Multi-scale modelling of woven composites accounting for manufacturing deformations

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


An automotive structure must perform a number of functions including all of those required for crash, NVH (noise, vibration, and harshness) and durability attributes. It is in the industrial interest that these attributes are assessed using virtual methods such as Finite Element models.

Due to their excellent mechanical and weight-efficient properties, carbon fibre composites are gaining popularity for structural applications. One of the most commonly used formats is textile reinforcement, due to its low manufacturing cost and ability to be formed into complex shapes. These composites are highly anisotropic, which means the material properties are dependent on the structural geometry of the fabric. Consequently, the properties might vary between components of different designs. This could be due to the carbon fibre alignment directions, the shape of the carbon fibre tows and the change in carbon fibre to resin ratio. All of this will have an effect on stiffness and strength locally in different directions. To optimise the use of this material, its properties must be understood. The traditional method, where physical experiments are conducted to characterise the material behaviour, is time-consuming, expensive and has poor repeatability as manufacture procedures may vary, and so the experimental work will need to be repeated if there are any changes to the material configuration.

Thus it is crucial to develop an innovative workflow, where these complex factors and changing variables can be accounted for using Finite Element simulations. This consists of a multi-scale modelling approach. Where the material properties are extracted from a meso-scale model, to be used in macro-scale component level models. The effect of shear and unit cell rotation during forming are also taken into account throughout the meso-macro workflow.

A forming simulation is first performed. A range of realistic textile unit cell geometries is then generated using an in-house digital element method for a discrete selection of shear angles, informed by the forming simulation. This realistic unit cell geometry is used to create a mechanical voxel mesh model, which is then subjected to displacement-controlled periodic boundary conditions. The homogenised material properties are obtained and mapped onto a macro-scale component shell model, where its material properties are varied based on the shear angle and unit cell orientation. Lastly, loading conditions are applied to obtain the final part performance.

An extensive experimental testing campaign has been carried out in support of the modelling work and is divided into three parts: To validate the realistic unit cell geometry from the digital element method using CT scans, to validate the homogenised material property predictions from meso-scale unit cell models using tensile coupon tests with DIC, and to generate numerical inputs for the forming simulation with shear stiffness and bending stiffness experiments. This PhD primarily focuses on a 2D twill woven carbon fibre prepreg material with a 3-minute cure resin system. Composite plates with various shear angles were manufactured for the tensile experiments. Both the prepreg and its dry fabric counterpart were characterised to compare the differences. While the focus is on a single ply fabric under room temperature, the effect of multiple ply lay-up and forming temperature is also explored.
Date of Award22 Mar 2022
Original languageEnglish
Awarding Institution
  • University of Bristol
SponsorsJaguar Land Rover
SupervisorStephen R Hallett (Supervisor), Bassam S F Elsaied (Supervisor) & Dmitry Ivanov (Supervisor)

Cite this