Sceptics, look away: a mostly renewable electricity grid is highly feasible

Australia’s electricity could be over 95% renewable by 2035, according to the latest release of Australia’s electricity market operator’s Integrated System Plan. But many people are sceptical. They regularly witness days with little wind and sun and deduce that relying primarily on wind and solar generation will lead to disaster. They firmly believe that 24×7 baseload generators are an essential ingredient in our energy mix.

Seeing countless arguments of this variety on social media, I decided that a useful contribution to this debate would be to model a near real-time simulation of a mostly renewable electricity grid. Whenever someone claimed that a day wasn’t particularly windy or sunny, I’d be able to point them to the day in question and they could see how the electricity supply may have looked on that day.

In August 2021 I started running a simulation of Australia’s main electricity grid, the National Electricity Market (NEM), with the results posted every week to social media. The goal was to show that a sufficient quantity of wind and solar, backed by existing hydro and 5 hours of storage (24GW / 120GWh) would enable the electricity grid to get very close to 100% renewable at a very affordable price.

Some rough optimisation experiments suggested 2-5 hours of storage is likely to be economically optimal. I chose the higher end which gives a higher renewables penetration.

The simulation uses actual wind and solar generation, rescaled to approximate what it might look like if we built enough to supply a little over 100% of annual demand.

After more than 2 years of weekly simulations, my model has achieved a 98.8% renewable electricity supply at an estimated average wholesale cost of $95/MWh, a similar cost to what it has averaged over the last 8 years.

There are a couple of caveats, but they don’t change the outcome of the modelling.

Some days aren’t particularly windy or sunny. And 5 hours of storage is clearly not enough to sustain supply during a long, windless winter night.

But Australia is lucky that wind generation tends to be above average during the night and during winter. Australia is also lucky that it has a reasonable amount of hydro power stations that are well suited to running primarily when there is a shortfall of wind and solar. Most of these stations are limited by annual water inflows, so they can’t be run at high output for a large number of hours in the year. But my simulation shows that if we run them primarily during nights with poor wind, they can make an extremely important contribution in firming a mostly wind and solar grid.

However, there remain some periods, nearly all in late autumn or winter, when the hydro and 5 hours of storage are not sufficient to fill these gaps in supply from wind and solar. It would be very expensive to build enough lithium batteries to fill these gaps. My model fills these gaps with what I call ‘Other’. In the short to medium term, ‘Other’ is likely to be gas or diesel-fuelled peaking plants, providing 1.2% of annual generation to complement the 98.8% coming from wind, solar and existing hydro.

Longer term, there are a range of possible solutions to avoid the occasional use of these fossil peaking plants. They could be replaced or converted to run on green fuels like hydrogen, methane, methanol or some other biofuel.

It is possible that new battery chemistries or compressed air storage may prove cheap enough to use for long term storage.

And “demand response” is likely to play a very important role. For example, in a future world where we are mostly driving electric vehicles, having owners incentivised to charge their cars primarily on days with plentiful wind or solar. The same applies to any somewhat flexible heavy energy user. It is important that there are financial rewards to encourage this behaviour.

Note that the need for ‘Other’ is not unique to a mostly renewable grid. Virtually all grids around the world use gas or diesel peaking generators. Many renewable sceptics are supportive of nuclear as an easier solution to managing supply. But a mostly nuclear grid would also need ‘Other’ to deal with extreme demand events, for example during a sustained summer heatwave, when demand can soar to more than 50% above average levels.

They would also be needed when multiple nuclear reactors are simultaneously offline for scheduled or unscheduled maintenance, or during unexpected trips. It is not economically sensible to build a nuclear plant that is only needed for a limited number of days per year. And filling these gaps with lithium batteries is unlikely to be cost effective.

Some important learnings from my simulation:

Over-generation is very important. It is far more economic to build enough wind and solar generation to supply more than annual electricity demand and have some curtailment, than it is to build lesser amounts and instead rely on huge amounts of storage for excess generation in spring and summer to supply shortfalls in late autumn and winter.

My simulation had enough wind and solar to supply 62% and 50% of annual demand respectively with 17% being curtailed.

There will be tremendous opportunities in a mostly renewable grid for flexible energy users. Anyone who mostly uses electricity during the peak solar hours is likely to be offered very cheap prices. Big discounts are also likely for anyone happy to reduce demand on challenging days.

A near equal mix of wind and solar takes advantage of the fact that these technologies are “anti-correlated” in Australia. Wind tends to generate more at night and during winter than at other times. A bias to wind requires less short-duration storage, but if solar continues to reduce in price faster than wind then that may swing the bias the other way.

Wind in Queensland is not correlated to wind in the other states. The performance of solar generation in winter is much better in Queensland than in the other states. As a result, there should be a bias to excess renewable generation in Queensland making it a large net exporter to the southern states.

Only about 10% of the generation needs to pass through storage. The vast majority is generated at the time it is needed, without the need of storage.

While ‘Other’ supplied only 1.2% of annual demand, on the most challenging day it was required to generate at 9 GW, providing about a third of that day’s demand. Australia’s NEM currently has approximately 8 GW of peaking gas & diesel generators.

To summarise, a mostly renewable grid is highly feasible and likely to be very cost effective for Australia’s main electricity grid. Existing hydro and a feasible amount of battery storage is likely to perform the bulk of the firming to ensure supply and demand match at all times. But there will be days with poor wind and solar where an alternative back-up supply is required. In the short to medium term this is likely to come from gas or diesel peaking generators, though these will eventually have to transition to a non-fossil alternative.