|
Home
Current Projects
Previous Projects
Publications
Presentations
CV
Outreach
|
Advanced Computational
Tools for Electric Power Systems
As the society’s dependence on engineered networks such
as the electric power and communication networks grows, the need for secure
and efficient operational standards for these systems becomes vital for the
economic, energy and national security of our country. The August 14, 2003
blackout in the Northeast depicted the possible catastrophic consequences
of a few broken power lines.
Robust operation of a power system requires the system
to stay intact after a small number of broken lines, and thus calls for
algorithms to detect small groups of lines that can cause a blackout if
they fail collectively. The flow of power can be modeled with a system of
nonlinear equations, and a blackout corresponds to the infeasibility of
these equations. Thus the vulnerability detection problem for the power
grid can be studied as a mixed integer nonlinear optimization problem,
which is a hard problem to solve by itself. We have reduced this problem to a pure combinatorial
problem, known as the network inhibition problem, by analyzing the
structure of an optimal solution to the mixed integer nonlinear
optimization problem. Our key observation was the relation between the
feasibility surface of the power flow equations and spectral graph theory. Instead of solving the NP-complete
network inhibition problem, we have introduced and studied the inhibiting
bisection problem, where we look for a bisection of the graph with a
minimum number of edges between the two parts, and a maximum production/consumption
mismatch within each part.
Relevant Publications:
A. Pinar, J. Meza, V. Donde, and B. Lesieutre, “Optimization
Strategies for the Vulnerability Analysis of the Power Grid, submitted to SIAM Journal on Optimization (pdf)
A. Pinar, A. Reichert, and B. Lesieutre, Computing
Criticality of Lines in a Power System, in Proc. 2007
IEEE International Symposium on Circuits and Systems, New Orleans, LA,
May 2007. (pdf)
V. Donde, V. Lopez, B. Lesieutre, A. Pinar, C.
Yang, and J. Meza, Identification of severe multiple contingencies in
electric power networks, to
appear in IEEE
Transactions on Power Systems. (pdf)
B. Lesieutre, A. Pinar, and S. Roy, Power
System Extreme Event Detection: The Vulnerability Frontier, Proc. Hawaii International Conference on System
Sciences, 2008. (pdf)
A. Pinar, Y. Fogel, and B. Lesieutre, The
Inhibiting Bisection Problem,
submitted to ACM 19th Symposium Parallel Algorithms and
Architectures (SPAA) 2007 (pdf)
B. Lesieutre, S. Roy, V. Donde, and A. Pinar, Power
system extreme event analysis using graph partitioning, Proc. of the North American Power Symposium,
Carbondale, IL, October 2006. (pdf)
V. Donde, V. Lopez, B. Lesieutre, A. Pinar, C.
Yang, and J. Meza, Identification of severe multiple contingencies in
electric power networks,
Proc. the North American Power Symposium}, Ames, IA, October 2005.(pdf)
Interconnection Networks
for Peta-scale Systems
As we enter the era of peta-scale computing,
system architects must plan for machines composed of tens or even hundreds
of thousands of processors.
Although fully connected networks such as fat-tree configurations
currently dominate HPC interconnect designs, such approaches are inadequate
for such ultra-scale concurriencies due to the superlinear growth of
component costs. Traditional
low-degree interconnect topologies, such as 3D tori, have reemerged as a
competitive solution due to the linear scaling of system components
relative to the node count; however, such networks are poorly suited for
the requirements of many scientific applications at extreme
concurrencies. To address
these limitations, we propose HFAST, a hybrid switch architecture that uses
circuit switches to dynamically reconfigure lower-degree interconnects to
suit the topological requirements of a given scientific application.
Relevant Publications
S. Kamil, L.
Oliker, A. Pinar, and J. Shalf, “Communication Requirements and
Interconnect Optimization for High-End Scientific Applications," submitted to IEEE Transaction on
Parallel and Distributed Computing. (pdf)
S. Kamil, A.
Pinar, D. Gunter, M. Lijewski, L. Oliker, and J. Shalf, Reconfigurable
hybrid interconnection for static and dynamic scientific applications, In Proc. 2007 ACM International
Conference on Computing Frontiers. (pdf)
Energy Efficient Disk
Systems for Scientific Loads
Scientific loads have a locality property,
both spatial and temporal. We
want to exploit this property to reduce the number of spinning disks, which
translates to huge savings in energy. We are
considering batch-job scheduling, data layout strategies, and access
pattern analysis in the context of this project.
Supernova Spectra
Analysis
This is another new
project; more will come soon.
|
