Periodic Reporting for period 1 - RELIANCE (REaL-tIme characterization of ANisotropic Carbon-based tEchnological fibres, films and composites)
Okres sprawozdawczy: 2023-02-01 do 2025-01-31
Improving Materials Manufacturing with Advanced X-ray Tools
The RELIANCE network aims to enhance the manufacturing and quality control of high-performance polymeric materials and composites by developing advanced X-ray imaging and scattering tools. These tools will enable real-time, automated characterization of materials at the nanoscale, ensuring precision, reliability, and efficiency in the manufacturing process.
Doctoral Training Programme and Industry Collaboration
RELIANCE trains 14 Doctoral Candidates in X-ray imaging, scattering, data reconstruction, machine learning, and automated analysis. The intensive network-wide doctoral training programme covers all aspects of these fields. Each research project is defined in collaboration with industry partners, addressing materials and problems closely related to their development programmes. This collaboration benefits industry partners through direct exploitation of results or access to a pool of knowledge and well-educated staff. Doctoral Candidates gain practical experience and knowledge, making them highly employable with skills aligned with Industry 4.0.
Addressing Key Manufacturing Needs
RELIANCE addresses the need for faster, more precise, and reliable real-time nanoscale characterization to ensure high-quality materials. Traditional methods relying on force and temperature control are insufficient for monitoring developing structures in-line during production. These industrial processes involve shaping materials by subjecting them to carefully adjusted processing parameters like force, temperature, and speed. While these parameters are easy to control, monitoring the developing structure in-line is challenging.
Impactful Outcomes
RELIANCE aims to achieve several impactful outcomes by integrating real-time data analysis with advanced X-ray scattering tools. This integration will provide high-speed characterization of nanoscale structures, allowing instant feedback to optimize manufacturing processes, identify defects, and improve material properties. This shift from in situ to in-line applications enables smart production lines with autonomous process control. The methodologies developed will equip production lines with smart sensors and detectors, combined with modelling and large-scale data handling.
Consortium of Leading International Experts
To achieve these ambitious goals, RELIANCE brings together a consortium of leading international experts in X-ray scattering, imaging, automated analysis of scattering data, 3D reconstruction algorithms, and automated analysis of imaging data. This consortium includes industrial leaders in manufacturing high-performance polymer materials and composites, and highly specialized X-ray instrumentation.
The RELIANCE network aims to enhance the manufacturing and quality control of high-performance polymeric materials and composites by developing advanced X-ray imaging and scattering tools. These tools will enable real-time, automated characterization of materials at the nanoscale, ensuring precision, reliability, and efficiency in the manufacturing process.
Doctoral Training Programme and Industry Collaboration
RELIANCE trains 14 Doctoral Candidates in X-ray imaging, scattering, data reconstruction, machine learning, and automated analysis. The intensive network-wide doctoral training programme covers all aspects of these fields. Each research project is defined in collaboration with industry partners, addressing materials and problems closely related to their development programmes. This collaboration benefits industry partners through direct exploitation of results or access to a pool of knowledge and well-educated staff. Doctoral Candidates gain practical experience and knowledge, making them highly employable with skills aligned with Industry 4.0.
Addressing Key Manufacturing Needs
RELIANCE addresses the need for faster, more precise, and reliable real-time nanoscale characterization to ensure high-quality materials. Traditional methods relying on force and temperature control are insufficient for monitoring developing structures in-line during production. These industrial processes involve shaping materials by subjecting them to carefully adjusted processing parameters like force, temperature, and speed. While these parameters are easy to control, monitoring the developing structure in-line is challenging.
Impactful Outcomes
RELIANCE aims to achieve several impactful outcomes by integrating real-time data analysis with advanced X-ray scattering tools. This integration will provide high-speed characterization of nanoscale structures, allowing instant feedback to optimize manufacturing processes, identify defects, and improve material properties. This shift from in situ to in-line applications enables smart production lines with autonomous process control. The methodologies developed will equip production lines with smart sensors and detectors, combined with modelling and large-scale data handling.
Consortium of Leading International Experts
To achieve these ambitious goals, RELIANCE brings together a consortium of leading international experts in X-ray scattering, imaging, automated analysis of scattering data, 3D reconstruction algorithms, and automated analysis of imaging data. This consortium includes industrial leaders in manufacturing high-performance polymer materials and composites, and highly specialized X-ray instrumentation.
Work package 1: Instrumentation and Data Acquisition (the work done by DC1, DC2, DCS1, DCU1)
Efforts focused on acquiring state-of-the-art data for aramid yarns and pultruded carbon fibre composites to establish "ground truth" for their nano to microscale structures. Lab-scale X-ray equipment configurations were explored for real-time, high-throughput measurements to identify material parameters and defects. Successful 1-second X-ray exposures distinguished materials with different fibre orientations, promising for industrial use. Simulations showed that sparse X-ray projections, reconstructed by plenoptic refocusing, could provide usable cross-sectional views of pultruded carbon fibre composites at high speed.
Work package 2: 3D Reconstruction (the work done by DC3, DC4, DCS2)
The open-source Python tool FibreSimulator was created to generate customizable fibre composite phantoms with defects, combined with deep learning for real-time reconstruction from limited-angle X-ray data. A real-time scattering tensor tomography algorithm using pre-computed filters was developed for sparse sampling. High-quality phantoms were created for accurate X-ray scattering simulation. A dynamic uniform sampling strategy improved limited-data X-ray scattering and was successfully applied to general samples, with optimization ongoing for fibre samples.
Work package 3: Data Analysis (the work done by DC5, DC6, DC7)
This work involved processing experimental data, developing digital twins of composite materials, and detecting resin-rich anomalies in fibre composites. Mechanical tests on pultruded carbon fibres showed variations in fibre properties, confirming that scattering data can estimate them. Challenges include distinguishing production parameter effects from external factors and adapting methods for automated, high-speed analysis. Digital twins were created using FibreTracker software for structural modelling. A digital twin for woven composites generated training data for segmentation algorithms. Super-resolution neural networks enhanced low-resolution images for large-scale defect identification, with a dataset of paired images being compiled. Initial focus is on resin-rich regions, with future work on more complex anomalies.
Work package: Modelling and Application (the work done by DC8, DC9, DC10, DCU2)
Two micromechanics-based models were developed to predict how fibre misalignment impacts compressive strength and kink band formation, using local fibre orientation and fibre/matrix properties as inputs. Voids in thermoplastics reinforced with carbon fibres were characterized for their influence on kink band formation under compressive loading using synchrotron X-ray CT imaging. Initial studies revealed new information about the physics and mechanisms controlling lithiation of carbon fibre yarns in structural composite batteries.
Efforts focused on acquiring state-of-the-art data for aramid yarns and pultruded carbon fibre composites to establish "ground truth" for their nano to microscale structures. Lab-scale X-ray equipment configurations were explored for real-time, high-throughput measurements to identify material parameters and defects. Successful 1-second X-ray exposures distinguished materials with different fibre orientations, promising for industrial use. Simulations showed that sparse X-ray projections, reconstructed by plenoptic refocusing, could provide usable cross-sectional views of pultruded carbon fibre composites at high speed.
Work package 2: 3D Reconstruction (the work done by DC3, DC4, DCS2)
The open-source Python tool FibreSimulator was created to generate customizable fibre composite phantoms with defects, combined with deep learning for real-time reconstruction from limited-angle X-ray data. A real-time scattering tensor tomography algorithm using pre-computed filters was developed for sparse sampling. High-quality phantoms were created for accurate X-ray scattering simulation. A dynamic uniform sampling strategy improved limited-data X-ray scattering and was successfully applied to general samples, with optimization ongoing for fibre samples.
Work package 3: Data Analysis (the work done by DC5, DC6, DC7)
This work involved processing experimental data, developing digital twins of composite materials, and detecting resin-rich anomalies in fibre composites. Mechanical tests on pultruded carbon fibres showed variations in fibre properties, confirming that scattering data can estimate them. Challenges include distinguishing production parameter effects from external factors and adapting methods for automated, high-speed analysis. Digital twins were created using FibreTracker software for structural modelling. A digital twin for woven composites generated training data for segmentation algorithms. Super-resolution neural networks enhanced low-resolution images for large-scale defect identification, with a dataset of paired images being compiled. Initial focus is on resin-rich regions, with future work on more complex anomalies.
Work package: Modelling and Application (the work done by DC8, DC9, DC10, DCU2)
Two micromechanics-based models were developed to predict how fibre misalignment impacts compressive strength and kink band formation, using local fibre orientation and fibre/matrix properties as inputs. Voids in thermoplastics reinforced with carbon fibres were characterized for their influence on kink band formation under compressive loading using synchrotron X-ray CT imaging. Initial studies revealed new information about the physics and mechanisms controlling lithiation of carbon fibre yarns in structural composite batteries.
Highlighted results that have progressed beyond state of the art, includes:
• The demonstration that differentiation of otherwise very similar materials with different structural properties can be accomplished with X-ray scattering in a second on laboratory scale equipment.
• Smart X-ray scanning strategies specifically tailored for challenging materials like fibres, significantly reducing the amount of data needed while maintaining high image quality.
• Dynamic, feedback-controlled data acquisition strategies directly embedded into the operational pipelines of experimental facilities, making these efficient techniques readily available.
• The first digital twin model (The FibreTracker) to generate fibre representations directly from tomographic data rather than synthetic approximations, significantly improving accuracy for structural modelling
• A woven composite digital twin that uniquely integrates voxel-based material properties with virtual X-ray projections to create high-fidelity training data for segmentation algorithms.
• A machine-learning-based pipeline demonstrating the feasibility of estimating fibre properties from scattering data, a novel approach that enhances material characterization.
• An application of super-resolution neural networks for large-scale anomaly detection in fibre composites represents a cutting-edge method for defect analysis in low-resolution imaging.
• The development of micromechanics-based models that take local fibre orientation into consideration expanding knowledge on the mechanisms involved in compressive failure of carbon fibre reinforced composites.
• The demonstration that differentiation of otherwise very similar materials with different structural properties can be accomplished with X-ray scattering in a second on laboratory scale equipment.
• Smart X-ray scanning strategies specifically tailored for challenging materials like fibres, significantly reducing the amount of data needed while maintaining high image quality.
• Dynamic, feedback-controlled data acquisition strategies directly embedded into the operational pipelines of experimental facilities, making these efficient techniques readily available.
• The first digital twin model (The FibreTracker) to generate fibre representations directly from tomographic data rather than synthetic approximations, significantly improving accuracy for structural modelling
• A woven composite digital twin that uniquely integrates voxel-based material properties with virtual X-ray projections to create high-fidelity training data for segmentation algorithms.
• A machine-learning-based pipeline demonstrating the feasibility of estimating fibre properties from scattering data, a novel approach that enhances material characterization.
• An application of super-resolution neural networks for large-scale anomaly detection in fibre composites represents a cutting-edge method for defect analysis in low-resolution imaging.
• The development of micromechanics-based models that take local fibre orientation into consideration expanding knowledge on the mechanisms involved in compressive failure of carbon fibre reinforced composites.