Fiber Optic Ice Sensor for Measuring Ice Thickness, Type and the Freezing Fraction on Aircraft Wings
Round 1
Reviewer 1 Report
Dear writer,
I think that the paper is interesting but some things could add more value to the paper. I will give you recommendations or opinions more than corrections. For example, in the introduction I would add more references for ice sensor technologies and I would compare your sensor with others (advantages and disadvantages). A common reference is Jackson and Goldberg paper:
Jackson, D.G., & Goldberg, J.I. (2007). Ice Detection Systems: A Historical Perspective.., & Goldberg, J.I. (2007). Ice Detection Systems: A Historical Perspective.
And additionally you can add more references of optic fiber ice sensors like:Icing Condition Predictions Using FBGS, M Gonzalez M Frovel
I will give you another recommendations:
1. In the paragraph of the line 98, there is a definition of the freezing fraction and its importance. I think, even thought I think you have explained it, to say why is important to discriminate between glaze and rime conditions.
2. In Line 215 Is A_overlap(x) no O_verlap.
3. Figure 3 is low quality. I would substitute it for other of higher quality
4. Line 246: Why in the case of rime ice Mie scattering is dominant? And glaze? I think a larger explanation is necessary.
5. I saw that you used the Messinger model for the heat flux in the surface. This model uses some a notation that is outdated (the paper is of the 50’s) I recommend you using the LEWICE software notation.
6. I don’t understand why the LWC uncertainties are of a 12%. The uncertainties of LWC depends on the measurement technique of the test. Why did you pick an external reference for your LWC uncertainty?
7. Figure 5 should have a higher resolution. It is difficult to read the legend.
8. Figure 6 needs more resolution as well. My recommendation is to express the results in 2D with different styles or colors.
9. In the line 493 is g/m3. It is a quite big LWC, why did you choose it?
10. Figure 8 needs more resolution. It is convenient a 2D plot and a more detailed description of the position each fiber number would be interesting.
11. In line 407, splashing and bouncing is a typical effect of SLD. The reason why the freezing fraction is less than one Is thermal for 20um droplets (the ones that you used). The horn are caused by other effect.
12. It would be interesting to compare the data obtained with glaze, mixed and rime ice
13. Why does the curves between 11a and 11b are so different?
14. Some parts of section 5.4 were already explained before, so they are redundant.
15. For the experimental results is necessary to say the dimensions of the tunnel and of the NACA 0012.
16. In the line 685 you said that you used the same collection efficiency than NASA Glenn tunnel with a identical NACA 0012. But the collection efficiency value depends on the airspeed and droplet size. It depends on the droplet distribution as well.
17. I would delete the figure 14a because is redundant. Figure 14 needs more resolution.
18. In line 727 you said that parameters a and b can be correlated with the freezing fraction. I don’t understand why this correlation happens and I don’t know the physical reason of it. Is it a semi-empirical correlation?
Lastly, the freezing fraction is a tool that can be used for ice accretion prediction, but it is not a physical variable as itself. There are many variables that can affect your results like droplet size in my opinion. For example you can have a rime test (freezing fraction of 1) and have different droplet distributions, and the ice layer surface will have different roughness.
Author Response
Reviser 1 Comments and Suggestions for Authors
I would like to thank the reviewers for reviewing this manuscript and their comments. All changes are indicated in red text.
Dear writer,
I think that the paper is interesting but some things could add more value to the paper. I will give you recommendations or opinions more than corrections. For example, in the introduction I would add more references for ice sensor technologies and I would compare your sensor with others (advantages and disadvantages). A common reference is Jackson and Goldberg paper:
-Jackson, D.G., & Goldberg, J.I. (2007). Ice Detection Systems: A Historical Perspective.., & Goldberg, J.I. (2007). Ice Detection Systems: A Historical Perspective.
-And additionally you can add more references of optic fiber ice sensors like: Icing Condition Predictions Using FBGS, M Gonzalez M Frovel
These are good suggestions and these references have been added (page 1,2)
I will give you another recommendations:
- In the paragraph of the line 98, there is a definition of the freezing fraction and its importance. I think, even thought I think you have explained it, to say why is important to discriminate between glaze and rime conditions.
Indeed it is expanded latter on in the paper but I think the reviewer has a good point, as it will highlight the concept and hence it has been addressed in this paragraph. (page 2)
- In Line 215 Is A_overlap(x) no O_verlap.
Corrected, thank you.
- Figure 3 is low quality. I would substitute it for other of higher quality:
This was probably my mistake as all the figures seem to be of very poor quality, (which was not the case with the word version of the summation). It is now corrected.
- Line 246: Why in the case of rime ice Mie scattering is dominant? And glaze? I think a larger explanation is necessary:
Although this is explained in more detail latter in the paper perhaps it is good to highlight the point here too. This was done in p. 7
- I saw that you used the Messinger model for the heat flux in the surface. This model uses some a notation that is outdated (the paper is of the 50’s) I recommend you using the LEWICE software notation. I used the notation from ref 20 (which uses Messiegers notation) :
I hope this is not a problem as the comparison is based on this reference (now [22] ) which was conducted in the NASA tunnel.
- I don’t understand why the LWC uncertainties are of a 12%. The uncertainties of LWC depends on the measurement technique of the test. Why did you pick an external reference for your LWC uncertainty?
In order to compare the results from D. N. Anderson, J.Ching Tsao Ref [22] I used the notation and results from this paper and calculated the FF for their wing which had almost the same dimensions as our wing. By virtue that the GKN tunnel had in the past, along with other European icing facilities, conducted a "round the mill" calibration tests using the same NACA 0012 wing and their icing tests results were very similar to those of the NASA-Glenn facility, it was decided to use their calculation method to estimate the FF for both wings. Therefore in this brief theoretical outline I mealy quote their results. This point is highlighted in p. 8
- Figure 5 should have a higher resolution. It is difficult to read the legend.
Again this was my mistake as all the figures seem to be of very poor quality, (which was not the case with the word version of the summation). It is now addressed.
- Figure 6 needs more resolution as well. My recommendation is to express the results in 2D with different styles or colors.
Again this was my mistake as all the figures seem to be of very poor quality, (which was not the case with the word version of the summation). This is a good point and the 3D has been converted to 2D.
- In the line 493 is g/m3. It is a quite big LWC, why did you choose it?
Although in the experiments we used LWC which ranged from 0.5 to 2 gm/m3 here we focus on our results for LWC of 1 gm/m3 to facilitate the comparison of the FF with those in [22]. To clarify this point…..
- Figure 8 needs more resolution. It is convenient a 2D plot and a more detailed description of the position each fiber number would be interesting.
Again this was my mistake as all the figures seem to be of very poor quality, (which was not the case with the word version of the summation). A 2 D version has been used and a better description is added.
- In line 407, splashing and bouncing is a typical effect of SLD. The reason why the freezing fraction is less than one Is thermal for 20um droplets (the ones that you used). The horn are caused by other effect.
I would like to thank the reviewer for this comment about splash back which is the dominate mechanism for SLD ice. (line 507). He is quite write and it is my mistake as the horns are due to the adiabatic expansion and cooling of the air above and below the leading edge causing ice to accrete in this regions of the wing . Its is corrected in the test (p.13)
- It would be interesting to compare the data obtained with glaze, mixed and rime ice
This is also a good suggestion and section 5.2 has been re-written with Fig 9a to d showing comparative graphs
- Why does the curves between 11a and 11b are so different?
One is at -200 C with the ice being nearly rime and the other at -250 C were the ice was totally rime. Some additional text highlight the point in page 13-14 and now Fig 11a,b are shown as a comparison in Fig 9a to 9d. This point is addressed in p.15
- Some parts of section 5.4 were already explained before, so they are redundant.
5.4 has been modified to eliminate repetitions and to address also point 18.
- For the experimental results is necessary to say the dimensions of the tunnel and of the NACA 0012.
The dimensions of the wing were given in section 5.4 p20 but there were added in section 3 p. 9 together with the dimensions of the icing tunnel (76 cm width by 51 cm height ).
- In the line 685 you said that you used the same collection efficiency than NASA Glenn tunnel with a identical NACA 0012. But the collection efficiency value depends on the airspeed and droplet size. It depends on the droplet distribution as well.
This is a good point which needs clarification. As it was mentioned in the experimental procedure section 3 we used very similar conditions in the icing runs (LWC 1g/m3, and 20 μm MVD droplets, and 150 Knots). Furthermore the shadowgraph laser beam could also used the measure the LWC (after suitable calibration) using the forward scattering of the beam from the cloud (Fig 4c). This technique can give a linear relation of the density of cloud, which directly related to the LWC and the droplet sizes. By scanning the shadowgraph beam along the leading edge of the wing in the vicinity of the fiber sensor it was possible to monitor qualitatively the cloud distribution in the beginning of icing runs. This technique was subsequently developed as a diagnostic tool for measuring the LWC in the tunnel but it is not reported in this work. To highlight the point a paragraph was added to in p. 10 and p. 19.
- I would delete the figure 14a because is redundant. Figure 14 needs more resolution.
14a has been deleted and all figures resolution has been resolved
- In line 727 you said that parameters a and b can be correlated with the freezing fraction. I don’t understand why this correlation happens and I don’t know the physical reason of it. Is it a semi-empirical correlation?
This is a good question and perhaps this point need better clarification. One of the aims of this work was to use the experimental ice growth curves to establish a relation between optical properties of ice, which are dependent on the way ice accretes on the wing, and the FF. Based on observation and the experiments on the optical diffusion, described in section 5.3, it was hypothesized that the district ice growth curves, shown in Fig 9, measure the partial reflection and diffusion of light reaching the fiber array. From fiber ray optic modeling, in section 4, and observations it was hypothesized that the two areas which are representative of the peak reflections and scattering contributions from the ice, are the points a and b, the choice of which is outlined in section 5.4. Furthermore as the FF is based on the physical process of freezing it was hypothesized (empirically) that it also affects the concentrations of the micro-bubbles trapped in the ice volume and this is the connection investigated in this paper. To clarify this point the text has amended throughout this paper and in particular in sections 4 and 5.
Lastly, the freezing fraction is a tool that can be used for ice accretion prediction, but it is not a physical variable as itself. There are many variables that can affect your results like droplet size in my opinion. For example you can have a rime test (freezing fraction of 1) and have different droplet distributions, and the ice layer surface will have different roughness.
I agree with the reviewer and the above statement is correct. One of the aims of the proof-of-concept paper is to investigate a way to determine the FF from the ice growth curves of this optical ice detector. Certainly more work is required to address this issue and perhaps investigate additional parameters which can improve the diagnostic capability of a system.
Reviewer 2 Report
The paper is about an optical fiber sensor for the ice measurement in terms of thickness, kind and Freezing Fraction (FF). The scientific soudness of this reasearch is really high since it opens a door in the tricky world of the aircraft anti-ice systems.
However, I found highly complicated to judge the manuscript since the most part of images are unreadable.
Due to the very poor quality (unacceptable for the journal) I can't associate the pictures to the text. This lead to me on a very difficult judgement on the manuscript contents even if it seems very promising. Moreover, also when the plots are in high quality, they are presented with typos, without any attention to details.
About the contents, I have a question that is more a curiosity. This sensor gives detailed information about the kind of ice. But, compared with a classic FBG sensor giving in output a temperature value and so just the presence or not about the ice, which is the difference? Isn't this latter enough?
Author Response
Reviewer 2
- The paper is about an optical fiber sensor for the ice measurement in terms of thickness, kind and Freezing Fraction (FF). The scientific soudness of this reasearch is really high since it opens a door in the tricky world of the aircraft anti-ice systems.
However, I found highly complicated to judge the manuscript since the most part of images are unreadable.
I would like to thank the reviewer for his effort and apologies for the quality of the figures. This was probably my mistake as all the figures seem to be of very poor quality, (which was not the case with the word and PDF version of the summation). It is now addressed.
- Due to the very poor quality (unacceptable for the journal) I can't associate the pictures to the text. This lead to meon a very difficult judgement on the manuscript contents even if it seems very promising. Moreover, also when the plots are in high quality, they are presented with typos, without any attention to details.
I must apologize for these omissions. They have been corrected.
- About the contents, I have a question that is more a curiosity. This sensor gives detailed information about the kind of ice. But, compared with a classic FBG sensor giving in output a temperature value and so just the presence or not about the ice, which is the difference? Isn't this latter enough?
This is a good question which is addressed in p 3. In the initial submission I have omitted the work on FBG as ice sensor which is now included as ref. [19] p.2. However temperature alone can only determine ice presence of but not its thickness or its type, both of which are important information to a de-icing system. For example one of the advantages of a de-icing system, with information on the ice thickness and type would be the ability to shed the ice rather than melt, thus reducing power conception. This is addressed in p. 3
Round 2
Reviewer 1 Report
I think the manuscript is right
Author Response
Thank you for our input
Reviewer 2 Report
The article contents are ok. The plots have been improved but in my opinion are still not compliant with the journal: there are a lot of typos and mismatches among axes title.
Author Response
The reviewer is quite write ant all have now been corrected and I would like to thank the reviewer for his input
