Comparing Like for Like

Since you’re reading this blog, you’ve almost certainly encountered this claim:

We don’t need nuclear because we can use renewables.

For renewable sources like geothermal and hydroelectric this may apply, since they can provide guaranteed generation around the clock. But the former has been abandoned in Australia and both can meet only a small portion of our future requirement for climate-friendly electricity.

But in truth the claim invariably refers to solar and wind, not all renewables. For a scalable technology like solar power to hypothetically meet such a demand profile, storage is implicitly included, or at least invoked upon inquiry. David Green of Lyon Solar described it well:

If we really want to address the penetration of large-scale renewables – and not just be able to satisfy the market you can connect large-scale batteries onto the grid – you need to be able to demonstrate that power generated from renewables can be dispatched with power from the batteries like base-load power, so it’s not creating problems.

However, the size of Lyon’s projects instead indicate a peak demand role in the power market. The megawatt hour (MWh) capacity of their batteries are too limited to supply constant overnight power (not to mention the unlikely economics of supplying at low overnight prices). So the question still remains, what would that look like, and how would it compare to the modern nuclear energy technology some believe it supercedes?

Simplified capital costs over time

In this thought experiment, we’ll use

By multiplying the number of 50 MW class solar plants to ensure that excess generation above this number equals overnight requirements, an idealised “solar+storage plant” can be modelled. Slightly more than 3 Broken Hill-sized plants would be needed but we’ll assume three for simplicity. Similarly, operational costs are excluded for both technologies.

Thus, we can compare assumed overnight capital costs for a NuScale plant, 60 year design life, and twelve solar+storage plants which would hypothetically match its nameplate capacity. As mentioned in the Finkel Review, the lifespan of lithium ion technology is 10 years so the cost of regular replacement has been factored in, in addition to renewal of the solar panels after 30 years (assumed to be half today’s cost).

When the capabilities of the two technologies are hypothetically levelised in this simplified way, it appears that the specific argument on cost is reversed.

Estimated required land area

The area of Broken Hill solar plant is 140 hectares. Thirty-six such plants will need about 5,000 hectares, only slightly smaller than the area of Sydney Harbour. However they don’t all need to be co-sited.

NuScale’s plant, which is now under formal design and licencing review by the US Nuclear Regulatory Agency, will cover a little over 36 hectares, including its maximum required emergency planning boundary. It can essentially be situated anywhere that would be suitable for an industrial facility, as water is not necessary for operational cooling. Notably, other options may well be available for the 2030 timeframe.

Material requirements levelised by generated energy

The US Department of Energy 2015 Quaternary Technology Review estimated various levelised material requirements for major electricity sources. Additionally, silver and uranium requirements can be authoritatively sourced. Charting these estimates illustrates the difference in amounts of materials needed by solar and nuclear, for the same amount of electricity produced.

This doesn’t include the materials like lithium, graphite and cobalt needed for the batteries, which aren’t a power source. It is assumed that materials needed for iPWR (intergrated pressurised water reactor) type SMRs are sufficiently similar to conventional PWRs.

This thought experiment attempts to match solar energy capability to that of nuclear. It hardly needs to be said that the reverse is a much less valuable exercise. Cyling a collection of SMRs daily between 0% and 100% output (with considerably less in poor weather) makes little sense in many ways, not least of which is the consequence of diminshed emissions abatement in a system still overwhelmingly supplied by coal and gas combustion. The whole benefit of including nuclear energy sources is they represent a drop-in replacement for dispatchable fossil fuel fired generators.

There are also commercial scale examples of battery storage paired with wind farms, such as the facility in Rokkasho, Japan. The particular battery chemistry used – sodium sulphur – was recently evaluated in California with sobering results.

We won’t compare the potential emissions savings since authoritative research puts solar and nuclear both at desirably low factors. However, the extra material intensity of batteries may contribute dramatically to lifecycle emissions, depending largely on their country of manufacture.

Solar plants and battery modules can be installed rapidly. In contrast, a certain first time regulatory cost and lead-time for that nuclear plant is unavoidable. Yet it isn’t necessary to overstate this hurdle. In its submission to the South Autralian Nuclear Fuel Cycle Royal Commission, Engineers Australia observed that ANSTO’s OPAL research reactor is of similar size but greater complexity than an SMR unit, and concluded:

The OPAL development at Lucas Heights provides an excellent management example for an SMR nuclear power station in South Australia. Extensive international guidance is available from the IAEA to assist in establishing a nuclear power program…

Australia already has a competent and very well managed regulatory regime with staff with wide international experience. Many of the ARPANSA staff have extensive experience in operating nuclear power plants both civil and military. There is no fundamental reason why the ARPANS Act 1998 cannot be amended to include the regulation of nuclear power in Australia.

The results illustrated here should not be taken as any reason not to build solar, especially paired with storage so as to shift generation to meet high demand, like Lyon Solar’s projects. The importance of this was underscored in the Finkel Review.

However, excluding nuclear energy, with its specific supply profile that can’t realitically be emulated by a variable source like solar, is probably unjustifiable on grounds of cost, land use, material intensity or regulatory challenges. This isn’t intended to downplay the regulatory and public education headwinds the technology faces, but rather to emphasise how important it is – considering the results here-in – to face them now and seriously begin the process. As the Engineers Australia submission noted:

The utilisation of a mix of all low emissions electricity generation technologies will be essential to achieve long-term greenhouse gas emissions targets.

What can be more serious than achieving targets that are aggressive as possible with everything available?


Speeding Up the Roadmap

*only these, but absolutely loads of them

Energy Networks Australia and the CSIRO have released the final version of their roadmap for transforming Australia’s energy supply, to the usual fanfare that these things receive. If you’re not keeping close track, yes it’s different to the efforts from ClimateWorks and from GetUp, amoung various others. Why do there need to be so many anyway?

Anyway, the headline chart illustrates the phase out of coal, then gas, with build up of solar and wind to simply supply all of the terawatt hours we’ll need in 2050 (a terrawatt hour, TWh, is a billion kilowatt hours). Well it certainly looks simple, and at the very least we can seperate out the wedges of renewable energy and take a closer, more critical look.

Analysed the old fashion way: pixels to TWhs.

Large solar PV

In other words, solar farms like the 102 megawatt (MW) Nyngan plant in NSW, which apparently generates about 230 million kWh per year. Judging by the shape of the dark blue wedge, enough of these need to be built by around 2035 to supply 45.6 billion kWh in that year. So that’s roughly 198 farms of that size. We already have Nyngan and a couple of other large solar farms which add up to at least the same output, so make it 196 solar farms in 18 years, or just about 11 per year. Starting now. Then, towards the end of the 2040s, we’ll need to roll our sleeves up again and start replacing these farms as they reach the ends of their expected service lives.

Wind onshore

Again, at around the 2035 mark the light blue share of wind energy is set to begin expanding fast. How fast? To about 183.3 billion kWh through 2050, supplied by the equivalent of 172 windfarms the size of the 420 MW MacArthur wind farm in Victoria (Australia’s largest) at the national annual capacity factor for wind. With 15 years left to build them, we’ll need the equivalent of eleven and a half per year. This is well over 10 times faster than wind has been built in the last decade. Perhaps we can count some of Australia’s existing windfarms at the start of this period, but the fact is most of them will be reaching or passing the end of their rated lifespans in 2035.

Rooftop PV

By the Australian Photovoltaic Institute‘s upper estimate, there was a total national installed rooftop solar capacity of 5,968.341 MW in March this year. Ignoring the need for replacement by 2050 (let’s face it, nobody’s thinking about that anyway) and at the normal 15% annual capacity factor for Aussie rooftops, this is set to grow to 90,650 megawatts (to account for an annual 119.1 billion kWh supply) within 33 years, representing a monthly addition rate of close to 415.5 MW (it’s presently a bit over 60 MW/month) which would look like this:

The arrow indicates today, when we need to start adding rooftop solar capacity almost three and a half times faster than we have in the last year. And not stop for 33 years. Data: APVI

The APVI also keeps track of the current proportion of Australian dwellings with rooftop solar by state. Simply scaling up these figures to roughly 100% for each state (i.e. tripling Queensland and South Australia, up to 10x for Tasmania and so on) yields 25,857 MW. That’s allAustralian rooftops with solar. Obviously we either need more houses or much bigger rooftop systems (probably both), however the CSIRO/ENA’s document is specific about assuming no further subsidies to incentivise addition, so all else being equal it’s not obvious why an individual household would install any more than the kW capacity that covers its own needs.

Two issues are left entirely unaddressed by the headline chart:

  1. A kWh of solar or wind doesn’t serve the same sort of demand as a “conventional” kWh, say from a gas power plant. The hundreds of billions of renewable kWhs appear to more than cover for coal and gas in 2050 at an annual timescale, but week-to-week, day-to-day supply is a different matter. Something more is obviously required when the weather won’t oblige.
  2. Storage of energy is the obvious solution on paper, and CSIRO/ENA foresee a plausible national capacity of 87 million kWh of batteries in 2050. Consider this figure against the 52,000 kWh installed in 2016, and the limited lifespan of these devices (even hoping for 15 years, this would require 5,800,000 kWh worth of battery capacity installed annually till 2050). This is precisely the approach critiqued in the recent review of 100% renewable energy scenarios by Heard and co-workers:

A common assumption is that advances in storage technologies will resolve issues of reliability both at sub-hourly timescales and in situations of low availability of renewable resources that can occur seasonally.

Battery storage is undeniably wrapped in buoyant optimism these days, even though recent large scale operational experience in California points to serious limitations. Additionally, the issue of lifecycle (generally only 10 years for lithium ion) emissions is almost universely neglected in “net zero carbon” scenarios which rely on battery storage. And ultimately, as recently stated by no less than Lazard:

Even though alternative energy is increasingly cost-competitive and storage technology holds great promise, alternative energy systems alone will not be capable of meeting the baseload generation needs of a developed economy for the foreseeable future. Therefore, the optimal solution for many regions of the world is to use complementary traditional and alternative energy resources in a diversified generation fleet.

To be entirely fair to the authors, the document contains some useful assumptions about future energy usage in Australia. It’s certainly worth a flick through. And they do try to account for the required build rate, however it isn’t quite as clear as starting a major new solar farm or wind farm virtually every month for the next three decades, and beyond, like I’ve elucidated here.

Is the future of 2050 sufficiently far from foreseeable? How close do we get before we critically and honestly examine our progress, or lack thereof, and potentially reconsider other energy resources we initially chose to exclude? And how ambitious is 2050 anyway – when including all low carbon resources now may well significantly speed things up, if history is any guide?

Fixing a Power Crisis with a Battery

Mira Loma battery facility, California

Last week the energy products VP of Tesla (supported by CEO Elon Musk) proposed the installation of 100 to 300 megawatt hours (MWh) centralised battery capacity in South Australia, consisting of banks of PowerWall 2 lithium batteries.

Based on figures from SolarQuotes, this represents a rough maximum of 37 to 111 megawatts (MW) of output for about 2 hours and 40 minutes. The amount of MWhs and MWs are very different quantities, routinely confused in commentary and the news; some articles have reported “100 MW”, and GetUp has appropriated the excitement in support of its questionable 100% national renewable energy ambitions:

Elon Musk has pledged to help fix South Australia’s power crisis by installing a 100 megawatt battery system in 100 days, or it’s free!

Exactly how it will fix a state’s power crisis hasn’t been quantified. The example cited in California, the recent Mira Loma facility (20 MW, 80 MWh):

will charge using electricity from the grid during off-peak hours, when demand is low, and then deliver electricity during peak hours to help maintain the reliability and lower SCE’s dependence on natural gas peaker plants.

Expert analysis of the broader Californian battery experience can be read about here.

While the excitement around the news was gripping social media on Friday, South Australian electricity demand looked like this:

The blue line is AEMO 30-minute demand data; the green line annotations simplify the day’s demand into an unseen bottom rectangle of baseload (1,200 MW for 24 hours: 28,800 MWh) and a peaky 7,200 MWh triangle corresponding to the normal daily demand fluctuation. Rooftop PV “behind the meter” consumption is added from APVI data, and is the estimated contribution from about 700 MW of total distributed capacity. The $/MWh price roughly follows this demand curve.

The advantages of battery storage are that it can be installed rapidly (regulations permitting) and can be switched on and off pretty much instantly. Based on the capabilities of a 100 MWh installation in South Australia, and the stated operation of the Californian example, no more than 37 MW could be suddenly supplied over the 2 hour 40 minutes of evening peak, as represented by the red line.

Does this look like it’s fixing a power crisis?

If this proceeds, the manner in which it will help keep state power prices from rising, or even begin to lower them, and how it will relieve the ever-growing reliance on South Australia’s interconnection with Victoria, must be primary considerations. As detailed in the SolarQuotes article, the 30% degradation in battery capacity from only 10 years of use and the limited operational lifespan thereafter needs to be highlighted: no other electricity grid infrastructure is expected to last such a short time. And perhaps most glaringly for many proponents, the potential environmental and social impacts from lithium production in other countries would never be tolerated here. If we’re were instead to pursue an Australian Made battery storage solution to our national power sector’s challenges, many vocal battery supporters need to work out why they prefer one massive foreign-owned hole gouged out of the earth to any other.

Greenbushes open cut lithium mine, Western Australia


Cooking the Comparisons

South Australia’s recent Nuclear Fuel Cycle Royal Commission examined many aspects of the complete fuel cycle, from mining to used fuel management and much in between. Considerable research was contracted to consulting firms. The task of comparing the economic viability of electricity generation in a mix of nuclear, gas and renewables was undertaken by DGA Consulting Carisway.

Sanmen AP1000 units in ChinaTwo nuclear options were included: a single AP1000 with a stated capacity of 1125 megawatts, and a six unit SMR plant rated at 285 MW (comprised of NuScale power modules of 47.5 MW each).

Various analyses were carried out to pit these nuclear options against gas and wind and solar, and the report’s details reveal the fundamental conceit utilised to shape the results: to ensure high value of wind and solar to the simulated scenarios, storage was assumed in unspecified but plentiful quantities which would smooth the output of these sources to whatever extent was required. Specifically, this was distributed batteries for rooftop solar (page 33), and grid storage paired exclusively with wind (page 34), justified only by the assertion that storage technology is advancing and will soon be economically feasible, alongside a reference to a newspaper article superficially covering laboratory work on graphene-based supercapacitors. The fact that one of the report’s authors is the chairman of a company connected to this graphene research – and an “international expert” in electricity grids based on high proportions of renewable energy and storage – is a minor detail compared to the absence of any accounting for the cost of this storage capacity.

Battery storage connected to rooftop solar has enjoyed spectacular coverage in Australia. As of the end of 2016, the closest thing to a guesstimate of current capacity is from page 34 of Solar & Storage Magazine – 31,000 installations, markedly lower than celebrated predictions of 100,000 from Morgan Stanley. But how much is this? If each installation is typified by the Tesla Powerwall, with a rated output of 2 kW and capacity of 6.4 kWh, then present distributed battery capacity in Australia might be about 62 MW and 200 MWh. It would take almost 5,000 times as much as this to fulfil the storage requirements in WattClarity’s 62 hour thought experiment.

In addition, the DGA report invokes an unspecified amount of vehicle-to-grid electric vehicle battery discharge to meet demand. This is now largely moot considering how complex and economically unviable it is understood to be.

oki07Recent thorough experience in california with proven battery storage technology at a larger scale – 2 MW, 14 MWh from sodium sulfur batteries – failed to demonstrate financial viability and anything close to the performance of the scheduled energy sources it is habitually touted to replace. Whatever the authors imagine to be coupled with wind in their models is even less proven. There are alternatives to battery storage of course, and seawater pumped hydro is often suggested, however the preeminent example – the distinctively octagonal Yanbaru plant on Okinawa – was this year decommissioned as uneconomical.

These criticisms are not intended to detract from the capabilities of battery storage technologies and their future roles. Storage should simply be respected instead of over-promised.

And why not allow nuclear output to charge these batteries in any of the modelling? The parallel analysis from Parsons Brinckerhoff also made reference to storage, and on page 19 observed:

…it should be realised that storage cost vs. benefit may well be more favourable for storage in a nuclear generation based systems than a renewable based system…

Without investigating the specific characteristics of South Australia, it is probable that the demand variability is more predictable than the supply variability. Since a nuclear-based system requires storage only to address demand variability, it is likely that the storage requirements to supplement a nuclear-based system and minimise the utilisation of fossil fuel-based assets is less than in a system that is highly renewable dependent.

The fundamental problems with the DGA report’s assumptions render it of limited usefulness, yet its ultimate shortcoming lies not in what it tried to achieve, but rather in what it intentionally didn’t. The modelled net present value of building and operating nuclear power stations was established as negative – unprofitable – when various market discount rates were assumed and applied to their costs. This isn’t surprising – reactors are up-front capital heavy, accruing expensive interest on associated loans, which nobody denies. When a lower social discount rate was applied (under direction from the royal commision) these projects abruptly flipped to profitable in all models (page 88). This lower rate is typical of government loan guarantees, public-private partnerships and straight-out public ownerships, which the authors surmise may be relevant if nuclear plants “were believed” to represent benefits for climate change action and the like which markets may not properly value.

There is no belief required here: it is simple scientific knowledge that nuclear energy is climate friendly. This fact has even been specifically supported in energy sector legislation in New York and Illinois this year. Unfortunately, the authors did not pursue the ramifications of finance-supported profitable nuclear capacity in their models and were content to let their chosen conclusions stand.

Yet, clearly this result should be the basis for further careful analysis of the economics of large modern reactors and small modular reactors in Australia, assuming that the recognition of societal benefits start promptly with the amendment of unjustified federal and state prohibitions. The current top-level review of Australia’s electricity sector can’t be excused from acknowledging the potential of using our uranium, not just exporting it, and where ultra-low emissions energy sources are to receive government support, this can be extended to nuclear with precisely the same justification. Indeed, on-going collaborative research by US National Lab experts in energy supply integration is revealing the benefits of supporting renewable and nuclear energy on the same grid.


With the urgency that serious climate action and electricity sector reform deserves, evidence-led inclusive analysis should begin as soon as possible to enable Australia to imagine, transition to and enjoy a clean, modern energy supply.


Note 1: This critique is also not intended to detract from the commendable work of the Nuclear Fuel Cycle Royal Commission, which can be examined here.
Note 2: Quotes and table reproductions from the DGA Consulting Carrisway report itself have been avoided due to the disclaimer at the foot of page 2.

Storage in the Cold Light of Day

battery-deadPeople want energy in modern society when they want it, and so you’ve go to have supply and demand matching. And, again, there’s a new delusion that’s spreading through the world at the moment which is, “oh yes, now solar is coming down in price, wind is coming down in price, and batteries are coming down in price as well.” People seem satisfied with these simple statements: the prices are coming down so it’s all going to be fine, but they haven’t done the numbers to think through actually how big the batteries would need to be if you wanted to do a solar-and-batteries-only solution.

A solar-plus-battery solution in a place like Las Vegas, I can definitely see playing a large role… Society still needs reliability, though… Society stops functioning if we don’t have a reliable electricity system going all the time, and so for a place like Las Vegas you’re still going to want other technologies in that mix as well. So, I’d advise Las Vegas to get a nuke, for example…

I’m delighted how the book has been helpful… but I’m also still irritated that these delusions about the easiness of getting by with a bit of renewables and a bit of batteries… I think there’s still a lot more to do.

~ Sir David MacKay

It was a relief to hear that sensible projections regarding the role of batteries in Australia’s near-term electricity supply challenge were authoritatively expressed at the meeting of energy ministers last month:

The AEMO told the recent COAG Energy Ministers meeting it may be 10-20 years until battery storage would be able to exert an influence on grid stability and support.

There’s understandable disappointment from some commentators. However AEMO’s sober assessment merely echoes that of the CSIRO.


AEMO itself expects approximately 6.6 gigwatt hours of battery storage distributed amoung rooftop solar capacity by 2035-2036, which sounds like a whole lot more than exists now.

Yet, to get a sense of perspective, 6.6 gigawatt hours would provide no more than 2/3 of one percent of the 62 hour becalmed period described by WattClarity (with a hypothetical ten-fold wind capacity connected to the present national market).

The COAG Energy Council and its independent review process must maintain this realism as it strives to “maintain the security, reliability, affordability and sustainability of the national electricity market” and integrate climate and energy policy. This should encompass a technology-neutral approach, and recognise that avoiding consideration of the future benefits of modern nuclear capacity – potentially available on a comparable timescale to batteries, but historically proven – serves only a diminishing, out-dated activism, when we really have a whole lot more to do.