Why is wind production being curtailed (or “dispatched down”, more strictly speaking)?

On Friday I posted these initial estimates of 65,746 MWh being curtailed from wind farms across South Australia, equating to more than 6% of the aggregate capability of the wind farms over that period (i.e. the effective capacity – down from the installed capacity because of the amount of available wind).

In Friday’s post I noted that I would provide this explanation – however have been unable to complete this article until today.  Other linked posts will have to wait until later.

It was inevitable that we would end up curtailing supplies from intermittent plant (such as the wind farms in SA over the past 8 weeks) when circumstances demanded.   It also seems almost certain that the rate at which this occurs is likely to increase into the futureparticularly if we continue with a certain set of policies* that are promoting this possibility by skewing incentives.

* The effects of these policies are, I believe, particularly important to understand.  I’ll post more about this later, and link in here…[LINK TO COME]

1)  It might have emerged under either of two scenarios

The emergence of (situational) curtailment of intermittent supplies was always just a matter of when and how much – it’s happening for wind now, and is highly likely to occur for solar, as well.

It might occur under each of two general scenarios:

Scenario #1)  Spillage, when supply exceeds demand

This scenario should be pretty obvious (and non-controversial):
(a)  The solar/wind harvest curve will never precisely match the demand shape for the contiguous NEM.
(b)  The NEM is also too far away from any other major energy consumption center to cost-effectively interconnect (hence we can’t rely on a “just make it bigger” approach possible in larger interconnected systems in Asia, North America and Europe, etc that can help to defer this challenge).

As the installed capacity of “run whenever you like” plant increases, there will be a need to do something with the surplus energy.  There’s essentially three things that could be done:
(a)  Spill the “surplus” generation (or “constrain down”, “dispatch down”, or whatever other terminology you like to choose);
(b)  Time-shift the production (i.e. through storage, assuming that the legacy “demand” shape does not change); or
(c)  Time-shift the legacy demand (e.g. by shifting when we consume to when it is extra sunny, or windy – for instance, by shifting “off peak hot water” to use it as a “solar sponge”).

In this respect, I have been puzzling about the scale of the numbers involved.  I performed some analysis about 2 years ago and posted here on WattClarity before talking at All Energy that year (now just 3 weeks away again), but have not had time to return to this (and update/expand) since that time.

It seems that this form of spillage won’t be necessary in the NEM for years (depending on growth rate of new capacity, etc) but that it will be very tricky to deal with, and possibly quite expensive.  Hence we really need to progress beyond “back of the envelope” calculations like mine quickly!

I am interested to hear of others who are actually looking at real numbers, from a systems perspective, (i.e. beyond first principles (or the “add storage” magic wand), and not just with respect to specific projects like the Kennedy solar/wind hub) to work through the scale of the challenge – if you know of people I can speak with, please let me know?

Scenario #2)  Curtailment, when network conditions require this

Even before scenario 1 emerges, however, there was always a good chance that there would be occasions when network (or market, or other) conditions would emerge where AEMO would be required to impose constraints on certain Semi-Scheduled plant in order to achieve outcomes like a more secure system, and so on…

1.2a)  Currently, because of AEMO’s considerations system security

Following the SA blackout of 28th September 2016, a huge volume has been written.  To help the interested observer, we’ve been compiling this listing of noteworthy analysis from various parties that has followed from this macabrely landmark event.

Following from the event, AEMO has taken a more proactive role in managing system security (particularly in South Australia, given the higher penetration of intermittent supplies and other factors).

1.2b)  How is this being done?

As I noted in Friday’s post, since 19th July 2017 the approach used has been through the permanent use (officially the term is “Invocation”) of a particular constraint equation (named “S_WIND_1200_AUTO”).

I will also posted here [RUN OUT OF TIME, SORRY, LINK TO COME] this brief explainer on how constraints work within the “NEMDE” market dispatch process.

1.2c)  Likely to see increasing amounts of this in future

The main point to understand is that, if we assume the same constraint equation is used into the future, it is almost certain that the “energy foregone” is likely to increase into the future – given the new projects that are being developed for the South Australian region (and given the high degree of correlation of their outputs over time).

2)  A prompt, to progress up the learning curve

We’re all on this journey together, ourselves included (and a big thanks for those readers (and others) who are helping us continue in this journey).  Occasionally we encounter prompts to remind us that we need to take the next big steps – for Semi-Scheduled generators, this curtailment is a big prompt.

2.1)  “Grace Period”

In this broader energy transition it has seemed that an important (but not commonly highlighted) objective has been to accelerate the entry of new entrants into the energy sector, both on the generation and retail side:

1)  On the generation side, the advent of the Semi-Scheduled registration category has worked in conjunction with other incentive structures to make it easier** for new generators to get started – which is a good thing.

2)  On the retail side we have seen a significant number of new retail licences being registered (Guest author, Connor, posted about this in January 2015 – so it would be timely to update this report), and this has been on top of the rise of the Retail Exemption(however retail is not he focus of this post)

Semi-scheduled generators have been (by the nature of the Semi-Scheduled classification) effectively provided a grace period in which they have not needed to learn** a number of the nuances of the way in which the dispatch process works (i.e. to the extent that they would have had they been Scheduled).

** I’ll post more later [LINK TO COME] about this apparent deficit of learning (which has also been exacerbated by the nature of implemented policy).

Whilst this speed-to-production has had some advantages (more MWh produced from renewable sources, keeping in mind the “what would have otherwise been” problem of course), it has has some intended consequences:

(a)  For generators this has led to a number of things:
i.  like difficulties in apportioning “Causer Pays” FCAS contributions (another big week recently), or big shock when receiving a very unexpected bill; and
ii.  surprise (and lack of understanding), when they find themselves being constrained down.  It’s possible, for instance, that this is adding to different forms of conspiracy theories in the absence of fuller understanding.

(b)  on the retail side, it may have contributed to some of the financial difficulties experienced by some of the newer retailers (with 2 having gone bust in recent times, and rumours of others in trouble);

2.2)  A signal that the “Grace Period” is ending

If a new generator were to emerge (say a peaking plant) and register as a Scheduled Generator, then that generator would have a keen incentive to get themselves up to speed very quickly in terms of the nuances of the bidding and dispatch processes in the market – for they would find themselves being dispatched on every dispatch interval.

Whilst one might raise questions about the specific merits of some of the logic in the “S_WIND_1200_AUTO” constraint equation, the broader point is that the NEM has become more complex for wind farmers than the more sheltered existence enjoyed beforehand.  This wake-up call should prompt a desire to learn more about the fuller nature of dispatch and the environment the generator operates within.

3)  Curtailment changes the economics of Semi-Scheduled projects (to some extent)

One of the biggest realisations in this respect should be that the curtailment we’re seeing now changes the economics of Semi-Scheduled projects.  I wonder how many of the current crop of proponents are realising this, and factoring into their business case – or are they just assuming unconstrained output forever into the future?

On the revenue side of the equation, there are essentially the two variables:

(a)  There is the volume of production, which will clearly suffer in the case of curtailment (to the tune of 6% or so over the past 8 weeks in South Australia);

(b)  The other variable is the unit price for production – which is comprised of spot and contract, plus price for LGCs and so on….  (specifically with respect to the “black” component provided of the NEM spot market, I have already noted about the “wind correlation penalty” clearly apparent in South Australia, and this seems to be getting worse).

However it may not be all negative.

If we are talking about a 6% reduction in volume from curtailment (but, at the same time, fewer prices down at –$1000/MWh because of wound-back output from wind farms), then the net financial position for the spot-exposed stakeholders of the wind farms may actually be improved.

That’s all I have time for on a Sunday…

Tags: , , , ,
  • Bookmark and Share

    | More

3 responses to “Why is wind production being curtailed (or “dispatched down”, more strictly speaking)?”

  1. Alex says:

    I couldn’t see it linked in the article but earlier this month AEMO published a report providing some background information and details on the modelling done to establish the system strength requirements that underlie the 1200 MW wind constraints, and the conditions under which it will constrain wind generation.


    It is fairly technical but worth a read.

    • Malcolm M says:

      It’s a good read. So the SA system needs to have enough synchronous momentum + instant demand response + battery to cover the tripping of the largest single gas generator operating at the time. As more batteries are added, fewer synchronous generators would be needed to back up the others.

      There was no modelling of a 100% renewable system on a sunny or windy day, where there would be sufficient ramp up from batteries and curtailed capacity that no unreliable gas generator needs to be “covered” by others.

      • chris baker says:

        I’d like to draw attention to two interesting things about the AEMO report. The first is that despite Kate Summer’s numerous analyses of generator dead band issues, there is no commentary on this issue, and no modelling of how the synchronous generators might respond with tighter deadbands, and the role that might play in limiting oscillations. Secondly, they refer to the possible role of synchronous condensers in supporting system strength, and that they will model this in future. It seems to me that these issues are closely linked. Whilst inertia picks up the immediate loss of a generator, additional power has to be added to the system and this can come only from the governor response or across the interconnectors (or maybe even from the wind farms, because under the modelled scenario they are running curtailed) If the governor response remains slow, it seems to me that wide oscillations will follow. The modelling indicates that they used benchmarking for the synchronous generators that was developed for the system black event. Does this mean AEMO continues to overlook this issue?
        Link to Kate’s article here http://www.wattclarity.com.au/2017/03/fast-frequency-service-treating-the-symptom-not-the-cause/

Leave a Reply

Your email address will not be published. Required fields are marked *