advances in testing and analytical simulation

Boeing 7th Unitized Structures Conf., Nov. 10, 2005 1

Advances in Testing and Analytical Simulation Methodologies to Support

Design and Structural Integrity Assessment of Large Monolithic Parts:

A New Perspective

R. J. Bucci, H. Sklyut, L. Mueller and M. A. JamesAlcoa, Inc.

D. L. BallLockheed-Martin

J. K. DonaldFracture Technology Associates

ASIP 200529 Nov – 1 Dec

Memphis, Tennessee


Acknowledgements:The work has many contributors over the span of 3+ decades

Gary BrayJohn Brockenbrough

Rich BrazillPete BrouwerJohn Dalton

Markus HeinimannMark JamesBill Kuhlman

Alcoa

Lockheed-MartinUSAFA (formerly Alcoa)Fracture Technology Assoc.

Mike KulakLarry Mueller

Mark NewbornBob Schultz Henry Sklyut

Steve WallaceJohn WattonGreg Wilson

Dale BallRalph Bush

Keith Donald


Abstract

The quest for lighter and more affordable airframes has accelerated demand for thicker/wider/shaped alloy products (plate, extrusions, forgings, castings) and manufacturing technologies (e.g., high-speed machining, weld-joining) to grow applications of unitized structure. Capturing the full benefit of these technologies requires that residual stress effects be accounted for in both material characterization and final part design. In the case of thick or shaped metallic products, residual stresses from thermo-mechanical processing can introduce bias and large scatter effects into coupon-based durability and damage tolerance property determinations, which in turn confounds the ensuing transfer to final design.

The presentation will describe efforts of Alcoa and others directed to developing improved fatigue crack growth rate data analysis methods and modeling tools that may be used to account for residual stress effects in testing and analysis. Two fundamental principles will be discussed at this seminar: advancements in fracture toughness and fatigue crack growth rate testing and analysis; and the way forward to account for the residual stress effect(s) in analysis and design of fatigue and fracture critical structures. Case study examples are presented to validate the recommended approach, and the presentation concludes with a vision for virtual design support to large monolithic part applications.


World's largest aluminum die-forgings:A380 Wing Spars (Alloy 7085)

330 in.

A


Outline

� Overview - Residual stress effect in fracture property testing- Toughness- Fatigue crack growth

� Advanced test methods/residual stress correction and the associated sensitivities

- Crack opening/closing methods� Pre- & post-notch displacement � Adjusted Compliance Ratio (ACR) method� Kmax sensitivity concept

- Case Studies:� Alcoa 7085-T7452 Die-Forging Experience � CT vs MT specimen effects� Sampling & location effects

� Process Simulation Tools/Vision for Virtual Design Support- Coupon-to-component transfer example- Sampling effect studies

� Recommendations & Summary


Thick/shaped product evaluations demand caution when residual stresses are involved

� Complete stress relief of thick and/or shaped product forms is seldom achievable

� Test coupons removed from such parts are likely to also contain residual stress- Coupon isolation partially relieves the original stress, and redistributes that remaining- While the isolated coupon residual stress state is generally less severe than that of the

original host, error potential in the ensuing test result can still be significant

� Residual stress can bias material property comparisons- Measured result may not represent the bulk material property- Property estimates can be non-conservative- Data pooling often yields overly conservative property minima (scatter effect)- Review of the literature reveals the problem is widespread

� The problem adversely impacts a number of promising technologies- New stress relief tempers; advanced materials- Net & near net product forms (forgings, extrusions, castings, spray form)- Low cost structure concepts (unitized and/or weld-joined structures)- Material replacement approvals for legacy aircraft and their derivatives


Fracture mechanics based damage tolerance assessment needs � large monolithic parts

� Pedigree material input- "True" property result

� Free of residual stress bias, sampling and/or geometry effects� Repeatable measurement

- Part-to-part- Low test scatter

- Coupon sampling and isolation effects understood

� Coupon-to-component transfer- Role of microstructure and process path understood- Host/part/coupon residual stress states understood- Crack drive solution(s) considering both internal & external forces- Process consistency controlled & understood (part-to-part)


Heat-treatable alloy residual stress profilesare strongly linked to thermal quench practice

Residual stress distribution of Al 7075 cylinder quenched in cold water spray

Post-quench residual stress states are generally compression at the surface and tension in the interior

Radial

Longitudinal

Center of cylinder

Surface of cylinder

Tangential

Distance from center of cylinder, in.

Res

idua

l str

ess,

ksi

σlong

σtan

σrad

CL


Residual stress induced clamping (or opening) can measurably impact crack tip stress intensity factor

Portrayal of common crack tip clamping occurrence associated with compact specimen isolation from an unfully stress-relieved host

(c) Clamping momentdeveloped aftermachining crackstarter slot.

(b) Isolated specimenlongitudinal residualstress distribution.

(a) Specimen locationwithin parent slab.

M

M


A simplified correction practice has beendevised for fracture toughness (Kic) testing

ASTMB-909

∆δ=0

P

P

Measure the specimen height before andafter machining the crack starter notch.

Kic test residual stress correction schematic

∆δ = δ2 − δ1

δ2

Load

Crack opening displacement

∆δ

Analyze test record withnew origin displaced by ∆δ

NewPQ

5% secant offsetline from new origin

δ1

∆ σ = − σ

∆δ

Clampingscenario


Fracture toughness test results frompartially stress relieved product can bemisleading without proper interpretation

Fracture toughness specimen size effect study (7050-T74 and -T7452 hand forging)Results showing interaction of internal stress state and specimen size on measured

fracture toughness from similar 4-in. thick billets, one stress relieved and the other not

20

24

28

32

36

25.8 26.0 26.0

27.528.4

30.9

24.325.7

24.8

1.5 2.0 2.5 1.5 2.0 2.5 1.5 2.0 2.5

Stressrelieved

Non-stressrelieved

Non-stressrelieved,

corrected Kic

Specimen width, W (in.)

W

Kic

, ksi

(in)1/

2(S

-L, T

/2)


The correction practice reverses when the notch machining results in crack opening

ASTMB-909

∆δ

∆σ = +σ

Measure the specimen height before andafter machining the crack starter notch.

Kic test residual stress correction schematic

∆δ = δ2 − δ1

δ2

Load

Crack opening displacement∆δ

Analyze test record withnew origin displaced by ∆δ

NewPQ

5% secant offsetline from new origin

δ1

∆δ=0

P

P


W

H

D-015365-RK

Location for Before and After Meas.

Alcoa 7085 Forging Green Letter Re-visited� Toughness data multiple shapes, multiple lots re-evaluated

per ASTM B=909� CT specimens, most all excised from web locations

Test Lot NotchReference Part Number S-Number Location Orientation W: H: B: Before After ∆ (After-Before) Change

MT# 20031015-025 Engine Support A8FH47 804628 L-T 2.0 2.4 0.25 2.4002 2.4005 0.0003 OpenedT-L 2.0 2.4 0.25 2.4001 2.4005 0.0004 Opened

Trunion A8FH50 804631 L-T 2.0 2.4 0.25 2.4002 2.3997 -0.0005 ClosedT-L 2.0 2.4 0.25 2.4004 2.3996 -0.0008 Closed

Terminal Fitting A8FH55 804639 Web L-T 3.0 3.6 0.25 3.5993 3.5996 0.0003 OpenedT-L 3.0 3.6 0.25 3.5993 3.5996 0.0003 Opened

A8FH56 804640 Web L-T 3.0 3.6 0.25 3.5983 3.5970 -0.0013 ClosedT-L 3.0 3.6 0.25 3.5992 3.5993 0.0001 Opened

A8FH57 804641 Rail L-T 3.0 3.6 0.25 3.5987 3.5988 0.0001 OpenedT-L 3.0 3.6 0.25 3.5989 3.5992 0.0003 Opened

MT# 20041110-210 Landing Gear Beam A8FZ17R1 805850 Web L-T 3.0 3.6 0.25 3.6008 3.5994 -0.0014 ClosedL-S 3.0 3.6 0.25 3.6009 3.6004 -0.0005 Closed

A8FZ19R1 805852 L-T 3.0 3.6 0.25 3.6007 3.6022 0.0015 OpenedT-L 3.0 3.6 0.25 3.6006 3.6014 0.0008 Opened

MT# 20031218-017 A8FZ18R1 805851 Web L-T-1 3.0 3.6 0.25 3.6001 3.6007 0.0006 OpenedL-T-2 3.0 3.6 0.25 3.6001 3.6009 0.0008 Opened

T-L-1 3.0 3.6 0.25 3.6006 3.6018 0.0012 OpenedT-L-2 3.0 3.6 0.25 3.6006 3.6015 0.0009 Opened

7085-T7452 Die Forging FCG EvaluationD-015365-RK C(T) Specimen H Dimension Measurements (Before and After Slot Fabrication)

H Dimension (in.)C(T) Specimen

7085-T7452 Die-Forging Toughness InterrogationC(T) Specimen H Dimension Measurements (Before and After Slot Machining)


Alcoa 7085 Green Letter Toughness Data Revisited7085-T7452 Die-Forging Kq Residual Stress Correction (ASTM B-909)

Toughness correction based on C(T) specimen height change after slot introduction

-4

-2

0

2

4

6

8

10

12

-0.004 -0.002 0 0.002 0.004 0.006 0.008 0.01

Cor

rect

edK

q–

Test

ed K

q(k

si-in

1/2 )

Landing Gear BeamEngine SupportTerminal FittingSide StayTrunion

∆δ = (δ2 � δ1), in

OpeningClosing L-T test orientation


Alcoa 7085 Forging Green Letter Toughness Data Revisited:Coupon size relative to host residual stress profile & failure mode can bias the result

Kq Toughness w/ & w/o Res. Stress Correction per (ASTM B-909)

20

25

30

35

40

45

50

55

60

65

0 2 4 6 8 10 12 14

Nominal Forged Section Thickness (in.)

Kq

(ksi

in1/

2 )

Original KqCorrected Kq

Greater CorrectionExcised spec. intercepts greater portion of host total res. stress profile

Higher KqNew failure mode, Delam. Toughening?

7085-T7452 Die-Forge Landing Gear Beam (CT Specimen, L-T orientation)


Residual stress influence can be greater in narrow products

� Machining distortion is the result of internal stress rebalance associated with metal removal

� A coupon's size relative to the principal dimensions of it's host determines the fraction of host residual stress profile intercepted

� It is thus reasonable to expect that residual stress bias will be greater from coupons having principal dimensions comparable to those of the host

MM

Boeing 7th Unitized Structures Conf., Nov. 10, 2005

FCGR correction for residual stress bias can be ascertained from the load-COD relationship

Load - P

∆Keff

Pop

Pmax

Pop

Pop

∆Keff ∆Keff

Effect of residual stress on load-COD traceCrack opening displacement - v

(a) (b) (c)Pop = crack openingload attributed to crackclosure effect (Elber).

FCG occurs for P > Pop

(a) Residual stress free.Pop linked to cracksurface roughness.

(b) Pop increasewith residualstress inducedclamping moment.

(c) Pop decreasewith residualstress inducedopening moment.

P

v


Precedent has been set for use of "closure-based" correction to remove residual stress bias in FCGR data

FCGR data from two partially stress relieved 7050-T7452 forgings(S-L orientation, R = 0.33, high humidity air)

10 -9

10 -8

10 -7

10 -6

10 -5

10 -4

10 -3

1 10

Specimen #1

Specimen #2

10 -10

10 -9

10 -8

10 -7

10 -6

10 -5

205

5 10 20

Uncorrected

∆K, ksi(in)1/2

da/d

N, i

n/cy

cle

∆K, Mpa(m)1/2

da/d

N, m

/cyc

le

10 -9

10 -8

10 -7

10 -6

10 -5

10 -4

10 -3

1 10

Specimen #1

Specimen #2

10 -10

10 -9

10 -8

10 -7

10 -6

10 -5

205

5 10 20

Corrected

∆Keff, ksi(in)1/2

da/d

N, m

/cyc

le

∆Keff, Mpa(m)1/2

da/d

N, i

n/cy

cle

(Bucci, ASTM STP 743, 1981 & Bush et. al., ASTM STP 1189, 1993)


Recent advances in closure correction methodology

� Inaccurate accounting of closure and the residual stress effect results in erroneous life prediction

- Important in near-threshold region where majority of structural life is spent- Important to reconciling coupon-to-component transfer issues, particularly

for thick, complex shape parts

� The current ASTM E647 closure measurement approach has deficits- Overly conservative, particularly ∆Keff thresholds (high weight penalty)- Mixed agreement against small flaw and variable R-ratio data sets- High measurement variability

� The adjusted compliance ratio (ACR) and Kmax sensitivity methods offer significant interpretive advantages over current ASTM-E647 ∆Kapplied and ∆Keff based practices

- Better threshold and near-threshold FCGR estimates - Better normalization of R-ratio effect- Better agreement between large and small flaw data- Better accounting of residual stress effects


Schematic of ASTM Opening Load Method

� Straight-line fit to upper linear portion of the curve to determine open-crack compliance

� Pop defined as the point where the slope deviates 2% from the fully open compliance

� Assumes only that portion of the load cycle between Pop and Pmaxcontributes to FCG

∆Keff = Kmax - Kop

Conceptually appealing but often problematic in practice� Kop depends significantly on measurement location and distance from

the crack tip and exhibits significant scatter for given location� Often poor correlation between ∆Keff and observed crack growth rates

∆Peff (ASTM)


Schematic of Adjusted Compliance Ratio (ACR) Method

� Developed by Keith Donald (FTA)� Experimental setup, instrumentation & data collection same as ASTM E647

� Pop marks a transition point below which applied force is no longer directly proportional to crack tip strain

� ACR method assumes there can still be a contribution to ∆Keff below Pop

� ∆Keff is related to the actual displacement range (δcl) to the displacement range that would have occurred in the absence of crack closure (δnc)

� ∆Keff = ACR · ∆Kapp

io

is

CCCCACR

−−

=CR = Cs / Co

∆Peff (ASTM)


Comparison of ∆Keff Curves Based on theACR and ASTM Opening Load Method

The ASTM opening load method typically yields lower ∆Keff thresholds and higher near-threshold FCG rates than the ACR method.

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100

∆K (MPa√m)

da/d

n (m

m/c

ycle

)

∆Kapplied

∆Keff - ACR

∆Keff - ASTM

Specimen = Middle Crack TensionW = 101.6 mm; B - 3.175 mmR = 0.10; Frequency = 25 Hz2ao = 5.08 mm

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100

∆K (MPa√m)

da/d

n (m

m/c

ycle

)

∆Kapplied

∆Keff - ACR

∆Keff - ASTM

Specimen = Middle Crack TensionW = 101.6 mm; B - 3.175 mmR = 0.10; Frequency = 25 Hz2ao = 5.08 mm

2324-T397075-T651

∆Kap

p

∆Kef

f (A

CR

)∆Keff

(AST

M)

∆Kap

p

∆Kef

f (A

CR

)∆Keff

(AST

M)


FCGR Response of Alloy 2324 Plate at R = -1.0, 0.1, 0.3, 0.5, 0.7

Ref. Bray & Donald, ASTM STP 1343, 1999

∆Kappl basis ∆Keff basis (ASTM)

∆Keff basis (ACR)

ACR method does best job of normalizing the R-ratio effect


Kmax sensitivity concept addresses the R-ratio shift associated with superposition of residual and applied forces


� Figures to the right show:Plate alloy 2324 FCGR response atR = -1.0, 0.1, 0.3, 0.5, 0.7 & const. Kmax= 6, 9, 14 & 22 ksi√in

� Best normalized response uses ∆Keff based on ACR method

Norm. Kmax sensitivity concept

Knorm = ∆Keff(1-n) · Kmax

n

� At n = 0, Knorm = ∆Keff; indicating that FCG rates depend only on ∆Keff

� At n = 1, Knorm = Kmax; indicating that FCG rates depend only on Kmax


∆Kappl basis

Knorm basis


ACR-Corrected Long Crack Data Matches Very Well with Small Corner Flaw Data

7075

long

& sm

all cr

ack data

2324

long

& sm

all cr

ack data

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100

∆K (MPa√m)

da/d

n (m

m/c

ycle

)

7075-T7651 ∆Keff - ACR M(T)

7075-T7651 ∆Kapplied - Corner Crack

2324-T39 ∆Keff - ACR M(T)

2324-T39 ∆Kapplied - Corner Crack

Middle Crack TensionW = 101.6 mm; B = 3.175 mmR = 0.10; Frequency = 25 Hz2ao = 5.08 mm

Open Hole Corner NotchedW = 38.1 mm; B= 3.175 mmHole Diameter = 6.35 mmR = 0.10; Frequency = 20 HzSmax - gross = 82.7 MPa

∆K (MPa√m)

da/d

N(m

m/c

ycle

)

Figure to right shows:

� Closure corrected long crack data (ACR method) matches the small corner flaw test result

� The small corner flaw data is presumed closure free in the purest sense (little crack wake to develop plasticity or roughness induced closure)


Representative 7085-T7452 Forging Fatigue Crack Growth Evaluation Study

Objective

� Evaluate sample and location effects of C(T) and M(T) specimens removed from a representative 7085-T7452 die-forging

� Evaluate various ∆Keff normalizing methods for removing residual stress bias from the data measurement

- Closure correction (ASTM opening load & ACR methods)

- Kmax sensitivity concept


7085-T7452 Die-Forging Test Program

Forged Part Evaluated:


Overview of Test Program

� Material: - 7085-T7452 Main landing gear beam die-forging- Sampling locations: web and rail

� Specimen type:- C(T), W = 3 in- M(T), W = 4 in

� Specimen orientation: L-T, T-L, S-L� Stress ratio, R=0.1; Freq = 25 hz� High humidity air (> 90% RH)� Crack length measurement: Compliance� Crack closure measurement: Compliance

- ASTM opening load (2% offset)- Adjusted compliance ratio (ACR)


C(T) and M(T) Specimen Sampling Locations

Sampling Locations for 7085-T7452 Die Forgings;GA319523 - Main Landing Gear Beam

Landing Gear Beam Web (S-No.805852)

(X-Sect. View)

Landing Gear Beam Rail (805850)

Landing Gear Beam Web (805852)L-

T (t/

2) &

L-S

(t/4

); C

TsL-

T (t/

2) &

L-S

(t/4

); M

Ts

Landing Gear Beam Rail (805850)


Salient findings - C(T) vs M(T) coupon FCGR response with residual stress present

� A unique da/dN-DK relationship is basic to fracture mechanics based life prediction (similitude concept)

� CT and MT specimens excised from the same host location often display notch-tip internal stress states of opposing sign (ref. prior chart)

- CT notch-tip compression (clamps); MT notch-tip tension (opens), or vice-versa- FCGR inconsistencies are often missed or mistakenly labeled as a geometry effect

� A residual stress profile resulting in notch-tip compression (clamping) introduces an additive closure effect into ensuing crack growth measurement

- Closure based ∆Keff approaches are useful for removing residual stress bias - ACR method the preferred closure correction approach

� A residual stress profile producing notch-tip tension (opening) increases cyclic R-ratio and Kmax sensitivity

- Useful approach to normalize Kmax sensitivity: Knorm = ∆Keff(1-n) . Kmax

n


7085-T7452 Die-Forging FCG Evaluation

0.0015 (Opened)0.0008 (Opened)

3.60223.6014

3.60073.6006

L-TT-L

805852(Web)

A8FZ19R1

-0.0014 (Closed)-0.0005 (Closed)

3.59943.6004

3.60083.6009

L-TL-S

805850(Rail)

A8FZ17R1

∆ (After - Before)AfterBefore

H Dimension (in.)OrientationS-NumberLot Number

C(T) Specimen H Dimension Measurements (Before & After Slot Fabrication)

The C(T) specimen is a sensitive residual stress indicator, whereas the M(T) specimen is not.


7085-T7452 Die Forgings: Fatigue Crack GrowthR=0.1, Frequency = 25 Hz., High Humidity (RH>90%), L-T, C(T)

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1 10 100 K (ksi √in)

da/d

N (i

n/cy

cle)

S-805850 L-T, C(T)

S-805850 L-T, C(T), DKeff (ACR)

S-805852 L-T, C(T)


S-805850 = RailS-805852 = Web

∆K (ksi√in)

Rail (L-T)CT clamps

Web (L-T)CT opens

� Large closure effect when CT clamps

� Closure free when CT opens� DKeff (ACR) gives best

agreement at low & mid ∆K� Kmax sensitivity when CT

opens


7085-T7452 Landing Gear Beam (Rail) Die Forgings: Fatigue Crack GrowthR=0.1, Frequency = 25 Hz., High Humidity (RH>90%), L-T Orientation, 4" M(T) vs. C(T)

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1 10 100-K (ksi √in)

da/d

N (i

n/cy

cle)

S-805850 L-T, M(T)

S-805850 L-T, M(T), DKeff (ACR)

S-805850 L-T, C(T)


S-805850 = Rail

∆K (ksi√in)

Rail (L-T)CT clamps

Rail (L-T)MT opens

� Large closure effect when CT clamps

� Closure free when MT opens� DKeff (ACR) gives best

agreement at low & mid ∆K� Kmax sensitivity when MT

opens


7085-T7452 Landing Gear Beam (Web) Die Forging: Fatigue Crack GrowthR=0.1, Frequency = 25 Hz., High Humidity (RH>90%), L-T Orientation, 4" M(T) vs. C(T)

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1 10 100-K (ksi √in)

da/d

N (i

n/cy

cle)

S-805852 L-T, M(T)


S-805852 L-T, C(T)


S-805852 = Web

∆K (ksi√in)

Web (L-T)MT clamps

Web (L-T)CT opens

� Large closure effect when MT clamps

� Closure free when CT opens� DKeff (ACR) gives best

agreement� Slight Kmax sensitivity

when CT opens


7085-T7452 Landing Gear Beam (Rail & Web) Die Forgings: Fatigue Crack GrowthR=0.1, Frequency = 25 Hz., High Humidity (RH>90%), L-T, 4" M(T)

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1 10 100-K (ksi √in)

da/d

N (i

n/cy

cle)

S-805850 L-T, M(T)


S-805852 L-T, M(T)


S-805850 = RailS-805852 = Web

∆K (ksi√in)

Web (L-T)MT clamps

Rail (L-T)MT opens

� Large closure effect when MT clamps

� Closure free when MT opens� DKeff (ACR) gives best

agreement at low & mid ∆K� Kmax sensitivity when MT

opens


7085-T7452 Die Forgings: Fatigue Crack GrowthR=0.1, Frequency = 25 Hz., High Humidity (RH>90%), 4" M(T) vs. C(T)

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1 10 100-K (ksi √in)

da/d

N (i

n/cy

cle)





∆K (ksi√in)

All L-T Data (ACR)� DKeff (ACR) collapses data

for all conditions tested - Web & Rail- CT & MT

� Some Kmax sensitivity at high DK (opening case)


Alcoa 7085 Forging Green Letter Re-visited- Multiple shapes, multiple lots re-evaluated w/ACR method- CT specimens, most all excised from web area- Findings consistent with prior study results

Test Lot NotchReference Part Number S-Number Location Orientation W: H: B: Before After ∆ (After-Before) Change

MT# 20031015-025 Engine Support A8FH47 804628 L-T 2.0 2.4 0.25 2.4002 2.4005 0.0003 OpenedT-L 2.0 2.4 0.25 2.4001 2.4005 0.0004 Opened

Trunion A8FH50 804631 L-T 2.0 2.4 0.25 2.4002 2.3997 -0.0005 ClosedT-L 2.0 2.4 0.25 2.4004 2.3996 -0.0008 Closed

Terminal Fitting A8FH55 804639 Web L-T 3.0 3.6 0.25 3.5993 3.5996 0.0003 OpenedT-L 3.0 3.6 0.25 3.5993 3.5996 0.0003 Opened

A8FH56 804640 Web L-T 3.0 3.6 0.25 3.5983 3.5970 -0.0013 ClosedT-L 3.0 3.6 0.25 3.5992 3.5993 0.0001 Opened

A8FH57 804641 Rail L-T 3.0 3.6 0.25 3.5987 3.5988 0.0001 OpenedT-L 3.0 3.6 0.25 3.5989 3.5992 0.0003 Opened

MT# 20041110-210 Landing Gear Beam A8FZ17R1 805850 Web L-T 3.0 3.6 0.25 3.6008 3.5994 -0.0014 ClosedL-S 3.0 3.6 0.25 3.6009 3.6004 -0.0005 Closed

A8FZ19R1 805852 L-T 3.0 3.6 0.25 3.6007 3.6022 0.0015 OpenedT-L 3.0 3.6 0.25 3.6006 3.6014 0.0008 Opened

MT# 20031218-017 A8FZ18R1 805851 Web L-T-1 3.0 3.6 0.25 3.6001 3.6007 0.0006 OpenedL-T-2 3.0 3.6 0.25 3.6001 3.6009 0.0008 Opened

T-L-1 3.0 3.6 0.25 3.6006 3.6018 0.0012 OpenedT-L-2 3.0 3.6 0.25 3.6006 3.6015 0.0009 Opened

7085-T7452 Die Forging FCG EvaluationD-015365-RK C(T) Specimen H Dimension Measurements (Before and After Slot Fabrication)

H Dimension (in.)C(T) Specimen

W

H

D-015365-RK

Location for Before and After Meas.


7085-T7452 Die Forging Fatigue Crack Growth Curves for Terminal Fitting, Trunion, Engine Support, and Landing Gear Beam, L-T Orientation: ∆KApplied

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100∆K (ksi √in)

da/d

N (i

n/cy

cle)

A8FH55, Terminal Fitting L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0003)A8FH56, Terminal Fitting L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Closed (-0.0013)A8FH57; Terminal Fitting L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0001)A8FH50; Trunion L-T, Applied; C(T) @ H=2.4, W=2.0, B=0.25, Closed (-0.0005)A8FH47; Engine Support L-T, Applied; C(T) @ H=2.4, W=2.0, B=0.25, Opened (0.0003)A8FZ18R1; Landing Gear L-T1, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0006)A8FZ18R1; Landing Gear L-T2, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0008)A8FZ17R1; Landing Gear L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Closed (-0.0014)A8FZ19R1; Landing Gear L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0015)

R = +0.1Freq = 25HzLab AirL-TK-gradC(T) Specimen

Highestopen

Highestclose

∆KApplied

∆Kapp, ksi√in


7085-T7452 Die Forging Fatigue Crack Growth Curves for Terminal Fitting, Trunion, Engine Support, and Landing Gear Beam, L-T Orientation: ∆KACR

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100∆K (ksi √in)

da/d

N (i

n/cy

cle)

A8FH55, Terminal Fitting L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0003)A8FH56, Terminal Fitting L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Closed (-0.0013)A8FH57; Terminal Fitting L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0001)A8FH50; Trunion L-T, Applied; C(T) @ H=2.4, W=2.0, B=0.25, Closed (-0.0005)A8FH47; Engine Support L-T, Applied; C(T) @ H=2.4, W=2.0, B=0.25, Opened (0.0003)A8FZ18R1; Landing Gear L-T1, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0006)A8FZ18R1; Landing Gear L-T2, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0008)A8FZ17R1; Landing Gear L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Closed (-0.0014)A8FZ19R1; Landing Gear L-T, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0015)

R = +0.1Freq = 25HzLab AirL-TK-gradC(T) Specimen

Highestopen

∆Keff(ACR)

∆KACR collapses data when notch clamped; Kmax sensitivity adjustment needed for case when notch opened

∆KeffACR, ksi√in


7085-T7452 Die Forging Fatigue Crack Growth Curves for Terminal Fitting, Trunion, Engine Support, and Landing Gear Beam, T-L Orientation: ∆KApplied

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100∆K (ksi √in)

da/d

N (i

n/cy

cle)

A8FH55; Terminal Fitting T-L, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0003)A8FH56; Terminal Fitting T-L, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0001)A8FH57; Terminal Fitting T-L, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0003)A8FH50; Trunion T-L, Appled; C(T) @ H=2.4, W=2.0, B=0.25, Closed (-0.0008)A8FH47; Engine Support T-L, Applied; C(T) @ H=2.4, W=2.0, B=0.25, Opened (0.0004)A8FZ18R1; Landing Gear T-L1, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0012)A8FZ18R1; Landing Gear T-L2, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0009)A8FZ19R1; Landing Gear T-L, Applied; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0008)

R = +0.1Freq = 25HzLab AirT-LK-gradC(T) Specimen Highest

open

Highestclose

∆KApplied

∆Kapp, ksi√in


7085-T7452 Die Forging Fatigue Crack Growth Curves for Terminal Fitting, Trunion, Engine Support, and Landing Gear Beam, T-L Orientation: ∆KACR

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1 10 100∆K (ksi √in)

da/d

N (i

n/cy

cle)

A8FH55; Terminal Fitting T-L, ACR; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0003)A8FH56; Terminal Fitting T-L, ACR; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0001)A8FH57; Terminal Fitting T-L, ACR; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0003)A8FH50; Trunion T-L, ACR; C(T) @ H=2.4, W=2.0, B=0.25, Closed (-0.0008)A8FH47; Engine Support T-L, ACR; C(T) @ H=2.4, W=2.0, B=0.25, Opened (0.0004)A8FZ18R1; Landing Gear T-L1, ACR; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0012)A8FZ18R1; Landing Gear T-L2, ACR; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0009)A8FZ19R1; Landing Gear T-L, ACR; C(T) @ H=3.6, W=3.0, B=0.25, Opened (0.0008)

R = +0.1Freq = 25HzLab AirT-LK-gradC(T) Specimen

Highestopen

∆Keff(ACR)

∆KACR collapses data when notch clamped; Kmax sensitivity adjustment needed for case when notch opened

∆KeffACR, ksi√in


Crack Growth Data Reduction for DesignSchematic of First Tier Validation Approach(ref. D.L. Ball, Boeing 7th Unitized Structure Conf., 2005)

Coupon Validation


Key Findings and Recommendations

� Residual stress can be responsible for large scatter in fracture property measurements developed from thick/large monolithic parts

� Guidelines and path presented to minimize the testing and interpretive problems- Warning signs and validity checks to recognize corrupted data sets- Initial proof-of-concept validated, and the technology ready for scaling to the next higher

(design feature and sub-component) level- Strong advocacy is needed for upgrading test/evaluation standards

� Corrective protocols established to purge residual stress bias from property data- The residual stress impact on specimen crack drive can be treated as an additive (or

subtractive) K-value capable of producing the equivalent residual stress induced opening (or closing) COD beyond the or residual stress free "neutral" (zero load) state.

- The concept of a normalized or "master" da/dN-∆Keff FCGR relationship can be used to untangle the residual stress effect

� da/dN-∆Keff data with and without residual stress bias fit the same normalized FCGR relationship� Best normalization results with DKeff basis the ACR method (handles crack clamping case)� Further normalization improvement with Kmax sensitivity adjustment (handles crack opening case)� A closure-based FCG model (e.g., FASTRAN) can be used to reverse-derive a family "true"

material da/dN-∆K curves (free of residual stress bias) for individual R-values, which in turn can be used to predict cyclic crack growth life in the usual way


Alcoa Internal Building Block R&DResidual Stress/Machining Distortion Tool Development;

Coupon-to-Component Transfer Know-how

� Objective: Deliver basic know-how, software, analytical methods to understand/quantify relationships between pre- and post-machining residual stresses in host product, final machined parts, test specimens and the associated impacts on material, manufacturing, and structural performance evaluations

� 2004/2005 Major software tool developments- Thermo-mechanical process and residual stress simulation- MDT (Machining Distortion Tool)- BARS-SP (Bi-axial Residual Stress � Specimen Placement)

� Library of standard specimen configurations � Ability to place a surface, corner or through crack into an arbitrary 3D

body and/or sub-structure detail- Analytical study of residual stress effects - Tool validation against actual measurement - in process


Alcoa Vision � Virtual Design Support for Monolithic Part Applications

Arbitrary 3D Body

BLOCKY

� Host Geometry� Quenching Simulation &

Stress Relief� Residual Stress Prediction � Microstructure Prediction � Machining Distortion

� Residual Stress in Finished Part� Microstructure in Finished Part

MDT (Part Isolation) MDT (Coup Isolation)

Effect of Microstructure & Residual Stress

K1c Correction Tool da/dN Correction Tool

S/N Correction Tool

Microstructure to Coupon Understanding

Coupon to Structure Understanding

Residual Stress Corrected Material Properties

3-D Geometry Microstructure Based Strength allowable

MDT(Machining Distortion Tool)

Design & Certification Forged Aerostructures

Process SimulationMachining Simulation Coupon Isolation &

Testing Simulation

GeometryResidual StressLocal Microstructure

Customer Support & Application Engineering

Customer Final

Geometry

3-D S/N Fatigue Small Flaw Model

Location Dependent S/N Allowable

CrackCrack Location Dependent Damage Tolerance Allowables

Customer Loads, Requirements, and

Boundary Conditions


7xxx Aluminum Die-Forgings – Old vs. New Process

Old Way

7050-O1 Quenched Forging

Rough Machine to 2-4in

Heat Treat/Quench/Age to -T74 Temper (No CW)

Intermediate Machine (flipping, resting, etc.)

Final Machine Part

Residual Stressesin Double Digits (ksi)

Residual stress & manufacturing cost improvement breakthroughs enabled via advance FEM Modeling

New Way

7085-T7452 ColdWorked Forging(at final temper)

Intermediate Machine(minimal flipping)

Final Machine Part

Residual Stressesin Single Digits (ksi)


Alcoa MDT Tool with Automated Crack Placement Feature

Part Geometry

Redistributed σxx

After MachiningDistortion Analysis

Output:- Stress- Crack Drive- Distortion

As Forged

Crack

Positioning of Crack

After MachiningDistortion Analysis


Vision for Virtual Design Support - Forging Example

Residual Stress - σ x (MPa) Residual Stress - σ z (MPa)Residual Stress - σ y (MPa)

Host bodyHost bodyForgingForging

x

y

z

Residual Stress Profile in Host Body


Machining Part Final Geometry from the Host - Position 1

Final Mach. Part

Host Forging


Machining Part Final Geometry from the Host - Position 2

Final Mach. Part

Host Forging


Residual Stress in Part Final Geometry

x

y

z

HostForging

RedistributedHost Pos. 1

Mapped

σx(MPa)

σz(MPa)

σy(MPa)

Host Pos. 2 Host Pos. 1 Host Pos. 2


Distortion Vector (mm) in Part Final GeometryHost Position 1 Host Position 2


We also want stress redistribution, distortion and K solution for cracking in the final machined part and in

test coupon isolated from said part

Crack

Specimen


Example C(T) toughness specimen isolation problem Read in final part geometry into BARS-SP tool. Select C(T) specimen and orientation from tool bar and position in final part detail.

Z

Y

X

C(T) SpecimenB = 0.12", W = 1"


Proper data interpretation requires that specimen sampling effect on residual stress induced crack drive be understood

Host Position 1

Host Position 2

0

0.02

0.04

0.06

0.08

0.10

0.12

0 0.1 0.2 0.3 0.4Residual stress induced K1, ksi-in1/2

Spec

imen

Thi

ckne

ss L

ocat

ion,

inHost Pos. 1Host Pos. 2

C(T) specimen crack drive as function of final machined part location within original host

C(T) spec., W= 1"distortion vector

Note: The relative smallness of the residual stress induced crack drive shown is attributed to smallness of the C(T) specimen (W = 1") relative to the part rib height. It is shown later that a larger W dimension would have resulted in a larger residual K.


FEA Simulation of CT Specimen Fracture Toughness TestResult shows predicted stress-strain condition in W=1" CT spec. with COD = 0.106 in

Host Position 1 Host Position 2


0

200

400

600

800

1000

1200

-0.01 0.05 0.11 0.17 0.23 0.29 0.35 0.41 0.47 0.53

COD (mm)

Load (Pascal)

FEA simulated load-COD diagram with & without KQ correction for residual stress; C(T) specimen (W = 1 in., B = 0.12 in.)

KQ5% (w/o corr.)= 24.8 ksi√in

KQ5% (w/o corr.) = 25.9√in

KQ5% (w/corr.) = 25.4 ksi√inKQ5% (w/corr.)= 25.4 ksi√in

Neutral notch

position Positi

on 2; Notch

Clamped

Position 1; N

otch Opened

Both tests yield identical KQ result after correction for residual stress bias

Note: the relative smallness of the residual stress correction shown is attributed to smallness of the C(T) specimen width (W = 1") relative to the part rib height. It is shown later that the needed K correction is greater for a larger W specimen.


Example: Sampling and location effectSpecimen Size & Notch Orientation

Specimen:C(T), L-T, B = 0.12",Host Pos. 1, Mid-Rib

W = 1"

W = 2"

W = 1"

W = 2"

Goal: Apply BARS-SP tool to demonstrate specimen residual K dependency on post-machining stress rebalance


Specimen size & notch orientation effect on residual stress induced crack drive

Notch orientation can also impact stress rebalance and the residual K as well.

0

0.02

0.04

0.06

0.08

0.10

0.12

-4.0 -3.2 -2.4 -1.6 -0.8 0 0.8 1.6 2.4 3.2

Spec

imen

thic

knes

s lo

catio

n, in

Residual stress induced crack drive, K1, ksi√in

W = 1"W = 2" W = 2"W = 1"

Notch clamping Notch opening

C(T) Spec., Mid-RibL-T, B = 0.12 in

Host Pos. 1


Example: Sampling and location effectPlacement of Final Part & Specimen within Host Machining Envelope

Specimen:C(T), L-T, Mid-RibW = 1", B = 0.12"

Host Pos. 2

Host Pos. 1Goal: Apply BARS-SP tool to demonstrate specimen residual K dependency on post-machining stress rebalance


Machined part placement within the original host also impacts the residual stress effect


Host Pos. 1 Host Pos. 1Host Pos. 2

Host Pos. 2

0

0.02

0.04

0.06

0.08

0.1

0.12

-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1.0

C(T) spec., Mid-RibL-T, B = 0.12 in, W = 1 in

Spec

imen

thic

knes

s lo

catio

n, in

Residual stress induced crack drive, K1, ksi√in



BARS-SP study to assess M(T) specimen type and effect of sample location on residual stress induced crack drive

Loc.#1

Loc.#2

Loc.#4

Loc.#3

Loc.#5

Dimensions in cmDimensions in cmMachined part excised from Pos# 1 and Pos# 2 within original host


M(T) specimen sampling studyBARS-SP prediction showing variation of residual stress induced crackdrive with part position (within host) and coupon location (within part)

Loc.#1

Loc.#2

Loc.#4

Loc.#3

Loc.#5

Loc.#1

Loc.#2

Loc.#4

Loc.#3

Loc.#5

Loc.#1

Loc.#2

Loc.#4

Loc.#3

Loc.#5

0

0.05

0.10

0.15

0.2

-3.0 -2.0 -1.0 0 1.0 2.0

K1, ksi-in1/2

Web

Thk

., in

Host Pos#1 Right CrackHost Pos#2 Right CrackHost Pos#1 Left CrackHost Pos#2 Left Crack

Location #1

0

0.05

0.10

0.15

0.2

-3.0 -2.0 -1.0 0 1.0 2.0

K1, ksi-in1/2

Web

Thk

., in



Location #1

0

0.05

0.10

0.15

0.20

-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8

K1, ksi-in1/2

Web

Thk

., in


Location #2

0

0.05

0.10

0.15

0.20

-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8

K1, ksi-in1/2

Web

Thk

., in



Location #2

0

0.05

0.10

0.15

0.20

-4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5

K1, ksi-in1/2

Web

Thk

., in


Location #3

0

0.05

0.10

0.15

0.20

-4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5

K1, ksi-in1/2

Web

Thk

., in



Location #3

0

0.05

0.10

0.15

0.20

-4.0 -3.0 -2.0 -1.0 0 1.0 2.0 3.0

K1, ksi-in1/2

Rai

l Thk

., inLocation #4


0

0.05

0.10

0.15

0.20

-4.0 -3.0 -2.0 -1.0 0 1.0 2.0 3.0

K1, ksi-in1/2

Rai

l Thk

., inLocation #4



0

0.05

0.10

0.15

0.20

-4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5

K1, ksi-in1/2

Web

Thk

., inLocation #5


0

0.05

0.10

0.15

0.20

-4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5

K1, ksi-in1/2

Web

Thk

., inLocation #5




Example: BARS-SP ToolSIF calculation for crack growth in final machined part

� Redistribution of residual stress (σy), distortion vector, and crack drive solution with crack propagation in final machined part

� Sensitivity to part placement in original host (Host Position# 1 & 2)

Crack

Deformed

Mapped


Host Position #1

Redistributed Residual Stresses σy &

Crack propagation, step#1.

Distortion vector (deformation scale = 40).

K1(max value) diagramdue to Residual Stress

-2000

0

2000

4000

6000

0 2 4 6 8

Crack Length, in

K1, Psi*in 0̂.5


Host Position #1





-2000

0

2000

4000

6000

0 2 4 6 8

Crack Length, in

K1, Psi*in 0̂.5


Host Position #2





-13000

-11000

-9000

-7000

-5000

-3000

-1000 0 2 4 6 8Crack Length, in

K1, Psi*in 0̂.5


Host Position #2





-13000

-11000

-9000

-7000

-5000

-3000


K1, Psi*in 0̂.5


6.136.49

6.02

3.21

-1.8

-9.61

-11.1

-8.49

-2.86

-12.3

-15

-10

-5

0

5

10

0 1 2 3 4 5 6 7 8 9

Crack Length, in

K1,

ksi

-in1/

2 Host Pos. 1

Host Pos. 2

Summary:Max. value of residual stress induced K1 vs. crack length

Crack

Deformed

Mapped

Crack

Deformed

Mapped


Summary

� Accounting for residual stress effects in damage tolerance evaluations will be essential to thick monolithic part design and approval

- Failure to account for closure and residual stresses accurately will result in erroneous conclusions and enormous inefficiencies in design, analysis and certification processes

- Even stress relieved product forms can be greatly affected

� Path established to minimize testing & data interpretation problems- Warning signs and validity checks available to identify potential for data corruption- Test/analysis methodology established to purge residual stress bias from property data

� Experimental protocol established and automated � Kic and da/dN data reduction and analysis practices established� Correction practice has basis in crack opening/closing methodologies

- A significant body of data has been generated to support initial proof-of-concept

� Coupon-to-component transfer path emerging- The technology is ready for scaling to the design feature and sub-component levels- Advance process simulation tools offer valuable new insights for characterization, analysis

and design of large monolithic parts- Strong advocacy is needed for upgrading industry test/evaluation standards & data bases

The Aeroelastic Design and Testing of the F/A-22

by William D. Anderson Lockheed Martin Aeronautics Company

Marietta, Georgia

Presented at the 2005 USAF Aircraft Structural Integrity Program Conference

Memphis, Tennessee29 November - 1 December 2005

© 2005 by Lockheed Martin Corporation. Lockheed Martin Aeronautics Company

229 Nov – 01 Dec, 2005 USAF ASIP Conference Lockheed Martin Aeronautics Company

Presentation Topics

• Flutter Development, Analysis Scope and Procedures• Flutter and Aeroelastic Design Criteria• Analysis Methods • Aeroelastic Tailoring Analysis Approach• Flutter Issues with Early F/A-22 Design

– Initial Modes of Concern– Initial Trades / Aeroelastic Design Optimization Results

• Design for Transonic Buzz and LCO• Flutter Model W/T Results / Issues• Aeroelastic Design Impacts• Verification, Validation and Certification

– Lab, Ground, and Flight Testing– Certification Analysis and Documentation


Overview of the F/A-22 Flutter Development

• Analysis– Doublet

Lattice– Zona 51

• L3 Press. Model• Flutter Model

• Stiffness Matrix

• Delta Ks• Delta Ms

• Mass Matrix • Stiffness• Response• Actuator

Bench Test

• Geometry• Mass• Stiffness

AeroelasticTailoring

FlutterAnalysis

ASEAnalysis

AeroData

TeamFEM

TeamMassData

ActuatorData

StoresData

Requirements to Team:•Stiffness – Actuator & Structural•Freeplay•Control Law Filters•Geometry•Etc.

YF-22 Design and Ground

& Flight Testing

Criteria

(Circa 1991)


YF-22/PAV Design, Ground and Flight Testing Influences

• YF-22 design basis for initial control loop stiffness sizing• From YF-22 Testing

– Identified 11 Corrections or Improvements for the Flutter Excitation System.• Lessons Learned applied to EMD Flutter Excitation System

Design.– Significant Horizontal Tail Journal Bearing Friction Effects

Identified on YF-22 GVT.• Result led to incorporation of low friction Horizontal Bearings on

EMD Flight Flutter Test Aircraft.– Only Subsonic Flutter Test Data Obtained due to Schedule

Constraints.• Lack of YF-22 Supersonic Data put emphasis on need for a

Supersonic Flutter Model for EMD.


Aeroelastic Design Criteria

• Classical Flutter– The Air Vehicle, including for all probable failures, shall be free from

flutter or other Aeroelastic Instabilities to 1.15 VL at constant altitude and at constant Mach.

– Probable failures included any single hydraulic system failure.• Damping

– The minimum damping of any potentially critical flutter mode shall be greater than the lesser of 0.03 or 1 percent above the GVT measured mode damping.

• Transonic Buzz and LCO– Control surfaces shall be free from buzz or LCO. A tailored criteria

was developed and applied to the F/A-22 design. • Aeroservoelasticity

– Any potential aeroservoelastically critical mode shall have a gain margin of 6dB, and separately a phase margin of +/-60 degrees.


Structural & Aeroelastic DesignVehicle / Airframe Design Team

Airframe A & ITeam

• Design Layouts

• Structural Sizing

• Design Drawings• Detail Design Schedule• Inputs from all Disciplines

– Manufacturing– Maintainability– Weights– Aero/Thermo– Structures– Etc.

• Sub-optimization

Airframe IPTs

Structures A & ITeam

• Structural Criteria, Policy & Methods

• Air Vehicle Loads• Team Finite Element Model• Air Vehicle Aeroelastic Analysis• Aeroelastic Optimization• Stiffness & Freeplay Requirements• Filters for ASE• Materials & Processes• Structural Development Tests• Internal Loads• Allowables• Vibration & Acoustics

Requirements & Sonic Fatigue

• Airframe Design Integration / Coordination

• Design Scheduling


Air Vehicle Finite Element ModelAir Vehicle Finite Element Model


Flutter Analysis Scope

Gross DesignSensitivities

Trend Data /Parametrics

Detail DesignSensitivities

Major Strength &Stiffness Updates

• Actuators• Rudder Plies• Skin Offsets• Fin Spars• Horizontal Shaft• Hinge

Placement• Super Element

Factors• L.E. Flap• Engine• Boom EI & GJ• Component

Substitution

• Control Loop Stiffness

• Rudder Plies• Fin Ply Sweep

Angle• Horizontal Ply

Sweep Angle• Horizontal Ballast

Weight• Boom Super

Element• Access Panel

Effectiveness• Wing Ply Sweep• Internal Fuel• Pylon Stiffness• Store Loadings

• 8 Major A/V FEM Updates

• 9 Less Major A/V FEM Updates

• Numerous A/V Aeroelastic Sizing Updates

• 5+ Engine FEM Updates

• 6+ Pylon and Missile Adapter FEM Updates

• H. S. Skins• Rudder Skins• Flaperon Skins• Wing Skins• Fin & Rudder

Spars & Ribs• Boom Panels &

Bar Elements• Control Loop

Stiffness• Freeplay! ∆ Wts at Asets• Missile Adapter

Stiffness, Damping & Geometry


F/A-22 Aeroelastic Tailoring / MDO Analysis Process

• Define Constraints and Objectives– Flutter Speeds– Hump Mode Damping– Ply Stack-up– Strength– Etc.

• Define Design Variables (800+ Design Variables Used)

• Compute Sensitivities/Derivatives

• Perform Optimization – to Achieve Objectives for Minimum Weight Design with Known Constraints.


F/A-22 Aeroelastic Tailoring / MDO Analysis Process

Key Elements of the Aeroelastic Tailoring Process:

• Parametric Analysis to Develop Understanding of Issues.

• Tools to make Rapid Selection and Definition of Design Variables.

• Meaningful Constraint Definitions Tightly Coupled to Design.

• Ability to Compute Accurate Design Sensitivities.

• Ability to Rapidly Update the Air Vehicle FEM and Mass Data bothfor Sensitivity Analysis and Aeroelastic Resize Analysis.

• Optimizer Capable of Handling many Design Variables, Sensitivities, and Constraints.


Aeroelastic Design Issuesand Flutter Critical Modes

• Classical Flutter– Wing, Fin-Rudder, Horizontal and Empennage

• Hump Modes– Coupled Horizontal and Flaperon

• Transonic Buzz– All Control Surfaces except Horizontal; Rudder most Critical

• LCO– Transonic and edge of envelope (all surfaces including horizontal)– Drivers - control surface inertia & stiffness and control loop

stiffness and freeplay.• Aeroservoelasticity

– Large, high inertia control surfaces coupling with vertical and roll / lateral modes of the fuselage.


Flutter Critical Modes –Early EMD Configuration

0

VL 1.15 VL

Altitude

Mach

52 Hz High Frequency Fin / Rudder Tip Mode

Potential for Transonic Buzz / LCO

(All surfaces except Horizontal - Rudder most

critical)

Wing Bending/ Torsion Mode

30+ Hz Horizontal Rotation Modes

30+ Hz Rudder Rotation Modes

Region of Potential Hump Modes

Increased Flutter and LCO Criticality

Flutter Boundaries


Initial Trades & Design Optimization Results- 52 Hz Coupled Fin / Rudder Tip Mode

0

VL 1.15 VL

Altitude

Mach

Early Strength Design

Design Trades & Tailoring:• 0.25 increase in t/c – Stabilized Mode but

did not eliminate mechanism. Adverse Aero Impact

• Upper hinge bearing lowered 6 inches –Eliminated flutter mechanism.

• 35 % Tip chord reduction – Favorable for all fin-rudder modes & loop stiffness. Adverse LO Impact.

• Aeroelastic Tailoring - Difficult to improve mode with tailoring alone

0.25 increase in t/c

Flutter Boundaries


Initial Trades / Design Optimization Results- 30 Hz Coupled Fin / Rudder Mode

0

VL 1.15 VL

Altitude

Mach

Initial Strength Design with Initial Control Loop Stiffnesses

Aeroelastic Tailored Design with Reduced

Control Loop Stiffnesses

Aeroelastic Tailoring very effective

Flutter Boundaries


Initial Trades & Design Optimization Results- 30+ Hz Coupled Horizontal Rotation Mode

• Primary stability drivers– Control Loop Stiffness– Control Surface Moment of Inertia (MOI)– Tail Boom torsion-plunge coupling (Actuator access door)– Skin stiffness distribution

• Impact on Design– Design changes to minimize MOI.– Control Loop Stiffness improvements– A change to the actuator access door location – Effectiveness of actuator access door due to maintainability

loose (high –clearance) fit fastener requirements was an issue.• With door on bottom of tail boom, tail boom pitch-plunge coupling

destabilized this mode.• Parametric trades showed moving the door to the inboard side of tail

boom eliminated the adverse coupling. • Design change implemented.


Design for Transonic Buzzand LCO – Empirically Based

M = 1.2, V = 220 KEAS, Vtrue = 1163 ft/sec

Reduced Frequency,

ωc/V

0.3

0.4

0.5

0.6

0.7

20 50 60 70 80 90 10030 40

% Loop Stiffness – Keffective/Knominal

Buzz Empirical Based Design Requirement

ωc/V > 0.40

Flaperon @ max FreeplayRudder @ max Freeplay

Aileron @ max Freeplay

LEGEND:AileronFlaperonRudder Rudder Most Critical


Flutter Model Vertical Tail Assembly Installed in 4X4 Tunnel

Mach

Dynamic Pressure

Accelerometer Response Flutter

0 1 2 3 4 5 6 7 8 9 10 11 12 13Time ~ Seconds

Magnitude

Mach

Dynamic Pressure

Accelerometer Response Flutter

0 1 2 3 4 5 6 7 8 9 10 11 12 13Time ~ Seconds

Magnitude

1000

500

0

500

-1000

Rud

der A

ccel

~gs

0 5 10 15 20Time ~ Seconds

Mach 1.41Q=2710 psf at FlutterFreq = 240 Hz Single Degree of

Freedom Flutter

Increasing Q

1000

500

0

500

-1000

Rud

der A

ccel

~gs

0 5 10 15 20Time ~ Seconds

Mach 1.41Q=2710 psf at FlutterFreq = 240 Hz Single Degree of

Freedom Flutter

Increasing Q

Rudder Tip Failure Total Fin

Failure

Rudder Tip Failure Total Fin

Failure

Failures after Flutter

Vertical Fin in Tunnel


Flutter Model Key Result / Issue

• Flutter Model Key result / issue

– Transonic Buzz of rudder measured at an ωc/V= 0.66

– Buzz Criteria used for design - ωc/V> 0.40

– To redesign to ωc/V> 0.66 would have been a major impact.

• Approach Taken– Used ENSAERO with 10 modes to correlate to flutter model at

model scale.

– ENSAERO then ran at full scale resulting in ωc/V= 0.47 at buzz.– Result indicated significant Reynolds number / scale effect.– With this result, decision was made to proceed with the then

current design into flight test.


Multiple Team FEM and Aeroelastic Analysis Requirements Updates

FEM ModelsBlock I• Model 638• Model 639• Model 640• Model 641• Model 642• Model 644• Model A645Block II• Model A645ARequirements Updates• Control Loop• Skin Sizing• Backup Structure• FreeplayTest Data / Correlation• L-3 W/T Pressure Model• Flutter Model• Actuator Bench Test

‘91 1992 1993 1994 1995 1996 1997 1998 1999 2000

3 updates

1 update 3 updates

2 updates 3 updates

3 updates

Steady Aero Correlation used to update Unsteady Flutter Correlation

Failure Modes Stiffness

Multiple Aeroelastic Sizing Updates Performed with

each FEM update


~ 245 Lbs for Flutter

Control Loop Stiffness &Freeplay Requirements(Substructure & Actuator Design)

Skin & Spar Sizing for BendingAnd/or Torsion Stiffness At Ply Level for Composites

Design Impacted by Flutter

MDO F/A-22 Structural Design – Flutter Design Impacts Overview


Summary – Aeroelastic Design

• The F/A-22 Presented many Aeroelastic Design Challenges – significant impact of LO on the control surfaces and edges.

• Several critical modes identified and were successfully addressed– Required an Integrated Approach using Parametric Analysis and Aeroelastic

Design Optimization• Parametrics key in identifying Hinge and Access Panel location changes.• Optimization used extensively for ply and substructure changes.

• Aeroelastic design optimization was successfully applied to achieve a minimum weight aeroelastic design for the F/A-22.

• A rigorous process in place of Mil-Std Criteria was used to establish freeplay allowables for the F/A-22 for control of LCO and Buzz.

• The integrated structural dynamic unsteady CFD analysis (ENSAERO) showed significant Reynolds Number effects for transonic buzz, with the wind tunnel scale result being very conservative.

• Multiple filters developed and incorporated to address aeroservoelastic stability.


Flutter Development / Certification Overview

• Analysis– Doublet

Lattice– Zona 51

• L3 Press. Model• Flutter Model

• Stiffness Matrix

• Delta Ks

• Mass Matrix• Delta Ms

• Stiffness• Response• Actuator

Bench Test

• Geometry• Mass• Stiffness

AeroelasticTailoring

FlutterAnalysis

ASEAnalysis

AeroData

TeamFEM

TeamMassData

ActuatorData

StoresData

Requirements to Team:• Airframe Stiffness• Actuator Stiffness• Free play• Control Law Filters• Geometry• Etc.

YF-22 Design and Ground

& FlightTesting

Criteria /Spec

Design (Circa 1991)

Verification Testing:• SICs• MOI• Stiffness & Free play• GVTs• Flight Flutter• Structural Coupling

Final Certification:• Correlation &

Analysis Updates• Final Analysis• Certification

Documentation• Final Reports

V&V & Final Certification


Flutter Verification / Certification Process

Basis of TISs, ETRs,AOLs, and Inspections

Preliminary Block IIFlutter and ASE

Analysis and Reports

Actuator Impedance

(ETR HB9205)

Aircraft SIC

(TIS ST0960)

HorizontalStab. SIC

(TIS ST0960)

Control LoopStiff & F.P.

(TIS ST0950

Control SurfInertia(ETRs)

Update VibrationPredictions

Vibration Correlation& Update Flutter &

ASE Analysis

Air Vehicle GVT

(TIS ST0930)

FES GroundTest

(TIS ST0940)

Flight FlutterTests

(TISs ST0010& ST0080)

Update Flutter, Buzz& LCO Predictions

Final Limitations,Certification Reports

and FSMP and IAT Updates

StructuralCoupling

(TIS FQ0900)

Interim Limitationsand Letter Reports

Store MOI &Cant PylonGVT (ETRs)


Ground Tests

Actuator Bench Impedance Tests

• Aileron, rudder, and horizontal tail actuators tested• Flaperon not tested due to similarity to rudder & aileron actuators• Powered and failed (Hydraulic System) conditions tested• Variations in hydraulic fluid temperature, mean load, oscillatory

load, and stroke position were tested

Results:

• The actuator stiffness less than predicted in the failed / compensator mode for several of the test conditions.

• Resulted in following flight manual requirements:– To slow to speed << than VL after any single failure condition.

– To slow to final approach speed when the compensator depleted ICAW enunciates.


Ground Tests (Continued)

Structural Influence Coefficient Test

• Aircraft 4001 and 4003 tested

• Included both distributed and for point loads–Point Loads on the wings, leading edge flaps, ailerons, flaperons, vertical tails, rudders, and horizontal tails.

• Load and deflection measurements were analyzed to obtain the Structural Influence Coefficients (SIC).

• A separate SIC test was conducted on isolated cantilevered horizontal tails



Control Surface Stiffness & Freeplay Tests

• Tests conducted on several aircraft to verify control loop freeplay and stiffness values.

• Tests were conducted with nominal actuators and with solid rods of known stiffness so that the backup structure stiffness could be determined.

• Freeplay for nominal control loop and control loop set to maximum freeplay was measured.

• Tests used to calibrate field freeplay check procedure.–Procedure has been incorporated into the Dash 6 / Tech Orders.



Full Aircraft Ground Vibration Tests (GVTS)• Tests were conducted on Aircraft 4001, 4003, 4005, and 4008 –

clean wing and with external stores. • Tests used to measure / verify the total aircraft structural vibration

mode shapes, frequencies, and damping values.• Individual control surfaces, doors, stores and launchers, and

landing gear were also tested.

Ground Vibration Test Setup


Other Ground Tests

• Control Surface Inertia / MOI

• Cantilever Pylon, fire missile adapter, and external store GVT.

• Flutter Excitation System (FES) ground test.

• Structural Coupling– Clean wing aircraft

– Aircraft with external stores


Ground Test Correlation

Correlation of Analytical Models to Ground Tests• Correlation of frequencies, mode shapes, and SICs performed

using GENESIS. • Changes made to the Finite Element Models (FEMs) were passed

on to the individual IPTs to help verify and improve their modeling techniques.

• Updated vibration and flutter predictions for flight test were made, using the correlated models.

• Final correlation data were used to develop ‘correlated’ models for the production representative FEMS for final certification analysis.


Flight Flutter Testing

• Flutter Verification Flight Testing Covered– Clean Wing on A/C 4001, 4003, and 4008

• 4003 Primary Flutter Test Aircraft• Basic clean wing envelope to VL, doors closed and open• Approximately 1190 TIS points were completed• TIS points included

– Mach and altitude– Elevated-g and sideslip testing– Control surface, symmetry, excitation mode (sweep, random, or burst)– Testing at nominal freeplay – And testing at maximum freeplay / zero hinge moment conditions

– 2-Tank Ferry on A/C 4002, and 4003– External Combat ‘Soda Straw’ envelope on Aircraft 4003


Flight Tests - Flight Flutter

• Clean Wing Flight flutter Mach / Altitudes Test Points – 69 closed door points– 31 main weapons bay door open points– 32 side weapons bay door open points– 19 EW and 19 IRCM door open points

• A total of 18 structural modes were excited and tracked.– A Flutter Excitation System (FES) was developed for the F/A-22 and

used to excite the critical in-flight aeroelastic modes. – Only the modes deemed to be critical at a Mach / altitude point were

excited at that point.• For door open testing, frequency and damping values were

obtained using random air turbulence to excite the doors.


Flutter Verification Flight Tests

• Real time data was transmitted simultaneously to the Ridley Mission Control Room (RMCR) at EAFB for safety of flight monitoring, and to Room 1060 in Marietta for near real time detail analysis and correlation– The analysis of real time data in Marietta proved invaluable in

providing flight-to-flight clearance, for updating flight card requirements, and for improving flight testing efficiency.

• The F/A-22 Flight Flutter Test Data used to: – Verify that the F/A-22 is free from flutter and other dynamic

instabilities throughout the structural design envelope– Measure sub-critical frequencies and damping for validation of the

aeroelastic stability analysis

• Measured sub-critical frequencies and damping compare favorably with predictions.


Flutter Analyses for Final Certification

• For both clean wing and with external stores, update– SIC/GVT correlation– Flutter correlation– Final Flutter, Buzz, and LCO analysis, including failure modes– ASE / Structural Coupling Analysis– Freeplay limits / inspection procedures– Control surface MOI limits

• The final flutter analyses used for the final flutter certification of the F/A-22.

• A total of 26+ reports / report volumes were prepared and submitted to the customer for final F/A-22 Aeroelastic Stability Certification covering:

– All ground and flight testing and correlation– All updated / final aeroelastic stability analysis– And updated inspections and limits

1

CONTRASTING FAA AND USAF DAMAGE TOLERANCE

REQUIREMENTS

2005 USAF Aircraft Structural Integrity Program Conference

Robert G. EastinFederal Aviation Administration

2

Overview• What Requirements? • Categories of fatigue• Background

– USAF– FAA

• Damage Tolerance Requirements– USAF– FAA

• Fail-safe– USAF– FAA

• Closure

3

What Requirements?USAF

MIL-A-83444, Airplane Damage Tolerance Requirements, July 2, 1974

FAA§ 25.571 Damage-tolerance and

fatigue evaluation of structure

(a) General

(b) Damage-tolerance evaluation

(c) Fatigue (safe-life) evaluation

(d) Sonic Fatigue Strength

(e) Damage-Tolerance (discrete source) evaluation

[Amdt. 25-45, 43 FR 46242, Oct. 5, 1978]

4

Fatigue Categories• Normal Fatigue

– Nominal part with no “surprises”– Every part has inherent characteristics– Expected, inevitable and predictable

– Probability of occurrence increases steadily with time

• Anomalous Fatigue– Off nominal physical condition (e.g. material defect, poor

fabrication quality, corrosion pit, ding, bang, scratch)– Unexpected and unpredictable occurrence– Some designs more vulnerable than others

• Unexpected Normal Fatigue– Nominal part with a surprise (e.g. external loading, internal

stress, usage, etc.) – “Postdiction” straightforward using normal fatigue methods

5

USAF Experience

1950 19701960 1980

Fatigue (8866/8867/Durability)

Damage Tolerance(83444)

1958

F-111

ACCIDENT

B-47

ACCIDENTS

1969 1974

BOTH

6

FAA Experience

1950 19701960 1980

Fatigue (Safe-life)

Fail-safe

1954

LUSAKA

ACCIDENT

COMET

ACCIDENTS

1956 1978

EITHER

1977

Damage-tolerance(Amdt 45)

Is DT Impractical?

No

Yes

7

Comparison of “Watershed” Events

Unexpected normalAnomalousCategory of Fatigue

Right Horizontal Stabilizer Aft Spar Upper Chord

Left Wing Pivot Fitting Lower PlateComponent Involved

.8.071Fraction of DL at Failure

16723 Flights/47621 Hours100 HoursTotal Time in Service at Failure

7079-T6 AluminumD6ac Steel (220-240 KSI)Material Involved

20,000 Flights/60,000 Hours6,000 HoursDesign Life (DL)

NoYes – 16,000 HoursFatigue Test

CAR 4.270 Fail-safeSafe-lifeFatigue Design Basis

B707-300F-111Airplane Model

May 14, 1977December 22, 1969Date

LusakaF-111

8

Outcome of USAF & FAA Experiences

• Create a new fatigue rule that requires quantification of crack growth and residual strength characteristics.• Require correlation between characteristics and any inspections established.• Provide an alternative if inspections are impractical.

• Effective inspections that can be relied on to ensure safety.

• Fail-safe approach does not adequately ensure that fielded designs fail safely.• Fail-safe approach addresses residual strength but neglects inspectability.

FAA

• Create supplemental design requirements that specify minimum crack growth life and residual strength attributes that a structure must possess with specified cracks assumed to be present as manufactured.

• A design that has a minimum level of tolerance to defects.

• Safe-life approach does not preclude selection of unforgiving materials, design concepts and working stress levels.• Safe-life does not adequately address potential defects.

USAF

STRATEGYOBJECTIVECONCLUSIONS

9

USAF vs. FAA DT Requirements

YesNoProvision for alternative approach if damage tolerance impractical?

NoNoUnexpected normal fatigue

YesYesAnomalous fatigue

YesNo(Addressed by Durability

Requirements)

Normal fatigue

Threats addressed:

Replace fail-safeReplace safe-lifeIncorporation Philosophy:

Maintenance Actions*Design attributes (& in-service inspections as required)Outcome:

Safety indefinitelySafety during design serviceObjective:

New airplane designs – safety of flight structure

New airplane designs – safety of flight structure

Applicability:

Fail-safe approach inadequateSafe-life approach inadequatePrimary motivation for:

FAAUSAF

* In-service inspections expected.

10

Design Attributes vs. Maintenance Actions“This specification contains the damage tolerance design requirements applicable to airplane safety of flight structure. The objective is to protect the safety of flight structure from potentially deleterious effects of material, manufacturing and processing defects through proper material selection and control, control of stress levels, use of fracture resistant design concepts, manufacturing and process controls and the use of careful inspection procedures.”

Scope paragraph of MIL-A-83444

“The purpose of the proposal was to establish an evaluation requirement rather than an absolute requirement for the strength, detail design, and fabrication of the structure”.

Response to comments in Preamble to Final Rule, October 5, 1978

“Based on the evaluations required by this section, inspections or other procedures must be established, as necessary, to prevent catastrophic failure……”

§ 25.571 (a)(3)

11

USAF vs. FAA DT Prescribed Requirements

YesYesResidual strength:

NoYesInspection intervals:

NoYesMinimum crack growth life:

NoYesCracking scenarios:

NoYesIn-service detectable crack sizes:

NoYesInitial crack sizes:

NoNoDesign Concept: (i.e. single or multiple load path)

FAAUSAF

12

Design Concept“It should be emphasized that while the “Fail Safe” concept appears to offer a larger degree of safety, it is the intent of the new criteria that structure qualified to either category have equal safety”.

Wood, H.W., Application of Fracture Mechanics to Aircraft Structural Safety, Engineering Fracture Mechanics, Vol. 7, 1975.

“… the applicant would be allowed to apply the damage-tolerance approach to both single load path and multiple load path structure. The FAA believes the applicant can, by sufficient analysis and testing, establish that a single load path structure has sufficiently slow crack growth properties so that, if a crack were to develop, it would be discovered during a properly designed inspection program.”

Preamble to Amendment 45 Notice of Proposed Rulemaking, August 15, 1977

13

Fail-safe

• Removed from § 25.571 with Amendment 45.• Integrated into the USAF requirements as an

optional design concept that, if chosen, must possess specified attributes based on inspectability.

• Past FAA and current USAF brands of “fail-safe”share similarities at the conceptual level but differ significantly at the detail level.

14

USAF vs. FAA Fail-safe

Design Attributes

Design Attributes*Outcome:

YesYesAssociated with Multiple Load Path Structure:

YesYesIncluded as Optional Approach:

FAA Pre-Amd 45USAF

* Plus inspections as required

15

FAA Pre-Amd 45 Fail-safe

“It must be shown by analysis, test, or both, that catastrophic failure or excessive structural deformation, that could adversely affect the flight characteristics of the airplane, arenot probable after fatigue failure or obvious partial failure of a single principal structural element. After these types of failure of a single principal structural element, the remaining structure must be able to withstand static loads corresponding to the following:………”

16

FAA Pre-Amd 45 Fail-safe Application

“Generally, manufacturers satisfying the requirements under the fail-safe concept merely substantiated the structures for failure of single principal elements under static loading conditions. Although it was recognized that inspections were necessary there were no specific requirements to determine safe inspection periods based on crack growth or remaining life of secondary structure in the event the primary member failure was not immediately obvious.”

Swift, T., Verification of Methods for Damage Tolerance Evaluation of Aircraft Structures to FAA Requirements, Proceedings of the 12th

Symposium of the International Committee on Aeronautical Fatigue, Toulouse, France, 1983.

17

Certified Fail-Safe Capability for Fuselage Structure in Longitudinal Direction

20”1 Crack stopper bay skin crack with center frame failed.L10112

12”12” Skin crack.B747

20”1 Frame bay skin crack.B727

20”1 Frame bay skin crack.B737

20”1 Frame bay skin crack.DC-9

40”2 Frame bay skin crack with central crack stopper failed but frame intact.DC-101

Skin Crack

size

“Fatigue failure or obvious partial failure of a single principal structural element”

Airplane Model

1. Crack stoppers located under frames.2. Crack stoppers located between frames.

18

USAF vs. FAA Fail-safe Prescribed Requirements

YesYesResidual strength.

Obvious during normal

maintenance.

Determined by manufacturer

Inspectability of stable load path failure or crack arrest.

NoYesMinimum crack growth life before and after stable load path failure or crack arrest.

NoYesCracking scenarios before and after stable load path failure or crack arrest.

NoYesIn-service detectable crack sizes.

NoOnly for “fail-safe

crack arrest”structure

Damage size after stable load path failure or crack arrest.

NoYesInitial crack size for intact structure.

FAA Pre-Amd 45USAF

19

CLOSURE

• USAF DT = FAA DT• USAF Fail-safe = FAA Fail-safe• FAA Fail-safe was removed in 1978.

A S I P 2 0 0 5

⊥ !

1

CONTRASTING FAA AND USAF DAMAGE TOLERANCE REQUIREMENTS

Robert G. Eastin*

Damage tolerance requirements were formally adopted by the United States Air Force (USAF) for the design of new airplanes and by the Federal Aviation Administration (FAA) for the certification of new large transport type designs in the 1970’s. The underlying reasons were different and it is therefore not surprising that the requirements adopted are different. The prescriptive nature of the USAF requirements is contrasted with the more objective nature of the FAA requirements. It is also noted that the outcome of each set of requirements is different. The USAF requirements result in structure with a specified level of tolerance to defects plus in-service inspections if necessary. The FAA requirements result in maintenance actions (i.e. “inspections or other procedures”) determined to be necessary to prevent catastrophic failure due to fatigue from all potential sources. The primary intent of this paper is to objectively identify similarities and differences between the two sets of requirements as they are written without passing judgment on them or getting into the nuances of how they have been implemented. This paper also examines “fail-safety” as included in the current USAF damage tolerance requirements and in the FAA fatigue requirements from 1956 to 1978.

INTRODUCTION

Two well known and widely applied sets of damage tolerance requirements are those that must be adhered to for the design of USAF aircraft and those that must be used for the certification of civil aircraft type designs in the United States. Although each set is commonly referred to using the words “damage tolerance” significant differences exist in intent and application. This paper examines some of these differences.

In conducting any comparison it is important to clearly define exactly what is being compared. The USAF damage tolerance requirements have been subject to revisions since they were first adopted and often custom tailored to specific aircraft systems. However the basic philosophy and intent has remained unchanged since

* Federal Aviation Administration, Los Angeles Certification Office

A S I P 2 0 0 5

⊥ !

2

the first requirements were published in 1974. Therefore for the purposes of this discussion the USAF requirements being compared are those in [1].

Extra care must be taken when identifying what FAA requirements will be compared. This is because somewhat different requirements have evolved over the years for small airplanes, transport airplanes, small rotorcraft and large rotorcraft. These requirements are contained in parts 23, 25, 27 and 29 respectively of [2] and the differences have been discussed by Eastin [3]. In the discussion that follows the FAA requirements that will be compared are a subset of those that were originally published for transport airplanes in [4]. This subset is included in paragraphs (a) and (b) of section 25.571 of [2] as amended by [4]. Other requirements are included in paragraphs (c), (d) and (e) of section 25.571. These are “Fatigue (safe-life) evaluation”, “Sonic fatigue strength” and “Damage-tolerance (discrete source) evaluation” respectively and are beyond the scope of this discussion since they have no similar counterparts in the requirements of [1].

CATEGORIES OF FATIGUE

The author believes it can be useful to separate fatigue into three categories. This was first proposed in [5] and this convention will also be used here to facilitate the discussion. The categories are normal, anomalous and unexpected normal, and are described below.

Normal Fatigue

Normal fatigue is the inevitable accumulation of damage with resultant cracking that can be expected to occur at some point in time in any structure that is subjected to cyclic loading of sufficient magnitude and frequency. It occurs in structure that is designed and fabricated without error, operated as planned, and serviced as expected. As defined, normal fatigue is predictable and the probability of it occurring is steadily increasing with time. Fatigue testing can be performed to characterize normal fatigue at the detail, component, and aircraft level. A normal fatigue event occurring in one aircraft can be expected to occur in others. In this sense the cracked aircraft is representative of the rest of the fleet.

Normal fatigue can occur locally when there are isolated areas that are significantly more fatigue sensitive than surrounding areas due to higher stress level, unique geometry, etc. Normal fatigue can also occur over large areas when similar details are subjected to the same stress levels. When large areas are subject to normal fatigue the term “multiple site damage” and “multiple element damage” are often used. The traditional strategy used to deal with normal fatigue is safety-by-retirement which is more commonly referred to as the “safe-life” approach. Safety-by-inspection may also be an effective strategy for normal fatigue provided

A S I P 2 0 0 5

⊥ !

3

inspection reliability is acceptable and eventual terminating action (e.g. modification, replacement) takes place based on inspection findings.

Anomalous Fatigue

Anomalous fatigue is the result of an off nominal physical condition. It is unexpected and unpredictable. Classic sources include material defects, tool marks and poor quality holes. Other sources include service induced damage such as corrosion pits and dings and scratches. All the sources mentioned above are by their nature unpredictable. Considerable effort is made during design and manufacture to mitigate the risk of introducing anomalous fatigue sources. Likewise controls are typically put in place once an aircraft enters service to minimize the risk of service related anomalies. Anomalous fatigue occurring in an aircraft is not, by definition, representative of the fleet. Swift [6] refers to such an aircraft as a “Rogue Flawed Aircraft” and others commonly use the term “rogue” to describe anomalous sources of fatigue.

Anomalies are, by their very nature, difficult to quantify before they occur. Tiffany has discussed this in [7] and questioned the validity of extrapolating equivalent initial flaw distributions although he also notes that this has been done. Anomalies tend to be singular events resulting in very localized fatigue cracking. This is reflected in the cracking scenarios that are specified for use by the USAF in [1].

The most effective strategy for anomalies is to design the structure to be tolerant of them. This is the essence of [1] as will be discussed in more detail below.

Unexpected Normal Fatigue

There are many examples of unexpected and premature fatigue that can’t be blamed on an off nominal physical condition. Some typical root causes include incorrect external loads and/or internal loads/stress, overly severe usage (as compared to design assumptions) and other shortfalls in our ability to accurately model the structure and predict the future. In hindsight this category of fatigue has to be considered “normal” and we typically do well at “postdiction” once we correct our input data. In most cases unexpected normal fatigue is representative of the fleet and should be addressed accordingly.

BACKGROUND

A review of key events leading up to the adoption of the requirements is considered helpful in understanding the differences that exist. As noted below the USAF and the FAA had uniquely different experiences that resulted in somewhat different conclusions, objectives and requirements.

A S I P 2 0 0 5

⊥ !

4

USAF

Key events and experience that lead to the adoption of damage tolerance requirements by the USAF have been reviewed by Lincoln [8], [9], [10]. A summary illustration is provided by Figure 1 below.

Up until 1958 the USAF had no formal fatigue requirements. According to Lincoln [8] aircraft were generally designed based on static strength considerations only and the factor of safety applied was expected to account for deterioration from usage and quality problems as well as uncertainties about loading and material strength. Based on this all three (normal, anomalous, and unexpected normal) of the author’s categories of fatigue should have been accounted for. Lincoln [9] attributes the success of this approach up through the mid-1940’s to conservative analysis methods, the inherent fatigue and fracture resistance of available and generally used airframe materials and the relatively low usage of USAF aircraft. These factors combined and resulted in aircraft designs that were inherently tolerant to fatigue and other kinds of damage in spite of the lack of any formal requirements.

However there were factors coming into play that resulted in an erosion of the inherent robustness of USAF aircraft. The advent of new high strength alloys, the increased importance of aircraft performance and more refined design tools were some of them. This loss of robustness resulted in an ever increasing number of structural integrity related problems. Lincoln [9], [10] specifically cites the fatigue problems experienced on the B-47 as being one of the primary drivers that led to the USAF adopting formal fatigue requirements, to be used in the design of future USAF aircraft, in 1958. These requirements specifically required that deterioration due to repeated loading in service be considered and minimized. This was accomplished in part by requiring full scale fatigue testing to a multiple of the specified service life. Any significant fatigue cracking that occurred during this test

1950 19701960 1980



1958

F-111

ACCIDENT

B-47

ACCIDENTS

1969 1974

BOTH

1950 19701960 1980



1958

F-111

ACCIDENT

F-111

ACCIDENT

B-47

ACCIDENTS

B-47

ACCIDENTS

1969 1974

BOTH

Figure 1 USAF Key Events

A S I P 2 0 0 5

⊥ !

5

had to be addressed such that it would not be expected in fielded aircraft during their service lives.

Although the new USAF safe-life requirement forced the aircraft designers to consider fatigue, in addition to static overload, as a threat to structural integrity it was soon realized that it did not prevent the use of low ductility materials operating at high stress levels. The example of this most commonly cited is the F-111. The F-111 experience painfully illustrated how such design decisions combined with an unexpected defect could be devastating. As part of the F-111 engineering development program a successful full scale fatigue test of the wing box was accomplished to 16,000 simulated flight hours. Accounting for test spectrum severity the USAF interpreted the results as demonstrating a safe-life of 6000 hours using a scatter factor of four. Nevertheless on December 22, 1969 an F-111 crashed as a result of a fatigue failure in the lower plate of the left wing pivot fitting. The total time in service at the time of the accident was 100 hours. This failure was attributed to a defect that was produced during manufacture of the forging that the plate was fabricated from. This and other service incidents convinced the USAF that the existing fatigue requirements needed to be augmented. It was reasoned that the requirement to fatigue test by itself could still result in designs that were not sufficiently tolerant to manufacturing and service induced defects. To achieve the desired tolerance something had to be done to positively affect the design relative to material choices, stress levels and design details. That something was determined to be prescriptive crack growth and residual strength requirements assuming that defects are present when the airplane first enters service.

In summary what motivated the USAF to adopt their damage tolerance requirements was the conclusion that the safe-life approach by itself had not delivered the overall structural integrity desired. Specifically they were missing a level of robustness largely due to unfortunate choices of materials and stress levels that were not influenced by the fatigue requirements that were on the books at the time. The added requirements directly influence material selection and stress levels at the design stage. It should also be noted that the USAF damage tolerance requirements were supplemental to the fatigue requirements already embodied in [11] and [12]. That is, the USAF did not get rid of the existing requirements but simply added to them to achieve the overall desired result.

FAA

A summary of key events that are important in the evolution of FAA damage tolerance requirements is provided by Figure 2 below.

A S I P 2 0 0 5

⊥ !

6

Fatigue requirements of some kind have been part of the civil aviation requirements for some time. For example if we go back to 1945 and look in the Civil Air Regulations (CARs) at section 04.313 we find a requirement that states that;

“The structure shall be designed in so far as practical, to avoid points of stress concentration where variable stresses above the fatigue limit are likely to occur in normal service.”

History indicates that, similar to USAF experience, fatigue was not a major issue early on with civil aircraft. The lack of major fatigue issues may be attributed in part to the existence of a formal requirement to consider fatigue. This should have resulted in more attention to fatigue by the civil aircraft manufacturers. However in the author’s opinion it is also due to many of the same factors at work in the design of early USAF aircraft that were mentioned previously.

As civil aircraft designs became more challenging (e.g. pressurized fuselages) fatigue events became more common place. Additionally it was recognized that even if normal fatigue is adequately addressed aircraft will always be vulnerable to anomalous and unexpected normal fatigue. It was reasoned that an alternative approach to dealing with fatigue might be to accept that fatigue cracking is inevitable and design the structure to crack gracefully. This concept was based on designing such that any cracking would be obvious during normal maintenance before it reduced the strength of the structure to an unacceptable level. This was generally referred to as the “fail-safe” approach.

1950 19701960 1980

Fatigue (Safe-life)

Fail-safe

1954

LUSAKA

ACCIDENT

COMET

ACCIDENTS

1956 1978

EITHER

1977


Is DT Impractical?

No

Yes

1950 19701960 1980

Fatigue (Safe-life)

Fail-safe

1954

LUSAKA

ACCIDENT

LUSAKA

ACCIDENT

COMET

ACCIDENTS

COMET

ACCIDENTS

1956 1978

EITHER

1977


Is DT Impractical?

No

Yes

Figure 2 FAA Key Events

A S I P 2 0 0 5

⊥ !

7

The key events that are considered the primary catalyst for the adoption of fail-safe requirements by the FAA are the Comet I airplane failures that occurred in 1954. These failures have been discussed in some detail by Swift [13] and will only be briefly reviewed here.

The Comet was designed and manufactured in the United Kingdom by De Havilland Aircraft Company. The Comet design was a major technological advance at the time. It was the first commercial jet and was designed for relatively high altitude operation. Shortly after entry into service a Comet flying at 30,000 feet disintegrated and crashed into the Mediterranean Sea. All airplanes were removed from service and were not returned until fleet modifications were made to correct what was thought to be the cause of the accident. However shortly thereafter a second Comet disintegrated at 35,000 feet and crashed into the Mediterranean. The accident investigation that followed included a full scale fatigue test of the fuselage and revealed fatigue critical locations at openings in the pressurized fuselage that had not been identified previously. It also was found that the critical crack size was relatively small and could not be expected to be detected during normal maintenance.

The Comet experience reinforced the thought that the fail-safe approach might be an acceptable and even superior alternative to the safe-life approach. Consistent with this the FAA revised the CARs in March 1956 [14] and added fail-safety as an option to the safe-life approach.

Fail-safe became the option of choice for the majority of large transport aircraft certified in the 1960’s and 1970’s. This included the Airbus A300; Boeing 707/720, 727, 737, 747; Douglas DC-8, DC-9/MD-80, DC-10; Fokker F-28; and Lockheed L-1011. The fail-safe approach was very attractive for several reasons. If a structure can be designed such that cracking will be readily detected before it becomes dangerous it can be reasoned that cracking in itself is not a safety issue. Additionally the knowledge of when cracking might be expected becomes an economic issue and is not necessary to insure safety. Consistent with this the fail-safe rule did not include a requirement to perform full scale fatigue testing or identify any special directed inspections to supplement normal maintenance. Compared to what safe-life required of both the applicant and their customers the attraction of fail-safe is easily understood.

Although the fail-safe option was widely applied there was an underlying concern by many relative to its effectiveness in the long term. Maxwell [15] discussed this and considered “…some of the potential dangers that have developed in the application of the fail-safe approach over the years”. One of the biggest concerns was the eventual loss of fail-safety as the airplane ages and normal fatigue cracking becomes more and more probable. This is because a structures’ fail-safe characteristics are dependent on successful redistribution of load from failed or partially failed elements to intact surrounding structure. In many cases success is

A S I P 2 0 0 5

⊥ !

8

dependent on the surrounding structure being in near pristine condition. At some point in the life of the structure normal fatigue wear out makes this an unrealistic expectation. It is at this point that the fail-safe concept can no longer be relied on for safety.

The concern over long term reliance on fail-safety for continued airworthiness became more widespread within the aviation community as the jet transports that had been originally certified using the fail-safe option started to approach their design service goals. Ultimately this concern is what prompted the Civil Aviation Authority (CAA), in the United Kingdom (UK), in the early 1970’s to limit the operational life of large transport aircraft that had been certified as fail-safe. For example all Boeing 707 airplanes in UK registry were limited to 60,000 flight hours. The British Authorities also announced that for these aircraft to be allowed to operate beyond the specified life limits something more would need to be done.

In the midst of all the concern over the long term effectiveness of fail-safety an accident occurred that is considered by many to be the key event that served to solidify and accelerate changes in civil aviation requirements and policies dealing with the threat of metal fatigue in primary airframe structures. This was the crash of a Boeing 707-300C, operating under British registry, during final approach to Lusaka airport on May 17, 1977. The details of this accident and its impact on airworthiness requirements have been discussed by Eastin and Bristow [16]. An extremely thorough accident investigation concluded that the crash was a consequence of the loss of the horizontal stabilizer due to undetected fatigue and subsequent failure of the aft upper spar chord. This was in spite of the fact that the design had been certified in accordance with the fail-safe rules of CAR 4b.270 by both the FAA and CAA. This is a classic example of structure certified as fail-safe that did not, in service, fail in a safe manner. The failure of fail-safety in this case was due to insufficient attention given to detectability, a lack of understanding of the external loads and incorrect assumptions made about the fatigue and residual strength characteristics of the structure.

As noted previously the Lusaka accident hastened major changes to civil aviation requirements that were already being considered. Consideration was already being given to requiring special directed inspections for fatigue cracking based on quantified crack growth and residual strength characteristics. This became know as the “damage tolerance” approach. Guidance for the use of this approach for protecting the safety of older aircraft was published by the FAA in [17]. Manufacturers of the fail-safe certified aircraft previously noted voluntarily followed the guidelines and produced Supplementary Inspection Documents (SIDs) that were mandated by Airworthiness Directives starting in the mid 1980’s.

Consistent with the change of philosophy for continued airworthiness for older aircraft was a change to the certification requirements for new type designs. Amendment 45 to part 25 was issued in 1978 [4]. This revision removed the fail-

A S I P 2 0 0 5

⊥ !

9

safe option completely and added damage tolerance as the approach that must be used unless shown to be impractical. In the past there has been some debate on whether or not fail-safety was actually removed and if so whether or not it was intentional. Some light is shed on these questions by the response to a comment on proposed deletion of the parenthetical expression “fail-safe” from the heading of section 25.571(b). The response is included in [18] and is as follows;

“….Fail-safe and damage-tolerance are not synonymous terms. Fail-safe generally means a design such that the airplane can survive the failure of an element of a system or, in some instances one or more entire systems, without catastrophic consequences. Fail-safe, as applied to structures prior to Amendment 25-45, meant complete element failure or obvious partial failure of large panels. It was assumed that a complete element failure or partial failure would be obvious during a general area inspection and would be corrected within a very short time. The probability of detecting damage during routine inspections before it could progress to catastrophic limits was very high. Damage-tolerance, on the other hand, does not require consideration of complete element failures or obvious partial failures, although fail-safe features may be included in structure that is designed to damage-tolerance requirements. A part may be designed to meet the damage-tolerance requirements of Sec. 25.571(b) even though cracks may develop in that part. In order to ensure that such cracks are detected before they grow to critical lengths, damage-tolerance requires an inspection program tailored to the crack progression characteristics of the particular part when subjected to the loading spectrum expected in service. Damage-tolerance places a much higher emphasis on these inspections to detect cracks before they progress to unsafe limits, whereas fail-safe allows the cracks to grow to obvious and easily detected dimensions.”

The author believes that this response underscores the fact that the “Fail-safe” option was removed and indicates that it was done intentionally.

In summary what motivated the FAA to adopt their damage tolerance requirements was the conclusion that the fail-safe approach as applied had not resulted in the level of safety desired. Specifically there had been a lack of attention given to making sure the detectability assumed was consistent with the actual crack growth and residual strength attributes of the structure. This was addressed by replacing the fail-safe requirements with damage tolerance requirements and retaining safe-life as a contingency approach if damage tolerance is shown to be impractical.

As previously noted there are watershed events that are commonly referenced as providing the major impetus for the adoption of damage tolerance requirements. For the USAF this was the F-111 accident and for the FAA it was the Lusaka accident.

A S I P 2 0 0 5

⊥ !

10

It is of interest to note how different these accidents were. Table 1 below summarizes some of the details for each. About the only thing they had in common was that metal fatigue was a factor and even then the categories were different.

In the case of the F-111 it was anomalous fatigue that resulted in the wing separation. As noted by Lincoln [9] the USAF could not reproduce the failure in the laboratory and did not see such a failure on another F-111 aircraft.

In the case of Lusaka unexpected normal fatigue lead to separation of the horizontal stabilizer. As noted by Eastin and Bristow [16] the failure was reproduced in the laboratory and the fatigue nucleation site was retrospectively identified as a fatigue critical location representative of the basic design. This was further validated by post accident inspections that detected cracks in the same local area on 7% of the fleet.

This again illustrates the fundamental differences between the USAF and FAA experience with fatigue and helps to explain some of the differences that exist in their approaches to fatigue that are reflected in their requirements.

THE REQUIREMENTS

At a high level there are some similarities between the USAF and FAA damage tolerance requirements. Both are applicable to new aircraft designs and compliance with them requires the quantification of crack growth and residual strength

Table 1. Comparison of Watershed Events












LusakaF-111












LusakaF-111

A S I P 2 0 0 5

⊥ !

11

characteristics. Additionally, when this is done analytically, fracture mechanics based analysis tools are used. However, at the detail level, there are significant differences. Some of the details are discussed below and Table 2 provides a summary comparison .

USAF FAA

Primary motivation for: Safe-life approach inadequate

Fail-safe approach inadequate

Applicability: New airplane design – safety of flight structure

New airplane design – safety of flight structure

Objective: Safety during service life Safety indefinitely

Outcome: Design attributes (& in-service inspections as required)

Maintenance actions (In-service inspections expected)

Incorporation philosophy: Replace safe-life Replace fail-safe

Threats addressed:

Normal fatigue No (Addressed by durability requirements) Yes

Anomalous fatigue Yes Yes

Unexpected normal fatigue No No

Provision for alternate approach if damage tolerance impractical? No Yes (Safe-life)

Prescribed requirements:

Design concept (i.e. single or multiple load path) No No

Initial crack sizes Yes No

In-service detectable crack sizes Yes No

Cracking scenarios Yes No

Minimum crack growth life Yes No

Inspection intervals Yes No

Residual strength Yes Yes

Table 2 Comparison of USAF and FAA Damage Tolerance Requirements

A S I P 2 0 0 5

⊥ !

12

USAF

A comprehensive description of the USAF damage tolerance requirements along with a discussion of the supporting rationale has been provided by Wood [19]. The following is limited to a brief overview.

The scope paragraph of [1] states that,

“This specification contains the damage tolerance design requirements applicable to airplane safety of flight structure. The objective is to protect the safety of flight structure from potentially deleterious effects of material, manufacturing and processing defects through proper material selection and control, control of stress levels, use of fracture resistant design concepts, manufacturing and process controls and the use of careful inspection procedures.”

It is clear that the subject requirements are intended to directly impact the design of the structure. For example these requirements, with some modifications, were imposed on the C-17A airplane and the design was significantly impacted as discussed by Eastin and Pearson [20]. In a number of areas on the C-17A the requirements had a direct affect on material selection, allowable stress levels and in some cases structural arrangement. Levying such requirements serves to insure that a minimum level of inherent robustness or tolerance to damage is achieved.

The manufacturer is given some latitude relative to design concept. Single or multiple load path designs are allowed however single load path structure without crack arrest features can only be qualified as “slow crack growth” while multiple load path structure can be qualified as either “slow crack growth” or “fail-safe”. Wood [19] offers some explanation for the allowance of this option when he notes that, “It should be emphasized that while the “Fail Safe” concept appears to offer a larger degree of safety, it is the intent of the new criteria that structure qualified to either category have equal safety”.

Once the design concept is identified the detail requirements are very prescriptive and specify certain crack growth and residual strength attributes that the structure must possess. Proposed designs not possessing such attributes must be changed. In general the requirements specify that a structure must exhibit a minimum amount of crack growth life, assuming an initial prescribed crack array, before its strength falls below a prescribed level. Additionally assumptions to be used about the cracking scenario are also prescribed.

The initial cracking array and subsequent cracking scenario has been characterized as representing an “escape” or “rogue” event. It is meant to approximate the occurrence of an unintentionally introduced defect or flaw in an otherwise nominal structure. Using the author’s fatigue categories this would be

A S I P 2 0 0 5

⊥ !

13

considered anomalous fatigue. It is something that is expected to be rare but possible. Swift [6] articulated this when he wrote “….the “Rogue Flawed Aircraft” needs to be accounted for. This is the one or two aircraft in the fleet having some kind of initial manufacturing damage not representative of the rest of the fleet”.

In accordance with prescribed initial condition assumptions, the initial cracking array, if holes are present in the structure, would typically consist of a singular .05” crack located on one side of the most critical hole along with .005” cracks located in all other holes. The subsequent cracking scenario to be assumed is also specified and addresses the growth of the “rogue” .05” crack and also growth of the .005” cracks. Assumptions to be made relative to continuing growth patterns are also included in the requirements.

The USAF requirements also allow the manufacturer some latitude relative to in-service inspection. Under certain circumstances it may be assumed that in-service inspections will occur. If this is done the requirements prescribe what size cracks should be assumed subsequent to inspection and how much crack growth life the structure must possess with those cracks present.

In all cases the structure must always retain a minimum level of strength. Residual strength requirements are specified as a function of the level of inspection required to detect the postulated cracking.

The USAF requirements leave little undefined or open to interpretation. They are intended to insure that the structure has a minimum amount of robustness relative to defects that might be unintentionally introduced. To achieve this the structure must possess specified crack growth and residual strength attributes. In this context they are design requirements.

FAA

It has been and is the general policy of the FAA not to dictate design. This is the case with the damage tolerance requirements and this was clarified in a response to public comments to the requirements as originally proposed. The Notice of Proposed Rulemaking [21] included text that could be interpreted to mean that the design had to have certain intrinsic properties. Several comments objected to the wording contending that it would impose an absolute requirement that would be impossible to comply with. In response to these comments in the Discussion of Specific Comments section of [4] the FAA noted that, “The purpose of the proposal was to establish an evaluation requirement rather than an absolute requirement for the strength, detail design, and fabrication of the structure”. Consistent with this the wording was changed for clarification.

Like the USAF requirements the FAA requirements leave it up to the manufacturer to decide on the design concept to be used. Both single and multiple

A S I P 2 0 0 5

⊥ !

14

load path structural designs are allowed. This was made clear in the Preamble Information section of [21]. Here it states that:

“… the applicant would be allowed to apply the damage-tolerance approach to both single load path and multiple load path structure. The FAA believes the applicant can, by sufficient analysis and testing, establish that a single load path structure has sufficiently slow crack growth properties so that, if a crack were to develop, it would be discovered during a properly designed inspection program.”

It is worth noting that the preceding statement is consistent with the remarks by Wood [19] that were previously referenced. It appears that at the time the USAF and FAA damage tolerance requirements were adopted there was the same philosophy regarding the merits of single load path versus multiple load path structure. It was believed that either design concept could be made equally as safe and therefore the choice was left up to the manufacturer.

The requirements state that fatigue from all potential sources must be considered. In terms of the author’s fatigue categories this would include both normal and anomalous fatigue. The requirements also state that crack growth and residual strength evaluations must be performed and based on the results inspections must be established unless shown to be impractical.

The detail requirements are very objective for the most part. There are no specific requirements relative to such things as initial crack sizes, in-service detectable crack sizes, inspection intervals or minimum acceptable crack growth life. The exception is residual strength. Levels of strength that must be maintained are specified.

In summary the FAA requirements leave many details undefined and open to interpretation. They are intended to result in the establishment of in-service inspections that will detect fatigue cracking from any potential source before the strength of the structure falls below prescribed levels. There is no design concept specified. There are no specific attributes that the structure must possess. There is only a requirement to perform an evaluation and establish inspections unless the applicant demonstrates that inspections are impractical. If it is determined that inspections are impractical the safe-life approach is allowed and safety is insured by retirement instead of inspection.

FAIL-SAFETY

As previously discussed fail-safe was completely removed from the 14 CFR part 25 requirements with amendment 45 in 1978 [4]. However it is still worth some discussion. This is because the subject of fail-safety has at times been a contentious issue and this has been due in part to differing views of what fail-safe “is”, “was” or

A S I P 2 0 0 5

⊥ !

15

“should be”. The intent of the discussion that follows is to clarify what the FAA requirements were and what that USAF requirements are.

As is the case with the damage tolerance requirements previously discussed similarities exist between the two different flavors of fail-safety when viewed at a high level. In both cases fail-safety was/is included as an optional approach and was/is associated with multiple load path structure. Additionally both versions of fail-safety share a similar requirement that the structure must retain a relatively high level of strength with a relatively large amount of damage present. Beyond that there are significant differences. Some of these differences are discussed below and Table 3 provides a summary comparison.

USAF FAA Pre-Amd 45

Included as Optional Approach: Yes Yes

Associated with Multiple Load Path Structure: Yes Yes

Outcome: Design attributes (Plus in-service inspections as

required) Design Attributes

Prescribed requirements:

Initial crack size for intact structure Yes No

Damage size after primary failure*

Only for “fail-safe crack arrest” structure No

In-service detectable crack sizes Yes No

Cracking scenarios before and after primary failure*

Yes No

Minimum crack growth life before and after primary failure*

Yes No

Inspectability of primary failure*

Determined by manufacturer.

Obvious during normal maintenance.

Residual strength Yes Yes

* Stable load path failure or crack arrest.

Table 3. Comparison of USAF and FAA Fail-Safe Requirements

A S I P 2 0 0 5

⊥ !

16

USAF

Fail-safety is fully integrated into USAF damage tolerance requirements as an approach that can be used for qualification of certain types of structure. The other approach is referred to as “slow crack growth” and can be used for all types of structure. In the context of the USAF requirements fail-safe is a design concept that must be matched with a degree of inspectability to identify a damage tolerance category. Detail requirements are prescribed, as previously discussed in the section on “Requirements”, and depend on the category.

If the fail-safe option is selected there are prescribed requirements for both the intact structure and the structure subsequent to a load path failure or crack arrest. This makes qualification of structure as fail-safe relatively onerous and since the selection of category is left up to the manufacturer it has been avoided in the past. It is noted in [22] that, at the time of publication of that document, there were no aircraft in the USAF inventory that had been originally designed and qualified to the USAF fail-safe requirements. The author believes that this still holds true today.

FAA

Prior to amendment 45 the fail-safe approach was included as an option to the safe-life approach. The requirements were include in 14 CFR, section 25.571, paragraph (c) Fail safe strength, where it stated the following:

“It must be shown by analysis, test, or both, that catastrophic failure or excessive structural deformation, that could adversely affect the flight characteristics of the airplane, are not probable after fatigue failure or obvious partial failure of a single principal structural element. After these types of failure of a single principal structural element, the remaining structure must be able to withstand static loads corresponding to the following:………”

The specified static loads were associated with design envelope type conditions.

Swift [6] succinctly summarized the generally accepted approach used for compliance with the above requirements when he wrote the following:

“Generally, manufacturers satisfying the requirements under the fail-safe concept merely substantiated the structures for failure of single principal elements under static loading conditions. Although it was recognized that inspections were necessary there were no specific requirements to determine safe inspection periods based on crack growth or remaining life of secondary structure in the event the primary member failure was not immediately obvious.”

A S I P 2 0 0 5

⊥ !

17

Swift [23] has also noted that “……reliance was placed completely on the correctness of the arbitrary selection of sites and the final size of damage chosen for residual strength substantiation”. Goranson [24] speaking to this same issue wrote that, “This would often lead to residual strength demonstration by analysis of defined obvious failures rather than showing that all the partial failures with insufficient residual strength were obvious”. What constituted a “fatigue failure or obvious partial failure of a single principal structural element” was a detail to be negotiated with the FAA and varied from manufacturer to manufacturer and even from airplane model to model for the same manufacturer. Table 4 below illustrates this. The information was taken from fail-safe reports that were submitted to the FAA to demonstrate compliance with the fail-safe requirement for basic fuselage shell structure.

Airplane Model

“Fatigue failure or obvious partial failure of a single principal structural element”

Skin Crack Size

DC-101 2 Frame bay skin crack with central crack stopper failed. 40”

DC-9 1 Frame bay skin crack. 20”

B737 1 Frame bay skin crack. 20”

B727 1 Frame bay skin crack. 20”

B747 12” skin crack. 12”

L10112 1 Crack stopper bay skin crack with center frame failed. 20”

If the fail-safe option was chosen by the manufacturer it was only necessary to submit a fail-safe report to the FAA that demonstrated by analyses and supporting tests that the structure was sufficiently “fail-safe”. There was no requirement to perform any fatigue testing or analysis or submit any corresponding documentation. Fortunately the manufacturers typically performed their own fatigue analyses and tests but it was not subject to review or approval by the FAA.

The past FAA fail-safe requirement can be best characterized as a design rule that resulted in multiple load path designs that could tolerate single element failures

1. Crack stoppers located under frames.2. Crack stoppers located between frames

Table 4. Examples of Certified Fail-Safe Capability for Fuselage Structure in

Longitudinal Direction

A S I P 2 0 0 5

⊥ !

18

or relatively large but somewhat arbitrary partial failures. It has its origins in the belief that structure could be designed such that it would always annunciate its distress loudly and clearly before anything catastrophic occurred. Given this it was reasoned that fatigue cracking, in itself, was not a safety issue since it would always be detected and corrected in the normal course of operation, before a catastrophic event could occur.

COMMENTS IN CONCLUSION

The use of the same words for different things can lead to confusion and needless debate. This has been the case with the words “damage tolerance” and “fail-safe”. It is hoped that this paper provides some clarification relative to USAF and FAA part 25 requirements for new airplane designs. Some of the more significant differences are summarized below.

For the USAF “damage tolerance” is a design philosophy that must be followed that results in a design that possesses prescribed crack growth life and residual strength attributes. It was adopted to address the threat of anomalous fatigue and is supplemental to other requirements that address normal fatigue.

For the FAA “damage tolerance” is a fatigue management strategy that must be used unless shown to be impractical. It relies on inspections to detect fatigue cracking before it becomes dangerous. If shown to be impractical another strategy is allowed.

For the USAF “fail-safe” is a design concept that may be selected for qualification of a design as damage tolerant. The level of inspection associated with it must be determined by the manufacturer and can range from obvious during flight to requiring a special directed depot level inspection. Structure qualified as fail-safe must also meet other fatigue requirements.

For the FAA “fail-safe” was a fatigue management strategy option that relied on designing the structure to crack in a manner that would be obvious during the course of normal maintenance and therefore detected and repaired before it became dangerous. Structure qualified as fail-safe did not need any special directed inspections and there were no other fatigue requirements that had to be met.

REFERENCE LIST

(1) Mil-A-83444 (USAF), Airplane Damage Tolerance Requirements, July 1974.

(2) Code of Federal Regulations, Title 14, Chapter 1 – Federal Aviation Administration Department of Transportation.

A S I P 2 0 0 5

⊥ !

19

(3) Eastin, R.G., A Critical Review of Strategies Used to Deal with Metal Fatigue, Proceedings of the 22nd Symposium of the International Committee on Aeronautical Fatigue, Lucerne, Switzerland, pp 163-187, 2003.

(4) FAR Final Rule, Federal Register: October 5, 1978 (Volume 43, Number 194), 14 CFR Part 25 (Docket No. 16280; Amendment No. 25-45).

(5) Eastin, R.G., Strategies for Ensuring Rotorcraft Structural Integrity, North Atlantic Treaty Organization Research and Technology Organization Meeting Proceedings 24 (RTO-MP-24), Corfu, Greece, April 1999.

(6) Swift, T., Verification of Methods for Damage Tolerance Evaluation of Aircraft Structures to FAA Requirements, Proceedings of the 12th Symposium of the International Committee on Aeronautical Fatigue, Toulouse, France, 1983.

(7) Tiffany, C.F., Durability and Damage Tolerance Assessments of United States Air Force Aircraft, Proceedings of the 9th Symposium of the International Committee on Aeronautical Fatigue, Darmstadt, Germany, pp. 4.4/1-4.4/31, 1977.

(8) Lincoln, J.W., Life Management Approach for USAF Aircraft, AGARD Conference Proceedings 506.

(9) Lincoln, J.W., Significant Fatigue Cracking Experience in the USAF, Proceedings of the 22nd International Congress of Aeronautical Sciences, August 2000.

(10) Lincoln, J.W., Damage Tolerance – USAF Experience, Proceedings of the 13th Symposium of the International Committee on Aeronautical Fatigue, Pisa, Italy, 1985.

(11) Military Specification, Mil-A-008866A(USAF), Airplane Strength and Rigidity Requirements, Repeated Loads and Fatigue, 31 March 1971.

(12) Military Specification, Mil-A-008867A(USAF), Airplane Strength and Rigidity Ground Tests, 31 March 1971.

(13) Swift, T., Damage Tolerance in Pressurized Fuselages, 11th Plantema Memorial Lecture, Proceedings of the 14th Symposium of the International Committee on Aeronautical Fatigue, June 10-12, 1987.

(14) Civil Aeronautics Board, Airplane Airworthiness Transport Categories, Part 4b-3 paragraph 270, March 1956.

(15) Maxwell, R.D.J., Fail-Safe Philosophy: An Introduction to the Symposium, Proceedings of the 7th International Committee on Aeronautical Fatigue Symposium, London, England, July 1973.

A S I P 2 0 0 5

⊥ !

20

(16) Eastin R.G., Bristow, J.W., Looking at Lusaka’s Lessons, Proceedings of the 2003 USAF Aircraft Structural Integrity Program Conference, December 2-4, 2003.

(17) FAA Advisory Circular No. 91-56, Supplemental Structural Inspection Program for Large Transport Category Airplanes, May 6, 1981.

(18) FAR Final Rule, Federal Register: July 20, 1990 (Volume 55, Number 140), 14 CFR Part 25 (Docket No. 24344; Amendment No. 25-72).

(19) Wood, H.W., Application of Fracture Mechanics to Aircraft Structural Safety, Engineering Fracture Mechanics, Vol. 7, 1975, pp. 557-564, Pergamon Press.

(20) Eastin, R.G., Pearson, R.M., C-17A Structural Development and Qualification, presented at 36th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, April 10-12, 1995, New Orleans.

(21) FAR Notice of Proposed Rulemaking, Federal Register: August 15, 1977 (Volume 42, Number 157), 14 CFR Part 25 (Docket No. 16280; Notice No. 77-15).

(22) Joint Service Specification Guide, JSSG-2006, Aircraft Structures, Department of Defense, 30 October 1998.

(23) Swift, T., “Damage Tolerance Technology – Phase I”, FAA Class Notes, 1999.

(24) Goranson, U.G., Damage Tolerance Facts and Fiction, 14th Plantema Memorial Lecture, 17th Symposium of the International Committee on Aeronautical Fatigue, June 9, 1993.