Files in this item

FilesDescriptionFormat

application/pdf

application/pdfHUQ-THESIS-2020.pdf (2MB)Restricted to U of Illinois
(no description provided)PDF

Description

Title:Improved performance of the hPIC particle-in-cell code by incorporating a new non-uniform implicit mesh based on the PUMI infrastructure
Author(s):Huq, Md Fazlul
Advisor(s):Curreli, Davide
Contributor(s):Ruzic, David N
Department / Program:Nuclear, Plasma, & Rad Engr
Discipline:Nuclear, Plasma, Radiolgc Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:M.S.
Genre:Thesis
Subject(s):hPIC
Particle-In-Cell
Implicit Mesh
Non-uniform Mesh
Speed-up
Large Plasma Domain.
Abstract:Uniform structured meshes are inefficient at capturing the high plasma gradients in a Scrape-Off-Layer (SOL). Especially in the magnetic and electrostatic sheaths forming in front of the material surfaces. The hPIC Particle In Cell code developed previously at UIUC has been modified to incorporate a new implicit, parameterized, non-uniform mesh using the Parallel Unstructured Mesh Infrastructure (PUMI) library developed at RPI to resolve the large field gradients in the plasma sheath region. The implicit nature of the mesh allows it to define meshes with the minimal number of parameters and generate all mesh information on-the-fly. The implicit mesh is particularly advantageous for large meshes, because it allows the avoidance of storing explicitly a large mesh, thus reducing the memory footprint, with a resulting speed up of the execution. The field solver for hPIC has been updated to comply with mesh non-uniformity. On top of the non-uniformity, the algorithm allows the splitting of the entire plasma domain into a number of blocks, termed as submeshes. The submeshes can be either uniform or non-uniform, of boundary layer type, with mesh size following a geometric gradation biased at the left or at the right node. The performance measurement of the non- uniform multi-block PUMI mesh based hPIC has been done for different domain size and mesh configurations. For a smaller domain size of 500 Debye lengths, a speed-up of nearly 2 is found, still maintaining an L2 norm of the error less than 1%. A speed-up of up to 16 is achieved with L 2 norm of error less than 1.1%. For a large plasma domain of 1.5 m, a speed-up of more than 100 is achieved. This work opens up the possibility of simulating large plasma domains in a full-orbit Particle-in-Cell at a reasonable computational time.
Issue Date:2020-08-21
Type:Thesis
URI:http://hdl.handle.net/2142/109460
Rights Information:Copyright 2020 Md Fazlul Huq
Date Available in IDEALS:2021-03-05
Date Deposited:2020-12


This item appears in the following Collection(s)

Item Statistics