3D crack propagation with XFEM cohesive elements

Similar documents
Modélisation d'interfaces linéaires et non linéaires dans le cadre X-FEM

Some improvements of Xfem for cracked domains

Abstract. 1 Introduction

FLOATING NODE METHOD AND VIRTUAL CRACK CLOSURE TECHNIQUE FOR MODELING MATRIX CRACKING- DELAMINATION MIGRATION

An Energy Dissipative Constitutive Model for Multi-Surface Interfaces at Weld Defect Sites in Ultrasonic Consolidation

Cohesive Band Model: a triaxiality-dependent cohesive model inside an implicit non-local damage to crack transition framework

Cohesive band model: a triaxiality-dependent cohesive model for damage to crack transition in a non-local implicit discontinuous Galerkin framework

Modelling Localisation and Spatial Scaling of Constitutive Behaviour: a Kinematically Enriched Continuum Approach

EDEM DISCRETIZATION (Phase II) Normal Direction Structure Idealization Tangential Direction Pore spring Contact spring SPRING TYPES Inner edge Inner d

Discrete Element Modelling of a Reinforced Concrete Structure

A Performance Modeling Strategy based on Multifiber Beams to Estimate Crack Openings ESTIMATE in Concrete Structures CRACK

Development of X-FEM methodology and study on mixed-mode crack propagation

6. NON-LINEAR PSEUDO-STATIC ANALYSIS OF ADOBE WALLS

Limit analysis of brick masonry shear walls with openings under later loads by rigid block modeling

COMPARISON OF COHESIVE ZONE MODELS USED TO PREDICT DELAMINATION INITIATED FROM FREE-EDGES : VALIDATION AGAINST EXPERIMENTAL RESULTS

Transactions on Engineering Sciences vol 6, 1994 WIT Press, ISSN

Finite element analysis of diagonal tension failure in RC beams

An Evaluation of Simplified Methods to Compute the Mechanical Steady State

CRACK GROWTH MODELLING: ENRICHED CONTINUUM VS. DISCRETE MODELS

NUMERICAL INVESTIGATION OF DELAMINATION IN L-SHAPED CROSS-PLY COMPOSITE BRACKET

MODELING OF THE WEDGE SPLITTING TEST USING AN EXTENDED CRACKED HINGE MODEL

Boundary element analysis of FRP-concrete delamination

TENSILE CRACKING VIEWED AS BIFURCATION AND INSTABILITY IN A DISCRETE INTERFACE MODEL

Checkerboard instabilities in topological shape optimization algorithms

Multi-scale approach of the mechanical behavior of RC structures Application to nuclear plant containment buildings

FRACTURE MECHANICS IN ANSYS R16. Session 04 Fracture Mechanics using Cohesive Zone Material (CZM) Model

Multi-scale digital image correlation of strain localization

Fracture Mechanics, Damage and Fatigue Linear Elastic Fracture Mechanics - Energetic Approach

Critical applied stresses for a crack initiation from a sharp V-notch

A 3D Discrete Damage Modeling Methodology for Abaqus for Fatigue Damage Evaluation in Bolted Composite Joints

Modeling of Interfacial Debonding Induced by IC Crack for Concrete Beam-bonded with CFRP

Nonlocal computational methods applied to composites structures

An exact reanalysis algorithm using incremental Cholesky factorization and its application to crack growth modeling

Comparison between thick level set (TLS) and cohesive zone models

Code_Aster. Reaction of the steel subjected to corrosion

PROPAGATION OF CURVED CRACKS IN HOMOGENEOUS AND GRADED MATERIALS

MESHLESS COHESIVE SEGMENTS METHOD FOR CRACK INITIATION AND PROPAGATION IN COMPOSITES

Aim of the study Experimental determination of mechanical parameters Local buckling (wrinkling) Failure maps Optimization of sandwich panels

Cracked concrete structures under cyclic load

An accelerated predictor-corrector scheme for 3D crack growth simulations

Numerical simulation of delamination onset and growth in laminated composites

Model-independent approaches for the XFEM in fracture mechanics

Snap-back analysis of fracture evolution in multi-cracked solids using boundary element method

Structural behaviour of traditional mortise-and-tenon timber joints

Finite element analysis of longitudinal debonding between fibre and matrix interface

A TIME-DEPENDENT DAMAGE LAW IN DEFORMABLE SOLID: A HOMOGENIZATION APPROACH

The Effects of Transverse Shear on the Delamination of Edge-Notch Flexure and 3-Point Bend Geometries

CHARACTERIZATION, ANALYSIS AND PREDICTION OF DELAMINATION IN COMPOSITES USING FRACTURE MECHANICS

MODELLING MIXED-MODE RATE-DEPENDENT DELAMINATION IN LAYERED STRUCTURES USING GEOMETRICALLY NONLINEAR BEAM FINITE ELEMENTS

Generalized fracture toughness for specimens with re-entrant corners: Experiments vs. theoretical predictions

The cohesive-frictional crack model applied to the bi-material interface between a dam and the foundation rock

3D dynamic crack propagation analysis with PDS-FEM

FRACTURE IN HIGH PERFORMANCE FIBRE REINFORCED CONCRETE PAVEMENT MATERIALS

Cracking in Quasi-Brittle Materials Using Isotropic Damage Mechanics

On characterising fracture resistance in mode-i delamination

CAST3M IMPLEMENTATION OF THE EXTENDED FINITE ELEMENT METHOD FOR COHESIVE CRACK

Application of a non-local failure criterion to a crack in heterogeneous media S. Bavaglia*, S.E. Mikhailov*

Studies of Bimaterial Interface Fracture with Peridynamics Fang Wang 1, Lisheng Liu 2, *, Qiwen Liu 1, Zhenyu Zhang 1, Lin Su 1 & Dan Xue 1

Comparison between a Cohesive Zone Model and a Continuum Damage Model in Predicting Mode-I Fracture Behavior of Adhesively Bonded Joints

Adaptive Analysis of Bifurcation Points of Shell Structures

Transactions on Modelling and Simulation vol 10, 1995 WIT Press, ISSN X

TIME-DEPENDENT MESOSCOPIC MODELLING OF MASONRY USING EMBEDDED WEAK DISCONTINUITIES

VERIFICATION OF BRITTLE FRACTURE CRITERIA FOR BIMATERIAL STRUCTURES

Simulation of Dynamic Delamination and Mode I Energy Dissipation

Shear Lag And Beam Theories For Structures

ANSYS Explicit Dynamics Update. Mai Doan

NUMERICAL SIMULATION OF FLUID-STRUCTURE INTERACTION PROBLEMS WITH DYNAMIC FRACTURE

Practical methodology for inclusion of uplift and pore pressures in analysis of concrete dams

5 ADVANCED FRACTURE MODELS

Size effect in the strength of concrete structures

Tensile behaviour of anti-symmetric CFRP composite

Interlaminar fracture characterization in composite materials by using acoustic emission

3D numerical modelling of twisting cracks under bending and torsion of skew notched beams

Numerical Modeling of Interface Between Soil and Pile to Account for Loss of Contact during Seismic Excitation

Modeling Fracture and Failure with Abaqus

A FINITE ELEMENT MODEL FOR THE ANALYSIS OF DELAMINATIONS IN FRP SHELLS

University of Sheffield The development of finite elements for 3D structural analysis in fire

NUMERICAL SIMULATION OF THE INELASTIC SEISMIC RESPONSE OF RC STRUCTURES WITH ENERGY DISSIPATORS

NUMERICAL SIMULATION OF THE NON LINEAR BEHAVIOUR OF REINFORCED CONCRETE STRUCTURES

Modelling of non linear soil-structure interface behaviour using M. Boulon & P. G arnica

Fracture Mechanics of Non-Shear Reinforced R/C Beams

TREBALL FINAL DE MÀSTER

202 Index. failure, 26 field equation, 122 force, 1

3D XFEM Modelling of Imperfect Interfaces

Transactions on Engineering Sciences vol 6, 1994 WIT Press, ISSN

Powerful Modelling Techniques in Abaqus to Simulate

Constitutive models: Incremental plasticity Drücker s postulate

FVM for Fluid-Structure Interaction with Large Structural Displacements

EXPERIMENTAL CHARACTERIZATION AND COHESIVE LAWS FOR DELAMINATION OF OFF-AXIS GFRP LAMINATES

Fracture Test & Fracture Parameters of Self Compacting Concrete using ANSYS. Zeel Vashi 1,Megha Thomas 2 I. INTRODUCTION

ON EFFECTIVE IMPLICIT TIME INTEGRATION IN ANALYSIS OF FLUID-STRUCTURE PROBLEMS

Numerical Modelling of Blockwork Prisms Tested in Compression Using Finite Element Method with Interface Behaviour

Models for dynamic fracture based on Griffith s criterion

Cracking in Quasi-Brittle Materials Using Isotropic Damage Mechanics

Fig. 1. Different locus of failure and crack trajectories observed in mode I testing of adhesively bonded double cantilever beam (DCB) specimens.

Instabilities and Dynamic Rupture in a Frictional Interface

ALGORITHM FOR NON-PROPORTIONAL LOADING IN SEQUENTIALLY LINEAR ANALYSIS

Non-Linear Finite Element Methods in Solid Mechanics Attilio Frangi, Politecnico di Milano, February 17, 2017, Lesson 5

Fonctionnalités de Code_Aster version 12

LIMIT LOAD OF A MASONRY ARCH BRIDGE BASED ON FINITE ELEMENT FRICTIONAL CONTACT ANALYSIS

Transcription:

Transactions, SMiRT-23, Paper ID 472 3D crack propagation with XFEM cohesive elements Patrick Massin 1, Guilhem Ferté 1,2, and Nicolas Moës 2 1 Institute of Mechanical Sciences and Industrial Applications, UMR EDF-CNRS-CEA-ENSTA 9219, Université Paris-Saclay, 828 Boulevard des Maréchaux, 91762 Palaiseau Cedex, France. 2 GeM, Centrale Nantes, Université de Nantes, UMR CNRS 6183, 1 rue de la Noë, BP 92101, 44321 Nantes, France ABSTRACT In case of brittle failure of metallic components or reinforced concrete structures, trajectories are often unknown in case of mixed mode loading. The XFEM cohesive model presented in this communication will address the issues of crack bifurcation and crack advance. A procedure in four steps is adopted: computation of the equilibrium state in the presence of cohesive forces with a given potential crack surface, detection of the updated crack front on the surface from the computed cohesive state, determination of bifurcation angles along the front, and update of the potential crack surface accordingly. The cohesive model that is used allows initial perfect adherence. It relies on the use of an appropriate XFEM space of Lagrange multipliers, on the use of a mortar formulation to write the cohesive law from quantities defined over this space in an appropriate manner, and finally on a lumping strategy leading to block-wise diagonal operators. The originality of the approach lies in the a posteriori computation of the crack advance speed that is naturally embedded in the cohesive model, while in most of the literature it is determined beforehand based on the stress state ahead of the front. Several numerical tests have been carried out in mixed mode I and II to reproduce 3D non planar crack paths and showed good accordance with previous results from the literature. Most situations are quasi-statics with load controlled strategies and extension to dynamics is also discussed. INTRODUCTION To evaluate the harmfulness of defaults detected in metallic components or concrete structures, EDF has developed advanced simulation tools such as Code_Aster, numerical software for finite element analyses in mechanical engineering. The class of phenomena aimed at in this communication concerns 3D crack propagation with a priori unknown trajectories, in fatigue but also in transient dynamics when unstable propagation occurs as in brittle rupture. To tackle this issue, we propose to associate the extended finite element method [1] with the cohesive approach [2,3,4]. Cohesive zones are defined by surfaces extended tangentially from an existing crack surface. Hence the cohesive behaviour will separate naturally adherent zones from completely opened crack surfaces. An original implicit update of the crack front is performed. It requires a robust treatment of non regular interface laws combined with XFEM in Code_Aster [6]. In statics, it relies on the use of an appropriate XFEM space of Lagrange multipliers [5,6], on the use of a mortar formulation to write the cohesive law from quantities defined over this space in an appropriate manner, and finally on a lumping strategy leading to block-wise diagonal operators. Then, a directional criterion using information on the cohesive zone behind the crack front is applied. It allows us to propose a completely automated crack propagation procedure for which the trajectory is not known a priori that is compared to experimental results of the literature.

In dynamics, an initially perfectly adherent cohesive law treated implicitly with an explicit time scheme is used which allows an analytical determination of cohesive forces with an appropriate discretisation. The formulation is first validated on a conic DCB test specimen before being extended to industrial applications. QUASI-STATIC PROPAGATION WITH X-FEM COHESIVE ELEMENTS In most strategies using cohesive elements with crack propagation, crack advancement is given in an explicit way, information being provided by the stress field ahead of the crack front. We propose here an alternate solution for which the cohesive law is applied on a large potential surface of crack propagation ahead of the current crack front in its tangential direction. The advance of the crack front is then provided in an implicit manner, relying on the behaviour of the cohesive law to determine by post-processing the status of the points belonging to the potential surface of crack propagation. Equilibrium and behaviour law will then determine naturally the separation between adherent zones and open zones of the cohesive surface. A necessary ingredient is the insertion of robust non regular interface laws within the XFEM framework. We can show for example that a cohesive law t c ([u]), [u] being the displacement gap on the interface and t c the cohesive pressure distribution, with finite initial stiffness, such as the one proposed in figure 1, does not allow describing appropriately large adherent zones. Figure 1. Penalised linear-softening mode-coupling cohesive law. Normal and tangential behaviours, with σ c the tensile strength and G c the fracture energy. If the penalisation parameter α 0 is too low, the solution is not correct and non physical opening is observed in adherent zones (zone 1 of figure 3). Inversely, if the penalisation parameter α 0 is too high, then numerical oscillations are observed (zones 2 and 3 of figure 4) which lead to misevaluation of the regime (adherence or debonding). This can be observed on the inclusion debonding test whose geometry and boundary conditions are represented on figure 2.

Figure 2. Inclusion debonding test. Figure 3. Normal opening for the inclusion debonding case. Figure 4. Normal tractions for the inclusion debonding case.

To take care of this inconvenient, several corrective solutions are applied: The introduction of an adapted Lagrange multiplier space λ, adapted to XFEM [5,6]. When equilibrium is reached λ is similar to the contact pressure. So as to improve robustness and performance, a «mortar» formulation may be adopted to assess the consistency of this space. The use of an infinite initial stiffness in the cohesive law t c ([u]; λ) expressed in the formalism of an augmented Lagrangian formulation, such that α 0 = 0 in figure 1. A diagonal expression of interface operators. The efficiency of the approach is illustrated with the pull out test of the inclusion previously described. Each one of these modifications allows, apart from reducing drastically oscillations, a reduction of the number of iterations of the Newton algorithm used to solve the equilibrium, as can be seen on table 1. Table 1. Evolution of the number of Newton iterations which are necessary to solve the debonding case in three time steps. The capability of the method to determine exactly the crack extension is based on the introduction of the cohesive law on a large potential crack surface. Crack propagation obeys then the following rules described on figure 5. Figure 5. Crack propagation procedure.

This procedure can be summarized as follows: A large potential crack surface from previous time step is present (5a) Equilibrium is computed with the large potential crack surface (5b). The propagated crack front is localised by post-processing cohesive variables (5c). Propagation direction along the crack front is determined using criteria such as the maximum principal stress [7] (5d). A new potential crack surface is determined (5e). The Lagrange multiplier space is modified accordingly as well as the internal variables of the cohesive law (5a). A new equilibrium is then computed to obtain the new crack front and so on (5b). In the localisation process of the crack front, a first rough crack front is determined using the intersection of the iso-zero of the internal variable α of the cohesive law with the faces of the mesh. Figure 6. Computation of a rough crack front.

From the cloud of points obtained at the previous step, we reconstruct with a regularizing operator the advance of the previous crack front up to the regularized positions as illustrated on figure 7. This advance regularized field is provided to a propagation algorithm and provides a new crack front much more regular than the initial one. The crack front is obtained as the intersection of the iso-zero of a normal level set which characterizes the distance to the crack surface and the iso-zero of a tangent level set which characterizes the distance to the surface orthogonal to the crack surface at the crack front. Then level set update algorithms are applied once the position of the new crack front is determined from the regularized operator. In this paper we used a geometrical update [8] of the level sets characterising the position of the crack surface and its front. Figure 7. Reconstruction of a smoothed crack front. The determination of the propagation crack angle is based on the criterion of the maximum of the circumferential stress provided by Erdogan and Sih [7], formulated in terms of stress intensity factors. Using stress intensity factors along cohesive zone models seems reasonable for concrete as suggested by [9,10] since experiments showed that the crack path, contrary to the load-deflection curve, is almost insensitive to the size of the cohesive zone, to the point where linear fracture mechanics is still valid. Moreover, since, with a cohesive zone, mechanical fields far away from the crack front have the same formulation than those of a free crack, we can define equivalent stress intensity factors [11]. We can show that these stress intensity factors can be deduced directly from the displacement gap and the cohesive stress using interface integrals [12]. There is therefore no need to construct contours surrounding the front and no need to compute auxiliary fields with this method.

This procedure is validated on three tests which can be found in the literature. The first one is the study of concrete crack propagation for a bracket specimen, described in [13]. Results obtained are in good agreement with the experimental ones, in terms of trajectory and load-displacement curve. It could still be improved using an exponential law for the cohesive zone rather than a linear or bilinear one. The second validation is performed with respect to a three point bending test of a Plexiglas specimen with a pre-crack at an angle with respect to the load direction [14,15]. In this case, only the crack trajectories are compared between the numerical simulations and the experiment, which show a good agreement. The last example focuses on the torsion of a parallelepiped concrete specimen pre-cracked at 45 degrees (see figure 8) [16]. The crack trajectory is complex, presented in figure 9, with an S shaped form. Once more the agreement between simulations and experiment in figure 10 is rather good from the point of view of trajectories and load-displacement curves, with a similar remark on the type of law used for the cohesive model. Figure 8. Prediction of a complex 3D trajectory for the torsion of a parallelepiped specimen pre-cracked at 45 degrees. Figure 9. S-shaped computed crack path and normal cohesive tractions for Brokenshire s test.

Figure 10. Load-deflection curve for Brokenshire s test. DYNAMIC PROPAGATION WITH X-FEM COHESIVE ELEMENTS Similarly to quasi-static propagation, we seek for a cohesive formulation to be applied on a large potential crack surface, which can describe correctly a partition in between large adherent zones and open ones. Once more, we will consider a cohesive law with initial infinite stiffness. It will be expressed in an implicit way within an explicit time scheme, the one of centered difference. This approach was proposed by Doyen et al. [17] for interfaces fitting the structural mesh: we propose here its extension to XFEM. Applying the specific space described before for the cohesive force with diagonal interface operators, cohesive forces can be solved analytically from scalar equations. The implementation is validated on a trapezoidal DCB specimen. In this experiment, a crack suddenly propagates dynamically before stopping inside the structure: its length when it stops is the quantity of interest to be studied. Several numerical simulations are performed with varying time steps as is often the case with explicit simulations: similar results are however obtained. It can be deduced from this result that the method is energetically reliable and allows determining correctly the length at arrest, even though some improvements could be performed. Figure 11. Trapezoidal Double Cantilever Beam. Load and Geometry.

For quasi-static loading, a load control technique can also applied to determine the critical load which triggers the unstable propagation. The one proposed by Lorentz [18] was chosen and extended to XFEM. It allows following instabilities during a quasi-static calculation, with the load intensity introduced as a new unknown in the system of equations to be solved, the maximal opening being prescribed in the cohesive zone. This method is equivalent to prescribing a maximal dissipation for which the load intensity is deduced. The crack length at arrest for the critical load obtained with this quasi-static method is smaller than the one predicted from the dynamic simulation, due to the fact that inertial effects are not taken into account (see figure 12). Figure 12. Load-deflection curve obtained with a path-following method and crack arrest for the Trapezoidal Double Cantilever Beam. Finally, the different steps to determine the trajectory of the crack during propagation are adapted from quasi-statics to dynamics. Even if the first results obtained to describe an unstable propagation are encouraging, the methodology seems difficult to be applied in case of transient phenomena involving shocks: in this case, the crack front is not clearly defined, being diffuse, which will make its update more difficult. Moreover, it is not easy to estimate a priori the cohesive process zone size as in quasi-statics, which could complicate the choice of calculation parameters to follow the crack path. CONCLUSION An original method is presented to predict crack path and load-deflection curve of brittle materials under quasi-static loading, combining XFEM and cohesive zone models. Instead of specifying the location of the crack front of the next load increment in advance, the cohesive law is defined over a large surface, so that the location of the crack front results from the equilibrium. This demands a robust insertion of cohesive laws with an exact enforcement of adherence with XFEM. We think the methodology could be extended to other interfacial behaviour such as contact-friction. In case of instability, a load control strategy may be used, but do not allow to represent the crack propagation correctly and more importantly its arrest, since inertial effects are not taken into account. Finally, using an explicit dynamics approach allows recovering a correct behaviour.

REFERENCES [1] Moës N., Dolbow J., Belytschko T. (1999), A finite element method for crack growth without remeshing, Int. J. Numer. Meth. Engng, 46:135-150. [2] Lorentz E. (2008). A mixed interface finite element for cohesive zone models. Comp. Meth. Appl. Mech. Engng., pages 317-320. [3] Wells G. and L. Sluys L. (2001). A new method for modelling cohesive cracks using finite elements. Int. J. Num. Meth. Engng., 50:2667-2682, 2001. [4] Mergheim J., E. Kuhl E., and P. Steinmann P. (2005). A finite element method for the computational modelling of cohesive cracks. Int. J. Num. Meth. Engng., 63:276-289. [5] Moës N., Béchet E., and Tourbier M. (2006) Imposing dirichlet boundary conditions in the extended finite element method. Int. J. Num. Meth. Engng, 67:1641-1669. [6] Géniaut S., Massin P., and Moës N. (2007). A stable 3d contact formulation for cracks using x-fem. Revue Européenne de Mécanique Numérique, 75:259-276. [7] Erdogan F. and Sih G.C. (1963). On the crack extension in plane loading and transverse shear. Journal Basic Engng., 85:519-527. [8] Colombo D. (2012). An implicit geometrical approach to level-set update for 3d non-planar x-fem crack propagation. Comp. Meth. Appl. Mech. Engng., 240:39-50. [9] Bocca P., Carpinteri A., Valente S. (1991). Mixed-mode fracture of concrete, Int. J. Solids Struct., 27:1139-1153. [10] Cendon D., Galvez J., Elices M. and Planas J. (2000). Modeling the fracture of concrete under mixed loading, Int. J. Fracture 103:293-310. [11] Planas J., Elices M. (1992). Asymptotic analysis of a cohesive crack: 1. theoretical background, International Journal of Fracture 55:153-177. [12] Ferté G., Massin P., Moës N. (2015). 3D crack propagation with cohesive elements in the extended finite element method, submitted for publication. [13] Wrinkler B.J. (2001). Traglastuntersuchungen von unbewehrten und bewehrten Betonstrukturen auf der Grundlage eines objectiven Werksto_gesetzes für Beton. PhD thesis, Universität Innsbruck, 2001. [14] Citarella R., Buchholz F. (2008). Comparison of crack growth simulation by dbem and fem for senspecimen undergoing torsion or bending loading, Engineering fracture mechanics 75:489-509. [15] Lazarus V., Buchholz F.-G., Fulland M., Wiebesiek J. (2008). Comparison of predictions by mode ii or mode iii criteria on crack front twisting in three or four point bending experiments, International journal of fracture 153:141-151. [16] Brokenshire D., Barr B. (1996), Torsional fracture tests, Ph.D. thesis, Cardiff's University. [17] Doyen D., Ern A., and Piperno S. (2013). Quasi-explicit time-integration schemes for dynamic fracture with set-valued cohesive zone models. Computational Mechanics, 52:401-416. [18] Lorentz E. and Badel P. (2004). A new path-following constraint for strain-softening finite element simulations. International journal for numerical methods in engineering, 60:499-526.