Hydrodynamic Interactions in Particle Suspensions: A Perspective on Stokesian Dynamics

K.W. Torre, J. de Graaf · 2025

Read Hydrodynamic Interactions in Particle Suspensions: A Perspective on Stokesian Dynamics by K.W. Torre, J. de Graaf (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-16083.txt.

Introduction

Kim William Torre and Joost de Graaf review a simulation method for tiny particles in fluid in this 2025 paper. Such swarms show up in paints, foods, and living tissue. The method they review blends far-field flow rules with close-range glide corrections, trading some exactness for speed.

The pair lay out the math base, the strengths and weak spots, and the compute hurdles. Recent speed gains come from lean solvers that skip full matrix work. They warn that cheaper shortcuts miss key effects in dense swarms. Their fix trains a neural network on the full method's data to patch the shortcut. In a simple test the patch matched the full method while running faster.

Catalog record

CREATORK.W. Torre, J. de Graaf
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE74 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-16083

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Hydrodynamic Interactions in Particle Suspensions: A Perspective on Stokesian Dynamics K.W. Torre, J. de Graaf Abstract Stokesian Dynamics (SD) is a numerical framework used for simulating hydrodynamic interactions in particle suspensions at low Reynolds number. It combines far-field approximations with near-field lubrication corrections, offering a balance between accuracy and efficiency. This work reviews SD and provides a perspective on future directions for this approach. We outline the mathematical foundations, the method’s strengths and weaknesses, and the computational challenges that need to be overcome to work with SD effectively. We also discuss recent advancements that improve the algorithm’s efficiency, including the use of iterative solvers and matrix-free approaches. In addition, we highlight the limitations of making stronger, albeit more cost-effective approximations to studying hydrodynamic interactions in dense suspensions than made in SD, such as the two-body Rotne-Prager-Yamakawa (RPY) approximation. To overcome these issues, we propose a hybrid framework that replaces SD’s full many-body computations with a neural network trained on SD data. That is, we correct the RPY approximation, while avoiding costly matrix inversions. We demonstrate the potential of this method on a simple system, where we find a close match to SD data while algorithmically outperforming RPY. Our work provides an outlook on the way in which large-scale simulations of particle suspensions can be performed in the foreseeable future. I Introduction Particle suspensions — solid particles in a continuous liquid medium [ 1 , 2 ] — make up a large fraction of household products [ 3 ] , play a key role in industrial processes [ 4 ] , are ubiquitous in rheology and soft-matter research [ 5 ] , and are relevant for biological systems [ 6 ] . Examples include personal care products such as shampoos, deodorants, and toothpastes [ 5 ] , foodstuffs like milk, fruit juices, and soups [ 7 ] , as well as biological fluids such as blood [ 8 ] , mucus, and synovial fluid [ 6 ] . Typically, the particles in these examples can undergo (slight) deformations. However, for many systems of interest these remain below 1\% of the particle diameter, making them practically negligible [ 1 , 2 ] . Then, particles can be effectively treated as rigid solids without significantly affecting the accuracy of the hydrodynamic model. Abstracting further, it is useful to consider suspensions of spheres with equal diameter, i.e. , monodisperse hard spheres. The motion of particles relative to the fluid generates dissipative forces that significantly influence the overall dynamics of the suspension [ 2 ] . A moving particle not only experiences drag from the fluid but also disturbs the surrounding liquid flow. This disturbance propagates throughout the medium, influencing neighboring particles, which in turn modify the velocity field experienced by others. The fluid-mediated couplings between particles are referred to as hydrodynamic interactions (HIs). These are inherently long-ranged, with the flow disturbance caused by a single translating particle decaying inversely with interparticle distance. Even weakly disturbed fluid regions can exert noticeable effects on nearby particles, leading to correlated motion across the suspension. As a result, particles do not move independently but collectively. This many-body coupling—arising from the interplay between self-induced motion, external forces, and fluid-mediated interactions—is central to the complex collective behavior observed in particle suspensions. The characterization of HIs has been a longstanding challenge in theoretical and experimental physics, with foundational studies dating back nearly a century [ 1 ] . The starting point for understanding these interactions are the Navier-Stokes equations [ 2 ] . In particle suspensions, dissipation — proportional to the dynamic viscosity \mu — dominates over particle inertia — proportional to the liquid density \rho . This is because the small particle size {a} (typically microns) and speed \|\mathbf{U}\| (microns per second), makes the Reynolds number \mathrm{Re}=(\rho Ua)/\mu for the system very small. It is therefore justified to drop the nonlinear term from the Navier-Stokes equations, reducing the governing equations to the time-dependent Stokes equations. When the unsteady fluid response can also be disregarded, which applies to many soft-matter and granular systems [ 1 , 2 ] , the fluid response is governed by the Stokes equations. These are a set of linear, steady-state equations, which can be elegantly summarized in two conservation laws \displaystyle\bm{\nabla}\cdot\bm{\sigma} \displaystyle=0; (1a) \displaystyle\bm{\nabla}\cdot\mathbf{u} \displaystyle=0, (1b) for momentum and volume, respectively. Here, \mathbf{u} denotes the fluid velocity field, while the stress tensor is given by \displaystyle\sigma_{ij} \displaystyle=-p\delta_{ij}+\frac{\mu}{2}\left(\partial_{i}u_{j}+\partial_{j}u_{i}\right), (2) where i and j index the spatial coordinates, p is the hydrostatic pressure and \delta_{ij} the Kronecker delta. To obtain a well-posed problem, the fluid velocity is typically required to satisfy no-slip boundary conditions at the surface of each immersed particle, ensuring a unique solution [ 9 ] . Additionally, since solving the full boundary-value problem for arbitrarily shaped particles is intractable, the spherical-particle approximation is commonly adopted as a necessary simplification. While real-life suspensions often contain irregularly shaped particles [ 6 ] , assuming spheres allows for analytical and numerical treatments that capture key qualitative behaviors [ 10 , 11 , 12 ] . Even with these simplifications, exact analytical solutions remain limited, with fully resolved results largely restricted to one- and two-sphere problems [ 1 , 2 ] . For many-particle systems, which are challenging to describe analytically even making significant approximations, numerical approaches have been developed over the last decades to handle many-body HIs [ 13 ] . The Stokesian Dynamics (SD) method, introduced by Brady and Bossis [ 14 ] , is a numerical approach for simulating HIs in low Reynolds number systems, which makes use of fundamental solutions. By combining far-field Green’s functions with near-field lubrication corrections, SD efficiently models the many-body HIs between particles [ 15 ] , offering a good balance of computational efficiency and accuracy. However, despite its strengths, SD never saw widespread adoption, due to its overall algorithmic complexity and the computational expense of propagating the particle dynamics. These aspects made it less accessible compared to, for example, lattice-Boltzmann (LB) [ 16 , 17 , 18 , 19 , 20 ] , multi-particle collision dynamics (MPCD) [ 21 , 22 , 23 ] , and Fluid Particle Dynamics (FPD) [ 24 ] . These methods gained popularity due to their versatility in handling complex geometries, thermal fluctuations, and capturing inertial effects. Nevertheless, these approaches struggle to achieve the same level of precision as SD both in the far- and near-field [ 13 ] . Recent advancements [ 25 , 26 , 27 , 28 ] have addressed SD’s historical bottlenecks, significantly improving its computational performance. Iterative solvers, preconditioning strategies, and matrix-free techniques have also extended SD’s ability toward increasingly dense systems. Fiore and Swan’s Fast Stokesian Dynamics (FSD) [ 28 ] represents a major breakthrough, achieving linear scaling \mathcal{O}(N) by incorporating spectral Ewald methods and optimized matrix-free algorithms. Thus, SD has become a competitive option for simulating HIs, particularly in bulk fluids and systems with simple boundary conditions. In this paper, we provide a comprehensive review of Stokesian Dynamics, outlining its mathematical foundations, computational methods, and limitations. In Section II, we derive the equations governing HIs between particles, and introduce the mobility and resistance formalisms that are at the basis of SD. We also discuss far-field approximations using multipole

Read the complete text