Will battery storage kill “duck curve” market for gas generators?

Print Friendly, PDF & Email

Australia’s big retailers are planning major investments in gas generation. But California illustrates how that market has been shuttered by battery storage.

share
Figure 3 QLD supply and pool price. Source: NEM Review
Print Friendly, PDF & Email

O ye’ll tak’ the high road, and I’ll tak’ the low road,
And I’ll be in Scotland a’fore ye,
But me and my true love will never meet again,
On the bonnie, bonnie banks o’ Loch Lomond”
Loch Lomond 1841

The case for gas, and the case against it.

Why you should read this note:

  • California is likely to retire about half of its existing largely gas peaking fleet of 21 GW by 2029, even as we in the NEM (Australia’s National Electricity Market) build possibly over 1.5 GW of gas generation mostly in NSW, but some in South Austalia. Can both strategies be correct? We don’t answer this question but instead look at the opportunities for batteries to make a “first cut” before bringing in the gas or pumped hydro.
  • The new buzzword for managing the expected fluctuation in daily average demand by time of day is “fatten and flatten”. Load shifting is largely in the “fatten” section; i.e. shifting South Australian hot water heater demand from midnight to midday.
  • Particularly where PV, and even more particularly behind the meter PV, provides a material portion of energy in the middle of the day there is likely to be a sharp ramp in other dispatchable generation culminating in a short peak around dinner.
  • The steeper and shorter that peak is, the more batteries are economically advantaged at flattening the peak. Batteries’ relative economic advantage over say pumped hydro is in the zero to two hour range. The same would go for battery v gas.
  • After knocking off/flattening the initial 30 – 60 minute peak it takes a lot more storage to make further reductions and batteries will be less advantaged at current prices.
  • If batteries are going to be coupled with solar plants a question is whether they should be on site or remotely located. If on site should they be AC or DC coupled? In practice we expect them to be on site because it’s the forward thinking developers who will actually make this happen. If State Govts got behind storage, as in South Australia, then behind the meter residential would probably lead the charge.
    • Fluence Energy’s new Sunflex product is DC coupled. This allows for capital savings on the inverter, efficiency gains and “free” energy storage via capture of PV output that otherwise would be clipped when panel output exceeds the inverter capacity. The extent of this free energy and the trade-off involved in being unable to island the system or go for most  ancillary revenues are the  issues we have just started to think about. DC coupled utility PV/battery  systems seem to be following the economic model first developed in the residential sector.
    • Most utility scale PV plants and virtually all single axis tracking plants have a panel capacity that exceeds the inverter capacity. A typical ratio is say 70 MW DC to 50 MW AC or about 1.3- 1.4. In general these ratios have increased in recent years.  The NREL’s Paul Denholm notes that clipped energy at conventional ILS ratios of say 1.3 may only amount to 1-3%, so ratios will go up if DC coupled storage comes into vogue.
    • We make some major guesses about clipping ratios and costs of a marginal MW of DC panel and battery cost to look at the economics of adding small amounts of DC coupled battery in NSW. In South Australia it might almost work.

More gas generation is likely in the NEM

Despite the high gas prices, but because of high electricity price forecasts, in turn partly due to high gas prices, we think that significant new  gas generation may be built in the NEM.

AGL has already announced the reciprocating engine 230 MW Barkers Inlet plant in South Australia. It’s true that this is notionally a replacement for TIPS A but the reality is its new generation that will run a lot harder.

AGL in its last results and as part of its NSW Liddell replacement strategy has announced its looking into a new 250 MW plant at Newcastle.

EnergyAustralia has also announced potentially reaching a finance investment decision later this year on as much as 1000MW of new gas generation at Tallawarra and Marulan in NSW

Snowy Hydro executives have mulled the possibility of a new gas plant being built in NSW prior to completion of Snowy 2.0 pumped hydro scheme.

AGL will lose market share in NSW generation, as it has lost it in gas retailing, if it builds just 250 MW and EA builds 1000 MW but that’s another story.

Neither AGL, nor EA, nor Snowy, nor Origin want to own renewable generation. All prefer to buy it from others, or in AGL’s case to take a minority equity investment in a renewable funding vehicle.

But by trying to play both ends of the candle, that is keeping prices high to customers but buying cheap from renewable developers, they run the risk of being cut out via the corporate PPA.

In particular they allow room for a renewables focused generator, eg Goldwind or Neon or RES, to back fill its renewable portfolio with firming capacity and start going direct to customers.

Long live the market, but again it’s another story. We should stick to gas today.

Fatten and flatten in California

Outside of the coal “Boom and Bust” report from Coalswarm, Greenpeace and the Sierra Club the most interesting piece of research we’ve read recently comes from NREL’s Paul Denholm, who apparently did some of the early investigation of the increasingly infamous California duck curve.

That curve development is four years ahead of its original schedule seen in 2010.

Figure 1 California electricity demand net of PV supply. Source: CAISO

It shows that California ramps up 15 GW of power.

So far California, having seen the duck curve coming, has dealt with it via flexible “shoulder” (or should it be “neck”) plant and by trying to shift demand.  In best American naming practice this is the “fatten & flatten” strategy.

Clearly though there is still a very steep ramp required. Everyone recognizes that the challenge of going from 50% renewables to 100% renewables seems to be tougher than getting to 50% renewables.

In California the challenge is arriving rapidly not just because the penetration of PV has been more rapid than originally forecast but also because a significant portion of the gas generation is likely to retire.

Gas generation will retire both because of age but also because of Californiaa’s “once through cooling” rule which will prohibit using sea or river water from cooling and then returning the heated water to the river or ocean.

The once through cooling rule will  require about 11 GW to be shuttered by 2029

According to Denholm’s recent NREL publication California [CA] has around 21 GW of peaking generation capacity. About 12 GW of this capacity is at least 40 years old and in the USA the average of shutdown gas plants is 44 years.

Figure 2 CA peaking plant by installation date. Source: NREL

The chart shows that very little plant was installed between 1980 and 2000 and for a grid of CA’s size not all that much even since 2000. The plant likely to be closed has capacity factors that average around 15%.

How can battery storage fit assist with “fatten and flatten”

One feature of the duck curve easy to overlook is the bump in the curve in the evening when peak demand falls off. That can be seen in Fig 1 up around 6 pm. However we can also see the same thing in Queensland over the past 12 months.

We start by looking at the daily average supply and pool price for the past 12 months.

This is the period when QLD has acted to keep prices down. You can see the price spike around the 6-7:30 PM

Figure 3 QLD supply and pool price. Source: NEM Review

If we then look at a closeup of the above graph and measure Qld thermal demand +  rooftop PV what we see is there is a peak in thermal demand around 7 PM that is 60 MW higher than 6:30 PM.

So a relatively small battery 60 MW/30 minutes could knock off that peak. 30 minutes is already in the area where batteries have economic advantage over say either gas or pumped hydro.

As it happens that half hour is about the half hour where daily peak average price has been at around $110 MWh.

Figure 4 Thermal demand in QLD 12 months to March 28, 2018. Source: NEM Review

However, the task for batteries gets more challenging as you try to flatten the thermal supply curve further. To knock of the 5PM to 8PM peak takes about 300 MW of power with 3 hours of storage or about 30X more expensive than just the peak half hour.

And then of course at the other end the recharge requirement will be longer and likely more expensive per MWh. Actually the chart below shows that there could be a case in QLD at the moment for 60 minutes of battery covering the 6PM to 7:00 PM slot and would take 110 MWh of battery.

That’s still in the battery sweet spot of price compared to the cost of pumped hydro or open cycle gas.

Figure 5 QLD peak first 30 minutes is easy. Source: NEM review

No doubt even more pointed examples could be found in South Australia.

Figure 6 “Operational” demand in Sth. Aust. 12 months to March 28, 2018. Source: NEM Review

At present you need about 26 -29 MW of battery for about an hour to flatten the peak. Again this is well with the time duration where batteries are economically advantaged compared to other forms of generation.

Equally demand management might be cheaper again. But actually we doubt it.

Figure 7 Flatten one hour in South Australia. Source: NEM Review

What is the optimal location and method of connecting the storage?

We think that batteries are most usefully located on the “fringe of the grid” that is behind the meter located right next to the point of consumption.

However, in this section we review a second Denholm paper that looks at the advantages and disadvantages of AC v DC coupling and on site v remote location for PV + Storage in the USA.

Most utility-scale PV plants and virtually all single axis tracking plants have a panel capacity that exceeds the inverter capacity. A typical ratio is say 70 MW DC to 50 MW AC or about 1.3- 1.4.

In general these ratios have increased in recent years. The main reason historically has been that panels only rarely produce  at their rated capacity.

However, in recent years a second reason has been attempts to produce output both earlier and later than fixed tilt systems. It costs more to put extra panels in but total output increases and some of the output is more valuable.

In  NSW the figure below shows that over the past 12 months Moree single axis tracking plant had about a  38% higher capacity factor (output expressed as index) and would have earned 40% more spot revenue.

However, even Moree couldn’t capture either the early morning or evening peak spot price properly

Figure 8 Rooftop v single axis tracking NSW. Source: NEM Review

Let’s assume that you want to add storage to a system such as Moree. A DC coupled system cannot access many ancillary revenue streams.

The following table shows some very, very, rough estimates of required revenues to earn an annual 6% return. Other ways of looking at this are LCOE or benefit to cost ratios.

Benefit to cost  ratios have the advantage of taking available revenues into account as well as costs. Still we just look at an accounting measure because the assumptions are so crude.

We assume the marginal MW of PV only faces the EPC construction cost of say A$1.10 -$1.20 watt. We assume that the battery, without inverter but with an MPPT can be installed for as little as A$0.8 m MWh.

Most crucially in some ways we assume that his marginal MW of DC capacity is clipped 40% of the time (so its free energy to charge the battery) but can sell its output into the spot PV market 60% of the time.

Under these assumptions the battery is almost economic in the current NSW spot market assuming a 6% annual return is required and that there are no incremental opex costs.

Figure 9 Guessing at storage and marginal panel economics at a solar farm. Source: ITK

The unclipped revenue gets the weighted average spot price of $89 MWh which means the battery has to get $109 MWh. Adding more storage would hurt the economics.

David Leitch is principal of ITK. He was formerly a Utility Analyst for leading investment banks over the past 30 years. The views expressed are his own. Please note our new section, Energy Markets, which will include analysis from Leitch on the energy markets and broader energy issues. And also note our live generation widget, and the APVI solar contribution.

Print Friendly, PDF & Email

3 Comments
  1. Mark Roest 11 months ago

    In the Moree single axis tracking plant example, what is the cost of MPPT (and what is it)? More to the point, what would be the economic value of batteries, installed with control system, that cost US$300/kWh, US$200/kWh, US$100/kWh, and deliver 5,000, 7,500, and 10,000 cycles respectively? It would be really great if you could give the results in ‘required revenues to earn an annual 6% return’, LCOE, and benefit to cost ratios. Thank you!
    Oh, one other question: what battery costs, for what years, have you been assuming in your model up to this point?
    Thank you!

  2. Ray Miller 11 months ago

    Thanks David for the interesting article. You are asking a number of questions which I would have thought the current engineering players would have been all over this (sarcasm)?
    Anyway I totally agree all the technologies should be evaluated for their benefits, disadvantages including location and scale-ability.
    With some minor tweaking, taking advantage of integrated PV and battery storage and a degree of coordinated response including managing load time shifting and efficiency.
    What is missing as we all seem to agree is the overall objective without this most of our efforts are somewhat ineffectual.
    I would also like to point out the urgent need to “flatten” that temperature sensitivity of the load, we do know a number of strategies but incentives are majorly lacking, I should say non existing.
    If we want markets to work (as you are a major fan) currently most of the users have limited to no incentive, baring the every increasing energy bills. This is one of the issues which do need priority without further burdening the significant numbers of the population with no or little capacity to pay anything more, while the high users seem to get added discounts.

  3. BushAxe 11 months ago

    Another interesting discussion David, the sweet spot for batteries seems to be in the 1-2hr peak window. Any longer run time seems to point to PHES or firing up the OCGT which seems more economical and complementary. Again SA will be the example as 200MW of batteries and 500MW of peakers (replacing TIPS A) come online in the next 2 years.

Comments are closed.

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