Вы находитесь на странице: 1из 11

FAILURE CRITERIA FOR POLYMER COMPOSITES UNDER 3D STRESS STATES: THE SECOND WORLD-WIDE FAILURE EXERCISE

A. S. Kaddour* and M J Hinton


*

QinetiQ, Ively Rd, Farnborough, Hampshire, GU14 0LX, UK. E-mail: askaddour@qinetiq.com

QinetiQ, Fort Halstead, Sevenoaks, Kent, TN14 7BP, UK. E-mail: mjhinton@qinetiq.com

Copyright QinetiQ Ltd 2009

SUMMARY The authors (hereafter referred to as the organisers) are coordinating a Second World-Wide Failure Exercise (WWFE-II) to establish the current status of theoretical methods for predicting structural failure in fibre reinforced composite materials when subjected to 3-D states of stress. The exercise runs in two parts. Part A is devoted to providing full details of the theories together with predictions, made by their originators, for a standard set of test cases. Part B is concerned with comparing the theoretical predictions with experimental results. This paper is directed at exposing some of the early lessons emerging from Part A. Particular attention is focussed on two Test Cases, the first being an isotropic material subjected to a range of triaxial compressive stress states and the second being a unidirectional laminate subjected to the same conditions. Theoretical predictions are presented and preliminary observations are drawn in regard to the degree of applicability of the current theories. Keywords: triaxial, failure criteria, 3D stresses, isotropic, through-thickness, hydrostatic pressure.

1 Introduction
In order to set this paper in context, it is important to provide the reader with the background to the work. In 1992, the organisers set out on a coordinated study (known as the World-Wide Failure Exercise or WWFE) to provide a comprehensive description of the foremost failure theories for fibre reinforced plastic (FRP) laminates that were available at the time, a comparison of their predictive capabilities directly with each other, and a comparison of their predictive capabilities against experimental data. In the exercise, selected workers in the area of fibre composite failure theories, including leading academics and developers of software/numerical codes, were invited to submit papers to a strictly controlled format. To make traction in this field, the organisers focused WWFE on the response of classical, continuous fibre, laminated, fibre reinforced polymer composites subjected to in-plane biaxial loading situations, in the absence of stress concentrations. WWFE proved to be a groundbreaking effort with many achievements :-

It established, for the first time, an open and objective way of working in order to compare, contrast and challenge disparate theories from around the world. It exposed the strengths and weaknesses of the current theories. It provided a stimulus for researchers to build upon the accurate theoretical features whilst making improvements to deal with the shortfalls that were exposed. It highlighted gaps in experimental data and in theoretical understanding, and preliminary recommendations were made in terms of prioritisation and approach to their resolution. It provided design engineers (the ultimate customers for such research knowledge) with recommendations on the preferred theories to use, together with evidence of the level of confidence and bounds of applicability.

WWFE was completed successfully in 2004, having generated numerous publications (but best summarised in Ref[1]). A high priority gap, identified in WWFE, was the need to examine the fidelity of failure theories when applied to components under 3-D (ie triaxial) states of stress. Such stress states are commonly induced in thick composite components (rotor blades, pressure vessels), during impact and ballistic conditions and as a result of stress concentrations (bolted joints et al). In order to meet this need, the authors launched a Second World-Wide Failure Exercise (WWFE-II), in 2007, building upon and employing the principles established during WWFE, with the objective of extending the assessment of predictive failure criteria from 2D to 3D states of stress. WWFE-II is being run in two parts, following the guidelines adopted in WWFE :Part A is devoted to providing full details of the theoretical models and failure criteria of the participants. Part B is concerned with comparing the theoretical results with experimental results.

The WWFE-II is organised to run logically through a series of activities. A description of these activities and the associated completion dates are shown in Table 1. Table 1 Timeline for WWFE-II
Activity Definition of the scope of WWFE-II (Selection of Test Cases and supporting data) Identifying suitable participants /gaining their agreement to participate Issuing Part A data to participants Receipt of Part A submissions Issuing Part B experimental data to participants Publication of Part A in special edition of a suitable journal Date of completion (*) Completed Dec 2006 Completed Dec 2006 Completed March 2007 Completed Dec 2008-April 2009 July 2009 Sept 2009

Receipt of Part B submissions Publication of Part B in special edition of a suitable journal Publication of WWFE-II text book

March 2010 June 2010 Dec 2010

(*): These timings are indicative The arrangement allows both a blind test and a further opportunity for participants to offer refinements to the theories. At the time of writing, the authors are nearing the completion of Part A and this paper is aimed at :Providing an overview of WWFE-II Identifying the participants and the theories employed Defining the Test Cases that have been chosen and the supporting rationale Previewing an initial slice of the data from Part A

2 The Participants and their associated theories


A guiding principle employed by the organisers has been to invite the originator of a leading theory to act as the participant in the exercise, rather than utilising unconnected experts who might stray in their interpretation of the theory from that intended originally. Where that has proven to be impossible, connections have been made between the participant and the originator to minimise any variations and/or identify the reasons for such. The organisers started WWFE-II by approaching the original participants in WWFE, many of whom had presented 3-D theories in the first instance. Six of the participants accepted the invitation, thereby providing valuable continuity between the two exercises. The six were supplemented by inviting those who were regarded as representative of contemporary modelling tools and methodologies currently in use by research and design institutes around the world. The participants represent some twelve institutions /groups /individuals from seven countries. Table 2 provides a summary of the participants, their institute affiliation and references to the theory that each has employed. Table 2: A list of the participants for the WWFE-II
I.D. No. Participants Name Organisation 1 Bogetti, Staniszewski, Burns, Hoppel, GillespieU.S. Army Research Laboratory (USA) and Tierney, Ref[13] 2 3 4 5 6 7 8 9 Wolfe, Butalia, Zand and Schoeppner, Ref[10] Nelson, Hansen, Mayes, Ref[5] Deuschle and Kroeplin, Ref[6] Carrere, Laurin and Maire, Ref[9] Cuntze, Ref[3] Pinho, Darvizeh, Robinson, Camanho, Ref[8] Rotem, Ref[13] Zhou and Huang, Ref[11] Ohio State University, AFRL, Wright-Patterson, AFB, Ohio (USA) Firehole Technologies, Wyoming University, Alfred University (USA) ISD, Stuttgart university (Germany) ONERA (France) Retired Scientist (Germany) Schuecker,Imperial College (UK), NASA (USA), University of Porto ( Portugal) Technion University (Israel) Tongji University (China)

10 11 12

Kress, Ref[4] Ye, Zhang and Sheng, Ref[7] Ha, Jin and Huang, Ref[15]

ETZ Zurich (Switzerland) Leeds University, Manchester University (UK), Hefei University (China) Hanyang University (S Korea)

3 Description of the Test Cases


The Test Cases in WWFE-II (described in Ref[2]) have been chosen carefully to stretch each theory to the full in order to shed light on their strengths and weaknesses. They are focused on a range of classical, continuous fibre, laminated, reinforced polymer composites subjected, in the absence of stress concentrations, to a variety of triaxial loading conditions. The key issues being explored are :The means by which the theories distinguish (if at all) between the effects of anisotropy and heterogeneity. The types of failure mechanism employed and the way that each is implemented within any given theory. The accuracy and bounds of applicability of each theory

Twelve Test Cases were identified for the purpose. They employ five lay-ups :(1)- a base resin with isotropic properties, (2)- a unidirectional laminate, (3)- a cross ply laminate, (4)- an angle ply laminate and (5)- a quasi-isotropic laminate. Six different fibre/matrix combinations were used and these are:(1)- an epoxy, (2)- T300/epoxy, (3)- E-glass/epoxy, (4)-S-glass/epoxy, (5)- A-S Carbon/epoxy and (6)-IM7/8551 materials. Full details of the Test Cases are provided in Table 3 (below). Table 3 Details of the Test Cases used in WWFE-II
Test Case 1 2 3 4 5 6 7 8 9 10 11 Laminate lay-up Resin 0 0 0 90 0 0 35 35 (0/90/45)s (0/90)s Material MY750 epoxy T300/PR319 T300/PR319 T300/PR319 E-glass/MY750 S-glass/epoxy carbon/epoxy E-glass/MY750 E-glass/MY750 IM7/8551-7 IM7/8551-7 Required predictions 2 versus 3 (1 = 3 ) envelope 12 versus 2 (1 =2 = 3 ) envelope 12 versus 2 (1 =2 = 3 ) envelope Shear stress strain curves ( 12- 12 ) (for 1 =2 = 3 =600MPa) 2 versus 3 (1= 3 ) envelope 1 versus 3 (2= 3 ) envelope 1 versus 3 (2= 3 ) envelope y versus z (x= z ) envelope Stress-strain curves (y -x and y -y) at z = x =100MPa yz versus z (y =x =0 ) envelope yz versus z (y =x =0 ) envelope

12

(0/90)s

IM7/8551-7

Stress-strain curves (z -z, z -x and z -y) for y = x =0

4 A Preview of an initial Slice of the Part (A) data


The present paper deals with a slice of the assessment of the theories - The attention here is focussed on two important Test Cases, referred to as Test Case 1 and 5 in Table 2. Test Case 1 deals with an isotropic material subjected to a range of triaxial compressive stress states and Test Case 5 is concerned with a unidirectional laminate subjected to the same conditions. Initial observations are drawn in regard to the degree of applicability of the current theories to both Cases. The Test Cases selected here are important building blocks in our understanding of traditional, high performance, continuous fibre, laminated composites which typically contain 60% fibre volume fraction in order to gain both stiffness and strength. It is accepted that a number of failure models rely on micro-mechanics to model the behaviour of the composite starting from that of the constituents, namely fibres and matrix. For this reason, a full understanding of the response of resin matrix is crucial for gaining an insight into how these micro-mechanics models tackle the behaviour of one of the main constituents of a composite material. The two Cases are interrelated insofar as the epoxy polymer material studied in Case 1 is the same resin matrix used in making the E-glass/epoxy composite laminate in Test Case 5. Hence, in choosing Test Cases 1 and 5 these will begin to illuminate the assumptions made in each theory regarding the treatment of material isotropy/anisotropy and material heterogeneity. The mechanical properties are provided in Table 4. Note that the epoxy exhibits isotropic stiffness but anisotropic strength properties, with a uniaxial tensile strength that is lower than the uniaxial compressive strength. Table 4 Mechanical properties for materials in Test Cases 1 and 5, Ref[2].
Test Case No Material Longitudinal modulus E1 (GPa) Transverse modulus E2 (GPa) In-plane shear modulus G12 (GPa) Major Poisson's ratio 12 Through-thickness Poisson's ratio 23 Longitudinal tensile strength XT (MPa) Longitudinal compressive strength XC (MPa) Transverse tensile strength YT (MPa) Transverse compressive strength YC (MPa) In-plane shear strength S12 (MPa) Through-thickness shear strength S23 (MPa) Test Case 1 Epoxy 3.35 3.35 1.24 0.35 0.35 80 120 80 120 54 54 Test Case 5 E-Glass/epoxy 45.6 16.2 5.83 0.278 0.4 1280 800 40 145 73 50

The loadings in both Test Cases involve the application of three direct stresses. The coordinate system used here is shown in Figure 1 where directions 1, 2 and 3 are those

2 2 1 1

in the fibre direction and transverse and through-thickness directions. For the polymer material, the same coordinate system is used to describe the three principal stresses. As there are many combinations of stresses that can be applied under 3D loadings, Test Cases 1 and 5 deal with a section through the 3D space and the state of stress consists of the following:

For Case 1: combined 2 and 1 (= 3): The stresses applied are such that those in 1 and 3 directions are equal while that in 2 direction varies proportional to that in 1 direction. For Case 5: Combined 2 and 1 (= 3) where 1 .is applied parallel to the fibre direction, 2 is applied in the through thickness direction and 3 is applied in the transverse direction (see Fig 1). In the present analysis, the application of equal triaxial compressive stresses is commonly referred to as hydrostatic compressive loading where : (-2)= (-1)= (-3)= -P where P is the equivalent hydrostatic pressure. The application of equal triaxial tension (ie (2)= (1)= (3) causes an analogous hydrostatic tensile state (ie +P).

Figure 1: Coordinate systems used in the Test Cases

5 Comparison between predictions


In order to avoid prejudicing the ethos of the overall exercise, for the purposes of this preview paper the data is presented without specific reference to the originating author (readers will need to wait until Part A is published, in full, for this information). Instead the theories employed are referred to as A, B, C, . and L (note that these letters have been assigned randomly). Though this reduces the impact somewhat, a number of very useful lessons can be drawn from the information. Predictions from two of the laminate configurations, defined earlier, have been selected and are discussed more fully below. Test Case 1 - MY750 Epoxy Figure 2 shows the failure envelopes predicted by different contributors for this configuration. The envelopes are superimposed in order to observe the general differences between the various predictions. The envelopes are split into groups to facilitate visualisation of differences and similarities between the curves. Test Case 5 - 90 E-Glass / MY750 Epoxy Figure 3 shows the failure envelopes predicted by different contributors for this configuration. The envelopes are superimposed in order to observe the general differences between the various predictions. The envelopes are split into groups to facilitate visualisation of differences and similarities between the curves.

The data from Test Cases 1 and 5 share similar features which raise a number of interesting issues regarding the assumptions used in the theories. (a) The interaction between the stresses is generally nonlinear, except in the models presented by Model F. In some portions of the envelope, other theories (models B, D, E) predicted linear interaction between the stresses. (b) In the tension-tension portion of the envelope, some theories predicted enhancement in the strength and others predicted no enhancement. For Test Case 1, it was noted that some of the participants introduced various forms of cut-offs in the envelopes in the tension-tension space. The models involved are B, C and I and those were introduced to indicate the dominance of different modes of failure. No equivalent cut-off was made in Test Case 5. (c) Table 4 provides a summary of the general shapes of the envelopes. These shapes are categorised as (a) open and (b) closed envelopes. In the last column, the models are described either as seamless or switched, depending upon the types of analysis used. The results in Figures 2 and 3 and those in Table 4 may be used to classify the models as follows: (I)- Theories that predict an open envelope under hydrostatic compressive stresses (all except theories A and F in Test Case 1 and all except theories A, B, C, E, F, G, I, K and L in Test Case 5). In other words, 2 out of 12 models predicted closed envelopes in Test Case 1 while 9 out of 12 models predicted closed envelopes in Test Case 5. (II)- Theories that predict an open envelope under hydrostatic tension stresses (just one theory (D) in Test Case 1, and no theories in Test Case 5). (d) The theories can also be classified according to the type of analysis used :(1)- Analysis Type 1 (Seamless Models): These theories employ identical equations for both the isotropic material (Test Case 1) and the anisotropic material (Test Case 5). These will be referred to as seamless models. They include A and F. In one of these models (Model F), the application of the composite criteria to isotropic material resulted in two sets of predictions, depending upon the interpretation of the stresses applied. Based on these two sets of prediction, the innermost (i.e. most conservative) envelope, resulting from the interaction between the two curves, was considered in the present work, as recommended by the participants. (2)- Analysis Type 2 (Switched Models): These theories use different equations, one for the isotropic material (Test Case 1) and a separate set for the anisotropic material (Test Case 5). These will be referred to as switched models. They involve 10 of the models employed (B, C, D, E, G, H, I, J, K and L). It can be then concluded that the majority of the participants used switched models.

1= 3 (MPa)
0.0 -1500.0 -1000.0 -500.0 0.0

A B

-500.0

2 (MPa)

C G J L

-1000.0

-1500.0

1= 3 (MPa)
0 -1500 -1000 -500 0

D E 2 (MPa) F H I K
-1500

-500

-1000

Figure 2: Failure envelopes for Test Case 1 for a pure resin epoxy material under triaxial stresses. All of the 12 curves were predicted by the participants of WWFE-II using their own models (A to L).

1=3 (MPa)
0 -1500 -1000 -500 0

2 (MPa)

B C G J

-500

-1000

-1500

Note: Model C predicted closed envelope at large stresses

1=3 (MPa)
-1500 -1000 -500

0 0

2 (MPa)

E F H I

-500

-1000

-1500

Figure 3: Failure envelopes for Test Case 5, describing the behaviour of E-glass/epoxy material under triaxial stresses. All of the 12 curves were predicted by the participants of WWFE-II using their own models (Models A to L).

Table 5 : States of the failure envelope under hydrostatic compressive stresses for a polymer and a UD composite obtained from the participants for the WWFE-II. Model used A B C D E F G H I J K L Test Case 1 Closed Open Open Open Open Closed Open Open Open Open Open Open Test Case 5 Closed Open Closed Open Closed Closed Closed Open Closed Closed/open Closed Closed Analysis Type(*) Seamless Switched Switched Switched Switched Seamless Switched Switched Switched Switched Switched Switched

(*) The word Seamless means the same equations are used for both polymer and composite materials. The word Switched means that the strength equations used for polymer are different to those for composite materials

6 Conclusions
(a) WWFE-II is now underway with twelve leading, and internationally recognised, groups taking part by employing their methods to solve 12 challenging Test Cases. WWFE-II is utilising the Part A /Part B format successfully pioneered via WWFE-I. (b) Initial results for two of Test Cases (Nos 1 and 5) from WWFE-II are presented in the form of predicted failure envelopes. The various models were employed to predict the failure of an isotropic un-reinforced polymer matrix material (Test Case 1) and an anisotropic, heterogeneous, E-glass/epoxy unidirectional lamina (Test Case 5) under triaxial stresses. (c) The overwhelming majority of theoreticians (9 out of 12) employed separate equations to delineate between isotropic and heterogeneous materials. This appears not to be an opaque feature of the models employed and rather it appears to require a conscious operator intervention to make that selection, based on an examination of the problem to be analysed. From a designers perspective, the preference would be for a black box modelling tool that contains sufficient resilience to provide accurate predictions in all circumstances. It remains to be seen if the theories employed within WWFE-II will satisfy that aspiration. (d)There was significant diversity between the theoretical predictions in terms of the shapes of the failure envelopes and whether or not the envelopes should be open under hydrostatic compressive and/or tensile loading situations. Clearly, Test Cases 1 and 5 indicate that some of the theories must contain incorrect assumptions. The

remaining Test Cases will undoubtedly provide further evidence of the resilience of the 12 theories.

References
[1] Hinton M J, Kaddour A S and Soden P D, Failure Criteria In Fibre Reinforced Polymer Composites: The World-Wide Failure Exercise, published by Elsevier Science Ltd, Oxford, UK, 2004. [2] Kaddour A S and Hinton M J, Input data for Test Cases used in benchmark triaxial failure theories of composites, to be published. [3] Cuntze R G, The predictive capability of failure mode concept - based strength conditions for laminates composed of UD laminae under static tri-axial stress states, to be published. [4] Kress G, Examination of Hashin's Failure Criteria for the Second World-Wide Failure Exercise, to be published. [5] Emmett E. Nelson, Andrew C. Hansen and Steven Mayes, Failure analysis of composite laminates subjected to hydrostatic stresses: A multicontinuum approach, to be published. [6] Deuschle H M and Kroeplin B-H, FE Implementation of Pucks Failure Theory for Fibre Reinforced Composites under 3D-Stress, to be published. [7] Ye J, Zhang D and Sheng H, Prediction of failure envelopes and stress strain curves of composite laminates under triaxial loads, to be published. [8] Pinho S T , Darvizeh R, Robinson P, Schuecker C and Camanho P P, Material and structural response of polymer-matrix fibre-reinforced composites, to be published. [9] Carrere N, Laurin F, and Maire J-F, Micromechanical based hybrid mesoscopic 3D approach for non-linear progressive failure analysis of composite structures, to be published. [10] Zand B, Butalia T S, Wolfe W E, and Schoeppner G A, A Strain Energy Based Failure Criterion for Nonlinear Analysis of Composite Laminates Subjected to Triaxial Loading, to be published. [11] Zhou Y X and Huang Z-M, A bridging model prediction of the ultimate strength of composite laminates subjected to triaxial loads, to be published. [12] Bogetti T A, Staniszewski J, Burns B P, Hoppel C P R, Gillespie, Jr. J W and Tierney J, Predicting the Nonlinear Response and Progressive Failure of Composite Laminates Under Tri-Axial Loading, to be published. [13] Rotem A, The Rotem Failure Criterion for Fibrous Laminated Composite Materials: Three Dimensional Loading Case, to be published. [14] Kaddour A S and Hinton M J, Comparison between the predictive capabilities of 3D failure criteria, to be published. [15] Ha S K, Jin K K and Huang Y. Prediction of composite laminate failure with micromechanics of failure, to be published.

Вам также может понравиться