Supporting large-scale transition to sustainable energy

The outcome

Meridian Energy Australia (MEA) and the University of Melbourne have developed and implemented a reliable wind forecasting model that can be integrated into the Australian Energy Market Operator’s (AEMO) real-time processes for managing the National Energy Market (NEM). This supports the transition to renewables.

Testing of the system continues at MEA’s Mt Mercer and Mt Millar Wind Farms in Victoria and South Australia respectively. The project is backed by $2.18 million from the Australian Renewable Energy Agency (ARENA). It aims to improve both generation forecasts used by the electricity market and the forecasting accuracy of Australian wind farms, so that wind can be a more reliable, secure and efficient energy system.

The highly innovative project, which will improve 5-minute ahead forecasts by optimising information from a number of data sources and forecast models, will help the uptake of renewables into the NEM. It will also allow AEMO to maintain system security and better balance electricity supply with demand, which should result in increased demand for clean and renewable energy generation.

Melbourne Energy Institute (MEI) Director, Professor Michael Brear, says the collaborative, long-term partnership between Meridian Energy Australia and the University has been instrumental to the success of the project so far.

“Meridian Energy Australia has been a wonderful partner of the Melbourne Energy Institute over several years. This project demonstrates that our close relationship can be a catalyst for real impact. Together we’ve produced state-of-the-art energy technology to make better wind farm forecasts that will lead to increased uptake of renewables into the market.”

The need

Wind is a great source of renewable energy, but its variable nature means it can be difficult to accurately forecast. This leads to difficulties in balancing supply and demand, resulting in either wasted power, power shortages, or greater expense, and therefore challenges in adopting it as a reliable, secure and efficient energy source.

As the transition to cleaner energy solutions continues, and the number of variable renewable resources in the overall energy mix increases, the challenge becomes even bigger, as AEMO has more sources to balance. Several studies around the world suggest that once wind and solar exceed 30-40 per cent of the overall generation mix (expected in Australia within the next 10 to 20 years), energy forecasts (in megawatts) become more uncertain than demand.

To overcome this and effectively transition at scale, we need more accurate forecasting of wind production. We also need to maintain a secure and efficient electricity system as we transition, and to make the most of existing assets to ensure any new system is sustainable.

The research

Bringing together the work and insights of MEA and the University of Melbourne, the project draws on interdisciplinary research from experts in meteorology, energy systems and machine learning. Dr Claire Vincent, Dr Grant Skidmore, Dr Dominic Davis, Dr Rachael Quill and Mr Mathieu Pichault are playing key roles in this research, which informs both methodology development and implementation of the forecasting system.

To provide an accurate wind forecast, the project combines data-driven and physics-based modelling with a knowledge of relevant atmospheric phenomena such as wind gusts, rain, thunderstorms and the onset of calm night-time conditions.

The team will then incorporate the data into new algorithms to calculate a 5-minute ahead forecast, which will then be sent to AEMO via an application program interface tool, also developed as part of this project.

The project will also review market benefits of the improved forecasts, to ensure real-world success.

Technology development history

In March 2019, Meridian Energy Australia and the University of Melbourne won support from the Australian Renewable Energy Agency (ARENA) to develop improved, real-time forecasts of wind farm power generation.

Using this support, the team began developing a methodology for accurate forecasting of wind generation, including refining an algorithm and combining it with state-of-the-art technology as well as complex data analysis techniques.

By mid-2019, the team had developed and implemented a wind forecasting system, which was submitted in real-time to AEMO’s real-time processes for managing the NEM.

While a working, integrated forecasting system at the MEA's two wind farms has been achieved, the project is ongoing as the team refines the system and quantifies its benefits.