Variability is an inescapable factor of energy usage. It doesn’t matter if your grid supplies 1, 10, 1000, or 10,000,000 people. Just today Alberta’s energy usage fluctuated by almost an entire GW over the course of the day. And this is a fairly nice spring day, not too much heat required and almost no AC unless you’re some kind of cold loving crazy person. So we can only imagine that this would be a fairly low variability day in the grand scheme.
How our energy grid will manage this variability is a very important question because it does need to be managed. Otherwise rapid shifts from brownouts cause by insufficient supply to blackouts caused by infrastructure damaging oversupply can be all too common. Luckily there are several ways to do this from the old reliable pumped hydro storage, the more cutting edge pressurized gas storage, and the current belle of the green revolution ball, grid scale batteries.
Now I might arouse some ire in the pro-nuclear community for this, but I love batteries. I think almost everything about them is extremely interesting from their chemistry to their design and all the little finicky bits that allow them to work the way they do. I have dozens of articles saved about them and the research that is happening at the very bleeding edge of their field and I remain giddily optimistic about the chances of batteries being able to power our cars, phones, assorted gadgets of life for weeks on end without worry.
What I do not think they will ever be capable of doing is powering our entire planet.
Let me demonstrate. Assume that the energy demand of Alberta over a nice spring day looks something like this:
Now say we had enough batteries that could allow us to get away with the smallest base-load capacity amount and that was our entire energy sector. Nothing but base-load and batteries. It would probably look something like this:
Base-load of exactly half of the split between max and min, so that when demand is below base-load it charges the batteries that are then discharged when demand is above base. The problem is is that there is no room for growth in this system. It is balanced on a razors edge and any increase in demand would require building more power plants, which means that new industry or commercial ventures are stymied until there is enough cause to build more. There is no give and take in this system. To maintain a growing economy we need a little slack to provide for new things to happen.
Second problem is one of scale. Sure we’ve minimized the base-load generation that is required, but we need almost 3 GWh of batteries to do it. And that is if we want to be fully charging and discharging them every single day, and that’s no bueno for battery life times. Now of course you could follow the 80-20 rule for max charge and discharge to extend the lifespan of the batteries but then you would need about 4.2 GWh to do the same thing.
So what would the dreaded zero-battery outcome look like?
This is actually a pretty good system, sure it’s a bit wasteful but it provides lots of room for growth of industry without needing more generation to be built. However this assumes that whatever is providing the base-load cannot load follow on the scale of a days variability, but even for current nuclear that is easy. And that looks like this!
So if load following is possible, Why am I still a fan of batteries? Because of when this happens:
It’s a tiny little thing in terms of the entire grid. Just a little 1% stumble because of something like a power line going down or a breaker tripping in the grid somewhere. But the energy that was going there needs to go somewhere, and if it doesn’t have someplace like a battery or other storage to go to it could overload another section and cause a bigger blackout somewhere else. But if there is a battery then that excess gets put away all nice and tidy and it is there when the initial problem is fixed and the pulse of demand goes up as the small section reconnects. The battery gives the energy back to the grid and the base-load power barely needs to be adjusted, if at all. This is called load shedding and required usually only short term management of power to maintain grid stability.
Smoothing out input variability is what most people think of when they think of batteries being used in an electrical system though. This is just as bad a system as the first example of the absolute minimum base-load power. Most people have probably seen some kind of graph for say solar generation that looks like this:
Sure you need like 5 GWh of batteries to cover that massive mismatch in generation/demand peaks for this kind of system, but that’s not that much worse than the minimum base-load scenario. Plus this way we are 100% renewable energy right? Well sadly I played a small trick on you with this that I’ve seen be played a few times out in the world. I hid the two vertical axis on the graph. In all the previous graphs I’ve only had to use one. And even those graphs have been shifted to make the variation per day more obvious. To showcase the difference, this is what the full range of data for the load following base-load scenario looks like compared to what I showed previous:
So that means that when I put the solar generation on a single vertical scale, it actually looks like this:
There is obviously no way 10.8GW of solar nameplate capacity can power all of Alberta. Even if the total energy demand of Alberta is only about 10.5 GW during periods of high load, the time where generation is above demand is vanishingly small compared to where demand is above generation.
Now Wind can generate at any time but there is no repeatable pattern to wind generation like there is for solar. I can’t just plot out a Gaussian distribution and compare it to a sine wave. However it is just a irregular as solar as shown here:
But if we want to really push for Solar in Alberta, here’s how much we would need to produce enough excess during a sunny day to power the rest of the day:
If we want to guesstimate how much that will cost, we can take those new proposed solar projects ( the Travers Solar Project and Vulcan Solar Project) in southeast Alberta as a yardstick. They run approximately $1.25 and $2.00 per Watt of capacity respectively. Meaning for the 38 GW that we need, it would cost between $47.5-76 Billion dollars just for the solar panels.
Then we also need to take into account the cost of the needed 10 GW, 130 GWh battery, which if we base things off of the South Australia Tesla battery ($66M USD for 100MW, 129 MWh) comes out to another $89.1 billion. And of course lets not forget that we need to take less than flawlessly sunny days into account.
And this is just one day. It doesn’t take into account seasonal variability, and any Albertan can tell you it is way darker during the winter here. While adding wind to the mix would probably produce some large savings from this current $136.6 – $165.1 Billion dollar price tag, but would it save more than the minimum $86 Billion dollars saving of going nuclear at $5000/kWh?