Resource Planning for Island Networks

In an island network, where the main resources are local generation, the simplest form of resource planning criteria is, perhaps, the loss of the largest unit with another unit out on maintenance. As the number of units, and load, increase, the criteria may take the form of a percent reserve on peak demand. Probabilistic forms of the criteria would specify an allowable energy not served or maximum loss of load expectation, such as one day in ten years.

In an island network, with import capability via transmission links, the criteria could take similar forms:

  • Deterministic – loss of an import link while one unit out on maintenance
  • Semi-deterministic – percent reserve assuming full capacity on the import link
  • Probabilistic – representing the individual outage and derating probabilities of the local generators and import links in either an analytical process or Monte Carlo simulation.

Analytical methods are typically applied to a load duration curve without consideration to hourly load variation. The advantage of a Monte Carlo simulation method is that the chronological nature of load is retained, and non-traditional resource options are modeled explicitly, such as back-up power, demand management, network upgrades, changes in maintenance practices, changes in pricing policy, among others.

Deterministic and semi-deterministic criteria have the advantage of being easy to calculate, but may not adequately capture the reliability characteristics of the resources. A Monte Carlo model for an island network could be relatively simple to develop. The model can be used for both near and long-term planning, and may be developed as the load increases and resources change. The specific advantages of a Monte Carlo approach:

  • Assess the composite reliability of local generation resource and imports
  • Calculate cost of unserved energy for various resource options
  • Capture the resource benefits of load control methods such as under-frequency load shedding or demand limitations
  • Handle different capacities on the interconnections reflecting changing operating conditions in the main grid
  • Handle switching of connectors to transfer resources and/or load
  • Handle backup power supply

One could further generalize the above to an urban area with limited access to outside resources, as for example, the city of San Francisco.  The same type of resource evaluation can be conducted without loss of accuracy.  In the same manner, an urban center such as Boston, Chicago or New York City, which has a large native load, with controlled links to external resources, may be analyzed in a similar fashion.

As power networks approach the type of zonal islanding that are implied by deregulation, the techniques described in this article become more applicable.