Benchmarking Wind Power Operating Costs in the United States: Results from a Survey of Wind Industry Experts
Ryan Wiser, Mark Bolinger, Eric Lantz, January 2019 (Lawrence Berkeley National Laboratory)
This paper draws on a survey of wind industry professionals to clarify trends in the operational expenditures (OpEx) of U.S. land-based wind power plants. The paper also highlights key drivers of those trends. We find that average all-in lifetime OpEx has declined from approximately $80/kW-yr (~$35/MWh) for projects built in the late 1990s to a level that is approaching $40/kW-yr (~$11/MWh) for projects under construction in 2018. Turbine operations and maintenance (O&M) costs—inclusive of scheduled and unscheduled maintenance—represent the single largest component of overall OpEx and the primary source of cost reductions over the last decade. We observe wide ranges of OpEx over time; for example, survey respondents cite a range in average expected costs for projects commissioned between 2015 and 2018 from $33/kW-yr to $59/kW-yr. Notably, these broad ranges include high levels of variability in both turbine O&M costs and non-turbine OpEx. Potential technical and strategic drivers of this variability are highlighted. We also use historical OpEx learning rates, showing a 9% OpEx reduction for each doubling of global installed wind capacity, to project a further $5–$8/kW-yr (12%– 18%) OpEx reduction from 2018 to 2040. When compared with the broader literature, these findings suggest that continued OpEx reductions may contribute 10% or more of the expected reductions in land-based wind’s levelized cost of energy. Moreover, these estimates may understate the importance of OpEx owing to the multiplicative effects through which operational advancements influence not only O&M costs but also component reliability, performance, and plant-level availability—thereby affecting levelized costs though OpEx reduction and by enhancing annual energy production and plant lifetimes. Given the limited quantity and comparability of previously available OpEx data, the data and trends reported here may usefully inform OpEx assumptions used by electric system planners, analysts, modelers, and research and development managers. The results may also provide useful benchmarks to the wind industry, helping developers and asset owners compare their OpEx expectations with historical experience and other industry projections.
The levelized cost of energy (LCOE) of wind power plants is driven by five primary parameters: upfront capital expenditures (CapEx), operational expenditures (OpEx), project performance, financing and tax assumptions, and project life. Among these factors, long-term OpEx has been understudied. While a robust and growing literature on turbine and component reliability exists (e.g., Echavarria et al. 2008; Spinato et al. 2009; Keller et al. 2016; Sheng 2017; Artigao et al. 2018), data on OpEx trends are limited.
More specifically, extensive literature on land-based wind CapEx has tracked trends over time and across countries (IRENA 2018; Wiser and Bolinger 2018; IEA Wind 2018), established data-driven costreduction trajectories based on learning curves (Wiser et al. 2011; Lindman and Söderholm 2012; Rubin et al. 2015; Samadi 2018), and developed engineering models to understand past and possible future cost-reduction options (Sieros et al. 2012). A growing literature also emphasizes improvements in wind project performance, especially as turbine rotor diameters and hub heights have increased (IRENA 2018; Wiser and Bolinger 2018). Facilitating the development of these literatures has been the availability of substantial project-level data on land-based wind CapEx and performance (IRENA 2018; Wiser and Bolinger 2017; IEA Wind 2018).
Project-level data on land-based wind plant OpEx, on the other hand, are not widely available (IRENA 2018; Wiser and Bolinger 2018; BNEF 2015a) owing to the proprietary nature of the data and the fact that lifetime OpEx data are only available after the full life of plants, which can be 20 years or more. Few plants have been operating for 20 years, and those that have are using turbine technology of vastly different scale and sophistication compared with modern projects. As a result, OpEx for early plants may not be relevant for estimating OpEx for newer plants (IRENA 2018; Wiser and Bolinger 2018; Poore and Walford 2008). A lack of standardization in both intra- and inter-firm data collection and management (e.g., limited tracking of specific costs that result from specific maintenance issues) has further hindered the development of OpEx datasets and intelligence (DNV KEMA 2018).
Even when wind OpEx data are available, they can be hard to interpret. In some cases, data are reported as actual realized costs; in other cases, as long-term cost expectations. The number of years covered by the data, relative to expected wind project life, may vary. Costs are often reported in $/kW-yr terms, but also as $/MWh, $/turbine, or $/project. Costs may vary by project size, location, and other factors. Turbine operations and maintenance (O&M) is sometimes contracted out to the turbine manufacturer or an independent service provider with varying servicing terms and durations. In other cases, O&M is self-provided by the wind plant owner. Turbines are typically under manufacturer warranty during the first years of operations, so costs due to unscheduled maintenance may be embedded in turbine purchase agreements, thereby reducing annual O&M costs for the project owner. Finally, a wide and diverse set of costs can be embedded within the OpEx category: turbine O&M (scheduled and unscheduled), balance of plant (BOP) O&M, land costs, property or other local taxes or payments, grid and electrical use, insurance, asset management and administration, and others. Less mature turbines have sometimes required extensive and costly in-field retrofits (e.g., gearboxes) due to premature component failures, which may or may not be considered part of OpEx. Absent clarity on what costs are included, establishing clean comparisons across various sources of OpEx data is impossible.
The result is not only a wide array of OpEx estimates in the literature but, more importantly, a general lack of fidelity and confidence in those estimates. Lacking solid data, for example, the U.S. Department of Energy and the National Renewable Energy Laboratory have assumed no change in land-based wind OpEx in the United States since 2014 (Stehly et al. 2017; DOE 2015). During the years leading up to 2014, their OpEx estimates rose as they were adjusted to account for anecdotal data suggesting that actual costs were higher than originally forecast, in part due to premature component failure for certain turbines (Tegen et al. 2013; DOE 2008). The U.S. Energy Information Administration has similarly assumed an increasing cost of wind plant OpEx in successive versions of its Annual Energy Outlook (e.g., EIA 2011, 2015, 2018), reflecting uncertainty in and lack of solid historical data on OpEx as well as recognition that realized OpEx was coming in higher than previous expectations.
Understanding past and current land-based wind plant OpEx is important for several reasons. First, OpEx represents a sizable and potentially growing share of LCOE, especially as wind’s LCOE declines owing to lower upfront costs and better performance. Ten years ago, analysts often attributed up to 20%–25% of land-based wind LCOE to OpEx (Blanco 2009; EWEA 2009; Walford 2006), associating approximately half of OpEx directly with turbine O&M (Blanco 2009; DNV KEMA 2018). Recent data suggest that OpEx accounts for 25% to more than 35% of overall LCOE (IEA Wind 2018; Stehly et al. 2017).
Second, operational practices and OpEx have important connections to other parameters that influence wind’s LCOE. Specifically, turbine O&M practices directly influence turbine component reliability and related downtime, turbine performance, and overall wind plant availability (Echavarria et al. 2008; Spinato et al. 2009; Keller et al. 2016; GL Garrad Hassan 2018; DNV KEMA 2018; Artigao et al. 2018; van Kuik et al. 2016), thereby affecting annual production and project lifetime. CapEx and OpEx are also related, because higher-cost, more reliable turbines may yield lower long-term OpEx, and vice versa.
Third, OpEx represents an important lever for wind plant LCOE reductions. IEA Wind (2018), for example, found that OpEx reductions accounted for 9%–11% of overall land-based wind LCOE reductions from 2008 to 2016 in Norway, Germany, and Denmark, 17% in Sweden, and 0% in Ireland. Wiser et al. (2016) reported on a survey of wind experts, who collectively anticipated that OpEx would decline 9%, on average, by 2030; the experts expected that the lower OpEx would account for 11% of the overall decline in land-based wind LCOE from 2014 to 2030, with plant lifetime extensions (related to OpEx, as noted above) accounting for another 14%. Dykes et al. (2017) forecasted a 25% reduction in OpEx for plants built in 2030, contributing to 13% of the projected overall LCOE reduction from 2015 to 2030; they also expected project lifetime extensions accounting for another 22% of the LCOE reduction.
Finally, OpEx for older plants can dictate the economics and timing of plant refurbishment and repowering, which are increasingly important as the wind fleet ages (Ziegler et al. 2018; Mertes and Milligan 2018; Rubert et al. 2018). Though past work has generally found OpEx decreasing over time— with new generations of wind technology—and with increasing turbine size, studies also show that OpEx can increase as projects age (Wiser and Bolinger 2018; IEA Wind 2018; Blanco 2009; EWEA 2009; BNEF 2018; Briggs 2017; Lemming et al. 1999; Rademakers et al. 2003; Vachon 2002; Hahn 1999; Lillian 2018).
Recognizing that wind plant OpEx is an important but sometimes overlooked driver of overall LCOE trends for land-based wind, this paper draws from a survey of senior members of the U.S. wind industry to clarify past and current trends in land-based wind OpEx as well as key drivers of those trends.1 We supplement the survey with a review of literature containing empirical OpEx data for U.S. wind plants. We compare our resulting estimates for average OpEx with other U.S. and global OpEx benchmarks. Finally, we extrapolate historical data to estimate future land-based wind OpEx, and we compare those estimates of potential cost reductions with other recent assessments.
Our core contributions to the broader literature are twofold. First, using an industry survey methodology, we seek consistent historical and recent data on OpEx and clarity on the drivers of OpEx. Given the limited quantity and comparability of previously available data, the data and trends reported here may usefully inform OpEx input assumptions used by electric system planners, analysts, and modelers. The results may also provide useful benchmarks to the wind industry, helping developers and asset owners compare their OpEx expectations with historical experience and other industry projections. Second, we project future OpEx based on historical learning rates. To our knowledge, ours is the first attempt to document a learning rate for OpEx, and to use those findings to forecast a future range in OpEx. These results too may help inform planners, analysts, modelers, research and development managers, and others—and can be compared to and inform other attempts to project future wind power OpEx.
Survey Methods…Land-Based Wind Power OpEx Trends and Drivers…Comparisons with Other Recent Benchmarks…Estimating Future OpEx for Land-Based Wind…
Wind plant OpEx is an important but sometimes overlooked driver of overall LCOE trends for land-based wind. This paper draws primarily from a survey of senior members of the U.S. wind industry to describe historical and current trends in land-based wind OpEx and to provide insights into drivers of those trends. We compare the resulting estimates for average OpEx with other U.S. and global OpEx benchmarks, and we extrapolate the historical data to estimate future land-based wind OpEx, comparing the resulting estimates with other recent assessments.
We find that average all-in lifetime OpEx in the United States has declined from roughly $80/kW-yr (~$35/MWh) for projects built in the late 1990s to levels approaching $40/kW-yr (~$11/MWh) for projects under construction in 2018. Turbine O&M costs—inclusive of scheduled and unscheduled maintenance—represent not only the single largest component of overall OpEx, but also the primary source of OpEx reductions over the last decade.
Reductions in OpEx are attributed to several factors. Wind turbines, wind plants, and owner-fleets have all increased in size, and each increase has reduced costs through economies of scale. In addition, wind technology and operational practices have matured, which has made components more reliable, made widespread the use of automated 24/7 monitoring and condition-based monitoring equipment, and improved predictive and preventative maintenance. Competitive forces, including a diversity of improved OEM service offerings and a growing market for third-party service providers and owner selfprovision of O&M, have also placed downward pressure on OpEx.
Actual OpEx for plants built from the late 1990s through about 2010 were substantially higher than expected OpEx at the time of plant commissioning, resulting in year-over-year increases in OpEx expectations even as actual OpEx declined. Premature component failures, especially gearbox failures, were a key cause of these discrepancies, particularly for some plants and specific turbines. It is believed that a convergence between actual and expected OpEx occurred around 2010.
Though all-in OpEx has declined over time, each point in time contains a wide range of OpEx estimates. For projects commissioned between 2015 and 2018, average lifetime expected costs are reported (often for large fleets of projects) to range from $33/kW-yr to $59/kW-yr ($9–16/MWh). This range is driven roughly equally by variations in turbine O&M costs and all other OpEx categories combined. Some drivers of OpEx variability are more technical in nature, including turbine, project, and fleet size; wind project location; turbine maturity and assumed rates of component failure; wind resource; and local tax rules. Other drivers are strategic in nature, including the choice between OEM versus self-provision of O&M services as well as tradeoffs between the cost and value of enhanced O&M practices.
The all-in OpEx values presented in this paper are often within the range of other recent U.S. and global benchmarks, but they may also inform upward or downward adjustments to some of these benchmarks where limited data are otherwise used. We find a 9% reduction in U.S. wind plant OpEx for each doubling of cumulative global installed wind capacity—that is, a learning rate of 9%. This OpEx learning rate is at the high end of the CapEx learning rate range (6%–9%), suggesting that historical advancements to reduce OpEx have been “doing their share” to reduce the LCOE of wind energy.
We apply the 9% historical learning rate to estimate future land-based wind OpEx reductions under business as usual conditions, finding a possible $5–$8/kW-yr (12%–18%) reduction in all-in OpEx from 2018 to 2040, which would reduce the LCOE of land-based wind by as much as $2/MWh. This estimate is broadly consistent with other projections, with notable exceptions.
These findings suggest that continued OpEx reductions—primarily related to turbine O&M—could contribute 10% or more of the overall land-based wind LCOE reductions expected in the future. Moreover, these estimates may understate the importance of OpEx owing to the multiplicative effects through which operational advancements influence not only O&M costs but also component reliability, performance, and plant-level availability—thereby affecting levelized costs though OpEx reduction and by enhancing annual energy production and plant lifetimes.
Given the limited quantity and comparability of previously available OpEx data, these findings can inform OpEx assumptions used by electric system planners, analysts, modelers, and research and development managers. The results may also provide useful benchmarks to the wind industry, helping developers and asset owners compare their OpEx expectations with historical experience and other industry projections. That said, the estimates presented here are not reliable or precise enough to enable detailed comparisons. Additional effort is clearly required to systematically collect standardized data on wind project OpEx to ensure the comparability of varying data sources and to better understand the differences that remain in OpEx expectations.