NSF Grand Challenge in Distributed Systems

A Proposal by Terence Dowling SIGOPS ACM #1181189

Distributed Power Grid Management with 
Massively Distributed Power Generation.

The current Power Grid has a relatively small number of generators and
a large and distributed consumer base. In the presence of more than token
distributed power generation, the control and regulation of the power
grid becomes much more complex. If the goal of a "Million Solar Roofs" is
approached then the control/regulation problem must be solved.

This is a problem in scope beyond that which is practical for the product
manufacturers and is also beyond the reach of individual power districts
or companies.

Some included sub-problems:

 Distributed real-time process control of disparate systems with differing
 time constants and constraints. Network configuration and connectivity is 
 dynamic and may change with growth and equipment failure. Reconnecting a
 power network after it has partitioned is a complex problem.

 Network security, reliability, management. Power network management is a
 national security issue and distributed management poses interesting issues.

 Data privacy issues. Aggregated data is needed to control the system but 
 individual data is somewhat sensitive (building occupancy patterns could be
 used for criminal purposes).

 Real-time capacity prediction with neighborhood reaction to clouds and
 changing wind patterns.

 Some power consuming loads can be opportunistic. Examples include agricultural 
 and municipal water pumping, domestic water heating, pumped storage hydro.

 End devices to monitor and control power production and consumption at the
 leaf nodes of the system.

 Network connectivity for all of the components.

 Life Safety controls and systems.

 Power consumption/production records for billing and modeling of reasonable
 strategies for service pricing.

A Potential Testbed - the Island of Hawaii:

 Stage 0: simulation - early - often - open. Model power sources, loads, controls,
 networks etc.

 Stage 1: An initial test location might be North Kohala. No significant local 
 generation. Two grid connections (Waimea 25 miles, Kona 55 miles). Good potential 
 for solar, wind, small hydro and pumped storage hydro.

 Stage 2: Add a significant population center. South Kohala - large power consumer
 hotel/resort area. Good area for solar. Significant water pumping.

 Stage 3: Add more of the island to include the areas that have geothermal
 generation and some industrial co-generation.

 Stage 4: Island wide power community.

 Why Hawaii? Good area for solar and wind power which have potential applicability
 throughout the US. An isolated community that decreases the initial problems with
 major power grid interconnection. Inherently expensive electricity due to the
 complete dependence on ship transport of coal and oil.