A Plan For Nuclear Power In Australia
There’s a persistent theme, particularly since the Texas power system dramas during the big cold, that nuclear power is some sort of magic bullet for power grids, and avoids the complexities of a majority solar and wind grid. I am frequently told that if we used nuclear in Australia we could avoid all these costly and time consuming transmission projects, greatly reduce the storage required and generally prosper. Problem is no one seems to have punched the numbers. Any of them. Not how much it will cost, not how many plants we’ll need and no estimate on whether or not we can actually do it without building additional transmission. This post will have a crack at all of those things.
Australia is just another power system, so I’ll size the system as if I was designing an industrial power system. Find the load profile, apply technology to meet that profile. See how many of each thing is required, add up the cost.
I’ve thought about this before, and the 2020 Forecast was a first step in this process. To size generators for a load, we need to know what the load is. To determine the load we need to know how much of the other technologies will be installed. This is where I got stuck after the forecast; I reckon by 2030 there will be no baseload market left in Australia. Zero. The residual demand on the NEM, that is grid demand less solar and wind production, will regularly go to zero. That’s a terrible use case for nuclear, or any baseload generator for that matter. So I think it’s a non-starter from the outset. But this will be a terrible post if I just say “nah not possible”, so I’ll need some assumptions. Let’s start by assuming that somehow no new renewables are added before 2030 when a nuke could feasibly start generating, and that the NEM load profile is unchanged. Essentially we’ll assume the nukes can be built overnight, some time next week.
There are a few ways to think about this. Nuclear is a baseload generator, so let’s start by meeting the entire energy demand for a year with nuclear. How much clean electricity do we need each year?
Here’s a summary of the last year in the NEM. This is electricity used per day, not the load profile necessarily. I’ll figure out how many nukes are required to meet various aspects of this load profile, replacing all the electricity delivered that is not currently wind, solar, biomass or hydro.
Add all that up and we need 145,663GWh of clean power in a year.
For this post I’ll assume that a nuke is a 1GW reactor, operating at 95% capacity factor, unless stated otherwise. So this theoretical plant can make 8,322GWh/year. If we assume that the demand the nukes are meeting is perfectly flat we can make that much electrical energy in a year with 17.5, 1GW nukes. Probably means we need 18 plants unless we can build half of one of these.
Clearly the demand won’t be flat. We can see that daily demand varies a fair bit. So how many nukes would we need to make enough energy to meet the demand of he highest day of the last year?
The peak day was almost 490GWh. My theoretical nuke can make 22.8GWh/day, so to meet the daily load that number climbs to 21.5 nukes to meet the daily load. Interesting that it’s a winter peak, not summer.
Going deeper, we all know that demand must equal supply exactly at all times. The absolute peak demand on the NEM goes closer to 35GW, but that includes some hydro, wind and solar generation. How much should we count on those? How much headroom should be in the system? Zero and zero, so I’m assuming no renewables contribution and no headroom. We can simply build 35 nuclear power plants to meet our peak demand.
How many plants do we need then? At least 18, and as many as 35 to meet NEM demand.
Transmission
Now, what about transmission? This is the big trump card from the nuclear supporters, claiming that since they are like the coal plants, steam plants in big lumps of capacity, we can simply put the nukes where the coal plants are/were and use existing transmission capacity. In comparison to an all renewables system, where we require significant investment in transmission lines to shift power around the country.
Again, I have problems with this assumption. To use the existing transmission capacity the plants must be built close to the existing transmission assets, in particular the big switching stations where the plants connect to the cross-state lines. To replace coal plants with nukes, we need to build nukes in parallel with the coal plants running, and connect to the same transmission assets. Because if we demolish the coal plants first, we have ten years to build the nuke when we don’t have enough electricity. This isn’t impossible, but gee it adds some complexity, and therefore cost. What if the owner of the coal plant doesn’t want to sell their connection assets? Is the land suitable for a nuke to be built on or is it a massive hole in the ground half full of water? Is the location of the existing coal assets an efficient place to build nukes or is it close to coal, when it should be close to ports and rail transport hubs or cement plants?
Take the now half-demolished Hazelwood power station as an example. Here’s an overhead image of the mine, with the power station and switching yard in the blue and orange boxes bottom right. It would be tricky to build a new nuke within this area.
The switchyard is shown below again in blue and the plant in orange. The switchyard is about 500m long.
Let’s pretend this difficulty doesn’t exist and we can build nukes on top of existing, operating coal plants and utilise their connection assets. Do we have enough existing connection assets to utilise? A quick summary of the challenge:
Luckily, we have a touch over 23GW of coal capacity operating now, suggesting there’s at least 23GW of existing transmission capacity to the locations where the biggest plants are:
This is okay if we are happy to just not deliver the last 12GW of demand on peak days, which of course we are not happy with. So it’s fair to say that if we build the minimum number of nukes to meet demand we probably don’t need additional transmission for the nukes, but if we have enough nuclear capacity to meet peak demand we need to add about 50% to existing transmission capacity to some big central generating hubs.
Storage
In the three scenarios; Annual, Daily and Peak, we can assume no storage for the Peak scenario. I doubt this would work technically, but this would look something like enough nukes running to cover our peak demand, and ramp them all down for the rest of the time. So no storage for the Peak scenario.
The Daily scenario is the most plausible for me, every day we can be pretty confident that the nukes will make enough electricity to meet our needs. But that’s averaged over the day, we’re 13.5GW short of meeting the annual peak demand; the Daily scenario has 21.5 nukes, peak demand is 35GW, we need 13.5GW of Something Else. A 13.5GW, 4-hour lithium ion battery would do a decent job of that, it’s basically a daily-cycling challenge so I don’t think the longer duration technologies are going to be best. 13gigs of new hydro would require 7 lots of Snowy 2, which doesn’t seem feasible with Australia’s geography. Of course the true system would be some combination of all of these, but I’ll assume the battery to make a point.
If we assume the Annual scenario, it’s a bit more complicated. Sure we save a few bucks by not building 4 nukes, but now we’re shifting electricity months into the future, not hours. In the graph below I show the GWh required each day in the blue columns, and a line for how much each scenario would generate each day. For the Annual scenario, we need to catch all those days when the blue columns are below the orange line, and shift them to days above the orange line. Inconveniently, it looks like we need to capture spring excesses and hold onto them until the peaks in winter. Some of that electricity needs to be stored for 6 months or more.
The power problem is worse with the Annual scenario too. We need 17.5GW of additional clean power, some of which will be powered by electricity stored for 6 months or more. That’s a bad use case for lithium, economically, that long duration suggests some sort of hydro will be the answer or perhaps some ammonia/hydrogen type storage. The optimum mix is again tricky to determine, but something like 2 Snowy Hydros in other pumped hydro projects across the NEM for 4GW of hydro, and the rest, 13.5GW in lithium, as in the Daily scenario.
Costs
The cost to build the first, and subsequent nuclear power plants in Australia is a gigantic unknown, and a pretty hotly contested topic. There have been a number of sincere attempts.
The UMPNER report, commissioned under Howard and delivered by Ziggy Switkowski was the first modern attempt, back in 2006. Despite many framing it as a stacked deck, Ziggy is a famous nuclear scientist and Howard didn’t care about climate change, I thought it was a good report that detailed the scale and complexity of not just building a few nukes and having a power sector, but actually initiating all the systems and processes required to run an entire nuclear sector.
They were fairly optimistic about nuclear’s prospects back in 2006, assuming a carbon price of $15-40/tonne would make it competitive with coal. They weren’t keen to name a capital cost for nuclear, preferring to look at LCOE forecasts for the output, but they do note CAPEX is likely to be 10-15% higher than the US to account for first of a kind costs in Australia.
More recently the South Australian government released the Royal Commission report into the nuclear fuel cycle, looking at mining, refining, power and waste storage. It’s a great report.
Chapter 4 of this report goes into the technical requirements of starting a nuclear power industry in Australia. One of the difficulties of nuclear in Australia is that there are large chunks of the industry that need to be created for the first plant, so that first plant looks very expensive. Things like port and fuel handling facilities, regulatory bodies, licencing in Australia etc. For this Plan I’ll assume that the cost of the first plant covers all of those things, and the first 1GW nuke will cost $8287/kW, or $8.3 billion dollars. 2014 dollars too mind.
They also have a stab at the cost of the first SMRs likely to become available, but have very low confidence in that estimate. It’s way higher than these so I’ll leave it out anyway. Further, they refer to a 2013 study that had much lower costs for nukes, possibly because they excluded some of the first-of-a-kind and other external costs.
From this study I’ll say that the second, third etc of a kind of these plants could be as cheap as $6,400/kW, as high as $9,300.
The most contemporary estimate is CSIRO’s famous GenCost report, with a summary of some recent cost estimates shown below. Sadly for the consistency of this Plan, CSIRO doesn’t see a future for the 1GW reactor that we plan to use, and has only included SMR costs. And they are really high. I really want this to decarbonise Australia, so I’m choosing the much lower estimates from the SA report as my modelled price.
I will however use the storage costs in GenCost to estimate the battery costs.
I want to model 4-hours of storage, so that requires 4kWh:1kW of power. So if I want 1GW of storage, I need 4GWh of batteries, at a cost of $450/kWh. So the 1GW standard battery for this Plan costs $1.8B
Add It All Up - Nuclear
Taking the costs from above and the three scenarios;
- Annual Smoothing uses the optimum number of nukes to make enough electricity in a year, plus two Snowy Hydros worth of pumped storage and batteries to meet the peak demand
- Daily Smoothing uses enough nukes to make enough electricity every day to meet the load on the highest use day, with enough batteries to meet the peak power demand
- Peak Demand is the frankly insane scenario of building enough nukes to cover the annual peak demand
And for a bit of fun, let’s make it an even bigger number and use the cost estimates from the most recent generator cost report, rather than the most favourable number I could find.
Yes they are big numbers, but we need to compare them to something. 5 NBNs for the Annual Scenario?
Add It Up - Windlab Study
Fortunately, more sophisticated modellers than me have had a look at the high proportion renewables scenario, so we can price that as a comparison. I really recommend this study from WindLab/ANU summarised by David Osmond on Twitter. Can’t find the original report any more sorry. Here’s the summary of generation.
And the storage required with these generators
Add all of these up and you get about $150b
You’ll note that this study requires 4% “Other”, which is basically very long duration storage with low utilisation. Could be ammonia storage or something like that. Chuck the cost of another Snowy 2 on top of that to cover it. $155b or so for this scenario. Plus transmission, which is not included in the nuclear scenario either.
Add It Up - Blunt Force
Lastly I’ll use the same blunt force approach I’ve used for nuclear above; figure out the energy and power requirements and multiply the numbers together. For this I’ll need to assume capacity factors for wind and solar, and based on this report I’ll use 0.3 and 0.2 respectively.
And we end up with a similar answer, about $174b. Less than $200b, excluding transmission.
Summarising
If Australia got excited about nuclear power and decided to go all in, it would cost at least $150b to meet our electrical energy and power needs, and using more recent estimates it could cost more like $280b - $480b.
I don’t see any reason to believe that using nuclear would reduce our transmission or storage requirements. Quite the opposite. One key thing this post shows me is that regardless of whether or not it’s technically possible to run a grid entirely on nukes, it will be heaps cheaper to include some storage, and probably quite a lot of it.
We could achieve the same outcome using renewables and similar amounts of storage, for a cost comparable to nuclear. And when I say “comparable” I mean:
- Using the cheapest possible nuclear price
- Without inflation
- Using no cost decreases in renewables over time
- And with recent estimates putting nuclear 2-3 times the price
At best then, I’m prepared to say that if things went very well for nuclear it could be delivered at a cost that is worth considering alongside the cost of a renewables system.
However, this ignores the utter impracticality of nuclear as a solution to Australia’s decarbonisation problems. It is currently illegal to build nuclear power in Australia, and advocates claim this is stopping development of any sort of plans for nuclear. We are not even at the starting gate. Before we can start the clock on the 10-15 year construction timeline for a nuke we need to change the law. The current government doesn’t even believe in climate change. If there is a miracle, we could have a nuke operating in Australia by 2035. How much will we need it by then? I think not at all, based on current deployment rates of renewables.
For nuclear to have any sort of chance we need to stop installing renewables. The grid we will have by the earliest that one could be built will not benefit from the expensive gigawatt of generation. It would have been good if Australia built nukes instead of coal in the 70s and 80s, but we didn’t, and now they look like a slow and expensive way to address an existential crisis. Australia has a an enviable opportunity to access renewables, and we can do it very quickly.
Comments
Post a Comment