Search

This work addresses the characterisation of the micro-structure of fibre bundles in reinforcement textiles for composites and its influence on resin flow in liquid composite moulding (LCM) processes. Random variations in local filament spacing result in a non-uniform flow velocity field. Merging flow fronts due to differences in local flow velocities can lead to gas entrapment in the matrix phase. This results in formation of micro-scale defects, which can significantly reduce the matrix-dominated mechanical properties of the finished composite. An automated image analysis method was developed to precisely determine local filament distributions within complete fibre bundles. Based on two-dimensional micrographs, filament arrangements were characterised statistically by means of nearest neighbour distance and angle distributions. It was observed that the micro-structure becomes more uniform with increasing level of compaction. A micro-structure generator to reconstruct filament arrangements was adapted incorporating these measurement data. Transverse permeabilities derived from numerical simulations of steady-state flow on automatically discretised model domains were found to be log-normally distributed. With increasing model size, average values and widths of the distributions decrease, converging to the permeability of a complete fibre bundle. Similarly, average values and scatter decrease with increasing fibre volume fraction. The transverse permeability of random filament arrangements was found to be significantly smaller than for uniform filament arrangements. The void content in composite specimens produced by resin injection along and perpendicular to the fibre bundles was characterised with a developed image analysis process. Due to the more uniform micro-structure at increased bundle compaction, a decrease in void content was observed in the case of transverse resin injection. A first step towards void content prediction in a fibre bundle by numerical simulation of transient flow through a randomised filament arrangement was made. It was hypothesised that steady-state flow results may be used for the prediction of transient flow.

This study focuses on the development of a flow monitoring system that can be integrated with the Liquid Composite Moulding (LCM) process to measure the resin frontal flow propagation. Today, finite element simulations are regularly used to design injection and cure kinetics in the LCM processes. However, purely predictive simulation suffers from issues related to uncertainty and variability in material state and numerous process variables. Thus, the online monitoring of the process is suggested to provide an estimation of the component's state and to predict potential defects, such as voids and dry spots. The research methodology of the developed system comprises the design of a flexible dielectric sensor and development of a measuring system for the liquid resin infusion process with simplified and complex composite parts. The analysis of the prototype dielectric sensor was performed with the development of an electrostatic model that simulates the behaviour of the sensor during the process and provides information on the sensor capacitance and sensitivity.

Utilisation of the Vacuum Assisted Resin Infusion (VARI) process has increased in recent years for the manufacture of large scale components for the transport industry. Quality control and repeatability are two key areas where the VARI process requires greater understanding due to the scale of the infusions. The marine industry has a history of being at the cutting edge of large scale composite manufacturing. This research focusses on utilising the processes and knowledge at Princess Yachts International (PYI) to monitor and analyse the process variability and uncertainty for large scale VARIs. A detailed process review was conducted at PYI to evaluate the level of process monitoring and variability as well as the defect controls that are used in industry. Core samples were extracted from infusions where surface visible defects had been recorded. The core samples were analysed using Micro Computed Tomography (µ-CT) to generate a three-dimensional reconstruction of the defects. The location and structure of the defects indicated matrix failure towards the mould surface of the component, an area which is not monitored during the VARI process. Having identified a requirement to monitor the manufacturing process, a review of available sensors was conducted identifying Fibre Bragg Grating (FBG) fibre optics as the most suitable monitoring method. The sensors measured the residual stress development during resin curing as well as the high loads exerted on the component during demoulding which were five times larger than during the infusion and resin curing process. The measured strains were lower than the failure strain of the resin suggesting the presence of defects that result in premature matrix failure. A novel approach to monitoring resin flow was then developed to analyse the through thickness resin flow progress using µ-CT. Both monolithic and sandwich laminates were analysed both in-situ and exsitu to evaluate the effect of variable laminate layup. The results further identified the high degree of variability during the infusion caused by changes in laminate schedule and the transition from sandwich to monolithic sections.

Automotive manufacturers are responding to increasingly stringent demands on fuel consumption by developing novel lightweight designs with new materials. Carbon fibre composites are seen to offer high specific properties – leading to low weight solutions, but their cost and manufacturing rate are currently unsuited to automotive high volume cost targets. This thesis presents a novel composites manufacturing process offering high production rates and low process costs. The process is termed ALCM, Advanced Liquid Compression Moulding and is a novel derivative of High Pressure Resin Transfer Moulding. In the two-stage ALCM process, features of both liquid moulding and compression moulding techniques are exploited. In the first stage the mould tool is held open by a small amount during the injection phase. This allows the fibre preform to expand into the mould cavity, increasing its permeability, thus reducing overall injection pressure requirements. In the second stage the tool is closed to finalise the resin infusion and form the finished component to the required geometry and fibre volume fraction. Two work packages are presented in this thesis: Objective 1 is to study a range of experimental parameters to characterise the ALCM process; Objective 2 is to simulate the ALCM manufacturing process using the Finite Element method. In Objective 1 a range of experiments explore the feasibility of the ACLM process to manufacture large structural components using carbon fibre NCF (non-crimp fabrics) with fast-curing epoxy resin. The experiments were performed on the newly-installed high pressure RTM injection equipment and large bed press at the National Composites Centre, UK. Characterisation of the ALCM process was undertaken using two mould tool geometries – a 1240x790mm rectangular flat plate tool and a top hat shape tool with the same planform area and 180mm depth. Generally, high quality mouldings were produced, with void content seen to be typically lower than 0.5%, and injection flow rate across the experimental range did not affect the void content. Flat plaque experiments were performed at resin injection flow rates between 35 and 85g/s and three different levels of tool opening during the first resin injection stage, leading to in-mould pressures of 100bar. Results from mouldings filled with between 18% and 100% of total resin volume demonstrated that, in the liquid moulding stage of ALCM whilst the mould tool is held open, the resin flow in the preform is in-plane. The in-plane flow through the preform follows Darcy’s law for flow through porous media, however it was seen that spaces around the preform inside the rigid mould tool were conduits for resin flow governed by Navier-Stokes flow. It was found that in the resin injection stage, after a minimum tool opening, the resin channels in the mould alleviate resin injection pressure to a greater affect than increasing permeability (through lower fibre volume fraction). In the flat panel mouldings, resin channels around the component perimeter determine the fill pattern. In the top hat shape mouldings, resin channels were present at corner bend radii and were seen to be advancing the flow front. Resin channels were found to be problematic through reducing the proportion of injected resin volume that remained in the preform. Injected resin volume determined by feedback from an in-mould pressure sensor, rather than prescribed volume of resin, demonstrated the process consistency across sixteen repeated flat panels. Preform/fibre wash was found to be a critical problem as a result of the high injection pressures. The effect varied from local wrinkle defects to complete wash out and was shown to be linked to the degree of preform stabilisation by binder or edge clamping and fibre tow wash to the fabric stitch and inlet configuration. Critically the extent of fibre wash was sensitive to small variations in preform geometry at the higher injection flow rates experimented. In Objective 2, the simulation methodology developed using the well-known University of Delaware, LIMS Liquid Injection Moulding Simulation software. The modelling methodology was successfully adapted to the ALCM process, employing a low FVF injection using a Finite Element Control Volume approach, followed by compression as each node becomes a resin injection gate. The simulation was used to predict fill pattern, mould pressure and press clamping force. Results from simulation models representing the flat and top hat shape panels correlated to experiments and found that the Darcy flow assumption is valid for predicting fill pattern provided that the preform remains fixed and resin channels are represented. Correct representation of changing resin viscosity is necessary to reliably calculate pressure and therefore press clamping force. A full-size vehicle floor was the technical demonstrator component. This large component, with surface area 3m2 was ultimately manufactured within a 5 minutes cure cycle. At the time of writing, this is believed to be the largest structural composite component manufactured at this production rate. It was found that the methodology developed in Objective 2 could be applied to give good correlation between simulation and experimental results for resin fill pattern and process pressure, with the additional requirement that resin outflow from the preform into the tool be characterised. The floor component is produced from three preforms. Fibre wash at the preform joint was observed at resin injection flow rates of 85g/s and above. An innovative design of Anchored Through-Thickness Reinforcement at the preform joint was incorporated and was seen to stabilise the preform joint, eliminating fibre wash The two part resin used in the process is a new formulation developed to have short cure time. At moulding temperatures between 1300C and 1100C the resin cure time was between 3 and 8 minutes. It was found that conventional rheometry characterisation techniques were not satisfactory for defining the resin viscosity in the initial reaction stages. The dielectric analysis technique was investigated for the first time in this application. Results found that this method is suitable for fast cure resin rheo-kinetics characterisation.

3D woven structures in composite applications are getting increasing attention due to their superior out-of-plane properties and ability to form near-net-shape preforms. However, the in-plane properties of 3D woven composites are often compromised due to manufacturing induced limitations. The through-thickness Z-binder holds the load-bearing warp and weft tows and gives the structural integrity to the 3D woven preforms. This research aims to investigate the role of Z-binder weave architecture and tension on the 3D woven preform and composite mechanical properties. 3D woven structures were developed with four different Z-binder weave architectures: 1x1 plain, 2x1 twill and 2x2 twill with orthogonal binding patterns and an angle interlock (AI) weave. The degree of binder interlacings was deliberately reduced to increase binder float length and reduce the binder waviness. Binders were incorporated in both warp and weft tow directions to alter the directions of the resin channels inside the structures. Two orthogonal structures were manufactured with different binder tensions. Dry preform compressibility and in-plane permeability were measured to explore the effect of binder in preform compaction and resin flow behaviour. Tensile and bending properties were characterised to investigate the role of Z-binder in composite mechanical properties and their failure mechanisms. Through-thickness binders were found to resist the transverse compaction of the dry preforms. Compressibility was improved by reducing the degree of binder interlacings. Preform permeability was primarily dominated by the resin channels which were created by the Z-binder interlacings. By reducing the frequency of binder interlacings, Z binder crimp % was reduced. This reduction also increased the fibre volume fractions of the composites. Both tensile and bending properties of the composites were primarily dependent on the degree of binder weave interlacings and fibre volume fractions. Angle interlock weave with minimum crimp % in Z-binder and least number of interlacings resulted in the highest modulus and strength in both tension and bending loads. Composite failure mechanisms were found highly sensitive to the loading directions. Damage modes were significantly distinctive in the binder way and its transverse directions. In both tensile and bending cases, the Z-binders were found the arrest the damage progression and delay the final failure of the composites.

Conventional fibre polymer composite materials consist of a single matrix system for a whole part and are selected on the dominant requirements of the part. A multi-matrix composite would provide an expanded design space for a composite part. However, there are few processes available to realise such a composite and these rely on modified forms of resin transfer moulding. These processes require active venting using an expensive vacuum system and the quality is reliant on bag quality. This research is devoted to developing a foundational principle for a processing strategy that uses the localised introduction of resin combined with an out-of-bag consolidation process containing a preheating stage that can form a multi-matrix composite. The resin flow in local deposition and consolidation processes is determined by a complex interplay between the resin and preform. Hence the chosen resin and fabric were characterised extensively. A range of viscosity-tailored consolidation schemes was implemented using a setup to estimate the in-line viscosity using the rheo-kinetics of the fast-curing resin system. By initiating the consolidation pressure at a higher resin viscosity, the voids were found to be suppressed. A simplified model to capture the dominant mechanisms during consolidation was developed. The dominant flow transitioned from a 2D yarn-scale to macro-scale flow on increasing the activation viscosity. The increased resin pressure associated with high resin viscosity during consolidation was found to be a dominating factor in reducing the voids. The model was then used to develop a simple analytical design tool for tailoring the consolidation parameters to minimise the void content. A feasibility study was conducted on combining different matrices to improve the mechanical performance of composites in an open hole tension test. The use of high stiffness carbon nanotube modified resin to locally manipulate the region around the hole increased the strength of the composite. The enhancement of multi-matrix composite was then qualitatively explained using a simple finite element analysis combined with classical laminate theory.

Any composite manufacturing method requires an application of a carefully designed consolidation process to ensure the suppression of voids in the laminate, establish bonding in laminate layers and prevent dimensional or fibre-path defects. The optimisation of consolidation processes relies on the characterisation of the composite precursors' deformability. There are multiple mechanisms occurring in consolidation and various experimental programmes have been suggested in the literature to describe these mechanisms and deduce relevant material properties. The selection of a testing methodology often relies on an initial hypothesis or prior knowledge regarding the deformation modes. This may be a source of significant errors. This research poses questions on the testing rationales, on subjectivity in material testing and on how data-rich programmes should be designed. The proposed solution to the consolidation characterisation problem is a real-time adaptive testing strategy that enables a "conversation with the material" - flexible autonomous steering of a testing programme reacting on the obtained output. Such system is implemented within this research and is called adaptive consolidation sensor framework. The system's algorithm is set to analyse composite precursor's feedback to compression in real time and to track characteristic signatures relevant to existing consolidation models. Based on that feedback the framework formulates a load programme for the test in a reactive manner. The developed testing system focuses on the identification of the underlying physical mechanisms rather than material properties identification in a rightly or wrongly assumed resin flow mode. To validate the proposed testing approach, a series of characterisation experiments for different material systems was performed. The obtained results highlight a way forward in terms of rethinking experiments for materials used in manufacturing and beyond.

The use of epoxy powder as the primary matrix in thick fibre-reinforced composite parts is investigated. The characteristics of three epoxy powders are assessed using several experimental techniques, focusing on their curing behaviour. At least one epoxy powder is shown to have advantageous characteristics for manufacturing thick-section composites. Material models are developed which can describe the processing behaviour (cure kinetics, viscosity change, etc.) of an epoxy powder. The cure kinetics model makes use of an additional rate constant to better describe the rate of cure at both high and low temperatures. The chemorheological model is based on an existing model for toughened epoxies. A one-dimensional simulation tool for manufacturing thick-section composite laminates is developed in MATLAB. The simulation tool employs a resin flow model for vacuum-bag-only prepregs to describe the infusion process and subsequent thickness change. This thickness change is coupled to a model for through-thickness heat transfer which can be solved numerically for various thermal boundary conditions. The model is used to explore the suitability of epoxy powders for the manufacturing thick-section composite structures. The aforementioned simulation tool is validated against experimental results for thick-section composite laminates. The experiments are carried out using a modified heated tool and test apparatus which apply known thermal boundary conditions. A linear variable differential transformer is used to measure the thickness change of each laminate during testing, while thermocouples are used to measure the temperatures at various positions within each laminate. The results of the tests show good agreement with the one-dimensional simulation tool. Additional simulations are performed to investigate the influence of material format, thickness change, and heating methods. Methods for reducing thermal and cure gradients are explored also. A method is outlined for implementing the process models within commercial finite element software, Abaqus FEA. User subroutines for heat transfer and thermal expansion are used to define the various process models. One-dimensional simulations are validated, and a convergence study is performed on time step size and element size. Simulations show the effect of in-plane heating for glass-fibre and carbon-fibre laminates, and the processing of a wind turbine blade root section is investigated. Overall, it is shown that thick-section composite structures can be manufactured using a low-cost commodity epoxy powder from the coating industry, and that these structures do not suffer from the risk of uncontrolable thermal events.

Many large composite structures are manufactured using sandwich laminates to achieve high specific bending strength and stiffness. Examples include wind turbine blades, where self-weight becomes increasingly important as blade size increases. Resin infusion of three-dimensional sandwich laminates can result in complex resin flow paths, and subsequent defect formation, which are difficult to predict. The core material used for sandwich construction and its interaction with liquid resins may also influence the formation of defects, and in the case of balsa this effect can be used to reduce defect severity. In order to evaluate the effect of cored sandwich laminate construction on the formation of defects, this thesis concentrates on the characterisation of commonly used core materials and their interaction with liquid resin under high vacuum conditions. It also considers two numerical flow-modelling packages which are shown to be effective at the prediction of flow front convergence for monolithic laminate, but over-estimate defect severity when modelling air- permeable cored laminates. For balsa core, experiments indicate that the available pore space can act as sink for trapped air, which can aid the reduction of defects where multiple flow fronts converge due to the complexity of flow in sandwich laminates. Empirical data for air absorption and desorption rates in balsa core were obtained using a custom-designed experiment. Using these data a theoretical model was developed that can indicate available pore space, which can inform optimum processing conditions, such as time under vacuum. The diffusion coefficients obtained for air absorption and desorption in balsa are very similar, and lie in the middle of published ranges for hard woods at around 2 x 10 -7 m2/s. The methodology developed for this research project represents actual behaviour of air absorption/desorption during resin infusion, whilst other techniques do not, merely measuring diffusion of air through a sample not allowing for finite pore space. In consequence, infusion strategies can be planned more precisely because core/resin interaction is better understood. Knit line defect formation could be predicted with greater accuracy with suitably modified flow-modelling programs.

Manufacturing defects are one of the main reasons for premature failure of wind turbine blades. This thesis investigates the effects of manufacturing defects on the fatigue performance of composite wind turbine blades. A strategic defect risk management protocol for wind turbine blade manufacture is proposed, in which the classification of defect characteristics, evaluation of the effects of defects, their criticality and implementation of manufacturing quality control strategies form the key phases. Based on published data, ply waviness, voids, delamination, adhesive failure and thickness difference are identified as the key manufacturing defects in composite wind turbine blades. These defects are attributed to poor quality raw materials and inadequate production process parameter control. Wet-layup, Prepreg, SPRINT-Prepreg and Vacuum Assisted Resin Transfer Moulding (VARTM) processes were considered for the analysis. Three small scale wind turbine blades were manufactured and a case study was conducted to determine the influence of VARTM process parameter variations to initiate manufacturing defects. It was identified that the resin flow of the VARTM process should be closely controlled to avoid defects.

Melding is a novel process which offers a promising route to creating seamless bonds, by partially curing two laminates in a controlled manner using a Quickstep chamber and subsequently co-curing them. Previous research has focused on melding lap joints in the x-y plane of a composite, whereas this study investigates through-thickness melding, or melding in the z-plane of a composite. In this process, two composite stacks were exposed to heat from one side and actively cooled on the other through the z-axis. The two semi-cured parts were then co-cured creating a monolithic part with a seamless bond.The initial stage of the project developed the semi-curing process. After unsuccessful attempts to produce a semi-cured part in a general purpose Quickstep chamber, due to excessive heat transfer, the process was moved to a hot press with independently controlled platens. The hot press succeeded because the platens were separated from each other by the composite plate, unlike the Quickstep bladders which, as they are designed to conform to the part, came into contact allowing for heat transfer. Thermocouples were embedded every 15 plies to quantify the temperature profiles generated through the laminate stack.The next stage of the project developed a process of joining the semi-cured panels to form a through-thickness melded part. The final process involved constraining the sides of the panel with cork edge dams and inserting woven glass fabric at the corners to allow for gasses to escape. However, the outer parts of the fully melded panel exhibited excessive porosity which had an adverse effect on mechanical properties. For example, whereas tensile and flexural moduli measured for material from the edges of the panels were comparable to values reported in literature, the properties of samples from the middle of the panels deteriorated significantly due to the porosity. Mode I interlaminar fracture energy was approximately 10% lower than values measured for panels fabricated in an autoclave.The entire curing process, from semi-curing to a fully melded panel, was characterized extensively. Differential scanning calorimetry was used to determine the degree of cure and values of glass transition temperature (Tg). The degree of cure of the material exposed to the hot side was approximately 50%, the middle 25%, whereas the cold side was only 15% cured. A corresponding Tg profile through the curing process was developed in which the Tg varied from 0 degrees C for the uncured resin to 245 degrees C in highly cured samples. After melding the sample, the degree of cure was found to be in excess of 99%. Rheological studies were carried out to determine the effects of the semi-curing process on resin flow during the melding cycle.Results showed that there was a large transition zone between uncured plies and solid (cured) plies.This thesis demonstrated the broad feasibility of through-thickness melding as a process to create thick composite laminates. However, the complexity of the process gives rise to thermal and rheological phenomena which affect the structural and chemical properties of the fully melded part. The process must therefore be engineered with these factors in mind in order to create a high quality part.

The process modelling of autoclave composites has received much attention over the years. This thesis concentrates on two types of processes namely the prepreg processing method and the resin infusion processing method. The work focuses on the modelling and simulation of the resin flow, heat transfer and cure processes of the composites during processing. The Hercules 3501-6/AS4 composite was chosen for the simulation and the data for its thermal properties was obtained from Loos and Springer [12]. The composite is considered as a multilayered system consisting of prepregs or dry fibre layers with alternate layers of resin. A similarity analysis for the prepreg process was carried out allowing the velocity field, in both the prepreg and the resin, to be analytically determined. This then permitted the temperature and the degree of cure to be computed numerically. A similar, but different analysis was then carried out for the resin film infusion process, allowing the temperature and rate of cure to be computed directly. The simulation results of the prepreg case of Hercules 3501-6/AS4 were compared to known experimental results and good agreement has been found. Experimental work was performed on the flow dynamics of the resin infusion case and good agreement has also been observed.

The compression of typical glass fibre woven fabrics, namely plain, twill, satin, and noncrimped stitch-bonded fabric, was studied experimentally by performing a series of mechanical tests on dry and resin impregnated assemblies of fabrics. This was followed by microstructural studies of corresponding laminates cured under different degrees of compression. The experiments included investigations on the effects of applied pressure, speed of compression, fabric orientation, number of plies and different resin systems on fibre volume fraction,Vf, of the compressed assembly. It was found that the compression of dry fabrics followed a power-law relationship between pressure and Vf where the power law index, b, was determined to be approximately equal to 10.3, 9.8 and 9.1 for assemblies consisting of plain, twill and 5 harness satin weaves, respectively. A mathematical analysis was performed for the viscoelastic compression of resin impregnated assemblies of fabrics and a model was developed incorporating the deformation of the fibre network and resin flow through the reinforcement. A methodology was devised for the geometrical representation of plain weaves in the microstructural analysis of cured laminates. The microstructural studies then provided data for the area and geometrical parameters of the yam cross-section; the mean amplitude, wavelength and phase angle of the yarn waveform and the distance between plies at different compression pressures. Cross-sections of laminates with each of the considered fabrics were compared in terms of fibre area fractions, porosity and void content, for the different compression pressures. The aim for the microstructural analysis was to elucidate the mechanism of compression and to follow the development of fibre and pore structure under different degrees of compression. It was concluded that the compression of resin impregnated woven fabrics could be considered as a combination of four modes of deformation; (a) the elimination of a resin rich interlayer between adjacent layers of fabric; (b) the nesting of layers of fabric by slipping while under compression; (c) the deformation of the yam waveform which results in the reduction of thickness of individual plies and (d) the compression and deformation of the cross-sections of individual yarns.

Resin Transfer Moulding (RTM), at present, is a semi-automated, low volume production process for fibre reinforced plastics with much work being undertaken to achieve full automation and medium volume production. The efficient and consistent impregnation of thermosetting resin into a reinforcement pack are key requirements in the development of R TM. For mass production and automation accurate predictions of mould fill times are critical in order to plan production. At present, the fill times are predicted by assuming Darcy's law and using experimentally determined permeability value. Permeability measurements display a large amount of scatter which affects the accuracy of simulations of mould filling. Commercial fabrics such as 'Injectex' are now available which achieve high permeability to resin flow through a modified microstructural architecture of fibre tows. This thesis seeks to examine the relationship between microstructure and permeability in these fabrics in the context of possible decreases in mechanical properties which may result from non-uniform fibre distribution. An image analysis technique is used to characterise and quantify the regions of flow within the fabric architecture which are then correlated with a series of permeability measurements determined experimentally. This then leads onto a mathematical model for the prediction of the permeability of the fibre pack.

In recent years interest has grown in the use of composite components within the automotive industry. Fibre reinforced plastic (FRP) components are of particular interest to the industry, since lower tooling costs and part consolidation can be utilised, whilst lighter, stiffer components can be produced. Several methods are available to produce FRP components at high volumes, including compression moulding (using dough and sheet moulding compounds), reinforced reaction injection moulding (RRIM) and liquid moulding processes (resin transfer moulding (RTM) and structural reaction injection moulding (SRIM)). RTM is a closed mould process, which is widely used to produce components economically in low volumes using matched moulds to produce two good surfaces. The absence of a high volume manufacturing technology, however, has impeded the acceptance and advance of RTM within the automotive industry. A research programme was established at the University of Nottingham to address the problems associated with the use of RTM for high volume manufacture. This programme has considered the topics of process technology, processing characteristics of polyester resin systems and fibre preforms, fibre wet-out and interfacial bonding, mould design, microwave pre-heating of reactive resin systems and process modelling. This thesis concerns the research which was undertaken to identify the causes of void formation during the impregnation and polymerisation stages of RTM, and methods of reducing the final void content within the component. The impregnation phase of the RTM process was identified as being the stage where the majority of voids were formed. A study of oil impregnation (having a similar viscosity to that of resin) into reinforcement was undertaken to determine the reasons for uneven flow and air entrapment. The dry reinforcements were studied to assess the microstructure of the preforms in order to determine reasons for obstruction of the resin flow. Fabric stitching, thermoplastic binder and size deposits were identified as potential causes of flow impediment. Fibre orientation and preform stacking were also assumed to assist in the development of uneven flow, leading to air entrapment. A major factor determining the formation of microvoids within fibre bundles was identified as the transverse impregnation of resin into high Tex fibre bundles. The major moulding process variables of injection pressure, vent pressure, fibre volume fraction, mould temperature and resin pre-heating have been assessed, to determine their effect on the void content within unidirectional and CFRM reinforced polyester laminates. It was observed that vacuum assistance during impregnation reduced void formation, although higher exotherm pressures and the possibility of monomer boiling arise from its use. A simple impregnation model was developed to assess the microscopic impregnation rates between fibre bundles, in the capillary between fibres and transversely into fibre bundles. The results from this model were compared with actual moulding histories. The findings of the overall work are discussed and suggestions proposed for the reduction of void content in RTM automotive components.