Multiagency Coordinating Committee for Combustion Research: Next Steps in Using Combustion Cyberinfrastructure

1
Multiagency Coordinating Committee for Combustion
ResearchNext Steps in Using Combustion
Cyberinfrastructure

• Phil Westmoreland
  Program Director,
  Combustion, Fire, and Plasma Systems NSF/ENG

Multiagency Coordinating Committee for Combustion
ResearchDoD - DOE - FAA - NASA - NIST - NSF
MACCCR Workshop on Advancing Cyberinfrastructure,
San Diego CA, March 28-29, 2007
2
Workshop agenda

• Wednesday, March 28, 2007, 130-530pm
• 130pm Review of Workshop context, agenda, and
  goals
• 150 pm Perspectives from the TNF Workshop, Rob
  Barlow, CRF
• 235pm PrIMe A Virtual Organization Phil
  Smith (Utah), Michael Frenklach (Berkeley), Greg
  Smith (SRI)
• 320pm BREAK
• 340pm Developing Cyberinfrastructure for
  Data-Oriented Science and Engineering Fran
  Berman, Director, SDSC
• 420pm Insights from the Pittsburgh Combustion
  Simulation Workshop Geo Richards, NETL
• 500-530pm Open Discussion
• Thursday, March 29, 2007
• 800am Panel-Oriented Discussion of
  Opportunities and Action Items
• 930am Breakout sessions to identify needs
• 1030-1130am Recap of breakout sessions
  Identification of Action Items

3
What is cyberinfrastructure - and what isnt it?
4
Combustion research has been a leader in using
cyber resources for modeling. For example
5
Automating the Generation of Detailed Chemical
MechanismsBill Green (MIT) - CBET-0312359

• Complicated chemical mechanisms abound in
  manufacturing, biology, environment, and energy
  pollutants from combustion.
• Ultimately, they are sets of individual
  reactions.

• Project goal To automate construction solution
  of combustion simulations.
• With XML data formats, maintain a large software
  package, used / modified by many researchers.
• Include unambiguous documentation of simulation
  assumptions.
• Generate simulation confidence limits along with
  the predictions.

6
(2) Using Full Chemistry in Large-Scale
Simulations (Turbulent Combustion) Pope, Chew,
Guckenheimer, Vavasis, Givi (Cornell
Pittsburgh)- CBET-0426787

• Using full mechanisms by brute force is not
  practical to model chemistry in complicated
  environments.
• However, the overall rate of the mechanism is
  often controlled by a few reactions, subsets of
  the full mechanism.
• Key Identify the Intrinsic Low-Dimensional
  Manifold (ILDM).
• May not be true species and individual reactions,
  but composites.

• May be different ILDM in different regions -- use
  adaptive chemistry.
• Rate still depends on temperature and chemical
  concentrations only.
• Collect each calculation and simply look up rate
  if T and concentration conditions recur, rather
  than repeating detailed calculation
• In Situ Adaptive Tabulation (ISAT)
• Use these strategies with turbulent LES/FDF model
  through parallel computing here, simple H2/O2.

7
However, cyberinfrastructure is a fairly
recent, useful organizing concept.

• It recognizes the coupling of infrastructure
• Computer data acquisition, processing, and
  storage
• Simulation, now accepted as one of the routine
  practical tools of conducting science and
  engineering
• Dramatic increases in computing power, including
  terascale speed, storage, local cluster
  computing, and Internet-enabled grid computing
• Ubiquity of the Web, linking people to each other
  and to information.
• Not long ago, none of these advances were
  established.
• Now, they set the stage for new approaches to
  combustion research and development.

8
In this light, combustion research is a pioneer
not only in computing but in use of
cyberinfrastructure.

• Combustion has been a natural.
• Diverse physical and chemical systems.
• Large volumes of data acquisition, archiving,
  quantified uncertainties, validation,
  visualization.
• Modeling from atoms to autos to the atmosphere.
• Cyberinfrastructure is aiding the combined use of
  computing, networks, and inter-researcher
  collaboration.

9
CI value cuts across agency missions, too.

• Breadth of molecular modeling codes and
  applications was seeded by sponsored research.
• Supercomputer centers for high-performance
  computing by NSF, DoD, DOE, NASA.
• Collaboration tools beyond teleconferencing
  beginning to be used.
• Companies, including DoD contractors, are using
  data and findings from combustion research.

10
How do we best use it?

• More powerful computing
• Grid computing
• Simulation-Based Engineering and Science
• Cyber-Enabled Discovery and Innovation
• Different ways of collecting data.
• Remote sensor networks.
• Remote experiments.
• More effective collaboration and information
  transfer
• Virtual organizations (gateways,
  collaboratories).

11
Virtual Organizations (VOs) can couple all three.

• Well hear from TNF and PrIMe as two examples.
• Not just websites

12
(No Transcript)
13
Useful, but compare to capabilities of
nanoHub.org
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Big or small, both can be useful.

• nanoHub.org has wide usage.
• In contrast, a virtual organization may be a
  small group of geographically dispersed
  collaborators
• Data storage, retrieval, visualization
• Models and model results
• Data vs model comparisons
• Conferencing (NetMeeting, etc.)
• Another possibility An interactive data resource
  like webbook.nist.gov/chemistry

20
One activity setting the stage for CI action in
combustion was an April 2006 workshop.

• Cyber-Based Combustion Science, Apr 19-20, 2006
• Report authors Trouvé, Haworth, J.H. Miller, Su,
  Violi
• See http//www.nsf-combustion.umd.edu/
• Three key themes
• High-performance computing and sensor-driven
  modeling
• Chemical data/software libraries and
  collaboratories
• Education

21
April 2006 NSF Workshop recommendations

• Quantitative, predictive capabilities for
  engineering-level simulations of combustion
  systems will require a cyberinfrastructure-enabled
  framework built around high-performance
  computing and collaborative science
  infrastructures
• Ongoing developments in numerical combustion,
  driven in part by continued access of combustion
  scientists to high-end HPC centers.
• Emergence of chemical digital libraries as
  data/software stores and collaboratories.
• Coordination of efforts in and across other
  sub-communities of scientists and engineers in a
  common framework.
• Use opportunities to improve combustion
  education
• Renewed emphasis on pedagogical ties between
  fundamentals and applications
• Promotion of combustion as a multi-scale
  discipline
• Integration of data science and scientific
  computing into the curriculum.

22
That then is our charge.

• We want to propose concrete steps forward to aid
  such approaches.
• Listen and comment on different aspects.
• Tomorrow, develop specific plans to propose.

23
Workshop agenda

• Wednesday, March 28, 2007, 130-530pm
• 130pm Review of Workshop context, agenda, and
  goals
• 200 pm Perspectives from the TNF Workshop, Rob
  Barlow, CRF
• 240pm PrIMe A Virtual Organization Phil
  Smith (Utah), Michael Frenklach (Berkeley), Greg
  Smith (SRI)
• 320pm BREAK
• 340pm Developing Cyberinfrastructure for
  Data-Oriented Science and Engineering Fran
  Berman, Director, SDSC
• 420pm Insights from the Pittsburgh Combustion
  Simulation Workshop Geo Richards, NETL
• 500-530pm Open Discussion
• Thursday, March 29, 2007
• 800am Panel-Oriented Discussion of
  Opportunities and Action Items
• 930am Breakout sessions to identify needs
• 1030-1130am Recap of breakout sessions
  Identification of Action Items