Content uploaded by Michael Hoefer
Author content
All content in this area was uploaded by Michael Hoefer on Dec 20, 2016
Content may be subject to copyright.
Wind Turbine Blade Recycling: An
Economic Decision Framework
Michael J. Hoefer
IE 503 Final Report
5/1/2015
Introduction
The objective of this project is to explore the economics behind wind turbine blade
recycling. The project will determine the current cost of disposing wind turbine blades, and
investigate the cost of pursuing other end-of-life alternatives. The goal of the project is to model
the economic factors that cause a wind farm owner to dispose of blades in a certain way.
Examples of economic factors would include cost of recycling equipment, distance from wind
farm to equipment, selling price of recycled output, etc. The project will involve studying the
cost of existing methods of wind blade decommissioning, as well as proposing costs for
recycling methods.
Background and Problem Statement
Wind energy is growing in the United States. Wind energy experienced 28% growth in
2012 and accounts for 36.5% of all new energy capacity installation in the US from 2007-2012
(AWEA Market Update). There are currently over 48,000 utility-scale wind turbines installed in
the United States (AWEA Facts), and this number has been growing (see Figure 1 below).
Figure 1. Incremental and cumulative growth of the wind energy industry in gigawatts.
Source: Mooney, 2015.
Each turbine has three wind blades, resulting in over 144,000 wind blades currently
installed in the United States. Wind turbine blades weigh about 7 tons each (Harnessing the Gulf
Stream) and are made of glass fibers embedded in an epoxy matrix. Balsa wood or foam is used
as a core material in the blade shell. Wind blades are a critical component of turbines with a
lifespan of 20-25 years (Larsen, 2009). Currently in the United States, the only option for wind
farm owners at blade end-of-life is to landfill (Figure 2).
Figure 2. Wind blades are often landfilled. Source: Rahnama, 2011.
As the volume of blades continues to grow, this poses an environmental risk including
pollution of land and water. While much of the turbine is recyclable, wind blades are the
component that pose the highest environmental risk and have the highest uncertainty on how to
handle, as shown below in Figure 3.
Figure 3. Chart of uncertainty and damaging effects of wind turbine components, based
on a survey of experts during a workshop. Source: Borup, Mads.
Plastics, such as wind turbine blades, are not easily decomposed and leach harmful
chemicals into the ground when landfilled (Knoblauch, 2009). This can cause negative health
effects on humans and animals. Rahnama mentions that the high heat capacity and recycling
potential of wind blades makes landfilling unwise (2011). A flowchart mapping the current state
of wind blades at the end of life is shown below. One item of significance is the second box on
the flowchart in Figure 4. Currently, wind farm owners must expend money and energy to crush
up the wind blades for transportation to the landfill (Murchison). An ideal future state would
allow for owners to spend this energy into created useful product, rather than toxic landfill waste.
Figure 4. Flowchart of the current state for processing wind blades at the end of life.
Source: Drafted in Microsoft Visio
An economically feasible technology for recycling wind blades has not been
implemented in the United States. Recycling composites is difficult, especially those with a
thermoset matrix, such as wind turbine blades. Thermoplastic composites can be melted down
and reused (Kasper, 2008), whereas thermoset composites will simply burn. In addition,
composite materials do not have isotropic properties. The variety of the length, strength, and
alignment of fibers within the polymer matrix make recycling difficult.
Despite the challenges of recycling wind turbine blades, multiple technologies have been
developed in an attempt to feasibly recycling wind blades. Potential technologies include
pyrolysis, chemical recycling, incineration, and mechanical recycling. Cherrington, et al.
provides more detail on the potential recycling solutions for large scale wind turbine blades. For
the purposes of this project, the specific process and technology is not studied. Instead, the focus
of this paper is on the cost of recycling and the factors that cause a wind farm owner to decide to
landfill or recycle a blade. For simplicity, this paper assumes a mechanical recycling system is
capable of handling the blade recycling with a one machine process.
The goal of this project is to provide the information necessary for a future state where
wind farm owners have the decision to recycle, as shown below in Figure 5.
Figure 5. Flowchart of a potential future state. Wind farm owners will make decisions on
remanufacturing and recycling. Source: Drafted in Microsoft Visio
This future state represent a significant improvement over the current state (Figure 4), as
wind farm owners have more options and can select the economically optimal path. Ideally wind
farm owners would be able to evaluate their blades and determine if remanufacturing is feasible.
If remanufacturing is not an option, then the next decision is to recycle. Assuming no
government intervention, the wind farm owners would only recycle if it were more cost effective
than putting blades in the landfill. The following section proposes a methodology for modeling
the decision to recycle or landfill a blade at end of life.
Methodology and Results
The first step in the project was to map out the current state and future state for wind
blades at end of life. These flowcharts were introduced earlier in Figure 4 and Figure 5. Cost
estimates of current disposal methods were gathered from decommissioning reports. Costs of a
future state recycling system were based on operating costs of large scale grinding machinery
used in other industries. Once the data had been gathered, two different mathematical models
were developed in Lingo to model the decision to recycle or landfill. Each of these models is
discussed in detail in the following sections. Finally, sensitivity analysis was conducted by
running various scenarios through the Lingo model using Microsoft Excel OLE embedded
models and VBA scripts.
Mathematical Models
Two models were investigated in this project. The first is a simple model that assumes the
decision to recycle is binary, and the percentage recovered is a parameter of the model. The
second model treats the decision to recycle as a continuous option between 0 and 100%. Wind
farm owners can either landfill the entire blade, or recycle the blade to get anywhere from 0 to
100% of the product out of the recycling process. The framework used in creating the model was
based solely on the economics faced by the wind farm owner. For this approach, the focus was
not on the intangible (or tangible, but difficult to measure) benefits to society. This approach
focuses on the cost to wind farm owners.
Assumptions and Constraints
●Assume the decision to recycle is made by one wind farm owner with one wind farm.
●Wind farm owners can choose to recycle (and recover, and sell) a certain percentage of
material from the blades. Model One treats this percentage as a parameter. Model Two
treats this as a decision variable.
●The relationship between percent recovered and recycling cost is cubic, with a base cost.
●The wind farm owner does not need to bear the cost of transporting the recycled material
from the recycling facility to the consumer.
●The fixed cost for the recycling setup includes transportation of equipment to the wind
farm.
●A two-phase mechanical recycling process is an acceptable estimate for the costs of
recycling wind blades.
●Selling price of the recyclate is constant, and the quantity sold is small enough that price
is not affected.
●Transportation cost will be applied to both landfilled waste and recycled waste, and is
therefore not included in either model.
●Rent of the recycling equipment is included in the operating costs.
●The maximum recycling cost is $300/ton.
●The global solver option is initiated in Lingo.
Model Parameters
Table 1 below shows the different inputs to both models, as well as the estimated value based on
decommissioning reports.
Table 1. Model parameters and estimated values for both Model One and Model Two.
Lingo input name
Equation
Meaning
Initial input
Source
(applicable model)
Notation
value
BLADEWEIGHT (1,2)
Bw
Weight of a single wind
turbine blade, in tons
7 tons
Harnessing
the Gulf
Stream
WASTECOST (1,2)
W
Cost per ton to
pre-process blade
material before landfill
$25/ton
Hassan,
2012
NUMTURBINES (1,2)
Bn
The number of turbines
in the wind farm
28
Hassan,
2012
SELLPRICE (1,2)
P
Selling price per ton of
recyclate
$10/ton
Half the
price of
crushed
rock, Braen,
2013
GOVTSUB
G
Government subsidy
given to wind farm
owners per ton of blade
material recovered
$0 (different
options
explored)
Used in
sensitivity
analysis
LANDFILLTAX (1,2)
L
Cost per ton of putting
wind blades in landfill
$71/ton
Murchison
RECYCLE_YIELD (1)
Y
Percent by weight of the
blade that can be sold to
the post-consumer
0.30
Larsen,
2009
RECYCLE_COST (1)
Cr
Variable cost per ton to
put blade material
through recycling
process
$60/ton
Based off
estimates of
Vermeer
equipment
(See
Appendix A
for
calculations)
RECYCLE_BASE (2)
Cb
Fixed cost per ton for
the recycling process,
representing the
minimum cost of
recycling a ton of wind
turbine blades in Model
$60/ton
Based off
estimates of
Vermeer
equipment
(See
Appendix A
2
for
calculations)
RECYCLECOST_VAR
(2)
Cv
Cost parameter that
scales cubically with the
recyclate yield percent
in Model 2
$300
Assumed to
be $300
Model Decision Variables
Table 2. Decision variables for Model One and Model Two.
Name (model)
Equation
Notation
Meaning
Type
RECYCLE (1)
R1
0 if the wind blades are
landfilled, 1 if the wind blades
are recycled
Binary
RECYCLE (2)
R2
Percent yield from the recycling
process. 0 if no recycling is
attempted, 1 if 100% of the
blade material is recovered
Continuous,
constrained to
0 ≤ RECYCLE ≤ 1
Model Outputs (for both models)
Table 3. Lingo output objects for Model One and Model Two.
Name
Equation Notation
Meaning
TOTALWEIGHT
Bt
Total weight of blades in wind farm
TOTLANDFILLCOST
TL
The total cost spent on landfilling blade material
TOTRECYCLECOST
TR
The total cost spent on recycling blade material.
This includes any revenue from selling recyclate
or from government subsidy, and can be
negative
COST_TOT
TT
The total cost of decommissioning the blades on
the wind farm
COST_PER_BLADE
Tb
COST_TOT divided by the number of blades
decommissioned
Model One Equations and Explanations
B
t
= 3B
W
B
n
Total blade weight equals the number of turbines, multiplied by three turbines per blade,
multiplied by the weight of each blade.
R
1
= Binary
Model One assumes the decision to recycle is binary.
T
R
= R
1
B
t
C
r
- R
1
B
t
Y(P + G)
Total recycling cost equals the recycling cost per ton times the total weight, minus any revenue
from selling the recyclate yield (including any government subsidy).
T
L
= R
1
B
t
L(1-Y) + |R
1
- 1|WLB
t
Total landfilling cost equals the amount of material that cannot be recycled times the landfill tax.
In the case where R1 = 0, the total weight multiplied by the pre-landfill processing cost (W) and
the landfill tax.
MIN T
T
= T
R
+ T
L
The objective is to minimize the total cost, which is the sum of the landfill cost and recycling
cost.
T
b
= T
T
/ (3B
n
)
The cost per blade is the total cost divided by the number of blades
The commented Lingo code and command script for Model One can be found in Appendix B.
Model Two Equations and Explanations
B
t
= 3B
W
B
n
Total blade weight equals the number of turbines, multiplied by three turbines per blade,
multiplied by the weight of each blade.
R
2
∈
(0,1)
Model Two assumes the decision variable, recycling yield, is continuous from 0% recyclate
obtained to 100% recyclate obtained.
T
R
= (
⌈
R
2
⌉
C
b
+ C
V
R
2
3
)B
t
- R
2
B
t
(P + G)
Total recycling cost equals the base recycling cost per ton (if R2 > 0) plus the variable cost that
scales cubically with the yield, minus any revenue from selling the recyclate yield (including any
government subsidy).
T
L
= B
t
L(1-R
2
) + |
⌈
R
2
⌉
- 1|WB
t
Total landfilling cost equals the amount of material is not recycling multiplied by the landfill tax.
In the case where R2 = 0, the total weight multiplied by the pre-landfill processing cost (W) is
added.
MIN T
T
= T
R
+ T
L
The objective is to minimize the total cost, which is the sum of the landfill cost and recycling
cost.
T
b
= T
T
/ (3B
n
)
The cost per blade is the total cost divided by the number of blades
The commented Lingo code and command script for Model Two can be found in Appendix C.
Model Results and Analysis
The Lingo code was integrated with Microsoft Excel using OLE (Object Linking and
Embedding), named ranges, and a VBA script (Visual Basic for Applications). This allowed the
model to be automatically ran on various parameter data in the Excel sheet. See Appendix D and
E for the VBA scripts used to automate running the Lingo models in Excel. Analysis was
completed on 30 runs of the model for each parameter analyzed, with all other variables held
constant.
Model One Analysis
Solving the model with the estimated parameters results in the wind farm owner choosing
not to recycle, with a cost per blade of $672. Given the linear nature of Model One, the
parameters have a linear effect on the cost per blade. For example, when the yield from recycling
is increased there is a threshold value where recycling becomes more economical than landfilling
the entire blade (Figure 6).
Figure 6. The total cost per blade broken out by landfill and recycling cost, as recycle yield
increases.
The effect of recycle yield on cost per blade is due to multiple reasons. First, there is less
material going to the landfill, so there is a lower overall landfill tax. Secondly, a higher yield
means there is more recyclate to sell for revenue. As the recycle yield goes towards 100%, the
landfill cost approaches zero and the total cost approaches the recycling cost.
Another example involves the use of a government subsidy. Figure 7 below shows the
effect of government subsidy on the cost per blade.
Figure 7. The effect of an increasing government subsidy on cost per blade.
The wind farm owner makes the decision to recycle once the government subsidy reaches
$36/ton. Additional subsidy past this threshold value results in cheaper recycling cost per blade.
The landfill costs remain the same, as the same amount of material is going to the landfill. This
indicates that in a situation where the recycle yield is constant, additional government subsidy
past the threshold value will not help divert additional waste from the landfill.
Similar analysis can be conducted for other parameters. However, it is more interesting to
examine what happens when firms control the yield they get from recycling, rather than simply
deciding to recycle or not. This requires analysis of Model Two.
Model Two Analysis
We start by looking at the effect of government subsidy on the recycling yield. The graph
below (Figure 8) was generated from the Lingo model ran 70 times as government subsidy
changes, holding all other parameters constant.
Figure 8. The effect of an increasing government subsidy on recycle yield
As is the case with Model One, there is still a threshold subsidy value where it becomes
more economical to recycle at least some of the wind blade. In this case, that value is a per ton
subsidy of about $55. This represents the price point where it is more economical to run the
entire wind blade through the recycling equipment, rather than the pre-landfill processing
equipment. Additional increases in government subsidy past $55 result in a significant increase
in recycling revenue for firms, and a slowing increase in recycling yield (see Figure 9 below).
Figure 9. Diminishing marginal returns are shown. As government subsidy increases linearly,
the marginal increases in recycle yield diminish.
There is a significant jump at the threshold subsidy value, and then interesting behavior
in the range of $55-$100. Another graph was created to zoom in on the complex area, shown
below in Figure 10.
Figure 10. Zoomed in graph of the effect of government subsidy on the change in recycle yield.
The graph implies that the choice of a government subsidy is not straightforward, and has
complex impacts on the changes in recycling yield. This relationship is likely due to the triple
role of recycling yield (R2) on landfill costs, recycling revenue, and recycling expense.
Now, we look at the effects of the scaling parameter on the recycling yield and cost per
blade.
Figure 11. The effects of a decreasing scaling parameter on the cost per blade and recycle yield.
If the government subsidy remains at zero, improvements in technology would allow
wind blade recycling to become more feasible. Increases in technology would be best indicated
through changes in the cubic scaling parameter. Figure 11 indicates that a significant decrease in
the scaling parameter (a significant improvement in technology) would be required to make
recycling feasible. Once the threshold “technology” is reached, additional improvements to
technology continue to lower the cost per blade and increase the recycling yield until 100% yield
is reached. Once 100% yield is reached, additional improvements in technology continue to
lower the decommissioning cost per blade. In an ideal future state, the base cost of recycling, Cb,
would be the only cost of running a system with 100% recyclate yield. The scaling parameter
would be 0.
Expected Impact and Future Work
Based on the results of this project, it can be seen that the decision to recycle or landfill a
wind turbine blade is based on multiple factors. Given the situation today, significant increases in
technology or a government subsidy would be required for wind blade recycling to be
economical from a wind farm owner’s standpoint. The effects of a government subsidy need to
be carefully studied before any policy is implemented.
The Excel worksheet, Lingo models, and VBA scripts developed in this project can be
used in the future by wind farm owners and recycling firms to perform additional sensitivity
analysis and encourage the creation of additional recycling technology. A screenshot of the
worksheet is shown in Appendix F, and is available upon request at mjhoefer@iastate.edu.
Future research is needed to obtain better estimates of the model parameters, and investigate
additional cost functions for recycling. Perhaps there is a more complex cost function, rather than
a base cost and cubic scaling factor.
Once an economically feasible recycling system is developed, it is apparent that the
decommissioning cost per blade will decrease (Figures 6, 7, 8 and 11). This will have secondary
effects on a wind farm owner’s decision to enter and exit the market. Min, et al. proposes a
model for the threshold operation and maintenance (O&M) cost for exit and entry, which
includes the “exit fee” as a parameter. As the exit fee decreases, the exit threshold O&M cost
decreases as well, meaning that firms will leave the market sooner and the expected life of the
wind farm decreases (Min, 2012). However, the lower exit fee also encourages firms to enter the
market, so the entry threshold O&M cost decreases. This will result in the creation of more wind
farms. Future research is needed to determine the overall environmental impact of more wind
farms staying around for a shorter period of time.
References
2008 Vermeer HG 8000 Horizontal Grinder. (n.d.). Retrieved from
http://www.apolloequipment.net/equipment-for-sale/horizontal-grinders/vermeer-hg80
00-2008-003049e
AWEA 2012 U.S. Wind Industry Market Update. (2013, May). Retrieved from
http://awea.files.cms-plus.com/FileDownloads/pdfs/AWEA%20U.S.%20Wind%20Ind
ustry%20Annual%20Market%20Update%202012_1383058080720_3.pdf
AWEA Federal Production Tax Credit for wind energy. (2013). Retrieved from
http://www.awea.org/Advocacy/Content.aspx?ItemNumber=797
AWEA Wind Energy Facts at a Glance. (2013). Retrieved from
http://www.awea.org/Resources/Content.aspx?ItemNumber=5059
Borup Mads and Andersen Per Dannemand, Recycling and removal of offshore wind turbines –
An interactive method for reduction of negative environmental effects, Risoe National
Laboratory, System Analysis Department, Technology Scenarios Programme
Braen, D. (2013, May 10). Crushed Stone vs. Pea Gravel: Prices, Sizes & Uses. Retrieved from
http://www.braenstone.com/2013/05/crushed-stone-vs-pea-gravel/
Cherrington, R., Goodship, V., Meredith, J., Wood, B., Coles, S., Vuillaume, A., . . . Kirwan, K.
(2012). Producer responsibility: Defining the incentive for recycling composite wind
turbine blades in Europe. Energy Policy,
47
, 13-21. doi:10.1016/j.enpol.2012.03.076
Harnessing the Gulf Stream. (1974). The American Biology Teacher,
36
(1), 44. Retrieved from
http://orionrenewables.com/wp-content/uploads/sites/2/2013/02/howitworks.pdf
Hassan, G. (2012). Suncor Energy Adelaide Wind Power Project. Retrieved from
http://www.suncor.com/pdf/Adelaide_Draft_Decommissioning_Plan_Report.pdf
Kaiser, M. J., & Snyder, B. (2010). Offshore Wind Energy Installation and Decommissioning
Cost Estimation in the U.S. Outer Continental Shelf. Retrieved from
http://www.fairtran.net/images/OffshoreEstimation.pdf
Kari Larsen, Recycling wind turbine blades, Renewable Energy Focus, Volume 9, Issue 7,
January–February 2009, Pages 70-73, ISSN 1755-0084,
http://dx.doi.org/10.1016/S1755-0084(09)70045-6.(http://www.sciencedirect.com/scie
nce/article/pii/S1755008409700456)
Kasper, A. (2008, February 10). Recycling composites: FAQs. Retrieved from
http://www.materialstoday.com/composite-industry/features/recycling-composites-faq
s/
Knoblauch, J. (2009, July 2). The environmental toll of plastics. Retrieved from
http://www.environmentalhealthnews.org/ehs/news/dangers-of-plastic
Min, K. J., Lou, C., & Wang, C. (2012). An Exit and Entry Study of Renewable Power
Producers: A Real Options Approach. Engineering Economist, 57(1), 55-75.
doi:10.1080/0013791X.2011.651566
Mooney, C. (2015, February 4). Report: Wind and solar energy have tripled since 2008.
Retrieved from
http://www.washingtonpost.com/news/energy-environment/wp/2015/02/04/report-win
d-and-solar-energy-have-tripled-since-2008/
Murchison, J. S. (n.d.). Bowers Wind Project: Decomissioning. Site Location of Development
Combined Application
.
Rahnama, B. (1990). Reduction of the Environmental Impact Effect of Disposing Wind Turbine
Blades. The Town Planning Review,
61
(2), 139-155. Retrieved from
http://www.diva-portal.org/smash/get/diva2:691565/FULLTEXT01.pdf
U.S. Energy Information Administration - EIA - Independent Statistics and Analysis. (n.d.).
Retrieved from http://www.eia.gov/petroleum/gasdiesel/
Appendices
Appendix A
Estimates for the recycling processing cost are based off the requirements of a Horizontal
Grinder (Model HG8000), shown below.
Horizontal grinder used to process logs. Source: Vermeer.com
Fuel requirements: 51.3 gph diesel (Source: Vermeer.com)
Cost of diesel: $2.70/gal (Source: US Energy Information Administration)
Total fuel costs per hour: $138.51
Throughput: 120 tons of asphalt shingles per hour (Source: Email from Vermeer Engineer)
Given the strength of composites, a conservative number of 30 tons/hour is used.
Labor: Assumed two operators at $20/hr for a total of $40/hr.
Maintenance cost assumed to be 5% of replacement value per year.
Value of HG8000, from Apollo Equipment, is about $500,000.
5% of $500,000 is $25,000 annual maintenance costs.
Assuming 50 weeks of operation in a year, and 10 hours of operation per week, the estimated
maintenance cost per hour is $50.
Total cost per hour: fuel (138.51) + labor (40) + maintenance (50) = $228.51/hr
With a throughput of 30 tons/hour, the estimated operating and maintenance cost per ton
processed is $7.62.
This number is then quadrupled to accommodate the rent of the capital equipment and overhead.
To obtain the estimate of Cr this number is then doubled to meet the assumption of a two-phase
mechanical recycling system. While this multiplication is somewhat arbitrary, it is important that
the recycling cost per ton be higher than the landfill option, as this reflects the reality of the
situation.
7.62 x 8 ⋍ $60/ton
Appendix D
VBA script used to run Lingo Model One on multiple parameter scenarios automatically.
Appendix E
VBA script used to run Lingo Model Two on multiple parameter scenarios automatically.
Appendix F
