CompositesWorld
Issue link: https://cw.epubxp.com/i/841219
JULY 2017 8 CompositesWorld DESIGN & TESTING » In the past 15 years, progressive damage analysis (PDA) for composites, implemented in finite element analysis software, has been under development. Some PDA proponents argue that "virtual testing" of this sort is, today, reliable enough to reduce or even replace the conventional "building block" approach to composite design that relies on physical testing. Although that could, in some cases, save significant time and cost in the development of composite structures, the results of a recent evaluation of the state of the art in this area, overseen by the Air Force Research Laboratory (AFRL, Wright-Patterson AFB, OH, US), indicates that comparative experimental fatigue data do not yet justify such confidence. Benchmarking PDA progress An AFRL benchmarking exercise titled, "Damage Tolerance Design Principles (DTDP)," conducted by Steve Engelstad of the Advanced Development Program at Lockheed Martin Aeronautics Co. (Lockheed Martin, Ft. Worth, TX, US) and myself from January 2014 through April 2015 asked the question, Can current PDA methods accurately predict initiation and growth of damage in composites? An intensive effort, it involved both static and fatigue blind predic- tions, using several PDA methods. To evaluate existing tools for composites damage progression modeling and prediction for future application of damage growth analysis needs, AFRL, as an impartial organizer, provided identical physical test data results to program participants who would use the data for PDA model calibration/ validation. e static data included 0° tension and compression, 90° tension and compression, V-notch shear, 90° three-point bend, Mode I double cantilever beam, and Mode II end-notched flexure. e fatigue data included 0° S-N, 90° S-N, ±45° S-N, 90° three-point bend S-N, Mode I double-cantilever beam fatigue, and Mode II end- notched flexure fatigue. Since brittle failure in 90° matrix-dominated tests resulted in low strength values and high fatigue scatter, AFRL is currently conducting follow-on research to improve test methods. Although there is not room here to examine the entire project, comparisons of fatigue predictions derived via PDA against physical open-hole fatigue test data for three different layups will serve to illustrate what we learned. e ultimate goal of this effort was to generate quality fatigue test results, along with high fidelity X-ray CT images that would enable us to assess the ability of the PDA methods to predict damage and residual strength after fatigue. e layups were of IM7 HexTow carbon fiber (Hexcel, Stamford, CT, US) and Cytec CYCOM 977-3 epoxy resin (Solvay Composite Materials, Tempe, AZ, US), in the following ply sequences: [0°/45°/90°/-45°] 2S , [60°/0°/-60°] 3S , [30°/60°/90°/-60°/-30°] 2S . It should be noted that in the static phase of the AFRL bench- marking exercise, nine analysis teams had already performed blind static predictions and then recalibrated the parameters in their models based on the results of 12 experimental load cases. Lessons learned during the program's static portion included more accurate How ready are progressive damage analysis tools? TABLE 1 Analysis teams and codes used for fatigue predictions Teams PDA Methods Vanderbilt University Eigendeformation-based Reduced order Homogenization (EHM) University of Dayton Research Institute (UDRI) B-Spline Analysis Method with Mesh Indepen- dent Cracking (BSAM with MIC) Global Engineering and Materials (GEM) Discrete Crack Network (DCN) NASA Glenn/University of Michigan (UM) Micromechanics Analysis Code with Generalized Method of Cells (MAC/GMC) Multi-scale Design Systems (MDS) Multi-scale Design System for Linking Continuum Scales (MDS-C, now part of Altair) AutoDesk/LM Aero Helius PFA (formerly ASCA) AlphaSTAR GENOA determination of material properties for some model inputs, appro- priate mesh sizes and orientations, and accurate representation of boundary conditions. Additionally, some PDA teams identified and corrected algorithm errors within their analysis codes. Seven of the teams then applied these lessons learned to their fatigue predictions during the benchmarking exercise's second phase. e teams and their PDA methods are listed in Table 1 above. Calculating correlations e benchmarking analysis protocol proceeded as follows: Each PDA participant initially submitted blind fatigue predictions for each of the three layups, after which the experimental (physical test) results were provided to each team, and then each participant was permitted to submit "recalibrated" fatigue predictions. For 10 samples of each layup, the fatigue cycling was stopped after a predefined number of cycles to allow measurement of the static residual stiffness and strength properties. Five replicates were tested in tension and five replicates were tested in compression. e goal was to impart a measurable amount of damage into the open hole composite specimens in a reasonable amount of time so that the ability of the PDA codes to predict the right type, amount and location of damage as a function of cycles could be assessed. As a result of screening tests, the maximum stress level in terms of percent of static ultimate strength for the [0°/45°/90°/-45°] 2S layup was 50%, for the [60°/0°/-60°] 3S layup was 80%, and for the [30°/60°/90°/-60°/-30°] 2S layup was 40%. Static ultimate strength was 554 MPa for the [0°/45°/90°/-45°] 2S layup, 543 MPa for the [60°/0°/- 60°] 3S layup, and 409 MPa for the [30°/60°/90°/-60°/ -30°] 2S layup. e residual tensile and compressive properties of the coupon with the [0°/45°/90°/-45°] 2S layup were measured after 300K cycles while the residual properties of the coupons with the other two layups were measured after 200K cycles. Residual tensile stiffness also was