Maintenance Optimization of Offshore Wind Turbines Based on an Opportunistic Maintenance Strategy
Round 1
Reviewer 1 Report
Reviewer's comments on Maintenance Optimization of Offshore Wind Turbine Based on Opportunistic Maintenance Strategy
The steps within the flowchart in figure 1 should be made clearer. Some steps are written just as equations without any context to them.
In sections 2 and 3, Weibull parameter estimation methods are discussed. It will help more to mention/discuss any tools which can be used to perform said estimations.
Why are some other estimations such as Method of Moments and Median Rank Regression not mentioned in sections 2 and 3.
There are too many distinct sections:
Sections 2 and 3 can be combined into one section
Section 5, 6 and 7 can come under one section
Table 2 and 3 should be much clearer so as to show which methods produce which results
The figure 2 should be improved so that the curves are distinguishable from each other.
Numbered lists, especially in section 6, should be arranged in list format for clarity
References in lines 378 and 458 should be corrected
The paper should be reviewed for grammatical errors especially those in lines 164, 673, 675, 678, 680, 688 and 691.
Some additional minor comments:
Line 85: “ the criteria of accessibility was proposed to evaluate the maintenance accessibility.”
Line 90: “…wind turbine was also optimised accordingly.”
In the introductory section, a sentence introducing chapter organisation will be useful.
In the introductory section Besides 2&3-parameter Weibull distribution, other possible methods can be discussed and a comparison for why the 3-parameter Weibull distribution is better/selected.
Line 136: “incomplete life…(what is meant by incomplete life?)”
Line 149: “…low stress levels…”
Line 153: “when the operation duration is beyond the …”
Line 159: “maintenance personnel may not get access…”
Line 160-161: “deficiency of knowledge on the faulty components and thus difficulties in analysing the operation reliability.”
Line 164: “The research work, shown in…”
Line 196: Perhaps explain the flow chart in more detail
Line 234: SCADA in bracket?
Line 260, 661: Table 3, 11 and 12 (not III, XI and XII)
Line 285: “For a 3MW wind turbine…”
Line 378, 673, 682, 691: is “21” a reference? Reference 8 and Reference 9.
Line 433: Equation number to be corrected
Line 457: Reference number(“literature27)
Line 521: “…after one component goes through …’
Line 667: “Combined with the influence of weather factors…”
After the simulation analysis, a discussion about the strengths and limitations of the method may be discussed.
Author Response
Response to Reviewer 1 Comments
Thank you for your serious and constructive comments on our manuscript (Manuscript ID: energies-519432). According to your suggestion, the manuscript has been revised as a letter to editor. The revisions we have made are as follows:
Point1. The steps within the flowchart in figure 1 should be made clearer. Some steps are written just as equations without any context to them.
Response 1: Thank you for your constructive and helpful suggestion. We have made relevant explanation in the context to the flowchart in figure 1 and shrink down some theoretical words. As shown in page 6 (line206-211) in the revised manuscript.
Point 2. In sections 2 and 3, Weibull parameter estimation methods are discussed. It will help more to mention/discuss any tools which can be used to perform said estimations.
Response2: Thank you for your constructive and helpful suggestion. We have made explanation about the tools which shows how we can perform the estimations. MATLAB is a mathematics software produced by MathWorks. It is a high-level technical computing language and interactive environment for algorithm development, data visualization, data analysis and numerical calculation, mainly including MATLAB and Simulink. The paper adopted Matlab software for data calculation and simulation. As shown in page 6 (line195-198) in the revised manuscript.
Point 3. Why are some other estimations such as Method of Moments and Median Rank Regression not mentioned in sections 2 and 3.
Response3: Thank you for your constructive suggestion, as a response, we added relevant description on Method of Moments and Median Rank Regression, as shown in page 4 (line135-146) in the revised manuscript. Please kindly find the following interpretation in relation to your comment.
The Weibull distribution is popular because it provides a useful description for many different kinds of data, especially in engineering applications such as reliability. Prior to desktop computers and reliability analysis software became available, engineers and statisticians commonly used probability plots to analyze censored life data. A non-parametric estimate of the fraction failing as a function of time, consisting of a point for each failure time, would be plotted on specially prepared papers designed such that a Weibull distribution would be represented by a straight line on the plot. If the plotted points did not deviate too much from a straight line, one would draw a line through the points to estimate the Weibull distribution.
After computers became available, it was possible to fit the line on the probability plot by using an objective analytical method. Most statisticians adopted the method of maximum likelihood because of its well-known distributional properties in “large samples” and also provides a visually appealing line through points. Simultaneously, another regular point estimator, method of moment featured by simplicity and practicability, was also applied by some engineers. During the engineering practice, maximum likelihood was usually considered in advance due to the prevailed accuracy as compared to method of moment which can be adopted once the maximum likelihood function was difficult to construct. In this paper, the description on maximum likelihood rather than method of moment was conveyed.
Meanwhile, Many engineers, however, used least squares method to draw the line on the probability plot because it was easier to program and more familiar and produces a visually appealing line through the points, when the points fall on a line. The most commonly used implementation of least squares probability plot approach uses median-rank plotting positions and the method is known today as median-rank regression. Practically, median-rank regression can perform well in case of the occurring extreme values when the number of failures is small. Generally, on the basis of least squares method, the mean variable derived from least squares method will be replaced by the median-rank variable so as to alleviate the estimated deviation caused by extreme value. In this study, the maintenance optimization was based on the small sample and the extreme value was carved out, therefore, least squares method rather than median-rank regression was expounded.
Point 4. There are too many distinct sections:
Sections 2 and 3 can be combined into one section
Section 5, 6 and 7 can come under one section
Response4: Thank you for your constructive and helpful suggestion. We have combined the Sections 2 and 3 and Section 5, 6 and 7 was arranged under one section in the revised manuscript.
Point 5. Table 2 and 3 should be much clearer so as to show which methods produce which results
Response5: Thank you for your constructive and helpful suggestion. We have combined table 2 and 3 so as to show which methods produce which results, as shown in page 9 (line258-line262) in the revised manuscript.
Point 6. The figure 2 should be improved so that the curves are distinguishable from each other.
Response6: Thank you for your constructive and helpful suggestion. We have made changes to the figure 2 and added an enlarged figure, as shown in page 10 (line 262-268) in the revised manuscript. Specifically, the color of the individual line has regulated and a partial enlarged figure was provided to clearly display the adjacent two line.
Point 7. Numbered lists, especially in section 6, should be arranged in list format for clarity
Response7: Thank you for your constructive and helpful suggestion. We have modified the questions raised by the experts, as shown in page 14 (line 375-377 and line 380-382) in the revised manuscript.
Point 8. References in lines 378 and 458 should be corrected
Response8: Thank you for your constructive and helpful suggestion. Due to our negligence, wording mistakes happened. It has been modified, as shown in page 14 (line 379) and page 16 (line 456) in the revised manuscript.
Point 9. The paper should be reviewed for grammatical errors especially those in lines 164, 673, 675, 678, 680, 688 and 691.
Response9: Thank you for your constructive and helpful suggestion. We carefully checked the above-mentioned contents. The grammatical errors were corrected and also inspected by native speakers. As shown in page 5 (line 181-183) and page 25-26 (line 692-710) in the revised manuscript.
Some additional minor comments:
Point 10. Line 85: “the criteria of accessibility was proposed to evaluate the maintenance accessibility.”
Response10: Thank you for your constructive and helpful suggestion. We have modified the questions raised by the experts. The sentence changed to “and proposed maintenance waiting time for offshore wind turbine”, as shown in page 2 (line 88-line 89) in the revised manuscript.
Point 11. Line 90: “…wind turbine was also optimised accordingly.”
Response11: Thank you for your constructive and helpful suggestion. We have modified the questions raised by the experts. The sentence changed to “and the preventive maintenance time and opportunistic maintenance time for main components of wind turbine was been optimized”, as shown in page 2 (line 91-92) in the revised manuscript.
Point 12. In the introductory section, a sentence introducing chapter organisation will be useful.
Response12: Thank you for your constructive and helpful suggestion. We have added a sentence introducing chapter organization, as shown in page 3 (line100-line108) in the revised manuscript.
Point 13. In the introductory section besides 2&3-parameter Weibull distribution, other possible methods can be discussed and a comparison for why the 3-parameter Weibull distribution is better/selected.
Response13: Thank you for your constructive and helpful suggestion. We have made discussion on other possible methods, as shown in page 4 (line 135-146). As for the reason for why three-parameter is prevailed, actually, we have made description of the 3-parameter Weibull distribution in line171-176 in the revised manuscript. As we discussed, the 3-parameter Weibull distribution has better characteristics of curve fitting as compared to two-parameters Weibull distribution. In Weibull distribution function, positional parameters is the threshold value for failure time of targeted devices and are used to estimate the earliest failure time and the condition with small sample data. We would like to make further explanation about this problem. As for 2-parameter Weibull distribution, such as the maximum likelihood estimation, under mild conditions, met in most common problems, maximum likelihood estimation have optimum properties in large samples. With modern computing hardware and software, maximum likelihood estimation is fast and easy to implement. As for least square estimation , using this method to fit the line would not seem to be a good idea because this application violates the assumptions under which least square is usually justified (constant variance and independent observations) as statistically optimum for estimation. The most commonly used implementation is least square estimation using median-rank positions, however, this method put large weight on the extreme observations. In this study, we focus on the offshore wind turbine that just put into operation on the early period of the life time with small sample data. Hence, in this sense, 3-parameter Weibull distribution is more suitable to deal with the short-term fault prediction which was proved in Figure 2.
Point 14. Line 136: “incomplete life…(what is meant by incomplete life?)”
Response14: Thank you for your constructive and helpful suggestion. As compared to the entire life-cycle of wind turbine. Incomplete life herein means a wind turbine that the time span after put into operation is less than the entire life of the wind turbine. For example, generally, the entire designed life of wind turbine is 20 years to 25 years. The practical time span after a wind turbine put into operation is 5 years. And Incomplete Life for this wind turbine is 5 years.
As an important and widely used point estimation method, under mild conditions, met in most common problems, maximum likelihood estimators have optimum properties in large samples. Experience, including many simulation studies, has shown that maximum likelihood estimators are generally hard to beat even in small samples. Meanwhile, the earlier we carry out the reliability prediction, the better the total cost of the wind turbine will be controlled. In some cases, the wind turbine was just put into operations for several years which has an incomplete life, so called newly engaged wind turbine. Under such condition, maximum likelihood estimators sometimes perform well in the point estimation process.
Point 15. Line 149: “…low stress levels…”
Response15: Generally, the two-parameter Weibull distribution is suitable for the fatigue test of high-stress metal materials such as rolling bearing of the mechanical devices. While, the three-parameter Weibull distribution is suitable for the fatigue test of low-stress metal materials of the mechanical devices, display better availability as compared to those of two-parameter Weibull distribution.
Point 16. Line 153: “when the operation duration is beyond the …”
Response16: Thank you for your constructive and helpful suggestion. The sentence was changed to“when the unit operation time exceeds the positional parameter, the unit starts to fail, as shown in page 5 (line176) in the revised manuscript.
Point 17. Line 159: “maintenance personnel may not get access…”
Response17: Thank you for your constructive and helpful suggestion. The sentence was changed to“maintenance personnel may can’t reach the wind turbine site for several months due to harsh weather conditions”, as shown in page 5 (line181-line183) in the revised manuscript.
Point 18. Line 160-161: “deficiency of knowledge on the faulty components and thus difficulties in analyzing the operation reliability.”
Response18: Thank you for your constructive and helpful suggestion. We changed the wording as follows: In addition, due to the particularity of the environment, when unit failure occurs in an offshore wind farm, maintenance personnel may can’t reach the wind turbine site for several months due to harsh weather conditions, which results in difficulty in obtaining component failure data and failure models. As shown in page 5 (line 182-183) in the revised manuscript.
Point 19. Line 164: “The research work, shown in…”
Response19: Thank you for your constructive and helpful suggestion. It’s a wording mistake, we are so sorry about that. As shown in page 5(line186) in the revised manuscript.
Point 20. Line 196: Perhaps explain the flow chart in more detail
Response20: Thank you for your constructive and helpful suggestion. We added some interpretation for the key parameter, as shown in page6 (line 206-211) in the revised manuscript.
Point 21. Line 234: SCADA in bracket?
Response21: Thank you for your constructive and helpful suggestion. We have modified the questions raised by the experts, as shown in page 8(line 255) in the revised manuscript.
Point 22. Line 260, 661: Table 3, 11 and 12 (not III, XI and XII)
Response22: Thank you for your constructive and helpful suggestion. We have modified the questions raised by the experts, as shown in page 10 (line 269) and page 25 (line 656 and line 659) in the revised manuscript.
Point 23. Line 285: “For a 3MW wind turbine…”
Response23: Thank you for your constructive and helpful suggestion. We have modified the mistakes raised by the experts, as shown in page 11 (line 294) in the revised manuscript.
Point 24. Line 378, 673, 682, 691: is “21” a reference? Reference 8 and Reference 9.
Response24: Thank you for your constructive and helpful suggestion. We are so sorry. It’s wording mistakes and we have corrected it.
Point 25. Line 433: Equation number to be corrected
Response25: Thank you for your constructive and helpful suggestion. We check it carefully and have modified the questions raised by the experts.
Point 26. Line 457: Reference number(“literature27)
Response26: Thank you for your constructive and helpful suggestion. We are so sorry. It’s wording mistakes and we have corrected it.
Point 27. Line 521: “…after one component goes through …’
Response27: Thank you for your constructive and helpful suggestion. We have modified the questions raised by the experts. We changed the wording as follows: when the component is repaired for a period of time, the maintenance rate is close to constant”, as shown in page 18 (line 519-520) in the revised manuscript.
Point 28. Line 667: “Combined with the influence of weather factors…”
Response28: Thank you for your constructive and helpful suggestion.We changed the wording as follows:Considering the influenced of weather factors such as wind speed and wave height”, as shown in page 25 (line 684) in the revised manuscript.
Point 29. After the simulation analysis, a discussion about the strengths and limitations of the method may be discussed.
Response29: Thank you for your constructive and helpful suggestion. We have added a discussion about the strengths and limitations of the method. Although the proposed model gains the competitiveness on unit maintenance cost, the limitations should be revealed as follows.
(1)The research on the maintenance cost was aimed at the key components of single wind turbine rather than the multiple wind turbines.
(2)Actually, the influential factors such as availability of the vessels, adequacy of the spare parts and meteorological surrounding conditions were linked to maintenance waiting time. In this paper, the wind speed and wave height was considered in the determination of maintenance waiting time, assuming the vessels and spare parts are available.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors address an important issue in wind energy production, which
is the maintaince procedures of off-shore wind turbines. In order to
reduce the costs of these maintainces strategies the authors propose a
strategy based in a modified Weibull distribution with a tunnable third
parameter representing the threshold value beyond which failure occurs.
As a quantitative output they find that the maintenance cost of opportunistic
maintenance strategy can be 10% less than the preventive maintenance strategy.
The work may be suitable for publication after addressing the following
remarks:
* Too much theory for markov chains, weibull etc.
* The aim of the present work is to develop an effective strategy to reduce
the maintenance cost of offshore wind turbine in consideration of the
accessibility of offshore wind farm. The idea of the extra parameter in
the Weibull distribution. However, little attention is given to the
fluctuating nature of wind speed increments (2-point statistics) and,
simultaneously, on the nonlinear connection between wind speed and
wind power (see e.g. power curve), which affects significantly the
effective cost. Previous work in this scope have been done in Energies
journal, e.g. Energies 10(12) 1944 and Energies 7(12) 8279-8293. Some
discussion in this scope would be in order.
Author Response
Response to Reviewer 2 Comments
Thank you for your serious and constructive comments on our manuscript (Manuscript ID: energies-519432). According to your suggestion, the manuscript has been revised as a letter to editor. The revisions we have made are as follows:
Point1. Too much theory for markov chains, weibull etc.
Response1: Thank you for your constructive and helpful suggestion. We have removed contents from the download submitted manuscript by online submission system and the detailed location can be seen from table 1 as below.
Table 1. Comparison of remove contents in submitted and revised manuscript
Download Submitted manuscript by online submission system | Position in revised manuscript |
line 109-110 | line 122 |
line 162-166 | line 186 |
line 182-186 | line 206 |
line 192 | line 215 |
Line353 | line 362 |
Line361-363 | line 366 |
Line383-386 | line 389 |
Line402-403 | line 404 |
Note: For example, line 109-110 herein means the line number of removed contents in download submitted manuscript by online submission system and line 122 means the line number where the reviewer can find what contents have been removed in the revised manuscript.
Point2. The aim of the present work is to develop an effective strategy to reduce the maintenance cost of offshore wind turbine in consideration of the accessibility of offshore wind farm. The idea of the extra parameter in the Weibull distribution. However, little attention is given to the fluctuating nature of wind speed increments (2-point statistics) and, simultaneously, on the nonlinear connection between wind speed and wind power (see e.g. power curve), which affects significantly the effective cost. Previous work in this scope have been done in Energies journal, e.g. Energies 10(12) 1944 and Energies 7(12) 8279-8293. Some discussion in this scope would be in order.
Response2: Thank you for your constructive and profound suggestion. We carefully probe into the references mentioned in the review comment. We think the novelties and comprehensive analysis give us an inspiration for our work. We’ve made relevant discussion in sector of conclusion from line 663 to line 673 ( the line number is located in the revised manuscript). The paper of Energies 10(12) 1944 and Energies 7(12) 8279-8293 was also cited from line 70 to line 72 ( the line number is located in the revised manuscript) to broad our viewpoint in this scope. Please kindly find the following response to your concerns.
The randomness of wind is reflected in the random variation of wind speed and wind direction. In this paper, the influence on the effective cost of offshore wind farm arising from the fluctuation of wind was considered in the estimation of maintenance waiting time. Practically, performing maintenance operations at offshore wind farms involves one major challenge compared with the onshore counterpart. All maintenance personnel and spare parts need to be transported from an onshore port or offshore station to the individual wind turbines by vessels(see different kinds of vessels from Fig.1 in the attached cover letter).
The vessels required for these tasks will constitute a major part of the maintenance costs for the offshore wind farms, and to reduce the cost of energy it is essential to keep an optimal or near-optimal vessel fleet for this purpose. Furthermore, offshore wind turbines are more exposed to breakdowns than their onshore counterpart due to facing a rougher environment. To enable growth in installed offshore wind capacity, offshore wind farms need to move further away from shore, to deeper waters. With this move, the cost of operation and maintenance will increase due to further travel distances, rougher weather conditions and greater logistic challenges. For example, to execute maintenance operations an operator needs a weather window that at a minimum covers the travel time and repair time at the wind turbine, and with a longer travel distance the length of the required weather window will increase. At this conditions, the random fluctuation of wind speed brings some difficulties to the maintenance vessels to reach to the wind turbine, and will prolong the weather window duration, i.e. maintenance waiting time. Generally, when the wind speed is higher than 10m/s, the maintenance vessels should be standby and are strictly forbidden to make navigation, which resulting in an increase loss cost arising from downtime and prolonged waiting time ( if the maintenance vessels was leased from manufacturer, the payment for lease will increase accordingly ). Therefore, the fluctuation of wind speed was considered to achieve the reasonable maintenance window time so as to reduce the maintenance cost to some extent (see Fig. 2 in the attached cover letter).
Generally, the total Life-Cycle Cost (LCC) of a Offshore Wind Farm is decomposed in the costs of each of the main phases of the process: wind resource evaluation cost (C1), design cost (C2), manufacturing cost (C3), installation cost (C4), levelized cost of energy(C5), maintenance cost(C6), financial cost (C7) and dismantling cost (C8). The total Life-Cycle Cost is shown as follows.
LCC=C1+C2+C3+C4+C5+C6+C7+C8 (1)
However, in order to obtain their main dependencies, each of these costs can be subdivided in the correspondent sub-cost(see Fig.3 in the attached cover letter). In particular, Operation and maintenance costs account for around one third of the life cycle cost of an offshore. Therefore it is essential to implement cost-effective operation and maintenance concepts and strategies to achieve a significant reduction of cost of energy from off re wind farm arising from the fluctuation of wind speed was considered in the determination of maintenance waiting time. From the perspective of reducing the entire life cost of offshore wind farm, we drew our attention on how much maintenance cost will be cut down with the introduction of the opportunistic maintenance strategy. Actually, in this sense how the power generation loss arsing from wind fluctuation will act on the unit cost is not the research priorities in this study. Indeed, the influence of wind fluctuation and grid impact on the unit is worth being intensively studied in further research. Therefore, we made a discussion about the strengths and limitations of the method in sector of conclusion (see line 671-681 in the revised manuscript).
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have correctly implemented almost all suggested comments. However the English writing is still not very good and has to be improved. Also some papers about opportunities maintenance in wind energy are still missing from the literature. Some examples are:
- Opportunistic maintenance for wind turbines considering imperfect, reliability-based maintenance
- An opportunistic condition-based maintenance policy for offshore wind turbine blades subjected to degradation and environmental shocks
- Opportunistic maintenance for wind farms considering multi-level imperfect maintenance thresholds
- Multi-objective opportunistic maintenance optimization of a wind farm considering limited number of maintenance groups
etc.
Author Response
Response to Reviewer 1 Comments
Thank you for your serious and constructive comments on our manuscript (Manuscript ID: energies-519432). According to your suggestion, the manuscript has been revised as a letter to editor. The revisions we have made are as follows:
Point1. The authors have correctly implemented almost all suggested comments. However the English writing is still not very good and has to be improved. Also some papers about opportunities maintenance in wind energy are still missing from the literature. Some examples are:
- Opportunistic maintenance for wind turbines considering imperfect, reliability-based maintenance
- An opportunistic condition-based maintenance policy for offshore wind turbine blades subjected to degradation and environmental shocks
- Opportunistic maintenance for wind farms considering multi-level imperfect maintenance thresholds
- Multi-objective opportunistic maintenance optimization of a wind farm considering limited number of maintenance groups
etc.
Response 1: Thank you for your approval on our revision work of Round 1. we applied to English editing service from MDPI English Editing to improved the language of our revised manuscript. As for the literature about opportunistic maintenance, we carefully read the recommended papers and made a review in the sector of introduction (please see line 78-89 in the revised manuscript).
Author Response File: Author Response.pdf
Reviewer 2 Report
The authors address all raised points. However bibliography should be formatted consistently (FAMILYNAME, Initials, avoid "et al", etc) and references to them in the main text should follow by naming last name of authors (not the first name), etc.
Author Response
Response to Reviewer 2 Comments
Thank you for your serious and constructive comments on our manuscript (Manuscript ID: energies-519432). According to your suggestion, the manuscript has been revised as a letter to editor. The revisions we have made are as follows:
Point1. The authors address all raised points. However bibliography should be formatted consistently (FAMILYNAME, Initials, avoid "et al", etc) and references to them in the main text should follow by naming last name of authors (not the first name), etc.
Response 1: Thank you for your approval on our revision work of Round 1. We referred to Guide for Authors of energies and the reference style of published papers issued by energies. The format of the references in the manuscript has been inspected and modified accordingly, as well as relevant format in the main text. Moreover, the language of our manuscript was edited by MDPI English Editing.
Author Response File: Author Response.pdf
Round 3
Reviewer 1 Report
The authors need to do another proofread for their papers before it goes for publication.
