How will the Australian grid of the future, one based on 100% zero carbon generation, differ from the current Australian grid which remains dependent on coal and gas fired power plants? For the purposes of this discussion let’s consider the National Energy Market (the NEM), which covers all of the country excluding Western Australia and Northern Territory. This is the largest electricity grid by area in the world and as shown below remains dominated by dispatchable generating assets – mostly coal and gas with some hydro from Tasmania and the Snowy scheme.
First a couple of important comments – the NEM, like all other electricity systems in the developed world, has a primary function of providing at all times enough electricity to meet customer demand. This is unlikely to change dramatically as generation shifts towards renewables. The NEM is however, different to many other global grid systems, in that it is essentially a standalone system which can not either import electricity if it has a shortfall or export power if it finds itself with an excess.
Current NEM design criteria
As most readers will know, coal and gas plants are regarded as being fully dispatchable, meaning generation can be turned on and off as required to meet customer demand. This being the case, the key design feature of a traditional baseload grid is satisfying annual peak demand – the point in time when customer demand for electricity is at its highest. Grid operators have a pretty good idea when peak demand will occur and based on historical data they also have a good handle on what the level of demand will be. Demand on the NEM peaks at about 35 GW each year during late summer heat waves, more specifically on late summer weekday afternoons when air conditioners are running flat out, industrial and commercial usage is high and household demand is ramping up with families starting to prepare for dinner.
Based on an expected peak demand of about 35 GW, it is reassuring to know that the current NEM has about 43 GW of installed dispatchable power. That gives it a reserve margin of about 8 GW (or 30%) to cover unexpected power plant outages (coal and gas plants do break down so are perhaps mostly dispatchable rather than fully dispatchable). Once peak demand is satisfied, meeting demand for the rest of the year is the relatively simple task of ensuring fuel supply contracts (ie coal and gas deliveries) are secure and maintenance programs are appropriately scheduled.
To complete the thumbnail sketch of the current NEM, it produces a total of 200 TWhrs of electricity each year. This is equivalent to a utilisation factor of 53% for the 43 GW of installed NEM capacity. In other words if all 43 GW were running for 24 hours/day x 365 days/year the annual output would be 375 TWhrs. This is a pretty typical dynamic for most large grids – in terms of capacity it is sized to meet peak demand as discussed. This means that throughout the year there are some smaller units that are not required on a regular basis and accordingly have a low utilization factor. Larger, lower cost units will run for most of the year to meet baseload requirements meaning that the average fleet utilisation is somewhere close to 50%.
If this is the way things are currently operated, how is the transition from coal and gas to wind and solar changing the way the grid is designed and managed?
The NEM with 100% renewable generation
Keeping things simple let’s imagine that the NEM demand remains constant – the highest demand remains at about 35 GW during late summer heat waves with a smaller peak of say 30 GW during a winter cold snap.
From a supply perspective it is expected that wind and solar will provide the bulk of the generating capacity – let’s assume 50 GW of each. With a 30% capacity factor this would provide 260 TWhrs/year, more than the NEM would need but obviously not always generated when required to meet demand. The NEM can also rely on about 8 GW (producing about 15 TWhrs/year) of dispatchable generation coming from hydro – mostly the Snowy scheme and assets in Tasmania. There is also a small amount from biomass that qualifies as both dispatchable and zero carbon power source.
The figure below shows that south eastern Australia has lots of world class wind and solar potential. There are even some good opportunities for geothermal energy across South Australia and Queensland, a resource that remains essentially untapped. Finding enough sites to build out 50 GWs of high quality (i.e. consistent, high capacity factor) wind and solar should not be difficult.
One advantage of a grid as physically large as the NEM is that generation is less likely to be impacted by localised climatic conditions not conducive to renewable power generation. For example, if it is cloudy in Queensland it will hopefully be windy in South Australia or vica versa. To capitalise on this generation diversity will mean significant investment in transmission to allow areas of excess generation to help in areas where generation is below demand. This is outside the scope of the current discussion but a couple of areas where government input would be valuable is firstly to encourage optimum geographical diversity of the growing renewable generating fleet and secondly to create the right signals for investment in transmission.
Electricity Storage
To enable the anticipated 100 % renewable generation fleet to meet expected customer demand, electricity will need to be stored during periods when conditions for wind and solar generation are favourable so that it can be used when they are not.
The first question that needs to be addressed is whether storage technology exists at the scale required by a 100% renewable NEM. As discussed in a previous JTZC article (1), pumped hydro is currently the most mature, proven option. This is a closed loop hydro system (see the figure below) that stores excess renewable electricity, generated when conditions are favourable, as gravitational energy by pumping water from a lower reservoir to an upper reservoir. When conditions for generation are not favourable, the water in the upper reservoir is allowed to flow through a turbine into the lower reservoir converting the gravitational energy back into electricity.
Pumped Hydro schematic
The amount of electricity stored by a pumped hydro facility is a function of the size of the storage reservoirs. The rate at which this electricity can be released will be a function of the elevation difference between the two reservoirs, the flow rate of water reaching the turbine and the size of the turbine itself. Globally, pumped Hydro is the most well developed storage technology representing over 90% of the existing storage capacity.
The other storage technology currently in use is the Lithium-Ion battery. These are most commonly used for small scale household battery systems but could also contribute to the creation of a larger grid resource. Given the importance of storage for a future that relies on renewables there is considerable research currently underway on these and other battery/storage technologies.
In high level terms, pumped hydro and to a lesser extent Lithium-Ion batteries are technically capable of supporting a 100% renewable NEM but this will require a considerable investment. Before this investment can be made, however, detailed studies will be needed to calculate exactly what is needed and how it should be operated.
So how would one go about estimating the amount of storage required – both in terms of capacity (maximum generation rate required from the storage system) and total electricity delivery (cumulative delivery of stored electricity back into the system before it can be recharged)?
Renewables in Summer
When electricity is mostly produced from wind and solar, dealing with summer peak demand should not be the biggest challenge the grid needs to meet. Given that long hot summer days mean plenty of sunshine, the grid should have ample daytime capacity even when demand is at its highest. With 50 GW of installed capacity, solar on its own will comfortably power the entire grid during daylight hours while providing excess to charge up domestic and small scale commercial scale Li Ion battery systems. Wind generation during the day can go straight into storage, perhaps preferentially into larger scale systems. At night the grid will run on wind, power stored during the day and hydro if needed.
The role of hydro under this scenario highlights a key change that will probably be needed as we move to a high reliance on renewables. If the grid needs, for example, to keep hydro capacity full to ensure some dispatchable power on summer evenings what incentives are needed to ensure this actually occurs? From a commercial perspective the obvious approach, one used in other countries, is to pay generators for providing back up capacity. Currently Australian generators are only paid for the power they actually generate. In the scenario outlined above, dispatchable hydro assets have value as a backup to wind during the evening. They should be compensated for not producing electricity during the day and hence being ready with full capacity, or as full as they can, in the evenings when solar generation starts to fade. If there is sufficient wind to run the whole grid overnight then the hydro system should continue to be paid not for competing with wind but for remaining full and able to backup the system when needed.
Winter Renewables Drought
If a grid based on 100% renewables generation should be able to cope reasonably well with the summer peak, what is the scenario that will cause the most concern? Modelling studies (2,3,4) have examined historic weather records to determine the likelihood and duration of periods where there is not enough sunshine or wind to meet expected demand. As one might expect, solar and wind generation shortfalls are typically observed in winter when there is an extended period of cloudy days with low wind over a significant portion of the NEM generation footprint. The magnitude of this renewable “drought” appears to be quite variable on a year to year basis, an observation which will make grid design more complex.
Analysis of weather data (2) between 2006 and 2015 showed that it was quite common for output from a hypothetical NEM with 50 GWs of wind and 50 GWs of solar to drop as low as 15 GW, well below the expected demand of around 30 GW. In the absence of demand reduction strategies this means the storage system would need to routinely supply over 50% of the electricity required by the grid. This is perhaps not unexpected given the intermittent nature of renewables and confirms that grid storage requirements will be significant.
What is perhaps more important is the duration of these supply shortfalls. During summer they were typically less than a day or two and were both preceded and followed by periods of excess generation allowing the storage to be replenished. In winter this is not always the case – there are typically extended periods of low generation. It is these periods which will set the total energy requirement of the storage system – the longer the shortfall, the longer the period during which the stored electricity reserve needs to run without being topped up. In a worst case it appears there can be multiple weeks when demand exceeds supply and stored electricity is needed to supplement wind and solar. While the shortfall during these periods is not always as high as 15 GW there is a consistent shortfall meaning no new electricity can be added to the storage system which is being continuously depleted.
This gets us to the result we were looking for – the specification of the storage system. Based on a study of a 10 year period, NEM capacity needs to be able to run at something like 15 GW for a month. This equates to being able to deliver 11 TWhrs ( or 5% of the annual NEM output) without needing to be replenished.
Summer Peak Demand vs Winter Supply Drought?
So what have we learned? The traditional grid based on despatchable coal or gas plants looks relatively simple – it is designed around meeting the summer demand which is pretty easy to predict. Next year it will be about 35GW give or take a few percent and will last for a day or two until cooler weather sees a bit of a drop in demand. Wait a few months until autumn weather starts and demand will drop to about 30 GW.
In terms of asset allocation 50% of generation needs to nominally run all the time with the other 50% is scheduled for intermittent operation. Some smaller units will only be needed for a few weeks each year and the grid operators have a pretty good idea when this will be.
Managing a winter supply drought with 100% renewables will be a lot more complex. In 2010 there was a month long drought that would have required a 15 GW/11 TWhr storage system. Perhaps in 2055 when we are trying to run with 100% renewables, the supply drought will last for 6 weeks or perhaps even 8 weeks and require a system twice as big. Unlike the traditional grid where the least used assets will still be needed every year or so, a hypothetical storage system will be sized to manage weather patterns that perhaps only occur every 25 years. This represents a lot of money being spent on something that runs for a few months every decade or so.
This, of course, may be part of the price we need to pay on our journey to zero carbon but one suspects the political classes will ask for a second opinion or two before they announce a string of pumped hydro systems up and down the length of the great dividing range.
