Periodic Reporting for period 3 - PRE-ECO (A new paradigm to re-engineering printed composites)
Reporting period: 2023-02-01 to 2024-07-31
Additive manufacturing and Automated Fibre Placement (AFP) processes brought to the emergence of a new class of fibre-reinforced materials; namely, the Variable Angle Tow (VAT) composites. AFP machines allow the fibres to be relaxed along curvilinear paths within the lamina, thus implying a point-wise variation of the material properties. In theory, the designer can conceive VAT structures with unexplored capabilities and tailor materials with optimized stiffness-to-weight ratios. In practise, steering brittle fibres, generally made of glass or carbon, is not trivial. Printing must be performed at the right combination of temperature, velocity, curvature radii and pressure to preserve the integrity of fibres. The lack of information on how the effect of these parameters propagates through the scales, from fibres to the final structure, represents the missing piece in the puzzle of VAT composites, which today are either costly or difficult to design because affected by unpredictable failure mechanisms and unwanted defects (gaps, overlaps, and fibre kinking).
This project is for an exploratory study into a radical new approach to the problem of design, manufacturing and analysis of tow-steered printed composite materials. The program will act as a pre-echo, a precursor, to: 1) implement global/local models for the simulation and analysis of VATs with unprecedented accuracy from fibre-matrix to component scales; 2) develop a (hybrid) metamodeling platform based on machine learning for defect sensitivity and optimization; and 3) set new rules and best-practices to design for manufacturing. A 5-year, highly inter-disciplinary programme is planned, encompassing structural mechanics, numerical calculus, 3D printing and AFP, measurements and testing of advanced composites, data science and artificial intelligence, and constrained optimization problems to finally fill the gap between the design and the digital manufacturing chain of advanced printed materials.
This project is for an exploratory study into a radical new approach to the problem of design, manufacturing and analysis of tow-steered printed composite materials. The program will act as a pre-echo, a precursor, to: 1) implement global/local models for the simulation and analysis of VATs with unprecedented accuracy from fibre-matrix to component scales; 2) develop a (hybrid) metamodeling platform based on machine learning for defect sensitivity and optimization; and 3) set new rules and best-practices to design for manufacturing. A 5-year, highly inter-disciplinary programme is planned, encompassing structural mechanics, numerical calculus, 3D printing and AFP, measurements and testing of advanced composites, data science and artificial intelligence, and constrained optimization problems to finally fill the gap between the design and the digital manufacturing chain of advanced printed materials.
Advanced multi-scale models for the static and dynamic, linear and nonlinear mechanical responses and multi-physics characterization of variable stiffness composite (VSC) beam/plate/shells structures have been implemented. These models have been utilized for the deterministic and non-deterministic manufacture-induced defects analysis of VSC parts. In detail, Continuous Tow Shearing (CTS)-induced thickness variation and the Automated Fiber Placement (AFP) manufacturing signature, viz. gaps and overlaps, have been studied. Furthermore, a stochastic-field-based method has been coupled with the in-house high-order finite element code to study non-deterministic manufacture-induced defects of VSC. Such stochastic fields have been employed alongside process simulations to model uncertainty defects at different scale levels, namely the micro- and meso-scale. In this regard, micro-scale models for multi-field applications (thermos-hygro-elasticity, thermal conductivity, moisture diffusivity, piezoelectricity) have been established.
A parallelized in-house hybrid optimization code has been developed, utilizing Genetic Algorithms (GA) and gradient-based methods to solve continuous and discrete problems efficiently. This tool enables using direct and/or surrogate simulations for enhanced computational efficiency. The code has been applied to multi-objective optimisation problems, targeting mass, strain energy, fundamental frequency, and buckling load as objectives. These optimizations integrate the manufacturing signature of the AFP process, explicitly accounting for gaps and overlap through machine simulations. Additionally, this approach is designed for potential applications in real-world structural scenarios, such as wing-box structural optimization, where design rules and various laminated composite patches are incorporated to further reduce structural weight, offering hope for practical and efficient solutions in structural engineering.
An advanced higher-order beam model employing the Node-Dependent Kinematics approach has been developed for the progressive damage analysis of composite structures. This scalable and efficient model reduces computational costs while accurately capturing damage propagation in critical zones.
Other activities include investigating the possibility of reducing the strain/stress concentrations in an open-hole plate using localized 3D printed carbon fiber reinforcements; using non-local elasticity models to conduct crack propagation analysis; solving inverse problems (e.g. detection of damage or printing issues) via algorithms based on artificial intelligence; cooperation with University of Tokyo on space antennas subject to thermal loading and investigate the effect of geometrical parameters, extending the developed methods to thin-ply composites for deployable structures. Moreover, the latest applications also include the development of high-order models for the accurate three-dimensional stress analysis at large strains of hyperelastic anisotropic materials, allowing the study of innovative composite configurations with isotropic silicone matrix of complex composite with immersed and distributed/dispersed fibers, with refined fully nonlinear models.
A parallelized in-house hybrid optimization code has been developed, utilizing Genetic Algorithms (GA) and gradient-based methods to solve continuous and discrete problems efficiently. This tool enables using direct and/or surrogate simulations for enhanced computational efficiency. The code has been applied to multi-objective optimisation problems, targeting mass, strain energy, fundamental frequency, and buckling load as objectives. These optimizations integrate the manufacturing signature of the AFP process, explicitly accounting for gaps and overlap through machine simulations. Additionally, this approach is designed for potential applications in real-world structural scenarios, such as wing-box structural optimization, where design rules and various laminated composite patches are incorporated to further reduce structural weight, offering hope for practical and efficient solutions in structural engineering.
An advanced higher-order beam model employing the Node-Dependent Kinematics approach has been developed for the progressive damage analysis of composite structures. This scalable and efficient model reduces computational costs while accurately capturing damage propagation in critical zones.
Other activities include investigating the possibility of reducing the strain/stress concentrations in an open-hole plate using localized 3D printed carbon fiber reinforcements; using non-local elasticity models to conduct crack propagation analysis; solving inverse problems (e.g. detection of damage or printing issues) via algorithms based on artificial intelligence; cooperation with University of Tokyo on space antennas subject to thermal loading and investigate the effect of geometrical parameters, extending the developed methods to thin-ply composites for deployable structures. Moreover, the latest applications also include the development of high-order models for the accurate three-dimensional stress analysis at large strains of hyperelastic anisotropic materials, allowing the study of innovative composite configurations with isotropic silicone matrix of complex composite with immersed and distributed/dispersed fibers, with refined fully nonlinear models.
Our vision beyond the state of the art is to develop a robust, flexible and hybrid simulation technology for the analysis and design of variable angle tow composite structures, consisting of (i) a high-fidelity, multi-scale model for micro-, meso- and macro-mechanics characterization, (ii) completed by existing and innovative optimization algorithms which take into account current manufacturing limitations, and (iii) embedded into a data intelligence, hybrid platform encompassing virtual experiments and tests.