"Baseload" Solar's Four Little Secrets

Reliability vs Curtailment

Electricity is a service. The power you require needs to be balanced with supply every second. My former colleague Chris Wright has likened electricity needs to demand for an Uber ride. Timing is everything.

There are two ways the Uber driver can fail to generate value: don’t ride at all or ride at the wrong time. Renewables, in a similar way, often do not serve the desired load at all or provide load at the wrong time.

The first problem is an issue when renewable systems are relatively small. An intermittent system that only works a 16-hour work week, like solar in Germany, simply does not work most of the time and requires backup from hardworking reliable sources.

But as solar has grown in relative capacity, the problem of generating power when it is not needed has become more prevalent. This is what we know as curtailment. And for so-called “baseload” solar, with the aim to provide a constant power continuously over the year, curtailment is a much bigger problem.

Why would curtailment be a “problem”? Utilities are already letting coal- and natural gas-powered generation sit when renewables show up for work. Isn’t this similar? Yes, these are similar issues and both are problems.

They are problems because these assets are only working part of the time they were supposed to work. That means their building cost has to be spread over fewer working power-hours. It means their mining and manufacturing carbon footprint is divided over fewer kWh’s. It means they require much more space due to overbuilding requirements. We will have a look at the impact of curtailment on all these parameters.

When utilities build a 1 Giga-Watt (GW) nuclear, coal or natural gas plant to generate electricity, this generally means that this power can be continuously provided for 90 - 95% of the time throughout a year, as these plants may be down for scheduled maintenance every now and then. For wind and solar, building a 1-GW plant does not mean that 1 GW of power is continuously provided. As you can see in Figure 1 for solar power generation in Texas and Germany, 1 GW is never achieved. In sunny Texas, at best there are a few hours during the year when 80% of that desired generation is achieved. Most of the time however, output is zero. On average, the output of a “1 GW” solar system in Texas is about 0.22 GW. In other words, the natural capacity factor for solar in Texas is 22%. The sun does not treat Germany as kindly, and the natural capacity factor there is less than 10%.

In addition to peak and average power being far lower than nameplate power, there are seasonal variations to the sun’s ability to generate power. In Texas, the natural capacity factor fluctuates between a monthly average of 7% and 30%; in Germany between 1.6% and 18%. To note here is the differential between seasonal minimum and maximum - about 4x in Texas and about 11x in Germany. This differential is the gap that requires bridging to size a reliable system.

Figure 1: Texas (left) and Germany (right) 2024 power output for a system with 1 GW solar capacity.

In a recent report, Ember claims that 6 GW of solar generation capacity with 17 GWh of battery storage can serve 97% of a continuous 1 GW load throughout the year in places like Las Vegas. They call this “baseload” solar. Yes, you read that correctly: to make up for the intermittency of sunshine, a solar system needs to be 6x overbuilt and have 17 hours of battery life to fight through the night and part of a cloudy day. And as you will see below, that system does not provide “continuous” power as desired.

We show in Figure 2 how a 6-GW/17-GWh system like that would fare in Texas, where the natural capacity factor for solar for 2024 was about 22%, and in Germany, where the natural capacity factor for solar was slightly lower than 10% that year. This is based on 2024 actual solar data from ERCOT and Germany’s Energy Charts.

In Texas, Ember’s preferred solar system reaches 91% load served while 32% of all generated electricity is curtailed. In Germany, its system only reaches 53% of load served while curtailment is minimal at 2%. In Texas, load served is near continuous throughout the summer. While this is also a better time in Germany, there are a few weeks during the summer where that is also achieved. But the flip side, especially in Germany, is that the provided load is nowhere near the desired 1 GW.

What these two examples show is that, if you build your weather-dependent system for summer, there will not be enough load served for winter; if you build for winter, there will be massive curtailment in summer. These seasonal changes will determine system’s ability to serve load and generate waste. And the extent of both are driven by the differential in seasonal electricity generation.

Figure 2: Texas (left) and Germany (right) 2024 power output for a desired 1 GW “continuous” solar power system using Ember’s suggested system with 6 GW of solar capacity and 17 GWh of battery storage capacity.

It is therefore fair to say that a solar system that is 6x overbuilt capacity and 17 hours of battery storage in Texas can provide “baseload” power. But that is in Texas, the sweet spot of solar. In places with less sun, not so much.

Back in Germany, however, addressing the lack of load served leads to challenges with curtailment. Figure 3 shows the load served vs the curtailment ratio, the ratio of curtailed electricity as a fraction of all electricity generated. Massive battery power is needed to overcome the impact of seasonal differences. Even a 200 GWh battery used in this graph, enough for about 8 days at 1 GW, is not nearly enough to get to 100% load served. That can only be achieved by overbuilding generating capacity - at a price of curtailing a lot of that power, even when massive batteries are in use.

The ideal system plots on the top left in this graph. That is an impossible combination for Germany. Systems without a battery all plot at the lower end of the rainbow-colored curves. As the sun shines mostly at the same time throughout Germany, the biggest gains in load served are in the smaller systems. Battery systems push the rainbow-colored curves to the left and the top, meaning batteries help to achieve more load served through nigh and dark spells while reducing curtailment.

What about achieving a modest goal of 90% load served, as is generally achieved by coal and natural gas power plants? As shown by the orange line below, this can be achieved with a system with 50 GW capacity and a battery of about 14 GWh. Yes, that means a 50x overbuild in solar capacity to achieve 1 GW continuously.

Figure 3: Germany load served vs curtailment ratio.

Does that sound reasonable? Wait until you see the impact on cost, power density and carbon footprint. This unused power in these oversized systems has consequences: it makes solar more expensive, it requires much more space, it reduces solar’s Energy Returned for Energy Invested (EROI) and it significantly increases solar power’s CO2 footprint.

Intermittency Impacts Economics

In a recent post (https://open.substack.com/pub/wirescrossed/p/solar-reliability-when-the-r-is-in?r=1r9eeo&utm_medium=ios), we have talked about the impact of increasing load served on Levelized Cost of Electricity (LCOE), mostly in an effort to define a LFSCOE - a Full System LCOE at various levels of “baseload”. Figure 4 shows an example of this for Germany.

To reference back to the previous two example - achieving 100% and 90% load served can be achieved in Germany for an LCOE of $1,400/MWh and ~$376/MWh, respectively. As an apples-to-apples reference, that compares to $110/MWh and $80/MWh for natural gas power in the United States for a similar load served.

Figure 4: German load served vs LCOE for various solar / battery systems

The ideal system is cheap and works all the time, and therefore plots near the top and as far as possible on the left. This graph makes clear you cannot talk about LCOE without incorporating dispatchability. For solar, there is a 4-15x difference between an unfirmed and a 90+% firmed solar/battery system.

Move Over, Biofuels

Waste from curtailment impacts more than cost. Overcapacity built to increase load served often just sits there and takes up space.

Several studies conclude that the average power density of solar panels in Texas is about 10 Watts per square meter (W/m2). This average incorporates the low natural capacity factor of solar, even in Texas, as it is averaged over a 24/365 period. The sun is never in the perfect spot. Simply by the nature of the lower natural capacity factor in Germany, this means it starts off around 4.4 W/m2 for a system where all power generated is consumed. This is shown at a single point in Figure 5 at a 9% load served and a solar power density of 4.4 W/m2.

When incorporating curtailment, the surface area needed for the consumed solar-generated electricity increases. If the curtailment ratio is say 50%, this implies that half the generated power was not generated at the right time, and that the surface area for the solar capacity was double as compared to a system where the curtailment was 0%. In other words, for this example, the power density of solar was also half of what it was for a system without wasted electricity.

Solar power is already very diffuse. Its 10 W/m2 power density is about 50x lower than coal-generated power, 70x lower than natural gas-generated electricity, and 100x lower than nuclear power. For baseload power in dark germany, power densities may drop to fractions of 1 W/m2, a few thousand times lower than natural gas power. On the inverse of power density is land use, which therefore requires thousands of times more land for the same amount of power generated as natural gas. This is similar to the average power density and land use of crops and biofuels.

It is often asserted that solar is so much more power dense than agriculture or forestry, where crops power density ranks in the range of 0.1 - 1.0 W/m2. But unlike solar, the fuel from these crops can be dispatched 24/365. Baseload solar approaches some of the extremely diluted power density of biofuels.

Figure 5: Power density for various “baseload” solar systems in Germany

Carbon Footprint

The main reason solar and wind were developed, and get subsidies, is that they have a smaller carbon footprint than natural gas and coal-generated electricity. Their supposedly low CO2 emissions have put them in the “green” category (never mind that more CO2 - not less - greens the planet, but that is another story). Also, some people categorize them as so-called “renewables”, while they build one-off products from fossil fuels with a limited life expectancy (but that is also another story).

When a lot of the power you generate is curtailed, and you have overbuilt to cater to serve a load, it turns out the carbon footprint in grams of CO2-equivalent (gCO2e) per kilo-Watt-hour (kWh) increases, as more grams of CO2 are emitted from mining and building solar panels, while disproportionally fewer kWh’s were produced.

We always hear that solar has a lower carbon footprint than electricity generation from natural gas or coal. The IEA currently puts it at about 41 gCO2e/kWh, vs about 500 to 1,000 gCO2e/kWh for natural gas and coal-generated power. But that is for unfirmed solar - just the panels, without batteries.

Figure 6 shows what happens when we overbuild to achieve a specific baseload with more solar capacity and battery storage. Ember’s suggested solar baseload system to continuously serve 1 GW with 17 GWh battery and 6 GW generational power has a carbon footprint of 405 gCO2e/kWh. That is about 10 times the advertised value we hear about all the time. That system ONLY gets you to 53% load served in Germany.

Figure 6: Carbon footprint for Various baseload solar systems in Germany

If you want to play in the adult-only area of power generation, where 90% or more of load is served, Germany’s solar carbon footprint is about 510 gCO2e/kWh - close to that of natural gas. But that is at a substantially higher price than natural gas, and a massively larger area needed than natural gas power generation.

Figure 7 summarizes these findings, and also includes what happens with baseload solar in sunnier Texas.

In summary, my assessment of baseload solar is that it is leads to overbuilding and waste through curtailment, that it is expen$ive, that it is definitely not “green”, and that it is going to occupy massive swatches of the Earth because it is so diluted.

Interestingly, when organizations brag about the possibility of achieving “baseload” solar, they forget the original promise that got solar in the door - its apparent low carbon footprint. The assessment that solar is “low cost” while insisting on a comparison between 10% unfirmed power and 90% firmed power has received at least some attention lately.

True cost, carbon footprint and area required need to be taken more seriously as our societies decide to further expand use of weather-dependent power sources.

Figure 7: Curtailment, LCOE, carbon footprint and power density for “baseload” solar in Germany and Texas

Conclusions

• While overbuilding capacity and battery backup can address Intermittency of solar power, it causes, sometimes to a large extend, curtailment of much of the extra electricity generated.

• Attempting to make unreliable energy sources more reliable is expensive. The Full System LFSCOE of baseload solar is about 5 - 15x higher than an unfirmed solar system.

• Curtailment leads to a much higher carbon footprint for baseload solar, in cases exceeding the carbon footprint of natural gas power plants. In the definition of “green” as in low CO2 emission, baseload solar does not qualify.

• Solar power is already a diluted power source, and baseload power dilutes it even further.