Spinning Out: Grid Inertia & System Stability

Uptake of renewable energy generation in energy markets around the world has long raised questions about system stability and this thing we call “grid inertia”, and when large-scale blackouts occur (like the recent blackout in the Iberian energy system which impacted Spain), this issue is brought to the forefront. It gets used by enemies of the energy transition as justification for retaining existing forms of generation, and leaves people uncertain about a renewable future.

This topic gets discussed a lot in the LinkedIn circles I inhabit, but when I went looking for an accessible post or blog that I could link to an interested person (not an engineer!) on Facebook, I just couldn’t find anything that seemed accessible enough.

Fellow energy nerds - I’m always happy for feedback if I’ve said something misleading, but keep in mind that I am also drastically simplifying many things to keep this article accessible to everyone.

1. Spinning Machines & Synchronicity

Let’s start with a car engine as an analogy for a coal generator. 
A coal station is like a car engine, except that it always rotates at the same speed (frequency). In Australia, that speed is 50Hz. Physically, the coal unit might be spinning at any RPM, but through a combination of mechanical and electrical "gearboxes", the output that we care about is a 50Hz sinusoidal wave. That means 50 peaks (and 50 troughs) in the voltage output every second.

A simpe way to visualise how a sinusoidal wave relates to something rotating at a constant speed. (Source: https://i.sstatic.net/pQ377.gif)

So if a coal generator is always at the same revolutions per minute (RPM), how does it vary output? Well, unlike a car, where we rev faster for more power, they just apply more "torque" (push) when more fuel is burnt, like a car on cruise control pushing harder to get up a hill, but not actually accelerating or revving any higher. If there isn't enough load in the grid to absorb that extra torque, then the coal generator will start to increase in RPM, and the frequency will climb. This is bad. The opposite is also true - too much load or not enough torque, and the frequency will drop. 

It is important that all loads and generators connected to any grid agree to use one common frequency - in Australia it’s 50Hz, and in some places it’s 60Hz. Any deviation from this agreed frequency is a bad thing.

An important word to understand ”synchronous” - anything that spins at a constant speed (coal, gas, nuclear) is referred to as "synchronous" and when we connect lots of these machines together on the grid, they all sync up together at the exact same frequency and timing.

Wind turbines are not synchronous, because their power output is linked to their rotational speed, which changes with wind velocity. To add them to a grid that runs at a constant AC frequency and voltage, we have to convert their output to 50Hz using inverters, similar to how we turn solar power into a 50Hz AC output. Anyway, back to synchronous machines...

Big heavy spinning things don't like to change RPM quickly. If you drop the clutch in your car, the engine RPM will drop, but then as the wheels of the car start to turn, engine RPM recovers and the car drives smoothly. In other words, if load in the NEM suddenly increases, we have a little bit of time before our coal/gas generators "stall", where they fight back against the change. This is "inertia".

Typically, if frequency starts falling, inertia gives us some time to shovel more fuel into the coal generators or re-balance the grid in another way. We didn't really intentionally build inertia into the electricity systems of the world - it was just an inherent property of things that are heavy and spinning, which is what made up the early grid.

If there is more inertia, the dial can't spin as fast. Without inertia, it might crash way out of bounds faster than we can re-balance demand & generation!

2. Balancing the Grid

There are a few ways we balance generation and demand in the grid, to ultimately keep frequency at 50Hz.

1. Every 5 minutes, AEMO estimates what demand will be, and instructs enough generators to generate, so that the demand should be met.
2. If they get this a little bit wrong, frequency will wander off-course a little. Generators in the NEM are required to provide a service called Primary Frequency Response (PFR), which means that if they see frequency wandering away, they all start gently generating more (to correct low frequency) or less (to correct high frequency), in order to sway it back the other way.
3. If frequency crosses certain thresholds, it triggers a series of services called Frequency Control Ancillary Service (FCAS).
   1. First, the gentler “Regulation FCAS” system is engaged, when frequency is 0.15Hz above or below 50Hz.
   2. Then if that doesn’t work, and the frequency reaches 0.2Hz away from 50Hz, the harsher “Contingency FCAS” service kicks in. This is like a knee-jerk reaction, where lots of generators or loads that can rapidly change their output will jump into action.
      The fastest of these services require service providers to respond within 0.5 seconds of frequency getting more than 0.2Hz away from 50Hz, which is pretty quick!

If frequency continues pushing out of bounds after all these mechanisms have fired, generators and loads will start to “give up” and automatically disconnect from the grid, to protect themselves, and to ensure they aren’t pushing electricity into a fault (like a fallen powerline).

3. Renewable Generators

Like I mentioned, wind and hydro turbines are not synchronous generators. Similarly, solar panels produce direct current electricity, which also needs to be converted to 50Hz alternating current by inverters. Modern inverters are a bit more “digital”. While synchronous machines are just big mechanical spinny things, comparable to an abacus, inverters are more like a digital calculator. So how do inverters work in an electricity grid?

Grid-Following Inverters

For a bunch of safety and design complexity reasons, grid-connected inverters have historically been designed as “grid following”, where they must first see an existing 50Hz grid operating, and then they synchronise their output to that frequency. They monitor this frequency in order to ensure they are adding power to a live, fault-free grid, and if they see frequency move too far from 50Hz, they will cease generating power. This is why installing solar panels alone doesn’t mean you are protected from a local power outage - your solar inverter is grid-following, and when that 50Hz signal disappears, the inverter shuts off to avoid pushing power back into a potentially unsafe electricity system.

Disconnection of a large amount of grid-following generation all at once can be a huge problem, as it can cause “cascading failure”. This is when a legitimate fault in one part of the grid can cause a sudden disturbance in frequency, and this disturbance then causes grid-following generators to also disconnect, making the total frequency disturbance (and demand-generation imbalance) even larger. When a disturbance gets to a certain point, safety mechanisms in the grid will try to prevent it taking down the whole grid, for example by disconnecting transmission lines, dropping entire regions into a blackout. An example of a cascading failure is when the Callide C power station had an unplanned rapid disassembly in 2021 - the unit wasn’t even providing any power to the grid at the time, but the jarring grid disturbance caused by a synchronised turbine exploding then caused a cascade of problems which resulted in power being cut to a large chunk of the QLD grid.

Standards for grid-following inverters have been updated in recent years, to give them clearer requirements for riding through short-duration disturbances, to minimise their contribution to cascading failure scenarios, but there are still concerns about whether inverter manufacturers have actually taken this onboard sufficiently.

In a grid dominated by existing synchronous machines, it has made a lot of sense that the addition of renewable generators has largely been done using grid-following inverters for the past few decades. However, as we look towards a grid that is increasingly dominated by inverters, we have to start improving what an inverter is capable of doing.

Grid-Forming Inverters

Grid-forming is a bit of an umbrella term, with different people having different definitions. However, the broad understanding is that grid-forming inverters are those which could perform a black-start of a grid - a start up with no existing 50Hz signal available to follow. This naturally means they can supply the initial “push”, which is similar to having inertia, but it’s not necessarily the same thing. Using this definition though, we could argue that a little 12V inverter from Jaycar is, on a small scale, a grid-forming inverter, but you couldn’t connect a bunch of those together and call it a grid.

Inertia is a different challenge, which we often group in as being a property of grid-forming inverters. Because we don’t have a big spinny thing inside an inverter*, we have to use something called “synthetic inertia”, which is basically applying code that pretends to be something with inertia, and tells the inverter hardware to push back against changes. In many cases, it involves literally running a physics model of something heavy spinning, and creating an equivalent electrical push-back. It can also be done more “mechanically” with transistor bridges, like this:

Model of how rapidly switching transistors can produce a 3-phase cyclical output. (Source: Nemo Bourbaki, https://www.linkedin.com/pulse/silicon-inertia-nemo-bourbaki/)

In the real world, instead of just 6 transistors, we could use a lot more to create a much smoother ‘rotation’, and then hopefully you can see how this animation could be combined with the earlier one, to produce a sinusoidal power output!

The cool part is that this can be dialled up and down to increase or decrease inertia. For example, we might want to tune systems like this to have less inertia if we want to be sure that, when a powerline falls over, we don’t continue feeding power into that line. The problem with electrical faults and disturbances, is that it can be tricky for a generator to know whether it should keep delivering power to try and resolve the disturbance, or whether there’s a serious problem and the generator should stop immediately.

Of particular interest is implementing grid-forming behaviours into the inverters used for batteries, because battery availability is a lot more certain than solar or wind. So far though, providing synthetic inertia is something that is still fairly new and unproven in largescale electricity grids, but existing research and early testing is proving hopeful that it will be just as reliable, if not more, than the natural inertia of spinning things. Another alternative is “synchronous condensers”, which are like giant flywheels that don’t generate any electricity, but spin in sync with the grid, and literally add mechanical inertia, but these are expensive.

*There are inverters that do have a spinny thing inside them, but you could argue that these are kinda just an inverter plus a small synchronous condenser, packaged into one box.

4. What Now, and What's Next?

Today, AEMO has to instruct spinning (fossil fuel) generators, particularly in South Australia, to continue generating, even when market conditions might not suit them, such as when prices are negative. This ensures enough inertia remains in the SA grid to ride through minor disturbances, but it also means those generators need additional compensation, and this is costly (not to mention emissions-intensive). Some synchronous condensers have also been installed to help with this, but they aren’t a complete solution. The ongoing uptake of renewable energy will need further innovation and understanding of inertia.

While a lot of blame for the largescale blackout in Spain has been placed on lack of inertia provided by renewable generators, investigations are ongoing, and it’s probably a little more complicated than that. Engineers have known about the need for inertia for several decades now, so it’s unlikely that the Iberian electricity system was completely ignorant of the need to manage inertia in their increasingly renewable grid. Perhaps they seriously underestimated how much inertia they needed, or perhaps even a high-inertia fossil-fueled grid wouldn’t have fared any better. More likely, grid-following inverters will be found to have contributed to a cascading failure that also involved other failures, and early indicators show that cyber-resilience of control systems may also have played a part. What is important, is that we can learn from events like this and continue to design the electricity grid of the future to be more resilient, even with renewable energy sources.

In Australia, we have already started on the path of technical specifications and integration of advanced inverter technologies. AEMO is doing work with industry on the largescale side of things, and at Reposit Power, we also have our own hopes about how smaller battery systems could provide synthetic inertia. If you look around, you’ll see that many definitions of a grid-following inverter say that they cannot participate in frequency control or ancillary services, but we’re proud to say that our Virtual Power Plants have been providing FCAS services for several years now. We use residential batteries and “grid-following” inverters, enhanced with some clever metering and control software designed here in Australia, and we’re hopeful that we could extend this to provide synthetic inertia soon too!

If you've read this far, thanks! If you have complex energy market questions you want answered in an accessible way, feel free to reach out to me on LinkedIn, and look out for me providing advice on the My Efficient Electric Home Facebook group or flick me an email. I am also always keen for feedback so that I can improve how I communicate tricky concepts clearly.