A tweet caught my eye recently about building a baseload electricity source with solar and batteries. By “baseload”, I just mean a source that can supply the same electrical power 24 hours a day, 365 days a year.
The tweet was based on a LinkedIn article based on another such piece.
Here’s the Press Release from the company concerned, Masdar, which probably fuelled the fire. Here’s the crux of the project.
“Delivering up to 1 gigawatt (GW) of baseload power every day generated from renewable energy, it will be the largest combined solar and battery energy storage system (battery energy storage systems (BESS)) in the world.”
“Located in Abu Dhabi, the project will feature a 5.2GW (DC) solar photovoltaic (photovoltaic (PV)) plant, coupled with a 19 gigawatt-hour (gigawatt-hour (GWh)) BESS, setting a global benchmark in clean energy innovation.”
You should be suspicious when people give a single number for a battery; there should be two. The maximum power and the energy capacity, which tells you how long the battery can supply that power for. So you can talk equivalently about a 1GW/2GWh battery as a 1GW/2-hour battery.
In this case, the 19GWh is the total energy storage capacity, but what is the maximum power output? It makes no sense for the maximum power to be more than 1GW, because that’s the baseload target; you don’t need more. So it’s a 19 hour battery with a maximum power of 1GW.
The whole thing reminded me of an anecdote from decades ago when I did a networking course with an expert who probably should remain anonymous. He told a story about running a training course in one of the Gulf States. After spending a couple of hours explaining some computer networking techniques that could vastly improve performance and save money, one of the attendees responded: “Saving money? Really? Where is the honor in that?”
If you think doing baseload with solar and batteries is cool, or honourable and you value that kind of honor, then go for it; praise the sheiks and, be very impressed.
But please don’t try this at home! I’ll explain why in due course.
The mechanics
Let’s just explain how it works.
The annual capacity factor of a solar farm is the average output over the year divided by what it says on the box; the maximum output. You can also talk about a monthly or daily capacity factor; the concept is the same. The capacity factor of a dark cloudy month will be lower than the annual figure, while that of your hottest brightest month will be above it. Easy.
Suppose you had a solar farm with a 25% annual capacity factor. If the farm had 4 GW of panels, then the capacity factor tells you that the average output is 1 GW.
Sometimes of course, you’d be producing close to the full 4 GW and sometimes nothing at all; the average is just that, an average. It isn’t the amount you can count on as being available. The idea of the 19 GWh of batteries is to store energy when you are getting more than 1 GW from the panels and release it when you are getting less. Thus smoothing the output. Sounds simple? The devil is in the detail.
It’s pretty sunny all year round in Abu Dhabi, the average capacity factor in the UAE for solar power was 26% in 2023. So why use 5.2 GW of panels? Surely 4 GW should be enough. Theoretically, if you have 25% capacity factor and enough batteries, you should be able to provide 1 GW continuously; regardless of the ebb and flow of dark and sunny days. Trust me for now, but if you tried to use just 4 GW of panels, even in Abu Dhabi, you’d need more batteries. I’ll present some data later. Needless to say, the Masdar engineers have worked this out and have very carefully chosen the sizing of 5.2 GW of panels and 19 GWh of batteries.
Suppose you copied the Abu Dhabi system in QLD. I’ve simulated this over 19 days of actual QLD 5-minute data on solar production from current QLD solar farms. The capacity factor over this period was 22%. Given the 26% UAE capacity factor, the simulation isn’t perfect but should be useful to understand the concepts. Here’s a chart showing the PV output before smoothing and (in cyan) and the actual flow to the grid given the batteries. I’m ignoring any economic details like price at various times of the day. The simulation has one goal, a reliable 1 gigawatt output; it tries never to produce more or less.
As you can see, once this system got over it’s initial starting problems, the smoothing process of the batteries is pretty bloody good. Just two small time periods where the output sank; to zero. Like I said, pretty bloody good; as long as you don’t expect grid level reliability.
The reliability of real grids, in rich countries, is much better. Australian Energy Market Operator (AEMO) expect 99.998% reliability. 97% would be a massive fail; heads would roll.
And what about the battery? Here’s a chart showing the state of the battery over time. At 19 GWh, it is perfectly sized. If you look very closely you can see a few little flat tops on some of the peaks. This is where some electricity had to be thrown out rather than stored.
Maximising waste
Think about it. In order to get 1 GW of baseload, this system required the construction of 24.2 gigawatts worth of stuff. And most of the time most of the stuff is doing sweet bugger all. It’s hard to imagine a more wasteful system.
I don’t have precise details, but we are talking about 9,500 hectares of land covered in perhaps 300,000 tonnes of panels and associated hardware. Plus perhaps 200,000 tonnes of batteries, made from a considerable tonnage of critical minerals. In a nuclear plant, by comparison, the bit that does all the work, the reactor pressure vessel, would be maybe 600-700 tonnes of steel. Plus a couple of thousand tonnes for the steam generators and generator turbines. The rest is pipes and concrete. Not all materials are created equal when it comes to production. Battery minerals are labelled as critical because their production is in mostly small quantities and fairly complex. The sheer amount of stuff you need to dig up for each tonne is also very different from what you need for steel and concrete.
A nuclear reactor will also be operating for decades after the solar farm has been dismantled and sent, hopefully, for recycling and turning into more stuff that will be doing sweet bugger all for a few more decades.
This is a remarkable achievement in the on-going renewable war against efficient energy production.
All of us buy stuff we don’t use very often. For example, I have a chain saw. I only use it once or twice a year. But it’s great to be able to just grab it and use it when needed rather than drive somewhere and hire one, use it and return it.
What makes this such an attractive proposition for me is that when I want it, my chainsaw will just work.
I don’t ever want a chainsaw with a 20% probability of working when I want it. Nor do I want to buy 5 chainsaws, each of which has a 20% chance of working when I want it; but which modelling tells me will yield at least one working chainsaw when I need it.
It isn’t just chainsaws. We don’t normally buy anything as unreliable as solar power; especially when we have a reliable alternative requiring less land, less mining; less of everything.
If your goal is maximising your honour by being the first on the planet to do something, regardless of how bloody silly and wasteful it is, then baseload solar is for you.
Don’t try this at home: Queensland
But what about building solar baseload in Australia? Could we perhaps beat Abu Dhabi and waste even more stuff for a 1 GW return?
Currently, QLD has about 3.8 GW of solar farms. What if we decided to run them baseload with the installation of a bunch of batteries?
How many batteries would we need? I’ve redone the simulation using the same data as above but without scaling up to the 5.2 GW used in Abu Dhabi. Instead of targetting 1 GW of output, I’ve just chosen to target the annual average. The maximum is 3.830 GW, the capacity factor is 22%, so the target will be 0.22 x 3830 megawatt (MW) = ~842 MW. We’ll call this number CFp … the average power output. I’ve assumed an infinite amount of battery storage is available because the whole point of this simulation is to estimate how much we need to buy.
Here’s the output chart. As you can see, the batteries work pretty well at smoothing the output.
How does the system work? Dead easy. When your panels are producing more than CFp you supply CFp to the grid and put the rest in storage. When you are short, you draw from the storage (if possible), but no more than you need to hit CFp. The cyan curve shows that the method works, but the initial conditions can cause some shortages.
Now comes the interesting bit. How much storage do we end up needing?
Here’s the state of the battery over time.
You can see that we max out at about 20,000 MWh of storage … (aka 20 GWh). That’s more than in Abu Dhabi, despite the smaller solar farm and lower average output.
Now you can see why they added more panels in Abu Dhabi; to reduce the battery requirements.
Summarising. We can get 842 MW of baseload power with 3.8 GW of panels and 20 GWh of storage. The Abu Dhabi plant was said to cost US$6.9 billion. At current (GenCost 2024) prices, our Queensland copy is about AU$14 billion.
Don’t try this at home: South Australia
But we aren’t quite done yet. I only used 19 days of data in the simulation above. In some parts of Australia, that 19 days would be reasonably representative of the whole year. But other parts of Australia (and the world) are not so blessed with unrelenting sunshine.
What happens if you try to build a baseload solar system in South Australia?
As it happens, I can simulate the kind of performance you’d get over a South Australian winter. Here’s data from 55 days in winter 2024. I’ve scaled it up simulate 3.8 GW of solar farms.
What’s happened to the cyan line? It’s much lower!
The capacity factor in SA during this dismal winter was just over 10 percent. The global average solar capacity factor for 2023 was 13%, so there are plenty of countries trashing large areas of land for this kind of dismal return. How does this drop in capacity factor impact the amount of storage you need? Here’s the chart.
The peak storage required more than doubled. Meaning that we now need about AU$21 billion worth of batteries.
Increasing your PV panels while sticking to the same target baseload figure will reduce the amount of batteries. That is what Masdar has done in Abu Dhabi. They have far more of everything than required; probably because they don’t want to fail. There isn’t much honour at being the first to do something incredibly stupid, but even less if you fail.
Comments