Introduction

One renewable energy source that has witnessed a significant growth in recent years is wind energy, with the installation of new wind farms around the globe. In the US in particular, installed wind capacity grew 13% in 2017, with wind energy producing 6.3% of US power (American Wind Energy Association 2017). Texas ranks first among US states in terms of wind power capacity, with over 22 GW of installed power as of 2017 (American Wind Energy Association 2017). One significant advantage of wind turbines is that they generate electrical energy from the wind’s kinetic energy without causing emissions or pollution in the conversion stage; however, this does not imply that the energy source lacks greenhouse gas or traditional air pollutant emissions altogether. Wind turbine manufacturing, disposal, and even operation can have environmental effects. For example, increased use of wind energy decreases loading on conventional thermal units, which may increase fuel consumption due to off-design operation (Stanek et al. 2018; Turconi et al. 2014). In order to compare effects of wind energy production with other energy resources, emissions and environmental impacts of wind power must be quantified over its entire life cycle.

A large number of previous studies have been conducted for evaluating environmental impacts of wind turbines. Wang and Wang (2015) reviewed 64 studies in terms of wind turbine impacts on noise pollution, bird and bat fatalities, greenhouse gas emissions, and land surface impacts. Davidsson et al. (2012) compared 12 life cycle assessments of wind energy systems in terms of methods used, energy use accounting, quantification of energy production, energy performance and primary energy, natural resources, and recycling. Leung and Yang (2012) reviewed 80 studies in terms of wind turbine impacts on noise and visual impacts, impacts on animals and birds, and greenhouse gases. Arvesen and Hertwich (2012) provide a more comprehensive review of the literature up to 2012, categorizing 59 studies by onshore versus offshore turbines, turbine size, turbine lifetime, geographic scope, and impact categories included.

For large (> 1 MW) onshore wind turbines, a number of life cycle assessments (LCAs) have been conducted which are limited to energy (China (Chen et al. 2011); Australia (Crawford 2007); Denmark, Germany, China (Guezuraga et al. 2012); Germany (Wagner and Pick 2004)), greenhouse gas emissions (China (Ji and Chen 2016); US (Kumar et al. 2016); Brazil (Oebels and Pacca 2013)), or energy and greenhouse gas emissions (US (Rajaei and Tinjum 2013; New Zealand (Rule et al. 2009)). In addition, Simons and Cheung (2016) developed a method for predicting energy production and greenhouse gas emissions from wind turbines over their lifetime. Such studies, however, do not comply with the ISO 14044 requirement to cover all relevant impacts of the assessed system (International Standards Organization 2006), including environmental impacts not related to climate change, human health impacts, and impacts on consumption of resources besides energy (water and materials).

Comprehensive LCAs for large onshore wind turbines, including multiple impact categories beyond energy and greenhouse gas emissions, have been conducted for Spain (Martínez et al. 2009), France (Tremeac and Meunier 2009), the United Kingdom (Ozoemena et al. 2018), and a mix of countries globally (Garrett and Rønde 2013), but none specifically for the US. Impacts for the US would be anticipated to be different from that of other countries, even using the same model wind turbines manufactured in Europe, due to:

  • Longer transport distances (turbines parts are shipped from Europe to the US via ship, whereas previous studies have generally included only truck transport within Europe),

  • Differing vehicle emission standards for heavy-duty diesel trucks used for transport of parts for assembly and dis-assembly: lower standards for carbon monoxide, hydrocarbons, and particulate matter in Europe, but lower standards for nitrogen oxides in the US (DieselNET 2018),

  • Differing end-of-life disposal practices (landfilling dominating in the US, versus combustion for many European countries) (Worrell et al. 2016). Cherubini et al. (2009), for example, found landfilling of waste with gas capture and utilization to produce greater net carbon dioxide emissions compared to combustion, but less acidification potential and dioxins.

In addition, the previous comprehensive wind turbine LCAs do not consider separately all phases of the turbine life cycle (materials acquisition/manufacturing, transportation, installation, operation and maintenance, and end of life). Martínez et al. (2009), Tremeac and Meunier (2009), and Ozoemena et al. (2018) lump raw material acquisition, manufacturing, and installation together into one phase. Ozoemena et al. (2018) and Garrett and Rønde (2013) include transportation with the other phases, not separately. Hence, this will be the first comprehensive LCA to treat all 5 phases of the life cycle explicitly.

In terms of turbine LCAs examining only greenhouse gas emissions and/or energy, installation and maintenance in particular are often omitted, or if included, are included only in part or not described in detail (Ji and Chen 2016; Oebels and Pacca 2013; Rajaei and Tinjum 2013; Rule et al. 2009; Martínez et al. 2015; Jungbluth et al. 2005; Proops et al. 1996). Particular attention will be given in this study to making the installation and maintenance phases complete and transparent.

A number of previous wind turbine LCAs have included an uncertainty/scenario/sensitivity analysis for various parameters, as summarized in Table 1. However, only one study (White 2006) has examined the impact of turbine lifetime, and this study was for medium-size turbines (< 1 MW) and included only energy and greenhouse gas emissions.

Table 1 Previous wind turbine LCAs including uncertainty/scenario/sensitivity analyses
Full size table

Accordingly, the goal of this study was to conduct a comprehensive life cycle assessment (LCA) for large wind turbines in the US, including all phases of the turbine’s life cycle separately (materials acquisition, manufacturing, transportation, installation, operation and maintenance, and end of life) and multiple impact categories (environmental, human health, resource consumption). Particular attention was given to making the installation and maintenance phases complete and transparent. Uncertainty in 2 LCA model inputs (turbine lifespan and wind speed) was also evaluated.

In particular, the study aimed to answer the following questions:

  • Which of the 5 life cycle phases produce the greatest environmental, health, and resource consumption impacts, so that their influence can be reduced?

  • Are the phases producing the greatest impact in the US similar to those for wind turbines in other countries?

  • Does comprehensive treatment of the installation and maintenance phases increase their impact, compared to other studies?

  • Does extending a turbine life span reduce its environmental impacts (does increased energy production offset increased maintenance requirements as the turbines age)?

  • What are life cycle emissions for wind energy in the US compared to other sources of energy?

The contribution of this study is that it is the first comprehensive LCA for large wind turbines in the US, including all phases of the turbine’s life cycle, and impact categories beyond energy and greenhouse gas emissions. For the US, different transport distances (including overseas transport of turbine parts), truck emission standards, and waste disposal practices will affect impacts, compared to LCAs conducted for other countries. It is also the first comprehensive LCA to examine separately all 5 phases of the turbine’s life cycle, as well as turbine lifespan as a sensitivity parameter.

Methods

According to the ISO 14040 and 14044 standards, an LCA is carried out in four stages (International Standards Organization 2006):

  1. 1.

    Goal and scope definition,

  2. 2.

    Inventory analysis: tabulating all inputs and outputs to the system,

  3. 3.

    Impact assessment: evaluating potential environmental, health, and resource consumption impacts associated with the inputs and outputs,

  4. 4.

    Interpretation: determining the significance of the potential impacts.

Goal and scope definition

As mentioned above, the goal of this LCA was to evaluate the potential life cycle environmental, health, and resource consumption impacts associated with the production of electricity from 2-MW Gamesa turbines located at a 400-MW onshore wind plant in the US comprised of 200 turbines. All phases of the turbine’s life cycle, including materials acquisition and manufacturing; transportation; installation; operation, maintenance, and replacement of parts; and end of life, were assessed.

Few studies of wind turbines have explicitly included maintenance, so this factor was of particular interest. Industry data regarding wind turbine life spans is limited due to the short time that most farms have been in operation. Several industry estimates along with studies have shown that an individual wind turbine’s life span before major maintenance is conducted is 20 years (Proops et al. 1996; Schleisner 2000; European Environment Agency 1998; Lenzen and Dey 2000). Therefore, the moving parts (generators, gearboxes, and rotors) are substituted after 20 years, while the turbines’ supporting systems and wind farm are not interfered with. The decision to operate a turbine longer than 20 years has some advantages and disadvantages. Longer life of the turbines might be a way to increase the revenue, but it means more maintenance than usual because the older the parts of the turbines mean more maintenance is needed. This study aims to determine whether extending the lifetime is beneficial in overall environmental, health, and resource consumption impacts.

In terms of physical scope, the study was conducted for 2-MW Gamesa onshore wind turbines G83 and G87 located at the Lone Star Wind Farm near Abilene, Texas, as well as the collector substation which increases voltage for transmission to the electric power grid. The wind farm contains 100 each of G83 turbines and G87 turbines, which are very similar, as seen in Tables 2, 3, 4 and 5. One “average” turbine was modeled: when the G83 and G87 turbine parts differed in terms of mass, an average value was used. Impacts per turbine were then divided by the total energy production over the turbine lifetime in kWh, so that the impacts are per functional unit, per kWh. Impacts of the G83 and G87 turbines were not modeled separately. The transmission lines themselves and distribution grid were not included.

Table 2 Gamesa turbine G83 and G87 parameter information (Gamesa 2013)
Full size table
Table 3 Nacelle components of Gamesa G83 and G87 turbines (Gamesa 2013)
Full size table
Table 4 Rotor components of Gamesa G83 and G87 turbines (Gamesa 2013)
Full size table
Table 5 Wiring, tower, and foundation components of Gamesa G83 and G87 turbines (Gamesa 2013)
Full size table

The Lone Star Wind Farm was chosen because it contains Gamesa G83 and G87 wind turbines, which are widely used and have publicly available data. As of 2016, over 2000 MW of capacity of Gamesa turbines had been installed worldwide, including over 400 MW in Texas, USA, around 500 MW in China, and over 1000 MW in India.

Figure 1 shows the flowchart of the wind turbine life cycle phases. The term “wind turbine” includes moving parts such as nacelle (housing, frame, main shaft, gearbox, brake system, generator, control system) and rotor (hub, nose cone, and blades), as well as fixed parts such as the tower and foundation. Activities associated with each phase are discussed in more detail below. The functional unit was defined to be 1 kWh of electricity delivered to the grid by a wind power plant (through the collector substation).

Fig. 1
figure 1

(adapted from D’Souza et al. 2011)

Wind turbine life cycle phases

Full size image

Inventory analysis

SimaPro Software version 8.3.2 was used to conduct the life cycle inventory and assessment (PRé Sustainability 2015, 2016). Data input into SimaPro for the modeled phases of the turbine’s life cycle are discussed below.

Data for raw material acquisition and manufacturing phase

Table 2 provides turbine parameter information for Gamesa turbines G83 and G87, and Tables 3, 4, 5 and 6 show material quantities for particular turbine components (nacelle; rotor; wiring, tower, foundation; and sub-station, respectively), along with the SimaPro categories chosen for modeling that material. All SimaPro categories were “{GLO}| market for | Alloc Def, U,” where GLO is global, with the exception of aluminum, primary, ingot, which was EU27 instead of GLO for all parts except the substation. Since the substation was constructed in Texas, rest-of-world (ROW) aluminum was used rather than European. Since the substation serves all 200 turbines at the wind farm, the mass values shown in Table 6 were divided by 200, to determine the impact of the substation per turbine.

Table 6 Substation components of Gamesa G83 and G87 turbines (Gamesa 2013)
Full size table

Energy used for manufacturing is provided in Sect. 2.2.6. Detailed processes (e.g., heating a material to a certain temperature, and then extruding and cooling it) were not modeled individually.

Data for transportation phase

Transportation of the materials and turbine components were done by sea and land (truck), both using diesel fuel, as shown in Table 7. Although several US databases were selected in SimaPro, a truck freight transport option with US emission standards unfortunately was not available. The generic “unspecified” category was chosen.

Table 7 Transportation information used in modeling
Full size table

Although several studies limited to greenhouse gas emissions and energy consumption have considered sea vessel transport (Oebels and Pacca 2013; Rajaei and Tinjum 2013), our study is unusual among comprehensive large-turbine LCAs in that it includes transport of turbine components via sea vessel. Other comprehensive LCAs only include transport of turbine components via truck within Europe (Martínez et al. 2009; Ozoemena et al. 2018, Garrett and Rønde 2013), with the exception of Garrett and Rønde (2013). Garrett and Ronde state that their study includes both truck and sea vessel transport and that their study was based on data from a number of Vestas manufacturing plants worldwide, but no specific details regarding transport are given and transportation is not broken out as a separate phase.

Data for installation phase

Table 8 compares information used for the installation phase for this study with other comprehensive wind turbine LCAs (Martínez et al. 2009; Tremeac and Meunier 2009; Garrett and Rønde 2013) and an additional study (Rajaei and Tinjum 2013). Studies that did not provide any specific information regarding the installation phase (Oebels and Pacca 2013; Ozoemena et al. 2018) are not included in the table. As shown in Table 8, this study is comparatively comprehensive in terms of the components included in the installation (foundation, substation, control building, underground cables, access roads), as well as the equipment specified (2 excavators, 2 loaders, bulldozer, grader, crane, and 3 heavy trucks). The foundation size of 450 m3 is equal to that of Ozoemena et al. (2018) and larger than the other studies. The amount of installation energy assumed in the Ozoemena et al. study (equivalent to 3.8 gallons of diesel) seems low for a 450 m3 foundation, indicating that impacts may be underestimated.

Table 8 Installation information included in various large wind turbine studies
Full size table

Data for operation and maintenance phase

The operation and maintenance phase includes inspection trips, change of oil, lubrication of gears and the generator, replacement of turbine parts, and repair of the turbines when they break down. Table 9 compares maintenance information used for this study with other comprehensive wind turbine LCAs (Martínez et al. 2009; Ozoemena et al. 2018; Garrett and Rønde 2013). SimaPro does not include a diesel passenger truck with US emission standards, so a passenger car RoW (rest of world) was used to model the inspection trips. Garrett and Rønde (2013) state that oil and replaced turbine components are included in their analysis, but details are not provided. Our study and that of Ozoemena et al. (2018) are comparatively comprehensive in treatment of the maintenance phase. It is interesting to note that the parts targeted for replacement vary among the studies listed in Table 9.

Table 9 Maintenance information included in various large wind turbine studies
Full size table

Data for the end-of-life phase

Open loop recycling (OLR) was considered in this study, because the materials comprising the components of the turbines are in most cases recycled to make other different products. This means that the impacts from the recycling were not included in the end-of-life phase, because the recycled materials are used to make a different product; the emissions are thus appropriately counted with the new product. Recycling percentages for the turbine parts were assumed based common recycling percentages of the materials and upon manufacturer recommendations.

Table 10 compares disposal percentages for turbine components assumed by this study with two other studies that provide detailed information. All three studies listed in Table 9 assume a 90% or greater recycling percent for metals, with the remainder landfilled. Two of the studies assume 90% recycling of plastics. In the study for Spain, Martínez et al. (2009) assumed that some wastes not recycled are combusted (prominent in Europe), but some are also landfilled. Interestingly, the study for France (Tremeac and Meunier 2009) assumes that all wastes not recycled are landfilled, despite the overall low percentage of wastes landfilled in Europe. This study also assumes that all wastes not recycled are landfilled, since landfilling is common in the US.

Table 10 Disposal information included in various large wind turbine studies
Full size table

In terms of studies not included in Table 10, Garrett and Rønde (2013) considered recycling of turbine parts based on industry data, but specific information was not provided. Ozoemena et al. (2018) included energy for dismantling the turbines by crane and transporting them to a disposal site, but did not consider impacts from disposal (recycling, landfilling, and/or combustion). Similarly, Ji and Chen (2016) considered energy for turbine pull-down and transport, but did not consider impacts from disposal.

Data for energy consumption

Table 11 provides energy consumption for raw materials acquisition and manufacturing the major parts of the turbine. In SimaPro, for manufacturing of the turbine parts (nacelle, rotor, etc.), which occurred in Europe, electricity, medium voltage (Europe without Switzerland), was chosen.

Table 11 Energy consumed for phases/parts of Gamesa turbines G83 and G87
Full size table

In terms of other comprehensive wind turbine LCAs, Tremeac and Meunier (2009) assumed operational energy consumption to be 15% of manufacturing and energy for decommissioning to be equal to manufacturing. Martínez et al. (2009) state that electricity information was obtained from the SimaPro database, but details are not provided. It is assumed that both Tremeac and Munier (2009) and Martínez et al. (2009) chose European power mixes, since the studies were for France and Spain, respectively. Garrett and Rønde (2013) used country-specific average energy mixes for various countries in the supply chain, but details are not provided. Ozoemena et al. (2018) based energy estimates on Chataignere and Boulch (2003): the total primary energy requirement of 379,734 MJ was split evenly between natural gas and electricity. The “electricity, medium voltage, production RER, at grid/RER U” option of the Ecoinvent database was used to represent average European electricity production.

Conversion of inventory data to functional unit

As mentioned above, the functional unit of this study is 1 kWh of the electricity that is generated and provided to the grid. The data in the inventory phase was converted into the functional unit by estimating the energy generated by the turbine over its entire life cycle using Eq. 1.

$${\text{Wind}}\;{\text{Power}}\,\left( {\text{watts}} \right) = {\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 2}}\right.\kern-0pt} \!\lower0.7ex\hbox{$2$}} \, \rho AV^{3} C_{\text{p}}$$
(1)

where ρ, air density = 0.91 kg/m3 without water vapor and 0.89 kg/m3 with water vapor; A, wind turbine rotor’s swept area = 5412.8 m2 for G83 and 5947 m2 for G87; V, wind speed (m/s); Cp, power coefficient = 0.59 (Royal Academy of Engineering 2016).

A wind rose for the Abilene Municipal Airport was used to calculate actual power production from the turbines at the Lone Star Wind Farm. Power production for a wind speed of 8 m/s was also estimated (optimal turbine performance), for three different turbine life spans, as shown in Table 12.

Table 12 Estimated energy production (GWh) for Gamesa turbines for different wind speeds and turbine life spans
Full size table

Impact assessment

The TRACI Tool for the Reduction and Assessment of Chemical and other environmental Impacts methodology developed by the United States Environmental Protection Agency (US EPA) was used in this study. Impact assessment categories included in TRACI are:

Environmental impacts

  • Global warming (kg CO2 eq)

  • Depletion of ozone (kg CFC-11 eq)

  • Tropospheric ozone (smog) formation (kg O3 eq)

  • Acidification (kg SO2 eq)

  • Eutrophication (kg N eq)

  • Ecotoxicity (CTUe)

Human health

  • Carcinogens (CTUh)

  • Non-carcinogens (CTUh)

  • Respiratory effects (kg PM2.5 eq)

Resource depletion

  • Fossil fuel depletion (MJ surplus)

  • Water depletion index (m3)

  • Cumulative energy demand (kWh)

The percentage of the inventory assigned to each turbine model was based on the mass of each type. At the Lone Star Wind Farm, there are 100 turbines of each model, with very similar total weights (1,483,937 kg for G83 and 1,484,781 kg for G87). Hence, 50% of the inventory was assigned to each model.

Interpretation

As part of the interpretation phase, a sensitivity analysis was conducted on two input variables: turbine lifespan and wind speed. The turbine life span was extended from 20 to 25 and 30 years. Extending the life span of the turbine is something that any wind farm owner or operator may consider to maximize the profits of their farm. Several factors were taken into consideration in modeling the longer life span of the turbines: additional maintenance, supplies, and the need to transport the supplies to the farm site. Additional production of energy was also considered.

In addition, two cases of the wind speed were tested. The first was using a fixed wind speed of 8 m/s. This wind speed represents the value where the turbines will perform the best (optimal performance) according to the manufacturer. Second, the wind rose which represents the actual wind speed in the area was used. The different wind speed averages were used in the equation to calculate the power production, which was used to put results in the form of the functional unit (per kWh of power generated).

Results and discussion

Impact assessment by phase of the turbine’s life cycle

Table 13 and Fig. 2 show the contribution of the various phases of the wind turbine life cycle to the TRACI impacts categories. As shown in the bottom row of Table 13, raw material acquisition/manufacturing ranks first in terms of impacts, followed by installation and then operation and maintenance. These three phases should thus be targeted in order to reduce overall life cycle impacts of wind turbines in the US.

Table 13 Environmental, human health, and resource depletion impacts by phase of wind turbine life cycle, for 2 MW Gamesa turbines in the US, per generation of 1 kWh of electricity
Full size table
Fig. 2
figure 2

Environmental, human health, and resource depletion impacts by phase of wind turbine life cycle, percent

Full size image

Table 14 compares the ranking of phases in terms of overall impact for this study with other wind turbine LCAs. Table 14 is discussed in more detail in Sects. 3.1.1 through 3.1.5, which examine each phase in turn.

Table 14 Ranking of life cycle phases in terms of overall impact, for various large wind turbine LCAs
Full size table

The most common impact category evaluated by previous studies is global warming, evaluated in CO2-eq. Surprisingly, the 4.9 g CO2-eq found in this study (not including benefits from recycling) is at the low end of the range, compared to other studies, despite the comprehensive treatment of the installation and maintenance phases; transoceanic transport of turbine parts; and use of landfilling, with higher CO2 emissions than combustion. Raadal et al. (2011) surveyed 63 LCAs conducted between 1990 and 2019 and found that GHG emissions from wind power varied from 4.6 to 55.4 g CO2-eq/kWh. The low value was for a 3 MW turbine; GHG emissions were found to decrease with increasing turbine size, reflecting economies of scale. Arvesen and Hertwich (2012) found an average of 19 g CO2-eq/kWh for 44 previous studies, ranging from small to large turbines. Garrett and Rønde (2013) conducted an LCA of the Vestas’ 2-MW GridStreamer wind turbines for a 50 MW onshore plant, using data from 20,000 wind turbines installed worldwide. They found global warming potential of 7–10 g CO2-eq./kWh.

In terms of other studies conducted for turbines in individual countries, Rajaei and Tinjum (2013) found CO2-eq. emissions of 16.9 for 1.8 MW turbines in Wisconsin, US. Tremeac and Meunier (2009) found 16 g CO2-eq/kWh for a 4.5 MW turbine in France. Oebels and Pacca (2013) found 7.1 g/kWh for a 1.5 MW wind turbine, based on the clean energy mix of Brazil (87% renewables).

Impact of raw material acquisition/manufacturing phase

As shown in Figs. 2, 3, and 4, raw material acquisition/manufacturing ranks first among the phases in terms of overall impact, also contributing the largest amount to each impact category (> 60% for all categories, and > 90% for most categories). Since the turbines were manufactured in Spain, and the European electricity mix was used. If the turbines had been manufactured in the US, the US electricity mix would have been used, and the impact of the manufacturing phase would have been even greater.

Fig. 3
figure 3

Environmental, human health, and resource depletion impacts by stage of transport, percent

Full size image
Fig. 4
figure 4

Environmental, human health, and resource depletion impacts by type of maintenance, percent

Full size image

The manufacturing phase consists of many activities which release pollutants. For example, the blade sealant contains high concentrations of chemicals such polycyclic aromatic hydrocarbons (PAHs) and acids. The tower is the largest part of the turbine and completely made of steel covered with zinc. Each part of the tower has a specific mold where the steel is heated to a very high temperature (2500 °F), which requires a large amount of energy. Burning coal or natural gas for electricity used in manufacturing generates nitrogen oxides, which are the primary contributor to eutrophication and ozone smog formation from the manufacturing phase, as well as a secondary contributor to acidification. After all pieces are cast, they are welded together; the welding process causes various pollutants to be released, including components of particulates like lead, nickel, zinc, iron oxide, copper, cadmium, fluorides, manganese, and chromium, and gases like carbon monoxide and oxides of nitrogen (Golbabaei and Khadem 2015). The global warming impact derives primarily from fossil fuel consumed during the manufacturing of the different types of steel for the tower and the nacelle and fiberglass for the rotor blades. Manufacturing was the most water-consuming phase, because of the heating and cooling processes involved, especially when casting the large parts of the turbine.

As shown in Table 14, other studies have likewise found the raw material/manufacturing phase (sometimes combined with the installation phase) to contribute the largest impacts to the turbine life cycle. Simons and Cheung (2016) also indicated that in wind turbine designs, the manufacturing phase contributes the largest impacts in terms of carbon emissions as well as cost.

Impact of the installation phase

Installation ranked second among the phases (behind manufacturing) in terms of overall impact. Its contribution to each impact category is second, with the exception of eutrophication, carcinogens, non-carcinogens (third), and ecotoxicity (fourth). The installation phase consists of burning diesel fuel in heavy construction equipment to do the installation, which contributes to the fossil fuel depletion and ozone smog formation categories. Diesel engines produce high quantities of the ozone precursor nitrogen oxides, due to their operation at high temperatures and pressures. Unlike electric power plants, which remove NOx using selective catalytic reduction or non-selective catalytic reduction controls, diesel vehicles typically do not have any NOx controls.

As shown in Table 14, Garrett and Rønde (2013) was the only other comprehensive LCA which separated installation from manufacturing. Their treatment of the installation phase was similar to ours in terms of elements included, according to Table 8. However, they found the impact of installation to be third, after manufacturing and operation. This may have been due to the size of the foundation that they analyzed being smaller than our foundation (the size was not specified). In terms of studies that analyzed greenhouse gases only, Rajaei and Tinjum (2013) also found installation to contribute the second largest share, even though their installation phase was limited to the foundation only, which was about half the size of the one included in this study. Oebels and Pacca (2013) found emissions from installation to be third, but their study provides no details about what was included in the installation phase. In summary, according to the studies listed in Table 14, the installation phase ranks second or third in importance, but there is little consistency in the phase ranking with comprehensiveness of elements included in the installation or size of the foundation.

Impact of transportation phase

Despite the inclusion of overseas transport of wind turbine parts, transportation ranked fourth among the phases in terms of overall impact. The transportation contributes its largest percent (5.9%) to ozone smog formation, due to diesel fuel creating large amounts of nitrogen oxides, an ozone precursor.

Figure 3 shows the contribution of each stage of transport (European truck, transoceanic ship, and US truck) to overall transportation impacts. The contribution from European truck transport, which ranged from 70 to 650 km, is the smallest. Surprisingly, the impact of US truck transport, which was 656 km, is in most cases comparable or even greater (ecotoxicity, non-carcinogens, and carcinogens) than that due to transoceanic ship transport, which was 8325 km. This indicates that transoceanic ship transport is much more efficient than truck transport.

As shown in Table 14, transportation ranks second, third, or fourth among the life cycle phases in terms of impacts, depending on the study. As mentioned previously, our study included overseas transport, which because of the long distance would be expected to increase the impact of the transportation phase. Tremeac and Meunier (2009) and Martínez et al. (2009) found transportation to contribute second and third to overall impact, despite the fact that overseas transport was not included. However, both of these studies lumped installation in with manufacturing, which reduced the total number of phases evaluated and thus elevated transportation in the rankings. Rajaei and Tinjum (2013), which included overseas transport of a few turbine components, also found the transportation phase to rank fourth; in addition, they found that eliminating transportation of components from overseas, via local manufacture, could reduce transportation GHG emissions by 22%. In summary, it is difficult to discern patterns in the ranking of the transportation phase based on whether overseas transport was included or not.

Impact of operation and maintenance

Despite the comprehensive treatment of maintenance in this study, operation and maintenance (O&M) ranked third among the phases (behind manufacturing and installation) in terms of overall impact. The largest contribution of maintenance is 6.8% to the non-carcinogens category, primarily due to the replacement of the control system after 10 years. As shown in Fig. 4, the largest impacts from maintenance come from replacement of the control system, followed by lubricant (which has the greatest contribution to fossil fuel depletion and ozone depletion). Transport for inspection trips contributes approximately 0%. Replacement of the brake is only substantial in its contribution to carcinogenics.

As shown in Table 14, O&M ranks second, third, or fourth among the phases in terms of overall impacts, depending on the study. Several studies in Table 14 (Ji and Chen 2016; Martínez et al. 2009; Tremeac and Meunier 2009; Ozoemena et al. 2018) lumped installation in with manufacturing, which would tend to elevate O&M in the rankings. However, among the other studies which did not lump installation in with manufacturing (Garrett and Rønde 2013; Rajaei and Tinjum 2013; Oebels and Pacca 2013), O&M ranked second, third, and fourth, respectively. These three studies did not provide detail about what was included in terms of maintenance. Differences in emissions from the electricity mix used for operation and maintenance (global, US, and Brazil) may account for differences in the rankings. Hence, discerning clear patterns in ranking of the O&M phase is difficult.

Impact of end-of-life

As shown in Table 13, recycling materials at the end of the turbine’s impacts by 52.9% to 97%, for the various impact categories. Since the manufacturing phase dominates impacts, recycling the materials used in manufacture reduces the overall life cycle impact substantially.

As shown in Table 14, in all studies, the impact of end-of-life consistently ranked last, regardless of whether the parts not recycled were combusted or landfilled. Tremeac and Meunier (2009), Martínez et al. (2009), and Garrett and Rønde (2013) similarly found impacts from the end-of-life phase to be negative across all categories because benefits of recycling the turbine parts were included.

Impact assessment by turbine part

Figure 5 below shows the impacts of each major part of the turbine, including both raw material acquisition/manufacturing and transport. Overall, manufacture of the tower has the largest impacts, contributing > 40% to all categories except fossil fuel depletion, followed by the nacelle. 95% of the greenhouse gases from manufacturing of the tower come from processing the steel, which must be heated to 2500 °F to be cast. Similarly, for a 1.5 MW turbine in France, Oebels and Pacca (2013) found the steel tower to contribute > 50% of the CO2 emissions from the manufacturing phase. Replacing the steel tower with a cement tower, with reinforcing steel, decreased CO2 emissions by 6.4% overall. Although manufacturing emissions decreased by more than 6.4%, transportation emissions increased, due to the greater weight of concrete. Garrett and Rønde (2013) found the nacelle and tower to have the largest impacts across all categories. In contrast, for a 2 MW turbine in Spain, Martínez et al. (2009) found the foundation to have the largest overall impacts, primarily due to the cement, with the rotor second, steel tower third, and nacelle fourth. Similar to Martinez, Rajaei and Tinjum (2013) found contributions to CO2 emissions, in order from greatest to smallest, to arise from the foundation, tower, rotor blades, and nacelle, respectively.

Fig. 5
figure 5

Environmental, human health, and resource depletion impacts by turbine part, percent

Full size image

Sensitivity analysis

Extension of the turbine life span

Table 15 shows the changes in impacts when the turbine life span is extended from 20 to 25 and 30 years, for the operation and maintenance phase (per turbine) and overall (per kWh energy produced). LCA studies typically use a life of 20 years, which matches the standard design life; however, in practice, lifetimes of 30 years and over have been achieved (Garrett and Rønde 2013). As expected, when the turbine lifespan was extended, impacts for operation and maintenance increased; however, this was more than outweighed by the additional energy production, so that the overall impact per kWh of electricity generated decreased. The decreases would be expected because most of the impacts are due to the manufacturing phase, and with a longer life span, the pollutants from manufacturing are distributed over more years.

Table 15 Impacts of operation and maintenance phase (per turbine) and overall impacts (per kWh) for 20-, 25-, and 30-year turbine lifetimes
Full size table

Table 16 compares the energy balance for the 20-, 25-, and 30-year life spans. Extending the life span means that more energy overall will be generated with the same devices. Energy consumption associated with manufacturing will be distributed over a longer life span, reducing the cumulative energy demand (CED) per year, and increasing the ratio of produced energy to CED. For the 20, 25, and 30-year life spans, the turbines produce 3.2, 4.2, and 5.1 times more energy than they consume over their life cycle, respectively. Garrett and Rønde (2013) found that 2 MW GridStreamer™ turbines returned 22 to 30 times the energy they consume over their life cycles. Their study, however, used much higher wind speeds (7 m/s, 8 m/s, and 9.25 m/s), compared to the wind rose speeds used in this study (average 5.3 m/s).

Table 16 Energy balance for different turbine life spans
Full size table

Variation of assumed wind speed

Table 17 shows the net energy balance per turbine for different wind speed scenarios (assuming 8 m/s fixed wind speed as a best-case scenario recommended by the manufacturer and using the annual wind rose at the wind farm site). Obviously, increasing the wind speed for the farm will increase the energy production. However, assuming that the wind speed is fixed for the whole life span and assuming that the turbines are operating 24/7 during its life span is far from reality. There are times when the turbines are not operating for different reasons such as the maintenance or malfunction in the system or even no wind at all to push the blades. In addition, the wind speed continually fluctuates. Assuming constant high wind speed represents a best-case scenario. The most realistic way is to use the wind rose for the area where the farm is operating.

Table 17 Energy balance for different wind speeds, 20-year life span
Full size table

By comparison, Martínez et al. (2009), Tremeac and Meunier (2009), and Rajaei and Tinjum (2013) found payback times of 0.4 years; 0.58 years (4.5 MW turbine) and 2.3 years (250 W turbine); and 1.03 years, respectively.

Life cycle assessment of wind turbines versus other energy sources

Table 18 compares the impacts of wind turbines with other energy sources. Widder et al. (2011) conducted an LCA of a coal-fired power plant in the US, using SimaPro software with the TRACI impact assessment method. Table 18 shows Widder et al. impact results for a 500 MW coal plant burning lignite, with and without carbon capture and sequestration (CCS), using mono ethanol amine (MEA) scrubbing for carbon dioxide removal. Table 18 also shows global warming CO2-eq. for natural gas combined cycle, biomass, and nuclear, estimated by Gagnon et al. (2002) for typical conditions in North America. Global warming CO2-eq. for PV was estimated by Zhang (2011) for thin film photovoltaic solar systems.

Table 18 Impacts of wind turbines versus other energy sources
Full size table

For all impact categories, the coal-fired power plant, even with CCS, causes substantially more environmental impacts than the wind turbines for the same amount of power produced. Water depletion index results in indicate that coal plants consume over 3000 times more water than the wind turbines. In coal plants, water is continuously used for the purposes of systems cooling, while in the case of wind turbine, water is not continuously needed; it is only used during manufacturing and installation. The use of the CCS technology lowers the global warming contribution of the coal plant, yet it is still over 12 times that of the wind turbine.

Stanek et al. (2018) found the thermo-ecological cost for wind turbines to be from 47 to 65 times lower than for coal-fired power plants and 35 to 48 times lower than for NGCC plants.

Conclusions and recommendations

In this study of life cycle environmental impacts of a large wind turbine in the US, manufacturing contributed the greatest impacts (> 60% for all impact categories), followed by installation. Even though maintenance was treated comprehensively, the operation and maintenance phase ranked third. Despite the inclusion of overseas transport, transportation ranked fourth. End-of-life ranked last, which means that the disposal method (landfilling or combustion) for turbine parts which are not recycled makes little difference in terms of the overall turbine life cycle. Future work should assess changes to manufacturing to reduce impacts, since this phase had the greatest contribution.

In studies for wind turbines in other countries with which this study was compared, manufacturing (with which installation was frequently lumped) consistently ranked first and end-of-life consistently ranked last. Installation ranked second or third, and transportation and O&M ranked second, third or fourth. Consistent patterns in rankings could not be discerned with comprehensiveness of the treatment of installation or O&M, or whether overseas transport was included in transportation. This indicates that other unspecified elements of the studies were more important in determining ranking of the phases.

Among the major parts of the turbine, manufacturing and transport of the tower produced the greatest impacts, due to steel manufacture. Replacement of the control unit produced the greatest impacts in terms of maintenance. Transport via transoceanic ship proved more efficient in terms of impacts per km travelled than truck transport.

For a 20-year lifetime, the turbines produce 3.4 times more energy than they consume. If the turbine life span is increased to 25 or 30 years, the turbines produce 4.2 and 5.1 times more energy than they consume, respectively. Extending the turbine life span lowers impacts per kWh of electricity produced because the impacts, which are due primarily to the manufacturing phase, will be distributed over a longer period of time.

The best-case wind speed recommended by the manufacturer, 8 m/s, overestimated electricity generation by a factor of 2.3 compared to using the wind rose at the farm site. Site-specific information should therefore be used in evaluating the potential for electricity production.