betaFeedback

WorksPhysics and Astronomy

A numerical framework for phoretic particles

Abstract

We develop a numerical a framework to study phoretic particle dynamics in two dimensions. The particles are modeled as chemically active rigid circles, which can emit or absorb a solute into surrounding fluid. The interaction between particles and solute induces a slip flow on particle surfaces, and the solute is advected by the fluid flow and diffuses with a constant diffusivity. The fluid-structure interaction is resolved by a boundary integral method accelerated by Ewald-like decomposition. The sharp resolution of moving boundaries for solute kinetics is performed thanks to an overlapping mesh method. The framework is validated separately for the Stokes problem and the advection–diffusion problem, reaching relatively high order of accuracy. Moreover, we employ the framework to more general problems, including particles in nearly infinite domain and straight channels, and multiparticle motions.

Introduction

In the past decades, phoretic particles (including rigid particles and droplets) have drawn increasing attention in experimental, theoretical, and numerical studies [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]. This problem involves an intimate coupling of fluid flow and solute transport with a moving boundary. More precisely, the fluid flow is induced by solute concentration gradient on particle surface, while the solute, besides diffusion and reactions, is advected by the flow [1, 2, 4]. In addition, the flow and solute concentration are affected by freely swimming particles and fixed boundaries [3, 6, 12, 13]. The interaction among multiple particles might lead to collective behaviors such as traveling lines, dynamic crystallites [7], caging [14], and rotating clusters [15]. In order to study these complex dynamics, a general numerical framework is needed. This constitutes the main motivation of the present work.

Numerical modeling of phoretic particles is not a trivial task, due to the nonlinear character of the solute advection by the fluid flows as well as to the presence of moving boundary conditions. Most numerical frameworks are designed for specific problems, which either deal with simple geometries or simplified models. For simple geometric configurations like the case of a single particle in an unconfined domain or the case of two-particle interactions, the full coupling of flow and solute field can be resolved using polar coordinates [1, 4, 5, 8, 10] or bi-polar coordinates [16, 17]. However, this does not apply to the case of more particles or to the case of a confined geometry, like in the presence of bounding walls (e.g. straight channels). To investigate collective behaviors for multiple particles, simplified models have been used. A simplification consisted of considering the particles as point sources [18], instead of being of finite size. Another simplification considered the case of rapid diffusion and rapid viscous transport [19, 20, 21]. In these simplified models, the hydrochemical coupling is discarded, and the solute transport reduces to either unsteady diffusion or Laplace problems. For this reason, some behaviors which result from hydrodynamic interactions, like particle attraction [12, 15], cannot be reproduced. Recently, some efforts have been directed towards accounting for more general geometries, larger numbers of particles, and nonlinearity of solute advection. By combining the embedded boundary method (EBM) with adaptive mesh refinement, the self-propulsion of an isotropic active particle in cylindrical pipes have been studied [22]. Others studies used the immersed boundary method (IBM) for both flow and solute field [23, 24], which transfers sharp boundary problems into diffuse boundary problems.

A full numerical framework for phoretic particles consists of two parts: the Stokes flow and the advection-diffusion problem for solute, which both contain moving boundaries. As particle motions are determined by solute distributions on their surfaces [1], the key point is to resolve boundary conditions on these moving boundaries. For the Stokes flow, the moving boundary problem can be solved by Green’s function, also know as boundary integral method (BIM) [25]. However, most boundary integral based methods are not appropriate for the present study due to following reasons. First, as the solute is advected by the flow, the boundary integral equation should be calculated on all sampling points in the fluid domain, not just on boundaries. Second, with the increase of particle numbers, the computational complexity scales as O(N2)O(N^2), where NN is proportional to the product of the number of particles and the number of sampling points on each particle. To overcome these limitations, several accelerated techniques have been developed, such as the particle-particle-particle-mesh (P3^3M) method [26], the particle-mesh-Ewald (PME) method [27, 28], and general geometry Ewald-like method (GGEM) [29, 30].

There are other numerical techniques for simulating fluid flow and advection-diffusion problems with moving boundaries. One candidate is the EBM or the cut-cell method (CCM), by which the Cartesian meshes are cut and reconstructed by moving boundaries [31, 32]. The reconstructed meshes are body-fitted in the vicinity of boundaries, which provide sharp resolution of boundaries and interfaces, while the remainder remains fixed Cartesian meshes. The overlapping mesh method (OMM) employs body-fitted meshes to several overlapping subdomains, and interpolates values on subdomain interfaces [33, 34]. Moreover, it allows these decomposed subdomains to move independently, suitable for moving boundary problems [35, 36, 37]. Another popular technique is the immersed boundary method (IBM) [38, 39, 40]. Boundary conditions are considered as source terms, and are spread to neighboring mesh points with a smoothed delta function, creating a smeared interface with finite thickness. To obtain sharp interfaces, the immersed interface method (IIM) applies interpolation or extrapolation to derive approximate boundary values or normal derivatives on neighboring mesh points [41, 42]. Similar ideas are applied to lattice Boltzmann method (LBM) as modified bounce-back boundary conditions [43, 44, 45]. Last but not least, particle-based methods are also used for studies of moving boundary problems, such as multiparticle collision dynamics (MPC) [46, 47], dissipative particle dynamics (DPD) [48, 49, 50], and smoothed particle hydrodynamics (SPH) [51, 52], to name but a few.

In preliminary studies, we first attempted to modify an IBM-based framework [53, 54] to model phoretic particles numerically. The Stokes flow was solved using the combined immersed boundary-lattice Boltzmann method (IBLBM) [55]. To account for low inertial effect, the Reynolds number was set to 0.05. The LBM with modified bounce-back boundary condition was employed for the advection and diffusion of solute [45]. Although this IBM-based framework could reproduce the symmetry breaking from stationary to straight motion for a circular phoretic particle, we found that the critical Péclet number PePe for straight motion was smaller (within 5 to 10 %) than the analytical finding [8]. Here PePe is the ratio of advection and diffusion. Moreover, we could obtain only qualitative agreement for meandering, circular, and chaotic motions. Further analysis showed that the IBM and the modified bounce-back boundary condition were only 1st-order accurate on the particle surface. In fact, quantitative disagreements have been observed in other studies [23, 24] when using IBM-based methods. For a circular phoretic particle, the critical PePe for straight motion was also smaller than the benchmark in Ref. [24], and meandering or circular motions were not reported in that work. For a spherical phoretic particle, chaotic motion was reported for Pe>15Pe > 15 in Ref. [23], while a more accurate result should be Pe≥24.2Pe \ge24.2 [56, 57]. These works indicate that a better alternative for the IBM should be proposed for quantitative studies of phoretic particles.

In this work, we develop a fast boundary integral method (FBIM) for arbitrary geometries, inspired by the GGEM [29]. The FBIM combines the advantages of BIM and IBM, being sufficiently efficient and accurate at the same time. For the advection-diffusion problem, we employ an OMM with second order accuracy in space. The subdomains attached to particles can move freely in the background fluid domain, resolving moving boundary conditions. The rest of this paper is organized as follows. In § 2, we first recall the governing equations for the motion of phoretic particles. Then we present the numerical implementation of the FBIM and the OMM in § 3 and § 4, respectively. The solution procedure is presented in § 5. In § 6, the FBIM and the OMM are validated separately, and convergence studies are performed. Lastly, the combined FBIM-OMM is employed to solve several general problems in § 7, and the effects of confined geometries and multiparticle interactions are explored.

Problem formulation

We consider a two-dimensional (2D) domain Ω\Omega of circular shaped particles of radius aa immersed in a Newtonian fluid of viscosity μ\mu. These particles are chemically active and can emit or absorb a solute into the surrounding fluid with an isotropic flux A\mathcal{A}. The solute diffuses with a molecular diffusivity D\mathcal{D} and is advected by the fluid flow. There is a chemical consumption/production proportional to the local solute concentration with coefficient B\mathcal{B}. The interaction of the solute molecules with the particles induces a slip tangential flow on the particle surfaces with a mobility M\mathcal{M} [4].

In the following, we adopt the non-dimensionalization of Ref. [1] where the length, fluid velocity, the solute concentration, and the pressure (or stress) are scaled by the characteristic values aa, ∣AM∣/D|\mathcal{A}\mathcal{M}|/\mathcal{D}, a∣A∣/Da|\mathcal{A}|/\mathcal{D}, and μ∣AM∣/aD\mu|\mathcal{A}\mathcal{M}|/a\mathcal{D}, respectively. The governing equations in dimensionless form are

∇p(x,t)=∇2u(x,t),∇⋅u(x,t)=0,(1)\nabla p(\boldsymbol{x},t)=\nabla^2\boldsymbol{u}(\boldsymbol{x},t),\quad\nabla\cdot\boldsymbol{u}(\boldsymbol{x},t)=0, \tag*{(1)}
∂c(x,t)∂t+u(x,t)⋅∇c(x,t)=1Pe∇2c(x,t)−βc(x,t),(2)\frac{\partial c(\boldsymbol{x},t)}{\partial t}+\boldsymbol{u}(\boldsymbol{x},t)\cdot\nabla c(\boldsymbol{x},t)=\frac{1}{Pe}\nabla^2c(\boldsymbol{x},t)-\beta c(\boldsymbol{x},t), \tag*{(2)}

where pp, u\boldsymbol{u} and cc are the dimensionless fluid pressure, velocity and solute concentration, Pe=a∣AM∣/D2Pe=a|\mathcal{A}\mathcal{M}|/\mathcal{D}^2 is the Péclet number which describes the ratio of advection and diffusion over the transport of solute, and β=aBD/∣AM∣\beta=aB\mathcal{D}/|\mathcal{A}\mathcal{M}| is the dimensionless consumption/production coefficient.

At the surface of a particle Γp\Gamma_p or a fixed boundary Γf\Gamma_f, the boundary conditions for the flow and the concentration field read

u(x,t)∣Γp=M∇sc(x,t)∣Γp+Up(t)+Ωp(t)×(x−Xp),u(x,t)∣Γf=0,(3)\left.\boldsymbol{u}(\boldsymbol{x},t)\right|_{\Gamma_p}=\mathcal{M}\nabla_s c(\boldsymbol{x},t)|_{\Gamma_p}+\boldsymbol{U}_p(t)+\boldsymbol{\Omega}_p(t)\times(\boldsymbol{x}-\boldsymbol{X}_p),\quad\left.\boldsymbol{u}(\boldsymbol{x},t)\right|_{\Gamma_f}=0, \tag*{(3)}
n⋅∇c(x,t)∣Γp=−A,n⋅∇c(x,t)∣Γf=0,(4)\left.\boldsymbol{n}\cdot\nabla c(\boldsymbol{x},t)\right|_{\Gamma_p}=-\mathcal{A},\quad\left.\boldsymbol{n}\cdot\nabla c(\boldsymbol{x},t)\right|_{\Gamma_f}=0, \tag*{(4)}

where Up\boldsymbol{U}_p and Ωp\boldsymbol{\Omega}_p are the translational and rotational velocities of the particle, Xp\boldsymbol{X}_p is the center of the particle, ∇s\nabla_s is the surface gradient operator, n\boldsymbol{n} is the unit outward normal, M=M/∣M∣\mathcal{M}=\mathcal{M}/|\mathcal{M}| is the dimensionless mobility and A=A/∣A∣\mathcal{A}=\mathcal{A}/|\mathcal{A}| is the dimensionless activity. In order to trigger self-propulsion in infinite fluid, A\mathcal{A} and M\mathcal{M} must have the same sign [1, 8]. The particle moves as a rigid body, and its center Xp\boldsymbol{X}_p and orientation angle Θp\Theta_p are given by

dXpdt=Up(t),dΘpdt=Ωp(t).(5)\frac{d\boldsymbol{X}_p}{dt}=\boldsymbol{U}_p(t),\quad\frac{d\Theta_p}{dt}=\boldsymbol{\Omega}_p(t). \tag*{(5)}

In the Stokes regime, there are two other constraints on each particle in the absence of external force: the force-free and torque-free conditions

∫Γpσ(x,t)⋅n dS(x)=0,∫Γp(x−Xp)×[σ(x,t)⋅n] dS(x)=0,(6)\int_{\Gamma_p}\boldsymbol{\sigma}(\boldsymbol{x},t)\cdot\boldsymbol{n}\,dS(\boldsymbol{x})=0,\quad\int_{\Gamma_p}(\boldsymbol{x}-\boldsymbol{X}_p)\times[\boldsymbol{\sigma}(\boldsymbol{x},t)\cdot\boldsymbol{n}]\,dS(\boldsymbol{x})=0, \tag*{(6)}

where σ=−pI+∇u+∇uT\boldsymbol{\sigma}=-p\boldsymbol{I}+\nabla\boldsymbol{u}+\nabla\boldsymbol{u}^{T} is the flow stress tensor, and SS is the curvilinear coordinate on the particle boundary Γp\Gamma_p.

Fast boundary integral method for fluid

The implementation of the FBIM is given as follows. We first recall the boundary integral equation, and then derive the analytical solution due to a point force based on Ewald-like decomposition. Later we apply the decomposed solution to calculate the single layer integral on boundaries. At last, a feedback forcing method is introduced to derive surface forces applied by rigid body on the fluid.

Boundary integral equation

The flow field is described by the Stokes equation (1) with Dirichlet boundary conditions (3). A general solution can be obtained by the boundary integral equation [25]

ui(x)=14π∑n=1Nb∫ΓnGij(x,X)Fj(X)dS(X),(7)u_i(\boldsymbol{x}) = \frac{1}{4\pi}\sum_{n=1}^{N_b}\int_{\Gamma_n}G_{ij}(\boldsymbol{x},\boldsymbol{X})F_j(\boldsymbol{X})dS(\boldsymbol{X}), \tag*{(7)}

where Γn\Gamma_n is the contour of the nnth boundary, X\boldsymbol{X} is a point on this boundary, GG is the 2D Green’s function, FF is the surface force applied by the boundary on the fluid, and the summations are over all the NbN_b boundaries. For an infinite domain, the free-space Green’s function [25] is used as

Gij(x,x′)=−δijln⁡(∣x−x′∣)+(xi−xi′)(xj−xj′)∣x−x′∣2.(8)G_{ij}(\boldsymbol{x},\boldsymbol{x}')=-\delta_{ij}\ln(|\boldsymbol{x}-\boldsymbol{x}'|)+\frac{(x_i-x_i')(x_j-x_j')}{|\boldsymbol{x}-\boldsymbol{x}'|^2}. \tag*{(8)}

The FBIM is derived based on Eqs. (7) and (8).

Decomposed solution due to a point force

We start from the simplest case, which is the flow due to a point force:

∇p(x)=∇2u(x)+gvδ(∣x−xv∣),∇⋅u(x)=0,(9)\nabla p(\boldsymbol{x})=\nabla^2\boldsymbol{u}(\boldsymbol{x})+\boldsymbol{g}^{v}\delta(|\boldsymbol{x}-\boldsymbol{x}^{v}|),\quad\nabla\cdot\boldsymbol{u}(\boldsymbol{x})=0, \tag*{(9)}

where gv\boldsymbol{g}^{v} is the force applied at a point xv\boldsymbol{x}^{v} (see Figure 1a). The resulted velocity is

Schematic of the point force and long-range density

Figure 1. (a) Schematic of the point force at xvx^v and the near neighbor region Ωc\Omega_c. (b) Long-range density DlD^l [see Eq. (17)] as the function of rr. Here the cutoff distance is set as rc=4r_c = 4.

ui(x)=14πGij(x,xv)gjv.(10)u_i(\boldsymbol{x})=\frac{1}{4\pi}G_{ij}(\boldsymbol{x},\boldsymbol{x}^{v})g_j^{v}. \tag*{(10)}

Inspired by other Ewald-summation-based methods [27, 28, 29], the idea of the FBIM is to split the Dirac-delta density δ(r)\delta(r) in Eq. (9) into a sum of a short-range density Ds(r)D^{s}(r) containing the singularity, and a sufficiently smooth long-range density Dl(r)=δ(r)−Ds(r)D^{l}(r)=\delta(r)-D^{s}(r). Here rr refers to the magnitude of the vector from any position to the pole of singularity, r=x−xvr=\boldsymbol{x}-\boldsymbol{x}^{v}. The short-range density is chosen in such a way that for any point beyond a certain cutoff distance rcr_c (i.e. r>rcr>r_c), the flow velocity due to the short-range force density gvDs(r)\boldsymbol{g}^{v}D^{s}(r) is zero.

The solution of the Stokes and continuity equation is also decomposed into two parts. For the long-range part, the solution ul(x)\boldsymbol{u}^{l}(\boldsymbol{x}) and pl(x)p^{l}(\boldsymbol{x}) due to the long-range force density satisfies the following set of equations

∇pl(x)=∇2ul(x)+gvDl(∣x−xv∣),∇⋅ul(x)=0.(11)\nabla p^{l}(\boldsymbol{x})=\nabla^2\boldsymbol{u}^{l}(\boldsymbol{x})+\boldsymbol{g}^{v}D^{l}(|\boldsymbol{x}-\boldsymbol{x}^{v}|),\quad\nabla\cdot\boldsymbol{u}^{l}(\boldsymbol{x})=0. \tag*{(11)}

This equation can be solved by any Stokes solver, and we will use a Fourier spectral method in this study. The short-range solution us(x)\boldsymbol{u}^{s}(\boldsymbol{x}) and ps(x)p^{s}(\boldsymbol{x}) satisfies the short-range set of equations

∇ps(x)=∇2us(x)+gvDs(∣x−xv∣),∇⋅us(x)=0.(12)\nabla p^{s}(\boldsymbol{x})=\nabla^2\boldsymbol{u}^{s}(\boldsymbol{x})+\boldsymbol{g}^{v}D^{s}(|\boldsymbol{x}-\boldsymbol{x}^{v}|),\quad\nabla\cdot\boldsymbol{u}^{s}(\boldsymbol{x})=0. \tag*{(12)}

This equation will be solved by using a short-range Green’s function for any points within the cutoff distance (r≤rcr\le r_c). Once the two problems are solved, the overall solution is obtained as

u(x)=ul(x)+us(x),p(x)=pl(x)+ps(x).(13)\boldsymbol{u}(\boldsymbol{x})=\boldsymbol{u}^{l}(\boldsymbol{x})+\boldsymbol{u}^{s}(\boldsymbol{x}),p(\boldsymbol{x})=p^{l}(\boldsymbol{x})+p^{s}(\boldsymbol{x}). \tag*{(13)}

Long-range solution

To solve the long-range problem, a proper long-range density Dl(r)D^l(r) should be chosen. The first constraint on Dl(r)D^l(r) is the finite cutoff distance rcr_c, such that

Dl(r)=0,∀r≥rc.(14)D^l(r) = 0, \forall r \ge r_c. \tag*{(14)}

Despite this, in order to have a smooth solution of ul(x)u^l(x), its derivatives should be continuous at r=rcr = r_c as

∂kDl∂rk∣r=rc=0,k=1,2,3,…,kmax⁡(15)\left.\frac{\partial^k D^l}{\partial r^k}\right|_{r=r_c}=0, k = 1, 2, 3, \ldots, k_{\max} \tag*{(15)}

where kmax⁡k_{\max} is the cut-off order of the derivatives.

The second constraint is related to the solution induced by the point force. As follows from the above, the solution driven by the long-range force density is the same as that driven by the point force for any point beyond the cutoff distance. Define Ωc\Omega_c as the near neighbor region around the singular point xvx^v within the cutoff distance rcr_c (see Figure 1a), then the following equation is valid for any point x∈Ω∖Ωcx \in\Omega\setminus\Omega_c:

uil(x)=14πGij(x,xv)gjv=14π∫ΩcGij(x,x′)gjvDl(∣x′−xv∣) dx′.(16)u_i^l(x) = \frac{1}{4\pi}G_{ij}(x,x^v)g_j^v = \frac{1}{4\pi}\int_{\Omega_c}G_{ij}(x,x')g_j^vD^l(\lvert x'-x^v\rvert)\,dx'. \tag*{(16)}

We found a 6th-order polynomial which satisfies the above constraints as (see Figure 1b)

Dl(r)={0,r≥rc,563πrc2[−252(rrc)6+54(rrc)5−90(rrc)4+70(rrc)3−452(rrc)2+1],r<rc.(17)D^l(r) = \begin{cases} 0, & r \ge r_c, \\ \frac{56}{3\pi r_c^2}\left[-\frac{25}{2}\left(\frac{r}{r_c}\right)^6+54\left(\frac{r}{r_c}\right)^5-90\left(\frac{r}{r_c}\right)^4+70\left(\frac{r}{r_c}\right)^3-\frac{45}{2}\left(\frac{r}{r_c}\right)^2+1\right], & r < r_c. \end{cases} \tag*{(17)}

The detailed derivation is given in Appendix § 10. Actually, it turns out that an expansion of Dl(r)D^l(r) as higher-order polynomial with only even degree terms leads to higher order accuracy. It should be noted that the choice of long-range density is not unique. Any functions that satisfy Eqs. (14)–(16) would be valid. Moreover, the long-range density DlD^l is strictly equal to 0 for any point beyond the cutoff distance. This is different from the screening function in GGEM [29], which is quasi-Gaussian and decays exponentially to 0.

To solve Eq. (11), the discrete Fourier series approximation is employed in the periodic xx and yy directions. The two components of the long-range velocity ul=(uxl,uyl)\mathbf{u}^{l}=(u^{l}_{x},u^{l}_{y}) is expressed as

uxl(x)+iuyl(x)=∑m=−Nx/2Nx/2∑n=−Ny/2Ny/2u^mnlei2πmx/Lxei2πny/Ly,(18)u^{l}_{x}(\mathbf{x})+iu^{l}_{y}(\mathbf{x})=\sum_{m=-N_{x}/2}^{N_{x}/2}\sum_{n=-N_{y}/2}^{N_{y}/2}\hat{u}^{l}_{mn}e^{i2\pi mx/L_{x}}e^{i2\pi ny/L_{y}}, \tag*{(18)}

where NxN_{x} and NyN_{y} are the number of modes in the corresponding series approximation, and u^mnl\hat{u}^{l}_{mn} is the complex amplitude of the harmonic. Similarly, the long-range force density fl(x)=gvDl(∣x−xv∣)\mathbf{f}^{l}(\mathbf{x})=\mathbf{g}^{v}D^{l}(|\mathbf{x}-\mathbf{x}^{v}|) is also represented by discrete Fourier series as

fxl(x)+ifyl(x)=∑m=−Nx/2Nx/2∑n=−Ny/2Ny/2f^mnlei2πmx/Lxei2πny/Ly,(19)f^{l}_{x}(\mathbf{x})+if^{l}_{y}(\mathbf{x})=\sum_{m=-N_{x}/2}^{N_{x}/2}\sum_{n=-N_{y}/2}^{N_{y}/2}\hat{f}^{l}_{mn}e^{i2\pi mx/L_{x}}e^{i2\pi ny/L_{y}}, \tag*{(19)}

where f^mnl\hat{f}^{l}_{mn} is the complex amplitude of the harmonic. The computational domain is a rectangle, and is discretized by means of equally spaced Cartesian mesh. The number of mesh points in xx and yy directions are set equal to the number of Fourier modes NxN_{x} and NyN_{y} respectively, so that the fast Fourier transform (FFT) can be used to transfer data between real and Fourier space. The FFT and reverse FFT are calculated by the cuFFT library [58], which is accelerated by GPU. In practice, Eq. (11) is first transformed into Fourier space using FFT. Then the harmonic’s amplitude u^mnl\hat{u}^{l}_{mn} is calculated as

u^mnl=12(m2+n2)(f^mnl−m2+i2mn−n2m2+n2f^−m−nl∗),mn≠0,(20)\hat{u}^{l}_{mn}=\frac{1}{2(m^{2}+n^{2})}\left(\hat{f}^{l}_{mn}-\frac{m^{2}+i2mn-n^{2}}{m^{2}+n^{2}}\hat{f}^{l*}_{-m-n}\right),\quad mn\ne0, \tag*{(20)}

where f^−m−nl∗\hat{f}^{l*}_{-m-n} is the complex conjugate of f^−m−nl\hat{f}^{l}_{-m-n}. The 0th mode u^00l\hat{u}^{l}_{00} corresponds to the average velocity in the computational domain, which is zero, unless there is an imposed flow (not considered here). Finally, a reverse FFT is employed to compute the long-range velocity ul\mathbf{u}^{l} in real space. For any point not lying on the mesh, the velocity is obtained by bi-cubic interpolation with fourth order accuracy in space.

Short-range solution

The short-range density is

Ds(r)=δ(r)−Dl(r).(21)D^{s}(r)=\delta(r)-D^{l}(r). \tag*{(21)}

The short-range velocity is then obtained as

uis(x)=14π∫ΩcGij(x,x′)gjvDs(∣x′−xv∣)dx′.(22)u^{s}_{i}(\mathbf{x})=\frac{1}{4\pi}\int_{\Omega_{c}}G_{ij}(\mathbf{x},\mathbf{x}^{\prime})g^{v}_{j}D^{s}(|\mathbf{x}^{\prime}-\mathbf{x}^{v}|)d\mathbf{x}^{\prime}. \tag*{(22)}

According to Eqs. (17) and (21), uis(x)=0u^{s}_{i}(\mathbf{x})=0 for any point x∈Ω∖Ωc\mathbf{x}\in\Omega\setminus\Omega_{c}. Hence Eq. (22) is solved for the points within the cutoff distance (i.e. x∈Ωc\mathbf{x}\in\Omega_{c}) only. We define the short-range Green’s function as

Gijs(x,xv)=∫ΩcGij(x,x′)Ds(∣x′−xv∣)dx′.(23)G^{s}_{ij}(\mathbf{x},\mathbf{x}^{v})=\int_{\Omega_{c}}G_{ij}(\mathbf{x},\mathbf{x}^{\prime})D^{s}(|\mathbf{x}^{\prime}-\mathbf{x}^{v}|)d\mathbf{x}^{\prime}. \tag*{(23)}

Then the short-range velocity reads

uis(x)=14πgjGGijGs(x,xv).(24)u_i^s(\mathbf{x})=\frac{1}{4\pi}g_j^G G_{ij}^{Gs}(\mathbf{x},\mathbf{x}^v). \tag*{(24)}

The derivation of the short-range Green’s function is given in Appendix § 11.

Single layer integral

The velocity due to a boundary Γ\Gamma is given by the single layer integral according to Eq. (7). Applying the same decomposition as § 3.2, we rewrite this equation in the following form

ui(x)=14π∫ΓGij(x,X)[∫ΩFj(X)δ(x′−X)dx′]dS(X)=14π∫Γ{∫ΩGij(x,x′)Fj(X)[Dl(∣x′−X∣)+Ds(∣x′−X∣)]dx′}dS(X).(25)\begin{aligned} u_i(\mathbf{x})=\frac{1}{4\pi}\int_{\Gamma}G_{ij}(\mathbf{x},\mathbf{X})\left[\int_{\Omega}F_j(\mathbf{X})\delta(\mathbf{x}'-\mathbf{X})d\mathbf{x}'\right]dS(\mathbf{X}) \\ &=\frac{1}{4\pi}\int_{\Gamma}\left\{\int_{\Omega}G_{ij}(\mathbf{x},\mathbf{x}')F_j(\mathbf{X})\left[D^l(|\mathbf{x}'-\mathbf{X}|)+D^s(|\mathbf{x}'-\mathbf{X}|)\right]d\mathbf{x}'\right\}dS(\mathbf{X}). \tag*{(25)} \end{aligned}

The contribution due to the long-range density is

uil(x)=14π∫ΩGij(x,x′)[∫ΓFj(X)Dl(∣x′−X∣)dS(X)]dx′,(26)u_i^l(\mathbf{x})=\frac{1}{4\pi}\int_{\Omega}G_{ij}(\mathbf{x},\mathbf{x}')\left[\int_{\Gamma}F_j(\mathbf{X})D^l(|\mathbf{x}'-\mathbf{X}|)dS(\mathbf{X})\right]d\mathbf{x}', \tag*{(26)}

and the long-range force density is

fl(x)=∫ΓF(X)Dl(∣x−X∣)dS(X).(27)f^l(\mathbf{x})=\int_{\Gamma}\mathbf{F}(\mathbf{X})D^l(|\mathbf{x}-\mathbf{X}|)dS(\mathbf{X}). \tag*{(27)}

Next, the contribution due to the short-range density is

uis(x)=14π∫ΓFj(X)[∫ΩGij(x,x′)Ds(∣x′−X∣)dx′]dS(X)=14π∫ΓFj(X)Gijs(x,X)dS(X).(28)\begin{aligned} u_i^s(\mathbf{x})=\frac{1}{4\pi}\int_{\Gamma}F_j(\mathbf{X})\left[\int_{\Omega}G_{ij}(\mathbf{x},\mathbf{x}')D^s(|\mathbf{x}'-\mathbf{X}|)d\mathbf{x}'\right]dS(\mathbf{X}) \\ &=\frac{1}{4\pi}\int_{\Gamma}F_j(\mathbf{X})G_{ij}^{s}(\mathbf{x},\mathbf{X})dS(\mathbf{X}). \tag*{(28)} \end{aligned}

Here we describe the numerical implementation of the above problems. The surface Γ\Gamma is discretized into NeN_e elements. For the long-range problem, we first need to know the long-range force density fl\mathbf{f}^l by computing the integral in Eq. (27). Its discretized form is written as

fl(x)=∑k=1Ne∫ΓkF(X)Dl(∣x−X∣)dS(X),(29)\mathbf{f}^l(\mathbf{x})=\sum_{k=1}^{N_e}\int_{\Gamma_k}\mathbf{F}(\mathbf{X})D^l(|\mathbf{x}-\mathbf{X}|)dS(\mathbf{X}), \tag*{(29)}

where Γk\Gamma_k denotes the kkth element. The above integral can be evaluated by a simple trapezoidal rule. Then the long-range velocity ul\mathbf{u}^l is solved for as described in § 3.2. For the shor-range problem, the boundary integral in Eq. (28) is discretized as

us(x)=14π∑Γk∈Ωc∫ΓkF(X)Gs(x,X)dS(X).(30)\mathbf{u}^s(\mathbf{x})=\frac{1}{4\pi}\sum_{\Gamma_k\in\Omega_c}\int_{\Gamma_k}\mathbf{F}(\mathbf{X})G^s(\mathbf{x},\mathbf{X})dS(\mathbf{X}). \tag*{(30)}

As the integrand in Eq. (30) includes logarithmic singularity at boundaries x=X\mathbf{x}=\mathbf{X}, a special quadrature rule [59] is used to ensure sufficient accuracy. It should be noted that the summation in Eq. (30) is calculated over elements within the near neighbor region of x\mathbf{x}, i.e. Γk∈Ωc\Gamma_k\in\Omega_c. The contribution of other elements has been included in the long-range solution.

Rigid body dynamics

In the present study, particles and geometric boundaries are considered as rigid bodies. The surface force F\mathbf{F} on these boundaries can be calculated by considering them as elastic bodies with a large elastic modulus. An alternative method is to apply a feedback force [38] as

F(X,t)=∫0tα[V(X,t′)−u(X,t′)]dt′,(31)\mathbf{F}(\mathbf{X},t)=\int_{0}^{t}\alpha\left[\mathbf{V}(\mathbf{X},t^{\prime})-\mathbf{u}(\mathbf{X},t^{\prime})\right]dt^{\prime}, \tag*{(31)}

so that the desired velocity V\mathbf{V} is enforced on the boundary. Here α\alpha is a positive coefficient. In numerical application, the time integral in Eq. (32) is computed in an iterative form as

F(X,t+Δt)=F(X,t)+α[V(X,t)−u(X,t)]Δt,(32)\mathbf{F}(\mathbf{X},t+\Delta t)=\mathbf{F}(\mathbf{X},t)+\alpha\left[\mathbf{V}(\mathbf{X},t)-\mathbf{u}(\mathbf{X},t)\right]\Delta t, \tag*{(32)}

where Δt\Delta t is the time step.

The desired velocity depends on the boundary conditions (3). For a fixed boundary with no-slip boundary condition, V=0\mathbf{V}=\mathbf{0}. For a freely-moving circular phoretic particle, the desired velocity is the summation of the slip velocity, rigid body translation, and rotation

V(X)=M∇sc(X)+Up+Ωp×(X−Xp).(33)\mathbf{V}(\mathbf{X})=M\nabla_{\mathrm{s}}c(\mathbf{X})+\mathbf{U}_{p}+\boldsymbol{\Omega}_{p}\times(\mathbf{X}-\mathbf{X}_{p}). \tag*{(33)}

As the surface force F\mathbf{F} equals to the hydrodynamic traction σ⋅n\boldsymbol{\sigma}\cdot\mathbf{n} on boundaries, the force-free and torque-free conditions in Eq. (6) read

∫ΓpF(X)dS(X)=0,∫Γp(X−Xp)×F(X)dS(X)=0.(34)\int_{\Gamma_{p}}\mathbf{F}(\mathbf{X})dS(\mathbf{X})=\mathbf{0},\qquad\int_{\Gamma_{p}}(\mathbf{X}-\mathbf{X}_{p})\times\mathbf{F}(\mathbf{X})dS(\mathbf{X})=\mathbf{0}. \tag*{(34)}

The 4 variables F\mathbf{F}, V\mathbf{V}, Up\mathbf{U}_{p}, and Ωp\boldsymbol{\Omega}_{p} can be obtained by solving the system of Eqs. (32), (33) and (34). Then the motion of each particle is integrated in time by a forward Euler scheme as

Xp(t+Δt)=Xp(t)+ΔtUp(t),Θp(t+Δt)=Θp(t)+ΔtΩp(t).(35)\mathbf{X}_{p}(t+\Delta t)=\mathbf{X}_{p}(t)+\Delta t\mathbf{U}_{p}(t),\quad\boldsymbol{\Theta}_{p}(t+\Delta t)=\boldsymbol{\Theta}_{p}(t)+\Delta t\boldsymbol{\Omega}_{p}(t). \tag*{(35)}

For circular particles, pre-conditioning is applied for better convergence. In a polar coordinate system (1,θ)(1,\theta) attached to the particle center Xp\mathbf{X}_{p}, the surface force F[X(θ)]\mathbf{F}[\mathbf{X}(\theta)], the surface velocity u[X(θ)]\mathbf{u}[\mathbf{X}(\theta)], and the surface concentration c[X(θ)]c[\mathbf{X}(\theta)] are all periodic, and can be expressed by Fourier series as

Fx(θ,t)+iFy(θ,t)=∑k=−∞∞F^k(t)eikθ,(36)F_{x}(\theta,t)+iF_{y}(\theta,t)=\sum_{k=-\infty}^{\infty}\widehat{F}_{k}(t)e^{ik\theta}, \tag*{(36)}
ux(θ,t)+iuy(θ,t)=∑k=−∞∞u^k(t)eikθ,(37)u_{x}(\theta,t)+iu_{y}(\theta,t)=\sum_{k=-\infty}^{\infty}\widehat{u}_{k}(t)e^{ik\theta}, \tag*{(37)}
c(θ,t)=∑k=−∞∞c^k(t)eikθ,(38)c(\theta,t)=\sum_{k=-\infty}^{\infty}\widehat{c}_{k}(t)e^{ik\theta}, \tag*{(38)}

where F^k\hat{F}_k, u^k\hat{u}_k, and c^k\hat{c}_k are the complex amplitudes. Applying force-free, torque-free, and velocity divergence-free conditions, one finds

F^0=0,Im⁡(F^1)=0,Re⁡(u^1)=0,(39)\hat{F}_0 = 0,\operatorname{Im}(\hat{F}_1) = 0,\operatorname{Re}(\hat{u}_1) = 0, \tag*{(39)}

where Re⁡\operatorname{Re} and Im⁡\operatorname{Im} denote real and imaginary parts, respectively. The value of Re⁡(F^1)\operatorname{Re}(\hat{F}_1) does not affect the flow field, and is set to 0. From time tt to t+Δtt+\Delta t, a small change of surface force is applied, inducing a change of surface velocity as

α^[V(θ,t)−u(θ,t)]=∫02πG(θ,ϕ)[F(ϕ,t+Δt)−F(ϕ,t)] dϕ.(40)\hat{\alpha}[\mathbf{V}(\theta,t)-\mathbf{u}(\theta,t)] = \int_{0}^{2\pi} \mathbf{G}(\theta,\phi)[\mathbf{F}(\phi,t+\Delta t)-\mathbf{F}(\phi,t)]\,\mathrm{d}\phi. \tag*{(40)}

Here α^\hat{\alpha} is a coefficient between 0 to 1. This equation can be rewritten using Fourier series as

Upx(t)+iUpy(t)+i∣Ωp(t)∣eiθ+∑k=−∞∞[M(1−k)c^k−1(t)−u^k(t)]eikθ=∑k≠0,1F^k(t+Δt)−F^k(t)4α^∣k∣eikθ.(41)U_{p_x}(t)+\mathrm{i}U_{p_y}(t)+\mathrm{i}|\boldsymbol{\Omega}_p(t)|e^{\mathrm{i}\theta}+\sum_{k=-\infty}^{\infty}[M(1-k)\hat{c}_{k-1}(t)-\hat{u}_k(t)]e^{\mathrm{i}k\theta}=\sum_{k\ne0,1}\frac{\hat{F}_k(t+\Delta t)-\hat{F}_k(t)}{4\hat{\alpha}|k|}e^{\mathrm{i}k\theta}. \tag*{(41)}

Solving the equation for the kkth Fourier harmonics, we obtain

Up(t)=[Re⁡[u^0(t)]−MRe⁡[c^1(t)]Im⁡[u^0(t)]+MIm⁡[c^1(t)]],(42)\mathbf{U}_p(t)=\begin{bmatrix}\operatorname{Re}[\hat{u}_0(t)]-M\operatorname{Re}[\hat{c}_1(t)]\\ \operatorname{Im}[\hat{u}_0(t)]+M\operatorname{Im}[\hat{c}_1(t)]\end{bmatrix}, \tag*{(42)}
∣Ωp(t)∣=Im⁡[u^1(t)],(43)|\boldsymbol{\Omega}_p(t)|=\operatorname{Im}[\hat{u}_1(t)], \tag*{(43)}
F^k(t+Δt)=F^k(t)+4α^∣k∣[M(1−k)c^k−1(t)−u^k(t)],k≠0,1.(44)\hat{F}_k(t+\Delta t)=\hat{F}_k(t)+4\hat{\alpha}|k|[M(1-k)\hat{c}_{k-1}(t)-\hat{u}_k(t)],\quad k\ne0,1. \tag*{(44)}

The surface force F(θ,t+Δt)\mathbf{F}(\theta,t+\Delta t) can then be calculated using Fourier series.

To ensure long-term stability of simulations, we perform a small correction of particle position when a particle approaches other particles or fixed boundaries. The correction is introduced thanks to the Morse potential [60]

Φ(d)=De[e2γ(d0−d)−2eγ(d0−d)],(45)\Phi(d)=D_e[e^{2\gamma(d_0-d)}-2e^{\gamma(d_0-d)}], \tag*{(45)}

where DeD_e is the surface energy, γ\gamma is the scaling factor, dd is the distance between two boundaries, and d0d_0 is the zero force distance, set to 3Δx3\Delta x for all simulations in the present work. Consider the mmth and the nnth particles located at Xpm\mathbf{X}_{p_m} and Xpn\mathbf{X}_{p_n}. The distance between the two particle surfaces is dmn=rmn−2d_{mn}=r_{mn}-2, where rmn=Xpm−Xpn\mathbf{r}_{mn}=\mathbf{X}_{p_m}-\mathbf{X}_{p_n}. The correction for the mmth particle due to the nnth particle is

Xmnc={0,dmn>d0,−Δt∂Φ(dmn)∂dmnrmnrmn,dmn≤d0.(46)\mathbf{X}^{c}_{mn}= \begin{cases} 0, & d_{mn}>d_0,\\ -\Delta t\frac{\partial\Phi(d_{mn})}{\partial d_{mn}}\frac{\mathbf{r}_{mn}}{r_{mn}}, & d_{mn}\le d_0. \end{cases} \tag*{(46)}

For the mmth particle close to the nnth fixed boundary, the distance dmnd_{mn} is the minimum distance between the particle surface and the fixed boundary, and rmn\mathbf{r}_{mn} is the normal vector of the fixed boundary pointing into the particle center. The correction is then calculated using Eq. (46). The position of the mmth particle after correction is

Xpm(t+Δt)=Xpm(t)+ΔtUpm(t)+∑m≠nXmnc(t).(47)\mathbf{X}_{p_m}(t+\Delta t)=\mathbf{X}_{p_m}(t)+\Delta t\mathbf{U}_{p_m}(t)+\sum_{m\ne n}\mathbf{X}^{c}_{mn}(t). \tag*{(47)}

Overlapping mesh method for solute

The detailed implementation of the OMM is given in this section. First we decompose the computational domain into two overlapping subdomains. Then the advection-diffusion problem is discretized in each subdomain. Last but not least, the treatment of moving boundary is introduced.

Domain decomposition

The main idea of the OMM is to divide a computational domain into several overlapping subdomains [33]. For the present work, we consider a circular particle immersed in a rectangular domain. The global solution domain Ω\Omega is decomposed into two overlapping subdomains, the fluid subdomain Ωf\Omega_f and the particle subdomain Ωp\Omega_p, so that Ω=Ωf∪Ωp\Omega= \Omega_f \cup\Omega_p, as shown in Figure 2a. For an annular subdomain like Ωp\Omega_p, the difference of radius between the inner and outer circles is defined as the subdomain size rpr_p. Boundary conditions (4) are employed on global boundaries (solid lines in Figure 2a). Solute concentration at interface boundaries Γfp\Gamma_{fp} and Γpf\Gamma_{pf} (dashed lines in Figure 2a) is equal to contiguous value in adjacent subdomains, leading to additional Dirichlet boundary conditions [as shown in Eqs. (51) and (54)]. As the particle swims, the subdomain Ωp\Omega_p moves with the particle, so do the interface boundaries. Here we use polar coordinates (r,θ)(r,\theta) in Ωp\Omega_p, where the origin is set at the particle center Xp\mathbf{X}_p. The mapping between two subdomains is written as

Schematics of the OMM

Figure 2. Schematics of the OMM. (a) Domain decomposition. (b) Polar mesh in the particle subdomain Ωp\Omega_p (c) Cartesian mesh in the fluid subdomain Ωf\Omega_f (d) Overlapped polar mesh and Cartesian mesh. For an annular subdomain, the difference of radius between the inner and outer circles is defined as the subdomain size rpr_p. The solid lines represent global boundaries, and the dashed lines represent interface boundaries.

r=x−Xp=r[cos⁡(θ+Θp)sin⁡(θ+Θp)](48)\mathbf{r}=\mathbf{x}-\mathbf{X}_p=r\begin{bmatrix}\cos(\theta+\Theta_p)\\\sin(\theta+\Theta_p)\end{bmatrix} \tag*{(48)}

where Θp\Theta_p is the orientation angle of the particle. The governing equations with boundary conditions in each subdomain read

∂c(x,t)∂t+u(x,t)⋅∇c(x,t)=1Pe∇2c(x,t)−βc(x,t),x∈Ωf(49)\frac{\partial c(\mathbf{x},t)}{\partial t}+\mathbf{u}(\mathbf{x},t)\cdot\nabla c(\mathbf{x},t)=\frac{1}{Pe}\nabla^2c(\mathbf{x},t)-\beta c(\mathbf{x},t),\quad\mathbf{x}\in\Omega_f \tag*{(49)}
n⋅∇c(x,t)∣Γf=0(50)\left.\mathbf{n}\cdot\nabla c(\mathbf{x},t)\right|_{\Gamma_f}=0 \tag*{(50)}
c(x,t)∣Γfp=c(r,t)∣Γfp∩Ωp(51)\left.c(\mathbf{x},t)\right|_{\Gamma_{fp}}=\left.c(\mathbf{r},t)\right|_{\Gamma_{fp}\cap\Omega_p} \tag*{(51)}

and

∂c(r,t)∂t+up(r,t)⋅∇c(r,t)=1Pe∇2c(r,t)−βc(r,t),r∈Ωp(52)\frac{\partial c(\mathbf{r},t)}{\partial t}+\mathbf{u}_p(\mathbf{r},t)\cdot\nabla c(\mathbf{r},t)=\frac{1}{Pe}\nabla^2c(\mathbf{r},t)-\beta c(\mathbf{r},t),\quad\mathbf{r}\in\Omega_p \tag*{(52)}
n⋅∇c(r,t)∣Γp=−A(53)\left.\mathbf{n}\cdot\nabla c(\mathbf{r},t)\right|_{\Gamma_p}=-A \tag*{(53)}
c(r,t)∣Γpf=c(x,t)∣Γpf∩Ωf(54)\left.c(\mathbf{r},t)\right|_{\Gamma_{pf}}=\left.c(\mathbf{x},t)\right|_{\Gamma_{pf}\cap\Omega_f} \tag*{(54)}

where up=u−Up−Ωp×r\mathbf{u}_p=\mathbf{u}-\mathbf{U}_p-\boldsymbol{\Omega}_p\times\mathbf{r} is the relative velocity in the particle subdomain Ωp\Omega_p.

Spatial discretization

The two subdomains are discretized separately. The particle subdomain Ωp\Omega_p is discretized with polar mesh, and the outermost mesh points from the interface boundary Γpf\Gamma_{pf} (see Figure 2b). The fluid subdomain Ωf\Omega_f is discretized with Cartesian mesh, and its interface boundary Γfp\Gamma_{fp} is set as the Cartesian mesh points around the particle (see Figure 2c). A schematic of overlapping meshes is shown in Figure 2d. As the particle swims in the fluid, the polar mesh translates and rotates with the particle, while the Cartesian mesh remains fixed. It should be noted that the particle interface boundaries may evolve due to particle movement, as shown in Figure 3.

Schematics of interface boundaries

Figure 3. Schematics of interface boundaries. (a) A particle close to a global boundary. (b) Two particles close to each other. The solid lines represent global boundaries, and the dashed lines represent interface boundaries.

In each subdomain, the finite difference method is employed to solve the advection-diffusion problem numerically. The second-order centered difference is used to approximate the first and second derivatives in Eqs. (49) and (52). For the Neumann boundary conditions (50) and (53), we apply the second-order forward or backward difference approximation. The interface boundary conditions (51) and (54) are obtained by interpolation from the other subdomain. Consider the situation shown in Figure 4 as an example, in which the concentration at point x(x,y)∈Γfp\boldsymbol{x}(x,y) \in\Gamma_{fp} is going to be interpolated from that in the particle subdomain Ωp\Omega_p. First, the point x\boldsymbol{x} is mapped from Cartesian coordinates to polar coordinates as r(r,θ)r(r,\theta), then the interpolation is calculated in a rectangular grid as

Interpolation of interface boundary conditions

Figure 4. Interpolation of interface boundary conditions. The point x(x,y)\mathbf{x}(x,y) in Cartesian coordinates is mapped to r(r,θ)\mathbf{r}(r,\theta) in polar coordinates.

c(r)=∑i=i0ir∑j=j0jrγijcij,(55)c(\mathbf{r}) = \sum_{i=i_0}^{i_r} \sum_{j=j_0}^{j_r} \gamma_{ij} c_{ij}, \tag*{(55)}

where γij\gamma_{ij} is the interpolation weight, cijc_{ij} is the solute concentration at grid points, and (i0,j0)(i_0,j_0) and (ir,jr)(i_r,j_r) are the lower left corner and the upper right corner of the interpolation stencil, respectively. To maintain second-order accuracy of the global solution, a bi-quadratic interpolation stencil is used [33] (see the red grids in Figure 4).

Moving boundary treatment

In the Cartesian mesh, some mesh points are covered by the particle, as shown in Figure 5a. At these points, concentration is set as 0. As the particle moves, these mesh points may become uncovered, and solute concentrations should be evaluated at these points (see the red dots in Figure 5b). Similar situations happen to polar mesh points when the particle get close to fixed boundaries in the fluid subdomain. To update concentrations at these newly appeared points, we employ interpolations from other subdomains using Eqs. (51) and (54) after advancing the particle.

Refilling of mesh points at two consecutive time steps

Figure 5. Refilling of mesh points (red dots) at two consecutive time steps. The solid lines indicate the particle, and the dashed lines indicate the interface boundaries Γfp\Gamma_{fp} of the fluid subdomain Ωf\Omega_f.

Another problem arises when there are multiple particles. As two particles get too close to each other, there might not be enough Cartesian mesh points for interpolation using Eq. (54). At this time, the interface boundary conditions of one particle are interpolated from the subdomain of the other particle. This ensures accuracy without local mesh refinement.

Solution procedure

The full numerical framework integrates the FBIM for fluid-structure interactions and the OMM for sharp-interface advection-diffusion problems. Given all values at time tt, the solutions at time t+Δtt+\Delta t are calculated by following procedures:

  1. Advance particles using Eq. (35);

  2. Refill solute concentrations at newly appered mesh points using Eqs. (51) and (54);

  3. Solve advection-diffusion problems Eqs. (49) and (52) in each subdomain with global and interface boundary conditions, and update solute concentrations on particles;

  4. Solve for surface force F\boldsymbol{F}, desired velocity V\boldsymbol{V}, translational velocity Up\boldsymbol{U}_p, and rotational velocity Ωp\boldsymbol{\Omega}_p.

  5. Solve the long range problem Eq. (11) with force density fl\boldsymbol{f}^{l} and the short range problem Eq. (12) using Eq. (30), and update flow field and velocities on boundaries (both particles and fixed boundaries).

Validations

In this section, the FBIM and the OMM are validated seperately by several numerical simulations. Their convergence behaviors are also presented. Here we define the relative error of any parameter CC as

εC=∣C(num)−C(ref)∣∣C(ref)∣(56)\varepsilon_C = \frac{\left|C_{(num)}-C_{(ref)}\right|}{\left|C_{(ref)}\right|} \tag*{(56)}

where C(num)C_{(num)} and C(ref)C_{(ref)} are the numerical and reference solutions, respectively.

Parabolic flow with circular boundary

We first validate the FBIM by considering a simple parabolic flow. The computational domain is a square of side length L=6.4L=6.4. Periodic boundary conditions are employed in xx and yy directions. A circular boundary Γ\Gamma of radius 2.5 is located at coordinates (3.0, 3.0), and the boundary condition reads

ux∣Γ=y2,uy∣Γ=0.(57)u_x|_{\Gamma}=y^2,\quad u_y|_{\Gamma}=0. \tag*{(57)}

The surface force F\boldsymbol{F} on the boundary is calculated by Eq. (32) with the desired velocity V=(y2,0)\boldsymbol{V}=(y^2,0). The velocity due to this surface force is computed at coordinates (3.2, 3.2) at t=20t=20.

Figure 6 shows the relative error of uxu_x and the computational time by varying the number of mesh points NxN_x and NyN_y between 24 to 512. With fixed cutoff distance rc=L/16r_c=L/16, we observe a 4th order convergence, and the computational time per step scales as 1/Δx1/\Delta x. Here N=Nx×NyN=N_x\times N_y. When setting rc=8Δxr_c=8\Delta x, the cutoff distance decreases linearly with the mesh size. On the one hand, decreasing rcr_c leads to a lower order convergence (1st order); on the other hand, the computational time per step is hardly affected. This indicates that most of the computational time is spent on the short-range problem depending on rcr_c. Taking both the accuracy and efficiency into consideration, the cutoff distance is set as rc=8Δxr_c = 8\Delta x in following simulations.

Validation and convergence study of the FBIM

Figure 6. Validation and convergence study of the FBIM. (a) Convergence of velocity uxu_x with respect to mesh size Δx\Delta x at coordinates (3.2,3.2)(3.2, 3.2). (b) Computational time per step with respect to mesh size Δx\Delta x and number of Cartesian mesh points NN. Here NN scales with Δx−2\Delta x^{-2}.

Single phoretic particle with finite system size

For the validation of the OMM, we consider a phoretic particle with limited system size. This problem has been studied analytically and numerically in Ref. [9]. As shown in Figure 7, the particle boundary Γp\Gamma_p and the outer boundary Γo\Gamma_o are concentric circles of radius 1 and R=3.25R = 3.25, respectively. Here RR is the system size. The governing equations are Eqs. (1) and (2) with 0 consumption/production rate (β=0\beta= 0). The boundary conditions are summarized as follows [8]:

Schematic of the computational domain for single particle dynamics

Figure 7. Schematic of the computational domain for single particle dynamics. The domain Ω\Omega is decomposed into three overlapping subdomains: the fluid subdomain Ωf\Omega_f which coincides with Ω\Omega, the particle subdomain Ωp\Omega_p, and the outer subdomain Ωo\Omega_o. The solid lines represent the global boundaries, and the dashed lines represent the interface boundaries.

u(x,t)∣Γp=M∇sc(x,t)∣Γp+Up(t)(58)\mathbf{u}(\mathbf{x},t)|_{\Gamma_p}=M\nabla_s c(\mathbf{x},t)|_{\Gamma_p}+\mathbf{U}_p(t) \tag*{(58)}
n⋅∇c(x,t)∣Γp=−A,c(x,t)∣Γo=0.(59)\mathbf{n}\cdot\nabla c(\mathbf{x},t)|_{\Gamma_p}=-A,\quad c(\mathbf{x},t)|_{\Gamma_o}=0. \tag*{(59)}

Moreover, the velocity attenuates to 0 in the far field. The flow field is calculated analytically by stream function [5, 8]. The solute concentration is computed by the OMM in a square domain Ω\Omega. As shown in Figure 7, the computational domain Ω\Omega is decomposed into three overlapping subdomains. The fluid subdomain Ωf\Omega_f coincides with Ω\Omega, and periodic boundary conditions are used in xx and yy directions. The particle subdomain Ωp\Omega_p is an annular domain comoving with the particle (see the red ring in Figure 7). Another annular subdomain (the outer subdomain Ωo\Omega_o, as shown by the blue ring in Figure 7) is attached to the outer boundary Γo\Gamma_o which also comoves with the particle. For the two annular subdomains, the global boundaries are represented by solid lines, and the boundary conditions follow Eq. (59), while the interface boundaries are labeled by dashed lines (see Figure 7).

The particle velocity ∥Up∥\lVert\mathbf{U}_p\rVert and angular velocity ωp\omega_p obtained by the OMM are compared with the analytical results from Ref. [9], as shown in Figure 8a. Here a non-zero angular velocity indicates circular motion. Good agreement between the two results is observed when setting the mesh size Δx=1/64\Delta x = 1/64. Convergence of the phoretic velocity ∥Up∥\lVert U_p\rVert with respect to Δx\Delta x is demonstrated in Figure 8b, as the Péclet number is Pe=5.72Pe = 5.72. To calculate the relative error ϵ∥Up∥\epsilon_{\lVert U_p\rVert}, the analytical result from Ref. [9] is used as a reference. As Δx\Delta x varies between 1/1001/100 to 1/321/32, a 3rd order accuracy is reached, thereby validating our implementation of the OMM.

Comparison between the analytical prediction and numerical simulation using the OMM

Figure 8. Comparison between the analytical prediction and numerical simulation using the OMM. (a) Phoretic velocity ∥Up(t)∥\lVert U_p(t)\rVert and angular velocity ωp\omega_p in stationary, straight and circular phases. The mesh size is Δx=1/64\Delta x = 1/64. (b) Convergence of phoretic velocity ∥Up∥\lVert U_p\rVert with respect to mesh size Δx\Delta x at Pe=5.72Pe = 5.72. The system size is set to R=3.25R = 3.25. The analytical prediction is obtained by Ref. [9].

Applications

In this section, we combine the FBIM and the OMM to explore new simulations for single particle and many particles dynamics. First we study particle swimming in a nearly infinite domain with periodic boundary conditions. Then particle dynamics in a straight channel is investigated. Finally we show examples of multiple particles in a rectangular domain with periodic boundary conditions to reveal some interesting phenomena. For all simulations, the comsumption rate is set to β=0.01\beta= 0.01.

Single particle in a nearly infinite domain

We first consider a similar case as presented in § 6.2, but with larger system size R=200R = 200, which is large enough to mimic an infinite domain [8]. The computational domain is a square and its size is L=409.6L = 409.6. The flow field is computed on a Cartesian mesh containing 8192 grid points in each direction, corresponding to a mesh size of Δx=0.05\Delta x = 0.05. The particle surface is discretized into Ne=128N_e = 128 elements. The subdomain sizes are set as rp=1.6r_p = 1.6 for the two annular subdomains Ωp\Omega_p and Ωf\Omega_f. The radial and tangential meshes contain 32 and 128 grid points for Ωp\Omega_p, and 32 and 2560 grid points for Ωo\Omega_o, respectively.

The phoretic velocity ∥Up∥\lVert U_p\rVert computed by the FBIM-OMM is in good agreement with the published data [8] as demonstrated in Figure 9. We further plot particle trajectories for Pe=6,12.8,13Pe = 6, 12.8, 13 in Figure 10. At short enough time scale, the trajectories obtained by the FBIM-OMM are similar to those in Ref. [8], as shown in the insets of Figure 10. However, at larger time scales, the particle does not move along a straight line at Pe=6Pe = 6 (Figure 10a), or along a circle at Pe=12.8Pe = 12.8 (Figure 10b), as reported in [8]. Moreover, the chaotic motion at Pe=13Pe = 13 is different from that in Ref. [8] (see Figure 10c). These deviations result from the finite computational domain size LL and the periodic boundary conditions. In Ref. [8], the flow velocity attenuates to 0 in the far field. In contrast, for the present work, the flow in the computational domain is affected by its images due to periodic boundary conditions. Although the domain size LL is 2 orders of magnitude larger than the particle radius, this effect is still non-negligible. For this reason, the particle trajectories are different in long scale.

Phoretic velocity as a function of Péclet number

Figure 9. Phoretic velocity ∥Up∥\lVert U_p\rVert as a function of PePe at R=200R = 200. The solid line represents the data obtained by Ref. [8].

Different trajectories obtained by the FBIM-OMM

Figure 10. Different trajectories obtained by the FBIM-OMM.

Single particle in a straight channel

In this subsection, we consider a phoretic particle immersed in a straight channel of length L=25.6L = 25.6 and width W=5W = 5, as shown in Figure 11a. Periodic boundary conditions are used at the left and right boundaries for both the flow and solute fields. At other boundaries, the boundary conditions follow Eqs. (3) and (4). Other simulation parameters are: mesh size Δx=0.05\Delta x = 0.05, particle surface element number Ne=128N_e = 128, and particle subdomain size rp=0.8r_p = 0.8. A convergence study with respect to Δx\Delta x and rpr_p will be presented later in this subsection.

Single particle swimming in a straight channel

Figure 11. Single particle swimming in a straight channel. (a) Steady-state relative solute concentration at Pe=2Pe = 2 and 12, where the arrows indicate translational and rotational directions. (b) Phoretic velocity ∥Up∥\lVert U_p\rVert, (c) rotational velocity ∥Ωp∥\lVert\Omega_p\rVert, and (d) lateral position ∣yp∣\lvert y_p\rvert at steady state.

As shown in Figure 11a, the particle swimms in a straight line (indicated by the white solid arrows), and its velocity is shown in Figure 11b. The phoretic velocity ∥Up∥\lVert U_p \rVert shows a transition from stationary state to straight motion around Pe≈0.6Pe \approx0.6, which corresponds to a pitchfork bifurcation. Moreover, a symmetry breaking is found around Pe≈8Pe \approx8, so that the particle stays at an off-centered position and rotates around its center (see Figure 11c and Figure 11d). Here ypy_p is the particle lateral position from the channel center. Indeed, when the particle locates in the center of the channel, the hydrodynamic interactions of the two fixed boundaries are equal, and no rotational motion occurs. As the particle gets closer to one solid boundary than the other, it rotates clockwise (counter-clockwise) if the closest boundary is on the left (right) in the frame moving with the particle (see the white dashed arrow in Figure 11a).

The convergence test is conducted at Pe=2Pe = 2. The relative error of phoretic velocity ϵ∥Up∥\epsilon_{\lVert U_p\rVert} is computed, and the numerical solution for Δx=0.0125\Delta x = 0.0125, Ne=512N_e = 512, and rp=2.0r_p = 2.0 is used as reference solution. First rp=0.8r_p = 0.8 is fixed, while Δx\Delta x is varied between 0.025 to 0.1 and NeN_e scales as 1/Δx1/\Delta x. As shown in Figure 12a, a second order convergence is obtained, in agreement with the nominal accuracy of the FBIM-OMM. Then we fix the mesh size at Δx=0.05\Delta x = 0.05 and the particle surface element number at Ne=128N_e = 128, while rpr_p is varied between 0.4 to 1.6. The error is slightly reduced with the increase of rpr_p, as demonstrated in Figure 12b. In fact, larger overlap sizes exhibit better stability, while the accuracy is determined by the interpolation scheme at interface boundaries [33, 61].

Convergence of phoretic velocity with respect to mesh size and particle subdomain size

Figure 12. Convergence of phoretic velocity ∥Up∥\lVert U_p\rVert with respect to (a) mesh size Δx\Delta x and (b) particle subdomain size rpr_p. The Péclet number is fixed at Pe=2Pe = 2.

Collective behaviors of multiple particles

In this subsection, a suspension of particles in a square domain is simulated with the proposed method. The side length of the domain is set as L=25L = 25, and 40 phoretic particles are distributed in the domain, as shown in Figure 13a. The corresponding area concentration is about 20%. Periodic boundary conditions are used in both directions. According to the convergence study in the previous subsection, the other simulation parameters are chosen: mesh size Δx=0.05\Delta x = 0.05, particle surface element number Ne=128N_e = 128, and particle subdomain size rp=0.8r_p = 0.8.

Collective behaviors of multiple particles

Figure 13. Collective behaviors of multiple particles. (a–c) Snapshots of particles and relative solute concentration at Pe=2Pe = 2, 4, and 6. The white dashed line in (a) is a guide for the eyes. The arrows in (b) and (c) indicate translational velocity directions. (d) Phoretic velocity ∥U∥\lVert U\rVert and (e) rotational velocity ∥Ω∥\lVert\Omega\rVert averaged over all particles.

At Pe≤2Pe \le2, the particles form a crystal-like structure at steady state, as shown in Figure 13a. Although the particles stay at equilibrium positions (i.e. ∥U∥=0\lVert U\rVert= 0, see Figure 13c), there is flow among them, since the surface concentration on each particle is not homogeneous. Moreover, some particles rotate around their centers due to hydrodynamic interactions. As a result, non-zero rotational velocity ∥Ω∥\lVert\Omega\rVert is observed in Figure 13d. It should be noted that ∥U∥\lVert U\rVert and ∥Ω∥\lVert\Omega\rVert are average values over all particles. The crystal-like structure loses its stability for Pe>3Pe > 3 in favor of a liquid-like regime, as shown in Figure 13b. Particles can swim in a short range, caged by other surrounding particles. For such a disordered (no noise is introduced) solution, the translational and rotational velocities are defined as their time-averaged values:

∥U∥=1T∫0T∥U(t)∥ dt, ∥Ω∥=1T∫0T∥Ω(t)∥ dt,(60)\lVert U\rVert= \frac{1}{T}\int_{0}^{T}\lVert U(t)\rVert\,\mathrm{d}t,\ \lVert\Omega\rVert= \frac{1}{T}\int_{0}^{T}\lVert\Omega(t)\rVert\,\mathrm{d}t, \tag*{(60)}

which is measured over a time interval TT. Increasing further PePe leads to a more chaotic regime (Pe≥5Pe \ge5), as shown in Figure 13c. In this gas-like regime, both translation and orientation orders are lost, and particles can travel freely for longer distance. Moreover, particles form traveling lines for Pe≥6Pe \ge6 (see Figure 13c). Similar behaviors have been observed experimentally [7]. With the increase of PePe, both ∥U∥\lVert U\rVert and ∥Ω∥\lVert\Omega\rVert show nonmonotonic behaviors, as shown in Figure 13d and e. These results lay the foundation for investigations of collective motions of phoretic particles. We leave detailed explanation and systemic studies in future works.

Conclusion

A full numerical framework for phoretic particles was presented. This framework consists of a FBIM for fluid-structure interactions in Stokes flow, and an OMM for advection-diffusion problems with moving boundaries. The FBIM decomposes the Stokes equations into short-range and long-range parts, so that the former is computed by boundary integral equation within a cut-off distance, while the latter is solved by a Fourier spectral method accelerated by FFT. To resolve moving boundaries in advection-diffusion problems, the OMM divides the computational domain into several overlapping subdomains, which are allowed to move independently. The FBIM and the OMM were then validated separately, showing relatively high order accuracy. We further applied this framework to more general problems, including single particle in nearly infinite domain, single particle in a straight channel, and collective behaviors of multiple particles. This framework can be further extended to non-circular rigid particles and even deformable particles, which will be our future research.

Acknowledgements

This work was supported by CNES (Centre National d’Etudes Spatiales) and the French-German university program “Living Fluids” (Grant CFDA-Q1-14). The simulations were performed on the Cactus cluster of the CIMENT infrastructure, which is supported by the Rhône-Alpes region (GRANT CPER07_13 CIRA).

Long-range density

A natural choice of Dl(r)D^{l}(r) is the polynomial, of which the high order derivatives are continuous. In practice, we chose a 6th-order polynomial Dl(r)=∑i=06airiD^{l}(r) = \sum_{i=0}^{6} a_i r^i and set

∂kDl∂rk∣r=rc=0,∂Dl∂r∣r=0=0,(61)\left.\frac{\partial^k D^{l}}{\partial r^k}\right|_{r=r_c}=0,\quad\left.\frac{\partial D^{l}}{\partial r}\right|_{r=0}=0, \tag*{(61)}

where k=0,1,2,3k=0,1,2,3.

The integral in Eq. (16) can be written in polar coordinates, and the following equation is obtained as

Gij(x,xv)=∫0rcDl(r′)r′[∫02πGij(r,r′)dθ]dr′,(62)G_{ij}(x,x^{v})=\int_{0}^{r_c}D^{l}(r')r'\left[\int_{0}^{2\pi}G_{ij}(r,r')d\theta\right]dr', \tag*{(62)}

where r′=x′−xv\boldsymbol{r}'=\boldsymbol{x}'-\boldsymbol{x}^{v} and θ\theta is the angle between r\boldsymbol{r} and r′\boldsymbol{r}'. Substitute Eq. (8) into the equation above and integrate over θ\theta, we obtain the following equation

Gij(x,xv)=2πGij(x,xv)∫0rcDl(r′)r′dr′+πri∗rj∗−rirjr4∫0rcDl(r′)r′3dr′.(63)G_{ij}(x,x^{v})=2\pi G_{ij}(x,x^{v})\int_{0}^{r_c}D^{l}(r')r'dr'+\pi\frac{r_i^{*}r_j^{*}-r_i r_j}{r^4}\int_{0}^{r_c}D^{l}(r')r'^3dr'. \tag*{(63)}

Here r∗\boldsymbol{r}^{*} is the vector perpendicular to r\boldsymbol{r}. To satisfy (A.3), we set

∫0rcDl(r′)r′dr′=12π,∫0rcDl(r′)r′3dr′=0.(64)\int_{0}^{r_c}D^{l}(r')r'dr'=\frac{1}{2\pi},\qquad\int_{0}^{r_c}D^{l}(r')r'^3dr'=0. \tag*{(64)}

By solving the system of (A.1) and (A.4), we obtain the polynomial coefficients, and the long-range density is given as

Dl(r)={0,r≥rc,563πrc2[−252(rrc)6+54(rrc)5−90(rrc)4+70(rrc)3−452(rrc)2+1],r<rc.(65)D^{l}(r)= \begin{cases} 0, & r\ge r_c,\\ \frac{56}{3\pi r_c^2}\left[-\frac{25}{2}\left(\frac{r}{r_c}\right)^6+54\left(\frac{r}{r_c}\right)^5-90\left(\frac{r}{r_c}\right)^4+70\left(\frac{r}{r_c}\right)^3-\frac{45}{2}\left(\frac{r}{r_c}\right)^2+1\right], & r<r_c. \end{cases} \tag*{(65)}

Short-range Green's function

Rewrite the integral in Eq. (22) in polar coordinate and substitute Eqs. (17) and (21), the short-range boundary integral equation becomes:

uis(x)=14πgjvGij(x,xv)−14πgjv∫0rcDl(r′)r′[∫02πGij(r,r′)dθ]dr′=14πgjvGij(x,xv)−A1gjvGij(x,xv)−A2gjv(ri∗rj∗−rirj)−A3gjvδij−A4gjvδij.(66)\begin{aligned} u_i^s(x)=\frac{1}{4\pi}g_j^vG_{ij}(x,x^v)-\frac{1}{4\pi}g_j^v\int_{0}^{r_c}D^{l}(r')r'\left[\int_{0}^{2\pi}G_{ij}(r,r')d\theta\right]dr' \\ &=\frac{1}{4\pi}g_j^vG_{ij}(x,x^v)-A_1g_j^vG_{ij}(x,x^v)-A_2g_j^v(r_i^{*}r_j^{*}-r_i r_j)-A_3g_j^v\delta_{ij}-A_4g_j^v\delta_{ij}. \tag*{(66)} \end{aligned}

where

A1=12∫0rDl(r′)r′dr′=283π[−2516(rrc)8+547(rrc)7−15(rrc)6+14(rrc)5−458(rrc)4+12(rrc)2],\begin{aligned} A_1 = \frac{1}{2}\int_0^r D^{l}(r')r'dr' \\ &= \frac{28}{3\pi}\left[-\frac{25}{16}\left(\frac{r}{r_c}\right)^8+\frac{54}{7}\left(\frac{r}{r_c}\right)^7-15\left(\frac{r}{r_c}\right)^6+14\left(\frac{r}{r_c}\right)^5-\frac{45}{8}\left(\frac{r}{r_c}\right)^4+\frac{1}{2}\left(\frac{r}{r_c}\right)^2\right], \end{aligned}
A2=14r4∫0rDl(r′)r′3dr′=143πrc2[−54(rrc)6+6(rrc)5−454(rrc)4+10(rrc)3−154(rrc)2+14],\begin{aligned} A_2 = \frac{1}{4r^4}\int_0^r D^{l}(r')r'^{3}dr' \\ &= \frac{14}{3\pi r_c^2}\left[-\frac{5}{4}\left(\frac{r}{r_c}\right)^6+6\left(\frac{r}{r_c}\right)^5-\frac{45}{4}\left(\frac{r}{r_c}\right)^4+10\left(\frac{r}{r_c}\right)^3-\frac{15}{4}\left(\frac{r}{r_c}\right)^2+\frac{1}{4}\right], \end{aligned}
A3=−12∫rrcDl(r′)r′ln⁡(r′)dr′=−283π[−25128(rrc)8+5449(rrc)7−52(rrc)6+145(rrc)5−4532(rrc)4+14(rrc)2−158331360+2516(rrc)8ln⁡(r)−547(rrc)7ln⁡(r)+15(rrc)6ln⁡(r)−14(rrc)5ln⁡(r)+458(rrc)4ln⁡(r)−12(rrc)2ln⁡(r)+3112ln⁡(rc)],(67)\begin{aligned} A_3 = -\frac{1}{2}\int_r^{r_c}D^{l}(r')r'\ln(r')dr' \\ &= -\frac{28}{3\pi}\left[-\frac{25}{128}\left(\frac{r}{r_c}\right)^8+\frac{54}{49}\left(\frac{r}{r_c}\right)^7-\frac{5}{2}\left(\frac{r}{r_c}\right)^6+\frac{14}{5}\left(\frac{r}{r_c}\right)^5-\frac{45}{32}\left(\frac{r}{r_c}\right)^4+\frac{1}{4}\left(\frac{r}{r_c}\right)^2-\frac{1583}{31360}+\frac{25}{16}\left(\frac{r}{r_c}\right)^8\ln(r)-\frac{54}{7}\left(\frac{r}{r_c}\right)^7\ln(r)+15\left(\frac{r}{r_c}\right)^6\ln(r)-14\left(\frac{r}{r_c}\right)^5\ln(r)+\frac{45}{8}\left(\frac{r}{r_c}\right)^4\ln(r)-\frac{1}{2}\left(\frac{r}{r_c}\right)^2\ln(r)+\frac{3}{112}\ln(r_c)\right], \tag*{(67)} \end{aligned}
A4=14∫rrcDl(r′)r′dr′=143π[2516(rrc)8−547(rrc)7+15(rrc)6−14(rrc)5+458(rrc)4−12(rrc)2+3112].\begin{aligned} A_4 = \frac{1}{4}\int_r^{r_c}D^{l}(r')r'dr' \\ &= \frac{14}{3\pi}\left[\frac{25}{16}\left(\frac{r}{r_c}\right)^8-\frac{54}{7}\left(\frac{r}{r_c}\right)^7+15\left(\frac{r}{r_c}\right)^6-14\left(\frac{r}{r_c}\right)^5+\frac{45}{8}\left(\frac{r}{r_c}\right)^4-\frac{1}{2}\left(\frac{r}{r_c}\right)^2+\frac{3}{112}\right]. \end{aligned}

Then the short-range Green’s function is obtained as

Gijs(x,xv)=∫ΩcGij(x,x′)Ds(∣x′−xv∣)dx′=(1−4πA1)Gij(x,xv)−4πA2(ri∗rj∗−rirj)−4π(A3+A4)δij.(68)\begin{aligned} G_{ij}^{s}(x,x^{v})=\int_{\Omega_c}G_{ij}(x,x')D^{s}(|x'-x^{v}|)dx' \\ &=(1-4\pi A_1)G_{ij}(x,x^{v})-4\pi A_2(r_i^{*}r_j^{*}-r_ir_j)-4\pi(A_3+A_4)\delta_{ij}. \tag*{(68)} \end{aligned}

References

  1. [1]S. Michelin, E. Lauga, D. Bartolo, Spontaneous autophoretic motion of isotropic particles, Physics of Fluids 25 (6) (2013) 061701. doi:10.1063/1.4810749.DOI
  2. [2]Z. Izri, M. N. van der Linden, S. Michelin, O. Dauchot, Self-propulsion of pure water droplets by spontaneous marangoni-stress-driven motion, Phys Rev Lett 113 (24) (2014) 248302. doi:10.1103/PhysRevLett.113.248302. URL https://www.ncbi.nlm.nih.gov/pubmed/25541808DOI
  3. [3]S. Herminghaus, C. C. Maass, C. Krüger, S. Thutupalli, L. Goehring, C. Bahr, Interfacial mechanisms in active emulsions, Soft Matter 10 (36) (2014) 7008–22. doi:10.1039/c4sm00550c. URL https://www.ncbi.nlm.nih.gov/pubmed/24924906DOI
  4. [4]S. Michelin, E. Lauga, Phoretic self-propulsion at finite péclét numbers, Journal of Fluid Mechanics 747 (2014) 572–604. doi:10.1017/jfm.2014.158.DOI
  5. [5]D. Sondak, C. Hawley, S. Heng, R. Vinsonhaler, E. Lauga, J. L. Thiffeault, Can phoretic particles swim in two dimensions?, Phys Rev E 94 (6-1) (2016) 062606. doi:10.1103/PhysRevE.94.062606. URL https://www.ncbi.nlm.nih.gov/pubmed/28085389DOI
  6. [6]C. Jin, C. Kruger, C. C. Maass, Chemotaxis and autochemotaxis of self-propelling droplet swimmers, Proc Natl Acad Sci U S A 114 (20) (2017) 5089–5094. doi:10.1073/pnas.1619783114. URL https://www.ncbi.nlm.nih.gov/pubmed/28465433DOI
  7. [7]S. Thutupalli, D. Geyer, R. Singh, R. Adhikari, H. A. Stone, Flow-induced phase separation of active particles is controlled by boundary conditions, Proc Natl Acad Sci U S A 115 (21) (2018) 5403–5408. doi:10.1073/pnas.1718807115. URL https://www.ncbi.nlm.nih.gov/pubmed/29735679DOI
  8. [8]W. F. Hu, T. S. Lin, S. Rafai, C. Misbah, Chaotic swimming of phoretic particles, Phys Rev Lett 123 (23) (2019) 238004. doi:10.1103/PhysRevLett.123.238004. URL https://www.ncbi.nlm.nih.gov/pubmed/31686429DOI
  9. [9]A. Farutin, M. S. Rizvi, W. F. Hu, T. S. Lin, S. Rafai, C. Misbah, A reduced model for a phoretic swimmer, Journal of Fluid Mechanics 952 (2022) A6. doi:10.1017/jfm.2022.870.DOI
  10. [10]G. Li, Swimming dynamics of a self-propelled droplet, Journal of Fluid Mechanics 934 (2022). doi:10.1017/jfm.2021.1154.DOI
  11. [11]S. Michelin, Self-propulsion of chemically active droplets, Annual Review of Fluid Mechanics 55 (1) (2023) 77–101. doi:10.1146/annurev-fluid-120720-012204.DOI
  12. [12]C. Jin, J. Vachier, S. Bandyopadhyay, T. Macharashvili, C. C. Maass, Fine balance of chemotactic and hydrodynamic torques: When microswimmers orbit a pillar just once, Phys Rev E 100 (4-1) (2019) 040601. doi:10.1103/PhysRevE.100.040601. URL https://www.ncbi.nlm.nih.gov/pubmed/31770913DOI
  13. [13]A. Izzet, P. G. Moerman, P. Gross, J. Groenewold, A. D. Hollingsworth, J. Bibette, J. Brujic, Tunable persistent random walk in swimming droplets, Physical Review X 10 (2) (2020). doi:10.1103/PhysRevX.10.021035.DOI
  14. [14]B. V. Hokmabad, J. A. Oguado-Canalejo, S. Saha, R. Golestanian, C. C. Maass, Chemotactic self-caging in active emulsions, Proc Natl Acad Sci U S A 119 (24) (2022) e2122269119. doi:10.1073/pnas.2122269119. URL https://www.ncbi.nlm.nih.gov/pubmed/35679341DOI
  15. [15]B. V. Hokmabad, A. Nishide, P. Ramesh, C. Kruger, C. C. Maass, Spontaneously rotating clusters of active droplets, Soft Matter 18 (14) (2022) 2731–2741. doi:10.1039/d1sm01795k. URL https://www.ncbi.nlm.nih.gov/pubmed/35319552DOI
  16. [16]K. Lippera, M. Morozov, M. Benzaquen, S. Michelin, Collisions and rebounds of chemically active droplets, Journal of Fluid Mechanics 886 (2020). doi:10.1017/jfm.2019.1055.DOI
  17. [17]N. Desai, S. Michelin, Instability and self-propulsion of active droplets along a wall, Physical Review Fluids 6 (11) (2021). doi:10.1103/PhysRevFluids.6.114103.DOI
  18. [18]K. Lippera, M. Benzaquen, S. Michelin, Alignment and scattering of colliding active droplets, Soft Matter 17 (2) (2021) 365–375. doi:10.1039/d0sm01285h. URL https://www.ncbi.nlm.nih.gov/pubmed/33169775DOI
  19. [19]T. D. Montenegro-Johnson, S. Michelin, E. Lauga, A regularised singularity approach to phoretic problems, Eur Phys J E Soft Matter 38 (12) (2015) 139. doi:10.1140/epje/i2015-15139-7. URL https://www.ncbi.nlm.nih.gov/pubmed/26701714
  20. [20]R. Singh, R. Adhikari, M. E. Cates, Competing chemical and hydrodynamic interactions in autophoretic colloidal suspensions, J Chem Phys 151 (4) (2019) 044901. doi:10.1063/1.5090179. URL https://www.ncbi.nlm.nih.gov/pubmed/33707557DOI
  21. [21]F. Rojas-Pérez, B. Delmotte, S. Michelin, Hydrochemical interactions of phoretic particles: a regularized multipole framework, Journal of Fluid Mechanics 919 (2021). doi:10.1017/jfm.2021.387.
  22. [22]F. Picella, S. Michelin, Confined self-propulsion of an isotropic active colloid, Journal of Fluid Mechanics 933 (2022) A27. doi:10.1017/jfm.2021.1081.DOI
  23. [23]Y. Chen, K. L. Chong, L. Liu, R. Verzicco, D. Lohse, Instabilities driven by diffusiophoretic flow on catalytic surfaces, Journal of Fluid Mechanics 919 (2021). doi:10.1017/jfm.2021.370.DOI
  24. [24]Q. Yang, M. Jiang, F. Picano, L. Zhu, Collective behaviors of isotropic phoretic microswimmers: from crystalline solids to active turbulence (2023). arXiv:2304.01593, doi:10.48550/arXiv.2304.01593.
  25. [25]C. Pozrikidis, Boundary Integral and Singularity Methods for Linearized Viscous Flow, Cambridge Texts in Applied Mathematics, Cambridge University Press, 1992. doi:10.1017/CBO9780511624124.DOI
  26. [26]M. Deserno, C. Holm, How to mesh up euqald sums. i. a theoretical and numerical comparison of various particle mesh routines, The Journal of chemical physics 109 (18) (1998) 7678–7693.
  27. [27]U. Essmann, L. Pereira, M. L. Berkowitz, T. Darden, H. Lee, L. G. Pedersen, A smooth particle mesh ewald method, The Journal of chemical physics 103 (19) (1995) 8577–8593.DOI
  28. [28]E. K. Guckel, Large scale simulations of particulate systems using the PME method, University of Illinois at Urbana-Champaign, 1999.
  29. [29]J. P. Hernandez-Ortiz, J. J. de Pablo, M. D. Graham, Fast computation of many-particle hydrodynamic and electrostatic interactions in a confined geometry, Phys Rev Lett 98 (14) (2007) 140602. doi: 10.1103/PhysRevLett.98.140602. URL https://www.ncbi.nlm.nih.gov/pubmed/17501260
  30. [30]A. Kumar, M. D. Graham, Accelerated boundary integral method for multiphase flow in non-periodic geometries, Journal of Computational Physics 231 (20) (2012) 6682–6713.
  31. [31]P. McCorquodale, P. Colella, H. Johansen, A cartesian grid embedded boundary method for the heat equation on irregular domains, Journal of Computational Physics 173 (2) (2001) 620–635. doi:10.1006/jcph.2001.6900.
  32. [32]L. Schneiders, C. Günther, M. Meinke, W. Schröder, An efficient conservative cut-cell method for rigid bodies interacting with viscous compressible flows, Journal of Computational Physics 311 (2016) 62–86. doi:10.1016/j.jcp.2016.01.026.
  33. [33]G. Chesshire, W. D. Henshaw, Composite overlapping meshes for the solution of partial differential equations, Journal of Computational Physics 90 (1) (1990) 1–64. doi:10.1016/0021-9991(90)90196-8.
  34. [34]W. D. Henshaw, D. W. Schwendeman, Parallel computation of three-dimensional flows using overlapping grids with adaptive mesh refinement, Journal of Computational Physics 227 (16) (2008) 7469–7502. doi:10.1016/j.jcp.2008.04.033.
  35. [35]N. C. Prewitt, D. M. Belk, W. Shyy, Parallel computing of overset grids for aerodynamic problems with moving objects, Progress in Aerospace Sciences 36 (2) (2000) 117–172.
  36. [36]W. D. Henshaw, D. W. Schwendeman, Moving overlapping grids with adaptive mesh refinement for high-speed reactive and non-reactive flow, Journal of Computational Physics 216 (2) (2006) 744–779.
  37. [37]B. E. Merrill, Y. T. Peet, Moving overlapping grid methodology of spectral accuracy for incompressible flow solutions around rigid bodies in motion, Journal of Computational Physics 390 (2019) 121–151. doi:10.1016/j.jcp.2019.01.048.
  38. [38]D. Goldstein, R. Handler, L. Sirovich, Modeling a no-slip flow boundary with an external force field, Journal of Computational Physics 105 (2) (1993) 354–366. doi:10.1006/jcph.1993.1081.
  39. [39]C. S. Peskin, The immersed boundary method, Acta Numerica 11 (2002) 479 – 517. doi:10.1017/S0962492902000077.DOI
  40. [40]X. Wang, X. Gong, K. Sugiyama, S. Takagi, H. Huang, An immersed boundary method for mass transfer through porous biomembranes under large deformations, Journal of Computational Physics 413 (2020) 109444. doi:10.1016/j.jcp.2020.109444.
  41. [41]Z. Li, M.-C. Lai, The immersed interface method for the navier–stokes equations with singular forces, Journal of Computational Physics 171 (2) (2001) 822–842. doi:10.1006/jcph.2001.6813.
  42. [42]W.-F. Hu, M.-C. Lai, C. Misbah, A coupled immersed boundary and immersed interface method for interfacial flows with soluble surfactant, Computers & Fluids 168 (2018) 201–215. doi:10.1016/j.compfluid.2018.04.013.
  43. [43]P. Lallemand, L.-S. Luo, Lattice boltzmann method for moving boundaries, Journal of Computational Physics 184 (2) (2003) 406–421.
  44. [44]J. Huang, Z. Hu, W.-A. Yong, Second-order curved boundary treatments of the lattice boltzmann method for convection–diffusion equations, Journal of Computational Physics 310 (2016) 26–44. doi:10.1016/j.jcp.2016.01.008.
  45. [45]H. D. Zhang, C. Misbah, Lattice boltzmann simulation of advection–diffusion of chemicals and applica- tions to blood flow. Computers & Fluids 187 (2019) 46–59. doi:10.1016/j.compfluid.2019.04.018. URL <GotoISI>://WOS:000470952600004
  46. [46]H. Noguchi, G. Gompper, Fluid vesicles with viscous membranes in shear flow, Physical review letters 93 (25) (2004) 258102.DOI
  47. [47]H. Noguchi, G. Gompper, Shape transitions of fluid vesicles and red blood cells in capillary flows, Proceedings of the National Academy of Sciences 102 (40) (2005) 14159–14164.pubmed.ncbi.nlm.nih.gov/16186506
  48. [48]D. A. Fedosov, B. Caswell, G. E. Karniadakis, A multiscale red blood cell model with accurate mechanics, rheology, and dynamics, Biophysical journal 98 (10) (2010) 2215–2225.DOI
  49. [49]C. Gao, P. Zhang, G. Marom, Y. Deng, D. Bluestein, Reducing the effects of compressibility in dpd-based blood flow simulations through severe stenotic microchannels, Journal of Computational Physics 335 (2017) 812–827.DOI
  50. [50]Z. Xu, C. Wang, F. He, P. Hao, X. Zhang, Coarse-grained model of whole blood hemolysis and morphological analysis of erythrocyte population under non-physiological shear stress flow environment, Physics of Fluids 35 (3) (2023).DOI
  51. [51]N. Tanaka, T. Takano, Microscopic-scale simulation of blood flow using sph method, International Journal of Computational Methods 2 (04) (2005) 555–568.DOI
  52. [52]S. M. Hosseini, J. J. Feng, A particle-based model for the transport of erythrocytes in capillaries, Chemical engineering science 64 (22) (2009) 4488–4497.DOI
  53. [53]H. Zhang, Z. Shen, B. Hogan, A. I. Barakat, C. Misbah, Atp release by red blood cells under flow: Model and simulations, Biophys J 115 (11) (2018) 2218–2229. doi:10.1016/j.bpj.2018.09.033. URL https://www.ncbi.nlm.nih.gov/pubmed/30447988DOI
  54. [54]Z. Gou, H. Zhang, M. Abbasi, C. Misbah, Red blood cells under flow show maximal atp release for specific hematocrit, Biophys J 120 (21) (2021) 4819–4831. doi:10.1016/j.bpj.2021.09.025. URL https://www.ncbi.nlm.nih.gov/pubmed/34547477DOI
  55. [55]Z. Y. Shen, A. Farutin, M. Thiebaud, C. Misbah, Interaction and rheology of vesicle suspensions in confined shear flow, Physical Review Fluids 2 (10) (2017). doi:ARTN10310110.1103/PhysRevFluids.2.103101. URL <GotoISI>://WOS:000412438800001DOI
  56. [56]T.-S. Lin, W.-F. Hu, C. Misbah, A direct poisson solver in spherical geometry with an application to diffusiophoretic problems, Journal of Computational Physics 409 (2020) 109362. doi:10.1016/j.jcp.2020.109362.DOI
  57. [57]W.-F. Hu, T.-S. Lin, S. Rafai, C. Misbah, Spontaneous locomotion of phoretic particles in three dimensions, Physical Review Fluids 7 (3) (2022). doi:10.1103/PhysRevFluids.7.034003.DOI
  58. [58]NVIDIA, cuFFT Library User’s Guide, NVIDIA (2022).
  59. [59]J. A. Crow, Quadrature of integrands with a logarithmic singularity, Mathematics of Computation 60 (201) (1993) 297–301.DOI
  60. [60]Y. Liu, L. Zhang, X. Wang, W. K. Liu, Coupling of navier-stokes equations with protein molecular dynamics and its application to hemodynamics, International Journal for Numerical Methods in Fluids 46 (2004) 1237–1252. doi:10.1002/d.798.DOI
  61. [61]Y. T. Peet, P. F. Fischer, Stability analysis of interface temporal discretization in grid overlapping methods, SIAM Journal on Numerical Analysis 50 (6) (2012) 3375–3401. doi:10.1137/110831234.DOI

Paper details

vlnc:W_6uqnbwu9
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents