Our research

Manchester has delivered a long list of developments in CFD over the past 4 decades, with significant focus on codes employing the finite volume method. Many of which have found their way into widely used computational engineering software; both commercial and opensource. The breadth of our work has enabled cross-fertilisation of ideas across sectors, and we have gained a world-wide reputation in this activity. 

Smoothed particle hydrodynamics (SPH) is a mesh-free method for simulating free-surface and multiphase flows. Manchester's SPH group excel have made a host of innovations that extend and improve the predictive accuracy and range of application of this approach, with a particular focus on coastal engineering and natural hazards.

Lattice Boltzmann methods (LBM) simulate fluid dynamics using particle distribution functions, which excel for complex geometries and multiphysics flows. At the University of Manchester, research in LBM has advanced predictive accuracy of turbulent flows, fluid structure interaction, and flows with complex multiphysics, including plasmas and multicomponent flows.

We are developing high-order meshless methods for direct numerical simulations (DNS) in complex geometries. These methods offer accuracy comparable to high-order finite differences, support arbitrary geometries, achieve up to 10th-order convergence, and provide variable resolution and dynamic adaptivity. Scalable to thousands of cores, they enable efficient, high-fidelity simulations for applications like turbulence and visco-elastic flows.

We enhance finite element methods (FEM) for complex material behaviour, covering multi-phase modelling, crystal plasticity, cohesive zones, and damage mechanics. Our work includes developing new element formulations and applying FEM to tough issues like quasi-brittle fracture, weld stress prediction, and composite failure. We emphasise parallel computing via ParaFEM, enabling large-scale simulations on high-performance systems.

We have pioneered a novel computational framework using combinatorial meshes, treating materials as discrete element assemblies. It employs variational formulations for transport phenomena on cell complexes, enabling element interactions across boundaries. Bridging particle and continuum modelling, it excels in complex structures, capturing topological changes like void formation and crack propagation. Applied to mass diffusion, heat conduction, fluid flow in porous media, and charge transport, our GeMS initiative extends these to engineering applications.

We develop peridynamic formulations for modelling discontinuous phenomena such as fracture, erosion, and multi-phase flow in porous media. Our work includes both theoretical advances in bond-based and state-based peridynamics, as well as practical implementations for problems involving phase transitions, heat transfer with solidification/melting, and coupled thermo-hydro-mechanical processes in geomaterials.

We employ quantum mechanical and atomistic simulations to study fundamental material behaviour and coupled thermo-chemo-mechanical problems. Our research spans from understanding irradiation damage in nuclear materials to predicting grain boundary migration in metals. We develop multi-scale approaches that bridge quantum mechanics, molecular dynamics, and continuum mechanics, providing seamless prediction capabilities across length scales.

We refine and develop Reynolds-averaged Navier–Stokes (RANS) models for steady and unsteady flows, including transitional flows, those with heat transfer and other phenomena such as magnetohydrodynamics.

We develop and routinely apply scale resolving methods such as large eddy simulation, direct numerical simulation and hybrid RANS-LES methods. These methods permit higher accuracy and more direct prediction of the unsteady phenomena relevant to thermomechanical fatigue and acoustics.

We develop models and methods for a range of heat transfer problems involving natural/forced convection, conjugate heat transfer and multiscale effects. A particular focus has been advanced wall treatments for turbulent flows.

We have worked on a range of methods to model the coupled response of the surrounding fluid with the immersed structure. This involves both rigid and flexible bodies. Methods include overset grid and the immersed boundary method.

We are world leaders in predicting welding-induced residual stresses and their effects on structural integrity. Our models capture the complex thermo-metallurgical-mechanical interactions during welding, including phase transformations, dilution effects, and post-weld heat treatment. We develop virtual prototyping tools for novel welding processes, including electron beam, laser, and arc-based techniques, supporting nuclear and aerospace manufacturing.

We specialise in modelling concrete, nuclear graphite, and other quasi-brittle materials using damage mechanics, fracture mechanics, and discrete approaches. Our continuum damage mechanics models predict crack initiation and propagation under complex loading conditions. We develop meso-scale models that explicitly represent material heterogeneity, enabling realistic predictions of localised failure and size effects.

We advance computational methods for polymer composites, from manufacturing process simulation to in-service performance prediction. Our work includes cohesive zone modelling of delamination, multi-scale analysis of fibre-matrix interfaces, and prediction of fatigue damage evolution. We develop models for composite bonded joints and repairs, addressing challenges in aerospace and renewable energy applications.

We combine algebraic topology, graph theory, and advanced computational methods to characterise and predict the behaviour of metallic materials. Our work reveals mesoscale features in alloys, disclination-induced stresses, and shear band formation. We develop multiscale tools for analysing millions of grains and structural elements, enabling next-generation microstructure characterisation and design of advanced alloys for extreme environments.

We work on the development of digital platforms which provide real-time updates, blending traditional engineering simulation with gaming platform physics to facilitate interactive analysis and ‘human in the loop’ interactive design. Live data sources can be incorporated for continuous on-site sensors.

Our expertise in design optimisation and uncertainty quantification leverages advanced computational techniques to enhance the performance and reliability of engineering systems. By employing sophisticated optimisation algorithms, we can explore high-dimensional parameter space in a far more complete manner than before. Uncertainty quantification then allows engineers to identify and explore key sensitivities and isolate critical aspects of the design space. It also permits a deeper understanding of a complex system compared to single state analysis.

We integrate artificial intelligence (AI) and machine learning (ML) with reduced order modelling (ROM) to create computationally efficient yet accurate simulations for complex systems. By training ML models on high-fidelity simulation data, we can enhance the predictive power of our models, enabling faster analysis and interpretation of results in CFD and solid mechanics applications.

Our team pioneers the adaptation of computational engineering to next-generation computing architectures, ensuring our methods scale with emerging technologies like exascale systems, GPU-based manycore platforms, and quantum computing. By developing algorithms tailored to these architectures, we address complex challenges in CFD and solid mechanics, preparing for the future of high-performance computing.