Why LCOE is not a good metric for renewables

The Levelized Cost of Electricity (LCOE) is a metric used to compare the costs of a power plant project to the electricity that it will generate. By summing up all the costs (CAPEX and OPEX) of the power plant project, discounting, and normalizing the costs one can obtain the golden value of what the electricity price must be for the project to be profitable over its lifetime.

Calculating such a metric requires some significant simplifications and assumptions. The simplicity of it has, however, made a very common argument for renewables, fitting into a single tweet.

Let’s have a look at different terms in the LCOE calculation. After dusting off some of my old textbooks from university I found the chapter on LCOE, which can be defined as

The first term relates to the CAPEX, which can be broken down into components such as project studies and design, development, procurement of land areas, equipment, taxes, and financing costs. Large construction projects such as building a power plant have a high risk of delays and unforeseen incidents during the construction phase. For projects going on for months and years the interest expenses could therefore add considerably to the CAPEX of the project.

The second term relates to the OPEX and is comprised of fixed and variable operations and maintenance costs, fuel costs, taxes, and emission costs. The fixed costs are normalized by the installed capacity and yearly electricity generation and the variable costs are defined with the same unit as the rest of the equation, here in Europe that is EUR/MWh.

Illustration of the Levelized Cost Of Electricity (LCOE).

Just by looking at what the different terms represent, one can feel the awe of the challenge of trying to predict all the values. The value of each term is indeed highly uncertain, so why do we often see the LCOE provided only by its expected value instead of a distribution? The distribution would still be suspect to the assumptions taken by the modeler, but it would at least provide some information to the reader on how sensitive the LCOE is. Just like IEA did in this report where they outline LCOE for different power plant technologies:

[embed]Projected Costs of Generating Electricity 2020 - Analysis - IEA This is an extract, full report available as PDF download Projected Costs of Generating Electricity - 2020 Edition is…www.iea.org

Since the construction of conventional thermal power plants are long-term projects that typically last for a couple of years, there are many things that can go wrong and the electricity prices in five years’ time are certainly extremely difficult to predict. Subsequently, lenders and investors require a risk premium that can significantly impact the interest cost accrued during construction. This is where renewables have a great proposition as construction of a wind power farm can be done in just a couple of months.

Without going any further into the details of the different terms of the LCOE I want to demonstrate what I believe is its main weakness, its inability to include additional costs that the power system would have to incur. Below is a figure showing historic electricity generation and installed capacity from wind and solar in the UK.

Wind and solar generation in the UK in June 2020.

With the intermittency of renewable energies, it should be evident that a megawatt-hour from renewable energy cannot be utilized as well as a megawatt-hour from a dispatchable power plant. So why are we comparing apples and oranges for LCOE?

A better solution would therefore be to see what the total costs for the entire power system are, and not just the individual plant. One could therefore define a “System LCOE” that can include these costs, where I break them down into balancing, grid security, and transmission costs.

These costs are related to how generation from intermittent energy sources coincides temporally and cannot, without curtailment, provide backup capacity for the grid. They also have a clear spatially correlation, as we can see from the plot above for the onshore and offshore wind generation. Due to wind and solars low marginal costs, dispatchable power plants can be pushed out of the market. The power system would, however, still need operating power plants that can be regulated in case of production and demand deviations, this is where the balancing costs come to play. The power system needs dispatchable energy sources.

The grid security costs come from the fact that wind and solar do also not provide a sufficient amount of inertia to the power grid. They don’t have huge rotating masses that can absorb and discharge energy when fluctuations between demand and production occur. There are methods to handle this by using batteries and capacitors to provide synthetic inertia, but this again comes at a cost for the power system that should be taken into account.

In the figure above on can see that it is rare for wind and solar to generate electricity near the installed capacity. However, when they do produce high amounts, the power system would have to consume this energy somehow. If the demand is not sufficient in that region one would either rely on the transmission grid or have some sort of energy storage in that region. Yet again, this is not free.

Illustration of the LCOE and System LCOE.

The calculation of the System LCOE is much more difficult than for the LCOE. How are you going to calculate the different system costs for each power plant? It is not a trivial task. The System LCOE would depend on the state of the power system. The more intermittent energy sources you already have on the grid, the more does it require for the power system to accommodate new sources.

Another issue here, and perhaps the largest one, is that it is the LCOE and not the System LCOE that power plant investors see. I would classify this as an inefficiency of the power market, as additional costs are not allocated to the ones who caused them. This inefficiency is further amplified as feed-in-tariffs and other schemes to provide additional revenue for renewables compared to other sources. For now, the additional system costs are consumed by Transmission System Operators (TSOs) that may further pass them on to consumers and producers via grid tariffs.

A more prominent concern for the investors is the cannibalization effect, which can lead to diminishing returns. A large share of renewables in a region can in hours with high generation result in low prices and thus less revenue for all the operating power plants. This can for example be seen in some periods in Northern Europe where there has been a lot of wind generation and power prices have subsequently become low, and even negative.

There are many different proposals of how the discrepancy of LCOE can be adjusted for. For instance, this great article from IEA discusses the growth of solar energy and shows how LCOE and “Value Adjusted LCOE (VALCOE)” compares between solar and coal power plants.

[embed]Is exponential growth of solar PV the obvious conclusion? - Analysis - IEA Solar PV has experienced exponential growth in recent years, with global installed capacity increasing ten-fold from…www.iea.org

In the article, we can see that in the years ahead, solar will be a clear winner over coal when it comes to LCOE. The reverse can be seen for VALCOE.

Are we doing the right thing when seemingly all our efforts go into wind and solar, and if so, are we aware of the additional costs we will occur going down that path? If only we have had an energy source that has similar operational characteristics as a coal power plant but the CO2 footprint of wind and solar..