How South Australia can function reliably while moving to 100% renewable power

The biggest challenge will be meeting the 2-3 hours of

peak demand during the evenings, when wind generation happens to be low. This will require a mix of different technologies and strategies, including solar, wind, storage, and possibly a new interconnector to New South Wales.
The issue is the variable nature of some  – wind turbines only generate electricity when there's sufficient wind, solar panels when the sun shines. But peaks in demand occasionally coincide with periods of low renewable generation, as was the case during the heatwave a few weeks ago. Although sufficient gas-fired generated capacity existed to pick up the slack, it was not all available at the time and short, localised blackouts were implemented.
Without strategic preparation, these events are going to be more difficult to handle in future as wind and solar farms grow, especially if the interconnector between SA and Victoria fails at a critical time.
But here are some of the things we can do in the short term (the next 2-3 years) and medium term (the coming decade) to create a reliable system on the path to 100% renewable electricity.
The short-term
The key point is that the challenging periods will be infrequent and only last for a few hours. Coal or nuclear power stations, which operate best when run continuously at full power, are too inflexible in operation to pick up the slack at peak demand. They would also be too expensive, hazardous and slow to construct.
In the short term, then, we need to install options that are flexible and dispatchable (i.e. able to generate when required). The options include open-cycle gas turbines (OCGTs), preferably each with a dedicated gas storage; concentrated solar thermal power with thermal storage (CST); and batteries. With the right policies these technologies could make significant contributions to peak supply within 2-3 years.


Read more at: https://phys.org/news/2017-02-south-australia-function-reliably-renewable.html#jCp