Since the first wave of nuclear reactors in 1970 to the on-going construction of Generation III+ reactors in Finland and France, nuclear power seems to be doomed to a cost escalation curse. If the curse is not stopped, nuclear power competitiveness will be compromised. On the one hand, construction expenses represent about 60% on the total cost of generation of this technology and on the other hand alternative sources of energy have experienced important decreases in their fixed costs. If the trends go on, nuclear power will become more expensive while competing technologies will become cheaper.
The US cost escalation has been widely studied and documented. The overnight costs of the first nuclear power plant in USD2010/MW were 7 times less than the last built (See Figure 1). Econometric studies found out that the expected cost reductions in building larger reactors were not achieved. Moreover, the learning effects were limited to a reduced number of utilities. Finally, all studies recognized that the stricter safety standards that the Nuclear Regulatory Commission set after the Three Mile Island accident were a key driver in the cost escalation.
Figure 1. Overnight Construction Cost for the U.S nuclear power fleet
Regarding the French nuclear fleet, the first cost assessment was done by Grubler (2011). He found that the units installed in 1974 were 3.5 times less costly than the post 1990 installed reactors (See Figure 2). In light of this result he concluded that learning effects were absent even when, unlike in the U.S, nuclear power in France enjoyed a favorable setting, i.e. centralized decision-making, high degree of standardization and regulatory stability. In other terms, even when the ideal conditions to lower costs are met, nuclear technology remains cost increasing.
It is important to mention that Grubler’s cost assessment used estimated rather than actual costs, because at his time these data were not publicly available. In a recent paper, we reexamine the cost escalation drivers for the French nuclear power fleet using the actual construction costs published by the Cour des Comptes in 2012.
Our findings are twofold.
Firstly we found that Grubler’s study overestimated the cost escalation. As we can see in the Figure 2, the cost estimates were accurate for the first reactors but overestimated for the latest ones. This gap introduces an important difference in the average annual growth rate: with Grubler’s estimated costs it is equal to 9% whereas with the actual costs the figure becomes 3.7% .
It is important to note that the average annual rate of the French Construction-cost index in this period was 4.8%
Figure 2. Grubler’s vs Cour des Comptes Construction Costs
for the French Nuclear Fleet
Secondly, we investigated the main drivers of the slow French cost escalation using a linear cost function. As explanatory variables we included: capacity to test scale effects, cumulative experience measured in number of completed reactors, experience within the same palier (i.e reactors of the same size) and same type of reactors. On top of that, we used two safety indicators to see if higher costs are somehow related with better safety performance. Finally, we also considered other control variables such lead-time and input price indexes.
Our analysis used principal component methodology to be able to overcome the severe collinearity among the main explanatory variables. The high correlation between capacity, cumulative experience and lead-time did not allow obtaining significant results under the linear regression framework. By using principal component analysis, we were able to identify the main drivers for the cost variation in France.
The main driver was the scale-up strategy. This means that increasing the size of the reactors ended-up in higher costs per installed MW. However, it is important to mention that nameplate capacity captures two variables, output and technological changes. Therefore it is not possible to reject scale effects, but if this were the case, it is clear that those gains were largely offset by the increased complexity embodied in bigger reactors.
We also found evidence of a learning curve within the same size and type of reactors. This result indicates that the French standardization strategy paid off. Constructing similar types of reactors made less severe the cost escalation. Finally, we found a positive relation between the safety indicators and costs. This last result means that although reactors have become more expensive, they also enjoy better performance in terms of safety.
These results allow us to derive some recommendations to overcome the nuclear cost escalation curse. The first one is that standardization is a good direction to look. As mentioned before, the experience gained in constructing the same type of reactors was what allowed to ease the cost escalation in France.
The second recommendation is to rethink the scale-up strategy. As in the U.S, the increase in reactors’ sizes resulted in more complex units that required longer lead-times and were subject to stricter regulatory scrutiny, which in turn meant more expensive reactors. This is reaffirmed with the costs announcements of the Generation III+ reactors, for instance, the EPR in Flamanville has an expected cost of € 5.100 per kW, that is much higher than the latest N4 reactors that had a cost equal to €1.450/kW.
One possible solution is to consider small modular reactors. Given that these reactors are smaller, they have shorter construction schedule, lower market risk resulting in a lower cost of capital. In addition, some cost savings can be achieved through off-site modules fabrication, as well as the learning by doing after the production of multiple modules.
Lina Escobar Rangel and François Lévêque, Mines-ParisTech