Engineering (Richard D. Braatz and Umberto
Ravaioli)
Need tools for coupling
scales
- Create algorithms,
software, & analysis tools for multiscale simulation
#1
challenge in simulation-based engineering science (NSF Blue Ribbon
Panel Report)
Example: CO2
Sequestration (LANL, MMI)
Energy
engineering policy decisions depend on
accurate simulation of all of these time and length scales
Example: Biological
Membrane Simulation Needed for Biomedical Engineering Apps
figure courtesy of
Dr. Mark Hill, Faculty of Medicine, University of New South Wales, Australia
Length
scales range from ions to lipids to proteins and ion channels to membranes
(to cells to tissues to organs)
Bottlenecks/Issues
to Achieving Objectives
- Need new numerical
algorithms for the simulation of systems that span widely varying time
and length scales
- Need algorithms
to analyze the effects of uncertainties in parameters and at interfaces
between physical domains on simulation outputs
- Need parallel/scalable
multiscale simulation algorithms
- Need algorithms
for applying such models to perform engineering
- Lack of long-term
funding (e.g., training grants, centers) large enough to support a cohesive
multiscale simulation effort across campus
- Large multiscale
simulation proposals to NSF handicapped by lack of faculty in multiscale
mathematics and applied quantum chemistry
- No courses
in multiscale simulation on campus to train people
- No books provide
a balanced coverage of multiscale simulation
- No educational
materials on “best practices” tools for multiscale simulation for
training people with varied disciplinary backgrounds
Cyberinfrastructure
Challenges in Reaching the Objectives
- Lack of software
that enables a non-expert in multiscale simulation to implement a numerically
accurate multiscale simulation algorithm
- Lack of software
to facilitate the validation of such a simulation
- Lack of scalable
software to analyze the effects of uncertainties in parameters and at
interfaces between physical domains on outputs computed from a multiscale
simulation code
- Lack of workflow
tools to facilitate running multiscale simulations on computers with
large numbers of processors
- Lack of support
staff with deep expertise in multiscale simulation
- Lack of support
staff with deep expertise in many key simulation methods, e.g., computational
fluid dynamics, KMC simulation
- These challenges
addressable only by computational researchers from many science and
engineering disciplines & new NCSA staff
Challenges, Barriers,
and Opportunities in Engineering Science (NSF
Blue Ribbon Panel)
- The Tyranny
of Scales: The Challenge of Multiscale Modeling and Simulation
- Verification,
Validation, and Uncertainty Quantification
- Dynamic Simulation
Systems, Sensors, Measurements, and Heterogeneous Simulations
- New Vistas
in Simulation Software
- The Emergence
of Big Data in Simulation and the Role of Visualization in SBES
- Next-Generation
Algorithms and Computational Performance
Need for carrying out
engineering tasks
Need for computational
efficiency
Multiscale Mathematics
Needs (DOE MMI)
- Multiresolution
Methods
- Hybrid Methods
- Closure Methods
- Adaptive
Methods
- Error Estimation
Methods
- Uncertainty
Quantification Methods
- Inverse and
Optimization Methods
- Dimensional
Reduction Methods
- Unifying
Mathematical Framework facilitate methods
development
- Mathematical
Software make available to engineers
Need for carrying out
engineering tasks
Need for computational
efficiency
Strategy: Multiscale
Simulation Education
- Create materials
for training in multiscale simulation
- Consider
a similar approach for other education efforts
Summer
School in Multiscale Simulation (2011)
Virtual School
of Computational
Science and
Engineering
University
course delivered via access grid (2012)
Textbook
Summer
School in Multiscale Simulation (2012)
Heath, Aluru, Braatz,
…
Blue Waters
Hires
Numerical
Simulation
Strategy: Meeting
Staff & Software Needs
NCSA
KMC expert
NCSA
CFD expert
Faculty
in applied quantum chemistry
Faculty
in multiscale math
Work
w/existing IACAT faculty & NCSA staff
algorithms,
software, & analysis tools for multiscale simulation
IGERT &
other proposals
Reference slide
- Multiscale
Mathematics Initiative: A Roadmap, U.S. Department of Energy, December
2004, http://www.er.doe.gov/ascr/Research/AM/Docs/MultiscaleMathWorkshop3.pdf
- Simulation-based
Engineering Science: Revolutionizing Engineering Science through Simulation,
NSF Blue Ribbon Panel Report, May 2006, http://www.nsf.gov/pubs/reports/sbes_final_report.pdf
- Community
Input on the Future of High Performance Computing, NSF HPC Task Force,
December 2009, http://www.nics.tennessee.edu/sites/www.nics.tennessee.edu/files/NSF-Workshop1-Report_FinDraft%20format.pdf