An Adelaide microgrid has answered the call to help provide grid security in South Australia.
The energy system is being transformed from a model where centralised sources supply a fractal tree of consumers to one where thousands and thousands of small power plants are largely left to do their own thing. It’s alarming, to put it mildly. To avoid stress in the system it makes sense to locate large energy users and encourage them to be as flexible about how they use electricity as possible. If supply is looking tight, can they cut demand?
In the case of the SA Produce Market in Adelaide, a hungry energy user, the solution extended to the addition of a microgrid that can also answer the call by supplying electricity into the grid, drawing on a 2.5MW solar system, 2.61MW/4.325 MWh of storage and a 2.4MW diesel generator.
The project is a fine example of what to expect as we shift towards a two-sided market, where consumers can become suppliers at a moment’s call. In this case it’s a strategy that was forged in the fire of catastrophe. The key driver for the produce market microgrid was the state-wide electricity blackout in 2016, says Schneider Electric vice-president of power systems James Hunt. “They lost about a million dollars worth of stock.”
In the wash up, where politicians left and right were looking for someone or something to blame, the South Australian government thought ahead and directed funding towards any solution that would improve grid resilience. The SA Produce Market microgrid project cost $10.5 million, including $2.25 million in government funds.
In Hunt’s mind, for a microgrid project to justify its existence it has to enable the entry of renewables, offer resilience to the grid and/or push energy costs down. “This one ticks all of those boxes,” he says. “They had experienced the blackout and lost some stock; they wanted to put in solar and make operations more renewable, and; as a direct market participant, they had an exposure they wanted to manage.”
A microgrid that included PV and storage could balance shortfalls in solar output and fluctuations in energy price. “They manage their own load and their own exposure,” Hunt says. “For them, it’s about managing their own risk in the market and optimising their own position.”
Although Schneider’s EcoStruxure Microgrid Advisor platform that underlies the microgrid is capable of responding to demand response signals and forecasting demand response opportunities, the market is not part of a virtual power plant or offering demand response services.
The platform decides on optimal output by forecasting solar generation, energy price and load shape. “It will adjust the storage and the grid energy component based on maximising the cost reduction to the customer,” Hunt says.
With a self-learning algorithm humming away in the background, the site essentially runs on “autopilot”, Hunt says. “They don’t have to set anything or make any changes to operations; it just runs in the background.”
The platform decides optimal energy mix on a 15-minute cycle but will revise its outcome in response to sudden swings in the market. “That’s more its reactive response as opposed to the optimisation engine that runs in the background.”
Hunt gives an example of a day where an engineer asked why the battery was charging when energy was costing $150/MWh. Well, it turns out that later in the day energy was costing more like $350/MWh. “It was still going to be the most optimal outcome,” he says. “Even though prices were relatively high at the time, they were going to be higher later.”
Every day is different and Hunt says there’s no point trying to anticipate the platform’s thinking. If prices go to negative, the battery will take on as much charge as it can and the solar will switch itself off. Through an agreement with SA Power Networks the market is allowed to export, although there are constraints.
Battery storage has been a slow burner in the world of commercial and industrial applications, with managers favouring the immediate value in solar. Where storage can counterbalance an evident risk in a business, the case for a battery becomes stronger. “It depends on the customer having another driver,” Hunt says. “The business case may stack up or it may be sub-marginal, but if there is another driver, such as loss of stock or business continuity, then that typically pushes it over the line.”
Decisions on forecasting are fine-tuned as the machine adds to its knowledge each day and refines its thinking. “If there was a person trying to do it, the question would be: why?” Hunt says. “Humans are notoriously poor at decision-making and forecasting. That’s why the optimisation engine runs every 15 minutes in the background, to update with new solar forecast, new market pricing, new load shape to optimise it. It’s not something you can set up and leave. And it’s not something you can codify in a set of rules that follow simple logic and get a good outcome.”
Operations at the market begin very early in the morning, when most of us are still asleep, and peak load typically occurs between 2pm and 3pm, Schneider says.
The Schneider platform optimizes the use of the battery against the volatile spot market, so it is doing most of its work during high or negative prices. The software also considers the operational constraints of the battery, so it is operated within the manufacturer’s requirements for warranty. This means the battery will work harder on high opportunity days and less on low opportunity days, Schneider says.
The engineering and construction of produce market project was delivered by Autonomous Energy as the principal contractor. The microgrid includes 6,412 solar modules and 25 Tesla Powerpack batteries. The solar and battery power is separate, so the system can have 2.34MW solar and 2.61MW batteries going at once.
Batteries can charge from solar power, grid power or from the back-up generator, supplying the site loads when the sun goes down or exporting to the grid when spot prices are high. In the event of grid outage, the system will automatically isolate from the grid and continue to supply the site’s full electrical load.
It’s good news for tenants, too, who saw their electricity costs fall 25% on the last financial year, compared with a 2% drop in consumption.