Battery-owners who underestimate the complexity of trading their asset’s services can expect to miss out on revenue rewards.
As variable renewables enter the grid they are learning the tricks about selling their product – electrons – in an open market. For solar and wind plants it’s a fairly simple case of trading in the energy market, with a little bit of nuance around frequency control ancillary services (FCAS) charges. Batteries, however, will trade eight FCAS markets, making nine products – and because a battery qualifies as generator and load the person making trading decisions must forecast an optimal trading mix for 18 markets.
And it’s not just a case of making that decision for the next five minutes. The battery-owner might be better off charging when power looks expensive – and edge-of-the-seat decision – based on the hope it can sell into an even pricier market many, many hours into the future.
“You very quickly get to hundreds of thousands of decisions you have to make every five minutes in order to optimally operate a battery in a dynamic market like the NEM,” says Matt Penfold, vice-president commercial of Fluence, a US-based company that last October acquired algorithmic trading and bidding optimization platform AMS.
Profitable operation of a wind or solar plant will involve looking ahead to see whether energy prices for the next half hour will be higher than the cost of generation. “If you’re coming out ahead, you want to generate,” says Penfold. “But if you think the price is going to settle below your breakeven price you want to bid to curtail and avoid generating.”
That’s the case for renewables. For a battery, it’s a lot more complicated.
A battery qualifies as generator and load, and the owner needs to manage the state of charge to make the most of future opportunities. This means guessing price outcomes well beyond the next half hour. Those forecasts are the basis for an optimisation program that returns the best outcome for a combination of energy and FCAS products the asset should participate in and when. “Once we’ve made that determination based on all the risks and uncertainties that can materialise, we then translate that target dispatch into a set of bids,” says Penfold, an Australian, speaking to EcoGeneration from his base in San Francisco.
This is all still an educated guess. It’s only when the market settles and the energy is traded that the result is known.
How accurate a forecaster, then, is the Fluence platform? For battery assets in Australia this can only be estimated using simulation, where bidding for an imagined asset is compared with how the market played out. The team has been running a virtual storage asset since 2018 and has seen simulated revenue reaching 80-90% the theoretical maximum revenue.
A simulation is mentioned here because the Fluence platform is not connected to any local large-scale batteries yet. “We’ve signed some contracts,” Penfold says. “There are only five utility-scale batteries in Australia at the moment … we’re on our way to the first one going live.”
In California, the platform is optimising dispatch of the 182.5MW/730MWh Elkhorn battery owned by PG&E at the Moss Landing gas plant.
Relative to a human strategy, an automated trading platform boosts revenue for a wind or solar plant sometimes by more than 10% over 12 months, Penfold says. When trading a battery’s services, the benefit of automation over human effort can be additional revenue in the range of 50% to more than 100%.
It took two years for AMS to sign up its first 2GW worth of large-scale generation in Australia. Since then, October 2020, AMS has been bought by Fluence and an extra 1.5GW is being traded via the platform. New customers signed up in Australia over the past six months include Spark Renewables (part of the Spark Infrastructure Group) and BJCE, bring the total number of assets using the platform to more than 3.4GW (contracted or deployed).
The company claims 18% market share of solar and wind plant capacity on the NEM and a 60% slice of the software bidding market.
As more batteries are connected to the grid – and there are about 3GW of projects in various stages of approval – the FCAS markets are expected to thin out. FCAS prices are a function of supply and demand, Penfold says, and a “big wave” of batteries are expected to connect by the middle of next year. This new supply will be balanced in the long term by the exit of coal generators, the traditional dominant participants in the FCAS market.
The steady replacement of dispatchable generation with variable renewables will see demand for ancillary services increasing, a trend predicted by the Australian Energy Market Operator. As it’s impossible to predict the next 20 years let alone the next five minutes, Penfold says market participants shouldn’t feel complacent. The past summer was easy going for the NEM but anything can happen.
“I don’t believe the FCAS prices and battery revenues we saw this past summer are going to be representative of what we’ll see moving forward,” he says.
The FCAS “goldrush” has a little bit left to run, and as coal is retired and the AEMC introduces additional FCAS markets battery owners will need to be nimble. “Fast frequency response and fast-ramping products can be quite valuable and important parts of balancing the grid.”
That a look at Ireland, which has implemented a two-second fast frequency response market with a scaling tariff for response to within 150 milliseconds. The move has accelerated investment in batteries to provide very high-speed response services.
Rapid uptake shows significant appetite for software to boost the value of renewable energy assets and/or utility-scale batteries.