Using ncode DesignLife for Fatigue of Composite Structures

Transcription

1 Using ncode DesignLife for Fatigue of Composite Structures Jeff Mentley HBM Prenscia October 5-6,

2 Agenda 3 1. Introduction 2. Short Fiber Injection Molded Composites Material Characterization Strain Energy Damage Parameter 3. Continuous Fiber Composite Failure Criteria 3 October 5-6,

3 Introduction 4 What is a fibre reinforced plastic composite material? A structural composite is a material system consisting of two or more phases on a macroscopic scale, whose mechanical performance and properties are designed to be superior to those of the constituent materials acting independently. (Daniel and Ishai, 2006) Ideal material when high stiffness to weight and strength to weight ratios are required For example: (Bloomberg, 2013) BMW i3 (fireballbase.com) Boeing 787 Dreamliner (1001crash.com) October 5-6,

4 Composite Materials 5 Categories of composite materials Particulate filler Discontinuous fibers Continuous fibers Other (e.g. sandwiches) Injection Molded Molding Compound October 5-6,

5 Composite Materials 6 Categories of composite materials Particulate filler Discontinuous Fibers Continuous Fibers Other (e.g. sandwiches) Unidirectional Woven fabrics Laminates Longitudinal direction Transverse direction Fill direction Warp direction Unidirectional composite lamina (Jones, 1999) Woven composite lamina (Jones, 1999) Unbounded view of a laminate (Jones, 1999) October 5-6,

6 Short Fiber Injection Molded Composites 7 October 5-6,

7 2016 ncode User Group Meeting Flow of material into mold controls microstructure 8 Courtesy of 2-D random Skin-core effect FLOW fibers aligned with flow fibers aligned transverse to flow Resulting material is: Inhomogeneous Anisotropic Temperature sensitive Environment sensitive (polyamides) Viscous component to behaviour October 5-6,

8 Process Modelling for Fatigue 9 Manufacturing simulation Basic material information and fatigue test data Fatigue material model Microstructure orientation, a ij Structural FE calculation Fatigue simulation Structural material model Loading Basic material information and physical test data Damage, LIFE October 5-6,

9 Fatigue characterisation challenges 10 Test coupons are: Inhomogeneous Anisotropic Leading to non uniform stress distribution spatially and through thickness We are not testing a material, but rather a complex structure Calibration of material models requires a measure of reverse engineering based on detailed FE analysis of coupons If we are creating SN curves What is stress? October 5-6,

10 Material characterisation and validation program 11 Nylon (PA66) with 30% glass fibers Injection molded plaques Determine microstructure Specimens cut from plaques 0, 45 & 90 degree orientations Environmental chamber Reduce environmental variations Room temperature (23 C) 50% humidity R = 1 loading Reduce viscoelastic behaviour Anti buckling guides Calibrate 3 fatigue damage models Simple interpolation method Digimat high cycle fatigue model Elastic strain energy approach October 5-6,

11 Experimental Measurement of Microstructure 12 0%W_40%L_Avg z/h A11 A12 A22 A33 A13 A E Orientation tensors evaluated at 9 points on plaque Histogram of length (um) Weibull Shape Scale N 669 At 13 points through thickness Averaged over 6 plaques Frequency Fiber length distribution determined at plaque centroid length (um) October 5-6,

12 Simulation Based Prediction of Microstructure 13 y z Courtesy of October 5-6,

13 Specimen preparation 14 Specimens are: Machined from plaques in 3 orientations Single specimen per plaque Gage region at plaque center October 5-6,

14 Stress analysis results of specimens 15 ANSYS Workbench 40,000 linear hex elements 13 layers through thickness Each layer assigned anisotropic properties based on measured microstructure Enforced displacements at gripped end tabs October 5-6,

15 Stress analysis results of specimens October 5-6,

16 Tensile tests and correlation 17 Mean effective modulus from tensile tests compared with FE solution Based on: Load cell and clip gauge Effective Modulus 0 Degree 45 degree 90 degree (MPa) FE Tensile test (mean) October 5-6,

17 Fatigue behaviour under test 18 Tests carried out in load control with strain monitoring using a clip gauge Elliptical hysteresis loops viscoelastic behaviour Cyclic softening throughout test (modulus change) Highly distorted hysteresis loop immediately prior to failure October 5-6,

18 Net section stress SN curves 19 Net section stress range October 5-6,

19 Strain Energy Density Damage Parameter 20 Damage equation W = (N f ) Strain energy history W(t) = ij (t) ij (t)/2 stiff not stiff flow direction = October 5-6,

20 WN curve based on local stresses and strains degree results conservative wrt W but based on very thin hot spot! Strain Energy Density Parameter W Net section stress life curve converted to local strain energy life curve based on the FE simulation results for each orientation October 5-6,

21 With simple averaging domain approach 22 Modified Strain Energy Density Parameter W Compensation for stress gradient effect by averaging strain energy density over a small volume (several elements) October 5-6,

22 Summary 23 Simple test coupons are actually complex structures. A proper understanding of material behaviour involves reverse engineering from physical tests, based on manufacturing simulations and detailed stress analysis of the coupons Strain energy density parameter shows some promise in correlating fatigue behaviour with material in different orientations Reduces fatigue testing by a factor of 3 Currently doing physical testing to develop and validate enhanced SFC methods October 5-6,

23 Continuous Fiber Composite Failure Criteria 24 October 5-6,

24 Composite Static Failure Criteria 25 Use of composite materials is still expanding Aerospace and Wind turbines largest users carbon fiber 43,500 tons used in 2012 Aerospace 18% Wind turbines 23% Automotive use of carbon fiber to be >$6 billion by 2020 overtaking aerospace Known technology already used within most FE solvers Utilises DesignLife's standard features to give the user an distinct advantage over what they currently have Combined with measured loads allows all load combinations to be analysed Current FE implementations require each load combination to be determined and solved separately October 5-6,

25 Failure mechanisms 26 Composite laminates fail in a variety of mechanisms (Herakovich, 1998) They all generally occur in combination: At the micro level, e.g.: Fiber fracture Fiber buckling Fiber splitting Fiber/matrix debonding Fiber pullout Matrix cracking At the laminate (macro) level: Fiber dominated failure Matrix dominated failure Delamination Failure modes in UD composites (Reinforced Plastics, 2003) Failure at the micro scale does not necessarily lead to immediate failure of a laminate Here we focus on failure at the macro scale October 5-6,

26 Failure criteria for composites 27 Metals Main assumptions: Linear elastic behaviour up to material yield strength, Isotropic behaviour 0 Load cases Same curve regardless of the loading direction, Typical curve October 5-6,

27 Failure criteria for composites 28 Metals For multiaxial load cases, a yield criteria is used to characterise the end of linear elastic behaviour Examples of yield criteria (Jones, 1999): Maximum stress Tresca Von Mises October 5-6,

28 Failure criteria for composites 29 Composites Main assumptions: Brittle material Linear elastic behaviour Anisotropic behaviour (x y coordinate system) 0 Load cases Typical curves The x y coordinate system is not convenient for defining a composite failure criterion because the x and y axes are not aligned with the principal material axes October 5-6,

29 Failure criteria for composites 30 Composites Main assumptions: Brittle material Linear elastic behaviour Orthotropic behaviour (1 2 material coordinate system) 0 Load cases Typical curves The 1 2 coordinate system is convenient for defining a composite failure criterion because the 1 and 2 axes are aligned with the principal material axes October 5-6,

30 Failure criteria for composites 31 Loading the material with various, combinations gives the following envelope (assuming 0for now) A failure criterion is a mathematical expression, under the form:, 1 giving a curve connecting the experimental points all together, 1 October 5-6,

31 Failure criteria for composites 32 The previous experimental envelope was utopian In reality, there is always a certain amount of scatter in the test results Examples of real experimental failure envelopes: ATJ S graphite (Tang, 1979) Satin weave glass/epoxy (Tang, 1989) October 5-6,

32 Failure criteria for composites 33 Influence of on the shape of the failure envelope,,,1 The envelopes are plotted when 0 100% of the material shear strength, 2D representation ( space) 3D representation ( space) October 5-6,

33 Composite Failure Criteria in ncode DesignLife 34 Limit or non interactive failure criteria Maximum stress (12.0) Maximum strain (12.0) Partially interactive failure criteria Christensen (12.0) Hashin Rotem (12.0) Hashin (12.0) Hashin Sun (12.1) NU (12.1) Fully interactive failure criteria Franklin Marin (12.0) Hoffman (12.0) Norris (12.0) Norris McKinnon (12.1) Tsai Hill (12.0) Tsai Wu (12.0) Custom (Python based), 1 1,,, October 5-6,

34 Futures 35 Failure criteria currently extended to fatigue Static failure envelope Fatigue failure envelope Approaches developed for counting cycles Fatigue property characterization studies currently inderway Beta versions currently installed at key OEM accounts October 5-6,

35 Questions? 36 October 5-6,

36 Jeff Mentley Connect with us on: linkedin.com/company/hbm ncode +ncode October 5-6,