FDS: Consistent kernel-based approximations for variable resolution particle simulations

FDS > Research > Consistent kernel-based approximations

Consistent kernel-based approximations for variable resolution particle simulations

Scientific Background and Motivation

Focal point of the project is in the field of the numerical methods for Thermo-Fluid Dynamics (Computational Fluid Dynamics). Starting point for the work are the so called kernel based approximations, on which the Smoothed Particle Hydrodynamics (SPH) method is based. Unlike other simulation methods, SPH is based on the decomposition or the discretisation of matter and not of space. No grid needs to be generated or adapted for the use of SPH, which leads in the practice to simplify the work process. Moreover, the method suits very well the use of efficient, cost-saving GPU Hardware, whose development is supported by the continous demand of computer based games and virtual reality animations in the market.

The application of SPH to large-scale problems and real-life simulations usually involves dealing with complex geometries and large computational domains, which increases the computational effort. By virtue of the high potential of this numerical method, there is considerable interest in improvement of the methodology. In particular, the ability to dynamically adapt the discretisation as well as the optimisation of the kernel function, the fundamental component of the SPH approximation, are objects of intensive international efforts.

Aims and objectives

The project deals with the implementation of simulation techniques and creates a connection between Engineering, Applied Mathematics and Informatics. The PhD project addresses these application areas and has three scientific objectives: the improvement of the approximation accuracy, the enhancement of the robustness and of the reliability of the SPH method, the increase of the method efficiency. Focus of the development is an innovative corrective procedure, which is able to increase the accuracy and the reliability of the SPH approximation [1]. The strategy leaves the door open to the use of an efficient, variable resolution model, which is going to be developed and implemented for the project [2].

In the implementation a great importance will be given to maintain the computational effort low, in order to make the use of the correction possible and easy in the practice. Furthermore, the applicability of the method is supported by efficient graphic cards. In order to achieve the widest possible availability and visibility of the results, the implementation is going to be done in a globally used, GPU-accelerated open source tool (DualSPHysics), mainly developed jointly by researchers at the University of Vigo (Spain) and at the University of Manchester (U.K.) [3].

Complex geometries and practical fluid flow problems, which feature large dynamics, can be simulated with the techniques here proposed. The derived methodological expertise will be applied in this project to fluid flow problems, as well as to neighbouring research fields, in particular to geotechnical, coastal and hydraulic engineering with fluid-structure interaction.

Present Focus

Present efforts are concerned with the following aspects

Explicit correction of kernel gradients and kernel function;
Variable resolution particle schemes;
Higher order boundary conditions;
GPU-based implementation.

References:

[1] M. Leonardi and T. Rung. Explicit Strategies for Consistent Kernel Approximations. Proceeding of the 9th SPHERIC Workshop, Paris, 2014.

[2] C. Ulrich. Smoothed-Particle-Hydrodynamics Simulation of Port Hydrodynamic Problems. Hamburg University of Technology. Institute for Fluid Dynamics and Ship Theory. Schriftenreihe Schiffbau, Bericht Nr. 671, ISBN 978-3-89220-667-5, 2013.

[3]A.J.C. Crespo, J.M. Domínguez, A. Barreiro, M. Gómez-Gesteira and B.D. Rogers. GPUs, a new tool of acceleration in CFD: Efficiency and reliability on Smoothed Particle Hydrodynamics methods. PLoS ONE,6(6), e20685. doi:10.1371/journal.pone.0020685, 2011.