Characterisation of Composite Materials in Railways for Structural Calculations

7/23/2019 Characterisation of Composite Materials in Railways for Structural Calculations

1/60

Deliverable D4.1 Characterization of composite material in railways for structural calculation - 1 -

Towards a REgulatory FRamework for the usE ofStructural new materials in railway passenger and freightCarbOdyshells

Grant Agreement no.: 605632

WP 4Characterization of composite material in railways forstructural calculation

Deliverable: D4.1

Due date of deliverable: M12

Submission date: 07/10/2014

Version: final

Project co-funded by the European Commission within the Seventh Framework

Program


2/60


REFRESCO Deliverable D4.1 was produced by AIRBUS DS and received

contributions from the following members of the consortium:

DLR for monolithic modelling procedure and failure criteria, fatigue modelling

CAF and CETEST (laboratory of CAF)

BOMBARDIER TRANSPORT for junctions (test matrix and modelling)

ALSTOM

This document should be referenced as:

REFRESCO- Characterization of composite material in railways for

structural calculation , Deliverable D4.1

Companies Status Names DatesDocument

issueVisas

WP4 Leader Lidia Joana ZUBIA 04/09/14 final

T4.1 leader Sylvain CLAUDEL 18/07/14 final

T4.1 contributor Jean Philippe LEARD 04/09/14 final

T4.1 contributor Frederic HALLONET 04/09/14 final

T4.1 contributor JankoKREIKEMEIER

04/09/14 final

cetest T4.1 contributor Mikel MURGA 04/09/14 final

T4.1 contributor David LENGERT 07/10/14 final

T4.1 contributor Patrick RICAUD 04/09/14 final


3/60


QUALITY CONTROL INFORMATION

Issue Date Descrip tion RevisingAuthorship

Draft Ed 1.1 11/04/2014 Draft version of REFRESCOD4.1 for TMT COMMENT

S. CLAUDEL /AIRBUS DS

Draft Ed 2 12/06/2014 Draft version of REFRESCOD4.1 for TMT COMMENT


Draft Ed 2.1 07/07/2014 Comments to DLR adds S. CLAUDEL /AIRBUS DS

Final forapproval

18/07/2014 Integration of BTs addsonjoints and finalization


Final forapproval

04/09/2014 Integration of BTs adds onjoints


Final forapproval

07/10/2014 Final comments and add toa conclusive chapter


DOCUMENT HISTORY

Issue Date Pages Comment

1 11/04/2014 All Initial issue

2 12/06/2014 All Draft Ed 2 includes CAF comments,

CETEST + AIRBUS DS adds2.1 07/07/2014 All Comments to DLR adds

final 04/09/2014 All Integration of BTs addson joints

Final 07/10/2014 All Final comments and add to aconclusive chapter

DISSEMINATION LEVEL

PU Public [X]

PPRestricted to other program participants (including the Commission

Services)

RERestricted to a group specified by the consortium (including the

Commission Services)

COConfidential, only for members of the consortium (including the

Commission Services)


4/60


TABLE OF CONTENTS

1. REFERENCE ............................................................................................................ 7

2. EXECUTIVE SUMMARY ........................................................................................... 7

3. GLOBAL CHARACTERISATION LOGIC FOR COMPOSITE MATERIAL ................. 8

4. STATISTICAL-BASED MATERIAL PROPERTIES .................................................. 10

5. PROPOSED TESTS MATRICES ............................................................................. 11

5.1 GENERAL CONSIDERATIONS.................................................................................... 11

5.2 STATIC PROPERTIES AND TESTS MONOLITHIC COMPOSITE (UD

LAMINA/LAMINATE) ....................................................................................................... 12

5.2.1 Screening test matrix ........................................................................................ 12

5.2.2 Qualification test matrix .................................................................................... 13

5.3 STATIC PROPERTIES AND TESTS - SPECIFICITIES FOR A SANDWICH

PANEL (UD LAMINA/LAMINATE) WITH CORE MATERIAL ........................................... 15

5.3.1 Screening test matrix ........................................................................................ 15

5.3.2 Qualification test matrix .................................................................................... 16

5.4 FATIGUE PROPERTIES AND TESTS ................................................................. 17

5.4.1 Fatigue characterization and test recommendations: ........................................ 18

5.4.2

Specificities for a sandwich panel with core material ......................................... 19

5.5 JOINTS ................................................................................................................ 20

6. MODELLING PROCEDURE .................................................................................... 24

6.1 STATIC MODELLING (*1) ................................................................................... 24

6.1.1 Monolithic composite ........................................................................................ 24

6.1.2 Specificities for a sandwich panel with core material ......................................... 32

6.1.3 Static Modelling of Joints .................................................................................. 37

6.2 FATIGUE MODELLING ........................................................................................ 39

7. CONCLUSION ........................................................................................................ 42

ANNEX 1- DETAILED TEST PROCEDURES DESCRIPTION ......44

ANNEX 2- STANDARDS FOR COMPOSITE TEST..52

ANNEX 3 COMPLEMENTARY REFERENCES..59


5/60


LIST OF TABLES

TABLE 1 LAMINATE AND STRUCTURAL ELEMENT - MECHANICAL TESTING, SCREENING

PROGRAMME ............................................................................................................................................................... 12

TABLE 2 LAMINATE AND STRUCTURAL ELEMENT - MECHANICAL TESTING, QUALIFICATION

PROGRAMME ............................................................................................................................................................... 13

TABLE 2 LAMINATE AND STRUCTURAL ELEMENT - MECHANICAL TESTING, QUALIFICATION

PROGRAMME (CONT) ............................................................................................................................................... 14

TABLE 3 SANDWICH - MECHANICAL TESTING, SCREENING PROGRAMME............... ................ .... 15

TABLE 4 SANDWICH - MECHANICAL TESTING, QUALIFICATION PROGRAMME........................... 16

TABLE 5 TEST MATRIX FOR STRUCTURAL JOINT ................................................................................... 22

TABLE 5 (CONT) TEST MATRIX FOR STRUCTURAL JOINT...ERROR! MARCADOR NO DEFINIDO.23

TABLE 6 - FAILURE CRITERIA AND CORRESPONDING F12COEFFICIENT............... ................ ............ 28

LIST OF FIGURES

FIG.1 - THE PYRAMID OF TESTS FROM MIL-HDBK-17-1F.......................................................................... 9

FIG. 2 - MATERIAL/STRUCTURAL QUALIFICATION....................................................................................... 10

FIG. 3 - LIFETIME DIAGRAM FOR A LAMINATE AT R = 0.1 ......................................................................... 18

FIG.4 - EXAMPLE OF GOODMAN DIAGRAM FOR DIFFERENT R RATIO................................................ 19

FIG 5. MECHANICAL JOINTS FRICTION AND FITTED GRIP JOINTS ILLUSTRATION................. 20

FIG. 6 - SAMPLES FOR ADHESIVE SHEAR TESTS ACCORDING TO REF [R5], VALUES IN MM.... 22

FIG 7: - PRINCIPAL SKETCH OF HOMOGENIZATION PROCEDURE........................................................ 26

FIG.8 - FLOWCHART OF FAILURE ANALYSIS OF A LAMINATE................................................................ 28

FIG. 9 - INTERACTION BETWEEN TRANSVERSE TENSILE STRESS AND SHEAR STRESS........... 29

FIG 10 - STRESS DISTRIBUTION ON THE ACTION PLANE AS INTRODUCED BY PUCK.................. 29

FIG. 11 - EXAMPLE OF BEARING STRESS/STRAIN CURVE........................................................................ 30

FIG. 12 - SANDWICH ELEMENT DEFINITION ................................................................................................... 32

FIG. 13 - LAMINATE FAILURE................................................................................................................................. 33

FIG. 14 - TRANSVERSE SHEAR FAILURE .......................................................................................................... 34

FIG. 15 - LOCAL CORE CRUSHING ...................................................................................................................... 34

FIG. 16 - GLOBAL BUCKLING................................................................................................................................. 34

FIG. 17 - SANDWICH FAILURE MODES ............................................................................................................... 35

FIG. 18 - BONDING BETWEEN TWO PIECES IN HYPERMESH ................................................................... 37


6/60


FIG. 19 - STANDARD BOLTED JOINTS WITHOUT INSERT IN HYPERMESH.......................................... 37

FIG 21 - TYPICAL STRENGTH AND STIFFNESS DEGRADATION IN COMPOSITE (SCHEMATIC).. 39

FIG. 22 MINERS SUM............................................................................................................................................. 40

FIG. 23 MINERS SUM AND LOAD ORDER ..................................................................................................... 40

FIG. 24: - PERCENT FAILURE RULE TO TAKE INTO ACCOUNT THE INFLUENCE OF THE

LOADING ORDER, THE NUMBER OF CYCLES AND THE ULTIMATE LEVEL. .............. ................ ......... 41

FIG. 25: - ELASTIC STIFFNESS DEGRADATION AS FUNCTION OF LOAD CYCLES.......................... 41

DEFINITIONS

DSC differential scanning calorimetry

DMA dynamic mechanical analysis

CAI compression after impact

CTE coefficient of thermal expansion

RH relative humidity

RT room temperature

TMA thermal mechanical analysis


7/60


1. REFERENCE

[R1] : MIL-HDBK-17-1F from June 2002 - Composite Material Handbook

[R2] : DOT/FAA/AR-04/24 from June 2004 - A Comparison of CEN and ASTM TestMethods for Composite Materials

[R3] : A UNIQUE CRITERION FOR DESCRIBING FAILURE OF FOAM CORE SANDWICHMATERIALS - A DESIGN ENGINEERING PERSPECTIVE - J. Feldhusen, S.Krishnamoorthy]

[R4] : ISO 13003:2003 (E) - Fibre-reinforced plastics - Determination of fatigue propertiesunder cyclic loading conditions, 2003

[R5] DIN 6701-3 (2002): Adhesive bonding of railway vehicles and parts Part 3:Guideline for construction design and verification of bonds on railway vehicles.

[R6] DIN EN 1465 (1994): Adhesives, Determination of tensile lap-shear strength ofrigid-to-rigid bonded assemblies.

[R7] ASTM D 3528-96 (2002): Standard Test Method for Strength Properties ofDouble-Lap Shear Adhesive Joints by Tension Loading

[R8] Zenkert, D. (1997): The Handbook of Sandwich Construction, EngineeringMaterials Advisory Services Ltd, UK

2. EXECUTIVE SUMMARY

The characterization of composite materials in railways for structural calculations has beenstudied and the findings have been included in the deliverable 4.1.

This document proposes:

- A guideline for determining the properties of a polymer matrix composites system fora structural application, in order to be able to perform analysis with finite elementmodeling.

- Several test matrices are proposed, based on the Aeronautic/aerospace state of arts.

- This document, and specifically the test matrix hereafter proposed, is mainly basedthe Composite Material Handbook in reference [R1] written by the US Department ofDefense which were modified by Airbus Group internal documents. It covers twokinds of structures using monolithic or sandwich material.

- A guideline for determining the modeling procedure of a composite structure as wellas failure criterion, considering both monolithic and sandwich architecture.

- A specific chapter is dedicated to junctions. A first test matrix is proposed, as well asmodeling guideline for composite joints.

It concludes that for the mechanical analysis on a composite structure, some changes on themethodology is needed, primarily due to the heterogeneity and anisotropy of the newmaterial.

Chapter 3 of this document gives a general overview on the global methodology that is usedfor the characterization and testing of composite structures in aeronautic (experimentalbuilding-block approach). The important link between the manufacturing process and thematerial properties is also underlined, as well as the effect of the environment.

Chapter 4 underlines specifically the variability of the sources in composite materials, whichjustifies the consideration of a statistical based method to establish the design values ofmaterial properties.


8/60


Chapter 5 proposes a selection of relevant tests to characterize the material propertiesneeded in the most standard structure analysis (in plane stress and strain and standardelastic theory). The test matrices proposed have to be considered for new materials andreduced test matrix are proposed for the screening phases of the project.

Chapter 6 proposes a number of preferred failure criterions to consider in the analysis of bothmonolithic and sandwich structures. It gives a brief overview on numerical methods availablefor calculation of composite structure in static and fatigue and advice for the calculation ofjoints with a finite element code.

3. GLOBAL CHARACTERISATION LOGIC FOR COMPOSITE MATERIAL

In comparison with common metallic materials, composite materials are characterized by theirheterogeneity (fiber + matrix at the ply level) and their anisotropy (depending from thestacking sequence or orientation of each ply at the laminate level).

Moreover, it is well know that the final behavior of a composite material is highly dependenton :

- its manufacturing process (prepreg, infusion, hand lay-up,) and manufacturingparameters (curing cycle, ), and not only from the semi-products (fibber, fabric andmatrix) which is made of,

- the unitary thickness of the elementary plyOther important specificities of composite materials are their sensitivity to out-of-plane loads,the multiplicity of failure modes and, finally, the lack of universal failure criteria.

As a consequence, the global justification/qualification logic of a composite structure isgenerally based on an experimental building-block approach. This building-block approachcan be summarized in the following steps:

1. Generate material basis values and preliminary design allowables.

2. Based on the design/analysis of the structure, select critical areas for subsequent test

verification.3. Determine the most strength-critical failure mode for each design feature.

4. Select the test environment that will produce the strength-critical failure mode. Specialattention should be given to matrix-sensitive failure modes (such as compression, out-of-plane shear, and bondlines) and potential "hot-spots" caused by out-of-plane loads orstiffness tailored designs.

5. Design and test a series of test specimens, each one of which simulates a singleselected failure mode and loading condition, compare to analytical predictions (andadjust analysis models or design allowables as necessary).

6. Design and conduct increasingly more complicated tests that evaluate more complicatedloading situations with the possibility of failure from several potential failure modes. Compare

tests results to analytical predictions and adjust analysis models as necessary.

7. Design (including compensation factors) and conduct, as required, full-scale componentstatic and fatigue testing for final validation of internal loads and structural integrity. Compareto analysis.

The building-block approach is shown schematically here below, usually called the pyramidof tests:


9/60


Fig.1 - The pyramid of tests from MIL-HDBK-17-1F

Inside this pyramid of test, five structural complexity levels have to be considered:constituent, lamina, laminate, structural element and structural subcomponents.

The material form(s) to be tested, and the relative emphasis placed on each level, should bedetermined early in the material data development planning process, and would likely dependupon many factors, including: manufacturing process, structural application.While a single level may suffice in rare instances, most applications will require at least twolevels, and it is common to use all five in a complete implementation of the building-blockapproach.Regardless of the structural complexity level selected, physical and chemical propertiescharacterization of the prepreg (or the matrix, if it is added as part of the process, as withresin transfer molding) is necessary to support physical and mechanical properties testresults.

The five structural complexity levels cover the following areas:

Constituent Testing:This evaluates the individual properties of fibers, fiber forms, matrix materials, and fiber-matrix preforms.

Key properties, for example, include fiber and matrix density, and fiber tensile strengthand tensile modulus.

Lamina Testing (elementary ply):

This evaluates the properties of the fiber and matrix together in the composite materialform. For the purpose of this discussion, prepreg properties are included in this level,although they are sometimes broken-out into a separate level. Key properties includefiber area weight, matrix content, void content, cured ply thickness, lamina tensilestrengths and moduli, lamina compressive strengths and moduli, and lamina shearstrengths and moduli.

Laminate Testing:Laminate testing characterizes the response of the composite material in a givenlaminate design.

Key properties include tensile strengths and moduli, compressive strengths and moduli,shear strengths and moduli, interlaminar fracture toughness, and fatigue resistance.

Structural Element Testing:This evaluates the ability of the material to tolerate common laminate discontinuities. Keyproperties include open and filled hole tensile strengths, open and filled hole compressive


10/60


strengths, compression after impact strength, and joint bearing and bearing bypassstrengths.

Structural Subcomponent (or higher) Testing:This testing evaluates the behavior and failure mode of increasingly more complexstructural assemblies.

At th is po int o f the document, we p ropose to consider that the smaller scale for fai luremodes and analysis/calculations will be performed at the lamina (or ply) level.We willnot consider datas for micromechanics calculations for instance, according to actual practisesconsidered in the aeronautic and space design offices.As a consequence, the test matrix and material data w il l focus on lamina testing (atthe elementary ply level), laminate testing and Structural Element Testing at thematerial level.

NOTE :- Constituent testing gives some material data useful for the process and/or for the

material specification acceptance values, but these properties are not directly usedfor the structure analysis.

- On the other side, the structural subcomponents (or components) testing are highlydependent from the application, and will not be considered in the framework of thisdocument.

The boundaries of the material characterization covered by this document is explicated by thegraphic below:

Fig. 2 - Material/structural qualification

4. STATISTICAL-BASED MATERIAL PROPERTIES

Variability in composite material property data may result from a number of sources includingrun-to-run variability in fabrication, batch-to-batch variability of raw materials, testingvariability, and variability intrinsic to the material. It is important to acknowledge this variabilitywhen designing with composites and to incorporate it in design values of material properties.

Procedures for calculating statistically-based material properties are not included in thisdocument. Nevertheless, details on these procedures could be found in the document in ref[R1] , Volume 1, Chapter 8

This statistical-based method justifies the realization of a minimum number of samples permaterial batch Ns and to test a minimum number of material batches Nb.

Constituent tests Lamina tests Laminate tests

Element

Subcomponent

Component

tests

Full scale test s)

Physical properties

DSC, DMA,

Basic properties

Strength, Stiffness, Environment

Laminate performance

Strength, Stiffness

Static/fatigue

Environment

Damage tolerance

Joint

Critical design verification

Boundary conditions

Secondary effects

Size effect

Static/fatigue

Damage tolerance

Framework of t his

document


11/60


Batch (or Lot) definition: For fibers and resin: a quantity of material formed during the same process and having

identical characteristics throughout. For prepregs/laminate : material made from one batch of fibber and one batch of resin

The following test matrices (see 5) give an indication of the number of samples to perform.The notation takes into account:- In the first number : the number of different material batch to test (= Nb),- In the second number : the number of samples to test for each batch (= Ns).As an example, 3x5 mean that it is proposed to perform tests from three different materialbatches, and with at a minimum five samples for each batch, that mean 15 samples are to betested.

5. PROPOSED TESTS MATRICES

The test matrices hereafter proposed have to be considered for new materials and for themost generic application.The tests to be performed should be simplified taking into account the existing

knowledge/available user data, and the specific requirements of the application.

5.1 General considerations

In the constitution of the test matrix, it is implicitly considered that the composite material ismainly loaded by in-plane loads (shell hypothesis). Indeed, this is considered as a bestpractice in the design of composite structures. As a consequence, most of the tests arefocused on in-plane mechanical characteristics of the material. In case the structure would beloaded by a high level of out of plane loads (3D state of stress), specific supplementary testsshould be considered to address this point.

As it has been said above in this document, the performance properties of compositelaminates are directly affected by the specific process used for their manufacturing process. It

is critical that the test specimens manufactured through the various levels of the buildingblock approach use the same process, representative of the one that will be used in themanufacturing of the Railway parts.

It is still important to evaluate the resistance of new polymer materials to fluids with whichthey might come in contact. In case the material is expected to be used in an applicationwhere fluid exposure occurs for significant time periods at a different temperature, it isrecommended that the test laminates be exposed to the above fluids at room temperatureconditions, and tested over the expected range of service temperatures.

Annex2 gives the list of the existing standards related to composite characterization.Concerning the standards proposed in the following matrices, AITM standards have not beenselected because they are under the copyright of AIRBUS INDUSTRIE. CEN/ISO standardshave been selected as much as possible. In case CEN/ISO standards are not available,

ASTM standards have been proposed according to document in reference [R1].For more detail, the document in reference [R2] performs a complete comparison betweenCEN and ASTM test methods for composite materials. Moreover, Annexe1 gives, forinformation purpose, some generic considerations on the test procedure.


12/60


5.2 STATIC PROPERTIES AND TESTS MONOLITHIC COMPOSITE (UDLAMINA/LAMINATE)

5.2.1 Screening test matrix

The objective of the screening process is to reveal key mechanical property attributes and/or

inadequacies in new material system candidates, while keeping testing to the minimumnumber of samples. The screening process identifies, for a particular composite materialsystem, the critical test and environmental conditions as well as any other specialconsiderations. Proper test matrix design enables comparison with current productionmaterial systems.The recommended minimum set of cured laminate mechanical properties for generalapplication is defined here after:

glass transitiontemperature

DMA EN 6032 1x5

fibber volume and resincontent

EN 2564 /

density ISO 1183-1 /

cured ply thickness / /

min T RT max T max T

Tensile strength, tensilemodulus and poisson ratio

0 tension EN 2561 B (or A) 1x5 1x5

Flexure strength andmodulus (**)

4 points flexure ISO 14125 1x5 1x5

In plane shear strengthand modulus tension EN 6031 1x5 1x5

Interlaminar shear strength(ILSS)

short beam shear EN 2563 1x5 1x5 1x5 1x5

bearing strength

Representative of thejunction (double lapshear, single lapshear, screw, rivet)

ASTM D5961procedure A and B

1x5

plain 1x5notch 1x5 1x5plain 1x5 1x5notch 1x5 1x5

Compression after impact(CAI)

EN 6038 1x5

(*) : [% at 0 / % at +45 / % at -45 / % at 90] or stacking sequence representative from the design

[0]n 1x5

(**) : not used for mechanical datas. Only to evaluate compressive strength sensitivity to environment (temperature/% moist)

1x5

[0]n

open hole tensile strength

0 compression

properties layuptest type and

directiontest method (1)

1x5

[0]n

N batch to be tested xN specimens for each batch

test condition

drysee EN2743

wetsee

EN2823

[0]n or (*) 1x5

[0]n or (*)

[0]n

50/20/20/10 (*)

(1) These are recommandations but not to be considered as exclusive test method

ASTM D6484

0 tension ASTM D5766

open hole compressionstrength

(*)

50/20/20/10 (*)

[45]ns

[0]n

25/25/25/25 (*)

Table 1 Laminate and structural element - Mechanical testing, screening program


13/60


5.2.2 Qualification test matrix

The recommended minimum set of cured laminate mechanical properties for generalapplication is defined here after:

glass transition

temperatureDMA EN 6032 5x3

fibber volume and resin

contentEN 2564 /

density ISO 1183-1 /

cured ply thickness / /

/

/

moisture diffusivity ASTM D5229 /

thermal diffusivity ISO 1159-2 /

specific heat DSC ISO 11357-4 /


Tensile strength and

modulus0 tension EN 2561 B (or A) 1x5 5x5 1x5 1x5

Poisson Ratio 0 tension EN 2561 B (or A) 1x5 1x5

Tensile strength and

modulus90 tension EN 2597 B 1x5 1x5 1x5

Compression strength

and modulus0 compression EN 2850 A (or B) 1x5 5x5 1x5 5x5

Compression strength

and modulus90 compression EN 2850 B 1x5 1x5

In plane shear strength

and modulusin plane tension EN 6031 1x5 5x5 1x5 5x5

Interlaminar shear

strength (ILSS)short beam shear EN 2563 1x5 5x5 1x5 5x5

1x5 5x5

1x5 5x5 5x5

1x5 5x5 5x5


1x5[0]n or (*)

1x5

1x5

dry

see EN2743

w et

see

EN2823

5x3

5x5

1x5

1x5

in plane coeff icient of

thermal expansion[90]n

ISO 11359-2TMA

[0]n or (*)

[0]n or (*)

[0]n

[0]n

5x3


25/25/25/25 (*)

10/40/40/10 (*)

50/20/20/10 (*)

bearing strength

N batch to be tested x

N specimens f or each batch

properties test method (1)

5x3

test type and

directiontest condition

[0]n

layup

ASTM D5961procedure A and

B

[90]n

[45]ns

[0]n

[0]n

[0]n or (*)

[0]n or (*)

[0]n

[0]n

[90]n

Representative of

the junction (doublelap shear, single

lap shear, screw ,

rivet)

Table 2 Laminate and structural element - Mechanical testing, qualification program


14/60



plain 1x5 1x5

notch 1x5 5x5

plain 1x5

notch 1x5 5x5 5x5

plain 1x5 1x5 1x5

notch 1x5 5x5 5x5

plain 1x5 1x5

notch 1x5 5x5

plain 1x5

notch 1x5 5x5 5x5

plain 5x5 1x5

notch 1x5 5x5 5x5

25/25/25/25 (*) notch 1x5 5x5

10/40/40/10 (*) notch

50/20/20/10 (*) notch 5x5 5x5

25/25/25/25 (*) notch 1x5 5x5

10/40/40/10 (*) notch 5x5

50/20/20/10 (*) notch 5x5 5x5Compression after impact

(CAI)EN 6038 5x11 (**) 5x11 (**)

G1c ASTM D5528 3x5 1x5

G2c PREN 6034 3x5 1x5

Resistance to agressive

fluid (***)

short beam shear

(ILSS)EN 2563


(**) : first 1 batch in both condition, other 4 batches at worst case condition

ASTM D6484

0 tension ASTM D5766

ASTM D6742

0 tension

test type and

direction

10/40/40/10 (*) 0 compression

layup


N specimens for each batch

test condition

open hole tensile strength

propertiesdry

see EN2743

wet

see

EN2823

open hole compressionstrength

25/25/25/25 (*)

10/40/40/10 (*)

50/20/20/10 (*)

25/25/25/25 (*)

test method (1)


filled hole tensile strength

ASTM D5766

modified according

to MIL-HDBK-17

section 7.4.2.2

0 compression

(***) : optional tests. These include testing of cured laminates after exposure of the laminates to solvents that the part will be subjected toin actual service.

(*)

50/20/20/10 (*)

filled hole compression

strength

Table 2 Laminate and structural element - Mechanical testing, qualification program(continuation)


15/60


5.3 STATIC PROPERTIES AND TESTS - SPECIFICITIES FOR A SANDWICHPANEL (UD LAMINA/LAMINATE) WITH CORE MATERIAL

5.3.1 Screening test matrix

The recommended minimum set of cured laminate mechanical properties for generalapplication is defined here after:


tensile strength (out of

plane) and core/skin

junction under tensile load

tensile ASTM C297 1x5

compresive strength and

modulus (out of plane) compression ISO 844 1x5

shear strength and

modulusshear ISO 1922 1x5 1x5

core/skin junction under

shear load (and shear

strength and modulus)

sandwich flexure

or

shear

ASTM C393

or

ISO 1922 with skin

1x5 1x5

plain 1x5 1x5

notch 1x5 1x5


(**) : instead of the open hole compression characterisation test of the composite skin alone (see matrix 4.2.1)

ASTM D 7249

core

core

[(*) / core / (*)]

properties layup (2)test type and

direction

[(*) / core / (*)]

test method (1)


N specimens for each batch

test condition

dry

see EN2743

wet

see

EN2823

skin strength and open

hole compression strength

(**)

sandwich long beam

flexure

(1) These are recommandations but not to be considered as exclusive test method(2) : If the material is designed to be self-adhesive to the core, then these tests should be conducted on cocured panels fabricated

without adhesive. If the material requires an adhesive layer for bonding to the core, then the tests can be conducted on either (or both)

cocured panels or precured skins secondarily bonded to the core, depending on the anticipated design and fabrication methods to be

used with the material.

[50/20/20/10] /

core /

[50/20/20/10] (*)

Table 3 Sandwich - Mechanical testing, screening program


16/60


5.3.2 Qualification test matrix

The recommended minimum set of core and cured sandwich mechanical properties forgeneral application is defined here after:


tensile strength (out ofplane) and core/skin

junction under tensileload

tensile ASTM C297 1x5 1x5

compresive strength andmodulus (out of plane)

compression ISO 844 1x5 1x5 1x5

shear strength andmodulus

shear ISO 1922 1x5 3x5 3x5

core/skin junction undershear load (and shearstrength and modulus)

sandwich flexureor

shear

ASTM C393or

ISO 1922 w ithskin

1x5 1x5

plain 1x5 1x5

notch 1x5 5x5

plain 1x5

notch 1x5 5x5 5x5

plain 5x5 1x5

notch 1x5 5x5 5x5

[25/25/25/25] /core /

[25/25/25/25] (*)notch 1x5 5x5

[10/40/40/10] /core /

[10/40/40/10] (*)

notch 5x5

[50/20/20/10] /core /

[50/20/20/10] (*)notch 5x5 5x5

(*) : [% at 0 / % at +45 / % at -45 / % at 90] or stacking sequence representative from the design(**) : instead of the open and filled hole compression characterisation test of the composite skin alone (see matrix 4.2.2)

(2) : If the material is designed to be self-adhesive to the core, then these tests should be conducted on cocured panelsfabricated w ithout adhesive. If the material requires an adhesive layer for bonding to the core, then the tests can be conductedon either (or both) cocured panels or precured skins secondarily bonded to the core, depending on the anticipated design andfabrication methods to be used w ith the material.


core

properties drysee EN2743

N batch to be tested xN specimens for each batch

wetsee

EN2823

test method (1)test type and

directionlayup (2)

test condition

[(*) / core / (*)]

[(*) / core / (*)]

core

filled hole compressionstrength (**)

sandwich longbeam f lexure w ith

open holeASTM D 7249

skin strength and openhole compressionstrength (**)

[25/25/25/25] /core /

[25/25/25/25] (*)

sandwich longbeam flexure

ASTM D 7249[10/40/40/10] /

core /[10/40/40/10] (*)

[50/20/20/10] /core /

[50/20/20/10] (*)

Table 4 Sandwich - Mechanical testing, qualification program


17/60


5.4 FATIGUE PROPERTIES AND TESTS

In the area of fatigue, however, no generalized methodology has yet been devised to predictlaminate behavior from unidirectional specimen data.

Hence, the development of fatigue design values becomes a unique problem for eachapplication lay-up.

Many studies have been undertaken, and much has been written concerning life prediction forspecific laminates under cyclic loading spectra. Even at this level, empirical methods havebeen favored due to the inadequacy of results obtained from cumulative damage models,fracture mechanics analyses, and other theoretical approaches.

Fatigue data is generated at the design critical test conditions (room temperature, or hot/wet).

The characteristics for fatigue resistance of materials shall be determined with experimentalmethods for cyclic loadings associated with static loads under representative serviceconditions.

The part of the fatigue curve which is characterized shall cover the domain of use (in term of

number of cycles, stress or strain amplitude and R ratio) see below:

With : R =max

min

; ampl= (max min) / 2

For instance, R= 0,1 correspond to a pure traction load, R= -1 correspond to atraction/compression load with moy = 0, R = 10 correspond to a pure compression load.

In general, composite structures are assumed to be less sensitive to dynamic loading thanmetallic structures, but general well accepted hypothesis of damage accumulation are stillmissing.

The damage itself is therefore strongly influenced by the number of cycles, the ultimateloading and the order of the loading, i.e. if high load levels are applied onto the laminatefollowed by lower leads to earlier failure than vice versa. This point has to be taken intoaccount when defining the test sequence of a composite structure/substructure under arepresentative cycling loading.

time

S

min

moy

max

amp


18/60


5.4.1 Fatigue characterization and test recommendations:

In ref [R4] (for instance), the principal procedure of fatigue investigation for composites isdescribed.

In the absence of todays existing guidelines for fatigue test specimens, the geometries ofthe specimens therefore is based on the standards for quasi static investigations, (e.g.according to ref [R4]):

for sample in tensile : DIN EN ISO 527-4 or DIN EN ISO 527-5 (for UD)

for samples in flexure samples : ISO 14125, with 4 points flexure preferred in thatcase

NOTE : The general lack of todays existing and applied test methods must be seenwithin the concentration on single mechanisms under investigation only, e.g.constitutive behavior at certain stress ratios and/or single damage phenomenaetc. The question of general transferability of results, is not solved yet.Furthermore, the principal phenomenology of fatigue damage on compositematerials is not understood. This is due to the very complex structure of

composite materials consisting of fiber roving, matrix material and adhesivelayers between fibers and matrix. In the near future, strong experimental efforthas to be spent for the investigation to answer these fundamental.

The determination of fatigue properties is commonly based on lifetime diagram (S-Ncurves), energy release rates, residual strength and/or residual stiffness properties, (DIN50100 - Dauerschwingversuch, 1978). With this, the fatigue strength and the fatigue limitare estimated by measuring the number of cycles for given stress ratios.

NOTE : Generally, the low cycle fatigue regime (


19/60


Fig.4 - Example of Goodman diagram for different R ratio

It would be very costly, in an experimental point of view, to perform test series for a largenumber of R ratio. As a consequence, it is usual to perform tests for a limited number.

Nevertheless, the number of R ratio to test for a material/structure depends on thematerial database available, on the cyclic load of the application and is finally linked tothe margin policy considered. As a consequence, it is difficult to define or propose to

consider a unique combination of R ratio to test for a generic application. For the mostgeneric application, it is proposed to test at a minimum R=0,1 and R=-1.

The frequency of the cycling test should not introduce temperature elevation which coulddamage the sample. As a consequence it is recommended to not exceed 5Hz.Nevertheless, it is possible to increase this frequency up to 20/30Hz with an adequatecooling device.

The failure can be considered at the complete sample separation or at a given level ofstiffness reduction. If the latest failure criterion is considered, a level of stiffnessreduction of 20% is a classical value.

5.4.2 Specific ities for a sandwich panel with core material

The reduction of stiffness or stability properties of sandwich structures due to cyclicfatigue should be considered. This reduction may be caused by:- a reduction in modulus of elasticity in the facings materials and/or in the core(s)

materials due to various types of damage, e.g. micro cracks- a local debonding between faces and core at the interface.

As a consequence, specific tests should be performed at the sandwich level, with 4 pointsFlexion test according to ASTM D7249 for instance.


20/60


5.5 JOINTS

5.5.1 Definit ion of Join ts

In order to limit the number of potential combinations all composite materials to be

considered as quasi-isotropic laminates. If it is not the case, tests should be performedfor each specific stacking sequence.

Bonded Joints (only pure adhesives considered, without any additives like short-fibers)

Both joints for composite vs. composite pieces and composite vs. metal pieces

Adhesives with high Youngs Modulus, low strains (i.e. known as thin-fi lm), bestproperties 0.05mm < t < 0.2mm i.e. for epoxy

Adhesives with low Youngs Modulus, high strains (i.e. known as thick-film)

Mechanical Joints

Friction Grip Joints (see illustration fig. 5 here below)For composite vs. composite pieces, composite vs. metal pieces, composite vs.

metal pieces covered with composite materialo Rivetso Huck-Bolt Type

Standard bolted joints with insert/reinforcement in laminate (with nuts)o Screws

Standard bolted joints without insert/reinforcement in laminate (with nuts) Standard bolted joints with insert/reinforcement in laminate (with nuts) Tapped thread joints

Fitted Grip Joints (see illustration fig. 5 here below)For composite vs. composite pieces, composite vs. metal pieces, composite vs.metal pieces covered with composite materialo Rivets (i.e. blind rivets)o Huck-Bolt Type

Standard bolted joints with/without insert/reinforcement in laminate (withnuts)

o Screws Standard bolted joints with/without insert/reinforcement in laminate (with

nuts)

Fig 5. Mechanical joints Friction and Fitted grip joints illustration


21/60


5.5.2 Test matr ix

The following test matrix is based on experiences and, if possible, based onstandards from railway engineering. It does not guarantee to be complete sincetesting procedures also depend on load cases, on the considered parts (structuralrelevant or not) and on the test laboratory. Before testing it has to be specificallydiscussed about testing budget/costs, load cases and agreed with the testlaboratory which tests are necessary and possible. Therefore, the followingstatements are proposals and tests can be omitted/added in specific cases.

Regarding failure criteria, if there are any restrictions or criteria mentioned in thestandard for a joint type then these are to use instead of/in addition to given failurecriteria in the table.

In general, for static load tests, 5-8 specimen are needed. For fatigue load tests,for every R-ratio, 15 specimen are necessary on different stress amplitudes inorder to define a Whler-curve.

For bonded joints, some specific points must be taken into account:

- Temperatures (i.e. 3 different levels) as well as other influencingmediums such as moisture have a significate effect on joint behavior. Asa consequence, the effect of environment must be considered in thecharacterization matrix

- Due to peak stress at the edge of the bonded joint, the shear strength isusually not proportional to the bonded length. As a consequence, thesample joint design should be as close as possible from the final design.In the case it is not possible, the local stress distribution along thebonded joint should be evaluated throw a calculation.

For mechanical joint with composite parts, it must be considered that composite

material is more sensitive to creeping than metal under compressive load. As aconsequence, the loss of pre-load in the screw/rivet have to be considered andcharacterized by tests. In the case of highly loaded screw/rivet, the admissiblesurface pressure has to be checked. A metallic insert can also be embedded inthe clamping area to improve those points.


22/60


Joint TypeSample

Dimensions

Standardsfor test

parameters

Load Type Static (with s chematic pictures) (*6)

Load Type

Fatigue, R=

Criteria for joint

verification (*1)

Useful

testing?Tensile/Compressi

onShear Combined

Other (i.e.creeping)

0.1 -1 (*3)

Bonded Joints (*2)

Adhesives withhigh YoungsModulus (*4)

DIN 6701, (high strength glue) DIN 6701-3 X X XX (Creep

tests)X X

Shear Strength,See DIN6701-3

X

Adhesives with

low YoungsModulus

DIN 6701 (low strength glue)

DVS 1618 (app. 3)see figure 5

DIN 6701-3 X X XX (Creep

tests)X X

Bimodal regulation(force regulation

amplitude), seeDIN6701-3

X

*1 In general for compo site laminates the followin g theories are commonly used for evaluation: a) Tsai Wu theory, b) Tsai Hill theory, c) Hoffmans theory, d) Maximum strain t heory. Moreover,critical str ains are useful for evaluation

*2 Also ASTM D 3528-96 (2002): Standard Test Method for Strength Properties of Double-Lap Shear Adhesive Joints by Tension Loading can be considered for co mposites and metal/compositecomponents

*3 In general, testing negative 'R'-ratios depends on th e parts which are considered and especially the loads which are applied, i.e. aerodynamical loads. Therefore, in some cases it is necessaryto consi der R=0, or 0.7 and in some rare cases R=-0.5 to -0.7 as well. For compression fatigue R-values (i.e. R=-1.0), the specimen needs to be stiff enough in order t o establish compressivestresses

*4 For structur al adhesives (thin-film) only a proposal is giv en because it is not often performed and therefore cannot generally be confirmed by BT.

*6 The small schematic pict ures are used to show how the load is applied in general but are not explicitly describing the sample geometries. These sample geometries vary depending on the joint.

Table 5 Test matrix for structural joint

Fig. 6 - Samples for adhesive shear tests according to ref [R5], values in mm


23/60


Joint TypeSample

Dimensions

Standardsfor test

parameters

Load Type Static (with schematicpictures) (*6)

Load TypeFatigue, R=

Criteria for joint v erification (*1)Usefultesting?Tensile/Co

mpressionShear Combin ed 0.5 -1 (*3)

Mechanical Joints (*5)

Friction Grip Joints

Rivets X X X X X For the joint: Loss of preload due to setting orcreeping, Measurement of preload either separately

or during static load testsFor the laminate: i.e. connection between cylindrical

insert and laminate, plastic deformation, fracture

X

Huck-Bolt Type X X X X XX or DVS/EFB

3435

Screws X X X X X (X) VDI2230

Fitted Grip Joints

Rivets X X X X X For the joint: Loss of preload due to setting orcreeping, Measurement of preload either separately

or during static load testsFor the laminate: i.e. connection between cylindrical

insert and laminate, plastic deformation, fracture,

bearing stress

X

Huck-Bolt Type X X X X XX or DVS/EFB

3435

Screws X X X X X (X) VDI2230

*1 In general for compo site laminates the followin g theories are commonly used for evaluation: a) Tsai Wu theory, b) Tsai Hill theory, c) Hoffmans th eory, d) Maximum strain t heory. Moreover,

critical str ains are useful for evaluation

*2 Setting should be investigated no matter if two pure composite parts are joint or if there are inserts (metal) with different surface properties

*3 In general, testing negative 'R'-ratios depends on th e parts which are considered and especially the loads which are applied, i.e. aerodynamical loads. Therefore, in some cases it is necessary

to cons ider R=0, 0.1 or 0.7 and in some rare cases R=-0.5 to -0.7 as well. For com pressi on fatigu e R-values (i.e. R=-1.0), the speci men needs to be stiff enoug h in order to establ ish co mpres sivestresses

*4 For structur al adhesives (thin-film) only a proposal is giv en because it is yet not performed and therefore cannot be confirmed by BT.

*5 Screws then can be evaluated according to VDI2230 (no tests necessary), Huck-Bolts according to DVS/EFB 3435 or tests. Strength of riv ets has to be proven by tests.

*6 The small schematic pict ures are used to show how the load is applied in general but are not explicitly describing the sample geometries. These sample geometries vary depending on the joint.

Table 5 (continuation) Test matrix for structural joint


24/60


6. MODELLING PROCEDURE

6.1 STATIC MODELLING (*1)

When modeling composites, two ways can be followed: solid elements or shell elements.

Since the thickness of composites is mostly very small compared to its remaining dimensionsmodeling laminates with a certain amount of plies by using solid elements is very expensiveconsidering model size and solving time.

Therefore in that case, using shell elements for the laminates is probably the best and mostefficient choice. Indeed, 3D models are reduced to their middle surface and meshed with shellelements. For laminates, it is defined that the laminate layers are bonded together in order toform a cohesive structure.

In the FE program, a component name for the laminate has to be chosen. After that theproperty for laminates is defined and a material is allocated. These properties generate shellelements where the following characteristics are defined and may be different for every ply:

- ply number,- amount of plies

- material ID of each ply

- thickness of each ply

- orientation of each ply (i.e. 0, 90 or +-45 degrees)

With this definition, laminates as well as sandwich structures can be mapped.

*1 Modeling composites is not commonly carried out by BT. Therefore, only a proposal isgiven here.

6.1.1 Monoli thic composite

6.1.1.1 Mixture Rules

The consideration of single fibers within the composite cannot be taken into account directly,due to extraordinary high computational effort. Instead, the single plies of the laminatedstructure have to be modeled as homogeneous continuum.

Homogenization methods or mixture rules are used to define the global behavior on asufficient level by neglecting local aspects. The mathematical analysis of a laminatedstructure demands the knowledge about the constitutive relations of the single plies, thedefinition of a reference placement, the transformation relations for the single plies and theassumptions about through thickness displacements.

In the aircraft design process the Classical Laminate Theory (CLT) is widely used. The CLTassumes homogeneous and orthotropic single plies, plane stress state conditions, the idealbonding between neighboring plies, linearly shaped membrane displacements and thenegligence of transverse shear strains.

The strains itself can directly be deduced from the membrane displacements viadifferentiation. The corresponding stresses are obtained by multiplying the layer wiseconstants ply stiffness with the strains. The complete constitutive relation of the laminate isdefined by integration of the layer wise material laws over the thickness of the laminate, whichresults in the well-known matrices of membrane stiffness A, coupling stiffness matrix Bandbending stiffness matrix D


25/60


xy

y

x

xy

y

x

xy

y

x

xy

y

x

D

DDSym

DDD

BBBA

BBBAA

BBBAAA

M

M

M

N

N

N

33

2322

131211

33321333

2322122322

131211131211

.

. (1)

The in-plane characteristics of the lamina EL, ET, GLTand nuLTneed to be define to performanalysis, where E and G are Youngs modulus and shear modulus resp. and the subscripts L,T denote the longitudinal and transverse fiber direction.

When the elementary properties of the fiber and the matrix are the only datas available, it ispossible to evaluate the properties of the lamina (elementary unidirectional ply) by using thesimplest assumptions for stiffness homogenization:

mfL EEE )1( (2)

mfLT )1( (3)

mfT EEE /)1(//1 (4)

Where the subscripts f and m denote the fiber and matrix correspondence respectively, andthe fiber volume fraction is denotes by .

Nevertheless, the properties in transverse fiber direction commonly need a correction. In caseof composites containing isotropic fibers, the following formulae for transverse stiffnessmoduli were established:

25,1

2

)1(/

)85,01(

fm

mT

EE

EE (5)

mf

TEE

E/)1(5,0/

)1(5,0

(6)

fm

mf

TEE

EEE

)1( (7)

)/1(1/ fmmT EEEE (8)General recommendations when using equations (5) to (8) can not be given. It must be noted,that the range of results can reach more than 10%. As a consequence, theses formulasshould be considered only unless material characterization of the lamina is available (see5.2).

NOTE : concerning GLT, it is assumed to be not so different from a material to another (forepoxy systems). As a consequence, for a early phase of project it is proposed to usepreviously characterized value as a first order of magnitude, unless characterization testresults will be available.

IMPORTANT :All the methodologies mentioned above are used to calculate the constitutiveproperties of composite materials in 2D conditions, only. The constitutive properties in

thickness direction are not covered by the CLT. For this, advanced mathematical tools, e.g.analytical and/or numerical homogenization techniques have to be applied.


26/60


6.1.1.2 Analytical and Numerical Homogenization Methods for Monolith icMaterials

The use of fibers reinforced composite materials in nowadays aerospace applications, i.e.glass and / or carbon fibers are embedded into a matrix material, demands a preciseknowledge about the constitutive behavior of the constituents. For this, different analytical andnumerical methodologies were developed during the past decades.

A considerable improvement for the estimation of elastic constants was reached by the worksof (Hashin & Shtrikman, 1962) and (Hashin & Shtrikman, 1963) by the definition of variationalprinciples for isotropic and anisotropic and heterogeneous multiphase materials. The stresspolarization to account for the difference between the true stress and the stress resulting fromthe true strain acting on a homogeneous trial material was introduced.

It must be noted, that for all the approaches mentioned above, the small strain assumptionholds.

In case of very complex microstructures, the use of a representative volume element (RVE) inconjunction with numerical homogenization scheme is recommended, (Bhlke, 2001),

(Kouznetsova, 2002), (Lubarda, 2002) or (Nemta-Nasser, 1999).Therefore the RVE must capture the main features of the microstructure and has to representa material point on the macrostructure at the same time. This is achieved if the dimensions onthe microscale are orders of magnitudes smaller compared to the structural dimensions,which is known as scale separation.

The aim of analytical as well as numerical methods to estimate effective material properties isto obtain macroscopically homogeneous properties from the microscopically veryinhomogeneous constituents, which can be used in structural analysis (see fig. 7 from Gross& Seelig, 2007).

It must be noticed that RVE method is considered as an advanced method not frequentlyused in industry for structure analysis.

Fig 7: - Principal sketch of homogenization procedure.


27/60


6.1.1.3 Failure Criteria for Mono lith ic Composite

Different failure modes can be observed at the failure of a composite material (see some

examples here below):

- matrix cracking , matrix/fiber debonding

- delamination,

- fiber failure,

- micro-buckling of f ibers (under compressive stress)

Failure criteria are used to evaluate the local or global structural behavior, taking into accountthe different failure modes as well as the multi axial state of stress of the structure.

6.1.1.4 Failure criteria - First ply failure/ last ply failure

In a composite structure nevertheless, the failure of one ply of a laminate doesnt correspondsystematically to the failure of the laminate. As a consequence, failure criteria of laminatescan be categorized into first ply failure criteria (FPF) and last ply failure criteria (LPF).

FPF are often used due to their very conservative predictions. This is due to the disregard offurther load carrying capabilities of the remaining plies.

Practically, this kind of failure prediction approach is favored in industry due to its simplicityand robustness.

In contrast to this, LPF predict the failure of the laminate if the strength limit of the lastremaining ply is reached. If the failure criterion of a single ply is reached, a recalculation withreduced stiffness values is carried out until the last ply of the laminate fails as well. A principalsketch of the failure analysis of a laminate is depicted in figure 8. The actual state of stress isassessed with respect to the failure criterion used. If fracture occurs, the user will decide if thedegradation is tolerable or if the stacking sequence has to be changed. The rearranged state


28/60


of stress is reassessed and evaluated again until the stress analysis is finished (Knops &Bge, 2006).

Fig.8 - Flowchart of failure analysis of a laminate

6.1.1.5 Failure cr iteria - Differential and non-differential failure criteria

Interpolation criteria take into account the general multi axial loading conditions of acomposite. In (Gol`denblat & Kopnov, 1965), a tensorial polynomial description wasintroduced, where in reality the restriction to fourth order strength tensors is commonly made,

1 jiijii

FF (14)

wherei

F and ijF denote strength tensors of second and fourth order, respectively.

It must be noted that a distinction between fiber cracking and matrix failure cannot be made inthose criterion : the onset of failure itself is predictable, only.

By different choices of the coupling coefficient F12numerous well established criteria can be

deduced from the general interpolation criterion (see table 1, where TTLL RandRRR ,,

denote the longitudinal and transverse tensile and compression strength values, respectively).

Criterion F12

Tsai-Wu 0

Hoffmann )/(5,0 LLRR

Norris )/(5,0TL

RR

Tsai-Hahn TTLL RRRR/5,0

Table 6 - Failure criteria and corresponding F12coefficient

Beside the interpolation criteria mentioned above, some other criterion where introduced inorder to perform a distinction between fiber fracture and inter fiber fracture. In that case theinter fiber fracture criterion was introduced, motivated by experimental observations where aninteraction of the transverse tensile stress and the shear stress was noticed (see figure 9).The fracture limit obviously is reached before the strength values.


29/60


Fig. 9 - Interaction between transverse tensile stress and shear stress

Here below will be detailed two of them: Hashin and Puck

In the Hashin criteria (Hashin, Failure Criteria for Unidirectional Fiber Composites, 1980),failure criteria to distinguish fiber tension (15), fiber compression (16), matrix tension (17) andmatrix compression (18) were developed

2

12

2

11

LT

t

fSX

F

(15)

2

11

C

c

fX

F

(16)

2

12

2

22

LT

t

mSY

F

(17)

2

1222

22

22 1

22

LCT

C

T

c

m

SYS

Y

S

F

(18)

Where the superscripts T and C and L denote tension, compression and longitudinal and X, Yand S are the longitudinal, transverse and shear strength values, respectively.

Puck, in his approach, introduced the concept of fracture process zone (see figure 10), whoestablished the most prominent criterion to take into account the very complex modes of interfiber fracture.

Fig 10 - Stress distribution on the action plane as introduced by Puck

Thus, three different inter fiber fracture modes can be distinguished:

1. Mode A: the fracture is caused by a tensile stress or by a longitudinal shear stresswhich leads to a degradation of the Youngs modulus and the shear modulus,respectively.

2. Mode B: the fracture is caused by a longitudinal shear stress. The transversecompression stress acts on the same fracture plane as the shear stress,


30/60


simultaneously. Hence there is no further crack opening but the fracture surfaces arepressed on each other.

3. Mode C: if the ratio of compressive normal stress at fracture and the transversecompressive strength exceeds a value of 0,4, the action plane of the external shearstress is no more the fracture plane but the fracture occurs on a plane inclined by an

angle 0FP

to the action plane of2

and21

.

From the physics point of view the Puck criterion should be preferred due to the introductionof the fracture process zone. In comparison, the Hashin theory seems to be moreconservative.

6.1.1.6 Failure criteria - Compression Strength for Monoli thic Composites

The compression strength of composite materials is dominated by two main reasons: fiberbuckling and composite delamination, (Martinez & Oller, 2009).

A shear buckling mode as well as the extensional buckling mode depending on the fiber

volume fraction was defined by (Rosen, 1965).Further improvements to describe the compression strength of composite materials can befound in (Jochum & Grandidier, 2004).

6.1.1.7 Failure cr iteria Bearing strength

The bearing stress correspond to the applied load P divided by the projected bearing areaonto the area orthogonal to the bearing direction, ie: the product of the nominal bolt diameterD and the specimen thickness t.

_bearing = P / (D x t)

An example of the resulting bearing stress/bearing strain curve is shown in Figure below.

The bearing strain was obtained by normalizing the displacement by the bolt diameter.

The offset bearing strength is the value to consider for calculations. Thus, the 2% offsetmeasurement, which is the default in the proposed standard (see figure below), correspond toa ovalisazion of 2% of the hole diameter.

Fig. 11 - Example of bearing stress/strain curve

Offset bearing strength

D

t

P


31/60


There is no general consensus as to what the value of the offset bearing strength should be.The usage in the aerospace industry varies from 1%D for stiff double shear joints, to 4%D forsingle shear joints (the latter being a standard for metal bearing tests).

6.1.1.8 Fracture Mechanical Approaches and Further Extensions for

Monolithic Composites

The characterization of cracks within structural components became possible by the definitionof stress intensity factors, (Irwin, 1957). The crack modes due to tension, planar shear andnon planar shear allow the definition of three different stress intensity factors. One drawbackmust be seen by the stress singularity at the crack tip where the stress value rise to infinity.

The definition of the J-Integral allows the assessment of linear elastic as well as inelasticmaterial behavior, (Rice, 1968). Due to the path independence of the integration, the stressintensity factor can be calculated as

221

IKE

J

(19)

To overcome the problem of stress singularity at the crack tip and to assess especiallydelamination phenomena, cohesive zone models can be used, (Dugdale, 1960), (Barenblatt,1962) or (Camanho, Dvila, & Ambur, 2001). The crack tip is extended to a fracture processzone where cohesive forces can cause a critical crack opening. Thereafter, the cohesiveforces degrade until the crack surfaces are stress free and the crack further propagates.

Beside the discrete fracture models described above, the continuum damage mechanicswhich uses a continuous degradation parameter to describe the loss of stiffness in a material,can be used. Therefore, the ratio of actual effective area without any defects to the initial areais assessed. Thus, the actual damage can be quantified in a smeared manner.

The numerical treatment of fracture and damage phenomena was pushed by (Mazars &Pijaudier-Cabot, 1989) by the introduction of a model to describe the damage localization.

Due to the smeared character not the crack itself but its action on the continuum is modeled.The use of discrete interface elements based on cohesive zone approaches was pioneeredby (Neeleman, 1987).

NOTE : For implementation issues of cohesive zone approaches into finite element program(Xu & Needleman, 1993), (Xu & Needleman, 1995), (Ortiz & Pandolfi, 1999), (Ghosh, Ling,Majumdar, & Kim, 2000), (Camanho, Dvila, & Ambur, 2001), (Camanho & Davila, 2002) or(Camanho, Davila, & de Moura, 2003) should be quoted.


32/60


6.1.2 Specific ities for a sandwich panel with core material

A sandwich panel is constituted of a light weight core (isotropic materials such as foams ororthotropic such as honeycomb, balsa ) embedded between two composite facesheets.

Fig. 12 - Sandwich element definition

The following sandwich parameters are used in the further analysis:

Ffacesheet properties

tf facesheet thickness

E1 youngs modulus in longitudinal direction

E2 youngs modulus in transversal direction

Ef youngs modulus geometrical average value equal to pE1 E2G12 in-plane shear Modulus

12 in-plane Poissons ratio

21 in-plane Poissons ratio

plasticity factor

waviness of facesheet

Ccore properties

tc core height

s cell size

Ec compressive modulus (in normal direction)

G13 core shear modulus in longitudinal direction

G23 core shear modulus in circumferential direction

c flatwise core compressive strength

d total height of sandwich (d=tc+ 2 tf)

6.1.2.1 Modeling

Depending on the level of through thickness stresses, choice of 2-D or 3-D elements shall bemade for sandwich panels analysis. If out of plane stresses may be neglected, in-plane 2-Danalysis (shell elements) may be used, otherwise,3-D elements should be used.

FEA of sandwich structures can be carried out with following element types or combinations:

a single layered shell elements for the entire sandwich material (for in-plane 2-Danalysis)


33/60


(layered) shell elements for the faces and solid elements for the core (for 3-D analysis)

solid elements for both faces and core (detailed 3-D analysis,).

The solid modeling could be restrained to local areas of complex geometry, load introduction where out of plane effects are more significant and require locally more complex modelling.

For the analysis of sandwich structures, special considerations shall be taken into account,such as:

elements including core shear deformation shall be selected

for honeycomb cores one shall account for material orthotropy, since honeycomb hasdifferent shear moduli in different directions

local load introductions, corners and joints, shall be checked

For many core materials, experimentally measured values of E, G and are not in agreementwith the isotropic formula. In that case, to assure that the shear response of the core will be

described accurately, the measured values for G and shall be used, and the E value shallbe calculated from the formula : E= 2 G (1+).

Modeling of skin laminates uses same methodologies than described chapter 6.1.1.

6.1.2.2 Failure modes and cr iteria

Failure of a sandwich panel can occur:

In the facesheets

In the core

at the core-facesheets interface

Failure in the facesheets

A laminate failure can occur in facesheets caused by an overstressing (Fig. 13).

Determination of strength allowables for the laminates uses same methodologies thandescribed chapter 6.1.1.

One can note that allowables material values used in facesheet analysis have to be fullyrepresentative of skin materials real strength, process effects on mechanical characteristicsshould be taken into account (for example, co-curing of composite skins on honeycomb maygenerate waves on the inner plies of skins and reduce strength) .

Fig. 13 - Laminate failure

Both faces shall be checked for failure, since they will be exposed to different stress states ifexposed to bending loads.


34/60


6.1.2.3 Transverse shear of core

This mode is driven by insufficient core shear strength or a low panel thickness (Fig. 14).

Fig. 14 - Transverse shear failure

In many cases the dominant stress in the core material is shear, causing shear yield, ultimatefailure, or tensile failure in 45 to the through thickness direction. In that case, it would bechecked that the through thickness shear stress does not exceed the shear strength.

When the stress state is more complex, a simplified version of Tsai-Wu criterion could beused for some closed cells foam material: see ref [R3]

Where F1d: compressive strength

F1z: tensile strength

S : shear strength

6.1.2.4 Local core crushing

This mode can be caused at locations with attachments to the panels by a low corecompressive strength (see Fig. 15)

Fig. 15 - Local core crushing

It would be checked that the local compression stress does not exceed the compressivestrength of the core material.

6.1.2.5 Global buckling of sandwichThis mode can be caused by an insufficient membrane or flexural stiffness of the sandwich oran insufficient core shear rigidity (Fig. 16).

Fig. 16 - Global buckling

The global buckling mode of sandwich panels is checked with same methodologies thanmetallic panels.

This analysis shall evaluate carefully:

boundary conditions effects


35/60


influence of geometrical imperfections

influence of high load is that could introduce partial damage in the structure (matrix

cracking or delaminations) , modify stiffness of sandwich panels and finally reduce the

buckling loads

6.1.2.6 local buck ling modes of sandwich

The five sandwich failure modes represented schematically in Fig. 17 have to be analyzes :

wrinkling lower bound (conservative) / wrinkling intermediate value

dimpling

shear crimping

flexural core crushing

Fig. 17 - Sandwich failure modes

Dimpling occurs only with honeycomb materials (intercellular buckling).

The allowable stresses of the single failure modes are given by :

wrinkling lower bound failure (pessimistic) :

wrinkling intermediate value failure :


36/60


where

dimpling failure

shear crimping failure

flexural core crushing failure (fcc)

6.1.2.7 skin /core debonding

This mode is driven by insufficient strength of the interface between skins and core of thesandwich panel.

It would be checked that :

the out of plane stress at the interface does not exceed the out of plane strength

the resultant shear at the interface2

23

2

13 does not exceed the shear strength

If it can be documented that the interface is stronger than the core, core properties can beused to describe the interface. For many sandwich structures made of foam core the interfaceis stronger than the core.


37/60


6.1.3 Static Modelling of Join ts

6.1.3.1 Bonded Joints

As mentioned before (see 6.1), laminates (as well as metal pieces) are in general modeled

using 2D elements in order to improve solving time.

When bonding two surfaces, no matter if metal or laminate pieces, the adhesive will bepreferably modeled with solid elements (usually 2 elements over the thickness for linearelastic evaluation), especially when thick-film bonding is used. This is shown in Fig. 18 herebelow:

Fig. 18 - Bonding between two pieces in Hypermesh

Between the surfaces, a so called freeze contact is defined. By this function, the bondingbetween adhesive and each adjacent layer/piece is defined by identical displacements of thenodes in this area.

Nevertheless, in case of large models (analysis on a full carbody for instance), simplifiedassumption such as no-bonded joint modelization (only perfectly tied) can be assumed inorder to safe calculation time. In that case, analysis of adhesive should be performed with adetail model.

6.1.3.2 Standard bolted joints without insert

In the first step modeling, standard bolted joints without insert (such as screws, bolts andrivets) can be done in the following way. 1D elements, with certain properties and crosssections, are connecting the layers where the connection shall be realized (Beam Elementwith increased thickness in order to prevent artificial bending moments). To map the contactface, which would usually be the bolt head, nut or an underlying washer, so-called 1D rigidelement, are used. This procedure is shown in Fig. 19 here after:

Fig. 19 - Standard bolted joints without insert in Hypermesh


38/60


After calculating, the bolt forces are obtained. With these forces, an evaluation with respect tothe according limiting standard (i.e. VDI2230) and values for screws, bolts or rivets can beundertaken. With respect to composite laminates this procedure is most likely for joints wheresmall loads are applied.

In a second step, if the bolt connection becomes critical or more relevant, preload to the bolts

and contact between the surfaces can be modeled in order to evaluate the joint moredetailed.

6.1.3.3 Standard bolted joints with Inserts

Joints with inserts in laminates are not commonly modeled by BT. Therefore, the followingsimple procedure is a suggestion and needs to be verified by tests as well.

As can be seen in the schematic drawing in Fig. 20 the insert is also modeled with 2D shellelements with metal material overlapping the 2D shell elements from the composite laminate.The bolt connection is again modeled as stated in 6.1.3.2 with 1D elements (Rigid 1Delements and 1D element BEAM). If large thicknesses (i.e. for sandwich structures) areexisting it is recommended to use 3D solid elements instead of 2D shell elements to map the

insert and laminate.

Fig. 20 - Schematic drawing of how to realize standard bolted joints with inserts


39/60


6.2 FATIGUE MODELLING

In principal, the distinction between endurance limit and fatigue strength can be made:

In case of endurance limit, the laminate must sustain the maximum load on a continuingbasis,

In case of fatigue strength, a damage accumulation occurs during the loading. Thedamage itself is therefore strongly influenced by the number of cycles, the ultimateloading and the order of the loading.

Fig 21 - Typical strength and stiffness degradation in composite (schematic)

Nowadays, there do not exist general fatigue calculation and assessment methods forcomposites, in contrary to metallic domain.

Due to this, the general design philosophy for composite structures in industries is based onfact that delamination during cyclic loading has to be prevented.

Some stress design limits (depending on material, stacking sequence, process, etc) based

on previous experience are considered in the early phase of the design, and then has to beconsolidated by fatigue tests.

For instance in the aeronautic field, the fatigue load (1 Mcycles) usually correspond to 20 to30% of the ultimate static load of the composite carbon material, and this level is prone to beacceptable without knockdown factor after residual strength. This kind of assumption can leadto conservative design strategies which act against to the lightweight potentials of modernfiber reinforced composites. Nevertheless, it shall not be considered as a general criterion.because it could be non-conservative in some other cases (other matrix, composite materialhealth, stacking sequences,).

In (Degrieck & Van Paepegem, 2001), three categories of fatigue models were defined:

1. Fatigue li fe models :These models use information from the S-N curves or Goodman-type diagrams topropose a fatigue failure criterion. Damage accumulation is not taken into account,whereas the maximum number of cycles for fatigue failure under fixed loading conditionsis predicted.

One fatigue life model usually considered is given by the Miners sum. In the Miners summethod, the results of a counting method and constant amplitude fatigue behaviordescription are converted into a damage parameter, D. Failure criterion of the laminateis considered when D>1 (see. Figure 22)


40/60


S

t

N1

N2 Ni

Fig. 22 Miners sum

The main limitations of this method are:o the degradation is assumed to be linear,o the potential effect of load order is not taken into account, that is

Fig. 23 Miners sum and load order

As a consequence, this method can be sometimes non conservative. Moreover, thevalue of the damage parameter only indicates whether or not failure occurred: it doesnot relate to a physically quantifiable damage.

Nevertheless, this method is today in use for the fatigue sizing of current area of windblades or even helicopter blade for instance, but with integrating specific coefficientbased on tests.

2. Phenomenological models to predict residual stiffness/strength :

These models predict the degradation of elastic properties during fatigue loading, where ascalar damage variable D=1-E/E0is commonly used to describe the loss of stiffness. Thedamage growth rate is then defined as the derivative dD/dN where N denotes the number

of cycles.

3. Residual strength models:The residual strength models are distinguished between sudden death models andwearout models. When the composite is subjected to high load levels within the low-cyclefatigue regime, the residual strength is initially constant and decreases drastically whenthe number of cycles to failure is nearly reached. For this, the sudden death model can beused to describe this phenomenon. If the composite undergoes a state of stress at lowload levels, the residual strength degrades more gradually and can be described bywearout models which incorporate the strength-life equal rank assumption, i.e. thestrongest specimen has either the longest fatigue life or the highest residual strength atrunout, (Degrieck & Van Paepegem, 2001).

n

i i

i

N

nD

1

DA= DB2

2

N

n

1

1

N

n

3

3

N

n

Loading A Loading B

1

1

N

n

2

2

N

n

3

3

N

n


41/60


Damage accumulationat high load level

Damage accumulationat low load level

Dama

geLevel

Number of Cycles

High Loads first

Characterisation of Composite Materials in Railways for Structural Calculations

Documents