Jongbae Moon, Byung-sang Kim, Kum Won Cho*

1
Development of e-AIRS System(e-Science Aerospace
Integrated Research System)

• Jongbae Moon, Byung-sang Kim, Kum Won Cho
• Korea Institute of Science and Technology
  Information
• Soon-Heum Ko, Jin-ho Kim, Jae Wan Ahn, Chongam
  KimSeoul National University
• Jung-hyun Cho, Eunhye Song, Yoonhee Kim
• Sookmyung Womens University

2
Current Status

• Just completed 1st Stage
• Participating organizations
• Korea Institute of Science and Technology
  Information (KISTI)
• Seoul National University(SNU)
• Sookmyung Womens University(SMWU)
• Industry and university collaboration
  network(Korea Aerospace-Net)
• Network based remote research environment
• Research and development for next generation
  aerospace vehicle
• Grid-based collaboration
• KISTI, KARI, KAI, Universities(7)
• International Collaboration
• NCSA
• Co-development of CoaxSim, Aerospace gateway WG
• The University of Manchester and BAE System
• Collaboration for Remote Wind Tunnel Experiment
• Wayne State University

3
e-AIRS Architecture
4
Construction of Web Portal

• Web Portal
• User Session Management
• Maintain user sessions in the portal
• Make use of existing Portlet API
• User Data Management
• Provides various types of data such as users job
  information, result file

• Implement e-AIRS Web Portal based on Gridsphere
  Framework
• Reusing the basic Gridsphere portlets
• Added e-AIRS portlets

5
Flow of Portal Services

• Service overview

parametric study service
multiple parameter generation
CFD data
simultaneous simulation
user
CFD service
result validation optimization
exp. data
remote experiment
computingresources
enlarged CFD data set
computational / experimental aerodynamic data
storage server
6
CFD Solver Service

• Features of CFD Solver
• Automatic job script generation in Grid
  environment
• Major input variables
• Mach Number
• Reynolds Number
• Angle of Attack
• Governing Equations Inviscid / Laminar /
  Turbulent
• Automatic partitioning independent to the number
  of processors

Surface Pressure around Smart UAV(Mach No. 0.84,
Angle of Attack 3.06)
Validation RAE 2822 Airfoil
7
PSS Parametric Study Service

• Integrated Analysis Environment
• Aerodynamic coefficients at variousflow
  conditions are supported bycomputing and
  experimental services
• Multiple results of experiments and computations
  plotted on the same frame
• Variation of aerodynamic coefficientswith the
  change of specific flow variablegraphically
  supported
• Automatic high throughput computingon selected
  flow range, giving detailedaerodynamic
  characteristics
• More valuable to the RD field

Simulation Results of Testcases
Plot Service of Aerodynamic Coefficients(PSS
Validation)
8
Flow of Parametric Study Service (1/2)

• Service overview
• Automatic parameter and job generation
• Job scheduling and dispatching
• Meta-data management

9
Flow of Parametric Study Service (2/2)

• Internal Operation Flow

PSSInterface
PSS Parameter Parser(PPP)
PSS Info Mgr
eAIRSPortalGUI
createSubcases

• Parameter Sweep Element, Case Info

caseinfo
PSS Job Scheduler (PJS)

• Create job directory and parameter file on PSS
  server

caseInfo
createJob

• Get available resources by using RIS
• Select a suitable resource
• Create RSL File to run jobs on selected resource

scheduleJob
resourceInfo
MMS
PSS Job Dispatcher (PJD)
prepareJob

• Create temporary directory in each node
• Transfer files to the nodes

jobInfo
executeJob

• Execute jobs and manage result data

jobInfo
10
Remote Conferencing Tools

• AG (Access Grid) TightVNC
• Add Shared TightVNC Service into AG as a Shared
  Application
• While existing Shared Desktop shares a full
  screen of remote server,
• the Shared TightVNC provides the function
  that shares a specific frame on the remote
  screen.
• Sharing the visible results through Shared
  TightVNC Client

11
CFD Monitoring Scheduling

• Runtime Monitoring
• Provides information such as status, resource,
  start time, and end time of selected Case
• Plan
• Implement a function that interrupt a selected
  job
• Convergence Graph
• Visualize a graph of interim results while job is
  running
• Using JFree Chart Library

12
Flow of CFD Service

• Overview of CFD simulation

user
e-AIRSview
computationalresult
monitoring service
job information
computationalresult
mesh file, boundary condition file
CAD data to mesh boundary condition
storage server
e-AIRSmesh
mesh generation boundary condition
users choice of mesh
mesh file
flow condition parameters
parameter file
computingresources
CFD solver service
13
Flow of Remote Experiment Service

• Service overview

requestinformation
request
save
data save
session request service
session managing service
data managementsystem
client
downloadof exp. condition
upload result
information
information
information
session information service
operator
exp. data
experiment
wind tunnel
14
e-AIRS Computing Resources
KREONET
15
e-AIRS Computing Resources

• Specification of computing facilities

Resources Spec Location
PortalServer P-4 3.2 GHz (Dual CPUs) KISTI, Daejeon
Storage Server Xeon 3.0 GHz (Dual CPUs) KISTI, Daejeon
DatabaseServer P-4 3.2 GHz (Dual CPUs) KISTI, Daejeon
ComputingNodes P-3 1.4 GHz, 20 Processors Seoul National University, Seoul
ComputingNodes Xeon 3.0 GHz, 8 Processors KISTI, Daejeon
ComputingNodes Xeon 3.0 GHz, 128 Processors PRAGMA (AIST, JAPAN)
ComputingNodes P-3 1.7 GHz, 16 Processors PRAGMA (KISTI, KOREA)
16
Performance enhancement of e-AIRS

• We checked several components
• - DB Access
• - Server
• - Response time (Transaction .vs. No. Virtual
  user, throughput test, page view and etc)
• Separate DB/Portal server and Storage Sever
• DB postgres ? mysql

17
Computational Components (1/3)

• CAD2MESH e-AIRSmesh
• CAD2MESH
• Extraction of geometryfrom CAD data (VRML file)
• Java applet based
• Middle tangent calculation methodology
• ? For extraction of indexed lines
• Ferguson curve fitting algorithm
• ? Mesh point generation algorithm
• e-AIRSmesh
• Mesh generator based on Java applet
• JOGL (Java API for OpenGL) library
• Easy to use Icon interface
• Easy display control - by mouse motion
• Transfinite interpolation technique
• ? For construction of mesh system
• PLOT3D data format
• Automatic production of NACA 4-digit airfoil

18
Computational Components (2/3)

• CFD Solvers
• Supporting three type of solvers for general
  purpose (finished)
• 3-D Turbulence solver
• 2-D Compressible solver
• 2-D Incompressible solver
• Two solvers for specific problems (on-going)
• For rocket simulation with internal flow analysis
• Combination of external flow analysis code and
  combustion simulation code
• All developed by Aerodynamic Simulation Design
  Lab. (S.N.U.)
• Validated by various problems

19
Computational Components (3/3)

• Methods of e-AIRS Visualization
• e-AIRSview
• Java applet based
• JOGL (Java API for OpenGL) library
• Visualization tool of temporary and final CFD
  data
• Easy display control by mouse motion
• Data Visualization by TecPlot
• Support of Tecplot (Commercial S/W) file format
• Download result and error history file

20
Utilization of e-AIRS Service

• Cyber Education
• Lecture materials for students (Seoul Natl
  University and Konkuk University )
• Lecture demo contents
• CFD simulation 2-Dimensional analysis through
  e-AIRS portal
• Experiment Through AG, remote lecture on
  introduction of wind tunnels

CFD analysis lecture and demo
Remote experiment lecture and demo
Main CFD analysis problems
21
Advanced Research Activities

• Experiment CFD simulation research activities
• Joint experimental research
• Support advanced rocket simulation
• Enabled only on the e-Science environment
• Example Integrated simulation of
  internal/external flow of a rocket
• Distributed research on the e-Science environment
• Simulation of rocket engine internal flows and
  rocker motion external flows and its integration

Integrated Simulation Example of Aerospace Field
on the e-Science Environment
22
Thank you very much for your attentionhttp//ea
irs.kisti.re.kr/gridsphere/gridsphere