Plenary Speakers

Fracture-scale fluid flow models for the simulation of flows in deformable porous materials

Dr Tim Hageman - UKACM 2020 Roger Owen Prize

Tim Hageman is an 1851 research fellow in at the University of Oxford, working on developing novel finite element schemes for a wide range of multi-physics and multi-scale problems (including hydraulic fracturing processes within ice sheets, hydrogen embrittlement and failure of metals, and corrosion processes). Prior to this, he was a research associate at Imperial College London performing research related to similar topics. Tim obtained his PhD at the university of Sheffield under the supervision of prof. René de Borst, developing multi-scale models for poroelasticity. His PhD thesis has earned the UKACM 2022 Roger Owen prize for best thesis in the area of computational mechanics in the UK, and the ECCOMAS PhD award for one of the two best theses related to computational modelling in Europe. 

Abstract

Fluid flows through deformable and fracture-able porous materials are dominant in many applications of engineering interest, such as underground geo-thermal and gas storage, spreading of pollutants, etc.. However, while these flows occur throughout the complete porous material (often in the size of km), the largest contribution to the flow comes from cracks and fractures with opening heights in the order of mm. This difference in length scales makes simulation of flows and deformations costly if all scales were to be directly captured. Here, two-scale approaches are presented which resolve the macro-scale fluid flow and deformations, coupled to a micro-scale formulation for the evolution of the fluid flow within the crevasses. By separating these scales, complex phenomena such as multi-phase and non-linear fluid flows can be captured at the small scale. Additionally, special attention is paid to obtaining a consistent coupling between this micro-scale and the macro-scale, resulting in consistent tangent matrices which greatly enhance the stability and convergence rate of the overall finite element scheme. The resulting equations are solved using a finite element scheme, comparing the use of standard Lagrangian elements with spline-based elements, and representing propagating cracks by inserting interface elements. Applications are shown for typical hydraulic fracturing benchmark cases, showing the capabilities of capturing travelling pressure waves and nonlinear fluid models.

Boundary elements – can they be useful to engineers?

Professor Jon Trevelyan

Jon Trevelyan is Emeritus Professor of Engineering at Durham University. After obtaining his PhD in Civil Engineering from Bristol (studying the dynamics of double curvature arch dams) he worked for over a decade in industry. The majority of this period was spent with the Computational Mechanics Group, developers of the BEASY software, and included seven years running the North American operations of the Group.  On his return to the UK, Jon made a career change into academia, first spending one year at the University of Brighton, and then in 1996 being appointed to a lectureship at Durham, where he spent the remainder of his career. His early years at Durham working alongside Professor Peter Bettess were formative ones, particularly in together developing enriched BEM formulations for acoustics problems. The idea of using non-standard basis functions, removing the adherence to piecewise polynomial elements, has been a continuing unifying theme in his research since then, and has led to promising developments in enriched BEM schemes for fracture mechanics and to isogeometric BEM approaches.  Jon’s BEM software has been used for over twenty years in the aerospace sector. He is a Fellow of the Institution of Mechanical Engineers, and has also served a four-year term as Head of the (then) School of Engineering and Computing Sciences in Durham. Jon formally retired in 2023 but has retained Emeritus Professor status so he can still access Matlab.

Abstract

The Boundary Element Method (BEM) has long been overshadowed by the Finite Element Method (FEM). In particular, since companies often like to maintain (and train their engineers in) just one analysis code, the more versatile FEM has found a much easier acceptance in industry. So what are the reasons why the BEM has continued to retain interest among computational mechanics researchers, and also to gain some usage in industry? In this lecture, Jon Trevelyan will explore some of those reasons.

He will focus on two areas: 

• the boundary-only modelling used by the BEM can bring significant levels of interactivity in mechanical design, enabling numerical stress analysis to move forward in the design cycle to the conceptual design stage. Real-time performance can be achieved in updating stress contours as an engineer dynamically alters his/her design geometry, allowing a speedy overview of the design space to arrive more quickly at a final design geometry on which to do more detailed analysis. Recent work will be described showing how this can be extended in the future to larger model sizes.

• the use of enrichment functions, no longer being tied to piecewise polynomial elements, can bring significant advantages to certain problems. It will be shown how such enriched BEM approaches can be highly effective in both acoustics and fracture problems, delivering accurate solutions from coarse meshes (FEM researchers familiar with XFEM will recognise some of these ideas). Further, with the additional introduction of the isogeometric philosophy, we will see how the basis functions selected for an analysis can be formed from bespoke combinations of polynomial, exponential, trigonometric and NURBS functions.

Engineering in the Exascale Era

Professor David Emerson

David Emerson is a Visiting Professor at the University of Strathclyde and Fellow of the Institute of Physics. He leads the Computational Engineering Group at STFC Daresbury Laboratory and his research interests cover numerical algorithms and high-performance computing applied to computational fluid dynamics with a particular interest in micro/nanofluidics, rarefied gas dynamics, and high speed flows. He is a founding member of the High-End Computing engineering consortia and was an expert member of the European Exascale Software Initiative and the International Exascale Software Project, the latter projects identifying the challenges to be addressed for the exascale era of computing. He has held numerous grants with EPSRC and has recently been involved in the ExCALIBUR (Exascale Computing Algorithms & Infrastructures Benefitting UK Research) and recently-funded activities include “Integrated Simulation at the Exascale: coupling, synthesis and performance” (EP/W00755X/1), which will investigate high-fidelity coupling to support extreme-scale multi-physics and multi-scale simulations, and “Turbulence at the Exascale: Application to Wind Energy, Green Aviation, Air Quality and Net-Zero Combustion” (EP/W026686/1). 

Abstract

Every 12 years or so, we witness a step change in computing capability. In 2008, we saw high performance computing (HPC) breakthrough the petaflop barrier, delivering more than 1,000,000,000,000,000 operations per second. This was achieved with IBM’s Roadrunner and was quickly followed by Cray’s Jaguar facility at Oak Ridge National Laboratory. As we approached realising a petascale capability, people were asking where we go next to deliver a thousand-fold improvement; is an exascale computer feasible? What are the challenges? What are the barriers? In a seminal report published in September 2008, and led by Peter Kogge, four major challenges were identified relating to energy, memory, concurrency, and resilience. The road to exascale computing had begun with many challenging and exciting research themes to solve. As a result of many contributions, exascale computing became a reality in June 2022 with Frontier delivering unprecedented computing capability at a relatively modest energy cost. In response to these technological advances, the UK, through The Future of Compute: Final report and recommendations [1], has provided a roadmap for HPC deployment in the UK across a spectrum of technologies and applications covering AI, quantum, and exascale computing. A £300M major investment in AI hardware is already taking place through the AI Research Resource (AIRR) with systems at Cambridge and Bristol. In addition, further funding (£500M) to support the UK’s AI infrastructure has been made available. As part of this investment, an exascale computing facility is planned with a tentative timeframe of deployment in 2026. In support of the UK’s exascale ambitions, the ExCALIBUR project has been looking at a broad range of activities that will target exascale computing. In the talk, we will look at the path to exascale computing and how the challenges were addressed. We will look at current and future investment in compute in both software and hardware and consider the challenges that need to be tackled. There are many applications that will benefit from this scale of computing including health. climate, materials, astrophysics, energy (fusion, nuclear, green), security, and others. Through the ExCALIBUR programme, we have been working on two complementary topics investigating code-coupling at scale, which will allow high-fidelity multi-physics simulations, and turbulence at the exascale, with a focus on green energy and environmental issues. The fundamental challenges associated with simulation at this scale, and the opportunity to deliver breakthrough simulations, will provide an exciting time for young researchers working in the field of computational engineering.

[1] https://www.gov.uk/government/publications/future-of-compute-review Independent Review of The Future of Compute: Final report and recommendations

Computational Modelling and AI assisted Design of Flexoelectric Metamaterials

Professor Dr. Xiaoying Zhuang

Dr. Xiaoying Zhuang’s key research area is computational solid mechanics. She has developed multiphysics and multiscale methods for materials design including nano composites, metamaterials and nanostructures. Dr. Xiaoying Zhuang obtained her PhD in Durham University, UK in 2011, which is followed by her postdoc in Norwegian University of Technology in Trondheim and then joined Tongji University as assistant professor and promoted to associate professor in early career stage. In 2015, she was awarded with the Sofja-Kovalevskaja Programme from Alexander von Humboldt Foundation that brought her to Germany and she focused on the modelling and optimization of polymeric nanocomposite. Her ongoing ERC Starting Grant is devoted to the optimization and multiscale modelling of piezoelectric and flexoelectric nano structures. In 2018, she was awarded with Heinz-Maier Leibnitz Prize and in 2020 awarded with Heisenberg-Professor Programme of DFG. 

Abstract

Flexoelectricity is a more general phenomenon than the linear change in polarization due to stress, the piezoelectric effect. In contrast to piezoelectricity, flexoelectricity exists in wider range of centrosymmetric materials especially nontoxic material useful for biomedical application. Flexoelectricity grows dominantly in energy density when scale reduces to submicro or nano, meaning the promise of enabling self-powered nano device such as body implant and small-scale wireless sensor. In Hannover, multiscale modelling and experimental characterization of flexoelectric materials and design of flexoeletric structures are being carried out. In this talk, we will present the machine learning assisted flexoelectric materials characterization and the topological optimization for single and multi-phase flexoelectric structure from atomistic to continuum scale. New formulation for onlinear topological optimization for flexoelectric structures accounting for the nonlocal stress and large deformation process will be presented. In the end, phononic metamaterials for enhancing the flexoelectricity is utilized and integrated in the design to outperform the current design of nano energy harvesters. Interesting phenomenon of topological insulating states and metaplates will be shown that can significantly enhance the performance of flexoelectric energy harvester.