Computationally efficient micromechanical model-aided woven fabric composite design based on genetic algorithms

Abstract

The manufacturing cost of woven fabric composites is low and this type of material exhibits high impact resistance and delamination resistance. Plain weave fabric composites are traditionally used on civil, offshore, marine and automobile structures but are increasingly utilised on aerospace structures in recent years. Triaxial weave fabric composites are new materials that are increasingly used in ultralight structures, such as the wings of unmanned aerial vehicles and deployable antenna on spacecraft. The need for better design of woven fabric composites is growing since having lower in-plane mechanical properties than unidirectional composite laminates becomes a problem which limits the use of woven fabric composites. However, the combination of modern manufacturing and optimisation techniques makes the wide use of bespoke high performance weave fabric composite become possible for aerospace applications. The yarn specification, including the undulation length, the width and the thickness of the yarn, significantly influences the mechanical properties and mass of these materials. Therefore, there is a chance to utilise optimisation techniques to search for the optimal design of these materials for a wide range of applications. Optimisation techniques are widely used across many different disciplines. They are often used to provide innovative solutions or to gain insights into complex problems. The literature review highlights that Genetic Algorithms (GAs) are one of the most popular categories of optimisation techniques for solving engineering optimisation problems as they are able to robustly find the entire set of optimal solutions for single-, multi- and manyobjective optimisation problems with large and complex search spaces. Genetic Algorithms are population, evolution and natural selection based optimisation tools inspired by Darwin’s theory of evolution and Mendel’s inheritance theory. The state-of-the-art Genetic Algorithms are developed and benchmarked with mathematical optimisation problems every year to find the best performing solvers for different types of problems. However, since the dominant characteristics of woven fabric composite optimisation problems are different from the mathematical benchmarking problems in the evolutionary computational literature, it is difficult to select a suitable Genetic Algorithm for solving a weave fabric composite optimisation problem. Therefore, state-of-the-art Genetic Algorithms are benchmarked in this research to determine the dominant characteristics of weave fabric composite optimisation problems and which solvers perform best. Using Genetic Algorithms is an effective method for solving weave fabric composite optimisation problems. However, it is obvious that using Genetic Algorithms requires high computational cost due to the large number of objective function evaluations, especially when the number of variables, objectives and constraints are increased. In addition, the computational time of evaluating each objective function significantly increases the total computational cost. Therefore, using computationally efficient methods to predict the mechanical properties of weave fabric composites becomes essential in the optimisation. This not only helps to reduce the computational cost, but also eliminates the limits on the optimisation techniques due to the difficulties of increasing the population size and generation number. The literature review of composite modelling methods highlights analytical methods as the most efficient methods when hundreds of thousands of different sets of weave patterns and yarn parameters are simulated to determine the optimal designs of weave fabric composites. However, the current available analytical methods are either imprecise or not robust when predicting the mechanical properties of woven fabric composites with different yarn specifications and material types. Therefore, two micromechanical models are developed in this work to achieve the computationally efficient predictions of the mechanical properties of weave fabric composites. This research aims to develop a methodology to better design woven fabric composite materials with the aid of a computationally efficient micromechanics-based analytical model. This research aim is achieved by conducting a state-of-the-art literature review of composite structures/materials optimisation and genetic algorithms to determine the optimisation process used in the current research; comparing the advantages and disadvantages of each woven fabric composites modelling method to find out the most appropriate method to be used as the objective evaluation method in the optimisation; developing and validating novel analytical models to perform computationally efficient prediction of the mechanical properties of weave fabric composites; benchmarking state-of-the-art Genetic Algorithms in the optimisation of 2D weave fabric composites to find out the optimal designs and determine the dominant characteristics of the optimisation problems for selecting the best practices of Genetic Algorithms.

More information

Identifiers

ORCID for Zhenzhou Wang: orcid.org/0000-0003-3926-070X

ORCID for Adam Sobey: orcid.org/0000-0001-6880-8338

Catalogue record

Export record