We simulate by lattice Boltzmann the steady shearing of a binary fluid mixture undergoing phase separation with full hydrodynamics in two dimensions. Contrary to some theoretical scenarios, a dynamical steady state is attained with finite domain lengths Lx,y in the directions (x, y) of velocity and velocity gradient. Apparent scaling exponents are estimated as Lx ∼γ −2/3 and Ly ∼γ −3/4 . We discuss the relative roles of diffusivity and hydrodynamics in attaining steady state.PACS numbers: 47.11.+jSystems that are not in thermal equilibrium play a central role in modern statistical physics, and arise in areas ranging from soap manufacture to subcellular biology [1]. Such systems include two important classes: those that are evolving towards Boltzmann equilibrium (e.g., by phase separation following a temperature quench), and those that are maintained in nonequilibrium by continuous driving (such as a shear flow). Of fundamental interest, and surprising physical subtlety, are systems combining both features -such as a binary fluid undergoing phase separation in the presence of shear. Here it is not known [2,3] whether coarsening continues indefinitely, as it does without shear, or whether a steady state is reached, in which the characteristic length scales L x,y,z of the fluid domain structure attain finiteγ-dependent values at late times. (We define the mean velocity as u x =γy so that x, y, z are velocity, velocity gradient and vorticity directions respectively;γ is the shear rate.) Experimentally, saturating length scales are reportedly reached after a period of anisotropic domain growth [2,4]. However, the extreme elongation of domains along the flow direction means that, even in experiments, finite size effects could play an essential role in such saturation [5]. Theories in which the velocity does not fluctuate, but does advect the diffusive fluctuations of the concentration field, predict instead indefinite coarsening, with length scales L y,z scaling asγ-independent powers of the time t since quench, and (typically) L x ∼γtL y [5]. In real fluids, however, the velocity fluctuates strongly in nonlinear response to the advected concentration field, and hydrodynamic scaling arguments, balancing either interfacial and viscous or interfacial and inertial forces, predict saturation (e.g., L ∼γ −1 or L ∼γ −2/3 ) [3,6,7]. Given these experimental and theoretical differences of opinion, computer simulations of sheared binary fluids, with full hydrodynamics, are of major interest.The aforementioned scaling arguments cannot really distinguish one Cartesian direction from another, but even in theories that can do so, a two dimensional (2D) representation, suppressing z, is expected to capture the main physics [5]. (Without shear, subtle non-scaling effects arise in 2D from the formation of disconnected droplets [8], but shear seems to suppress these [9].) Performing simulations in 2D is therefore a fair compromise, especially given the extreme computational demands of the full 3D problem [3,10]. But, apart from [9,11]...

Abstract. We present a progress report on our work on lattice Boltzmann methods for colloidal suspensions. We focus on the treatment of colloidal particles in binary solvents and on the inclusion of thermal noise. For a benchmark problem of colloids sedimenting and becoming trapped by capillary forces at a horizontal interface between two fluids, we discuss the criteria for parameter selection, and address the inevitable compromise between computational resources and simulation accuracy.

No abstract

No abstract

We simulate by lattice Boltzmann the steady shearing of a binary fluid mixture with full hydrodynamics in three dimensions. Contrary to some theoretical scenarios, a dynamical steady state is attained with finite correlation lengths in all three spatial directions. Using large simulations we obtain at moderately high Reynolds numbers apparent scaling exponents comparable to those found by us previously in 2D. However, in 3D there may be a crossover to different behavior at low Reynolds number: accessing this regime requires even larger computational resource than used here. PACS numbers: 64.75.+g, 47.11.Qr Systems that are not in thermal equilibrium play a central role in modern statistical physics [1]. They include two important classes: those evolving towards Boltzmann equilibrium (e.g., by phase separation following a temperature quench), and those maintained in nonequilibrium by continuous driving (such as a shear flow). Of fundamental interest, and surprising physical subtlety, are systems combining both features -such as a binary fluid undergoing phase separation in the presence of shear. Here a central issue [2,3] is whether coarsening continues indefinitely, as it does without shear, or whether a nonequilibrium steady state (NESS) is reached, in which the characteristic length scales L x,y,z of the fluid domain structure attain finiteγ-dependent values at late times. (We define the mean velocity as u x =γy so that x, y, z are velocity, velocity gradient and vorticity directions respectively;γ is the shear rate.)Our recent simulations, building on earlier work of others [4,5], have shown that in two dimensions (2D), a NESS is indeed achieved [6]. In 3D, the situation is more subtle. Fourier components of the composition field whose wavevectors lie along the vorticity direction feel no direct effect of the mean advective velocity [2,7]. Therefore it might be possible for coarsening to proceed indefinitely by pumping through tubes of fluid oriented along z [3]. Another crucial difference is that in 2D fluid bicontinuity is possible only by fine tuning to a percolation threshold at 50:50 composition (assuming fluids of equal viscosity) so that the generic situation is one of droplets. (Indeed, for topological reasons, droplets are implicated even at threshold [4].) In contrast, in 3D both fluids remain continuously connected across the sample throughout a broad composition window either side of 50:50.In 3D experiments, saturating length scales are reportedly reached after a period of anisotropic domain growth [2,8]. However, the extreme elongation of domains along the flow direction means that, even in experiments, finite size effects could play a role in such saturation [9]. Theories in which the velocity does not fluctuate, but does advect the diffusive fluctuations of the concentration field, predict instead indefinite coarsening, with length scales L y,z scaling asγ-independent powers of the time t since quench, and (typically) L x ∼γtL y [9]. As emphasized in [6], in real fluids, however, the velocit...

No abstract

We describe some scaling issues that arise when using lattice Boltzmann methods to simulate binary fluid mixtures -both in the presence and in the absence of colloidal particles. Two types of scaling problem arise: physical and computational. Physical scaling concerns how to relate simulation parameters to those of the real world. To do this effectively requires careful physics, because (in common with other methods) lattice Boltzmann cannot fully resolve the hierarchy of length, energy and time scales that arise in typical flows of complex fluids. Care is needed in deciding what physics to resolve and what to leave unresolved, particularly when colloidal particles are present in one or both of two fluid phases. This influences steering of simulation parameters such as fluid viscosity and interfacial tension. When the physics is anisotropic (for example, in systems under shear) careful adaptation of the geometry of the simulation box may be needed; an example of this, relating to our study of the effect of colloidal particles on the Rayleigh-Plateau instability of a fluid cylinder, is described. The second and closely related set of scaling issues are computational in nature: how do you scale up simulations to very large lattice sizes? The problem is acute for systems undergoing shear flow. Here one requires a set of blockwise co-moving frames to the fluid, each connected to the next by a LeesEdwards like boundary condition. These matching planes lead to small numerical errors whose cumulative effects can become severe; strategies for minimising such effects are discussed.

This work presents the application of reinforcement learning for the optimal resistive control of a point absorber. The model-free Q-learning algorithm is selected in order to maximise energy absorption in each sea state. Step changes are made to the controller damping, observing the associated penalty, for excessive motions, or reward, i.e. gain in associated power. Due to the general periodicity of gravity waves, the absorbed power is averaged over a time horizon lasting several wave periods. The performance of the algorithm is assessed through the numerical simulation of a point absorber subject to motions in heave in both regular and irregular waves. The algorithm is found to converge towards the optimal controller damping in each sea state. Additionally, the model-free approach ensures the algorithm can adapt to changes to the device hydrodynamics over time and is unbiased by modelling errors.