The long-term dynamics of the electrical grid are examined in new studies
conducted by Ben Carreras and his colleagues at Oak Ridge National Lab,
the University of Wisconsin, and University of Alaska.
Engineers at the utilities are of course always looking for ways to
make their systems better, especially in the aftermath of large blackouts,
such as the event on August 14, 2003. These post-mortem studies typically
locate the sources of the outage and suggest corrective measures to
prevent that kind of collapse again, often by strengthening the reliability
of specific components. Carreras argues that a more effective approach
to mitigating electrical disasters is build more redundancy into the
system.
And to do this, he says, you need to look at how the electrical grid,
considered as a dynamic system subject to many forces, behaves over
longer periods of time. And to do this one needs to build into any grid
model social and business forces in addition to the physics forces that
govern the movement of electricity.
Thus the Oak Ridge model not only solves the equations (governed by
the so-called Kirchoff laws) that determine how much power flows through
specific lines in a simulated circuit, but also build in the strain
on the system over time caused by an increasing demand for power, the
addition of new generators and transmission lines, and even elements
of chance in the form of weather fluctuations and the occasional shorting
caused by warm, sagging lines contacting untrimmed trees.
The model proceeds to let the grid evolve, and for each "day" it computes
possible solutions---in the form of successful combinations of power
generation levels and subsequent transmission of that power over existing
lines, some of which come in and out of service---for the continued
running of the grid. The model derives a probability curve for blackouts
which matches pretty well the observed outage data for North America.
The Oak Ridge scientists believe that their model could be used by
utility companies to test grid behavior for various network-configuration
scenarios, particularly those where the grid is operating dangerously
close to a cascade threshold. (Carreras et al., Chaos,
September 2004; carreras@fed.ornl.gov)