THEEFFECTOFYAWERRORONTHERELIABILITYOFWINDTURBINEBLADES V2.0 RoozbehBakhshi

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/305000849

The Effects of Yaw Error on Reliability of Wind Turbine Blades

Conference Paper · June 2016


DOI: 10.1115/ES2016-59151

CITATIONS READS
15 1,014

2 authors:

Roozbeh Bakhshi Peter Sandborn


University of Maryland, College Park University of Maryland, College Park
17 PUBLICATIONS   131 CITATIONS    284 PUBLICATIONS   3,446 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Servitization; manufacturing focused work View project

Dynamic Performance Model for a Closed-loop Supply Chain System with Uncertain Demand and Returns View project

All content following this page was uploaded by Roozbeh Bakhshi on 07 July 2016.

The user has requested enhancement of the downloaded file.


Proceedings of the ASME 2016 Power and Energy Conference
PowerEnergy2016
June 26-30, 2016, Charlotte, North Carolina

PowerEnergy2016-59151

THE EFFECT OF YAW ERROR ON THE RELIABILITY OF WIND TURBINE BLADES

Roozbeh Bakhshi Peter Sandborn


University of Maryland-College Park University of Maryland-College Park
College Park, Maryland, USA College Park, Maryland, USA

ABSTRACT P Power production


The rise of energy prices, concerns over climate change and r Reliability improvement relative to 0 degrees (fraction)
geopolitical issues have brought special attention to renewable T Failure time
sources of energy and wind energy in particular. Based on NREL V Wind speed
projections, the United States has more than 32,000 TWh of α Yaw error
onshore and 17,000 TWh of offshore potential for wind power β Weibull shape parameter
generation, which is far beyond its 11,000 TWh of current annual 𝜂 Weibull scale parameter
electricity consumption. However, there are a number of γ Weibull location parameter
efficiency challenges that must be overcome in order to turn this ρ Air density
potential into actual production. One area that can potentially σ Equivalent stress
improve the energy production of wind turbines is the correction
of yaw error. Yaw error (also referred to as yaw angle or yaw INTRODUCTION
misalignment) is the angle between the turbine`s rotor and the With concerns over climate change and geopolitical
wind direction. A yaw error reduces turbine`s power production conflicts, countries have been shifting toward renewable energy
at wind speeds below the rated speed. sources to address the emission of carbon dioxide, and at the
Besides impacting the power producing ability of a turbine, same time secure stable sources of energy. Wind energy, both
yaw error also affects the reliability of critical subsystems in onshore and offshore, have been a central focus in renewable
wind turbines. Variation in yaw error (at any wind speed and not energy strategies. US National Renewable Energy Lab (NREL)
only below the rated speed) affects the loads on the components projects the US potential wind energy electricity production as
and the subsequent mechanical stresses. These mechanical 32,000 TWh for onshore and 17,000 TWh for offshore, which is
stresses change the damage accumulation for components and far beyond the current annual US electricity consumption of
sub-assemblies, which ultimately affects their reliability. About 11,000 TWh [1].
17 to 28% of wind project costs attribute to O&M costs, which Energy production from a wind turbine is a function of many
are directly affected by the reliability. parameters such as wind speed, the distribution of wind speeds,
In this study, we investigate the effects of yaw error on the uptime (the number of hours the turbine is operational), etc. One
reliability of blades by performing load and stress analysis for parameter that affects the energy production of a wind turbine is
various yaw errors. We then use the results of these analyses to the yaw error (also referred to as yaw angle or yaw
adjust the Weibull parameters used for predicting the failure time misalignment). Yaw error is the angle between wind direction
of blades. Finally, we will use a stochastic cost model to show and rotor`s central axis line (defined in the horizontal plane).
how correcting the yaw error can avoid maintenance costs in Yaw error reduces the power production at wind speeds less than
wind farms. the rated speed. The relation between the yaw error and power
production is:
1
NOMENCLATURE 𝑃 = 𝜌𝐴𝑉 3 𝑐𝑜𝑠 3 (𝛼) (1)
2
A Sweeping area of rotor
where:
B Coefficient depending on material properties
P: power production
m Exponent depending on material properties
ρ: air density
N Number of cycles to failure

1 Copyright © 2016 by ASME


A: sweeping area of rotor occurrences of lower wind speeds do not result in instantaneous
V: wind speed failures but they introduce damage to the component and over
α: yaw error time, this damage accumulates and results in a failure. Cyclic
The traditional way to measure and correct the yaw error is loads are the primary cause of fatigue (which is a wearout failure
through a vane mounted on the nacelle of the turbine. This mechanism) in turbine sub-assemblies, [7-8]; wearout is the
method always results in some remainder yaw error. The average focus of this paper. In order to study cyclic loads, we have to
reported yaw error for a wind turbine using vanes is about 7 understand which wind speeds are experienced most frequently.
degrees [2] and some studies report it as high as 10 degrees [3]. In this work, we focus on four main cyclic loads and how
There have been several technological improvements to address they yaw error affects them. These loads are edgewise shear
this issue. For example, light detection and ranging (LIDAR) force, flapwise shear force, edgewise bending moment and
uses laser beams and their reflection from particles in the wind flapwise bending moment. These loads are a result of shear wind,
flow to measure wind speed and direction in front of the turbine. where the wind speed at the top of the rotor is different from
Using LIDAR, yaw error can be reduced to as low as 1 degree. wind speed at the bottom of it.
Several studies have focused on the benefits of yaw error We study the variation of these four loads at constant wind
reduction to the revenue generation of a wind turbine. For speeds for different yaw errors. These speeds are the most
example, a 7 to 1 degree reduction results in an extra 2% power probable wind speeds that a wind turbine will experience
generation [4]. The important aspect of using LIDAR systems in throughout its life. For example, wind speed probability density
wind farms is how often the yaw correction is made. Ideally, each function (pdf) in the Delaware Bay [9] is shown in Figure 1. The
turbine in a farm would have its own dedicated LIDAR, which most probable wind speed is 8 m/s. Here, we focus on wind
keeps the yaw error at a minimum at all times, however this speeds of 6, 7 and 8 m/s.
approach can be expensive. Another approach is to move a single
or multiple LIDAR units between all the wind turbines in a farm.
In this approach, the LIDAR stays on a particular turbine for a
specific period of time, corrects the yaw error then moves to the
next turbine. The corrected yaw error (to about 1 degree)
gradually regresses back to its original uncorrected state until the
LIDAR comes back to that particular turbine.
The studies referenced above [2-3] only consider power
generation efficiency improvements and neglect the effects of
yaw error on the reliability of wind turbines. The misalignment
between the wind direction and rotor puts extra cyclic loads on
some critical components such as the blades and gearbox. These
elevated loads produce elevated stresses on the components.
These stresses cause additional cyclic damager to accumulate
and eventually results in a shorter useful life and a subsequent
failure of turbine components. Load alleviation through yaw
error reduction has been shown to improve the reliability of sub-
assemblies [5]. As a result, not only the cost of maintenance is
reduced but also, the uptime (operation time) of the wind turbine
increases, which results in extra revenue. In this paper, we focus Figure 1- Wind speed pdf for year 2011 at Delaware Bay, using data
on the effects of yaw error improvements on turbine blades, from [9]
quantify the reliability improvement as a function of yaw error
reduction and investigate the cost benefits of yaw error Load analysis can be done on any location on the blades;
reduction. however, the most susceptible site for failure is the blade root,
where the blade is attached to rotor. Research shows that most
LOAD ANALYSIS fatigue related failure of turbine blades occur at the junction
Load analysis in wind turbines, includes forces and where blade attaches to the rotor [10].
moments applied to components and sub-assemblies. These In our investigation, we use the load analysis data provided
loads are either extreme loads or cyclic loads. Extreme loads are by [11]. The analysis was done on a generic 2MW wind turbine
due to extreme conditions such as high or unusual winds speeds. with hub height of 65 meters, cut-in wind speed of 4 m/s, cut out
Cyclic loads come from attributes such as the rotation of the of 25 m/s and rated wind speed of 12.8 m/s.
rotor, yaw movement, uneven wind profiles on the rotor (shear Variation of the four main loads, flapwise bending moment
wind), etc. Extreme loads result in overstress failure mechanisms (Flap BM), edgewise bending moment (Edge BM), flapwise
whereas cyclic loads result in wearout failure mechanisms [6]. shear force (Flap SF) and edgewise shear force (Edge SF) as a
For example, a single occurrence of an extreme load at a very function of yaw error are plotted in Figures 2-4. Assuming a yaw
high wind speed may result in an instantaneous failure whereas error of 0 degrees as a point of reference, vertical axis shows how

2 Copyright © 2016 by ASME


much loads at each yaw error change relative to the loads at 0 Another important observation is the sensitivity of loads to
degrees. These plots represent loads at the blade root. yaw error variations. The most sensitive load is flapwise bending
moment where a drop of 8% to 11% at different wind speeds is
observed when yaw error changes from -20 to 0 degrees.

STRESS ANALYSIS
Cyclic loads create damage that accumulates over time. The
accumulated damage results in a fatigue failure mechanism. This
damage accumulation can be modeled using several methods.
Miner`s rule is the most common model for damage equivalent
stress. Through the use of S-N curves, the equivalent stress can
be transformed to the number of cycles to failure. S-N curves
depend on the components` material properties. Equation (2)
shows the mathematical form of an S-N curve.
Figure 2-Variation of loads with yaw error at blade root for wind
speeds of 8 m/s, using data from [11] 𝑁 = 𝐵𝜎 𝑚 (2)
where:
N: number of cycles to failure
B: coefficient depending of material properties
m: exponent depending on material properties
σ: equivalent stress

The number of cycles to failure calculated through this


method corresponds to the point where 63.2% of the population
have failed. Equation (2) can also be written as a ratio like
Equation (3) in order to compare the number of cycles to failure
for cases where stress changes.
Figure 3- Variation of loads with yaw error at blade root for wind 𝑁2 𝜎
speeds of 7 m/s, using data from [11] = ( 2 )𝑚 (3)
𝑁1 𝜎1

The equivalent stress (σ) in Equation (3) is the stress at a


specific point on a structure caused by all the loads at that
particular point. However, if there is a dominant load, which
contributes the most to the variation of stress, it can replace the
stress in Equation (3). In our study, flapwise bending moment, as
discussed in the previous section, is the load that reacts the most
to variations of yaw error and therefore can replace the stress.
We assume that there is a linear relation between load and stress.
Huang et al. [11] investigates the loads at yaw errors
between -20 to 20 degrees in 8-degree step intervals. In order to
Figure 4-Variation of loads with yaw error at blade root for wind see the effects of yaw error on the number of cycles to failure,
speeds of 6 m/s, using data from [11] we take the loads at 0 and 8 degrees yaw errors and calculate the
ratio in Equation (3). Table 1 shows the loads for different wind
In all the scenarios shown in Figures 2-4, the changes in speeds.
edgewise bending moment and edgewise shear force have a
direct reverse relation with changes in yaw error, i.e., they Table 1- Flapwise bending moment (units in kNm), data from [11]
decline as the yaw increases and they never reach a minimum.
The flapwise bending moment and shear force show a 8 m/s 7 m/s 6 m/s
different behavior. They have a deflection point where the load Flap BM (Yaw=8 degrees) 2024.39 1876.34 1706.44
reaches a minimum. The deflation point is different at different Flap BM (Yaw=0 degrees) 2074.26 1892.9 1687.22
wind speeds. For the flapwise bending moment, the minimum
occurs at 0 degrees of yaw error for wind speeds of 7 and 8 m/s,
whereas at 6 m/s it doesn`t reach the minimum until a much Calculating the change in N for a blade made out of
larger yaw error. composites with a material property exponent m = -10, and at the

3 Copyright © 2016 by ASME


wind speed of 8 m/s, which is the most probable wind speed number of cycle to failure from an S-N curve. By using Equation
based on Figure 2, we obtain: (5), for a 7 to 1 degree yaw improvement, the reliability of blades
improves 19.8%. The subsequent improved Weibull scale
𝑁2 𝜎2 𝑚 𝑙2 𝑚 𝐹𝑙𝑎𝑝 𝐵𝑀 𝑁2 2024.39 −10 parameter will be 12.366 years. These parameters (shape and the
=( ) ≅( ) → =( ) (4) new scale) form a new Weibull distribution, which represent the
𝑁1 𝜎1 𝑙1 𝑁1 2074.26
= 0.9759 −10
= 1.2762 time to failure of blades in the case where the yaw error is 1
degree. Figure 5 shows the Weibull distribution for blade time to
This implies that for an 8 degree correction of yaw error, the failure before and after the yaw error correction for a total
number of cycles to failure improves by 27%. support time of 20 years.
Using the data in [11] for yaw errors ranging from 0 to 20
degrees and performing a similar analysis, we developed a third
degree polynomial function that shows the relation between
reliability and yaw error change:

𝑟 = 0.000139(𝛼)3 − 0.001553(𝛼)2 + 0.037151(𝛼) +


0.004266 (5)
where:
r: reliability improvement relative to yaw = 0 degrees (fraction)
α: yaw error
The formula calculates the reliability when the corrected yaw is
0 degrees. In order to use it for corrected yaw values other than
0 degrees, a ∆r value has to be calculated where r1 and r2
represent the yaw errors before and after correction.

RELIABILITY OF BLADES
Reliability of wind turbine and its sub-assemblies can be Figure 5-Weibull distribution for blades failure before and after yaw
modeled using a two or three parameter Weibull distribution. error correction
Equation (6) shows the probability density function (pdf) of 3-
parameter Weibull distribution. COST MODEL
𝑇−𝛾
𝛽 𝑇−𝛾 𝛽−1 −( 𝜂 )𝛽
Yaw error reduction has multiple benefits: extra energy
𝑓(𝑇) = ( ) 𝑒 (6) production due to higher efficiency (Equation (1)), and fewer
𝜂 𝜂
where: maintenance events due to improved reliability. In this section
T: failure time we focus on cost benefits due to fewer maintenance events, i.e.,
β: shape parameter cost avoidances and the extra revenue generated due to
𝜂: scale parameter performance improvements.
γ: location parameter Each failure costs money. Maintenance could be corrective
(break-fix) or preventive (replace before failure). In either case,
The location parameter refers to failure free time. In many cases, an improved reliability extends the life of a component and
γ is assumed to be zero and Equation (6) turns into what is known results in fewer maintenance events over the total life cycle of
as a 2-parameter Weibull distribution. the wind farm (nominally 20 years). Component replacement is
Studies such as [12-14] studied field failure data for turbines only one of the contributors to the maintenance cost. Other costs
in Britain and Sweden while [15] performed reliability analysis include transportation cost, labor cost, inventory management
to generate to 2-parameter Weibull distributions for wind cost, power loss, etc.
turbines in Germany. The investigation in [15] was for the cases Our approach is to model the maintenance cost in a
that a failure of a blade occurred and the subsequent maintenance stochastic discrete-event simulator [17-19]. The analysis is
was a blade replacement. The data provided in [15] does not stochastic because the failures follow a probability distribution
specify the yaw errors, however, since the population is large (on and do not occur on pre-determined dates. The system
the order of 1000s of turbines), and the data goes back as far as considered is a wind farm consisting of individual wind turbines
1994 where LIDAR yaw correction was not common, it is where the state of each turbine (which is ‘functional’ or
assumed that the yaw error was the common 7 degrees ‘stopped’) only changes at discrete points in time. Failures and
mentioned in [2]. Here we use Weibull parameters calculated by their subsequent maintenance are events that change the state of
[15] in which blades follow a 2-parameter Weibull distribution the system.
with scale parameter (η)=10.323 years and shape parameter We focus only cost avoidance due to fewer replacements
(β)=1.042. In a Weibull distribution, the scale parameter (component costs) and do not include other maintenance costs
represent 63.2% unreliability, which is the same concept as the (labor, transportation, power loss, etc.). The average cost of a

4 Copyright © 2016 by ASME


single blade is assumed to be $200,000 [20], and we assume that
every maintenance event we generate through our discrete-event
simulator results in a blade replacement since the inputs to the
model were calculated based on blade field failure data.
Maintenance events that result in blade repair (as opposed to
blade replacement) are not included since we don`t have enough
data to investigate them. For a wind farm with 30 turbines and a
LIDAR device on each turbine, the total cost of maintenance due
only to blade replacement with yaw errors of 7 and 1 degrees for
20 years of operation are shown in Figure 6.

Figure 7-Cumulative revenue for a 30-turbine wind farm over 20 years


(discount rate = 0)

Summing the cumulative blade replacement cost avoidance


(Figure 6) and cumulative revenue increase (Figure 7) we get $4
million (referred to as the Total Cost Benefits), which still falls
short of the LIDAR purchase and maintenance cost over 20 years
($7.2 million from Table 2). Figures 6 and 7 are the results for a
single run of the stochastic model. Figure 8 shows the results of
using Monte Carlo analysis (100 runs).

Figure 6- Cumulative cost of maintenance for a 30-turbine wind farm


over 20 years, steps indicate blade replacement maintenance events
(discount rate = 0)

The cost avoidance due to correction of yaw error is about 2


million dollars for the wind farm (the difference between the two
lines in Figure 6 at 20 years). Each LIDAR device costs
$120,000 and maintaining the LIDAR is 10% of the cost of the
LIDAR device every 2 years [2]. Table 2 shows the total cost of
LIDAR ownership for a 30-turbine farm.
Table 2- LIDAR cost of ownership for a 30-turbine farm over 20 years
(discount rate = 0)

Total Cost of
Purchase Cost ($) Maintenance Cost ($)
Ownership ($)
Figure 8-Histogram of cost benefits due to correction of yaw error
3,600,000 3,600,000 7,200,000 after 20 years

Considering only the cost avoidances due to blade In the best case scenario, the maximum cost benefit is about
replacement ($2 million from Figure 6), implementation of $5 million, which is less than the $7.2 million cost of ownership
LIDAR on every single turbine in a wind farm ($7.2 million from of the 30 LIDARs on 30 turbines for the whole operation period
Table 2) is not financially justified; however, LIDAR has other of the farm.
cost benefits. There is the extra revenue generation because of Considering all cost benefits and the best-case scenario,
performance improvements. As Equation (1) shows, the energy implementing LIDAR on every single turbine to avoid
production of a turbine has a direct relation with the yaw error of maintenance costs due to blade replacements does not have a
the turbine. Figure 7 shows the revenue generated by a wind farm financial justification. However, it is important to note that
over 20 years consisting of 30 turbines with a rated power of 6 LIDAR does not just affect the reliability of blades. It also
MW per turbine assuming the cost of electricity is $0.144 per concurrently affects the reliability of other critical and expensive
kWh of energy. components such as the gearbox and generator (component costs

5 Copyright © 2016 by ASME


of around $300,000 and $150,000 [20]). We have not Lastly, the policy of installing a LIDAR device permanently
investigated the reliability improvements of other components. on every single wind turbine in a wind farm may not be the
We are also ignoring the cost avoidances due to maintenance optimal solution. In future studies, we will investigate other
events that do not necessarily result in a component replacement policies where one or multiple LIDAR devices circulate in the
but require only a repair. A comprehensive cost model, which wind farm. In these scenarios, the LIDAR capital and operation
included repair and replacement of all critical components, will costs reduce significantly. It`s also important to point out that the
give a more thorough understanding of true cost benefits of cost of LIDAR is in a downtrend and in 10 years, the prices might
LIDAR systems. be a fraction of what they are today. In addition, the cost of the
components relative to the cost of the LIDAR is a significant
DISCUSSION driver. Cost of blades and other major components for large
In this paper, we showed that correction of yaw error offshore turbines vary depending on the size of the turbine,
reduces the stress, improves the fatigue life and subsequently which may change the business case for putting a LIDAR on
leads to a better reliability. However, it`s important to point out every turbine.
the simplifying assumptions in this work. The first assumption
to address is how load calculations are done. The loads are ACKNOWLEDGMENTS
investigated at only three wind speeds with 8-degree yaw errors Funding for this work was provided by Exelon for the
steps. In order to get better results, loads calculations should be advancement of Maryland's offshore wind energy and jointly
in smaller intervals of yaw errors at a broader range of wind administered by MEA and MHEC, as part of “Maryland
speeds. Offshore Wind Farm Integrated Research (MOWFIR): Wind
Another assumption is in the calculation of stress. In reality, Resources, Turbine Aeromechanics, Prognostics and
stress is not a function of only one load and it may or may not Sustainability”.
have a linear relation with load(s). A finite element analysis
(FEA) model investigates the effects of multiple loads on REFERENCES
stresses in a more detailed way. After calculating equivalent [1] A. Lopez, B. Roberts, D. Heimiller, N. Blair, and G. Porro,
stresses, in order to transform these stresses into fatigue life, one “U.S. Renewable Energy Technical Potentials: A GIS-Based
would have to choose a specific or reference wind turbine with Analysis.” National Renewable Energy Laboratory, July 2012.
known material properties such as mass and stiffness. [2] E. Marin and H. Pedersen, “Pointing to the Right Direction,”
The maintenance cost model used in this paper is very presented at the EWEA, Malmo, Sweden, December 2014.
simple - corrective maintenance that considers the cost of spares [3] K. A. Kragh, M. H. Hansen, and T. Mikkelsen, “Precision
as the sole contributor to maintenance cost. More sophisticated and Shortcomings of Yaw Error Estimation Using Spinner-Based
maintenance models could be used [21-23]. Here we are Light Detection and Ranging,” Wind Energy, vol. 16, no. 3, pp.
assuming that every turbine in the farm has a dedicated LIDAR 353–366, April 2013.
installed on it, and every maintenance event results in a blade [4] F. Rebeyrat, “Turbine-Mounted Lidar for Performance
replacement. Basically it`s a comparison between two cases Optimization,” Wind Energy Update: 6th International Wind
where the turbines operate at a constant 7-degree yaw error for O&M Forum, Hamburg, Germany, January 2014.
20 years versus a case where turbines always operate at 1-degree [5] K. A. Kragh and M. H. Hansen, “Load Alleviation of Wind
over the life of the wind farm. Turbines by Yaw Misalignment,” Wind Energy, vol. 17, no. 7,
CONCLUSION pp. 971–982, July 2014.
It is clear that correction of yaw error has benefits beyond the [6] R. Bakhshi, S. Kunche, and M. Pecht, “Intermittent Failures
extra revenue. Lower stresses and subsequent improved fatigue in Hardware and Software,” Journal of Electronics Packaging,
life leads to an increased reliability. The improvements reduce vol. 136, no. 1, pp. 011014–011014, February 2014.
the number of failures and essentially maintenance events. With [7] H. Li, Z. Hu, K. Chandrashekhara, X. Du, and R. Mishra,
high costs of components, cost avoidance from improved “Reliability-Based Fatigue Life Investigation for a Medium-
reliability is on the order of millions of dollars. Scale Composite Hydrokinetic Turbine Blade,” Ocean
We performed the analysis only on blades. Extra loading due Engineering, vol. 89, pp. 230–242, October 2014.
to yaw error concurrently affects other components in a wind [8] C. Kong, T. Kim, D. Han, and Y. Sugiyama, “Investigation of
turbine such as gearbox, generator and rotor. A thorough analysis Fatigue Life for a Medium Scale Composite Wind Turbine
of all the components affected by yaw error would demonstrate Blade,” International Journal of Fatigue, vol. 28, no. 10, pp.
the true cost avoidances of yaw error correction. Just considering 1382–1388, October 2006.
cost avoidances due to only blade replacements does not make a [9] National Data Buoy Center, “Station 44009 (LLNR 168) -
business case for implementation of LIDAR. In future studies, DELAWARE BAY 26 NM Southeast of Cape May, NJ.”
development of a comprehensive model will demonstrate how [Online].Available:http://www.ndbc.noaa.gov/station_page.php
yaw error correction affects the reliability of all components, ?station=44009. [Accessed: November 2014].
which experience extra loading due to yaw error. The model has [10] C. C. Ciang, J. R. Lee, and H. J. Bang, “Structural Health
to incorporate both component replacement and repair. Monitoring for a Wind Turbine System: a Review of Damage

6 Copyright © 2016 by ASME


Detection Methods,” Measurement Science and Technology, vol.
19, no. 12, p. 122001, 2008.
[11] L. Huang, A. Cordle, and G. McCann, “Fatigue Load
Calculations for ROMO Wind to Assess Sensitivity to Changes
in 10-min Mean Yaw Error.” GL-Garrad Hassan, 2012.
[12] J. Carroll, A. McDonald, and D. McMillan, “Failure rate,
repair time and unscheduled O&M cost analysis of offshore wind
turbines,” Wind Energy, January 2015.
[13] J. Carroll, A. McDonald, O. Barrera Martin, D. McMillan,
and R. Bakhshi, “Offshore wind turbine sub-assembly failure
rates through time,” in Scientific Proceedings of EWEA Annual
Conference and Exhibition 2015, Paris, France, 2015, pp. 112–
116.
[14] J. Ribrant and L. M. Bertling, “Survey of Failures in Wind
Power Systems With Focus on Swedish Wind Power Plants
During 1997-2005,” IEEE Transactions on Energy Conversion,
vol. 22, no. 1, pp. 167–173, March 2007.
[15] F. Spinato, P. J. Tavner, G. J. W. van Bussel, and E.
Koutoulakos, “Reliability of Wind Turbine Subassemblies,” IET
Renewable. Power Generation, vol. 3, no. 4, pp. 387–401,
December 2009.
[16] P. J. Tavner, J. Xiang, and F. Spinato, “Reliability Analysis
for Wind Turbines,” Wind Energy, vol. 10, no. 1, pp. 1–18,
January 2007.
[17] R. Bakhshi, P. Sandborn, X. Lei, and A. Kashani-Pour,
“Return on Investment Modeling to Support Cost Avoidance
Business Cases for Wind Farm O&M,” Proceedings of the
EWEA Offshore, Denmark, 2015.
[18] T. Jazouli, P. Sandborn, and A. Kashani-Pour, “A Direct
Method for Determining Design and Support Parameters to Meet
an Availability Requirement,” International Journal of
Performability Engineering, vol. 10, no. 2, p. 211, March 2014.
[19] P. Sandborn, Q. Cui, X. Zhu, and A. R. Kashani-Pour, “A
New ‘Availability-Payment’ Model for Pricing Performance-
Based Logistics Contracts,” NPS Report, June 2014.
[20] J. Nilsson and L. Bertling, “Maintenance Management of
Wind Power Systems Using Condition Monitoring Systems -
Life Cycle Cost Analysis for Two Case Studies,” IEEE
Transaction on Energy Conversion, vol. 22, no. 1, pp. 223–229,
March 2007.
[21] X. Lei, P. Sandborn, R. Bakhshi, A. Kashani-Pour, and N.
Goudarzi, “PHM Based Predictive Maintenance Optimization
for Offshore Wind Farms,” Proceedings of the IEEE Conference
on Prognostics and Health Management (PHM), pp. 1–8, 2015.
[22] X. Lei, P. Sandborn, R. Bakhshi, and A. Kashani-Pour,
“Development of a Maintenance Option Model to Optimize
Offshore Wind Farm O&M,” Proceedings of the EWEA
Offshore, Denmark, 2015.
[23] F. Besnard, M. Patrikssont, A.-B. Strombergt, A.
Wojciechowski, and L. Bertling, “An Optimization Framework
for Opportunistic Maintenance of Offshore Wind Power
System,” in PowerTech, 2009 IEEE Bucharest, pp. 1–7, 2009.

7 Copyright © 2016 by ASME

View publication stats

You might also like