Managing frequency in a modern electricity system

The electricity grid is changing as we use more renewable energy like wind and solar power and less conventional generation such as coal and gas.

Infrastructure investors are embracing the new over the old and consumers are actively taking part in the clean energy revolution.

Everyone knows that wind, solar and storage do not work the same way as gas or coal. These technologies are not the same and advocates for the status quo love to point this out. Some claim that renewable energy doesn’t provide ‘inertia’ to the grid, although this is often not followed by an understanding of what ‘inertia’ actually means.

A new Clean Energy Council paper released this week investigates the issue of inertia, how it works with the power grid and how renewables and storage can help using currently available technology.

Inertia relates to the power system’s response to unexpected shocks, and its ability to provide time for other supporting controls to react and restore the system’s balance.

It has traditionally been provided by thermal gas, coal and hydro generators, and there are some other options available through conventional technology as well.

What is not so well understood is the interaction between mechanical inertia and the power grid. In short, the main job of inertia is to slow down the rate at which frequency changes after a grid shock, such as the failure of a large power plant or a transmission line – the larger the shock or the lower the inertia the faster the rate of change of frequency.

In the event shown below, South Australia’s Northern Power Station (this coal-fired plant closed in 2016) failed in 2005, the Heywood interconnector linking Victoria and SA tripped offline. And this happened before there was any wind in the state[1].

The red line shows the rate of change of frequency. The rapid frequency decline was stopped by turning off around a third of the state’s customers leading to major blackouts.

Picture1

The example above is a stark reminder that the grid has never been perfect. In these extreme conditions all inertia is doing is buying time by slowing down the effects of the shock (i.e. the interconnector going offline), ensuring there is enough time for other emergency controls to kick in.

In the example, these controls included switching off lots of customers and the automatic powering up of other operating South Australian generators.

Those who advocate for the status quo because of the inertia provided by synchronous generators should be aware that these technologies are far from perfect. For example, they can become unstable at low power output. And there is simply no information available on how effectively these generators can respond to fast rates of change of frequency if they started operating before 2007.

Another concern is that, in comparison to other markets around the world, the National Electricity Market (NEM) is designed to have a very heavy reliance on the inertia provided by synchronous power generators such as coal and gas plants, instead of brining in a controlled automatic generator response more rapidly. There are relatively easy fixes to this issue that will improve outcomes for customers and align the NEM to good industry practice.

Other technologies like wind turbines and batteries can also assist in arresting frequency changes, thus supplementing inertia.

Capability of modern renewable energy and energy storage

Frequency is controlled by the balance of supply and demand. This is analogous to a pool of water in a stream where the water level stays the same if the flow in is equal to the flow out.

Injecting power into the grid will push the frequency (water level) up while drawing power from the grid will pull it down. Arresting a sharp fall in frequency requires a fast-acting injection of the right amount of power to stop it falling.

Modern wind turbines can draw on the kinetic energy in their rotating blades to deliver a fast-acting power injection into the grid if trigged by an event. They can also be flexibly controlled to deliver the correct response to suit the local grid conditions and requirements.

Hydro Quebec has been utilising this technology since 2011 with hundreds of wind these turbines now installed and operating. Wind farms can also turn down their output very quickly if this is expected by grid operators.

Large- and small scale-batteries can also detect a rapidly-changing frequency and either inject into or draw power from the grid. This ‘Fast Frequency Response’ solution is already being implemented in other countries.

It is not hard to see a day where homes around Australia have battery systems that don’t just help with their bills and the use of their solar power they also help to secure the entire energy system from major shocks.

Designing a 21st century grid

No one is pretending that these responses from wind turbines and storage are the same as the inertia from a synchronous generator. But the solution is to provide enough flexibility in the energy system to ensure new technologies are providing these services in a power system which is changing dramatically.

The grid has to move with the times – it has to embrace these differences and utilise every opportunity to deliver the right outcomes for consumers. The following actions will be needed to bring ours into the 21st century:

  • Establish appropriate standards for frequency conditions that apply to all technologies following major events, with a focus on the speed and accuracy of their contribution to arresting the change in frequency following a disturbance.
  • Accelerate trials of fast frequency response from inverter-based technologies to further prove this solution in the context of the NEM.
  • Undertake a review of the existing synchronous generator fleet to understand its performance in response to frequency changes during normal operation and in response to major frequency disturbances.

The paper Arresting Frequency Changes in a Modern Electricity System is available on the Clean Energy Council website.

[1] National Electricity Code Administrator, Report into power system incident on 14 March 2005 in South Australia, 2005.

Tom Butler is Clean Energy Council Director of Energy Transformation

Comments

17 responses to “Managing frequency in a modern electricity system”

  1. AllanO Avatar
    AllanO

    No sign of the report on the CEC website despite the last para in the article

    1. harleymackenzie Avatar
      harleymackenzie

      I cant find it either

  2. George Michaelson Avatar
    George Michaelson

    Inertia is understood by the more engineering-minded to mean kinetic, rotational inertia. The idea that the frequency is a derivation of a system and can be corrected and asserted “digitally” is challenging. The analogy would be iron-core transistors, for voltage supply, vs switched-mode power supply. If you know the former, the idea of the latter is very strange.

    As a former trainspotter, I too love mechanical contrivances which have giant lumps of iron rotating very fast: this is a system which inherently has inertia, and has properties which do not demand intelligence to understand why it maintains hysteresis in a system: because it spins, it continues to spin at the (broadly) same rate (of course, as a function of reactive power changes, it probably slows down, but this is missed in the simple analysis)

    I don’t understand how it works, but I am perfectly capable of believing electrical engineers telling me that non-mechanical, non rotational mechanisms can work well.

    I also believe (from what I read) that the existing investment in generation can be re-purposed to become what is called a ‘condenser’ and run solely to act as a mechanical energy store, to provide this rotationally sourced inertia. I’ve read briefly of it, and it feels like something which offers former generation investment at least some return on sunk capital, for marginal-cost spend to re-engineer.

    Isn’t that worth talking about? a graceful exit from the industry could include a change of role.

    1. Vox Avatar
      Vox

      Part of the problem that Tom points out is how the term “inertia” captures the imagination with large, heavy rotating object that counter rapid frequency changes.
      The best way to look at it is by distilling the outcome we are trying to achieve: a grid without large, fast changes in frequency.

      Having physical inertia is obviously a simple, mechanical counter to that issue. However, you can use fast-response battery systems to provide the same result of a stable frequency without actually using physical inertia.

  3. MG Avatar
    MG

    “Other technologies like wind turbines and batteries can also assist in arresting frequency changes, thus supplementing inertia… Arresting a sharp fall in frequency requires a fast-acting injection of the right amount of power to stop it falling.”

    All true, but fails to consider the role of the demand side will play in future provision of frequency control services. There’s two equal ways to address the challenge, and for too long we’ve only focused solving problems using the supply side… expect the demand side to innovate in this space in the near future.

  4. DJR96 Avatar
    DJR96

    Everyone, including the author, has completely overlooked the obvious. Why not design a whole system that doesn’t depend on inertia and has a completely fixed frequency?
    Frequency doesn’t need to vary. It varying is only a legacy side-effect of synchronous generators varying in speed due to unbalanced systems.

    But in not too many decades there will be hardly any synchronous generation left, and they won’t be able to control and maintain frequency at all well.

    So the future network MUST rely on something else. We can’t continue clinging on to a soon to be obsolete model.

    What you ask?

    You all know off-grid inverter/battery systems provide stable effective power. The inverter forms the voltage and frequency with great accuracy. Even if it is overloaded the voltage may sag but the frequency remains constant no matter what. And it will keep going for as long as there is charge in the battery and something charging it. Often solar of course. And it can be scaled up as big as needed.

    So why not simply install a heap (5-6GW) of inverter capacity and 30 minutes worth of storage for that? If it operated in a coordinated way by using GPS time to keep synchronised, it could provide the grid forming and stability function of the grid. It would replace all of the ancillary services, allow any form of generation to contribute to the network, and allow them to ramp up or down in a controlled manner to suit demand.

    Cost? We seem to be happy to blow a couple of billion on a new coal-fired power station, by why do we baulk at spending the same on a solution that actually solves a whole raft of problems? Instead of what would become a stranded asset that only perpetuates the problems.

    1. Chris Avatar
      Chris

      It has been argued elsewhere that system stability has been compromised since the market has been segmented into generators and FCAS suppliers.

      I would have thought that with ability to easily maintain time across a network to better than a microsecond the most simple approach is to simply synchronise all equipment and dispense with specialised, and expensive, FCAS.

      The term ‘synchronous’ is well past it’s use by date. All equipment connected to an AC network is synchronous! It is just that some equipment, eg rotating equipment, can have problems maintaining synchronicity.

      1. DJR96 Avatar
        DJR96

        System stability has been eroded when frequency standards were relaxed which allowed generators to essentially disconnect governing mechanisms on turbines. Relying on others in the network to maintain frequency. It just seems lazy, in an effort to save a few dollars.

        The network is “formed” by the big rotating equipment (synchronous), everything else is just a barely tolerated supplement to the network that by regulation isn’t allowed to contribute towards forming the network. But as we know that synchronous equipment has it’s own inherent characteristics that adversely effect the networks stability. Seems like a good opportunity to turn the tables and approach the whole thing from the opposite side. Thus solving a multitude of issues.

        There is a clear need to have a means of maintaining synchronicity more accurately than using network frequency alone.
        For a multitude of inverter banks to provide clear network visibility for everything else to follow, they would need to be coordinated much more accurately than milliseconds. One AC sine wave is only 20 ms, and 1ms is 18 degrees of rotation or cycle of a sine wave. That’s where using GPS time becomes effective and it is readily available. And it’s accurate to nanoseconds. If it isn’t done with the utmost accuracy, the sine wave from all the inverters won’t be providing a clear single output, making it too vague and vulnerable to disturbances in the network. Or at least no better than what we have now.

        1. Tom Avatar
          Tom

          While I like your idea I don’t think your suggestion considers voltage phase change due to network impedance. While the first source would establish a network voltage according to the gps signal the voltage seen at the second source to connect will be out of phase with the gps signal.

          1. DJR96 Avatar
            DJR96

            GPS time signal is only used to synchronise the frequency, which is universal grid wide. Even now the frequency at Cairns is the same as at Port Lincoln, even if it is a few degrees lagging or leading. My proposal would eliminate even that (assuming the installations are spread out across the network). This mechanism doesn’t have anything to do with voltage.

          2. Tom Avatar
            Tom

            Thanks for elaborating i have re-read your original post and can see benefit in what you propose. I assume the power electronic controllers would generate their own reference frequency locally anyway and the gps would add a level of accuracy. I can’t see how the problem can persist if the mechanical generation is reduced.

          3. DJR96 Avatar
            DJR96

            Thanks Tom. And yes, it would mean the entire network can run perfectly well without any synchronous generation. Ultimate flexibility.

  5. Peter F Avatar
    Peter F

    This report (which I did find on the website) is just the start. Gas turbine based systems are in fact far more vulnerable than steam turbine systems so the advantages of synchronous gas compared to wind and solar are minimal.

    Turbines have what is known as an inertia constant i.e. the ratio of rotating inertia to maximum power in seconds. Thus the whole rotary inertia of a turbine is equivalent to X seconds worth of power. For gas turbines that is about 3 vs 9 for steam turbines. Because frequency should only vary by about 0.15Hz or 1 Hz in emergencies then a gas turbine can only give up 1-(49/50)^2 of its inertia i.e.4%of its inertia. In turn the total inertia is equivalent to 3 seconds worth of rated power so in effect the maximum inertial response is Pmax x .04 x 3 MW.s for a 200MW gas turbine that is equivalent to 8MW for 3 seconds whereas a steam turbine of the same rating could provide 24MW for the same time.
    In the above example the nadir happened after about 2 seconds so a 200MW gas turbine might contribute 12MW maximum inertial response.

    Whether the inertia comes from a flywheel or thermal turbine or wind turbine it must be replaced so the extra fuel from the governor is first used not just to compensate for lost power elsewhere in the grid but to spin the generator backup to speed.

    Typically a gas turbine can increase output at a rate of about 3-5% per minute, this means governor response on a 200MW turbine will be less than 10MW/minute 160kW/sec. Therefore just to recover the 12MW.s it has just given up should take about 12 seconds. Due to the nature of the control loops and thermal inertia of the system it is probably more like 15-18 seconds before the generator could add supply to the system in the event of a persistent power shortage.

    However that is just one part of the problem. If the frequency does drop, the gas turbine compressor speed also drops, so the airflow into the combustion chambers falls and the output initially falls usually by something like the square of the speed change. Thus the 200MW gas turbine running at a nominal 150MW sees power fall by roughly 6MW so that adds another 20-30 seconds to the response time.

    In the meantime all the other parts of the system are responding at their own rate so you can end up with serious system oscillations which can trigger further generation and/or load disconnects.

    Inertial response is theoretically instantaneous, but governors are set up with a dead band of around 100mHz to minimise stresses on the mechanical system, thus it can be 0.3-1 second before the governor even starts to act which means that the power response when it happens has to be that much greater.

    Alternatively a 12-20MW battery/ultracapacitor system could supply the same “inertial” response because it can ramp up to full power in less than 0.2 seconds. Then the governor on the gas turbine can start adding power to the grid within a few seconds so the system is beginning to add additional power after 1-5 seconds rather than after 30 seconds without the batteries.

    A further potential refinement is that the lack of “real” inertia in the batteries/capacitors means that the storage control system could be set up with a much smaller dead band perhaps 1/10th to 1/5th of the mechanical system so the storage system can respond even more quickly and thus provide better frequency regulation.

    One key benefit of batteries, ultracapacitors etc is that a system of sufficient size can be paired with gas or wind turbines to provide both primary and secondary reserves. For example a 100MW gas turbine supplying secondary reserve has to be running at around 30MW just to be stable. It will have two revenue streams a) the FCAS or reserve capacity payment and the power it is actually delivering. Power prices at times may be less than the fuel cost so if the turbine could have instant start it would be better off to be turned off and not sell any power at all. A 10-30MW short duration battery system can provide both inertial and primary reserves and provide plenty of time for the gas turbine to start from cold so saving fuel and improving system stability

    1. Chris Avatar
      Chris

      You have correctly identified that, in particular, modern rotating generators have very little to offer in terms of ‘inertia’. The problem that is claimed to be solved by rotating equipment is actually caused by the rotating equipment. The frequency control problem is an artifact of the modest ‘intertia’ provided by the very same rotating equipment. Get rid of rotating equipment and the problem of frequency control goes away immediately as non-rotating equipment can guarantee a stable frequency. Ok, so you might be left with voltage sag under overload conditions but we don’t have the problem of systems losing synchronicity. I don’t have much confidence that AEMO has the slightest appreciation of the issues involved in system stability.

      Whilst I agree with the general theme of everything you say, I gas turbines cannot start and synchronise in 30 seconds – more like 10 minutes.

      1. Ian Avatar
        Ian

        A large part of the load would be electric motors – spinning things. How much would these be contributing to the conversion of voltage sag to frequency drops in the overload condition?

        1. Chris Avatar
          Chris

          An overload scenario is when there might be a transient short circuit. One advantage that rotating equipment has over inverters is the ability to cope with significantly greater overload currents. I don’t think that normal loads are important in this case. The problem with generators is that the rotational velocity changes under transient change of load – hence the frequency control issue. This doesn’t have to happen with inverters because they can be programmed to maintain the desired nominal frequency – though the voltage may sag.

Get up to 3 quotes from pre-vetted solar (and battery) installers.