We are too timid

We are too timid

On 28 September 2016, a devastating storm knocked over electricity transmission towers in South Australia. This, coupled with a cascade of automatic trips of protection systems in nine the state’s wind turbines, caused a sequence of events resulting in the complete shutdown of the South Australian electricity system.

By mid-2017 (winter in Australia) the government of South Australia wanted to prepare for a coming hot summer. South Australia is a very hot state in summer, with a high electricity demand due to heavy air conditioner use.

Concerns about the reliability of the electricity grid had encouraged the already innovative state government to look for additional security for its grid.

Eventually, after a well publicised campaign including Atlassian’s Mike Cannon-Brookes and Tesla boss Elon Musk, South Australia signed a contract for a new 100 MW / 129 MWh lithium-ion (Li-ion) battery to help secure the State’s grid. This became known as the Hornsdale Power Reserve after the Hornsdale wind farm it supports.

There were many public statements against the battery. Most seemed to misunderstand (perhaps deliberately) the purpose and intent of the battery. It wasn’t intended to replace the entire grid’s supply in the event of another system black event.

For me, the key lesson from the Tesla battery story isn’t the technology itself. It’s what we’ve all been able to learn from its implementation. The original plan was to smooth out the power production from the Hornsdale wind farm. It has achieved this aim, and what’s happened since its commissioning has been much more interesting.

Now that a utility-scale battery is in service, we are seeing the interactions between technology and the market. The battery has cornered much of the Frequency Control and Ancillary Services (FCAS) market in South Australia, and it doing that work so well that it’s likely batteries will begin to take over that aspect of the market elsewhere. It’s also providing a case study for how we can transition to renewables being a substantial fraction of the energy supply for an electricity grid.

None of this would have been possible if the battery hadn’t been installed. Without courage and the willingness to forge ahead, we would still be arguing and discussing and analysing whether a battery like this is a good idea. Building a single example is an excellent way to cut through the critics. It’s easy to object to something if you interpret uncertainties in your analysis as sources of risk.

We need to start treating our energy transition more like scientists and less like accountants. Make experiments, and be prepared to build sizeable examples. It only takes one example of a new technology to show exactly what it can do. Real operating data will quickly clarify any uncertainties in pricing, performance, efficiency and so on, and embolden others to follow the path.

Of course, not every technology will work. That’s the point. Instead of dithering endlessly with talk and reports and analysis, the failures will enable us to put down the poor performing ideas and move on with the more promising ones.

We need to empower our people to make these bold choices without fear of losing their livelihoods if things don’t work out.

We’re fearful and we’re using endless analysis as an excuse to remain timid. It’s time to overcome those limits with action. We come from a world where capital was scarce and time was plentiful. Now things are the other way around.

Don’t throw away sophisticated analysis, but recognise it has limits. At some point, further analysis won’t aid decision making - it just provides an excuse for inaction.

We are too timid. It’s time to be bold. Future generations are counting on it.

Modelling Systems - Comparing conventional system modelling with machine learning

Modelling Systems - Comparing conventional system modelling with machine learning

Engineers rely heavily on models to do the work they need to do. Whether it’s designing a new piece of equipment like a pump or a cooler, right through to modelling a full oil refinery to enable us to optimise its performance, models are essential.

Conventional engineering models are based on fundamental laws of the universe (e.g. the laws of thermodynamics, or Maxwell’s equations for electromagnetic phenonena) as well as empirical relationships based on experiment. They are very powerful and have helped us to build the world we live in today.

In recent years Machine Learning (a branch of artificial intelligence) has emerged as a new alternative for building models. This uses big data to develop purely empirical models of systems. It’s fundamentally a very different approach to conventional modelling.

There are two videos below. The first part discusses how modelling is done more generally. The second part will cover specifics of machine learning vs conventional modelling.

Carbon Capture and Storage Part 3 - how to store CO2 once you've captured it

In this final (3rd) part of my Carbon Capture and Storage #WithASharpie series, I give an overview of some options for carbon dioxide storage (aka “carbon storage”).

The key options covered are geosequestration: saline aquefers, enhanced oil recovery, depleted oil or gas wells; and mineral capture: turning CO2 into solid carbonates (“mineral”).