2017 activities poster available here

The 2017 ZCAM program is ready

The 2016 ZCAM program is now finished!

Reunión de Usuarios y Desarrolladores de Métodos de Simulación de Aragón

Multiscale Modeling of Simple and Complex Liquid Flow Using Particle-Continuum Hybrids

Rafael Delgado-Buscalioni, Universidad Autonoma de Madrid
Ignacio Pagonabarraga, Universidad de Barcelona

DATE OF EVENT : 05/10/2011       DURATION : 3 day(s)

LOCATION : ZCAM Campus Actur C/ Mariano Esquillor s/n Edificio I+D 50018 Zaragoza
URL : http://www.cecam.org/workshop-573.html


This workshop intends to merge researchers in applied science (engineering), theoretical modelling and applied mathematics to pose questions and share perspectives on the directions of future research lines in the field. Also, we would like to stimulate connections, collaborations and synergies between existing methodologies, developed by different (and scarcely overlapping) scientific communities.

We now briefly comment some of the questions the workshop aims to highlight.

Type A hybrids have experienced a significant advance in recent years and the state-of-the art is now flourishing. Some important problems have been already solved, such as hybrids allowing chemical specificity, hydrodynamic fluctuations, mass transfer, compressibility, and large length scale disparities. However, type A hybrids have been so far applied to rather academical problems (such as the contact line problem, cavity flows, tethered polymer dynamics...). There is a clear need to determine the pool of applications that really demand hybrid molecular-continuum approaches via domain decomposition. To this end it is quite important to discuss technical requirements coming from applied science simulations [17]. Some of these challenges (and applied problems) are

Energy exchange across the hybrid interface (hybrid description of melting or solidification fronts) [18,19] Insertion of large molecules [20] (grand canonical simulations of confined systems, films, sound-soft matter interaction, shear in polymeric fluids)Multispecies (osmotic effects) [21]Electrostatic interactions (micro-devices, nano-fluidics, bio-molecules).Hybrid treatment of liquid surfaces and membranes (surface tension, deformation of water interfaces, flow-membrane interactions).
On the computational side, the choice of the optimal hybrid model depends on the relevant scale under study: either molecular or macroscopic (mean flow). For instance, hydrodynamic fluctuations play an important role in describing how the atomic structure of a growing crystal is altered by the mean flow advection, or might be also relevant to understand how flow affects the structure and dynamics of a macromolecule. However for many CFD based research, fluctuations are only a nuisance that should be averaged out, because the main problem there is the effect of molecular singularities or defects on the mean flow [2,19]. The coupling strategy (either flux based [1] or state-coupling [2]) is sensitive to these facts. These computational aspects are necessary for the sake of developing a general purpose hybrid interface in molecular and fluid dynamics packages; this is in fact a work under construction for some open tools such as ESPRESSO [22] and OPENFOAM [17]. This work will require solving technical aspects such as:General formulation of molecular dynamics simulations in open systems, allowing for mass and energy exchanges in either flux- or state-coupling.Technical requirements to hybridize a molecular package (variable number of molecules, external forces, buffer density control)Flexible coupling geometries, moving hybrid boundariesParalelization aspects

Type B hybrids (constitutive molecular modeling) need to impose non-equilibrium states into many molecular boxes where the local stresses are measured. In this respect, type B approaches might benefit from the methodologies developed for type A hybrids. Generalized open boundary conditions for molecular dynamics [18] can be an alternative to the standard Lee-Edwards or SLLOD algorithms. Such a formulation can avoid periodic boundaries and work with open boxes allowing mass fluctuations and energy transfer in compressible flows, an issue which remain unsolved in Type B schemes. At the physical boundaries, complex fluids exhibit complex dynamics (such as slip-stick in polymer melts) and an alternative approach to account properly for them consists in using type A hybrids to solve the molecular detail of the boundary nodes of type B schemes [8]. Finally, space and time correlations are relevant for polymeric liquids and discussions on the applicability and limitations of existing algorithms are necessary. Specific, relevant, open issues include assessing the impact of current transmission between adjacent MD nodes to capture spatial correlations arising from molecular origin [8] as well as the limitations of multiple time coupling [23,6] to accelerate type B schemes assuming separation of microscopic and macroscopic time scales.

Type C methods group several schemes depending on the scale under consideration. At micron or submicron lengths, the Reynolds number is expected to be small and the force between the particles and the liquid are usually modeled via the Stokes friction term applied at a point representing the center of particle (point particle approximation). However, applications of polymeric and colloidal suspensions in micro-flow devices may require to apply relative large flow rates or involve velocity gradients of the same order of the particle radius, for which the Stokes friction limit is not valid. Another example of the limitation of the Stokes force is the treatment of ultrasound-matter interactions at micron or submicron scales, with important technological applications [24]. A way to avoid the Stokes friction assumption is to measure the particle-liquid force directly enforcing no-slip boundary condition at the particle surface. This idea is used by the Immersed Boundary method [12-14], originally developed to solve larger bodies embedded in a liquid. Recent works try to implement the boundary force coupling within the point particle approach, but still consistently keeping some information of the particle size. A proper interpolation between fluid nodes is essential for this sake. A similar problem is found in Lagrangian-Lagrangian schemes (SPH-MD) [16]. However, a prior requirement for consistency between all these approaches is to get some knowledge on the effective friction constant, whose value depends on the particle-fluid boundary condition (e.g. the no-slip assumption is not guaranteed in all cases or scales). Eulerian fluctuating hydrodynamics solvers are rapidly evolving [25,26]. Novel schemes allow to solve non-Newtonian liquids [26]; a step which has been recently solved in the Lagrangian counterpart (SPH) [16]. This opens a new route to treat colloidal suspensions in non-Newtonian solvents (mixtures of colloid and polymers in aqueous solution). However, generalizations of the boundary force or on the effective friction coefficient would be probably necessary for these complex liquids with memory and oscillatory damped stress time-correlations. To conclude, some other important issues concern comparisons and benchmarking between type C and type B hybrids and discussions on their range of applicability and parallelization aspects [27], with particular emphasis on GPU architectures.