Jacinta Bowler
Like managing traffic lights, or making sure there’s enough toilet paper for a lockdown announcement, creating enough energy for future demand is a fine art. And renewables mean we have to start thinking about this art in a completely different way.
One of the ways being considered is a “microgrid” – a smaller, self-contained version of the traditional, large-scale grid system, usually with the ability to store power in batteries.
Traditional electricity grids are huge. On Australia’s east coast, we have the National Electricity Market, which is made up of around 40,000 kilometres of transmission lines and cables and connects all five of the eastern states together – from Queensland all the way down to Tasmania.
But the grid doesn’t store this electricity, it just passes it from the producers – coal power stations, gas turbines, wind, solar and hydro – to your house. If the grid needs more power the only way to make more is to add more energy to the system from one of the producers. The price the electricity market will pay to the producer goes up when lots of power is needed, and goes down when it’s not.
Think of a sprawling network of powerplants, electrical substations and power lines connecting almost every house, retail and industry to the grid.
Now, a microgrid at Monash University is helping scientists find solutions to these problems by giving researchers huge amounts of data on the way we create and use power.
“Australia has the highest uptake of solar globally, with around 30% of homes with rooftop solar,” said Monash electrical engineer Dr Reza Razzaghi in an email.
“Microgrids provides a great way for communities to generate power from renewable energy resources and trade energy with their neighbours.”
These microgrids are usually thought of as “islands” that are unplugged from the regular grid – they can be super helpful for areas where there’s not a traditional power grid set up already.
Think of a sprawling network of powerplants, electrical substations and power lines connecting almost every house, retail and industry to the grid.
For example, Lifou, an island in New Caledonia with around 10,000 residents, isn’t able to connect to a larger grid because of its remote location. In 2019 the island switched on a microgrid that includes generators, solar panels, wind and bio-energy plants and batteries to store the power.
Without microgrids, many island or remote communities have to rely on generators, which can be inefficient and costly to run.
“Sometimes you don’t want to have to run like 200 kilometres of wire to just connect five houses [to the grid]. You could instead put rooftop solar, a couple of batteries and connect just them together,” says Monash data scientist Dr Christoph Bergmeir.
“But why Monash? Obviously, we are connected to the main grid.”
The Clayton campus where the microgrid is installed is well within the City of Melbourne and doesn’t need to be an electricity island. But having a microgrid helps the university in other ways. The 1 MW of solar panels and 1 MW hour of battery storage is part of their ‘Net Zero Initiative’, allowing them to use less electricity from the wider electrical grid.
But what Bergmeir and his colleague, optimisation data scientist Dr Frits de Nijs are really excited about is what you can do with the data.
“The quality of the solution you get out is really dependent on the quality of the input,” de Nijs explains. “Because no AI system is going to magically work without the sufficient and correct kind of input data.”
Bergmeir explained that although data science relies on large amounts of data it’s not always easy to get. A lot of energy companies are privately owned and don’t want to splash their data out for the world (and their competitors) to see.
“It allows us to start asking really interesting questions and to publish that data without having to talk to companies, to ask them whether they want to publish that data, or make it available for students,” he said.
“And now with this microgrid we know exactly which building it is, we know which solar panel.”
Monash already has much of this information online, with dashboards showing individual building power demands and generation of power. The microgrid is connected to 20 buildings as well as electric vehicle charging stations, and the data is being recorded every 15 minutes from six of those buildings.
Without microgrids, many island or remote communities have to rely on generators, which can be inefficient and costly to run.
Although 20 buildings are only a small section of the large Clayton campus, the buildings include the sports facility, one of the libraries and a number of lecture halls. For each of the six buildings, a meter records how much electricity is being used. This is just to the building level, although Bergmeir suggests room level would be a good next step.
You might think that a university – with lots of classes during the day and limited movement at night – might not seem like the most relevant place to collect data when it comes to energy consumption. However, the way the microgrid has been set up makes it a little like a ‘micro city’ – with campus residents, businesses and industry all being part of the trial.
Bergmeir, de Nijs and a number of other researchers have started a competition to see exactly what we can do with this microgrid data. The competition – aptly titled ‘Predict+Optimize’ asks data scientists to create an optimal battery and lecture schedule using forecast and optimising techniques for the entire month of October 2020. They’re allowed to use the data provided from the Monash microgrid up until September 2020, weather reports and other easily available public energy data.
“If successful,” the researchers wrote to the entrants, “you will not only help making renewable energy more reliable and affordable, thus playing your part in the fight against climate change, but the proposed technical challenge may be applicable in many other fields facing similar problems of optimal decision-making under uncertain predictions.
Although the two researchers are both data scientists, their specialties vary greatly. Bergmeir is a forecaster – a person who uses data to predict what will happen in the future and De Nijs is an optimisation expert – a person who ‘optimises’, or works out the best option from a set of criteria. Neither of which have much experience in each other’s disciplines.
Both are required to make accurate schedules, and yet, there’s been little research into how optimisation and forecasting interact.
This is a problem throughout the whole energy industry. Both these skills are required to make accurate schedules, and yet, there’s been little research into how optimisation and forecasting interact.
“There aren’t really good solutions out there because it’s very difficult and you need these different skill sets,” says Bergmeir.
“It’s really the core problem that we try to solve [in this competition] … We want to see if people come up with better solutions.”
Because solar and wind are only producing power when the sun shines and wind blows, making sure we know how much power we’re going to have and when that power is needed is crucial to be able to limit power outages and use renewables most effectively.
Imagine a really hot day where everyone is blasting their aircon to keep cool. The energy on that day is significantly more than usual demand, and power grids are actually sensitive to high temperatures as well. These hot days are the ones you’d most likely expect a power failure.
“If you look at summertime in California or just recently in Europe, there’s been requests by grid operators and politicians to minimise consumption during these extreme heat waves,” says de Nijs. “They ask people to turn up [the thermostat on] their air conditioning because otherwise there will be black outs.”
Making sure we know how much power we’re going to have and when that power is needed is crucial to be able to limit power outages and use renewables most effectively.
But the hope with this technology into the future is that it would no longer have to be something we all personally have to decide to do. Microgrids like Monash might one day provide a way for researchers to not just forecast, but to tweak the way we use energy.
In the competition for instance, the team are asking the participants to create an ideal timetable schedule. If you have lots of solar energy during the day, it might be best to have lectures when the sun is shining instead of at night. These small changes in how we use energy could allow the university to use more solar and rely less on fossil fuels from the larger grid. It flips the system from using as much energy as we want, whenever we want it, to trying to use energy when it’s available and stored.
But there could also be much larger changes. For example, what if your ‘smart’ air conditioner knew when to adjust the temperature a few degrees to get the best use of your stored solar power?
“Ideally, if we can invest enough [into developing the technology], we can do this without impacting your comfort – you might not even notice,” says de Nijs.
“You just put clothes in the washing machine and four hours later they’ll be washed, but the washing machine started two hours after you pressed the button. Or you plug in your electric vehicle in the socket, and you say ‘I need it tomorrow morning to drive to work’. When it charges [within that period], that’s just up to the system.”