New manufacturing techniques for optimised fibre architectures

Home > Research > Core Projects

New manufacturing techniques for optimised fibre architectures

Host Institutions: The University of Manchester, The University of Nottingham

Start Date: 1st February, 2017

Duration: 3.5 years

Lead Investigators: Andrew Long, Prasad Potluri

Co-Investigators: Mikhail Matveev, Shankhachur Roy, Vivek Koncherry

Aims

This project aims to discover new 3D textile preform architectures, using computational modelling or “virtual testing” to evaluate the utility of different textile designs within an optimisation framework. This framework will not be constrained to architectures that can be produced using existing manufacturing technologies, such as weaving or braiding.
Optimum textile preforms will be realised either by modifying existing textile processes or, where potential benefits justify, by developing entirely new, bespoke manufacturing technologies. This will result in a step change in performance, leading to significant weight reductions and lower cycle times through routine use of automated manufacturing technologies. A CIMComp feasibility study previously demonstrated that, for a specific
application, a weight saving of at least 50% can be achieved by relaxing constraints on binder path and in-plane fibre orientations. Here we will further relax constraints on the fibre architecture, aiming to identify and manufacture a number of classes of improved material forms.

The project aims to discover new forms of 3D fibre reinforcements, enabling composites with higher specific properties to be manufactured compared to conventional 3D reinforcements. These new reinforcements will complement and extend the currently available class of 3D textiles, such as orthogonal weaves or layer-to-layer weaves. A computational framework will evaluate properties of various composites designs, and together with an optimisation algorithm, will select the best solution. The computational framework will implement a building-block approach where new models can be added at any stage to evaluate more reinforcements and resulting composites. Optimisation algorithms used within the framework will enable prediction of the best possible solution or a range of optimal solutions (a Pareto front).

A series of case studies, developed through collaboration with industrial partners, will be used to demonstrate potential weight-savings or performance improvements by preform optimisation.

Figure 1a. Meso-scale model of an orthogonal 3D weave.

Figure 1b. Meso-scale model of a multiaxial 3D preform

Progress

The computational framework has been implemented for the development of new preforming technologies and the project is moving towards implementing this for some industrial case studies.  The framework implements a multi-scale approach for modelling composite structures, which ultimately reduces the entire structure down to a meso-scale representative volume element of the material. It creates parameterised meso-scale unit cells of the fibre reinforcements using TexGen, the University of Nottingham’s textile pre-processor (Figure 1). Mechanical properties of the unit cells are calculated using Abaqus finite element solver. A new meshing algorithm has been implemented to improve the efficiency of this part of the framework. The meshing algorithm combines octree-based local refinement/coarsening of meshes and surface smoothing. The algorithm is embedded within TexGen and will be published as part of the open-source program.

The framework optimises the meso-scale geometry of fibre reinforcements according to the selected criteria, e.g. minimal weight, within selected constraints, e.g. component target stiffness. The design space (possible configuration of yarns/ layers) for the optimisation of the meso-scale geometry is larger than that of conventional 3D textiles. In addition to changing the total number of layers, the spacing between yarns and the binder path, off-axis yarns and changes in fibre orientation are allowed to every layer of the optimise preform. A multi-objective genetic algorithm performs a search over the design space and selects the best solutions according to selected criteria.

One of the selected case studies is the optimisation of fibre reinforcement for a section of a vehicle floor pan, in collaboration with the Advanced Manufacturing Research Centre (AMRC). Two load cases, bending and torsion, are modelled in Abaqus using material properties calculated at the meso-scale. A multi-objective optimisation of the fibre reinforcement predicts multiple optimal solutions, as shown by the Pareto front in Figure 2. This highlights the relationship and trade-offs between the selected optimisation criteria and helps an end-user to select a solution according to the weighting of the requirements.

Novel manufacturing techniques are currently under development in order to validate the optimisation algorithms. A technique to place off-axis yarns for 3D preforms has been developed, initially for a flat preform (Figure 3) and subsequently for a tubular preform. The preforms made with this technique are not limited to having most of the layers in a particular direction like orthogonal 3D woven preforms.

Figure 2. 65 mm thick 3D multiaxial orthogonal preform.

Figure 3. Optimisation of fibre reinforcement for a demonstrator: 1. The geometry of the component; 2. Evaluation of the component’s performance in bending and torsion; 3. Pareto front of optimal solutions.

 

Key Achievements

  • Computational framework for multi-objective optimisation of fibre reinforcements has been developed.
  • The computational framework has been applied to an automotive demonstrator component.
  • A novel meshing technique has been developed, which will be published as part of the Texgen open-source code.
  • A novel multiaxial preforming concept has been demonstrated, which will be realised by developing new textile machinery based on these concepts.
  • A novel preforming technique has been developed to manufacture flat and tubular multi-axial 3D fibre preforms.
Copyright © 2017 CIMComp