Technische Universität München (Germany)
Local Project ID:
HPC Platform used:
SuperMUC of LRZ
A team of scientists from Germany, UK, US and Spain have developed a multiscale particle methods framework based on Smoothed Particle Hydrodynamics (SPH) and the stochastic Smoothed Dissipative Particle Dynamics (SDPD) to simulate the complex dynamics of submicron-sized colloidal and large non-colloidal particles suspended in Newtonian and non-Newtonian fluids.
Accurate modelling and simulation of particles suspended in simple and complex media is critical to predict collective flow behaviour and to forecast quickly the rheology of novel materials. Important industrial applications involve processing of filled plastics, ceramics, paints but also many of our daily-life products in cosmetics, pharmaceutics and especially in the growing areas of food- and bioengineering deal with particle suspensions. In the area of complex materials/fluids, understanding the mechanical response of particulate systems under flow is crucial for the design of novel products with optimized target properties. For example, it is well known that addition of nanoparticles to a simple fluid matrix can reduce its viscosity at large applied shear-rates, whereas keeping it high at low shear. This phenomenon, known as shear thinning, is typically used to functionalize colloidal fluids for paints which are required to flow easily under large stresses and to exhibit solid-like behaviour at rest. Under specific circumstances, an opposite viscosity increase under large deformation (shear thickening) can happen, which is seen often as an undesirable deteriorating effect when dealing with transport of concentrated suspensions under confinement, limiting pumping, coating and spraying operations at large flow rates .
A team of scientists from Germany, UK, US and Spain have developed a multiscale particle methods framework based on Smoothed Particle Hydrodynamics (SPH) and the stochastic Smoothed Dissipative Particle Dynamics (SDPD) to simulate the complex dynamics of submicron-sized colloidal and large non-colloidal particles suspended in Newtonian  and non-Newtonian fluids .
The SPH and SDPD based particle methods used in this project have been parallelized using optimized particle-particle-mesh (PPM) libraries developed at the ETH Zürich . This platform offers a fully parallelized environment which allows linear scalability up to thousands of CPUs. Access to supercomputers such as those available at the High Performance Computing of the Leibniz Supercomputing Centre (LRZ) allowed the researchers to perform direct numerical simulations of thousands of suspended particles retaining their precise morphology as well as all the relevant many-body features of the solvent hydrodynamics (Fig. 1).
The main target of this project is related to the simulation of the dynamics and rheology of multi-particle systems with particular emphasis on the physical understanding of the so-called shear thickening behaviour which typically takes place at very large solid particle concentrations. To reach these conditions, ability to perform stable simulations of nearly contacting solid particles is mandatory. This is, however, a numerical challenge due to the short-range diverging hydrodynamic behaviour of inter-particle lubrication forces, which poses a serious time-step limitation for standard explicit integration schemes. During this project, the scientists developed a novel splitting integration scheme for the dynamics of solid particles suspended in a Newtonian liquid, which allows to run simulations stably and efficiently up to very large solid volume fraction .
The novel technique has been applied to study the rheology of dense suspension of non- colloidal particles. Continuous hydrodynamic shear thickening was observed, whose strength was directly related to the microscopic distribution of hydrodynamic particle clusters . The investigation of the complex interplay between the suspension micro- structure, macroscopic suspension viscosity and specially the effect of external confinement on the suspension flow behaviour are key aspects of our research. Fig. 2 shows a snapshot of a typical simulation of a particulate system undergoing shear flow: a total number of 4,096,000 fluid particles (blue) is currently used to simulate ~4,000 suspended solid particles (grey) at a solid volume fraction 0.5.
 N. J. Wagner and J. Brady, Phys. Today, 62(10), 27 (2009).
 X. Bian, S. Litvinov, R. Qian, M. Ellero, N. A. Adams, Phys. Fluids, 24 , 012002 (2012).
 M. Ellero , A. Vazquez-Quesada, P. Espanol. Microfluidics and Nanofluidics 13, 249-260 (2012).
 I. Sbalzarini et al. J. Comp. Physics 215 (2), 566 –588 (2006).
 X. Bian, M. Ellero, Comp. Phys. Comm., 185, 53-62 (2014).
 X. Bian, S. Litvinov, M. Ellero, N. J. Wagner, J. Non- Newt. Fluid Mech. 213, 39-49 (2014).
LRZ's support for providing computing time on HPC system SuperMUC for this research project is gratefully acknowledged.
Assoc. Professor M. Ellero (principal investigator), Dr. A. Vazquez-Quesada, Dr. X. Bian.
Technische Universität München
Lehrstuhl für Aerodynamik und Strömungsmechanik
D-85748 Garching bei München/Germany.
Swansea University, Zienkiewicz Centre for Computational Engineering (ZCCE)
College of Engineering, Swansea University, Singleton Park, Swansea SA2 8PP, UK.
e-mail: Marco.Ellero@aer.mw.tum.de, M.Ellero@swansea.ac.uk