rss
31

Customers missing out as rule changes miss the point on inertia

Print Friendly

The Australian Energy Market Commission’s (AEMC) has made newly announced rule changes, which are intended to ensure sufficient inertia is available as the electricity system transitions away from coal and towards renewables.

The goal is admirable as the changes ultimately aim to put mechanisms in place that support the transition. But does this work align to the bigger picture and the broader systemic view that the Finkel Review proposed?

tom-butler

Clean Energy Council’s Tom Butler

There are many unanswered issues that are not addressed in the AEMC’s work. The Clean Energy Council recently provided two submissions to the AEMC on the proposed rule changes.

  • Submission to the draft rule for managing the rate of change of power system frequency
  • Submission to the draft rule for managing power system fault level

The inertia ‘problem’ is a missed opportunity

The Clean Energy Council understands the importance of carefully considering a changing power system in the transition towards renewables, energy storage and smarter demand management.

A fragmented and uncoordinated approach to the transition has led to declining inertia being the focus of one of the AEMC’s rule changes in question (more information on inertia and renewable energy can be found on our energy security page).

Reducing the size and number of large gas and coal generators certainly reduces the inertia in the system. However, there are underlying trends observed in the current power system that do not reflect the expect behaviour of a system with declining inertia.

Available evidence shows that declining inertia leads to increased rates of change of frequency (or ROCOF). Because the system is ‘lighter’ it reacts more to a fault or transmission failure (let’s call this a ‘bump’).

But, if the system reacts more to a small bump, it should only take a small bump to push it back into line. In other words a lighter system should be easier to control, not harder. As explained by leading power system engineer Bruce Miller, the system’s controls should have to work less, not more.

However, the AEMC and the Australian Energy Market Operator tell us that the system is getting harder to control, not easier. The grid frequency is getting harder to manage and is leaving the normal operating range more than ever.

Figure 1 – the frequency is leaving the normal range more than ever [2]

figure 2

To illustrate the point we can contrast the difference in the control of a Mack Truck and a motorcycle. The truck’s mass makes it hard to slow down and steer, requiring lots of work and power steering to push the wheels and the truck in the right direction.

A motorcycle has low inertia. The front wheel moves the bike and rider easily with a slight twist of the handle bars. In summary, a high inertia system requires more effort than a low inertia system, for the same outcome.

In a power system, lower inertia means smaller, faster and more accurate controls could be used to manage the system frequency.

It follows that a well-designed frequency control regime should be delivering lower costs to consumers because less frequency control service is needed. This doesn’t appear to be the case in the National Electricity Market (NEM) frequency control costs have skyrocketed in just two years.

Figure 2 – frequency control costs are off the chart [3]

FCAS regulation recovery cost

 

Something is fundamentally wrong in the NEM. Frequency control costs have risen dramatically, but the quality of the service has dived.

The issue was first highlighted earlier this year by the engineering team from Clean Energy Council members, Pacific Hydro Ltd [4].

It is now being progressed by the Australian Energy Market Operator and this work which is increasingly demonstrating that it is the controls in the NEM’s coal generators that are driving the decline in performance and subsequently the increase in cost.

More recent analysis from the Australian Energy Market Operator shows that this problem and its costs will become much worse in the coming years [5], and confirm that there is no relationship between the deployment of renewable energy and current frequency control issues across the market.

Figure 3 – No relationship between renewable energy deployment and poor frequency control

 rsz_wind_and_solar_deployment_by_year_in_australia

The potential opportunity of a lower-cost, higher performing frequency control regime has been lost in the AEMC’s draft rules. Lower inertia would be better managed and costs of controlling it would be lower if the frequency control regime called upon fast frequency response solutions available from demand management, energy storage and renewable energy.

The AEMC’s work on inertia has not drawn this link. It focuses on the goal of setting thresholds and standards within the current paradigm where poor power control dominates.

Putting the cart before the horse in this way means that the AEMC’s solution for transmission businesses to make investments that secure inertia will likely be excessive and overly conservative.

Old thermal coal generators are a liability to the security of the system

While older thermal generators are playing havoc with the grid frequency, the AEMC has also reaffirmed the fact that they are an unreliable source of inertia.

There is no evidence that they would remain in operation through high speed frequency changes, even those expected today. Prior to 2007 there was no standard for generators to remain in operation for high speed frequency changes. All of the NEM’s thermal coal generators, except Kogan Creek, were commissioned prior to 2007.

Unlike the advanced technologies deployed in renewables which can withstand major disturbances and frequency changes easily, there is no basis on which customers should have confidence in the performance of most of the thermal generators in the NEM. These generators may simply shut down during major power system disturbances, having a major impact on energy security. Relying on them to provide inertia to support the grid is untenable.

The AEMC’s final ruling must make it clear in the National Electricity Rules that only generators with clearly stated and known performance during power system events can provide services that support these events. Testing and certification of this capability must be a requirement to register to provide the service.

[1] http://reneweconomy.com.au/inertia-power-system-dont-actually-need-much-65691/

[2] Australian Energy Market Operator, ‘FREQUENCY MONITORING – THREE YEAR HISTORICAL TRENDS’, December 2016, page 4.

[3] Australian Energy Market Operator, ‘Causer Pays Workshop 2 (presentation)’, April 2017.

[4] K. Summers, ‘Fast Frequency Service – Treating the symptom not the cause’, February 2017.

[5] Contact details can be found here https://www.aemo.com.au/Stakeholder-Consultation/Industry-forums-and-working-groups/Other-meetings/Ancillary-Services-Technical-Advisory-Group

Tom Butler is head of policy for the Clean Energy Council

  

Share this:

  • George Michaelson

    Personally, I believe this narrative. There is no strong evidence of a smoking gun pointing to renewables as the source of incoherency in the management of frequency.

    But I also believe (regrettably) that you’re going to fail to bring a significant community of newspaper columns with you on this: the story is going to be far easier told blaming renewables, absent any proof.

    It would be good to think the regulators and industry bodies know to say whats going on. Again, I am very unsure they see a need to comment, or a benefit. Shame on them.

  • Chris Fraser

    I must have some knowledge to catch up on before understanding the system like the CECs expert views. Skeptics like me think the current FCAS market is an excuse to overcharge.Inertial-type generators from long ago have been excused from needing properly governed output in terms of voltage, frequency and phase acuity – how did we get here ? It’s not like the technology needed disappeared. If anything, technologies such as inverters have provided additional solutions.If a spinning generator wants to take some of the load, why should it not be solely responsible for properly cooperating with the system ?

  • Mike Westerman

    IMO at present there is not enough control over loads to make full use of FFR. In a mini-grid or behind the meter embedded generation plus load centre, you can use very fast reacting load shedding schedules to not have to worry about inertia, but currently in our system short of shedding whole feeders or selected large loads, we don’t have access to that facility, even tho’ we should be getting on with obtaining it (and maybe the AEMO EOI process recently is a start).

  • Aluap

    I don’t trust the analogy between the Mack Truck and the motorcycle. Why not take a few more words to explain exactly what is happening in this short article?

    • itdoesntaddup

      It’s wrong. If the analogy has any value it would be to say that you don’t need a big generator with massive inertia to supply a small village (though it will need to cope with large percentage changes in load in short order), but for sure you do when you are supplying a big city and its surrounding industries.

  • Chris

    Whilst the reference to Mack trucks and motor cycles may not inspire confidence, the article [1] should as it makes reference to some straight forward control systems theory. I agree with the this and the earlier article. There seems to be a misplaced “belief” in the importance of inertia, when inertia is in reality pretty useless in frequency control. Whilst inertia does provide dampening to step changes of load, it comes at a cost of response time of a governed system.

    Frequency control is a problem caused by rotating generators, not solved by them. Inverters used in solar and wind do not change frequency as load changes – rotating generators do.

    Rather than a system where governed, and ungoverned generators attempt to follow a FCAS, all inverter based generators should be phase locked to a standard frequency source, and all rotating generators should simply follow – as they are not synchronous, they must.

    • itdoesntaddup

      You fail to take account of the law of the conservation of energy. If you maintain the frequency, but the supply/demand balance changes, then the voltage will change to reflect the lower or higher supply from wind turbines and solar panels, or changes in demand side uses. Inertia allows an immediate injection of stored energy or absorption of surplus quite automatically, at the temporary expense of frequency (there are outfits trying to market flywheels as sources of inertia for renewables heavy grids). Energy supply at the generator can then be adjusted to restore frequency. Only if the change is abnormally large and unexpected might additional capacity be needed.

      Wind and solar cannot load follow unless they are connected to ancillary devices, and even then their ability to do so is limited by the weather. Loads have to follow the available supply – ie. you end up with load shedding.

      • Chris

        I think you are overly focussed on rotating generators. Frequency is irrelevant (nothing to do with the law of conservation of energy) for inverters, but you are right that in the absence of stored energy the voltage must change. Inverters do have a small amount of stored energy – enough to hold up over a few cycles. Wind turbines have more physical inertial energy than modern gas turbines, on a per MW basis, and can supply or sink power for longer. And modern wind turbines can be operated below maximum output to allow them to automatically increase/decrease output on demand. FF generators also must be run below maximum capacity if they are to supply more power.

        But you would probably know that the amount of inertial energy in the grid is so modest that it can only provide frequency “stability” (the stability comes as a cost of a frequency change) for seconds – far too shorter time before a governor can respond to a large event.

        The point of the article was that the more inertia in a system, the slower the feedback mechanism must be to respond to changes, and this does not help maintain a stable system.

        And my point is the frequency problem is caused by the presence of rotating generators – not solved by them. If there were no rotating generators there would be no frequency problem. Put an inverter between a rotating generator and the grid and there is no frequency problem and no potential disconnection problem that would occur if the rotational velocity became out of sync with the grid frequency – they, otherwise, must disconnect in such a circumstance.

        I agree that the introduction of intermittent energy sources does bring issues that historically have not been well addressed, and many prefer to ignore. The points you make should not be ignored, but if you remove your team colours for a moment you might be able to see that as each problem is identified a solution is inevitably implemented. It seems that on a MW basis (though dispatchable might be different) renewables are becoming economically dominant.

        • itdoesntaddup

          You have a problem in SA that you can be a third dependent on one source of supply – the Heywood connector – a level of risk that normally only occurs with small island systems. Lose that, and you are left with very little inertia on the generating side of the system and little chance of adequate backup, and system black beckons. The grid in Victoria or NSW has more than adequate inertia to handle normal variations in grid load – or even whole power station trips – and the ability to recover after the event. The risks in NSW are simply because of a lack of capacity, as occurred last summer in February.

        • Mike Westerman

          I don’t think that is at all correct: you will always have rotating machines as loads, and as long as you are using AC you will have time constants in the inductive and capacitive elements of the network. Many of these are quite non-linear with respect to frequency and load, hence the network looks more like a sea of waves than like a ripple on a pond. That’s where the motorcycle analogy falls down: there is significant hysteresis and non-linearity in the system, and control systems need 10s or 100s of milliseconds to discriminate and act in an orderly and predictable fashion.

          Hopefully the AEMO trials on synthetic inertia will be positive, and active control systems can be brought into the challenge but rotating machines like steam sets in solar thermal and hydro generators need to be brought in to compensate for loss of inertia in FF generators and loads.

  • Peter F

    I believe one of the reasons it is leaving the normal range more than ever is that the deadband has been opened up at the request of the coal and gas plants. This makes it easier to handle small mismatches in supply and demand but it also means that the response rate of the system (stiffness if you like) is lower so when a larger frequency error occurs it is harder to correct it. It is like trying to drive a car with lots of play in the steering

  • stalga

    The truck/motorcycle analogy worked for me. It sounds like a high renewables grid would be better “tuned” in future, as well as easier to stabilise the bumps.

  • Les Johnston

    Control system modelling of the power grid suggests that grid stability parameters have changed in response to fast response elements introduced through renewable technology. Hence there is truth in the claim that renewables have “caused” the problem. However, the actual problem is the system rules which have not been designed around responsive fast acting generators in the form of renewables. The failure to change the rules is the problem and this has cost electricity consumers millions of $. A helpful analysis.

  • Mark W

    The truck/motorcycle analogy (or actually just the explanation of it) misses a very important element of control systems which is the round trip delay of the control loop. Lighter and faster-acting systems need tighter control loops but the propagation speed of changes across transmission lines and (probably more importantly) the speed of reaction of legacy control elements still in the system remains the same, so you inevitably push the poles of the system towards the right half-plane and instability.

    It’s no accident (excuse the pun) that deaths per motorcycle mile far exceed deaths per mack truck mile.

    I have no doubt that new technologies can solve these issues and that if you were building a new system from the ground up it would not be that hard. But I’m also pretty sure that the transition of the system is more difficult than just bolting some fast-acting elements onto a slow-reacting whole. Some caution and small, careful steps are required.

    • Mike Westerman

      Mark you beat me to a response but you explained the problem very well, along with Les below. We are going to have all those legacy components on the network for some time, but will always have a dynamic configuration in the network itself with changing reactance that makes accurate modelling impossible. We will always need to set gains conservatively, and test subtransient characteristics of generators empirically. Dynamic control of loads in concert with inertia-less generation will gradually become more prevalent but I don’t see a day when inertia can be dismissed so casually.

    • Chris

      I agree dynamic modelling is a non trivial task where you have generator frequency dependent on dynamic loads and network distances spanning sub wavelengths. I would argue that it is unrealistic to expect stability in such a geographically dispersed network where frequency is expected to be controlled by a diminishing number of rotating generators. It becomes much simpler if the frequency is fixed, and the network split up into smaller segments. The frequency would be fixed by locking all inverter based generators and having the rotating generators simply following the grid, as most currently do, or simply interposing an inverter between the rotating generator and the grid.

      You would no longer have a frequency problem. Of course there would still be a voltage stability problem! If you are still stuck on the idea of inertia, an (ideal) inverter with a fixed frequency can be modelled as a generator with infinite inertia.

  • Ray Miller

    Thanks Tom, and I like many others have Zero confidence in the AEMC to come up with any rules which govern the NEM’s technology. The AEMC have repeatedly demonstrated they are not fit for purpose.
    The issues of management of the technical complexity of the grid’s operation should be removed from AEMC rule makers and what we need is new engineers with new thinking capable of solving 21st Century problems, not more from the existing mob unable to determine what problem to solve.
    I’m sure in Australia we must have at least a couple of engineers capable to defining what the technical problems of the NEM are, and then come up with possible long term least cost solutions? And no I would not ask any of our politicians for solutions, nor would I listen to any utterance by them, that would be insane.

  • itdoesntaddup

    Grid inertia is derived from all the rotating machinery connected to the grid – motors as well as generators. It is amazingly blasé to suggest that if renewable generating devices are made tolerant to grid frequency changes that user equipment is unaffected. You can wreck it, or shut it down, potentially playing havoc with manufacturing processes, or requiring expensive investment in power conditioning equipment to isolate the bad grid beaviour.

    There is no way that high inertia generators can be a source of bigger grid frequency problems. On the other hand the spaghetti chart here:

    http://anero.id/energy/wind-energy

    • Ren Stimpy

      These fluctuations you are so worried about won’t be a problem as more and more grid storage power packs make their way onto wind and solar farm sites. Even a small amount of grid battery storage (relative to the output of the site) is enough to smooth out renewable power supply of the site to avoid those issues. This is why it was recommended by our chief scientist, but it also makes economic sense, as grid batteries become cheaper by the year (neodymium supply is plentiful) and wind and solar generators want to see their generation stored rather than curtailed when the wind is blowing off the handle and/or the sun is blazing like a bastard.

      Without storage this only becomes an issue when there is higher than 40% renewables penetration on the grid. Beyond that, storage (that can be instantaneously discharged) is required, and we are seeing the significant uptake of those grid batteries now in South Australia as they go beyond 40% renewables.

      • itdoesntaddup

        Grid batteries and associated electronics are untested as a system to support a major renewables grid on a long term basis with no other support. We do not know what their cost will prove to be in such a role, nor do we have much idea about the behaviour of the whole system when adopted.

        • Ren Stimpy

          Yes South Australia is actually doing real innovation, whereas our politician PM Turnbull is just fapping on about it, AND criticising SA for striving for real innovation in the same breath.

          • itdoesntaddup

            Blackouts and load shedding – the new innovation in electricity supply.

          • Ren Stimpy

            Storage of the type South Australia are introducing reduces the risk of blackout during large storms. As I have previously told you, the coal-fired states here as more prone to blackout from storms than SA, just look at the history. Load shedding during storms and load shifting generally to balance demand can both be achieved intelligently with smarter systems to bolster the reliability of networks.

            We seem to get locked in these circular arguments every time, you and I. If only we could sell the inertia!

    • Chris

      It is a pretty horrible chart, but did you notice that the average of all is remarkably stable?

      Furthermore, any frequency change that might occur is of course the fault of rotating generators – as inverters don’t change their frequency on changes of load.

      • itdoesntaddup

        You assume there are no inertial loads on the system. Perhaps when you have killed all industry that will be true, but then you may not be able to afford much electricity anyway.

  • DJR96

    Sooner or later the NEM must be able to operate in a stable, secure state with very little or even no synchronous generation at all. Has anyone else come up with a solution to that?
    It is entirely technologically feasible to be able to reform the NEM such that frequency is permanently fixed precisely and unwavering at 50Hz. It does not need to be a variable at all. Think about how many issues become obsolete, redundant and irrelevant when that happens.

    • Mike Westerman

      I doubt that is true – as long as the reactance of lines and loads is so variable. You may improve greatly on what we have by controlling load.

      • DJR96

        Controlling loads for the purpose of maintaining stability, to me, is approaching a problem from the wrong end of the stick.
        What customers willingly want to disrupt their processes for the benefit of the network.
        How about ensuring the network is robust enough to cope without having to depend on customers?
        There is a place for load control, but it should only ever be a minor role. Or an option of last resort/emergency action.

        • Mike Westerman

          If you can’t control loads and supply you can’t stop one driving the wrong response in the other. We have a stable system by making it insensitive to power-frequency relationships in the network by adding inertia. If you want an inertia-less system then the supply control has to know what the load controls will do, otherwise they chase each other. You cannot eliminate time and phase lags in a geographically vast system.

          • DJR96

            What control over loads do we currently have? The very biggest customers (smelters, mines etc) are dispatchable loads and participate directly on the NEM. ie. They have to advise AEMO when they want to switch on/off. Most of these same customers also have contracts whereby they can be directed by AEMO to load-shed if absolutely necessary. The Olympic Dam mine site is a good example of that 12 months ago.
            [ As an example, in WA there is/was a wood chip mill whose giant pencil sharpener is driven by a multi-megaWatt electric motor. Normally it is a 24/7 operation, but if they have to restart it they have to call Muja power station so that they can ramp up beforehand.]

            Then there is the off-peak HWS’s. Which traditionally would turn loads on during the night when overall demand is lower. This helped level the load throughout the day and legacy generation didn’t have to ramp up and down so much.
            SA’s system is notorious for creating a demand spike when it turns on at 11pm. But now that State has so much renewables generation, it would be more beneficial to have it turn on at 11am and use the solar power being generated then. The use of it needs to factor in the generation source, not be just a dumb timer.

            Beyond that, no one really wants to be inconvenienced by load control systems. We just want/expect power to be available when we flick the switch.

            Besides, network-wide the actual energy demand normally doesn’t swing too dramatically and can be modelled and predicted quite accurately. This is all used by AEMO to help determine dispatching generation. That’s the way it works just fine now.

            So again, load control should really only be an option of last resort. For when a major unanticipated generation or transmission event occurs.
            ——————————————————————
            Inertia is the inherent resistance to change in frequency of synchronous generation units.
            Note that nothing can respond to frequency until there is a change in frequency. By very nature it introduces a delay in responding to an event.
            Also, the frequency standard requires it to be kept in a very narrow band (I think I calculated it to be +/-0.05% once) which doesn’t provide much manoeuvring room to act within.
            Whereas, if frequency was fixed and voltage was used to monitor the supply/demand balance, you’d have a +/-4% band to work in. And you could more easily tell where the imbalance is, whereas frequency is network wide so is only an overall indicator.
            Time and phase lags can be greatly improved too.

      • DJR96

        Why not? Are you thinking far enough ahead? What synchronous generation will we have in 2040-2050? Many of the coal-fired units will have retired, and those that technically could last that long may have been pushed out of the market unable to compete. So there may only be the hydro-generators, some gas peaker units and the begasse units running on sugar cane waste. Which really could be less than 20% of the total capacity. And that won’t be enough to maintain and control frequency or keep the network in a stable, secure operating state.

        Then what?

        [See that’s the sort of vision that is absent today]