Damage Tolerance Analysis of Repaired Composite Structures: Engineering Approach and Computational Implementation Mark R. Gurvich, Vijay N. Jagdale United Technologies Research Center 411 Silver Lane, East Hartford, CT 06108, USA GurvicMR@utrc.utc.com, JagdalVN@utrc.utc.com 2012 SIMULIA Community Conference Providence, RI, May 15-17, 2012 1
Agenda Introduction Motivation Major Ideas Objectives Concept and Approach Computational Implementation 2D Examples Statement Effect of Computational Parameters Effect of Physical Parameters 3D Examples Statement Parametric Studies Conclusions 2
Introduction Continuum Mechanics Perfect bonding assumption Continuum elements Max stress/strain criteria Max energy density criteria Sensitivity to mesh refinement Stress or strain at a distance Low computational cost Major ideas of FEA modeling Fracture Mechanics Presence of interfacial crack Crack propagation criteria Stress intensity factor approch Mix-mode scenarios Challenge of material data Complex experimental validat. Moderate computational cost Progressive Damage Specifically for FEA Damage initiation and growth Cohesive elements as solution Zero-volume interface Potential combination with continuum mechanics Robust FEA implementation Progressive Damage approach* provides advantages vs. Continuum & Fracture Mechanics methods. Simplification of CE-based analysis for practical engineering applications is focus of this study. * Kafkalidis & Thouless, 2002; Campilho et al., 2009a; Campilho et al., 2009b; Li et al., 2006; Pinto et al., 2010; de Moura et al., 2008; etc. Objectives 1. Develop engineering computational approach of DT assessment for repaired areas, capturing processes of both damage initiation and growth. 2. Demonstrate its robustness in Abaqus on representative 2D and 3D examples of composite repair designs. 3
Concept and Approach Typical rotorcraft structure under repair* Scheme of approach Original Composite Plies Repair Plies Extra Ply Previous repairs Areas for new repairs Original Core Repair Core Repair Zone Damage b Core/Skin Damage c a Interlaminar Damage d The approach captures both damage initiation and damage growth: a) scheme of actual repair [Ref. 8]; b) FEA statement of analysis; c) expected damage scenarios; d) model of all damage mechanisms; e), f) examples of models with selected individual paths of damage propagation. Cohesive-element based FEA is applied to model damage process e f * Kumar, Gurvich, et al. 67 th AHS Forum, Virginia Beach, VA, May 3-5, 2011. 4
Assumptions A. Interfacial properties between connected parts and interlaminar properties between individual layers are considered the weakest links of repair zones. to ignore consideration of cross-laminar damage and, more generally, complex damage networks, since interfacial and/or interlaminar damage alone seem to be severe enough for definition of the load limit. B. Structural integrity of repaired zone is controlled by the very worst scenario of damage initiation and growth. provides a convenient opportunity to consider only a few the most dangerous damage paths instead of numerous mathematically possible combinations of damage growth scenarios. C. Potential imperfections of repaired interfacial contacts are taken into account through introduction of small initial cracks in critical locations. 5
2D Examples Parametric statement of the problem and FEA model a b c d e Analyses in Abaqus/Standard 6.10-1 allowing large scale, nonlinear deformation to occur. Plane strain 4-node bilinear quadrilateral elements (CPE4). Cohesive and interlaminar damage by 4-node 2D cohesive elements (COH2D4). Cases of damage: A,B,C, D (D - numerous interlaminar damages C ). Load conditions: tension; bending (both directions), combined tension with bending. Displacement-controlled loading conditions. Representative material properties CFRP [0] 20 : Hexcel 8552 IM7 (unidirectional prepreg) Selection of the examples was motivated by work of Campilho et al., 2009b. 6
2D Examples Effect of FE mesh density (case A; a/w = 0.25) a b c d e m y m x Total elements Error in P max prediction (%) Computational time (%) 20 40 800 35.86 0.3 40 80 3,200 11.25 1.5 80 160 12,800 2.89 29.4 100 200 20,000 1.72 45.1 160 320 51,200 0.39 62.0 200 400 80,000 0.00 100.0 7
2D Examples Computational convergence and solution stabilization *Static, stabilize = dissipated energy fraction, allsdtol = accuracy tolerance w/o stabilization w/t stabilization Case Force and displacement results P max P max with without ΔP stabiliza stabiliza ma x (%) tion tion (N/mm) (N/mm) 8 δ max (mm) with stabilizat ion Time (min) without stabilizat ion ratio A 11,481 11,487 0.052 0.0318 33.80 6.42 5.3 B 16,995 17,003 0.047 0.0418 3.83 3.55 1.1 C 143,740 143,740 0 0.3294 29.47 11.02 2.7 Dissipated energy fraction = 0.0002; Accuracy tolerance of 0.05
2D Parametric Strain Analysis Distribution of strains Tension Bending Effect of geometry Effect of geometry 9
2D Parametric Limit Load Analysis Tension Bending Effect of damage mode Effect of repair geometry 10
3D Examples Statement of the problem: scarf repair Cohesive Zone a b c d Analyses in Abaqus/Standard 6.10-1 allowing large scale, nonlinear deformation to occur. 3D solid 8-node linear brick elements (C3D8). Cohesive damage by 8-node 3D cohesive elements (COH3D8). Displacement-controlled tension loading conditions. Representative material properties: Hexcel 8552 IM7 (unidirectional prepreg) 11
3D Examples Distribution of stresses and damage parameters in the cohesive zone S33 S13 S23 Stresses QUADSCRT SDEG Damage 12
3D Examples Distribution of QUADSCRT as function of applied load (%) 25% 43% 70% Damage initiation 81% 96% 100% 13
3D Examples Effect of initial crack (a/w) on distribution of damage parameters a/w=0.05 a/w=0.15 a/w=0.25 QUADSCRT a/w=0.05 a/w=0.15 a/w=0.25 SDEG 100% of failure load 14
3D Examples Definition of limit load and effect of repair imperfections Force-displacement diagram Dependence on initial cracks Introduced repair imperfections can be potentially used for safety assessment 15
Damage Tolerance (DT) of Repaired Composite Structures 1. A convenient engineering approach is suggested and demonstrated for DT assessment of repaired zones in composite structures. 2. Efficient FEA implementation of the approach is achieved through robust Abaqus capabilities using cohesive elements. 3. Limit loads can be predicted relatively easy by analysis of major expected damage paths and the follow-up worst case scenarios. 4. Repair imperfections can be taken into account through initial suggested cracks and potentially used for analysis of safety margins. 5. Successful 2D and 3D demonstrations suggest similar applications for analysis and optimization of a wide range of repair designs. 6. The follow-up efforts are focused on experimental validation of the approach and its expansion for fatigue life analysis. 16
Acknowledgments The authors thank UTRC for permission to present these results aerospace systems Sikorsky Carrier Pratt & Whitney power solutions UTC Power Hamilton Sundstrand UTC Fire & Security building systems Otis 17