Exploring the AEMO ISP
Back in July, the Australian Energy Market Operator released the results of some rather intensive modelling for the National Electricity Market: the ISP. With the benefit of a little time to consider what it does and doesn't cover, an examination of some of the key components is in order.
Dr. Oscar Archer for Bright New World
The Integrated System Plan is a modelling assessment of electricity supply by the Australian Energy Market Operator, prepared as part of the Finkel Review and in light of the transformative changes to the supply of energy currently underway.
Depending on your priorities, the ISP either signals the end of coal-fired power, or its importance for meeting demand, or the need to keep it around for as long as possible. It's the latest version of an existing model, or it's a revolutionary roadmap to 50% solar power.
The ISP is certainly a detailed effort, with much of that detail released to the public. It would be an interesting and useful collection of work for expanding the literature on electricity network modelling. However, it is instead intended to guide how Australia's power supply changes over the next 20+ years. As such, any deficiencies will tend to steer us towards a suboptimal destination.
This article isn't intended as a critique but rather as a set of suggestions and observations to further align the goals of rapid and sustained emissions reduction alongside assured system security and reliability. These are:
Gas and its persistent role in meeting demand
Solar: transmission and location requirements, limits to relative penetration in other countries
Wind’s relatively diminished role, turbine lifespan and market contribution factor
Efficiency and Rebound
Electric Vehicles, their future market share and known charging patterns
Energy storage: an attempt to illustrate the scale and the lifecycle emissions of what is proposed
Nuclear energy’s proven and potential role.
As the emissions trajectory specified in the Fast scenario is the most ambitious and closely aligned with that anticipated by Bright New World, this aspect of the ISP will be the focus. It incorporates a reduction in emissions of 52% on 2005 levels by 2030, and 90% by 2050. This target and the replacement of fossil fuel combustion is the right starting point. It's only how we get there which we need to discuss, since, as AEMO itself writes:
"There are extents of [variable renewable energy] penetration that have not been explored, and for which there is no experience globally."
Gas-powered generation, GPG, is a prominent feature of the modelling. GPG is:
"Expected to produce less energy overall, but continue to provide a reliability and security role to complement variable renewable energy. Renewable energy will erode in the next decade the need for GPG to operate during the day, but GPG production is forecast to recover to near existing levels in the second decade as coal retirements increase."
By referring to the 2018 Generation Outlook spreadsheets, this expectation is demonstrated by the presence of 11,823 megawatts of GPG capacity in 2039-40, with a modelled output meeting 5% of annual demand, compared to 11,078 megawatts in 2018-19 which will supply 9.2%.
Solar dominates the supply mix by mid-2040, at 51.5% of the financial year's supply. Three quarters of this is expected from solar farms. To enable this the model requires a) more storage, and b) more transmission. Storage will be covered below, but the Fast scenario assumes all new transmission options go ahead. AEMO's interactive map gives an idea of the scale:
The missing context is that anywhere which isn't desert is probably going to need right of way corridors, like this one west of Sydney.
Now, even assuming the impacts are inconsequential, it is stated clearly that:
"there is high solar energy correlation across the NEM for all REZs."
Substantially more solar capacity can undoubtedly be built and connected in the short term, but in other nations there are now clear logistic trends due to various factors more or less halting the expansion of solar's share. Data also suggests appreciable year-to-year variation in generation at high share - some years are cloudier than others.
How safe is the assumptions that solar proportions in Australia can blast past what other nations have managed, with high correlation and substantial new transmission requirements? And even if factors such as falling costs can act to enable this, will it be sufficient to stay ahead of the increasingly recognised declining value of the energy observed in other regions?
We'll get to the matter of energy storage shortly.
Wind energy has edged out hydro as the NEM's top renewable source, even though 2017 generation was noticeably lower than that of 2016 with 547 new megawatts connected, "the third highest amount added in the history of the Australian wind industry."
Nevertheless, the Fast change scenario does not anticipate much acceleration in addition, and solar dwarfs wind by 2040 in the model. Its share isn’t quite 23%.
This is the last 20 years of Australian wind addition:
This recent paper on wind capacity in Europe suggests an average turbine design service lifespan of twenty years.
"As a consequence, the wind industry needs to prepare for upcoming challenges, such as maintenance of aging assets, assessment of structural integrity, lifetime extension decision making, and decommissioning of turbines.”
By 2040 a full quarter of the NEM's modelled wind capacity will be at this stage. The ISP unfortunately neglects to account for this, even though retirements for coal (and gas) generation are explicit: "It is assumed that approximately 15 GW of generation (14 GW of coal fired and about 1 GW of GPG) will reach its end of technical life by 2040 and retire."
20 years may well be conservative, but as observed at Renewable Energy Focus,
"The tower and foundation of the turbine are... obviously critical to operation, as they can’t be easily replaced. Cracks and corrosion are real threats, and while they can be modelled in design stages, exactly how a turbine will be impacted by the natural environment can’t be accurately predicted. Other potential failures may lie in the electronics – responsible for 13% of failures at present – or in other crucial components like the generator, gearbox and pitch control."
Whatever wind capacity isn't decommissioned at end of design service life will tend to face steadily increasing maintenance costs.
In the past, AEMO has put notably more emphasis on analysing the capacity credit or contribution factor of wind generation. It is dealt with obliquely in the ISP, but this multi-year contribution has been published elsewhere:
The shape of this curve doesn't change with increasing capacity no matter how far capital costs fall.
Energy efficiency improvements are stated as "Strong" for the Fast change scenario. The basis of this improvement is not obvious and remains elusive within the ISP's supporting information. However, ambitious renewable energy-dominated scenarios have a history of assuming markedly dramatic gains through efficiency or intensity of energy use. In 2014 Loftus and co-workers reviewed the existing literature, observing:
"Historical annual changes in global energy intensity... declined by 0.8%/year on average over the last 40 years. ...Even [sustained business-as-usual growth] scenarios require sustained improvements in energy intensity of −1.5 to −1.8%/year, matching the highest annual rates seen over the last 40 years."
More fundamentally, do these strong efficiency improvements neglect to account for efficiency rebound? Rebound is the demonstrable phenomenon of increased energy consumption resulting from below-cost efficiency improvements, i.e. not all of the saved energy is left unused.
This manifests for transportation, heating and cooling, industry, household appliances, and multi-factor productivity improvements. There are substantial benefits in continuing to improve the efficiency of sectors of the economy such as household energy use, but no modelling is served by overstating them or assuming they are without barriers or costs.
The Strong uptake of battery electric vehicles predicts that they will comprise 53% of the fleet in NEM states by financial year 2037-38, consuming 12% of annual electricity (net of rooftop solar output). The Fast scenario has expanded the total number of electric cars 260-fold compared to today. The new material demands of this expansion need to be kept in mind.
Recent research in California, where electric vehicles now number close to 400,000, has demonstrated that people predominantly charge them overnight at home.
This is obviously anti-correlated with solar, but much more closely correlated with scheduled thermal generation which can operate as needed, day and night.
The operation of grid scale storage in the form of batteries has been trialled in California, and the results are relevant to energy planning scenarios which assume an unprecedented scale of bulk energy shifting.
"Charging when electricity is cheap and discharging when it is expensive is what first comes to mind as an obvious use of electricity storage. This time-shifting of generation to match consumption peaks involves techniques such as peak shaving and load levelling; these are easy to envision and model and optimize when looking at yesterday’s load and price curves, but very difficult to do in real-time when the load and price are varying stochastically and neither the height nor timing of the actual load peak can be known or recognized till well after the fact. In practice, energy arbitrage only generated enough revenue to barely cover operating expenses."
Furthermore, "the trial also revealed how different batteries are from actual generation resources." Despite this, the Fast scenario apparently treats close to 22,500 megawatts of utility scale storage as a functional replacement for conventional generation capacity.
Pumped hydro storage (PHS) will inarguably dominate the world's total energy storage capacity for the foreseeable future, and the ISP also cites it as a major component. But this is where simplified assumptions about costs per kilowatt hour, even falling ones, can start to erode a scenario's robustness. Like every natural resource, PHS sites can be ranked in terms of cost effective development of the resource. As such, there are probably real limitations. Though batteries and hydro can't directly substitute for one another, at the massive scale which is being proposed every megawatt and megawatt hour that isn't PHS will have to be batteries.
And while the falling cost of batteries is invariably treated as inevitable, it is also just an assumption: in 2016, the reported installed cost of a 7 kWh battery was "nearly $10,000", while today the average is $10,560 for 8 kWh. Perhaps economies of scale will allow grid-sized batteries to avoid these stubborn costs to some extent, but this so far doesn't help distributed energy resources.
If the assumptions around the costs of storage are common, the assumptions about emissions intensity are universal. Batteries are highly engineered devices full of extremely purified material: a lot of energy goes in to producing them. As they don't produce any energy of their own, the full lifecycle emissions intensity of the electricity they supply has to be added to the energy source which charges them. Even charging with solar and wind, this takes the combined intensity above 100 gCO2e/kWh according to this fairly comprehensive analysis.
While a great improvement on the intensity of today's power network, it's dramatically different to the 0 operational emissions claimed for renewable energy in the Finkel Review final report.
It's not AEMO's role to advocate, in any form, for the inclusion of a prohibited technology, and nuclear energy's absence from the ISP is consistent and correct. This is despite the inclusion of modern modular reactor designs in the Blueprint for the Future, the final report from the Finkel Review.
Decadal deployment rates for nuclear compared between countries, per generated kilowatt hour and levelised for population, indicate a formidable promise for displacement of fossil fuelled electricity. To date, solar and wind addition rates have been modest. Growth of solar output in particular in politically committed Germany has peaked at less than a quarter of the rate proposed by the ISP Fast scenario (the rate so far in Australia has been a fifth of what's hypothetically required).
Every objection to inclusion of nuclear energy can be robustly debated on technical grounds, with many of them finding little traction during the South Australian Nuclear Fuel Cycle Royal Commission process. For example, even with the entire fuel cycle, construction and decommissioning considered, nuclear's lifecycle emissions intensity is as low as that of wind. One result of modelling carried out for the royal commission in 2015 conservatively estimated the timeframe for reactor build in Australia:
'Low' and 'high' refer to capital and operating cost scenarios. The analysis provided a good starting point at the time; today it's obviously in need of fundamental revision and expansion, as part of a comprehensive evaluation of the economics, operation and regulation of modern nuclear in Australia. As the royal commission concluded:
"The Commission did not find that nuclear power is ‘too expensive’ to be viable or that it is ‘yesterday’s technology’. Rather, it found that a nuclear power plant of currently available size at current costs of construction would not be viable in the South Australian market under current market rules."
We are currently in the process of working out how to evolve the electricity market and its rules in arguably fundamental ways. One of those is low emissions. Another is reliability. Again, according to the royal commission:
"Storage and power-to-fuel technologies also offer the potential to displace capital expenditure on the transmission and distribution networks. However, if the expected reductions in the cost of these technologies are not realised, the potential for nuclear power to provide reliable generation capacity to balance the intermittency of wind and solar would be improved."
At Bright New World, we assert that the proven potential of nuclear energy is more than sufficient to warrant immediate, equitable inclusion in the design of Australia's future energy supply network, even if optimistic forecasts for the costs of other technologies prove accurate. The path to similarly cost-effective modern nuclear capacity has recently been enumerated - widely publicised excessive costs aren't inevitable or even particularly common.
Most importantly, the Chief Executive, Audrey Zibelman, of the Australian Energy Market Operator is fully aware of the value of nuclear energy, having unrelentingly supported it's continued operation in New York state:
“Due to the vital importance of affordable, reliable and secure power as the engine of a strong economy, care must be taken now more than ever to manage this transformation in order to minimise costs and risks and maximise value to consumers.”
The recently abandoned National Energy Guarantee model presented an opportunity for a solid long term policy framework, as both its mandate and designers understood the importance of a) reducing emissions per unit energy generated, and b) requirements for system strength. It encouraged long-term thinking, which AEMO’s ISP certainly is, but in contrast the ISP is prescriptive and fails technology neutrality.
It's time for actual technology neutrality, with emissions being the primary decider. We need all the tools, and we need to take the utmost care.