Allowing for Wind Power
Aucklanders may remember the power blackout in 2000 that closed businesses, and a ship in the harbour had to generate power for the central city. This was a reminder of how finely-tuned industrial power systems are, and how long it takes them to recover from rolling blackouts. New Zealand generates between five and ten percent of its power from wind, a high proportion internationally. Mathematicians are working out ways to deal with the uncertainty this power source adds to the national electricity system, to prevent future blackouts. Jenny Rankine investigates.
- Download article: Allowing for Wind Power (IMAges Issue 9: October 2010)
University of Auckland mathematician Golbon Zakeri explains:“The country’s power system is regulated in 30-minute blocks. Market traders in Meridian, TrustPower and other generators submit offers every half hour to Transpower, which operates the national grid.The offers are made up of five quantities at different prices. And every 30 minutes,Transpower solves an optimisation problem to determine generation quantities from each supplier.
“So they say to Mighty River Power, which owns the Waikato River hydro chain, ‘Generate this quantity or at this rate over the next 30 minutes’. So Genesis runs Huntly at a specific level, and Meridian provides a specific amount of hydro power from Lake Manapouri, and so on.”
Transpower’s optimisation software, called Scheduling Pricing and Despatch, aims to minimise the cost of electricity while meeting demand and satisfying transmission constraints.
The problem with wind power is that it is much more unpredictable than hydro or geothermal power, even over 30 minutes, and there is very little international research about how it can be efficiently integrated into national power grids.
Javad Khazaei’s PhD is exploring ways of adding stochastic optimization, which takes into account uncertainty, to the deterministic optimization used by Transpower. The current system copes with variability by using one generation station in each island to monitor the rate of demand, and feed in more or less power so that supply is maintained close to 50MHz. But it is too expensive to use more stations to regulate power in this way.
Khazaei used repeated simulations of past market data with stochastic optimisation that allows for high, medium, and low scenarios of wind during each half-hour.
In most cases this programming performed better and increased consumer and producer welfare.“It is a tuning, a natural extension, to enhance the current optimisation and cater for higher levels of uncertainty,” says Zakeri. “It will also work for solar and photovoltaic power,
An implemented stochastic process would have more bids, more information within the 30 minutes, and include deviations with penalties if the wind isn’t blowing. Khazaei’s theoretical analysis also asks whether this mechanism is susceptible to power generators taking advantage of these penalties, as New Zealand’s market is not highly regulated like those in other countries.