen, C.M., "Designinly for Airline Safety," AerospaceEngineering, September 06P.
49. Stewart, C.B. , "Aircraft Manufacturer's Air Safety Program- TheSST System Safety Program" Air Safety Symposium, Journal of theAir Law and Commerce, Vol 34, 1968.
96