university of illinois at urbana-champaign university of illinois at urbana-champaign (uiuc)...
TRANSCRIPT
University of Illinois at Urbana-Champaign
UNIVERSITY OF ILLINOIS at URBANA-CHAMPAIGN (UIUC)
NATIONAL CENTER FOR SUPERCOMPUTING APPLICATIONS (NCSA)
ASC PET CSM REVIEW and ASC PET CSM REVIEW and CY4 & CY5 PLANNINGCY4 & CY5 PLANNING
Briefing Document October 7, 1999
Harry H. HiltonProfessor Emeritus of Aeronautical & Astronautical Engineering, UIUC
ASC CSM Senior Academic Lead, NCSA(217-333-2653 [email protected])
University of Illinois at Urbana-Champaign
NCSA CSM ASC Personnel
Harry Hilton - Senior Academic Lead (25%)
Cristina Beldica - Research Scientist (95%)
LeRay Dandy - Senior System Programmer (15%)
Dave O’Neal - On-Site Lead (100%)
Seid Koric - Graduate Student (50%) *§
Gyuseok Kwak - Graduate Student (50%) *
Youngjin Woo - Graduate Student (50%) *§_______________* Funded by NCSA§ Till July 1999
University of Illinois at Urbana-Champaign
NCSA CSM ASC Personnel
Tom I. PrudhommeSenior Associate Director
E. J. GrabertASC / ARL Program Manager
Harry H. HiltonSenior Academic Lead
Cristina E. BeldicaResearch Scientist
LeRay T. DandySr. System Programmer
David C. O’NealOn-Site Lead
CSM - THE THREE COMPONENTS
University of Illinois at Urbana-Champaign
STRUCTURAL / SOLIDMECHANICS
COMPUTATIONALSCIENCE
MATERIALSCIENCE
University of Illinois at Urbana-Champaign
CSM - PET Program Benefits to DoD
University of Illinois at Urbana-Champaign
ASC & NCSA Personnel Collaborations
UIUC faculty is engaged in leading edge research and offers world class new ideas and solutions to DoD, ASC, ARL and the industry
Development of and modifications to existing codes for portable parallel computing and integrated visualization
New and unique analytical research and computational protocols in fundamental areas with applications to materials and composite flight structures under aeroelastic loads
University of Illinois at Urbana-Champaign
ASC & NCSA Collaborations (cont’d)Promoting and executing unified cross CTA research and development projects through multidisciplinary analytical modeling, analyses and computational simulations
Advancing levels of interaction between groups, directorates and MSRCs through integration of multiprocessor application codes that traverse and ultimately span MSRCs (metacomputing)
Training, short courses and workshops tailored to specific CSM interests of DoD, ASC, ARL, HBUC & MI researchers, scientists and engineers
None of this work is being done elsewhere
University of Illinois at Urbana-Champaign
ASC CSM CY4 & CY5 PROJECTS
CSM 1 Outreach, Training & Support (Funded)
CSM 2 Visualization (Funded)
CSM 3 Tool Development (Funded)
CSM 4 A 3 D Structural Solver for FDL3DI CFD Code (F)
CSM 5 Light Weight Advanced Composite Structures (F)
CSM 6 Portability and Performance Improvement of the Micromechanics Damage Models (Proposal)
CSM 7 Multidisciplinary Outreach Program UIUC & UNM (P)
University of Illinois at Urbana-Champaign
Summary of CY 3 - 5 Cooperative Projects
• Code Parallelization for Damage Model (Materials Lab)
• Multidisciplinary Collaboration CSM-CFD Code Integration (Flight Dynamics) • Air Vehicle Survivability Study (VAVS)
• Investigation and Implementation of Multidisciplinary Techniques (AFRL)
• Analytical and Computational Modeling of Advanced Composite Flight Structures and Materials (AFRL)
• Diversified CSM On-site Training Efforts, Conferences, Workshops, Short Courses and Symposia (ASC/ARL)
University of Illinois at Urbana-Champaign
Current Cooperative NCSA - ASC CSM Projects
• Axisymmetric Damage Model (Materials Lab) Multithread porting project
Performance optimization through integration into iSightDoD Users Group Conferences and SC’98 presentation
• Multidisciplinary Collaboration (Flight Dynamics) Integration of 3D FEM solver into CFD research codeParametric finite element model of an aircraft wing
• Air Vehicle Survivability Study (VAVS)Parametric finite element analysis of structural damage
in aircraft wings due to internal explosionsGrand Challenge project proposal for prediction of
aircraft wing damage due to ballistic impacts
University of Illinois at Urbana-Champaign
• Investigation and Implementation of Multidisciplinary Techniques
Software evaluation (CAPTools, Physica, MDICE, iSight)Organized MAPINT’98 & ‘99
• Modeling of Advanced Flight Structures & Materials (AFRL)Performance evaluations of massively parallel visco- elastic FE / FD computations for compositesStructural reliability & probability of failure analysesStructural health monitoring of flight structuresAnisotropic thermo-viscoelastic FE / FD analyses of
composites (manufacturing & service conditions)Piezoelectric control of aero-viscoelastic structural
phenomena (divergence, flutter, control surface effectiveness, aerodynamic noise, damping)
Analytical material characterization (optimum properties)Analytical & computational design of material property
determination experiments
University of Illinois at Urbana-Champaign
CSM Outreach, Training & Conferences
• Diversified CSM On-site and Cross-site Training Efforts
• Workshops, Short Courses, Conferences and Symposia
Presentations at conferences organized by DoD or with direct DoD involvement
Participation in conferences outside DoD sponsored by national & international scientific societies
Papers published in archival scientific / engineering journals & conference proceedings
Posting research publications and results on the CSM PET web site
Millennium NCSA-PET Workshop on CSM Strategic Planning for the next Quarter Century
University of Illinois at Urbana-Champaign
UIUC/NCSA CSM ASC Research Activities
1997 1998 1999 Archival journals andproceedings papers
Published or accepted 16 25 23
Submitted, not yet accepted 11
Presentations at nationaland international 11 17 12Conferences (not all trips charged to PET)
University of Illinois at Urbana-Champaign
CSM Core Support Ongoing Activities
Web Pages Development and Support • Promote user productivity through web based access to
information, tools and training
• Local interrelated sites on NCSA, ASC, ARL, DoD, HBUC & MI web servers
• Showcase for PET projects and accomplishments
• Provide links to resources for DoD researchers, scientists and engineers
Software and hardware available at NCSA and the MSRCsCalendar of conferences, workshops and training eventsPublications and presentations by NCSA PET members
University of Illinois at Urbana-Champaign
• SoftwareCommercial codesFree research softwareScientific librariesManualsPerformance evaluations
• PublicationsAuthored by members of the PET CSM groupCollaborations with ASC, ARL, industry scientific
staffs and university faculties at Clark Atlanta, Delaware, Florida Atlantic, New Mexico, Ohio State, Rice, UIUC, Greenwich, Singapore
Original papers in archival journals & proceedings
• Other web resources
ASC CSM MULTIDISCIPLINARY PROJECTS
AEROELASTICITY & MATERIAL DAMPING
STRUCTURAL 3D SOLVER FOR FDL3DI CFD CODE
PIEZOELECTRIC & STRUCTURAL CONTROL
MULTISCALE CCM & CSM MATERIAL MODELING
MULTIDISCIPLINARY OUTREACH PROGRAMS at UIUC & UNM and UIUC & CAU
University of Illinois at Urbana-Champaign
Cristina Beldica, Harry Hilton, Seid Koric & David O’Neal - UIUC NCSA
Nicholas Pagano - AFRL
David Veazie - Clark Atlanta U
Jack Vinson - U of Delaware
Sung Yi - Nanyang Technological U, Singapore
CSM 5: AEROELASTIC, PIEZO-VISCOELASTIC &
STRUCTURAL CONTROL OF COMPOSITE
FLIGHT STRUCTURES
University of Illinois at Urbana-Champaign
• Objective: To make use of smart composite materials to regulate deformations, stresses & failure probabilities through structural, material and piezoelectric control
• Methodology: Multiphase and multidisciplinary analytical formulations and computational simulations of viscoelastic material and piezoelectric constitutive relations and actual scientific and engineering problems
• Results: Deterministic and probabilistic stress, defor-mation and failure solutions for flight vehicle metal and composite structures to avoid excessive vibrations, noise transmission and undesirable aeroelastic effects and to predict structural survival times. (See web site)
• Significance: Provide realistic modern analytical and computational models and protocols to dampen un-wanted deformations and noise responses and to furnish structural survival probabilities and times to failure to be used by DoD, ARL and ASC researchers and designers
CSM 5: Viscoelastic Material & Piezoelectric Damping
-25
-20
-15
-10
-5
0
-0.2 0 0.2 0.4 0.6 0.8 1
PROBABILITY OF DELAMINATION ONSET
PIEZO CONTROL @ 800C
NO CONTROL @ 200C
PIEZO CONTROL @ 200C
NO CONTROL @ 800C
LO
G (D
ELA
MIN
ATIO
N P
RO
BA
BIL
ITY
)
TIME (t / )
0.017
0.018
0.019
0.02
0.021
0.022
0.023
0.024
0.025
Defl
ecti
on
(in
)
0 20 40 60 80 100
Time (sec)
00.2
0.40.6
0.81
0.4
0.6
0.8
10
0.2
0.4
0.6
0.8
1
J1 / J1maxJ2 / J2max
J3 /
J3m
ax
Fig. 3 INVARIANT FAILURE SURFACE
• Objective: To formulate and evaluate performances of parallel computational protocols for linear and nonlinear viscoelastic stress, failure and structural control analyses
• Methodology: Commercial codes such as ANSYS and ABAQUS with modified subroutines are used for solving real engineering problems. For nonlinear viscoelasticity, new codes have to be written because no available commercial software has such capabilities
• Results: ABAQUS scales well for large number of DOF, but needs modifications to run efficiently on more than 16 parallel processors (See web site)
• Significance: Provide realistic modern efficient compu-tational protocols for structural analysis, survival probabilities and times to failure. Of interest to DoD, ARL, ASC & industry engineers and scientists
CSM 5: Massively Parallel Piezoelectric / Viscoelastic Simulations
0
1000
2000
3000
4000
5000
6000
-1
0
1
2
3
4
5
0 5 10 15 20
PERFORMANCE EVALUATIONS FOR 201,000 DOF
CPU TIME - BENDINGCPU TIME - TORSION
SPEED UP - BENDINGSPEED UP - TORSION
TO
TA
L TIM
E (s
ec)
SP
EED
UP
NUMBER OF CPU
201,000 DOF FEM
1000
1500
2000
2500
3000
3500
4000
4500
-25
-20
-15
-10
-5
0
5
10
-4 -3 -2 -1 0 1 2
Fig. 4 CANTILEVER BEAM TOP FIBER STRESSES
11
PIEZO
11
BEAM
12
PIEZO
12
BEAM
BEN
DIN
G S
TR
ESS
ES
SH
EA
R S
TR
ESS
ES
LOG (TIME)
• Objective: To investigate applicability of viscoelastic damping & piezoelectric effects, torsional divergence, flutter, aerodynamic derivatives, aileron effectiveness and aerodynamic noise in composite structures
• Methodology: Employ analytical tools and large scale computational simulations to conduct feasibility studies of light weight, low power and inexpensive piezo devices to control undesirable aeroelastic effects including aero-dynamic and structural coupling
• Results: Studies demonstrate ability to control aeroelastic effects and flight vehicle motion by application of simple sensing and actuator piezoelectric devices. (See website)
• Significance: Provide realistic modern analytical and computational models for aeroelastic and structural control of composite structures. Of interest to DoD, ARL, ASC & industry engineers and scientists
CSM 5: Aeroelasticity with Piezo-Viscoelastic Control
90.0
100.0
110.0
120.0
130.0
140.0
150.0
160.0
10 100 1000 10000
MEASURED DELTA IV NOSE CONE NOISE LEVELS
EXTERNAL
INTERNALDAMPED
AC
OU
STIC
LEV
EL (d
B)
FREQUENCY (Hz)
0
2
4
6
8
10
12
0 5 10 15 20
Fig. 3 ELECTRIC POTENTIAL FOR FIXED ELECTRIC DISPLACEMENT
D = 2.5 degD = 5.0 degD = 7.5 degD = 10 degD = 25 deg
VO
LTA
GE E
(V)
TIME (t/1)
0
5
10
15
20
25
0 1 2 3 4 5
Fig. 1 MAXIMUM ANGLE OF TWIST
VISCOELASTIC NO CONTROL
VISCOELASTIC PIEZO CONTROL
ELASTIC NO CONTROL
TW
IST A
NG
LE
(t)
/ E
TIME
V < VD
E
• Objective: To investigate influences of ramp loading functions and rise time to full load on experimental material characterizations through computer simulations
• Methodology: Employ analytical tools and large scale computational simulations to conduct feasibility studies of experimental procedures for viscoelastic deterministic & random material characterizations
• Results: Loading patterns distinctly affect the determi-nation of viscoelastic material parameters, leading to possible erroneous or misleading characterizations. Procedure for real time (not Laplace Transform) characterization is developed. (See website)
• Significance: Provide realistic and proper analytical and computational tools for material characterization of advanced composite structures. Of interest to DoD, ARL, ASC & industry engineers,researchers and scientists
CSM 5: Analytical & Experimental Material Property Determinations
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0
0.2
0.4
0.6
0.8
1
1.2
-3 -2 -1 0 1 2
Fig. 1 LOADING FUNCTIONS & RELAXATION FUNCTIONS
RELA
XA
TIO
N FU
NC
TIO
NS
LO
AD
LOG (TIME)
L1 L
2
A(t)
B(t)
C(t)
t1
t0
0.5
0.6
0.7
0.8
0.9
1.0
1.1
-8
-7
-6
-5
-4
-3
-2
0 50 100 150 200
Fig. 2 TRUE AND APPARENT YOUNG'S MODULUS
TRUE E0
APPARENT E0
FOR t1 > t
0
t0
NO
RM
ALIZ
ED
YO
UN
G'S
M
OD
ULU
S (E
0)
LO
G (t0 )
TEMPERATURE (oC)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
-4 -3 -2 -1 0 1 2
Fig. 3 LOAD RAM FUNCTIONS AND RELAXATION FUNCTION
LOAD = H ( t ) / F0
[t0 = 10-2 t
1 = 0]
t / t1 [t
0 = t
1 = 10-2]
.5 [1 - cos (* t / t1)] [t
0 = t
1 = 10-2]
( t ) / O
NO
RM
ALI
ZED
LO
AD
&
R
ELA
XA
TIO
N FU
NC
TIO
NS
LOG (TIME)
• Objective: Analytical formulations & computational simulations to determine optimum anisotropic viscoelastic designer material properties to be manufactured to meet specific structural service requirements
• Methodology: Solutions for composites are approached by an inverse method through examining individual effects of portions of relaxation modulus curves on creep and relaxation responses
• Results: A mechanistic understanding of viscoelastic responses to loading, moisture and temperature histories has been achieved which allow the designer to select tailored composite materials for actual service conditions. Protocols for proper experimental determination of material properties have also been formulated. (See website)
• Significance: Provide modern composite property models to allow ASC, ARL, DoD and industry engineering designers to select optimum tailored materials for specific structural service conditions
CSM 5: Viscoelastic Designer Materials
0.0
0.2
0.4
0.6
0.8
1.0
1.2
-10 -8 -6 -4 -2 0 2
Fig. 1 - MASTER RELAXATION MODULUS
NO
RM
ALIZ
ED
R
ELA
XA
TIO
N M
OD
ULU
S
LOG (TIME)
A
B
C
DE
0
2
4
6
8
10
12
14
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1
Fig. 5 MODIFIED RELAXATION MODULI (C0
11) IN DIRECTION (1,1)
110% C011REF
120% C011REF
130% C011REF
140% C011REF
150% C011REF
RELA
TIV
E D
IFFER
EN
CE IN
D
ISS
PA
TED
EN
ER
GY
(%
)
LOG (TIME) (sec)
-5
-2.5
0
2.5
5
-4 -2 0 2 4 6
Fig. 8 MODIFIED RELAXATION MODULI IN DIRECTION (I,J)
(1,1)
(2,2)
(3,3)
(1,2)
(1,3)
(2,3)
RELA
TIV
E D
IFFER
EN
CE D
ISS
IPA
TED
EN
ER
GY
(%
)
LOG (TIME) (sec)
(SLOPE)NEW
= 0.5 (SLOPE)OLD
University of Illinois at Urbana-Champaign
CSM Predictions of Aircraft Wing Damagedue to Ballistic Impact
Grand Challenge Proposal
J. Calcaterra - VAVS
R. Hinrichsen - Anteon Corp.
A. Palazotto - AFIT
B. Baron - AFRL
H. Hilton - UIUC NCSA
University of Illinois at Urbana-Champaign
• Numerical simulations for evaluation and prediction of end damage due to ballistic impact
• 3-D hydrocode model coupled to an accurate structural solver
• Comprehensive failure criteria
• Interactive framework for design optimization
Fluid response at t=0.0002 sec after explosion
Prediction of Aircraft Wing Damagedue to Ballistic Impact
University of Illinois at Urbana-Champaign
CSM 7: UIUC / NCSA & UNM / AHPCCMultidisciplinary CSM Outreach Program
A CSM Project Proposal
Harry Hilton, David O’Neal - UIUC NCSA
Andrew Pineda - UNM AHPCC
University of Illinois at Urbana-Champaign
CSM 7: UIUC / NCSA & UNM / AHPCCMultidisciplinary CSM Outreach Program
A two stage approach will be used to accomplish the project objective.
• First, a one day symposium will be organized featuring presentations by AHPCC and NCSA CSM teams augmented by research groups from AFRL Phillips AFB & WPAFB. Targeted applications will include aeroelasticity, composite structures, fracture mechanics, aerodynamics, multidisciplinary optimization and multiscale models.
• Secondly, a cross site project plan will then be completed by the NCSA and AHPCC teams to benefit research groups at Phillips and Wright-Patterson AFBs.
University of Illinois at Urbana-Champaign
CSM 6: Portability and Performance Improvement of the
Micromechanics Damage Models (MDM)
A CSM PTES Project Proposal
Cristina Beldica, David O’Neal - UIUC NCSA
Richard Luczak - Rice U
Nicholas Pagano - AFRL
University of Illinois at Urbana-Champaign
CSM 6: Portability and Performance Improvement of the
Micromechanics Damage Models (MDM)
• Execution of the MDM code suite has been restricted to HP plat- forms due to a fundamental portability problem. A solution to this problem was recently discovered by the NCSA CSM team and this project’s primary objective is to assist the AFRL development group in implementing portability onto high performance computers.
• Focused efforts will resolve portability problems while simulta- neously yielding significant performance improvements. This is a collaborative project involving AFRL developers of the MDM codes, NCSA researchers and the PET CSM and PTES site leads at ASC.
Cristina Beldica, Harry Hilton, David O’Neal - UIUC NCSA
David Veazie - Clark Atlanta U.
Margaret Hurley, Gerry Lushington, Richard Pritchard -
Ohio State U. OSC
Multiscale Materials Modeling
A CCM-CSM Proposal for Multidisciplinary and Multisite Collaboration
University of Illinois at Urbana-Champaign
• Objective: Construction of a prototyping tool capable of predicting continuum effects in composite materials from the molecular properties of the constituent materials.
• Methodology: The key to this project lies in the design of the mechanism used to bridge atomistic and continuum scale analysis tools. A materials database will also be used to validate new results, as well as catalog information associated with previous studies.
• Results: This project is still in planning. Related proposals for resources have been completed. Study of existing multiscale bridging techniques has been initiated.
• Significance: Multidisciplinary collaboration involving CCM and CSM CTAs and Clark Atlanta University. Huge potential associated with virtual prototyping.
CSM: Multiscale Materials Modeling
Microscopy experiments
University of Illinois at Urbana-Champaign
CSM 5: THE AXISYMMETRIC DAMAGE MODEL (ADM)
Cristina Beldica, D. O’Neal - UIUC NCSA
N. Pagano, G. Shoeppner - AFRL
G. P. Tandon - AdTech
K. Flurchick - OSC
University of Illinois at Urbana-Champaign
CSM 5: THE AXISYMMETRIC DAMAGE MODEL (ADM)
• Brittle Materials Lab code assist • Used to establish design properties of experimental composite materials • Current models have cut run time costs in half and enhanced post
processing capabilities • Potential for additional order of magnitude reductions in run time costs
University of Illinois at Urbana-Champaign
CSM 4: Structural Solver for FDL3DI CFD Code
L. Dandy, D. O’Neal - UIUC NCSA
M. Visbal, R. Gordnier, R. Melville - AFRL
H. Thornburg, B. Soni - Mississippi State U
University of Illinois at Urbana-Champaign
CSM 4: Structural Solver for FDL3DI CFD Code
• Objective: To integrate a compact portable 3D FE structural solver into the FDL3DI CFD code in order to predict flexible aircraft responses to aerodynamic loads and to properly couple aeroelastic phenomena
• Methodology: A linear elastic FE solver is being integrated into a high fidelity CFD code replacing the previous 1D modal solver
• Results: Linear elastic solvers are much more accurate than modal solvers in predicting structural responses of aircraft subjected to aerodynamic loads
• Significance: Provide accurate and realistic structural responses for aircraft under various aerodynamic loads and structural stiffnesses
Convergence Plot (S-S Plate w/ Conc. Load)
0.0590.06
0.0610.0620.0630.0640.0650.0660.0670.0680.069
16 36 144 400
Number of Elements
Disp
lace
men
t (in
.)
Exact
FEA
Convergence Plot (S-S Plate w/ Uniformly Distributed Load)
0.096
0.098
0.1
0.102
0.104
0.106
0.108
16 64 144 256
Number of ElementsDi
spla
cem
ent (
in.)
Exact
FEA
University of Illinois at Urbana-Champaign
CSM 2: Visualization and
Job Monitoring of FDL3DI
L. Dandy, D. Semeraro, D. O’Neal - UIUC NCSA
M. Visbal, R. Gordnier, R. Melville - AFRL
University of Illinois at Urbana-Champaign
CSM 2: Visualization and Job Monitoring of FDL3DI
• Objective: To build a portable infrastructure for monitoring the progress of FDL3DI jobs during simulations
• Methodology: CUMULVS will be integrated into the FDL3DI code to provide analysts with snapshots of job progress during simulations
• Results: Many codes (i.e., EPIC and FDL3DI) require many days to perform a typical simulation. With CUMULUS jobs may be monitored throughout the computation to determine accuracy of results or to restart with different physical parameters.
• Significance: Provides a substantial increase in HPC resource efficiency and allows the analyst to improve results by check pointing and restarting a job based on real time examination of results
University of Illinois at Urbana-Champaign
CSM 4: Parametric Finite Element Model
of an Aircraft Wing
L. Dandy, G. Kwak - UIUC NCSA
M. Visbal, R. Gordnier, R. Melville - AFRL
University of Illinois at Urbana-Champaign
Sample Calculation of Wing FEA Model - MSC/PATRANCompletely automatic with internal structure
University of Illinois at Urbana-Champaign
CSM 4: Parametric Study of Structural
Damage due to Internal Explosions
L. Dandy, Y. Woo - UIUC NCSA
A. Mayer, G. Czarnecki, J. Calcaterra - VAVS
R
t1
L3
half span
L1 L2
t3
t2
Load
p2
p1
kspar2
Parametric Model
L4 - span to next spar
kspar1
NOT TO SCALE
University of Illinois at Urbana-Champaign
CSM 4: Preferred damage path before and after optimization simulation
David O’Neal - UIUC NCSA
Ruth Pachter - ASC MSRC
Geoffrey Fox - Syracuse U
Richard Luczak - Rice U
CSM 3: MAPINT’99 Symposium
University of Illinois at Urbana-Champaign
CSM 3: MAPINT’99 Symposium
• Methodology: The MAPINT’99 agenda again promoted a cross-section of contemporary research projects and emerging technologies.
• Results: Collaborations with ARL MSRC, the University of Greenwich, and Engineous Software, Inc. have resulted directly from NCSA ASC CMS personnel participation in this event.
• Significance: The DoD supports the development of multidisciplinary frameworks and environments targeting more realistic simulations and analyses. The MAPINT Symposium provides a platform from which strategic academic, industrial, and government researchers may communicate their ideas and achievements.
• Objectives: Provide an overview of leading multidisciplinary programs, initiate interaction between developers and users and explore advanced computing systems where such applications are likely to be executed.
University of Illinois at Urbana-Champaign
David O’Neal - UIUC NCSA
Mark Cross, Constantinos Ierothrou - U of Greenwich, UK
Michael White - Ohio Aerospace Institute
Richard Luczak - Rice U
CSM 3: CAPTools Project
University of Illinois at Urbana-Champaign
CSM 3: CAPTools Project
• Methodology: Semi-automatic dependency analysis and parallelization of FORTRAN source codes using the Computer Aided Parallelization Tools. Evaluation nearly completed. Steer development in a manner consistent with DoD HPC program goals. Produce portable message passing and OpenMP source codes.
• Results: Preliminary results describing performance and scaling attributes associated with a selected set of test applications were presented at the 1999 DoD HPC Users Group Conference in Monterey. Final results were presented at MAPINT’99. Five new multiprocessor codes (2 DoD) have been completed.
• Significance: Huge potential for legacy code modernization projects. Effects on utilization of HPC resources could be dramatic as improvement in job turnaround will encourage consideration of larger simulations.
• Objectives: Identify, evaluate, and deploy software tools that address the needs of the DoD research community. Transition CSM legacy codes to the high performance multiprocessor computing systems characteristic of the MSRCs.
LeRay Dandy, David O’Neal - UIUC NCSA
Jeffrey Calcaterra, Nicholas Pagano - AFRL
Juan Carlos Chavez - HPTi
Mark Ondracek - Engineous Software, Inc.
CSM 3: iSIGHT Project
University of Illinois at Urbana-Champaign
CSM 3: iSIGHT Project
• Results: Acquisition of network licensed product supporting two concurrent users was recently finalized. Approval of funding for purchase of additional seats by ARL is imminent. A combined network licensing arrangement will be served from ASC, thus providing a model for cross-MSRC resource allocation. Two DoD related test applications were also completed during the evaluation phase.
• Significance: Benefits of combining discipline-specific analysis tools into coupled physics models will support simulation of the behavior of real-world weapons systems with unprecedented accuracy.
• Objectives: Justify and acquire iSIGHT framework software and optimization logic. Create cross site multidisciplinary applications. Document individual achievements in a public forum, e.g. the PET CSM web site and promote collaborations.
• Methodology: Facilitate implementation of user defined problem solving environments by fostering a collaborative approach and providing relevant training.
University of Illinois at Urbana-Champaign
ARL & ASC JOINT CSM PROJECTS
University of Illinois at Urbana-Champaign
University of Illinois at Urbana-Champaign
CSM 1: Outreach, Training & Short Courses
• Objective: To disseminate newly developed information on structural analyses and large scale computational protocols as rapidly as possible to as wide and as diverse an audience as budgets permit through short courses and website course listings
• Methodology: 1 - To provide information and training through use of computerized instructional materials simultaneously to DoD and industry audiences at multiple locations through distance learning.2 - To provide current information on new research, development and associated results through seminars and short courses
• Results: The training and educational programs will be pervasive and cover numerous topics based on audience demand and new developments in universities, government laboratories, commercial codes, etc.
• Significance: This program will keep DoD personnel informed on current state of ongoing research efforts and on high performance computational protocols. It will also provide training for users to effectively execute advanced computer codes and to apply modern research results in a timely fashion
University of Illinois at Urbana-Champaign
CSM 1: ARL & ASC Year 3 Training and Support
– CoMeT seminar July 7, 1998
– Aeroelastic Design for CFD Engineers July 13-14, 1998– Introduction to ANSYS January 19-21, 1999– Viscoelastic Constitutive Relationships February 22, 1999– Equation of State (EOS) Course March 17-19, 1999– Using ParaDyn/DYNA3D March 23-24, 1999– Introduction to LS-DYNA3D July 26-30, 1999– INGRID September 14-15, 99
Other Core Support
– Support for ARL & ASC software committees
– On-going user support and consulting
CSM 1: Computational Software Seminars & Short Courses
Topic Tentative Date Location
NASTRAN TBD ASC/ARL PATRAN TBD ASC/ARL CAP Tools Training (via D. L.) TBD ASC/ARL Advanced ANSYS TBD ARL/ASC EOS/CTH Course TBD ARL/ASCCTH - DICE training TBD ASC/ARL Advanced Paradyn trainingTBD ARL/ASC ISight (via D. L.) Fall 99 ASC/ARLGraphics & GUIs with MATLAB TBD ASC/ARL
University of Illinois at Urbana-Champaign
University of Illinois at Urbana-Champaign
Structural/Solid Mechanics Seminars & Short CoursesTopic Tentative Date LocationProbability & Statistics TBD ASC/ARLDeterministic & Probabilistic Failures TBD ASC/ARL Composite Delamination TBD ASC/ARLComposite Manufacturing TBD ASC/ARLComputational Optimum Control TBD ASC/ARLIntroduction to BEM TBD ARL/ASCBEM & Failure Analysis Fall 99 ARL/ASCFracture Mechanics (theory) Fall 99 ARL/ASCComputational Fracture Mechanics Spring 00 ARL/ASCHigh Strain Rates Fall 99 ARL/ASCStructural Health Monitoring Spring 00 ASC/ARLStructural Control Fall 00 ASC/ARLComposite Aging TBD ASC/ARL
Nonlinear & Stochastic FEM TBD ASC/ARL Smart Materials & Characterization TBD ASC/ARL
University of Illinois at Urbana-Champaign
CSM for CY4 & Beyond
University of Illinois at Urbana-Champaign
YR 1OUTREACH
YR 2 TRAINING
YR 3 PROJECTS
YR4 ADVANCEDPROJECTS
YR5META
SYSTEMS
YR5 ADVANCEDPROJECTS
YR 4 PROJECTS
YR 5 PROJECTS
YR 3 TRAINING
YR 4TRAINING
YR 5 TRAINING
PROGRAM YEARS03/29/99 Grabert / Smith
YR 2OUTREACH
YR 3OUTREACH
YR 4OUTREACH
YR 5OUTREACH
0 1 2 3 4 5 t
ARL and ASC PET STRATEGY&
5 - YEAR DEPLOYMENT
• Objective: 1 - To provide multidisciplinary computational protocols and codes to analyze and design next century lightweight structures for flight, surface and naval vehicles based on probabilistic and deterministic reliabilities, including aeroelastic capabilities. 2 - To make results widely available through training and/or short courses by electronic transmissions and web postings for all MSRCs and selected HBUCs & MIs.
• Methodology: Cross CTA and MSRC multiphase and multidisciplinary analytical formulations and massively parallel computational simulations of viscoelastic composite materials with or without piezoelectric structural control for linear and nonlinear structural and aeroelastic problems, including stress and failure analyses.
• Results: Deterministic and probabilistic stress, deformation and failure solutions for flight vehicle metal and composite structures including optimum material selection and material characterization based on massively parallel computational protocols for real material solid/fluid interaction problems.
• Significance: Provide DoD, ARL, ASC and industry researchers, engineers, scientists and designers with modern analytical and computational models and tools to give them the ability to create realistic structures based on survival probabilities and times to failure. To make software & results universally available through grids. (metacomputing)
CSM for CY4 and Beyond
University of Illinois at Urbana-Champaign
National Computational Science Alliance
(The Alliance)
QuickTime™ and aGIF decompressor
are needed to see this picture.
University of Illinois at Urbana-Champaign
National Computational Science Alliance
The National Science Foundation Partnership for Advanced Computational Infrastructure initiative supports the development of a powerful computational problem-solving environment for national scale, multidisciplinary, collaborative work.
The National Computational Science Alliance (Alliance) is a partnership among more than 50 U.S. universities and research institutions to prototype the computational and information infrastructure of the next century. The Alliance's National Technology Grid will consist of a broad range of high-end parallel computing systems located at NCSA and other leading-edge facilities within the Alliance -- the super nodes of the Grid.
University of Illinois at Urbana-Champaign
Alliance Technology Roadmaps
• Building the Grid
• Capability Computing
• Science Portals
QuickTime™ and aBMP decompressor
are needed to see this picture.
QuickTime™ and aGIF decompressor
are needed to see this picture.
University of Illinois at Urbana-Champaign
Creating the Virtual Machine Room -Alliance National Scale Enterprise Testbeds
• High Speed Networking
• Connecting Multiple Vendor Supercomputers
– from Maui to Boston
• Common Web Interface to User
• Security/Authentication
• Accounting/Metacomputer Director Service
• Scheduling
• Access to Files and Distributed Data
• Remote I/O
• Quality of Service Reservations
University of Illinois at Urbana-Champaign
• A 5-year Plan That Identifies the Services Necessary to Prototype the Grid
• The Alliance Grid Strategy Is Designed to Achieve the Following:
Seamlessly Integrate Alliance Hardware, Software, Data and People
Allow for Extensions to Other Grids Allow for Easy Incorporation of New Technologies
• The Grid Can Be Dissected Into Two Logical Grids: Computational Grid Access Grid
Building the Grid
University of Illinois at Urbana-Champaign
Access Grid
Enabling Groups to Interact With Grid Resources
• Mural Displays • Cost Effective Multiple Projectors Base System < $50K Front or Back Projected Driven by PCs Software to Drive System Desktop Accessible • Audio Stereo, Quad, 3-D
Collaborations and Visualization
• Collaborations
Multiple Participants Share: Applications, Chat, White Boards, Presentations• Distance Training/Education• Visualization
Group Analysis of: Scientific Data, Virtual Reality, Instrument Steering
University of Illinois at Urbana-Champaign
What is a Science Portal? A New Generation of Alliance Workbenches
• Built on an Alliance “Common Portal Architecture” Defined by Alliance Teams Using: Emerging Web Distributed Object, Component Technologies & Information Repositories Standards• Web Access to: Remote Computational Resources Data Discovery and Analysis Tools Alliance Collaboration Infrastructure• Drawing on Emerging Commercial Technologies Personalization Integration of Services Security
University of Illinois at Urbana-Champaign
Science Portals are Gateways for the Computational Scientist
• Allow HPC Users to Interact With NCSA Queues & Globus Connected Supercomputers
• NCSA Is Developing a System to Report to Portal: System Allocations System Status (Up/Down, Capacity, Queues) Mass Store Status (Up/Down, HIPPI Connection Working?) Job Status (Where Job Is in Queue, Projected Run Date) Queue Information (Various Presentations) Account Statistics Project Statistics Usage and Patterns Billing Data • NCSA D2K is a Working Model of Science Portal
University of Illinois at Urbana-Champaign
FOR ADDITIONAL INFORMATION PLEASE CONSULT:
http://www.asc.hpc.mil/PET/CSM/frame.html
http://www.ncsa.uiuc.edu/SCD/Science/PET/publications/
http://www.ncsa.uiuc.edu/SCD/Science/PET/presentations/
http://www.ncsa.uiuc.edu/SCD/Science/PET/conferences/
University of Illinois at Urbana-Champaign
For further information please contact:
HARRY H. HILTON
Voice 217-333-2653 Fax 217-244-0720
Cellular 217-840-0358
h-hilton @ uiuc.edu
http://www.uiuc.edu/ph/www/h-hilton