Life cycle assessment of wind turbine blade recycling approaches in the United States

Most wind turbine blades reaching end-of-life are sent to landfill where embedded cost, energy, and materials are lost. To avoid landfilling future blades, a broad range of recycling and material recovery approaches have been proposed as solutions in the U.S., each with benefits, challenges, and varying levels of technical maturity. The approaches include 1) cement co-processing, 2) mechanical recycling, 3) pyrolysis, 4) microwave pyrolysis and 5) solvolysis. While these approaches are all capable of recovering various forms of materials for use in secondary markets, there are trade-offs between material circularity, reducing harmful environmental emissions, and cost-effectiveness for the U.S. market. Life cycle assessment (LCA) is a critical step needed to compare these trade-offs and determine where future research and development should be focused. As a result, some previous LCA has been performed on recycling approaches. However, attempts to quantify and compare greenhouse gas emissions across a broad range of technologies have been limited, particularly within the U.S. market where landfill availability and costs do not hinder disposing of wind blades. This work addresses this limitation by presenting a detailed comparison of LCA greenhouse gas emissions and material yields from a range of wind turbine blade recycling approaches in the U.S. The LCA presented in this work includes baseline results, as well as a variety of sensitivity and scenario analyses that look at the impact of process modelling uncertainty, future energy mixes, and other critical input parameters. Overall, results show that mechanical recycling and microwave pyrolysis have the lowest net greenhouse gas emissions. However, the value of mechanically recycled materials is highly uncertain, as mechanical recycling generates a mixed feedstock that may underperform compared to virgin materials. Cement co-processing has higher net emissions than mechanical recycling or microwave pyrolysis but does generate a value-added feedstock that offsets virgin material from mining for cement production. Other advanced thermal and chemical recycling methods such as pyrolysis and solvolysis have higher net emissions due to increased energy consumption but are also highly sensitive to thermal energy sources within the model.


Page 1:
In the Abstract section, the following text appears: "Overall, results show that mechanical recycling and microwave pyrolysis have the lowest net greenhouse gas emissions.However, the value of mechanically recycled materials is highly uncertain, as mechanical recycling generates a mixed feedstock that may underperform compared to virgin materials.Cement co-processing has higher net emissions than mechanical recycling or microwave pyrolysis but does generate a value-added feedstock that offsets virgin material from mining for cement production."This should read: "Overall, results show that cement co-processing, mechanical recycling and microwave pyrolysis have the lowest net greenhouse gas emissions.However, the markets for recycled materials from mechanical recycling and microwave pyrolysis are highly uncertain.Cement co-processing does have slightly higher net emissions than mechanical recycling but generates a value-added feedstock today that offsets virgin material from mining for cement production."In the Results section, the following text appears: "Following mechanical recycling and microwave pyrolysis, the next lowest emitting approach is cement co-processing.Given that that cement co-processing is already occurring in the U.S., it seems like a sensible near-term solution for avoiding wind turbine blades in landfills.However, by looking at the results from this analysis, some limitations appear with respect to minimizing future greenhouse gas emissions.This should read: "In addition to mechanical recycling and microwave pyrolysis, another low emitting approach is cement co-processing.Given that cement co-processing is already occurring in the U.S., it seems like a sensible near-term solution for avoiding wind turbine blades in landfills.However, by looking at the results from this analysis, some limitations appear with respect to the long-term benefits of cement co-processing."

Introduction
Wind turbine blades are constructed primarily using fiber reinforced polymer composites due to their light weight and durability, allowing blades to survive 20-30 years in extreme conditions and variable IOP Publishing doi:10.1088/1757-899X/1293/1/012027 2 climates.However, the material properties that provide these advantages also present challenges when blades reach the end of their service life.The primary challenge is that thermoset polymers are difficult to separate from fiber reinforcement.As a result, most blades currently reaching end-of-life are sent to landfill where the embedded energy and costs of those blades are left unrecovered [1].Recognizing this issue, researchers and industry have started to propose a range of approaches for recycling wind turbine blades [2].While many of these options can recover materials or energy from the wind turbine blades, each approach has a unique set of economic and environmental trade-offs.Evaluating these trade-offs at an early stage of development is critical to ensure that solutions are optimized to increase the overall sustainability and circularity of wind turbine blades.
Addressing the trade-offs of different wind turbine blade recycling approaches requires quantifying a range of metrics.One set of metrics that are especially important are the emissions released to the environment during recycling.Quantifying these emissions requires performing a detailed life cycle assessment (LCA) of each recycling approach.Previous efforts have performed LCA on select recycling approaches specifically for wind turbine blades [3][4][5].Other efforts have performed LCA in the more general context of recycling fiber reinforced polymers [6][7][8].However, there is still a research gap where additional detailed LCA is needed across multiple recycling approaches.Furthermore, the majority of LCA that has been performed is focused on Europe, where current and future laws regarding landfilling are accelerating the need for near-term solutions.The large amount available landfill space and relatively low cost associated with landfilling in the U.S. poses a unique challenge and has specific implications for performing LCA on blade recycling in the U.S.
This paper highlights an ongoing effort to perform a U.S. specific LCA of multiple wind turbine blade recycling approaches.This effort has consisted of three major phases to date.First, a broad literature review was conducted to gather qualitative and quantitative data about existing recycling approaches [9].Second, data from the literature review and discussion with industry experts were used to develop detailed process models which track all materials and energy required by each recycling approach.Third, the process modelling data were combined with life cycle inventory (LCI) databases to perform a baseline gate-to-gate LCA of cement co-processing, mechanical recycling, pyrolysis, microwave pyrolysis, and solvolysis, as well as landfilling for comparative purposes.Following this baseline assessment, a series of scenario analyses were performed to quantify uncertainty and identify the impact of different energy mixes on greenhouse gas emissions.

Process Modelling
Process models for each recycling approach were developed to track energy and mass entering and leaving the system boundary.Process modelling data were sourced from published literature, as well as discussions with industry and fellow researchers regarding the current standard practices within the U.S. [9].Using this information, the baseline scenario was developed around a land-based turbine being decommissioned in Texas, U.S with the properties shown in Table 1.Each wind turbine blade is disassembled, downsized on-site, and transported using the parameters shared in Table 2.Although not explicitly defined in the modelling, any metal components are assumed to be removed from the blade during the cutting and shredding steps to avoid contamination in downstream recycling processes.
The packing density of blade segments or shredded materials greatly impacts the efficiency of transportation.The current practices around downsizing and transportation logistics vary widely across industry, leading to a range of possible outcomes for packing and transportation efficiency [9].Based on this range, a middle estimate of packing efficiency was selected to represent a conservative improvement of on-site shredding, while avoiding an overly optimistic result.In the baseline, shredded material fills 100% of a standard truck's payload.Meanwhile, blade segments are assumed to have a packing density of 50% due to volumetric constraints.In addition to packing density considerations, the transportation distance has a significant impact on the process modelling.Since few commercial scale options for recycling blades exist in the U.S., estimates for varying transport distances were selected to capture the potential differences across approaches, while maintaining an overall conservative approach.After transportation, the blade materials arrive at a landfill, cement co-processing facility, or recycling facility.Select process modelling parameters associated with each of the approaches are displayed in Table 3.The two simplest process models developed for this stage were landfilling and mechanical recycling.In landfilling all blade material is treated as mixed plastic waste at a sanitary landfill.For mechanical recycling there are a broad range of potential options including downsizing for reuse as aggregate in construction materials or simple grinding to recover resin rich fibers and powders.In this process model, simple grinding was selected to represent mechanical recycling due to the higher likelihood of market adoption.
In contrast to landfilling and mechanical recycling, cement co-processing represents the most complex process model developed.In the model, a wind turbine blade is assumed to be further downsized, then mixed into a cement kiln where the resin is burned as fuel and the remaining constituent materials from fiberglass are used as feedstock for cement clinker.The key materials in the blade required for clinker are CaO, SiO2, Al2O3, and Fe2O2.These materials must be balanced with typical cement clinker feedstock materials such as limestone, sand, clay, and ore.In addition to these mass balance considerations, the kiln requires a significant amount of thermal energy that comes from different fossil fuels such as coal, petroleum coke, and natural gas, as well alternative fuels such as the organic materials of wind turbine blades considered in this analysis [10].Pyrolysis is also a relatively complex process, especially considering the range of potential operating conditions and corresponding mix of output products.As a result, the baseline process model was developed using data from multiple publications [2,5,11,12], as well as feedback from industry experts.In pyrolysis, downsized composite material enters a reactor where pressure and temperature are increased to thermally remove the epoxy from the fiber.Recovered fiber represents the primary output product of the pyrolysis system, while co-products of the system include the oils and non-fiber solids.Gases generated through pyrolysis are assumed to be recirculated and combusted within the system to offset a portion of the thermal energy demand.Given the energy demands of pyrolysis, several alternative approaches have been proposed to reduce the energy demand to recover fiber.One promising approach is microwave pyrolysis, which uses microwave radiation in place of traditional thermal energy sources.The microwave pyrolysis model was primarily based upon reports from a previous project [13,14].Since all energy provided to the microwave pyrolysis is electrical, any oil and gas are treated as coproducts that would be sold to external customers.
When modeling solvolysis there are a broad range of options relating to the type of solvents, temperature, pressure, and other operating parameters.In this work the solvolysis model was developed based upon a previous publication with experimental results [15].In this model a combination of acetone and water are used as solvent.The solvent is raised to an elevated temperature via natural gas heating to expedite dissolution.At the end of the process solvent is recovered and recirculated through distillation also heated by natural gas, which corresponds to a full scale, continuous process.After dissolution, the fiber is recovered while the resin and a small fraction of spent solvent are removed from the system and combusted to produce thermal energy that offsets a portion of natural gas demand.

Life cycle Assessment and Scenario Analyses
LCA was performed in accordance with guidance from ISO14040 and 14044 [16,17].Based on this guidance the four standard phases of LCA are 1) goal and scope definition, 2) LCI analysis, 3) life cycle impact assessment (LCIA), and 4) interpretation and decision making.Within this specific LCA the current goal is to compare the greenhouse gas emissions of multiple wind turbine blade recycling approaches.In the future, additional environmental impacts such as human toxicity, ecotoxicity, particulate matter, and water consumption should be considered to obtain a full set of environmental trade-offs.The system boundary is held consistent across all recycling approaches and is considered "gate-to-gate".It begins when a blade reaches end-of-life on a wind turbine and ends when the blade has been transformed into a new material intended for remanufacturing.To enable a reasonable scope for this paper, life cycle stages outside of the system boundary such as blade manufacturing and remanufacturing of recycled materials have been excluded from the LCA.The primary functional unit considered in this analysis is one kilogram of recycled material leaving the system boundary.However, since landfilling does not result in any recycled material, comparisons with landfill are shown on a per blade basis.Primary LCI data were sourced from the process modelling.Secondary LCI data were sourced from Ecoinvent 3.9.1 [18].When possible, Ecoinvent LCI data were selected based upon specific geographic regions within the U.S. If these data were unavailable, "rest-of-world" or "global" options were selected to represent a best possible approximation.Within Ecoinvent only "cut-off unit process" data were selected to avoid any duplicate accounting for recycling processes.
To account for the benefit of recycled materials, substitution credits were taken for any mass of cement clinker, fiberglass, powders, oils, and combustible gases produced during recycling.Table 4 displays the credit taken for one kilogram of each material.The amount of credit given to a recovered material was based upon the properties of the recovered material compared to the properties of a similar virgin material.Cement clinker was assumed to meet all requirements for standard quality clinker and receives a credit equal to the emissions of producing and equivalent amount of standard cement clinker.Recovered fiberglass was given a credit based upon the emissions of producing virgin fiberglass.However, that credit was weighted by the estimated tensile strength of recovered fiberglass compared to the tensile strength of virgin fiberglass.Recovered oil or gas were given credit based on standard production of light fuel oil and natural gas.These credits were assumed proportional to their higher heating value.Recovered powder was assumed to have properties equivalent to mortar used in construction processes and therefore received a credit equal to the emissions of producing virgin mortar.
Once all LCI data were collected, LCIA was performed using climate change emissions factors from the U.S. Environmental Protection Agency's Tool for Reduction and Assessment of Chemicals and Other Environmental Impacts (TRACI v2.1) [19].As mentioned previously, the scope of this paper was limited to assessing the greenhouse gas emissions (kg CO2eq).These results were compared across a range of scenarios to understand the impact of different process modelling assumptions and parameters.The first two scenarios considered were an optimistic and conservative scenario.These scenarios were 6 developed to capture the simultaneous effect of adjusting all process model input parameters to their best or worst value.While these scenarios are unlikely to represent real world outcomes, they are useful in defining likely upper and lower bounds of uncertainty for each technology.The next two scenarios considered were based around lower carbon energy sources for both electricity and thermal energy.The first scenario looks at the potential emissions reductions based on electricity being supplied by a projection of the 2035 U.S. electrical grid mix [20].The second scenario considers this same electrical grid mix, but also analyzes the impact of introducing a low emissions fuel source such as green hydrogen or biomethane for all thermal energy demand.Based on data from Ecoinvent, this fuel source is estimated to have emissions of 0.015 kg CO2eq per MJ of heat delivered, which is also in line with optimistic estimates for green hydrogen production via renewable energy [21].

Results
Figure 1 displays the blade mass yields for each approach considered.All five recycling approaches lose 5-6% of material due to system inefficiencies.These include dust lost to the environment during downsizing and the metal that is removed from the blade before recycling.Four of the approaches include incineration of organic material within the system boundary.This includes initial incineration of the resin when it is heated, as well as recirculation of certain oils or gases for secondary combustion to provide thermal energy back into the process.In addition to this incineration, pyrolysis and microwave pyrolysis also result in an export of some oil or gas, which is not utilized within the system boundary.The remaining solid recycled materials are broken into two categories.One is the primary solid recycled material, which consists of either recovered fiber or cement clinker.The other category is secondary solid material, which consists of powders resulting from the recycling processes.The baseline results for greenhouse gas emissions of recycling approaches and landfilling are displayed in Figure 2.These results show the positive greenhouse gas emissions from decommissioning, size reduction, transportation, and recycling or disposal.They also show an estimate for credits due to substitution of recycled materials for cement clinker, virgin fiberglass, mortars, light fuel oil, and natural gas.The sum of the positive emissions and the negative credit represents the net greenhouse gas emissions of each recycling approach.To enable comparison of net emissions to landfill, the results are shown per one wind turbine blade entering the system boundary.The net greenhouse gas emissions of mechanical recycling and microwave pyrolysis are negative, meaning the credit taken for producing recycled materials is greater than the emissions of the recycling process itself.All other processes result in net positive greenhouse gas emissions, meaning that the emissions of recycling a wind turbine blade are greater than the credit for recycled material.In four out of six approaches the greenhouse gas emissions of decommissioning, on-site size reduction, and transportation are small compared to the recycling process.Across the approaches there is a small but notable difference in transportation emissions due to truck packing efficiency and transportation distance.
Net greenhouse gas emissions results from the optimistic and conservative scenarios are shown in Figure 3, with a functional unit of one kilogram of solid recycled material.These results show the total cumulative uncertainty of net greenhouse gas emissions for each recycling approach, and inherently account for the mass yield of solid recycled materials.Within the figure, all approaches show a wide range of uncertainty due to the breadth of different approaches available in literature.While the general trends across recycling approaches remain similar between baseline, optimistic, and conservative scenarios, some findings do change dependent upon the scenario considered.For example, microwave pyrolysis shows a net-negative emissions result in the baseline and optimistic scenario, but net-positive in the conservative scenario.Based on the wide uncertainty bands, a sensitivity analysis was also performed to identify which individual input parameters are most sensitive to uncertainty.Figure 4 displays the top six most sensitive parameters to uncertainty for pyrolysis and solvolysis.In both approaches, energy consumption and fiber yield are highly sensitive to uncertainty.Figure 5 shows the impact of a 2035 electrical grid mix that is mostly decarbonized and low emission fuel sources.The 2035 electrical grid reduces greenhouse gas emission from 1%-23% across different recycling approaches.The largest impacts are on mechanical recycling and microwave pyrolysis which see reductions of 12% and 23%, respectively.Simultaneously adding a low emissions fuel for thermal energy results in total reduction of emissions from 8%-50% across all recycling approaches.The largest reductions are seen in solvolysis and pyrolysis which are reduced by 40% and 50%, respectively.Across all approaches the decommissioning, on-site size reduction, and transportation steps are estimated to operate via standard diesel power and are not impacted by the 2035 grid or low-emissions fuel scenarios.

Discussion
Recovery of wind turbine blade materials at their end of life is currently challenged by increased cost and emissions relative to landfilling in the U.S.However, there are recycling approaches that show promise as a responsible alternative to disposal.For example, this paper has identified promising results for mechanical recycling in terms of both circularity and greenhouse gas emissions in the baseline scenario.This is largely due to modelling a simplified grinding process, which does not incinerate any material, and requires a relatively small amount of electrical energy to recover fibers and powder.However, it is important to note that the resin rich fiber recovered through this process will likely have significantly degraded properties compared to virgin fiberglass.While the credit taken for this recovered fiber does account for a 50% reduction in tensile strength, it does not consider the market demand for this reduced strength material.As a result, more work is needed to research potential markets for mechanically recovered fiber and assess actual value beyond a ratio of tensile strength.Microwave pyrolysis also shows promise in terms of low net greenhouse gas emissions.In this case, the microwave pyrolysis benefits from both the efficiency of the microwave pyrolysis process, as well as the source of energy being electricity.Not only does this result in low baseline emissions, but it also opens the potential to see significant emissions reductions with future U.S. electrical grid mixes.The current limitation of microwave pyrolysis is that it is an early-stage technology for recycling wind turbine blades at scale.While there have been pilot scale efforts outside the U.S., most research in the U.S. is still at laboratory scale, meaning unforeseen challenges may arise in the future and commercialization is not near-term.Following mechanical recycling and microwave pyrolysis, the next lowest emitting approach is cement co-processing.Given that that cement co-processing is already occurring in the U.S., it seems like a sensible near-term solution for avoiding wind turbine blades in landfills.However, by looking at the results from this analysis, some limitations appear with respect to minimizing future greenhouse gas emissions.First, cement co-processing burns the epoxy resin of the wind turbine blade for energy.While this process does capture embedded energy that would otherwise be lost to landfill, it is also a direct source of CO2 emissions.Second, cement co-processing of wind turbine blades currently benefits from avoiding the direct carbon emissions associated with calcination of standard feedstock materials.Given the current focus on decarbonizing U.S. cement production emissions with carbon capture or alternative feedstocks [10,22], the future benefit of co-processing wind turbine blades may be reduced.
Pyrolysis and solvolysis have the largest net greenhouse gas emissions in the baseline case.However, they also have significant uncertainties and the greatest potential for reduction if low emissions fuel sources can be implemented.Ongoing research on pyrolysis of wind turbine blades in the U.S. is likely to further optimize processing and decrease the overall greenhouse gas emissions.Efficient use of oil and gas byproducts from pyrolysis will be especially important to this reduction.One other major consideration for pyrolysis is the value of the recovered fibers.Research on cleaning or recovering the properties of fibers after pyrolysis will be critical to enabling this approach.Currently, this step is often the most proprietary part of the pyrolysis process and gaining additional data for future modelling may prove challenging.Like pyrolysis, solvolysis can also benefit from further research, however the added complexity of using harsh solvents at elevated temperatures is a challenge for use with existing thermoset resins.Future thermoplastic or reversible thermoset materials creates a greater opportunity for effective use of less abrasive solvents at lower temperatures.
It is important to note that the scope of this analysis was limited to the fiberglass composites that will make up most near-term blade waste in the U.S. Therefore, it does not explicitly consider the greenhouse gas emissions associated with processing other materials, most notably carbon fiber.If carbon fiber was considered it would have significant impacts on the overall results.Specifically, the credit received for recovered fiber would be higher, as the greenhouse gas emissions of producing virgin carbon fiber are much higher than that of virgin fiberglass.As a result, pyrolysis, microwave pyrolysis, and solvolysis would all see reduced net greenhouse gas emissions when processing carbon fiber due to the increased credit.In contrast to these approaches, the presence of carbon fiber in materials destined for cement coprocessing is unacceptable, as it cannot be processed in most cement kilns.As a result, any increase of carbon fiber content in future wind turbine blades will pose an added challenge for cement coprocessing.With mechanical recycling, carbon fiber would have a mixed impact on greenhouse gas emissions.Energy loads and the time for size reduction would likely increase due to carbon fiber's increased strength.At the same time, the approach may benefit from a higher value of mixed-fiber recovered material.For mechanical recycling, as well as the other approaches, understanding the net impact of carbon fiber on greenhouse gas emissions requires its own dedicated analysis.
The overall results of this paper indicate that there are multiple approaches for wind turbine blade recycling that could result in net greenhouse gas emissions similar to or below that of landfilling.The results also show that there are many ways to reduce the net greenhouse gas emissions of these approaches, including cleaner electricity grid mixes and low-emission fuels for thermal energy.While this analysis helps enable a comparison of different recycling approaches, there are remaining research and market challenges to be addressed.First, the value of recovered products is critical to this analysis and requires more detailed investigation for representative secondary markets to increase certainty.This includes understanding potential U.S. markets for recycled products, as well as the physical characteristics that are important in those markets.Second, many of these approaches are at a low technology readiness level and have only been demonstrated at laboratory or pilot scale.Optimizing these approaches while scaling them up will be critical for their success.Last, greenhouse gases are just one of many important metrics that must be considered when assessing the trade-offs of different recycling approaches.Ultimately, more analysis is needed to quantify and compare the costs, circularity, and other environmental impacts of each approach to ensure future sustainability.This paper describes objective technical results and analysis.Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes.

Figure 1 .
Figure 1.Wind turbine blade mass yields through different approaches.

Figure 2 .
Figure 2. Greenhouse gas emissions of recycling approaches compared with landfilling.

Figure 3 .
Figure 3. Optimistic and conservative scenario net greenhouse gas emissions compared with the baseline results.

Figure 4 .
Figure 4. Sensitivity analysis of adjusting individual input parameters by 10%.

9 Figure 5 .
Figure 5. Greenhouse gas emissions for different recycling approaches with a 2035 grid mix and lowemission fuel source for thermal energy.

Table 1 .
Wind turbine blade properties used in analysis.

Table 2 .
Key parameters for disassembly, on-site size reduction, and transportation of blade.

Table 3 .
Key parameters used in process modelling of recycling approaches.

Table 4 .
Credit given for production of one kilogram of clinker, fiber, filler, oil, and gas.