AEMO: Australian Energy Market Operator
ESB: Energy Security Board
GSOO: Gas Statement of Opportunities
GW: gigawatt, 1,000 billion watts
ISP: Integrated System Plan
NEM: National Electricity Market
NREL: United States National Renewable Energy Laboratory
VRE: variable renewable electricity
Errata: In earlier posts in this series, I said that the Australian Energy Market Operator (AEMO) Integrated System Plan (ISP) contained 14% gas. Thanks to David Osmond (@DavidOsmond8) for pointing this out as wrong. The ISP does envisage more gas generators, up from 7 gigawatt (GW) to 10 GW, but less actual use. There is still a substantial amount of gas predicted to be used in the ISP, but I wrongly assumed it was for electricity. For details see the AEMO Gas Statement of Opportunities (GSOO). The gas used will be for other things like heat and industrial uses. So it will still generate greenhouse emissions, but just not for electricity. Figure 39 of that GSOO report predicts a drop from about 330 peta-joule (PJs) to 250 PJs. By 2032, Australia is predicted to be short on gas. By 2040, we will be short by about 50 PJs. Note, of course, that nuclear can supply heat but not all reactors run at the same temperature. If you need very high temperatures, then you need high-temperature reactors; such as those being built in China. The waste heat from most reactors, after electricity is generated, is still substantial and adequate for many uses; like desalination and district heating. Nuclear plants can make hydrogen and simultaneously supply heat to make low-carbon synthetic oil. Heat is the Swiss Army Knife of the energy world, and nuclear can provide it. “Electrify everything” is a simple and memorable slogan, but “Think before acting” is better.
Nuclear power and reducing total grid costs
This post will dig deeper into the mechanisms of using nuclear reactors to reduce system costs. Earlier posts mentioned the study by Jesse Jenkins and co-authors into the benefits of adding nuclear to a grid. This study threw technology constraints into an optimisation algorithm and the algorithm planned, in detail, the optimal hour-by-hour use of various types of technology. The inputs contained details on capacity payments; these flow from markets allowing you to get paid not for supplying electricity, but for guaranteeing to supply it at a particular time. So a nuclear plant can, for example, promise to be available between 6pm and midnight on some day and be paid for that promise, even if it is windy on the day in question and wind farms actually supply the power. The optimal operation of a nuclear plant in such situations isn’t obvious, but can be calculated by the algorithm. The existence of a capacity payments stream is a game changer that can reduce overall costs. More on this below.
Nuclear load following and grid costs
Any modern nuclear reactor can reduce or increase its output quite rapidly (up to its maximum) as required. It’s not just some theoretical possibility, they’ve been doing it for decades. Increasing the power is called “ramp up”, and decrease is “ramp down”.
You may have heard that nuclear is only a baseload technology with fixed output. Whoever told you this is either deliberately lying or simply didn’t bother to check. The fact that some nuclear plants operate this way is a function of the way markets work, not the way reactors work.
Typical ramp rates for a reactor are 2-5 percent per minute. For a 1000 MW reactor it’s 20 to 50 megawatts. This certainly won’t match the best performance of a modern gas turbine, but it can cope unaided with many ramp requirements.
The problem for grids with high variable renewable electricity (VRE) penetration is that wind and solar power can decline at the same time that demand for electricity ramps up; during the evening peak, for example. This creates a ramp rate far greater than anything grids have ever needed to cope with in the past. The optimal strategy may be to slowly curtail the wind and solar ahead of time to reduce the ramp rate that needs to be handled. But that’s just my intuition. You can only know the optimal strategy if you have a good model.
Fast frequency response
Ramps up or down can also be short and sharp, rather than long and steep.
In the South Australian grid (and elsewhere in the world), batteries are being used for what is called fast-frequency response, to handle short, sharp ramps. Skip the next paragraph if you understand this term.
When there’s a break for oranges at the football and there’s a simultaneous rush by people to boil kettles, then you need more electricity. As the kettles increase the load on the system, there is a drop in the frequency of the alternating current; to something below the target (in Australia) of 50 cycles per second (CPS).
This drop has to be rectified by adding more electricity generation. If it isn’t rectified, some very expensive stuff breaks. With coal, gas, or nuclear power plants, their heavy spinning turbines act as a flywheel to reduce the speed of the frequency drop. This buys time for extra electricity generation to be activated or existing generators to increase their output. If you don’t have those spinning turbines, then a big enough battery can do the job and buy that time. But such batteries are expensive and can’t buy very much time. So you still need sufficient dispatchable electricity to be available. Wind and solar aren’t dispatchable. It’s not like you can order the wind or sunshine to be increased. The problem is asymmetrical, it’s easier to dump excess electricity than to find extra, but it doesn’t happen by magic – you need to design for it.
Having nuclear in your grid can provide considerable capacity to deal with short and long-term ramps. If you are smart about it, nuclear can also provide operational reserves.
If you want a nuclear reactor to ramp much faster, meaning as fast as the best gas turbines, then that’s possible also, but you need to separate the turbine from the reactor with heat storage. So instead of the reactor feeding the turbine directly, you run the reactor at full power and store the heat output – it doesn’t matter how – there are many ways of storing heat. Then you connect the gas turbine to your heat storage and this will allow you to achieve very fast ramp times. Imagine the heat storage is a tank of salt, then the ramp rate is determined by the pump that moves the salt to the heat exchanger/s connected to the turbine/s. A heat exchanger is like a car radiator, the trick being to have a large surface area connecting the hot and cold stuff.
It’s much cheaper to store heat than to store electricity, because you don’t need highly processed metals and chemistry; crushed rocks will do it.
So, nuclear power provides mechanisms to deal with both capacity reserves and ramping, and that’s just the start of the advantages.
But you can only ramp up if you are running below your maximum output. Why would you do that unless you get paid for it?
Removing coal and gas plants has created problems dealing with both short and long ramps. The result has been the creation of markets where operators can bid to supply solutions. The operator might be a battery, or perhaps a gas turbine that will sit online ready to supply power and be paid for being available regardless of whether it is used. Or it might be a nuclear reactor running at half power.
In Australia’s National Electricity Market (NEM), there are no less than eight frequency control markets that have been created to try and compensate for increased wind and solar penetration. The capacity problems which have also emerged has led to the Energy Security Board (ESB) launching a project to create a capacity mechanism; most likely another market.
Ironically, a capacity mechanism provides an opportunity for nuclear plants to make the same (or more) money by supplying less electricity,while still reducing total grid cost.
How? Because these extra system services can be worth serious money. Australia’s first “big” battery made $A88 million in its first two and a half years of operation, not bad when you think it cost $A90 million to buy.
The Jenkins paper, mentioned at the start of this post, considered how nuclear power could use capacity and frequency markets. The result was clear. A reactor could simply run at less-than-full capacity and earn money providing reserve capacity, inertia, and frequency services. Having the capacity to ramp up as required helps decrease the amount of wind and solar electricity that has to be thrown out because it is generated at the wrong time.
The Sepulveda summarised it clearly:
“[The results showed that] VRE [wind and solar] and batteries are only weak-capacity substitutes for firm low-carbon resources…”
Batteries only appear in optimal solutions if their costs are at a level they aren’t projected to get even close to until about 2050 (using projections from United States National Renewable Energy Laboratory (NREL)). And in the real world, increasing battery production is hitting serious hurdles; both due to factory and, more critically, mining bottlenecks.
Reactors operating flexibly like this are, in a sense, a lot like storage, but cheaper and better. Batteries don’t last long and are always empty when you need them most (Murphy’s law of batteries).
These days, the collection of renewable deficiencies is typically called “Essential System Services”, and nuclear can do these cheaper than other technologies while providing a host of extra benefits.
The Sepulveda and Jenkins modelling shows that it is better to drop terms like “baseload” (when nuclear is involved) and view wind and solar as fuel-saving technologies which allow reactors to reduce output and provide operating reserves, inertia, and frequency services. As long as you pay them for running slow and ramping up and down, they will be happy.
Not only does Gencost (GENCOST) miss all of these possibilities by its superficial approach, it has misled many investors by presenting results that many have interpreted to mean that Australia can have cheap reliable electricity without nuclear power.
It's not clear at all that anybody can build a good reliable 100% renewable energy grid. The problems in the Denholm paper, with which this series began, from renewable energy experts, range from trivial to profound. The Jenkins and Sepulveda studies raise the obvious question; why bother? There are better, more eco-friendly and cheaper ways to get reliable electricity. And if nuclear reactor regulatory structures are reformed and both costs and build times are slashed? It shouldn't be a question of "if", but when this happens.
Part V will look in more detail at Gencost. Not only is it lacking in ambition, to put it mildly, it's biases are simply extraordinary.