Application of GPU processing for Brownian particle simulation

Way Lee Cheng, Ali Sheharyar, Reza Sadr, Othmane Bouhali

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Reports on the anomalous thermal-fluid properties of nanofluids (dilute suspension of nano-particles in a base fluid) have been the subject of attention for 15 years. The underlying physics that govern nanofluid behavior, however, is not fully understood and is a subject of much dispute. The interactions between the suspended particles and the base fluid have been cited as a major contributor to the improvement in heat transfer reported in the literature. Numerical simulations are instrumental in studying the behavior of nanofluids. However, such simulations can be computationally intensive due to the small dimensions and complexity of these problems. In this study, a simplified computational approach for isothermal nanofluid simulations was applied, and simulations were conducted using both conventional CPU and parallel GPU implementations. The GPU implementations significantly improved the computational performance, in terms of the simulation time, by a factor of 1000-2500. The results of this investigation show that, as the computational load increases, the simulation efficiency approaches a constant. At a very high computational load, the amount of improvement may even decrease due to limited system memory.

Original languageEnglish
Pages (from-to)39-47
Number of pages9
JournalComputer Physics Communications
Volume186
DOIs
Publication statusPublished - 1 Jan 2015

Fingerprint

Fluids
Processing
simulation
Program processors
fluids
Physics
Heat transfer
Data storage equipment
Computer simulation
Graphics processing unit
heat transfer
physics
interactions
Hot Temperature

Keywords

  • CFD
  • GPU
  • Multiphase flows
  • Nanofluids
  • Simulations

ASJC Scopus subject areas

  • Physics and Astronomy(all)
  • Hardware and Architecture

Cite this

Application of GPU processing for Brownian particle simulation. / Cheng, Way Lee; Sheharyar, Ali; Sadr, Reza; Bouhali, Othmane.

In: Computer Physics Communications, Vol. 186, 01.01.2015, p. 39-47.

Research output: Contribution to journalArticle

@article{5f7253c6653c445089587636d36f3264,
title = "Application of GPU processing for Brownian particle simulation",
abstract = "Reports on the anomalous thermal-fluid properties of nanofluids (dilute suspension of nano-particles in a base fluid) have been the subject of attention for 15 years. The underlying physics that govern nanofluid behavior, however, is not fully understood and is a subject of much dispute. The interactions between the suspended particles and the base fluid have been cited as a major contributor to the improvement in heat transfer reported in the literature. Numerical simulations are instrumental in studying the behavior of nanofluids. However, such simulations can be computationally intensive due to the small dimensions and complexity of these problems. In this study, a simplified computational approach for isothermal nanofluid simulations was applied, and simulations were conducted using both conventional CPU and parallel GPU implementations. The GPU implementations significantly improved the computational performance, in terms of the simulation time, by a factor of 1000-2500. The results of this investigation show that, as the computational load increases, the simulation efficiency approaches a constant. At a very high computational load, the amount of improvement may even decrease due to limited system memory.",
keywords = "CFD, GPU, Multiphase flows, Nanofluids, Simulations",
author = "Cheng, {Way Lee} and Ali Sheharyar and Reza Sadr and Othmane Bouhali",
year = "2015",
month = "1",
day = "1",
doi = "10.1016/j.cpc.2014.09.013",
language = "English",
volume = "186",
pages = "39--47",
journal = "Computer Physics Communications",
issn = "0010-4655",
publisher = "Elsevier",

}

TY - JOUR

T1 - Application of GPU processing for Brownian particle simulation

AU - Cheng, Way Lee

AU - Sheharyar, Ali

AU - Sadr, Reza

AU - Bouhali, Othmane

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Reports on the anomalous thermal-fluid properties of nanofluids (dilute suspension of nano-particles in a base fluid) have been the subject of attention for 15 years. The underlying physics that govern nanofluid behavior, however, is not fully understood and is a subject of much dispute. The interactions between the suspended particles and the base fluid have been cited as a major contributor to the improvement in heat transfer reported in the literature. Numerical simulations are instrumental in studying the behavior of nanofluids. However, such simulations can be computationally intensive due to the small dimensions and complexity of these problems. In this study, a simplified computational approach for isothermal nanofluid simulations was applied, and simulations were conducted using both conventional CPU and parallel GPU implementations. The GPU implementations significantly improved the computational performance, in terms of the simulation time, by a factor of 1000-2500. The results of this investigation show that, as the computational load increases, the simulation efficiency approaches a constant. At a very high computational load, the amount of improvement may even decrease due to limited system memory.

AB - Reports on the anomalous thermal-fluid properties of nanofluids (dilute suspension of nano-particles in a base fluid) have been the subject of attention for 15 years. The underlying physics that govern nanofluid behavior, however, is not fully understood and is a subject of much dispute. The interactions between the suspended particles and the base fluid have been cited as a major contributor to the improvement in heat transfer reported in the literature. Numerical simulations are instrumental in studying the behavior of nanofluids. However, such simulations can be computationally intensive due to the small dimensions and complexity of these problems. In this study, a simplified computational approach for isothermal nanofluid simulations was applied, and simulations were conducted using both conventional CPU and parallel GPU implementations. The GPU implementations significantly improved the computational performance, in terms of the simulation time, by a factor of 1000-2500. The results of this investigation show that, as the computational load increases, the simulation efficiency approaches a constant. At a very high computational load, the amount of improvement may even decrease due to limited system memory.

KW - CFD

KW - GPU

KW - Multiphase flows

KW - Nanofluids

KW - Simulations

UR - http://www.scopus.com/inward/record.url?scp=84922239518&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84922239518&partnerID=8YFLogxK

U2 - 10.1016/j.cpc.2014.09.013

DO - 10.1016/j.cpc.2014.09.013

M3 - Article

VL - 186

SP - 39

EP - 47

JO - Computer Physics Communications

JF - Computer Physics Communications

SN - 0010-4655

ER -