Yet another study has come out, this one for Australia, concluding that all-renewable electricity is feasible while maintaining current standards for reliability. The difference here is the authors, the Australian Energy Market Operator (AEMO). These are conservative professional engineers not green activists. They prepared the report (technically still in draft) because their political masters told them to do it, not out of enthusiasm. As John Quiggin points out :
The AEMO is the body that manages the electricity market on a minute-to-minute basis, so it has the expertise to assess this claim, unlike the many amateurs who have tried their hands. And, since it might have to do the job, it has no reason to understate the difficulties of a renewables-based system.
The extra costs are significant but not absurd. Quiggin again:
Second, the estimated cost of $111 to $133 per megawatt-hour represents an increase of $60-80/MwH on current wholesale prices, or 6-8c/Kwh on retail prices. That’s much less than the increase we’ve seen thanks to the mishandling of electricity market reform.
AEMO modelled two scenarios: slow demand growth and rapid technological progress, and higher demand and slower progress.
The resulting generating mixes are these:
CST means concentrating solar power with storage (example). The capacity numbers are in Tables 15 and 16 in the report.
It’s striking how much despatchable capacity you need (CST, pumped hydro, geothermal, bio) to ensure reliability with majority intermittent sources (PV solar and wind). In the conservative Scenario 2, in 2050:
Solar PV and wind capacity 74 GW 58%
Despatchables 53 GW 42% (of which 18 GW CST and 15 GW biogas)
Total 127 GW
The optimistic Scenario 1 has similar proportions within a smaller total, but CST is largely replaced by geothermal. AEMO assumes successful development of EGS, but not of large-scale battery storage.
The more useful scenario to me is the conservative, little-progress one. It represents a solid worst case. But there are many ways in which things could go better, that is cheaper, which are incompletely handled by the models.
The modelling excluded nuclear by design. This is just as well, as AEMO used a cost database from another government agency (BREE) which is completely out of date on nuclear. To quote commenter Nick on Quiggin’s post:
Unfortunately, on nuclear, BREE’s estimates were in fantasy land, estimating that nuclear plants could be built now for as little as $55-$60/MWh in the low cost scenario – about one quarter of the price quoted to the UK government by the world’s leading nuclear developers.
I would add that the UK has a much longer history with nuclear power than Australia, and hence a far deeper pool of expertise. The likelihood is very remote that real nuclear costs could come down enough to displace any of the other technologies, all of which have normal rather than perverse learning curves.
AEMO did not carry out a sensitivity analysis for say 95% renewable, with the remaining 5% fossil gas. This must make a big difference to the costs. Electricity grids must keep a large reserve of capacity (20% or so) to meet exceptional peak loads and breakdowns, a reserve which is hardly ever used. Gas generators are very cheap. In the pursuit of overall carbon neutrality, it may be much cheaper to offset some gas generation by sequestration, fleets of electric buses, or something than to ensure virgin-pure electricity generation.
The selection of good news in Scenario 1 looks arbitrary. Is working EGS really more likely than cheap grid batteries? A lot more effort and money is going into the latter. (The new US Energy Secretary, Ernest Moniz, recently went out of his way to be nice about geothermal, which his predecessor neglected, so research funding is likely to shift in the US.)
Technical progress in wind and solar, a pretty sure thing, is more or less irrelevant to the capacity mix. Look at this graph from a utility spokesman of the dramatic impact of solar PV on daily load curves in subtropical Queensland in the last four years:
Click for clearer image.
A similar projection for cloudy Germany. It’s clear that solar PV will eat up the midday load pretty well everywhere at $1 a watt, just with incremental improvements along the supply chain. But it can’t do more without storage. 50c a watt with plastic panels or nanotubes just lowers the cost, without changing the result. You still have to buy the same amount of expensive despatchables in some form.
Most countries will face the same problem, but the solution will vary; Japan has poor conditions for CSP, but geothermal resources and mountains for pumped storage; the USA could follow the Australian pattern, with long-distance transmission from the Southwest; Denmark already draws on Norway’s ample hydro. God knows what we Brits can do.
A generating mix that looks anything like AEMO’s models has another implication. They include a very large amount of capacity lying idle most of the time with zero marginal cost: around 25 GW on average in the most challenging week of the year in Scenario 2. Marginal cost pricing will therefore break down, probably long before we reach 100% renewables – the pressures are already evident in conflicts between utilities and solar householders. Electricity pricing will have to shift largely to a capacity basis.
This immense idle capacity will be available at no marginal cost (as long as it’s fully interruptible) for large-scale sequestration or atmospheric synfuel. We don’t yet quite have the technology for these, but will surely need both. Moniz could fund the research by taking money away from the now strategically irrelevant solar and nuclear labs.