Session: 01-06-02: Impact, Fatigue, Damage and Fracture of Composite Structures

Paper Number: 107576

107576 - Determination of Composite Design Allowables Based on Mechanics of Structure Genome and Multiscale Simulations 

Composite materials have been increasingly used in many industries, including aerospace, automotive, and wind energy. However, composites are anisotropic      , heterogeneous, and associated with numerous uncertainties during manufacturing. These features pose a great challenge in the design and certification process of composite materials and structures. The current approach requires mechanical characterization at the structural component level and determination of the design allowables, as defined in the Composite Material Handbook (CMH-17) [1]. The design allowable is a statistical measurement of the material strength, which can be used in the design of composite structures to assure safety. The allowables are usually represented as A and B basis. The A basis is defined as a strength at which 99% materials will not fail with a 95% confidence. Similarly, the B basis is defined as a strength at which 90% materials will not fail with a 95% confidence. Notably, the statistical characterization of the allowables requires a tremendous amount of experimental tests. Different loading conditions and structures (notched, un-notched, and more complex structures) must be determined separately according to the methods specified in CMH-17. Therefore, various simulation-based approaches are proposed to reduce the number of tests needed. Progressive failure models are used in FEA analysis to predict the strength, which is a statistical value with uncertainties from constituent material stiffness, strength, the volume of fraction of fiber, etc. Due to the multiscale nature of composite materials, a multiscale approach could help further reduce the simulation costs in predicting allowables.

 

Mechanics of structure genome (MSG) [2] is a unified approach for homogenization and dehomogenizaiton of composite materials and structures, based on the concept of structure gene (SG). An SG is defined as the smallest mathematical building block of a structure. For instance, a 1D SG is enough for a composite laminate, while a 3D SG is needed for a woven composite. MSG is capable of calculating the global stiffness and obtaining the local stress field based on the global structural responses. MSG also enables us to treat different macroscopic models (beam, plate, and solids) and microscopic models (woven, laminate, honeycomb, etc.) in a unified way. MSG-based approaches have the potential to be applied in a wider scope without many modifications.

 

In this study, an MSG-based multiscale approach is proposed for predicting allowables of composite materials and structures. Fiber and matrix properties are calibrated from lamina-level experiments. Variability of the constituent material property and volume of fractions are sampled as required by CMH-17. In each virtual test, a progressive failure model is employed to simulate the damage of a test coupon. Statistical methods are then used to obtain the allowables based on the simulation results. The key part of this workflow is the multiscale simulation with a progressive failure model. Taking a coupon made of composite laminate as an example, the steps of the simulation are described as follows:

1)    Obtain the plate property from MSG theory and the constituent material property and distribution.

2)    In FEM analysis, load the test coupon incrementally. The structural response is then fed back to MSG to obtain the local stress field.

3)    A progressive failure model is implemented to determine the damage variable of the constituent materials. The materials are then degraded by the damage variables. Go back to Step 1) and iterate.

 

The main contributions and advantages of this work include:

1)    Enabled MSG with progressive failure analysis capability.

2)    Greatly reduced computation time by using a multiscale simulation framework.

3)    Established a unified framework for virtual allowable prediction of different types of composite materials.

Reference:

[1] Amsc, N., & CMPS, A. A. (2002). Composite materials handbook. Polymer matrix composites materials usage, design, and analysis.

[2] Yu, W. (2016). A unified theory for constitutive modeling of composites. Journal of Mechanics of Materials and Structures, 11(4), 379-411

Presenting Author: Haodong Du Purdue University

Presenting Author Biography: Haodong Du is a PhD student in Prof Wenbin Yu's group at Purdue University. His principal interests include damage and fatigue of composite materials, multiscale simulations, and applications of machines learning in composite simulations.

Authors:

Haodong Du Purdue University
Wenbin Yu Purdue University