A5049218014- Mechanical Behavior of Composites
- Probabilistic and Robust Engineering Design
- Model Reduction and Neural Networks
- Numerical methods in engineering
- Markov Chains and Monte Carlo Methods
- Advanced Materials and Mechanics
- Composite Structure Analysis and Optimization
- Vibration and Dynamic Analysis
- Statistical Methods and Inference
- Manufacturing Process and Optimization
- Elasticity and Material Modeling
- Advanced Multi-Objective Optimization Algorithms
- Gaussian Processes and Bayesian Inference
- Advanced Numerical Methods in Computational Mathematics
- Autonomous Vehicle Technology and Safety
- Epoxy Resin Curing Processes
- Composite Material Mechanics
- Reservoir Engineering and Simulation Methods
- Structural Analysis and Optimization
- Electromagnetic Simulation and Numerical Methods
- Aerospace and Aviation Technology
- Metal Forming Simulation Techniques
- Air Traffic Management and Optimization
- Groundwater flow and contamination studies
- Nonlocal and gradient elasticity in micro/nano structures
University of Exeter
2017-2024
Phillips Exeter Academy
2024
Turing Institute
2019-2023
The Alan Turing Institute
2018-2023
King's College Hospital
2023
British Library
2021-2022
University of Bath
2012-2019
In this paper we address the problem of prohibitively large computational cost existing Markov chain Monte Carlo methods for large-scale applications with high-dimensional parameter spaces, e.g., in uncertainty quantification porous media flow. We propose a new multilevel Metropolis--Hastings algorithm and give an abstract, problem-dependent theorem on estimator based set simple, verifiable assumptions. For typical model subsurface flow, then provide detailed analysis these assumptions show...
An efficient surrogate modelling framework is proposed for full-field predictions of stresses and cracks in composite material microstructures. The comprises two sequential convolutional neural networks (CNNs), predicting the elastic stress fields local crack maps, respectively. Training test data are created from high-resolution fracture simulations randomly generated representative volume elements (RVEs), including geometric variabilities such as fibre fraction porosity. This work shows...
Offshore wind turbines are complex pieces of engineering and are, generally, exposed to harsh environmental conditions that making them susceptible unexpected potentially catastrophic damage. This results in significant downtime high maintenance costs. Therefore, early detection major failures is important improve availability, boost power production, reduce article proposes a supervisory control data acquisition (SCADA) data-based Gaussian process (GP) (a data-driven, machine learning...
In this paper we address the problem of prohibitively large computational cost existing Markov chain Monte Carlo methods for large-scale applications with high-dimensional parameter spaces, e.g., in uncertainty quantification porous media flow. We propose a new multilevel Metropolis--Hastings algorithm and give an abstract, problem-dependent theorem on estimator based set simple, verifiable assumptions. For typical model subsurface flow, then provide detailed analysis these assumptions show...
Novel Design and Analysis of Generalized Finite Element Methods Based on Locally Optimal Spectral Approximations
A new framework for optimising the process of forming dry textile materials using finite element (FE) analysis and Gaussian Process (GP) regression is explored in this work. FE models were generated to simulate double diaphragm non-crimp fabric over a hemisphere tool. GP emulator was developed regress dataset by model, then used optimise process. Importantly simulations can capture formation wrinkles during under different configurations boundary conditions. Rigid blocks (risers) introduced...
This paper presents a lightweight, open-source and high-performance python package for solving peridynamics problems in solid mechanics. The development of this solver is motivated by the need fast analysis tools to achieve large number simulations required `outer-loop' applications, including sensitivity analysis, uncertainty quantification optimisation. Our software toolbox utilises heterogeneous nature OpenCL so that it can be executed on any platform with CPU or GPU cores. We illustrate...
.We develop a novel Markov chain Monte Carlo (MCMC) method that exploits hierarchy of models increasing complexity to efficiently generate samples from an unnormalized target distribution. Broadly, the rewrites multilevel MCMC approach Dodwell et al. [SIAM/ASA J. Un‐certain. Quantif., 3 (2015), pp. 1075–1108] in terms delayed acceptance Christen and Fox [J. Comput. Graph. Statist., 14 (2005), 795–810]. In particular, is extended use arbitrary depth allow subchains length. We show algorithm...
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI
In this research, a Gaussian process (GP) surrogate modelling framework for the forming of dry carbon-fibre textile was investigated. A particular focus work is development dimension reduction algorithms, allowing to solve high-dimensional sparse optimisation problems. The concept active subspace adopted find principal space problem. Then, low-dimensional (i.e., active) can be obtained by selecting directions with highest explained variance. kernel-combined GP format developed. This takes...
The large deformation Cosserat continuum model presented shows the potential of capturing internal buckling instabilities within layered media at a fraction computational cost when compared with conventional modelling methods. A homogenized elasto-plastic tensor is derived for periodic arrangements orthotropic elastic layers separated by weak interfaces that exhibit modified Mohr–Coulomb friction law. Focus given to physical interpretation measures, derivation and implementation using finite...
Quantifying the uncertainty in model parameters and output is a critical component model-driven decision support systems for groundwater management. This paper presents novel algorithmic approach which fuses Markov Chain Monte Carlo (MCMC) Machine Learning methods to accelerate quantification flow models. We formulate governing mathematical as Bayesian inverse problem, considering random process with an underlying probability distribution. MCMC allows us sample from this distribution, but it...
In the deformation of layered materials such as geological strata, or stacks paper, mechanical properties compete with geometry layering. Smooth, rounded corners lead to voids between layers, while close packing layers results in geometrically-induced curvature singularities. When are penalized by external pressure, system is forced trade off these competing effects, leading sometimes striking periodic patterns. this paper we construct a simple model geometrically nonlinear multi-layered...
In this paper, we present a generalization of the multilevel Monte Carlo (MLMC) method to setting where level parameter is continuous variable. This (CLMC) estimator provides natural framework in PDE applications adapt model hierarchy each sample. addition, it can be made unbiased with respect expected value true quantity interest provided converges sufficiently fast. The practical implementation CLMC based on interpolating actual evaluations at finite number resolutions. As our new...
The emergence of additive manufacture (AM) for metallic material enables components near arbitrary complexity to be produced. This has potential disrupt traditional engineering approaches. However, AM exhibit greater levels variation in their geometric and mechanical properties compared standard components, which is not yet well understood. uncertainty poses a fundamental barrier users the material, since extensive post-manufacture testing currently required ensure safety standards are met....
In this paper we derive an obstacle problem with a free boundary to describe the formation of voids at areas intense geological folding. An elastic layer is forced by overburden pressure against V-shaped rigid obstacle. Energy minimization leads representation as nonlinear fourth-order ordinary differential equation, for which prove there exists unique solution. Drawing parallels Kuhn–Tucker theory, virtual work, and ideas duality, highlight physical significance equation. Finally, show that...