It is important than ever that our SPH community finds new ways to meet, discuss and share the latest developments and ideas. The SPHERIC Steering Committee is therefore delighted to announce a brand new online seminar series on SPH! These monthly seminars are open to everyone and free to attend.
To register for these seminars, please fill out the form here. The joining details will be sent by email shortly before each seminar.
Also, please note that most of these seminars are recorded. You can find the video of each seminar by clicking the titles or visiting the SPHERIC YouTube channel here.
Future Speakers (Titles and dates are tentative)
(Previous talks can be viewed on the SPHERIC YouTube Channel)
Dr. Matthieu Schaller
Date and time: Thursday, March 3, 2022 - 10:00 UTC
Title: HPC: Task-based parallelism for SPH
Abstract: Cosmological simulations present unique challenges for numerical methods. They exploit a challenging combination of large volumes and high-resolution to cover all the scales of interest. This is compounded by the coupling of the SPH method to a gravity solver and, often, to a large network of sub-resolution models to represent stars and black holes. The low arithmetic intensity of SPH also leads to difficulties harnessing the raw calculation capabilities of modern HPC systems. In this talk, I will introduce the key challenges in the base SPH algorithms (neighbour finding, etc.) used for cosmological simulations. I will then discuss the need for modern load-balancing methods such as task-based parallelism. Finally, I will sketch the road towards multi-trillion-particle simulations and ideas for the path to exa-scale.
Bio: Dr. Matthieu Schaller is a lecturer in numerical cosmology at the university of Leiden (The Netherlands), where he works on astrophysical simulations of the large-scale structure of the Universe. He is the lead-developer of the SWIFT simulation code, which is designed to the tackle the challenges coupling hydrodynamics with gravity and other sub-resolution processes on large HPC systems. His research focuses on the one hand on the analysis and applications of SPH simulations to astrophysical problems, and, on the other hand, on the development of modern parallelization methods and numerical methods.
Prof. David Le Touzé and Dr. Salvatore Marrone
Date and time: Thursday, February 3, 2022 - 13:00 UTC
Abstract: Water flows obsessed Leonardo Da Vinci throughout his life. One of his most famous drawings represents a water jet impacting on a pool (RCIN 912660 The Windsor Collection). and has often been used by many fluid dynamicists as the first important document concerning turbulent flows. The word “turbulence” itself, one of the most important phenomena in fluid dynamics, was used for the first time by da Vinci in the “Atlantic codex.” Our aim is to reproduce one of Leonardo's drawings with in-house solvers developed by our French-Italian group to see if through the analysis of the simulations we are able to recognize the descriptions written by Leonardo in his notes, 500 years after his death. After a detailed study of different historical documents, we sought to reproduce the flow drawn in the sheet RCIN 912660 using the δ-LES-SPH model in order to try to answer some of the questions behind this drawing.
Bio: David Le Touzé (top photo) is a Professor of Fluid mechanics at the Ecole Centrale Nantes, where he is currently the Director of the LHEEA laboratory. Since 1999, he has worked on different numerical methods (HOS, SPH, FV, DVH and LBM) and the coupling between methods, among which SPH is his main expertise which he and co-workers apply to ocean engineering and automotive industries, and to cardio-vascular flows. He is part of a collaborative framework for the development of the industrial code SPH-Flow.
Salvatore Marrone (bottom photo) is a permanent researcher at CNR-INM, the Institute of Marine engineering of the Italian National Research Council. Since 2008, the main focus of his research is on computational aspects of mesh-free numerical methods (SPH, DVH), including coupling with mesh-based methods, and their applications in the naval and marine engineering contexts. He is part of a collaborative framework for the development of the industrial code SPH-Flow.
Dr. Giuseppe Bilotta
Date and time: Wednesday, January 12, 2022 - 14:00 UTC
Abstract: Graphic Processing Units (GPUs), programmable video cards equipped with massively parallel compute units and originally designed for 3D graphics and video games, paved the road to a revolution in scientific computing, offering a low-cost high-performance parallel computing alternative to traditional CPU clusters. Weakly-compressible formulations of SPH are particularly amenable to implementation on GPU, with an intuitive mapping between numerical particles and computational work-items, but maximizing performance without sacrificing accuracy and robustness requires attention to detail and the adoption of some non-trivial programming strategies.
Bio: Dr. Giuseppe Bilotta is a Researcher at the Italian National Institute of Geophysics and Volcanology (INGV), where he works mainly on the modelling of geophysical flows for hazard assessment purposes. The computational challenges posed by the simulation of these complex fluids on real topographies have brought him to appreciate the benefits of Lagrangian meshless methods, leading to the development of GPUSPH, the first implementation of weakly-compressible SPH to run entirely on GPU.
Dr. Marcus Ihmsen
Date and time: Wednesday, December 15, 2021 - 10:00 UTC
The SPH method has been carried over from research labs in academia over to the industry, enabling engineers to solve real world problems. This is one of the grand challenges announced by the SPHERIC community. A change in the industry to simulate more and do less hardware testing is a perfect match for mesh-free methods that simplify workflows and reduce the computational effort by solving only for the liquid phase.
In this talk, Dr. Markus Ihmsen from FIFTY2 Technology, will present how and why the implicit SPH-based Preon solver has an edge over traditional CFD methods for water management applications. This is showcased on the application of water wading. Besides comparisons to a VOF-based method, simulation results are compared to physical wading tests, demonstrating that the SPH-based solver is able to qualitatively predict the results of the experiments.
Bio: Dr Markus Ihmsen is Managing Director of FIFTY2 Technology. He obtained his PhD from Albert-Ludwigs-Universität Freiburg im Breisgau (Germany) in 2013 and has been researching and developing SPH for over 10 years making many key advances in its theory and application from boundary conditions, to implicit incompressible SPH to parallelisation. Markus is now bringing SPH to the world of industrial application with the PreonLab software. FIFTY2 is a growing tech start-up and think tank in the field of computer science, mechanical engineering, computer graphics, numerics, and mathematics. With PreonLab, their particle-based simulation software for free-surface fluids, FIFTY2 empowers engineers to develop sustainable, innovative, and safe products at lower costs, in shorter time, and with minimal ecological impact.
Dr. Jean-Christophe Marongiu
Date and time: Thursday 15 April 2021, 12:00 UTC
Abstract: SPH-ALE is a variant of the weakly compressible SPH approach that makes use of an Arbitrary Lagrange Euler description. It also adopts some numerical ingredients derived from the Finite Volume Method. It is well adapted for free surface flows interacting with moving, complex-shaped solids. This talk will present a review of the main features of SPH-ALE, followed by examples of applications in the field of hydraulic turbines. Extensions to multiphase flows and fluid-structure coupling will also be briefly introduced. Finally some directions for future investigations and improvements will be drawn.
Bio: Dr. Jean-Christophe Marongiu works in the ANDRITZ Hydro industrial company as Head of the R&D department in Vevey, Switzerland. He has been driving the development of an in-house, SPH-based software package to simulate water flows and support the design of hydraulic turbines.
Dr. Renato Vacondio
Date and time: Wednesday 3 March 2021, 13:00 UTC
Abstract: As a meshless method, SPH has been adopted to simulate a broad range of applications from astrophysics to free-surface flows, to complex industrial flows. However, the SPH method still requires development to address important elements which prevent more widespread use. This talk will introduce the most relevant aspects of the SPH Grand Challenges as they have been identified by the members of the SPHERIC community. I will also briefly discuss what possible future work threads are needed to ensure that SPH further improves in the near future.
Bio: Dr. Renato Vacondio is an Assistant Professor in Civil Engineering at the University of Parma. His research is focused on challenging problems of developing numerical modeling for flood risk management using both Shallow Water and Navier–Stokes equations. The research on meshless Lagrangian methods has led him to become a developer of the open-source codes SPHysics and DualSPHysics.
Prof. Moubin Liu
Date and time: Wednesday 3 February 2021, 09:00 UTC
Abstract: SPH, as a truly Lagrangian and meshfree particle method, is very attractive in modeling fluid-structure interaction (FSI) problems. This talk reports some recent developments of SPH method in modeling FSI problems with rigid, elastic and flexible structures, with granular materials, and with extremely intensive loadings.
Bio: Moubin Liu is a Professor at the College of Engineering of Peking University, China. His areas of interest include computational fluid dynamics and computational fluid-structure interactions, using particle-based methods including SPH, and particle-grid coupling approaches including SPH-FEM, DEM-CFD, etc.
Dr. Steven Lind
Date and time: December 10th. 12:00 PM UTC
Abstract: The mesh-free method Smoothed Particle Hydrodynamics (SPH) is often applied in weakly compressible form. Incompressible SPH is an alternative approach with an ability to predict accurate pressure fields without the use of empirical equations of state, artificial sound speeds, or excessive numerical diffusion. This talk will provide an overview of recent developments undertaken at the University of Manchester for the incompressible SPH approach, including examples of its application in quite diverse areas of fluid mechanics. New methodologies for achieving stability and very high accuracy will also be discussed.
Bio: Dr. Steven Lind is a Reader in Smoothed Particle Hydrodynamics in the Department of Mechanical, Aerospace and Civil Engineering at the University of Manchester, UK. His research interests concern the development of the SPH numerical method and its application to diverse problems across engineering.
Prof. Daniel Price
Date and time: November 12th. 12:00 PM UTC
Abstract: Smoothed particle hydrodynamics is a mesh-free method for solving the equations of fluid dynamics on a set of particles that move with the fluid. My aim in this talk is to provide a set of guiding principles for you, the SPH developer. Namely, how does one derive robust and accurate algorithms for simulating physics in fluids and gases? I will highlight similarities and differences compared to more standard numerical methods, and include practical examples from blood to black holes.
Bio: Daniel Price is a Professor in the School of Physics and Astronomy at Monash University in Melbourne, Australia. Prior to this he was a Royal Society University Research Fellow at the University of Exeter, before which I also held a PPARC/STFC Postdoctoral Research fellowship at Exeter. He completed his PhD at the Institute of Astronomy at the University of Cambridge. His research interests are broadly in Computational Astrophysics — generally involving star and planet formation, accretion discs and the Smoothed Particle Hydrodynamics (SPH) method. He has been in iso for most of 2020.