Kutz reported in early 2020 that Steffen et al. tackled a challenging topic to quantify and project costs for operation and maintenance (O&M) of solar and wind power plants in Germany. They note that O&M costs can be 20%–25% of overall costs and find that such costs have decreased, resulting in common overestimates of O&M costs. Experience curves (plotting cost as a function of cumulative production on a log-log scale) have been valuable tools for projecting expected cost reduction. Experience curve projections have been especially useful for solar panels as their costs have decreased relatively faster.
Kutz commented that an analysis of costs is challenging, as you can easily imagine if you try to name “the” price for a loaf of bread. Both costs and prices depend on many things such as the quality of the product and the size of the purchase. Attempting to quantify the cost of O&M is even more challenging in my experience. The definition of O&M can vary from “fixing it when someone complains that it’s not working” to “having an engineer on site full time and constantly checking on things.”
Steffen et al. have overcome much of this challenge by first defining which tasks they include in O&M and then limiting the dataset to contracted O&M. Smaller systems (e.g., rooftop solar) may be maintained by the owner or by a building engineer that is providing O&M for all aspects of a building, making it difficult to identify the actual costs. By limiting the dataset to contracted O&M, Steffen et al.1 avoid many potential ambiguities in the data and obtain the data that are most relevant to someone who is thinking of purchasing a wind or solar plant and planning to contract out for the O&M. In this case, the “price” of an O&M contract translates directly into a “cost” to the plant owner.
The authors report variability in the data, which Kutz anticipated for different designs of plants or plants in different locations. Just as some cars end up in the shop more frequently than others, we expect that some power plants will have more problems than others. Additionally, the weather can have a huge effect. Although this would not be found in Germany, where Kutz lives now in California, it is common to have no rain for 6 months straight. During this time solar panels get quite dirty, but the grass doesn’t grow! The costs of mowing or other weed control and of washing can be large (this study suggests about one-third of the solar O&M cost) and highly variable.
In Germany, the weather is relatively consistent compared to the range of climate zones in California, but there is still substantial variation of wind and solar resources as well as differences in local temperatures. Also, as noted by the Steffen et al., the distance between plants can be very important. A common problem with solar plants is needing to restart an inverter, which is a fairly simple thing unless you need to drive for an hour to reach a remote site.
All in all, Kutz was impressed that they were able to reduce the scatter in the data and identify some clear trends. It was especially useful that they were able to separate different categories of O&M costs in a meaningful way, finding, unsurprisingly, that wind systems require more maintenance and repair, while solar systems require more site maintenance costs (like mowing or washing). Perhaps the most significant historical trend they identified was the large reduction in solar O&M costs.
Kutz wanted to see them compare the O&M costs of solar and wind directly, but Steffen et al. used the conventions of reporting wind O&M costs in Euros per kWh and the solar O&M costs in Euros per kW per year. Kutz compared these more directly by noting that the average solar plant in Germany generates ∼1,000 kWh per kW per year. Converting their numbers then to consistent units, the O&M costs for solar dropped from about 5 Euro cents per kWh in 2005 to about 0.7 Euro cents per kWh in 2017.
In the same time frame, they reported that wind O&M costs dropped from 1.75 to 1.32 Euro cents per kWh. The much larger decrease for the solar O&M costs enabled Steffen et al. to quantify the experience rate with much more confidence for solar than for wind. Both O&M costs and levelized cost of energy (LCOE) for solar have been decreasing much more quickly than for wind, reflecting both the steeper experience curve for solar as well as the greater maturity of the wind industry. From their analysis, solar O&M costs are now smaller per kWh than wind O&M costs by about a factor of 2, and their analysis suggests that this difference will continue to increase.
The authors’ analysis of the reasons for the reductions in the costs is quite interesting but may not be useful to readers who are not focused on events in Germany, as most of the unusual observations were directly linked to changes in Germany. Nevertheless, the themes they cite may be relevant everywhere. For example, they observed that market competition drove prices (and profit margins) down very quickly, which is an effect that has been seen in many parts of the world.
Kutz wholly supports the conclusions of Steffen et al. that there is further room for reduction in O&M costs. Not only will the increasing density of plants facilitate lower costs, but I see many opportunities for innovations and refinements. Plant owners are experimenting with strategies that reduce costs. Adjusting the tilt of a tracked solar system during a rain can increase the amount of dirt that is washed off during even a small rain, avoiding the large cost of washing a system. Some have found that sheep are an inexpensive means for keeping vegetation trimmed, while goats may be too destructive to be attractive.
Steffen et al.’s study is a useful starting point for a question that is of increasing importance to the solar and wind industries. The primary conclusions are not surprising: O&M costs are coming down for both technologies but more quickly for solar. However, the data that caught Kutz's attention were those shown in Figure 5 suggesting that O&M costs are substantially (up to a factor of 7) lower than have been assumed historically. Assumptions of outdated high costs may contribute to the consistently low projections of solar growth. As Steffen et al. noted, accurate modeling of how to solve the world’s energy problems requires accurate cost data. But are today’s models assuming such high prices?
The data in Figure 5 are mostly from the 2010–2012 time frame. A 2016 study that wasn’t included in Figure 53 suggests $7 per kW per year for solar O&M costs, which is much closer to the current reported results. So are modelers making realistic assumptions? I don’t know, but I hope they will take a look at Steffen et al.’s paper to help improve their assumptions and their projections. Though this dataset is limited to Germany, it provides an excellent starting point and a basis for expanded studies. Improved understanding of projected costs will enable the world to make wiser decisions as it seeks clean-energy solutions.
[Bias Alert: Fact Check underway and will report here on results .....
Some results: Two separate 2021 analyses from Origis Services and Wood Mackenzie suggest that cost-cutting now for maintaining solar projects leads to greater O&M expenses down the road (more)
kWh Analytics provides discussions and reports on solar risk assessment (more)
In a January, 2021 blog by Hanawalt, he discusses solar O&M predictions ....but look for bias here too (more)
In a new Statistics plot on "Fixed and variable costs for the operation and maintenance of new power plants in the United States in 2020*, by technology type" (here)]
Managed & Built by I2M Consulting, LLC | © 2010-
