Isabelle Blanc, Center for Energy & Processes (CEP), MINES ParisTech, BP 207-1 rue Claude Daunesse, F-06904 Sophia Antipolis Cedex France. Email:Isabelle.email@example.com
A full life cycle assessment (LCA) is usually a time, energy, and data-intensive process requiring sophisticated methodology. Our meta-analysis of life cycle greenhouse gas (GHG) emissions of wind electricity highlights several key, sensitive parameters to provide a better understanding of the variability in LCA results, and then proposes a methodology to establish a simplified, streamlined approach based on regressions built on these key parameters. Wind electricity's environmental performance can be linked to three essential components: technological (e.g., manufacturing), geographical (e.g., wind speed), and LCA methodology (e.g., product lifetime).
A regression has been derived based on detailed LCA results from a representative sample of 17 industrial wind turbines manufactured and recently installed in Europe on average land configurations. Simple GHG performance (i.e., emissions) curves depending on average on-site wind speed and wind turbine lifetime are proposed. Whatever the system power, considering the full range of possible wind speeds in Europe (4 to 9 meters per second [m/s]) and a lifetime of 10 to 30 years, emissions vary from 8.7 to 76.7 grams of carbon dioxide equivalent per kilowatt-hour (g CO2-eq/kWh) when the wind speed is less than 6.5 m/s, and from 4.5 to 22.2 g CO2-eq/kWh when the wind speed is 6.5 m/s or greater. This second situation with a turbine lifetime of 20 years is assumed to be most realistic based on economic criteria.
This research presents simplified models as an alternative to detailed LCA. The methodology has been applied as a first trial to wind electricity and could be applied to other energy pathways.
Wind energy production has been widely studied for the last three decades. Wind turbine technology has greatly improved and drastically increased the average nominal rated power, from an average power of 730 kilowatts (kW) in 2002 to 1.85 megawatts (MW) in 2009 (Le Jannic and Petitjean 2010).1 Thus, in 2009, 37.4 gigawatts (GW) of new onshore wind turbines (WTs) were installed worldwide, representing an annual 42% growth in installed capacity (Global Wind Energy Council 2010).2
Recent reports on renewable energies promote the relative greenness of these technologies compared with fossil fuel-based systems (European Wind Energy Association 2009; Greenpeace and EREC 2008). Most of the time, electricity generated from WTs is considered carbon dioxide (CO2) free over its use phase (Global Wind Energy Council 2010). However, if examined beyond direct emissions and from a life cycle perspective, WTs do have environmental impacts (such as “indirect CO2”) during, for example, the manufacturing or building phase of the system. It is therefore necessary to consider impacts over the full life cycle of wind power technology, especially when aiming to compare different energy pathways (Greenpeace and EREC 2008).
Life cycle assessment (LCA) is a useful tool dedicated to the assessment of environmental impacts over all the life stages of a product, providing a “cradle-to-grave” environmental profile. LCAs have been used since the 1970s, and have been widely used since the methodology was standardized through guidelines of the International Organization for Standardization (14040; ISO 2006a, 2006b). Many wind power environmental assessments based on LCAs focus on primary energy consumption and greenhouse gas (GHG) emissions.
The environmental performance of WTs is highly variable. For instance, Lenzen and Munksgaard (2002) show through literature review an emissions range of 7.9 to 123.7 grams carbon dioxide equivalent per kilowatt-hour (g CO2-eq/kWh).3 Such a wide range of results is obviously linked to the variable assumptions and parameters used by different authors. Many reasons explain such variability, such as the dynamic development of this industry and the broad array of technologies it represents. However, the GHG emissions from wind power electricity remain far below the values observed for fossil fuel electricity, such as coal electricity, which ranges from 755–1,309 g CO2-eq/kWh (Spadaro et al. 2000). Renewable energy systems’ environmental performance is also highly geo-dependent (Blanc et al. 2008) and driven by external factors that influence electricity production; in our particular case, wind characteristics (European Environment Agency 2009). This wide variability in environmental performance can lead policy makers to consider LCA as an inconclusive method (Reap et al. 2008). To improve confidence in LCA results, it is thus necessary to make these wide-ranging results comprehensive and explain the sources of variability through meta-analysis by identifying the main parameters influencing and characterizing the system's environmental performance.
Another important step to expand the use of LCA in industry is to provide a simpler, more efficient approach to assessing environmental impacts. Simple and easy access to robust environmental assessments would be useful, for example, to identify the most suitable locations for new WTs. We propose a methodology to establish a simplified approach to analyzing onshore WTs using regressions built on key sensitive parameters. This approach does not substitute for a full detailed LCA, but relies on some key recursive parameters that can explain most impacts. Bala and colleagues (2010) have already presented the possibility of decreasing the use of detailed LCAs by developing reduced inventories comprising the main parameters influencing GHG emissions. Our methodology shares the same goal by providing a simple and robust tool to estimate GHG performance in the wind power industry. However, unlike the work of Bala and colleagues (2010), we do not focus on reduced LCAs, but on regressions based on a representative set of detailed LCA results. Thus, in this article, we present the methodology for producing a reduced model capable of deriving the environmental performance of any existing onshore wind power electricity system. This generic approach has been developed for one single impact indicator related to climate change (GHGs expressed in g CO2-eq/kWh).
The methodology is described in the following sections:
2A detailed analysis of the origin of wind turbine impacts is provided through thorough assessment of one WT inventory from the Ecoinvent database. The Ecoinvent (Ecoinvent 2009) WT LCA analysis and the literature survey highlight the importance of three parameters types (technological, geographical, methodological) and support the identification of key sensitive parameters, which are responsible for the main impacts related to wind power electricity generation.
3A representative sample of current WT technologies is selected (also based on the technical data available) and their related GHG performance are calculated. Then, regressions are generated between the key sensitive parameters identified and the GHG emissions results.
4The regressions are then applied to a new range of WTs and their impact performances are extrapolated. Results are compared to the related detailed LCAs in order to assess the validity and reliability of the simplified approach.
5The limits of the relevance of this type of approach derived from LCAs for the wind turbine sector are then examined.
Variability of the Environmental Performance of Wind Turbines from Published Life Cycle Assessment Studies
Figure 1 illustrates the distribution of WT carbon performance with a chart showing the frequency in literature of various CO2 equivalent emissions per kilowatt-hour.
Most impacts range from 5 to 35 g CO2-eq/kWh. The average GHG performance per kilowatt-hour is 13.5 g CO2-eq/kWh. The same trend and results are observed by Weisser (2007), confirming WT electricity as a low CO2 equivalent emission technology. Through analyzing results in these publications, it is possible to determine the importance of some parameters and life cycle phases, which could explain their influence on the results distribution:
• The manufacturing phase is the main impact over the life cycle. Different studies highlight the significance of the system's manufacturing phase. The impacts are mainly caused by WT steel content, but also, to a lesser extent, by the composite material used for the nacelle encapsulation and the blades (Martínez et al. 2009). The concrete used for the foundation also represents a significant share of total WT impacts.
• Most studies demonstrate that impacts during the building and operation phases are negligible (Jungbluth et al. 2005).
• Maintenance represents a significant share of the impact. This maintenance is often defined as replacing 15% of nacelle components and one blade during a WT lifetime (Chataignere and Le Boulch 2003).
• The load factor, which characterizes electricity production performance, has a great influence (Lenzen and Wachsmann 2004). Load factors are assumed to be in the range of 20% to 30%; the average value in the publications covered was 27.4%, but no explicit calculations or complete sensitivity analyses are provided for this specific parameter.
• The lifetime of the entire WT is assumed to be 20 years; only one publication considers a lifetime of 30 years (Hondo 2005).
Identification of Key Parameters Through Wind Turbine Life Cycle Assessment Disaggregation
Following this first survey of WT LCA performance, and in order to attain deeper knowledge of the contribution of each subsystem, we performed a disaggregation of one specific WT system inventory then analyzed its related impacts to understand and identify the most influential parameters.
We selected the Nordex N50/800 (800 kW) life cycle inventory (LCI) from the Ecoinvent database version 2.1 (Ecoinvent 2009), as this WT is representative of the most common WT installed in 2003 (Le Jannic and Petitjean 2010). No more recent WT inventories for onshore WTs in the Ecoinvent database are available, and we consider this inventory a valid reference, as no significant change in technology has appeared since then. For this specific inventory, the flows were provided by the manufacturer (Dones et al. 2007).
The assumed lifetime is 20 years for moving parts (rotor, nacelle, generator, etc.) and 40 years for fixed parts (tower and foundation). The connection to the grid is included in the moving parts. Figure 2 presents the relative GHGs per kilowatt-hour of electricity produced throughout the different WT phases.
From figure 2 it is possible to conclude that transportation (for the maintenance phase) and lubricating oil represent a negligible share of the overall impact. Considering the maintenance phase, compared to the literature survey (Chataignere and Le Boulch 2003), some nacelle component replacements are missing in this inventory. Maintenance is therefore assumed to be negligible and not fully considered.
The moving parts (i.e., rotor and nacelle) represent more than 67% of the overall impact. The fixed parts have less impact (32% of the overall impact) partly because their assumed lifetime is twice as long as that of the moving parts.
Going a step further in the disaggregation of the fixed and moving parts, we observe that WT impacts are only driven by a limited number of materials and processes, mainly composite, concrete, steel, and their processing (see the supporting information available on the Journal's Web site for more information). This confirms the need to focus on these flows instead of putting intensive effort and time into obtaining information on flows with a minor influence on the final WT GHG impacts.
Sample Selection and Characterization of Wind Turbines for the Regressions
In order to calculate the GHG performance regressions of wind power electricity, we must define a representative WT sample and build associated WT inventories. The sample needs to represent the current industrial WT market in terms of power installed and technology (characterized by materials, power curve profiles, swept area, etc.). We chose 17 industrial WTs from different manufacturers: three 800 kW (Dones et al. 2007; Enercon 2010a, 2010b), three 850 kW (Gamesa 2010a, 2010b; Vestas 2007), one 1.65 MW (Vestas 2008), nine 2 MW (Gamesa 2008a, 2008b, 2009; Vestas 2006), and one 4.5 MW (Chataignere and Le Boulch 2003) WT. The WT characteristics can be found in the supporting information available on the Web. For these WTs we were able to obtain their mass and power curve profiles, which are the minimum data we need to build their inventories. A survey on wind power development (Le Jannic and PetitJean 2010) has shown that 800 kW WTs were the standard type of WT installed in 2003, and that 2 MW WTs are installed as of 2010. Moreover, in 2008 the manufacturers we selected (Nordex, Enercon, Gamesa, and Vestas) represented 43% of the wind power market (Lopez 2009). Our study is therefore assumed to be representative of the wind power market from the beginning of its “boom” period (2002) to now. WTs with higher rated power do exist, but discussion with experts concluded that they are not used in onshore configurations in Europe (Maupu 2010). The regression we develop is based on this sample.
The main differences between the WTs in our sample are the tower and nacelle weights, the blade lengths (and thus the swept area), the power curves, and the hub heights. Technical characterizations for some WTs from the sample are given in the supporting information on the Web (see tables S1–S4).
Assumptions for Modeling Wind Turbine Inventories
The next step is to build the 17 selected WT inventories to calculate GHG performance. The only one of these inventories currently available in the Ecoinvent database is for the 800 kW WT. We shall now describe the set of assumptions used to extrapolate new inventories from the initial 800 kW inventory.
In the Ecoinvent database itself, all material flows for the moving parts or a 2 MW offshore WT were estimated using the material distribution of the 800 kW onshore WT nacelle (Dones et al. 2007). The same material extrapolation assumption is applied to the full set of WT inventories; details of the material shares can be found in the work of Dones and colleagues (2007). We selected a 100% steel, low-alloy composition for the tower. Regarding the 2 MW WT foundation, it was decided to use 338 cubic meters (m3) of concrete and 30 tonnes (t) of reinforcing steel, as selected by Sutcliffe and colleagues (2010).4 It is also necessary to estimate the connection to the grid. In a first approximation we consider Ecoinvent for information on the connection to the grid of onshore 800 kW WTs (Dones et al. 2007). The associated GHG impact is 6,950 kilograms (kg) CO2-eq over a 20-year lifetime. The section on approach limitations discusses the sensitivity of GHG emissions from wind power electricity to this subsystem.5
As seen in figure 2, transport for maintenance and oil disposal is of minor importance. Thus it is not considered in our WT inventories.
The energy/transport requirements (for manufacturing) are difficult to obtain. Our work aims at reducing the time/energy necessary for the GHG performance calculation; we thus had to define an approach to consider their contributions. We decided to use the data provided by the Ecoinvent database (for an 800 kW WT) for these flows, and to assume a proportionality hypothesis for the other nominal power rate. Thus energy/transport flows represent 4.5% of the overall WT GHG impact. Sensitivity to this assumption is discussed in the section on approach limitations.
Finally, from our literature review and maintenance considerations, we decided to include a 15% replacement rate of nacelle components and one blade replacement during the WT lifetime. The lifetimes of both the moving and the fixed parts are assumed to be identical and set to 20 years, as found in the literature. These modifications to the initial 800 kW inventory and its extrapolation scheme were necessary to model our sample of 17 WTs.
Analysis of Modeling Parameters for Wind Turbine Environmental Impacts
The next step in setting our regressions involves the identification of key parameters explaining the GHG WT results based on sensitivity analyses.
We define GHG environmental performance as the ratio between the overall WT GHG impact (impactsWT) and electricity production over the WT life cycle:
8,760 = total number of hours in a year;
L= load factor (ratio of the operating hours at nominal power divided by the total hours in a year);
A= availability factor, which takes into account periods when the WT should be producing but for various reasons is not;
LT= WT lifetime; and
P= WT nominal power.
We identify three types of parameters related to three dimensions: a technological dimension, with parameters related to the type of technology and materials used for the WT; a geographical dimension, directly related to wind conditions through the load factor (L); and a methodological dimension with the lifetime (LT) parameter.
Sensitivity studies are performed first for GHG impacts, then for GHG performance, varying these three parameters to identify which is most influential.
Sensitivity of Greenhouse Gas Emissions to Technological Parameters
We investigated the GHG emission variations of several 2 MW WTs currently installed. The weights of the different components were obtained from WT manufacturers’ datasheets (Gamesa 2008a, 2008b, 2009; Vestas 2006).
Impacts expressed in GHG for five 2 MW WTs show a relative standard deviation of 9% (figure 3). Because the material distribution is assumed to be the same for all WTs, the differences are due to the variation in the components’ weights (e.g., tower weight and nacelle weight). The variability of these 2 MW WT GHG impacts is fairly limited under identical conditions of wind turbine lifetime and wind speed. The next step is to investigate the influence of geographical parameters (wind regime) for the same WT sample on GHG performance.
Sensitivity of Greenhouse Gas Emissions to Geographical Parameters
Using the production calculations for wind power electricity described in the supporting information on the Web, the power curves from product datasheets and PelaFlow consulting (2010), and equation 1, wind power GHG performance per kilowatt-hour is presented as a function of wind speed for the same set of 2 MW WTs. Moreover, GHG performance is calculated with an availability factor of 95%.
The results presented in figure 4 show great variability in relation to the average wind condition. For low wind speed conditions, the average GHG results are around 35 g CO2-eq/kWh, but for high wind speeds this result drops to 7 g CO2-eq/kWh, or five times lower. Indeed, for high wind speeds, WT performance depends less on the power curve profile, as the occurrence of wind corresponds more closely to the nominal power of the WT.
Figure 4 also shows that for medium-to-good wind site locations (above 6 meters per second [m/s]), GHG emissions per kilowatt-hour are very close for each WT.6 For instance, at 7.5 m/s average wind speed, the maximum difference between all technologies is 1.5 g CO2-eq/kWh (with a relative standard deviation of 7.5%). For very low wind speeds, the biggest difference in CO2 equivalent emissions is more significant (up to 11 g CO2-eq/kWh, relative standard deviation of 29%, for a wind speed of 4 m/s).
We identified possible sources of variability for wind power GHG performance based on equation 1: lifetime, load factor, WT GHG impact from manufacturing, and power.
As a first step, we have excluded lifetime from our variability study and will consider it at a later stage. We go on to analyze the other three sources of variability in more detail:
• Load factor variability: the load factor represents electricity production as a function of wind speed frequency by giving an equivalent percentage of operation at nominal power (see the supporting information on the Web). It is strongly correlated with WT GHG performance because it is defined as a function of the wind speed at the WT hub height (TheWindPower.org 2010). Other parameters defining the load factor function are land roughness and the power curve profile (see equations in the supporting information on the Web). Land roughness is independent of technology and is a geographical parameter type. We assume that we will set this parameter to a standard class of land (agricultural land) to simplify the study and not consider this parameter variability.
• WT GHG impact variability from manufacturing: We have assumed similar material distribution for the nacelle within the WT sample; a fairly low variability for the mass of moving parts is found within the sample (see tables S1–S4 in the supporting information on the Web), which induces marginal variability for GHGs.
• Size effect of the WT power: in our study, this parameter is intrinsically linked to the others presented above; we assume this effect to be negligible and consider linearity between the different WT nominal powers. While significant, this assumption has already been made where the size of the turbine does not appear to be an important factor in optimizing the life cycle energy performance (Crawford 2009). WT power might only have an impact when considering small equipment, less than 500 kW, as shown by Tremeac and Meunier (2009).
The wind power electricity GHG performance is therefore identified as a function of two major parameters: the WT lifetime and the average wind speed on the site.
Wind Turbine Greenhouse Gas Performance Regression
So far we have identified the parameters to which the results (in GHG performance) are most sensitive: wind speed has been demonstrated to be highly influential on performance, with a factor of five between the lowest and highest wind conditions (figure 4). Conversely, the GHG impact from WT manufacturing, presented in figure 3, can be considered as low (relative standard deviation of 9% for the five 2 MW samples). These findings lead us to define a regression from the 17 WT sample, relating the wind electricity GHG performance to a single parameter: wind speed. We derive this GHG performance curve through a polynomial interpolation as a function of wind speed with a fixed lifetime (LT) initially assumed to be 20 years.
Table 1 and figure 5 confirm that GHG performance is highly dependent on a site's wind conditions, which is directly linked to electricity production. As a consequence, optimizing economic profitability is a win-win situation, as it minimizes GHG emissions: the more electricity produced the less CO2 content per kilowatt-hour produced. Indeed, electricity production performance is the main parameter influencing economic and GHG performance.
Table 1. Greenhouse as (GHG) performance for WTs as a function of wind speed (lifetime = 20 years)
Average wind speed (m/s)
GHG performance (g CO2-eq/kWh)
Relative standard deviation (%)
Notes: m/s = meters per second; g CO2-eq/kWh = grams carbon dioxide equivalent per kilowatt-hour.
However, in low wind conditions (typically less than 5 m/s), as the confidence interval gets larger, GHG performance may depend on a wider set of parameters. Thus, for very low wind speed conditions (worst possible case in Europe), emissions are found to be 38.3 g CO2-eq/kWh (±14.8%). For average wind conditions (∼6 m/s) GHG emissions are 13.0 g CO2-eq/kWh (±12.2%), which is very close to the results in the literature. For high wind speeds (greater than 8 m/s), the GHG performance of WT electricity ranges from 5 to 10 g CO2-eq/kWh (∼±10%). The confidence interval also decreases as the wind speed condition increases, because the sample results become less dispersed. Finally, the coefficient of determination (R2) of the regression is 0.91.
Overall Reduced Life Cycle Assessment Results Including Lifetime
As described in equation 1, LT is an LCA methodological parameter that needs to be assessed separately from technological and geographical parameters. Indeed, this factor strongly influences total electricity production because LT can change with the type of material used in WT fabrication and depends on climatic conditions at the installation. Furthermore, LT is an arbitrary LCA assumption, but it is critical to resulting sensitivity. Standard LT figures considered range from 10 to 30 years; they are complex to define and important when considering their variability.
We included this parameter and derived a set of regressions as a function of the two main factors: lifetime and wind conditions (figure 6 and table 2).
Table 2. Greenhouse gas (GHG) performance as a function of lifetime (LT) and average wind speed (meters per second)
Average wind speed (m/s)
GHG performance (g CO2-eq/kWh)
10 year LT
20 year LT
30 year LT
Notes: m/s = meters per second; g CO2-eq/kWh = grams carbon dioxide equivalent per kilowatt-hour.
Table 2 and figure 6 show the strong influence of lifetime on GHG performance. The selected lifetimes approximately represent lower, average, and upper real values. When forecasting the performance of a new WT installation, it is necessary to estimate the upper and lower possible environmental value for the kilowatt-hours produced, including the variation of one particularly important methodological parameter: lifetime.
Figure 6 gives the full overview of possible GHG performance for actual onshore wind power electricity. The distribution of literature results can now be explained. The confidence interval remains unchanged compared to the results in figure 5 and table 1. Indeed, in figure 6 and table 2, results are presented as a function of the average wind speed and lifetime, which does not induce any modification in the confidence interval definition.
Finally, in order to consider the actual wind power context, we added an economic criterion: the box in the bottom-right corner of figure 6. Nowadays most onshore WT systems are installed with a 20 year lifetime and an annual 50 m height wind speed of at least 6.5 m/s; see Archer and Jacobson (2005) for a wind atlas and ADEME (2011) for WT locations in France. In this restricted interval we thus obtained a GHG performance range from 4.5 (±10.6%) to 11.1 (±11.7%) g CO2-eq/kWh. This restricted interval presents the upper and lower limits of the GHG performance considering the economic criterion. Nevertheless, the full GHG performance interval (for a lifetime between 10 and 30 years and an annual wind speed of 4 to 9 m/s) is presented to explain the variability in the literature and because
• economic feasibility criterion can evolve depending on WT market costs, and
• financial incentives can lead developers to install WT in less windy sites.
Usefulness and Validity of the Approach Developed
The validity of the reduced model must be estimated, and a comparison with the literature survey is now presented. In the general literature survey, the GHG performance of wind electricity was found to be 13.5 g CO2-eq/kWh for “average conditions.” Applying the regression gives a result of 13.0 g CO2-eq/kWh ± 12.2% (at 6 m/s and considering a 20 year lifetime). Thus the results obtained fully match the literature survey interval. To check the validity of the regression, we now apply it to some specific WT situations. In the study by Jacobson (2009), GHG performance of wind power electricity ranges from 2.8 to 7.4 g CO2-eq/kWh (for a 30 year lifetime and an average wind speed between 7 and 8.5 m/s). These specific cases were not in the sample used to develop the model. By calculating GHG performance of wind power electricity using the newly defined regression for the same lifetime and identical average wind speed conditions, the regression results range from 4.8 g CO2-eq/kWh (±0.5 g) to 6.5 g CO2-eq/kWh (±0.7 g). Considering the range of uncertainty, both studies give comparable results, confirming the validity of our regression compared to detailed LCA results.
Nevertheless, as explained previously, this work does not aim to substitute for a complete LCA. The approach presented offers two major advantages:
• it presents an explicit description of system variability, enabling us to understand the variability of LCA results observed (presented in literature); and
• it provides an estimation of GHG performance of wind power depending on the main parameters, including the differentiation between geographical and methodological parameters.
Approach Limitations and Error Assessment
In order to perform the regression for GHG performance of wind power electricity, we had to make several assumptions to build the WT inventories sample. The regression quality is thus linked to the quality of the chosen hypothesis, and its validity is restricted to the sample characteristics. Possible errors related to these assumptions need to be underlined.
The grid connection and the energy/transport requirement contributions have been estimated to represent 4.5% of the overall WT impacts. This hypothesis could induce errors. However, such errors would remain at a low level: an increase of 1.5% if grid connection material needs doubled, and 5% if energy/transport requirements also doubled. Compared to lifetime and wind speed, the transport contribution presents a small influence in the European case, but could have a significant influence if the WTs are produced or used outside Europe (Lenzen and Wachsmann 2004). The authors have shown that transport inside Europe (Germany) represents less than 2% of the GHG emissions for WTs manufactured and installed in Europe. Our selected sample corresponds to this case. Moreover, sensitivity analysis of WT impacts have shown variability linked to the energy and transport requirements during the WT life cycle (Martínez et al. 2010). It appears that, as much as possible, WTs are produced locally to avoid long distance transportations; this trend is observed both in Europe (Vestas 2011) and in China (ALEF – International Energy post-master program 2011).
In the WT nacelle flow quantification, another error could be caused by choosing the wrong steel in the inventories (low-alloy steel versus chromium steel). Most types of steel present a comparable CO2 equivalent impact per kilogram. Chromium steel, which emits three times more CO2 equivalent per kilogram, is the exception. Thus if the correct steel is not assigned to part of the nacelle components, this would induce an error. For instance, if 10% (of the nacelle weight) is made of chromium steel instead of low-alloy steel, the final error induced would be around 3.8%.
In addition, the end of life of WTs was not taken into account in the work because most of the studies presented in the literature survey (Ardente et al. 2008) and the Ecoinvent (2009) database assume a high recycling rate for the metals; only the composite is supposed to be incinerated or partially landfilled. The concrete is partially removed (and landfilled) or left in the ground. The recycling process is partially described in the Ecoinvent database and it supposes no CO2 impacts (the old material is used again to replace the new material, without any loss of properties or treatment). Thus only the composite incineration has an impact, but this is estimated to be low (less than 2%) and is therefore omitted.
The land roughness parameter influence has been partially assessed. This parameter is used to calculate the wind speed at the WT hub heights (see supporting information on the Web). We selected a standard value for agricultural land, which is supposed to be representative of most WT installation sites. The roughness factor could have an influence if urban sites are considered. However, because of policies, installations in those sites seem unrealistic for industrial WTs. Considering the roughness in coastal areas, for a 2 MW WT the extrapolated wind speed at hub height would increase by 3%. The results would be modified by approximately 5%.
Finally, we can roughly estimate a maximum 15% total error associated with the assumptions we made for the WT LCIs. However, this 15% error is an accumulation of all possible errors and should be considered a worst possible case.
The main goal of this study was to develop a methodology toward a simplified approach to quantifying the GHG performance of WTs as an alternative to performing a full LCA. First, throughout our contribution to a meta-analysis of life cycle GHG emissions, this present work highlights the importance of several key sensitive parameters, and thus provides the user a better understanding of the variability of results in the published literature on LCAs. A regression was then derived based on detailed LCA results for a representative sample of 17 industrial onshore WTs recently installed in Europe on average land configurations (agricultural). GHG performance curves for WTs are based on the most dominant geographical and methodological parameters: wind speed and wind turbine lifetime. An estimation of greenhouse gas emissions from the electricity produced by a wind turbine during its lifetime is therefore possible for onshore WTs using these regressions. The regression results range from 3 to 77 g CO2-eq/kWh, depending on the main parameters. Uncertainties associated with the regressions vary according to the range of wind speed considered: ±15% for low wind speed conditions, ±12% for average conditions, and ±11% for good conditions. For standard wind conditions in Europe, the mean results are similar to the variability range of results in the literature. For low wind conditions, the derived LCA regression proposed in this article shows a greater level of uncertainty. For low average wind speeds, electricity production is a critical factor, and technological parameters such as hub height, or most probably power curve profiles, might have a greater influence on GHG performance results. This should be considered to further refine this set of regressions. Regardless, however, the GHG performance (i.e., emissions) of wind electricity remains far below that of fossil fuel-based electricity.
This research presents a new methodology for creating an alternative model to detailed LCA for ascertaining environmental performance. The methodology to develop this lightweight approach has been applied as a first trial to wind power electricity GHG performances, and it could be applied to other energy pathways.
This article has been made possible thanks to the on-going international post-master program (ALEF) jointly organized by the MINES ParisTech and Tsinghua University. Special thanks go to Romain Richard and Vincent Maupu from EDF and Dr. Thierry Ranchin from MINES ParisTech for their great expertise in this topic, as well as to the three anonymous reviewers.
One kilowatt (kW) ≈ 56.91 British thermal units (BTU)/minute ≈ 1.341 horsepower (HP). One megawatt (MW) = 106 watts (W, SI) = 1 megajoule/second (MJ/s) ≈ 56.91 × 103 British thermal units (BTU)/minute.
One gigawatt (GW) = 109 watts (W, SI) = 1 gigajoule/second (GJ/s) ≈ 56.91 × 106 British thermal units (BTU)/minute.
One kilowatt-hour (kWh) ≈ 3.6 × 106 joules (J, SI) ≈ 3.412 × 103 British thermal units (BTU). Carbon dioxide equivalent (CO2-eq) is a measure for describing the climate-forcing strength of a quantity of greenhouse gases using the functionally equivalent amount of carbon dioxide as the reference.
One cubic meter (m3, SI) = 103 liters (L) ≈ 264.2 gallons (gal). One metric ton (t) = 103 kilograms (kg, SI) ≈ 1.102 short tons.
One kilogram (kg, SI) ≈ 2.204 pounds (lb).
One meter (m, SI) ≈ 3.28 feet (ft).
About the authors
Pierryves Padey is a Ph.D. student and Isabelle Blanc is a professor in the Center for Energy and Processes at MINES ParisTech in Sophia Antipolis, France. Denis Le Boulch is a research engineer at EDF in Les Renardières, France. Zhao Xiusheng is a professor at INET at Tsinghua University in China.