Author: Benedict D. Rogers, Mechanical and Aeronautical Engineering
The open-source DualSPHysics software aims to simulate the continuously evolving challenge of fluid flows in industrial processes and environmental applications. Conventional well-established and mature commercial software use a computational mesh, but it is increasingly apparent that this is inadequate for the progressively sophisticated demands of modern industry and research. The Smoothed Particle Hydrodynamics (SPH) method is meshless using particles to represent the moving fluid and solids. Developed originally for astrophysics, SPH has seen rapid development in the last two decades for application to engineering problems. The meshless nature of SPH makes it ideal for applications with large deformation, multiple phases, moving interfaces, floating and deformable objects.
First released in 2011, DualSPHysics has two versions: one optimised for multi-core CPUs and another for fast execution on Graphics Processing Units (GPUs). Taking advantage of the NVIDIA GPUs and the CUDA programming environment, the GPU version was designed to be fast and energy efficient. Without the need for massive supercomputers DualSPHysics allows simulating 10s million particles at runtimes of hours on a desktop PC, instead of months using multi-core CPUs.
Use of DualSPHysics has grown, with over 100,000+ downloads supported by a GitHub repository. There are now Training Days and regular DualSPHysics Users Workshops with the 5th Workshop taking place online in 2021 with 226 registered participants from 38 countries in industry and academia. A DualSPHysics YouTube channel already exists featuring tutorial playlists with 10,000s of views.
Applying open research practices
DualSPHysics is a consortium of 6 universities around Europe and the US, with a larger network of additional contributors. Releasing the code using the Lesser General Public License (LGPL) license allows companies to exploit the open-source code commercially at the same time as allowing us as researchers to continue developing the source code and keep it open. The development of the code follows a cycle of initial idea, development and validation, peer-reviewed publication, code preparation and release. The major code releases are supported by peer-reviewed open-access publications (Domínguez et al. 2021, Crespo et al. 2015, Crespo et al. 2011 - see Top Tip below). Like all open-source software, the code is available on GitHub.
The biggest challenge is coordinating the simultaneous contributions of multiple code authors (PhD students and postdocs) and incorporating radically new developments. The secret to overcoming this is using code repositories, regular meetings of the core development team, and engagement with the user community.
Benefits of using these open research practices
- The biggest benefit to academics is that it has led to a huge range of new collaborations with both industry and other universities around the world leading to exciting new ideas, developments, opportunities for research and application.
- Working as an international consortium has given the members access to a far greater range of expertise, support and larger team in leading the development of our research and state-of-the-art open-source code.
- The biggest benefit to users is that they get access to a state-of-the-art code that is continually updated with a community co-design approach.
Write an open-access peer-reviewed article that describes the code and its applications - this has led to many new collaborations and a large number of citations (600+ citations).
Open-source physics solver for engineering flows
YouTube channel with tutorials and examples.