Embedded Files
Generalised Deep Learning Model for Prediction of Multiphysics Properties of Composites
Generalised Deep Learning Model for Prediction of Multiphysics Properties of Composites
A generalized material-encoded Convolutional Neural Network (CNN) is developed for fibre-reinforced composite materials with a ratio of elastic moduli of the fibre to the matrix between 5 and 250 and fibre volume fractions between 25 and 75%, which spans the end-to-end practical range. A dataset containing 18000 microstructures was used whose effective thermomechanical properties are obtained using variational asymptotic homogenisation. To make the predictions physically admissible, models are trained by enforcing Hashin–Shtrikman bounds which led to enhanced model performance in the extrapolated domain.
Dataset distribution (top), Training and predictions (bottom).
Impact Damage Modelling & Experiments in Aerospace Composites
Impact Damage Modelling & Experiments in Aerospace Composites
An integrated experimental-computational approach was developed to quantify damage in composites under impact. Rate-dependent cohesive and continuum damage models were developed and calibrated using extensive experimental campaign involving tests across scales , modes and strain rates. Split Hopkinson Pressure Bar, Hydraulic testing machine and gas gun tests are used for characterisation and validation. The project was conducted in close collaboration with Rolls-Royce and funded by Innovate UK.
Lifetime Prediction and Uncertainty Quantification of Self-Healing Thermal Barrier Coatings
Lifetime Prediction and Uncertainty Quantification of Self-Healing Thermal Barrier Coatings
Funded by the Netherlands Innovation Programme and EU Horizon FP7 Grants, the project adopted an integrated computational-experimental approach to design and develop novel self-healing thermal barrier coatings (SH-TBC). Lifetime prediction models were developed utilising a multiscale and multiphysics approach. Data-driven surrogates were developed to quantify uncertainties in the lifetime of SH-TBC.
Multifunctional Piezo-Battery Composite for Next-Gen Aircraft
Multifunctional Piezo-Battery Composite for Next-Gen Aircraft
The project was led by Dr Sathiskumar Ponnusami with researchers from the US, UK, India and the Netherlands, as part of the Airbus Fly Your Ideas Global Challenge. Dr. Dineshkumar Harursampath was the academic mentor for the team. With novel integrated multifunctional composites for energy recycling and storage, it was awarded £30,000 first prize out of 500+ teams from over 100 countries. The project was evaluated by international aerospace experts from Airbus and other aerospace industry/academia.
A multifunctional composite aircraft wing and fuselage
Strain-gradient beam stiffness (top-left), Model Comparison (top-right), Cross-sectional nonlinear deformation (bottom-left) and 3-D deformed CNT (bottom right).
Asymptotically-Accurate Strain-Gradient Plate and Beam Models
Asymptotically-Accurate Strain-Gradient Plate and Beam Models
A novel approach to model length-scale effects in micro and nano beams and plates by using the dimensional reduction technique, Variational Asymptotic Method (VAM). Traditional continuum approaches have been advanced by accounting for the length-scale effects, which have gained significant importance in micro- and nano-scale applications like MEMS and NEMS devices. Asymptotically-correct modified strain gradient beam and plate theories has been developed by incorporating higher-order length-scale parameters through VAM, integrating the strain gradient theory within the VAM framework for the first time. A systematic investigation has been conducted to assess the effect of the choice of the order of the material length-scale parameter, providing considerable insights into the appropriateness of the choice regarding the beam behaviour. The developed theory captures the stiffening effect of the length-scale parameters and establishes an asymptotically-accurate beam model that includes higher strain gradient effects.
Gaussian Process-based Performance Prediction and Bayesian Optimisation of Screw Compressor Systems
Gaussian Process-based Performance Prediction and Bayesian Optimisation of Screw Compressor Systems
The present research investigates the potential of Gaussian Process Regression (GPR) and Bayesian optimization for the prediction and optimization of the performance of an oil-flooded screw compressor. Specifically, the GPR-based surrogate model is developed to predict the compressor performance characteristics based on its geometrical design parameters. The resulting surrogate model is then used to optimize the compressor design parameters using Bayesian optimization. The results are compared with optimization using Genetic Algorithm (GA) and physics-based multi-chamber thermodynamic model. It was shown the proposed approach results in similar optimal design parameters but with a significantly less optimization time by a factor of 7.
Google Sites
Report abuse