Why does optimisation matter?
Optimiser this, optimiser that - you hear it everywhere when batteries are mentioned. But why is an optimiser important? Who should be optimising for whom? What even is optimisation? We will look to answer these questions below, using Australia’s NEM (National Electricity Market) as our area of focus.
Battery Energy Storage Systems (BESS) have been a hot topic in the energy industry for a while. They hit the mainstream when our best friend Elon made his big promise on Twitter all those years ago.
Source - Australian Broadcasting Corporation
To his credit and thanks to what was probably a lot of frenzied and very hard work from many people, he came good, and the Hornsdale Power Reserve was energised in late 2017. At the time, it was the world’s biggest BESS.
What is probably the most ludicrous (yet true) statement about it is the fact that it effectively paid for itself in under 3 years. This was likely achieved thanks to some chunky government contracts which gave it some revenue certainty, but effective optimisation played a massive part in its success too.
A BESS can respond quickly to market signals and import or export power (acting from the perspective of both the market and the grid as both a load and a generator). This makes it eligible for pretty much every market going. What follows are a series of generalisations that will serve to illustrate the point.
All AEMO-procured markets in the NEM have a floor price (-$1,000/MWh for energy and $0/MWh for FCAS) and a market cap of $17,500/MWh, with the cap set to increase to an eye-watering $20,300/MWh from 1st July 2025.
Source - Australian Energy Market Commission
This is a hell of a spread.
Negative prices are common in the energy market during periods when energy generation is plentiful (usually the middle of the day when there is a lot of PV being pumped into the grid), because large generators would rather pay the market and keep running than shut down and then start up again, especially given the 5-minute settlement periods we have in the NEM. During these price periods, a BESS can get paid to charge.
In the evenings, as grid demand remains high but solar generation starts to drop off (no prizes for guessing why this happens), prices start to go up, and more importantly, volatility starts to increase. A fully-charged BESS can respond quickly to volatility, discharging to take advantage of high price periods and idling/charging if prices drop.
A BESS may still charge during a high-ish price period if a higher price is anticipated further down the track. A common misconception is that a BESS should only charge when prices are negative. This is not true. What matters more is the spread that a BESS can exploit. Buy low, sell high.
Prices low? Charge!
Prices high? Discharge!
Big spread? A R B I T R A G E.
The famed value stack
A decent-sized BESS in the NEM is usually monetised in at least 11 different markets, each in 5-minute intervals. Some of these are complementary, whilst participation in some effectively excludes the BESS from others. Market prices are changing all the time, and the decision-making process can vary depending on a particular BESS owner’s exposure to market volatility.
Some batteries, depending on their connection voltage and technical configuration, can also be liable for network charges. This introduces a 12th ‘market’ which must be considered to ensure that revenue is maximised.
This is why optimisation is important, and machine learning has become even more important in this area over the last few years. It helps if you have good software to help you run this process from end to end, and the optimisation component alone has huge importance. Running a multi-BESS, multi-region, round-the-clock portfolio with just people is expensive and does not scale well. It is easy to justify running one 200MW BESS with a team of traders. It is not so easy to justify running 40 x 5MW BESS with a team of traders, unless you have good optimisation software to help them. Gets pretty expensive like.
Source - Utility Dive
What’s important?
If you’ll pardon what appears to be a contradiction here, people have a tendency to over-egg the importance of optimisation. Don’t get me wrong, it is important. However, there are a few factors which, if ignored, can render even the best optimisation utterly meaningless.
Site connection/integration - it is very important that your optimisation commands can effectively be delivered and acted upon at site. I have worked on too many projects where the optimisation gets all the attention during development, and the site integration is not given anywhere near enough. Guess what happens if you create the best optimisation plan but your site integration doesn’t work? Nothing, that’s what. Your plan won’t go anywhere, and your battery won’t be optimised.
Observability - there are some prominent recent examples of optimisation software ‘losing their connection’ to key data inputs during major market events. Left unchecked, one BESS even charged from the grid when the price of electricity was at market cap. Rather than capitalising on these high prices, the BESS paid the high price and charged. ‘The market price feed broke’ was the excuse. Market price is one of the biggest, if not the biggest driver(s) of optimisation outcomes. Wouldn’t you expect your optimiser to sound the alarm if your market price feed broke, if not multiple alarms if it broke when market prices were on their way towards market cap?
Energy-only markets - high price periods can sustain for a long time. Based on the volatility insurance products available today, any price above $300/MWh is theoretically considered to be ‘high’. Most good traders don’t want to rely on AI when there’s a major market event underway (e.g. an interconnector tripping), not least because most batteries make most of their money on a small number of days in the year (especially the case if major market events occur). During such an event, traders will want to turn the robots off and rely on their own knowledge, holding their nerve and dispatching batteries in accordance with what they expect to happen. It’s very difficult to get robots that can substitute for years, sometimes decades’ worth of market experience. A combination of the two is ideal.
There’s a strong argument to be made that the fastest path to building a big, robust BESS portfolio is to focus on the non-scheduled segment. Increasing energy market volatility aside, a combination of simpler project development regulations, asset diversification across locations, plummeting equipment costs, and ready access to contracted offtakers (even easier where projects are co-located with Commercial & Industrial loads) makes this an incredibly attractive proposition for investors.
Where we fit in
Put simply, Hachiko has been built to focus on enabling the key players in this segment to achieve scale. We’re a small team, but we’re set up to succeed thanks to our combination of core knowledge, years in the trenches, and a good amount of scar tissue from multiple first-of-a-kind projects in this area.
The name symbolises loyalty. Our success will drive the success of those whom we serve. We’re very excited to be here, and can’t wait to meet you all.
