Aging Aircraft 2003 MMPDS Presentation

8/10/2019 Aging Aircraft 2003 MMPDS Presentation

1/32

Equipment Developmentand Mechanical Systems

10/13/2003 1aeromat-2001-pmp.ppt

Development of MMPDS HandbookAircraft Design Allowables

September 10, 2003

7th Joint DoD/FAA/NASA Conference on

Aging Aircraft, New Orleans, LA

by

Richard C. Rice and Randall J. GoodeStructural Integrity Projects Office

Battelle; Columbus, OH

John G. Bakuckas, Jr.Federal Aviation Administration

William J. Hughes Technical Center

Atlantic City International Airport, NJ

Steven R. ThompsonU. S. Air Force Research Laboratory

Materials and Manufacturing Directorate

Wright Patterson Air Force Base, OH


2/32


10/13/2003 2aeromat-2001-pmp.ppt

Presentation Outline

Acknowledgements

MIL-HDBK-5 Historical Milestones and Background Transition from MIL-HDBK-5 to MMPDS

Design Allowables Terminology

Current MMPDS Design Allowable Applications and Issues Impact of Skewed Mechanical Properties Reconciling S-Basis and A- and B-Basis Properties Worldwide Coordination of Aircraft Design Allowables

Statistically-Based Fastener Design Allowables Statistical Treatment of Notch Effects on Fatigue Allowables Statistically-Based Crack Propagation Design Limits

Summary of MMPDS Activities and Issues


3/32

Equipment Developmentand Mechanical Systems10/13/2003 3aeromat-2001-pmp.ppt

Acknowledgements Government

Federal Aviation Administration

Dr. John Bakuckas, FAA Technical Center, AAR-431 Mr. Robert Eastin, FAA, National Resource Specialist Mr. Jon Hjelm, FAA, Airframe Certification Engineer

Air Force Mr. Steve Thompson, AFRL/MLSC Mr. Neal Ontko, AFRL/MLSC

Army

Walter Roy, APG


4/32


Acknowledgements Industry

MMPDS Industrial Steering Group (ISG)

Mr. Steve Fantle, Boeing (Chair) Mr. Pete Brouwer, Alcoa (Vice Chair)

The aircraft metallic material suppliers and users that haveparticipated in the ISG since its inception in 1997:-Alcoa - Bell Helicopter - Boeing- Cessna - Corus Aluminum - Howmet

- Lockheed - McCook - Northrop Grumman

- Pechiney - Textron Aero Fasteners - Universal Alloy The many dedicated contributors, reviewers, and users of

MMPDS (formerly MIL-HDBK-5) throughout the world


5/32


MIL-HDBK-5 Historical Milestones

1937First published as ANC-5 (Army-Navy-Commerce,Handbook 5)

1954Battelle began coordination of the Handbook undercontract with the Air Force

1956Converted from ANC-5 to MIL-HDBK-5

1971 Incorporated detailed guidelines for statistical analysisof data

1985 Incorporated Weibull analysis methods to account forskewed strength distributions

1988Completed replacement of all constant-life fatiguediagrams with statistically based equivalent stress plots


6/32


Transition from MIL-HDBK-5 to MMPDS 1997

Substantial reductions in Air Force funding available for Handbook

coordination announced Battelle, in collaboration with AF and FAA, set up Industr ial SteeringGroup (ISG) to help support Government Steering Group (GSG) andongoing Handbook coordination

1999 Initial version of Handbook numeric databases released to ISG/GSGmembers

2000

All Air Force funding for Handbook coordination lost After a brief hiatus FAA picked up transitional funding through AFcontract

Initial, internet-accessible version of Handbook design allowablessoftware released to ISG/GSG members


7/32


Transition from MIL-HDBK-5 to MMPDS

2002 FAA established new technical coordination contract with Battelle to

continue longstanding AF legacy of MIL-HDBK-5 technicalcoordination support

FAA changed name of Handbook to Metallic Material PropertiesDevelopment and Standardization (MMPDS) Handbook

2003 Final version of MIL-HDBK-5 published, Revision J First version of MMPDS published, Revision 01 Public version of ISG website established mmpds.org

2004 First major revision of MMPDS-01 to be published MIL-HDBK-5J will be designated non-current


8/32


MMPDS Design Allowables Terminology


9/32


MMPDS Design Allowables Terminology

A- and B-Basis design values are statistically based and, bydefinition, must meet or exceed specific requirements Primary Structure: A-Basis >= 99% exceedance, 95% confidence Secondary Structure: B-Basis >= 90% exceedance, 95% confidence

By comparison, S-Basis design values do not have reliable

statistical significance Generally calculated to approximate an A-Basis value to cover earlyproduction

A-Basis Value in MMPDS are defined as lower of

T99 (99% exceedance, 95% confidence) S-basis Value


10/32


Handling of Skewed Mechanical Property

Distributions Key Issue

0

50

100

150

200

250

300

350

400

450

500

75 80 85 90 95 100 105 110 115 120

Normalized Streng th

RelativeFrequenc

yofOccurren

ce

Skew ness = -1.00

Skew ness = -0.60

Skew ness = -0.20

Normal

Skew ness = 0.20

Skew ness = 0.60

Skew ness = 1.00

Mode = 100

Std. Dev. = 5.0

Critical zone for design allowable calculations


11/32


Observed Skewness in Actual Metallic Material

Receiving Inspection Data

In 1980s MIL-HDBK-5 coordination group examined 57metallic material tensile and yield strength data sets.

Sample sizes ranging from 25 to over 8000 observations.

Results revealed sample skewness levels ranging from as lowas -1.0 to as high as 1.0.

Only 26% of these data sets displayed insignificant skewness. Remaining 74% of data sets were significantly skewed with

either a long lower tail (negative skewness) or a long upper tail(positive skewness).

Note: Possible effect of secondary variables, such as thicknesson the material properties, was eliminated before performingthese calculations.


12/32


Use of Normal Statistics on Skewed Properties can

Lead to Significant Errors in A-Basis Properties

-6.00%

-4.00%

-2.00%

0.00%

2.00%

4.00%

6.00%

-1.00 -0.80 -0.60 -0.40 -0.20 0.00 0.20 0.40 0.60 0.80 1.00

True Skewness

Estima

tedConservatis

mi

nDesignAllowabl

n = 20

n = 30

n = 50

n = 100

n = 300

n = 1000

Sample Size


13/32


Additional Observations Regarding Effects of

Skewness on Minimum Mechanical Properties

Ability to accurately quantify skewness in mechanical

properties is low with small samples (< 30 observations) Influence of skewness on accuracy of B-basis properties is

about one-half that for A-basis properties

Recent adoption of Pearson Type III procedures in MMPDS toaccount for skewness has simplified calculations comparedwith older 3-parameter Weibull procedure Pearson calculation depends only on sample size, mean, standard

deviation, and skewness Use of 2-parameter Weibull does not allow accurate

representation of skewness in most metallic materials


14/32


Reconciling S-Basis and A- and B-Basis

Design Allowable Properties

All materials included in MMPDS must be covered by a publicspecification to ensure requirements for production of thematerial do not change over time.

As a result, most materials in the Handbook are covered byAerospace Material Specifications (AMS) published by the

Society of Automotive Engineers (SAE). However, the S-basis value defined in a public specification

for an aerospace material does not have a known statisticalsignificance.

The specification limit typically represents a lot-releasestrength level, above which a supplier may sell the material toa user.


15/32




As a result it has been standard practice within MIL-HDBK-5for many years to define an A-basis design allowable as thelowest of either the statistical T99 value or the S-basis value.

This sometimes has led to a situation where the A-basis valueshown in the Handbook has fallen well below the statistically

computed T99 value. However, in these same situations the estimated B-basis

design allowable has not been downgraded, creating anartificially large statistical difference between the published A-

and B-basis values. This approach has led to two possible scenarios, neither of

which is particularly desirable.


16/32




Scenario # 1: Material properties remain constant over time

Effect # 1: The airframe designer must use an artificially lowA-basis value in non-redundant primary structure, whichtranslates into excess weight in the aircraft.

Scenario # 2: The disparity between the A- and S-basisvalues allows degradation over time in actual T99 values fromthe old T99 value to the S-basis value.

Effect # 2: The true T90 value would likely decrease (and notbe documented in the Handbook), leading to unconservativedesign allowables for redundant primary structure.


17/32




Increased, close collaboration between MMPDS and

SAE/AMS coordination groups has been very helpful SAE/AMS has agreed to pursue updated

specifications in situations where A-basis value is

shown to fall well above S-basis value Bob Steffen at Raytheon has provided key link from AMS

to MMPDS coordination groups

Jana Jackson at Battelle has provided key link fromMMPDS to AMS coordination groups


18/32


Worldwide Coordination of Aircraft Design

Allowables There are only two widely accessible, approved sources of

design allowable properties for aircraft and aerospacematerials MMPDS (formerly known as MIL-HDBK-5) and ESDU 00932.

The first of these is the de facto standard in the United States,while the second is the de facto standard in Europe and GreatBritain.

Recent examinations of these two documents have shown

significant differences in the guidelines for collecting andanalyzing strength data for computation of design allowableproperties.


19/32


Worldwide Coordination of Aircraft Design

Allowables

In the short-term, the problem is limited becauserepresentatives of both coordination groups have found only afew cases where design allowables are published in bothdocuments for identical alloys, tempers, and product forms.

In the long-term, both groups have agreed to increase their

interaction to eventually resolve or reconcile differences inapproaches between the two documents.

In recent years, worldwide coordination of MMPDS has also

been enhanced with participants from Japan, Russia,Canada, and South America.


20/32


Statistically-Based Fastener Design

Allowables

Design properties for fasteners in MMPDS (formerly

MIL-HDBK-5) have traditionally been estimatedultimate and yield strength design values, with nostatistical basis

In 2002 MMPDS coordination group approved a newstatistical procedure for determination of B-basisdesign allowables for fastener yield and ultimatestrength

In 2003 focus has moved to approval of a sunsetclause, which will require periodic (~ every 7 year)validation of MMPDS fastener allowables


21/32



Allowables B-basis lower-limit design curves will be based on the

following regression equation:

where P = test load,

D = fastener diameter

t = sheet thicknessln = natural logarithm

Ai

= regression coefficients

+

+=

D

tA

D

tAA

D

Pln2102


22/32


10/13/2003 22aeromat-2001-pmp.ppt


Allowables

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

4.50

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80

t / D

Py

/D

2,x10-4

Individual Data

Group Averages

New Average

Old Average

T90 Allowable

Aluminum xxxxxxx Flush

Shear Head Solid Rivets

in Clad 2024-T3 Sheet


23/32


10/13/2003 23aeromat-2001-pmp.ppt

Development of Statistically-Based Crack

Propagation Design Limits Crack growth data in MMPDS are currently represented by

visually best fit mean curves that have no quantified statistical

significance. It has been recognized for a number of years that it would be

advantageous within MMPDS to identify mean and upper-bound crack growth trends through quantitative procedures: To account for effects of stress ratio Represent variability about mean trends and in sufficient detail, that

individual organizations could construct their own statisticallybaseddesign limits

A similar statistical approach was adopted in MMPDS for load andstrain control fatigue data over 15 years ago.


24/32


10/13/2003 24aeromat-2001-pmp.ppt


Propagation Design Limits For example, an inverse hyperbolic tangent model was proposed over

30 years ago by Collipriest.

where Ko = lower asymptote,Kc = upper asymptote,

Kmax = maximum stress intensity,

R = stress ratio (stress ratios less than zero set to 0),

m = optimized exponent between 0 and 1 (to accountfor stress ratio effects, and

C1, andC

2= regression coefficients.

log tanhlog[ / ( ( ) )

log( / )

maxda

dNC C

K K K R

K K

c o

m

o c

= +

1 2

1

21


25/32


10/13/2003 25aeromat-2001-pmp.ppt


Propagation Design Limits Example: Ti-6Al-4V Castings

> 40 Specimens

> 4400 da/dN Measurements

Effectively Consolidates CrackGrowth Data for Stress Ratios

from 0.50 to 0.80 Accounts for Upper and Lower

Limit Effects

Includes Approx. +/- 2 SigmaLimits on Mean Curve1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

1 10 100 1,000

Effective Stress Intensity Range, ksi-in0.50

da

/dN

,inc

hes

/cyc

le

Lab Air, Base, R = 0.10


Regression Mean+/- 2 Sigma Bounds

Lab Air, Base, R = -0.50



26/32


10/13/2003 26aeromat-2001-pmp.ppt

Prediction of Notch Effects on Fatigue Life Equivalent stress procedures in MMPDS allow quantitative

definition of mean trends and statistical variability.

However, comprehensive procedures have not been available toconsolidate load-control fatigue data generated on unnotched andnotched specimen geometries.

Therefore, separate analyses and data presentations have been

made for unnotched data and each available notch concentration. This has limited the usefulness of the information for actual notch

concentrations different from those in the Handbook.

In a few cases it has also led to inconsistencies where longerfatigue lives have been predicted for more severe notchconcentrations.


27/32


10/13/2003 27aeromat-2001-pmp.ppt

Prediction of Notch Effects on Fatigue Life Recent analytical efforts to improve interpretation and

statistical representation of notch effects on fatigue life have

produced promising results, as will be shown for 2024-T3sheet.

The approach accounts for crack initiation, Ni, and crackgrowth, Np, cycles in total life, Nf Ni/Nfratios near unity for unnotched specimens Ni/Nfratios below 0.01 for some sharp notches

Predicted cycles to crack initiation (or nucleation) based on a

local strain analysis at the notch tip Neuber analysis used to estimate local stresses and strains

from nominal stress conditions via cyclic stress-strain curve


28/32


10/13/2003 28aeromat-2001-pmp.ppt

Prediction of Notch Effects on Fatigue Life

0.001

0.010

0.100

1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 1.E+08

Estimated Crack Initiation Cycles

Equ

iva

lent

Stra

in,

in./

in.

Kt = 1.00Kt = 1.50Kt = 2.00Kt = 4.00Kt = 5.00Mean+/- 3 Sigma

2024-T3 Sheet, Load Control Fatigue


29/32


10/13/2003 29aeromat-2001-pmp.ppt

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 1.E+08Ac tual Total Life

PredictedT

otalLife

Kt = 1.00Kt = 1.50Kt = 2.00Kt = 4.00Kt = 5.00

Mean+/- 3 Sigma

Prediction of Notch Effects on Fatigue Life2024-T3 Sheet, Load Control Fatigue


30/32


31/32


10/13/2003 31aeromat-2001-pmp.ppt

Summary of Benefits of Ongoing MMPDS

Coordination Avoids duplication of effort between material suppliers, users and

certifiers responsible for establishing design allowables and maintaining

safe and reliable aircraft Promotes use of proven and standardized test methods

Ensures consistency in statistical basisof design allowables used bydifferent airframers

Ensures consistency in analysis proceduresused by material suppliersand users to develop design properties

Small aircraft material suppliers and userscan reduce material testing

burden to build reliable aircraft ALL material suppliers and usershave consistent avenue for building

and maintaining affordable, yet reliable aircraft structures


32/32


10/13/2003 32aeromat-2001-pmp.ppt

Thank You for Your Attention!

Questions?

Documents

Aging Aircraft 2003 MMPDS Presentation