Stress gradient failure theory for textile structural composites

Micromechanical methods for stiffness and strength prediction are presented, the results of which have led to an effective failure theory for prediction of strength. Methods to account for analysis of multi-layer textile composites are also developed. This allows simulation of a single representative volume element (RVE) to be applicable to a layup of an arbitrary number of layers, eliminating the need for further material characterization. Thus a practical tool for failure analysis and design of a plain weave textile composite has been developed. These methods are then readily adaptable to any textile microarchitecture of interest.; A micromechanical analysis of the RVE of a plain-weave textile composite has been performed using the finite element method. Stress gradient effects are investigated, and it is assumed that the stress state is not uniform across the RVE. This is unlike most models, which start with the premise that an RVE is subjected to a uniform stress or strain. For textile geometries, non-uniform stress considerations are important, as the size of a textile RVE will typically be several orders of magnitude larger than that of a unidirectional RVE. The stress state is defined in terms of the well-known laminate theory force and moment resultants [N] and [M]. Structural stiffness coefficients analogous to the [A], [B], [D] matrices are defined, and these are computed directly using the Direct Micromechanics Method (DMM), rather than making estimations based upon homogenized properties.; Based upon these results, a robust 27-term quadratic failure criterion has been developed to predict failure under general loading conditions. For multi-layer analysis, the methods are adapted via three techniques: direct simulation of a multi-layer composite, an adjustment of the data output from single-layer FEM simulation, and an adjustment of the quadratic failure theory (without the requirement of determining a new set of failure coefficients). The adjusted single-layer data analysis and the adjusted quadratic failure criterion show 5.2% and 5.5% error over a variety of test cases.; The entire body of work is then applied to several practical examples of strength prediction to illustrate their implementation. In many cases, comparisons to conventional methods show marked improvements.