Development of failure models and criteria for laminated composites

复合材料作为结构部件的潜力目前受到我们缺乏理解和预测故障发生和传播的能力的阻碍。这一领域的发展可以带来更快、更经济的设计，以及更轻、更有效的结构。本项目将致力于提高预测由单向层组成的层压复合材料失效的能力，重点是（i）每种失效模式的实验表征，（ii）失效机理的分析解释和（iii）失效传播的数值模拟工具的开发。对每种失效模式的实验研究将特别详细。侧重于定性和定量描述导致每种失效模式及其相互作用的事件序列。这些事件最初发生在微观力学尺度上，然后逐渐增长，导致结构失效。这方面的一个例子是纤维压缩扭结。在失败的复合材料中观察到的扭结带是一系列事件的结果，其中包括基体开裂，纤维弯曲失效，并最终纤维微屈曲，并依赖于纤维错位和基体的非线性行为在剪切。在这些事件中，例如基体开裂，其本身是基体微裂纹的生长和合并的结果。实验研究的目的是在这些过程中产生广泛的，详细的，明确的信息。它将需要设计试验台，并将利用密集的仪器（例如声发射、摄影测量和应变仪）以及光学和扫描电子显微镜，实验结果将成为发展分析模型的基础，这些模型描述导致故障的物理事件的顺序及其相互作用。这些分析模型将形成每种失效模式的物理、机械解释。它们可以被理解为每种失效模式的物理理论，将观察到的事件转化为涉及材料（例如弹性，强度，韧性）和几何（例如纤维直径，典型的纤维错位幅度和分布，典型的基质微裂纹尺寸和分布）特性的数学表达式。这些模型的结果将是一组方程，这将表示为失效准则，直接用于设计。为了精确分析复杂的结构，数值模型必须考虑。为此，将开发一种先进的数值失效模型，包括以前的失效标准，用于商业有限元软件。为了准确地模拟故障传播和避免虚假的网格依赖性，数值模型将基于一个待开发的涂抹裂纹方法，适合于层压复合材料中的各种故障问题。这些都与复合材料可以表现出的多种失效模式以及每种失效模式如何影响材料响应有关。例如，基质开裂将导致断裂平面上的牵引矢量的剪切分量减少到零，以及法向分量（如果为正）;计算模型应该能够准确地再现这一点，以及正确地解释该过程的断裂能。最后，数值模型将验证实验数据获得的这种效果在项目期间，以及在文献中发表。

The potential of composite materials as structural parts is currently hindered by our lack of ability to both understand and predict failure initiation and propagation. Developments in this area can lead to faster and more economical design, as well as lighter and more efficient structures. This project will contribute to the capability of predicting failure of laminated composites consisting of unidirectional plies, by focusing on (i) experimental characterisation of each failure mode, (ii) analytical interpretation of the failure mechanisms and (iii) development of numerical simulation tools for failure propagation.The experimental investigation into each failure mode will be particularly detailed, focusing on the qualitative and quantitative description of the sequence of events leading to each failure mode and their interaction. These events take place initially at a micromechanical scale, and then progressively grow to cause structural failure. An example of this is fibre compressive kinking. The kink bands observed in failed composites are the outcome of a sequence of events which include matrix cracking, fibre-bending failure and eventually fibre micro-buckling, and are dependent on fibre misalignments and matrix nonlinear behaviour in shear. Within these events, matrix cracking for instance, is itself the result of the growth and coalescence of matrix micro-cracks. The experimental investigation aims at producing extensive, detailed, univocal information on these processes. It will require design of test rigs, and will make use of intensive instrumentation (e.g. acoustic emission, photogrammetry and strain gauges) as well as optical and scanning-electron microscopy.The experimental findings will be the basis for the development of analytical models describing the sequence of physical events leading to failure and their interaction. These analytical models will form a physical, mechanist, interpretation of each failure mode. They can be understood as a physical theory for each failure mode, translating the observed events into mathematical expressions involving material (e.g. elastic, strength, toughness) and geometric (e.g. fibre diameter, typical fibre misalignments magnitude and distribution, typical matrix micro-cracks size and distribution) properties. The outcome of these models will be a set of equations, which will be expressed as failure criteria, for direct use in design.For the accurate analysis of complex structures, numerical models have to be considered. For this reason, an advanced numerical failure model including the previous failure criteria will be developed, to be used within commercial finite elements software. In order to accurately model failure propagation and avoid spurious mesh dependency, the numerical model will be based on a to-be-developed smeared-crack methodology appropriate for the variety of issues of failure in laminated composites. These have to do with the multiple failure modes composites can exhibit and how each failure mode affects the material response. For instance, matrix cracking will result in the shear components of the traction vector on the fracture plane being reduced to zero, as well as the normal component if positive; the computational model should be able to reproduce this accurately, as well as correctly accounting for the fracture energy of the process. Finally, the numerical model will be validated against experimental data obtained for this effect during the project as well as published in the literature.