Session #2 Abstracts

Towards Adaptive Mesh Refinement (AMR) and Particle-in-cell (PIC) for Plasma Simulation

Bei Wang, Greg Miller, Phillip Colella

PIC has been successfully used for plasma simulation in the last two decades. The extension of PIC with AMR has been explored recently, however, with the observation of large self-force on the particles around the refinement interface. Although the approach to control the self-force is proposed for one-particle model, robust choices of numerical parameters and numerical tests for many-particle model haven't been addressed. Our starting point for addressing those questions is the solving of Poisson equation with infinite domain condition, since we currently concentrate on electrostatic PIC model.
We employ Jame-Lackner's boundary potential method for infinite Poisson problem. In this method, the infinite Poisson problem is represented by a Poisson problem with appropriate Dirichlet boundary condition. The Dirichlet boundary values can be found by solving two Dirichlet problems, plus a boundary-to-boundary convolution. For efficiency, we implement this by solving one of the Dirichlet problems with fast multigrid-based Poisson solver and another one with fast Fouries transformation. The boundary-to-boundary convolution is calculated using fast multipole method. The code has been developed based on Chombo library, a software package applying finite difference methods for the solution of partial differential equations on a hierarchy of locally refined grids. We compare the exact solution with the computed solution in both 2D and 3D test cases. Second order accuracy in L1 has been demonstrated in the tests.
With the infinite domain Poisson problem successfully solved on a hierarchy of refined grids, our next step will be coupling particle properties with grid ones. This includes depositing the charge defined on particles to grids and interpolating the force calculated on the grids back to the particles. Carefully approach will be used to control the self-force on the particles located in the neighborhood of coarse-fine interface. After rigorously analyzing the errors and running a bunch of numerical tests, the mechanism of choosing robust parameters will be developed. At last, the algorithms will be used for simulating real beam problem and verified through experiment results.

A Higher Order Approach to Fluid-Particle Coupling in Microscale Polymer Flows

Bakytzhan Kallemov, Greg Miller, D.Trebotich

The dynamics of a continuum fluid with discrete embedded polymers is important for certain microfluidic applications,(e.g., so-called lab-on-a-chip PCR reactors)and for modeling viscoelastic phenomena in the dilute limit. Toward this end we proposed a fluid-particle coupling strategy that uses Brownian dynamics to approximate molecular-level fluid--polymer interactions. To simulate polymer flows in microscale environments we have developed a numerical method that couples stochastic particle dynamics with an efficient incompressible Navier-Stokes solver. Here, we examine the convergence properties of the stochastic particle solver alone, and demonstrate that it has second order convergence in both weak and strong senses, for the examples presented. In this work we consider the framework of a freely-jointed chain (no polymer-polymer interactions)and the fluid velocity field to be prescribed.

Towards Single Crystals Using Strain Induced Grain Growth

Corentin Guebels, Tien Tran, Ben Fell, Joanna Groza

The microstructural evolution and kinetics of thermal and strain-induced grain growth were analyzed using a two dimensional Monte Carlo Potts (MCP) model. MCP simulations have captured various grain growth behaviors; however, the stochastic nature of the MCP time-step limits its verification with physical data. An energy-controlled relationship between MCP and physical time-steps was established using thermal grain growth simulations and published experimental data. The case of static strain- induced growth is considered next. An internal energy term is introduced to model the elastic stresses caused by the strain-induced dislocations in the microstructure. The modeling strategies for the strain-induced energy, recovery and nucleation and their implementations are discussed in connection with experimental evidence.

Back to conference program
Back to conference main page