Expensive Nuclear is No Where Good Enough

I'm always looking for ways to repeat the mantra: expensive nuclear is nowhere good enough. Recently, I'm been playing with some 3D bar charts to try and make this point. Did not come out nearly as well as I had hoped; but, for what it is worth, here are some graphs based on the GKG study of the options facing Germany.

To review, the GKG model comes up with that combination of power source and storage capacities which meets a given hourly demand profile at minimum social cost, where social cost is the sum of the market grid cost plus the cost of the CO2 emissions at a user supplied CO2 price. By varying the CO2 price, we can expose the efficient options available to society.

The GKG has exercised this model on Germany. Since there is so much difference between nuclear's should-cost and nuclear's did-cost in large parts of the planet, we varied both the unit social cost of CO2 and nuclear's overnight cost over wide ranges. The hourly demand profile used was actual German demand, for the 8 year period, 1993 to 2000. Wind/solar hourly capacity factors for this period are from the ENTSO database.

Figures 1 and 2 summarize the results. In these runs, the costing assumptions were based on somewhat dated 2020 numbers. These are greenfield runs. We pretend we are starting out with a blank slate.

Figure 1. Unit Grid Cost, Germany, 1993-2000 (2020 USD)

Figure 2. Carbon Intensity, Germany, 1993-2000.

If the social cost of atmospheric CO2 is zero, (the far left column), Germany can have electricity at about $50/MWh wholesale, regardless of the cost of nuclear. But the resulting carbon intensity jumps from 59 gCO2/kWh if nuclear's overnight cost is $2000/kW to 717 gCO2/kWh if nuclear's overnight cost is $4000/kW or more. Figure 3 shows the capacities that the model installed for the zero CO2 social cost column. The total height of each bar is the installed nameplate capacity. The opaque portion of the bar is actual average power actually marketed. The ratio of the opaque portion to the total height is the capacity factor for the 8 year period.

Figure 3. Installed Capacities, Social Cost of CO2 = $0/ton}

Figure 3 is hard to read and frankly not very interesting. At a nuclear cost of $2000/kW, nuclear supplies almost all the power, coal none, with gas doing some peaking. This combo resulted in the 59 gCO2/kWh, 7% of which was produced by nuclear. At a nuclear cost of $4000/kW and higher, coal supplies almost all the power, nuclear none, with gas doing some peaking. The model invests in no wind or solar at all at zero CO2 cost. The program finds a niche for a tiny bit of batteries, but makes no hydrogen.

Notice the vertical scale. The average demand over the study period was 62.5 GW. The peak hourly demand was 101.2 GW. The program installed about 60 GW of nuclear or coal, which it employed at a capacity factor in the low 80's, about 20 GW of Combined Cycle Gas Turbine, which it employed at a capacity factor in the low 30's, and about 30 GW of Open Cycle Gas Turbine, which it employed at a capacity factor of less than 2%.

Now let's go to the other extreme, Figure 4, and assume the social cost of CO2 is $1600 per ton, the far right column in Figures 1 and 2. This is far higher than anything I've seen in the literature. Actual carbon prices have rarely gone above $100/ton CO2.

Figure 4. Installed Capacities, Social Cost of CO2 = $1600/ton. Vertical scale totally different from Figure 3.

If nuclear's overnight cost is $2000/kW, the model's solution is not all that different from the $2000/kW solution in Figure 3.. However, the model does install more nuclear (84 GW rather than 58), which it uses at a lower capacity factor of 0.67, and less CCGT (5 GW rather than 21) and OCGT( 17 GW rather than 30), which it uses at a tiny capacity factor, less than 1%. But overall the grid cost only goes from $51/MWh to $57/MWh while the carbon intensity drops from 59 g/kWh to 6 g/kWh. Almost all the CO2 produced is from nuclear. Outlawing gas peaking in this situation would increase CO2 emissions.

The $4000/kW installed capacities are not that different from the $2000/kW; but the grid cost has shot up to $82/MWh. Germany may be able to afford this. Most of the planet cannot. Remember everything is in 2020 dollars.

At $8000/kW, the model installs 15 GW of onshore wind, but most of the power is still coming from nuclear at a prohibitive $127/MWh. At a social cost of $1600 per ton CO2, the program is prepared to pay a very high price to keep the CO2 below 10 g/kWh.

At $16000/kW, the program installs a massive amount of wind and solar; but about half the power is still coming from nuclear. The grid cost is a civilization sapping $185/MWh and the CO2 intensity is up to 19 g/kWh. This is the kind of success that the Vogtle defenders would have us celebrate.

At $32000/kW, the model finally gives up on nuclear, installs 394 GW wind/solar (six times average demand), and depends on 32 GW of gas and some hydrogen to bridge the dunkelflauten. The grid cost drops slightly to $168/MWh and CO2 emissions jump up to 37 g/kWh.

There are two ways of looking at all this:

1) Even if nuclear's cost is 4 or 5 times its should-cost, we can still sell a few plants to the wealthy, if we can convince them that the social cost of CO2 is extremely high. A few very expensive plants will support us nicely; and we don't have to worry much about competition.

2) Unless we get nuclear's did-cost down close to nuclear's should-cost, we will solve neither energy poverty nor global warming.

Expensive nuclear is nowhere good enough.