Australia is in the midst of an electricity network stability crisis, with major impacts from “system strength” issues .
At present, the solution appears to be painstaking, time-consuming startup procedures for new generators and the deployment of multi-million dollar “synchronous condensers” to provide “inertia”.
A Bit of History
So what is the fundamental system strength or stability issue? To understand this, we first need to review just how the grid is controlled. Historically, electricity was generated by steam-driven generators.
Steam is generated by boiling water using heat from burning coal or gas or even nuclear decay and then the steam is fed through a turbine to drive a generator. Regardless of the energy source, all of these generators are large spinning machines and have a fundamental response to load.
If the electrical load is too high and exceeds the power supplied by the steam, then the generator slows down and the output frequency of the electrical power decreases.
Conversely, if the generator is under-loaded and has more steam power than electrical load, then the generator speeds up and the output frequency increases.
For historical reasons, frequency is used as a proxy for over- or under-load. Therefore the fundamental method developed for controlling the grid has been to control the frequency.
Note that these steam-driven turbines and generators are huge, massive machines – the rotor in a 1000MW generator weighs thousands of tons.
The nature of a heavy object to resist changes in speed is called inertia. When we talk about grid inertia, we are talking about the tendency of massive spinning machines to resist changes in speed as the electrical load changes.
The large inertia of huge spinning generators has allowed a relatively “hands-off” approach to control.
In addition to the primary market for energy, AEMO operates secondary markets to ensure grid stability.
These secondary markets allow generators (and large loads) to bid to provide frequency modifications – that is, to raise or lower the frequency. AEMO accepts these bids as required to ensure the grid frequency remains close to 50Hz.
There are raise and lower markets for small adjustments (so called frequency regulation) for three different time periods – 6 seconds, 60 seconds and 5 minutes.
In control systems, these time periods are extraordinarily slow – for example, anti-lock brakes update the control signals up to 10 times a second.
For the enormous mass of traditional, large generators, 6 second updates works OK – but things are changing.
Renewables and Grid Stability
As opposed to traditional generators, renewable generators all connect to the grid via an inverter. The inverter consists of computer-controlled transistors that switch and modulate the constant voltage (DC) from the generators into an alternating, sine-wave signal (AC) compatible with the grid.
The computer estimates the current frequency of the grid (and the time offset or phase) and matches the inverter output to the grid. There are two critical things to note here: grid-following inverters have no inertia and inverters are directly controlled by a computer and can achieve cycle-by-cycle control.
The lack of inertia is contributing to grid stability issues – inertia-less generators do not fit in the historical model.
As the fraction of renewable generation has risen, and the fraction of the grid without inertia has risen, it is inevitable that grid stability issues arise. Control of an inertia-less grid via frequency simply cannot work.
Growth of Smaller Generators
Another factor contributing to grid stability issues is the growth in the number of generators – particularly smaller generators.
Recent fossil-fuel projects have been smaller, more agile plants – due to the changing generation landscape. Also, renewable projects are on a much smaller scale than traditional GW-scale generators.
The result is that over the last decade, there’s been considerable growth in smaller generators. See below:
Rooftop solar is another major driver here: every rooftop solar installation has it’s own inverter. There are over 2 million rooftop solar installations in Australia. Australia’s NEM has gone from 200 generators late last century to millions of generators now.
This explosion in smaller-scale generation is a classic example of “oo many cooks spoil the broth”. All of these smaller generators are attempting to estimate the current frequency, phase and voltage and exert their own control over these things. The smaller generators (cooks) are all fighting against each other, without effective overall control (head chef).
To extend the analogy further: what we have is like “too many cooks playing Chinese whispers to spoil the broth”.
In the grid, the generators don’t have good information – they are connected via long transmission lines that introduce delay and uncertainty.
So how can we possibly solve these problems?
To give a pointer to the solution, let us imagine a future grid that is 100% renewable. In such a grid, there are no machines to exhibit the frequency response.
If we overload such a grid, the frequency will not drop. The frequency of the inverters is computer-controlled and has no relationship to the load.
Instead, for an overloaded renewable grid, the voltage will drop. So it seems clear that we need to transition to new methods for managing the grid.
Load will need to be managed via voltage – if the system voltage drops, then the grid is overloaded and if the voltage rises, then the grid is under-loaded.
One more consideration in this imaginary 100% renewable grid is that the inverters will need to be synchronised. They must all be generating their 50Hz AC waveform at exactly the same time.
If they get out significantly of step, then they will be driving different voltages – and “fighting” each other, rather than supplying power to grid loads. It is crucial that the inverters are synchronised.
Here we see the three core components of a future grid:
- Load control via voltage control
- Inverter time synchronisation
- Centralised control of frequency, by broadcast to all generators
To ensure that generators are not fighting against one another, and are instead working together to deliver power to the system loads, they need to be closely synchronised.
Traditional generators achieve synchronisation by following the grid (which is partly why the grid frequency varies). If we have an external synchronisation method, then it needs to be quite accurate.
For example, 10 millionths of a second (microsecond) of error would lead to at most 0.3% voltage mismatch at 50Hz. Aiming for 1 microsecond accuracy seems desirable.
A cheap, ubiquitous accurate time source is available: the GPS system delivers worldwide timing to within 25 nanoseconds (25 billionths of a second) see here.
This is 40 times better than is required. For robustness, a secondary time synchronisation method should be developed and deployed to protect the grid against cyber attack.
One option would be wired (or optical fibre) connections to each generator.
An Immediate Solution to Frequency Stability Issues
If renewable inverters have good time synchronisation, then a new tool for grid frequency stabilisation is available: remote control of inverter frequency.
An immediate, simple scheme for stable renewable generation would be to place a sensor at the nearest conventional generator that transmits frequency control to the renewable inverter. Under this scheme, the inverter would be perfectly in step with the conventional generator. It would be as if the conventional generator had been increased in size/inertia.
Note that both the sensor and the inverter need to share the same time reference (GPS).This shared time reference allows the use of low-cost communication networks for transmitting the frequency and phase. Alternatively, a dedicated, deterministic communications link could be used.
The basic process is:
1. Periodically (let’s say once per second), the frequency and phase of the
conventional generator are measured (with respect to GPS time).
2. That information is transmitted to the renewable inverter controller.
3. The inverter generates AC outputs to perfectly match the conventional generator.
If the grid is overloaded, the conventional generator slows down and it’s output frequency decreases. Then the renewable inverter follows suit. The same with an under-loaded grid, except the frequency increase. Existing frequency control techniques can be employed to stabilise the grid.
The current situation with grid stability issues with the five renewable projects near the north-western corner of Victoria is a classic case of “too many cooks”. It is predictable that grid stability issues will crop up first in distant regions with high renewables percentages.
Here, the voltage is fluctuating many times a second.
The solution is easy to see from our analogy – if the cooks are fighting then bring in a head chef (or promote one). In this case, the five renewable projects should have their inverters upgraded to receive remote commands (this is mostly a software change) and a new computer deployed to control the five inverters in harmony.
Alternatively, one of the five renewable projects could be nominated as the master and it could provide commands to the other four projects.
The current solution of deploying a synchronous condenser will cost tens of millions and is ignoring the root cause of the problem. It’s true that the synchronous condenser will slow down the fighting and reduce the magnitude, but there will still be fighting. (In our analogy, a synchronous condenser is akin to making the pot bigger – the inputs from each of the cooks will be diluted and averaged out.)
As the grid evolves, it is expected that clustering will be increasingly used. From a control perspective, having a single cluster for the entire grid would be the most accurate.
However, from a reliability standpoint, having the grid able to keep operating when communication lines are down would recommend multiple clusters. State-wide clusters could be deployed, or clusters based on locality and interconnectedness.
Australia’s grid is in crisis, with stability issues cropping up with increasing frequency. However, by considering the future, 100% renewable grid, we have proposed a scheme for stabilisation of the grid that is both low-cost and can be deployed very rapidly.
In fact, the cost is likely to be less than a single day’s losses from the currently curtailed renewable projects! The grid of the future will have inverters that can be remote-controlled. With remote-controlled inverters, the lack of system inertia can be addressed and the proliferation of small-scale inverters can be arrested (by clustering).
Dr David Austin is director of MadJ Innovations, a Canberra consultancy company that specialises in complex systems. He comes from a control systems and robotics background and has, in recent years, led the development of a wide range of decentralised systems, involving both electronics and software.