Scour Propagation Rates around Offshore Pipelines Exposed to Currents by Applying Data-Driven Models
Round 1
Reviewer 1 Report
Various machine learning techniques are applied to experimental data of scour around pipe lines in order to generate predictive equations for (non-dimensional) scour propagation based on key parameters (such as relative depth of pipeline, Shields parameter, Froude number, etc.). While there may be some utility in the final expressions derived, these appear highly complex and the range of applicability is not clear, and furthermore often have numerous fitting parameters. Some techniques are clearly not very successful, but there is little critical analysis of why some techniques work better than others. There is limited discussion of the practical application of the results, statements of what might be regarded as the “best” expressions to use, or suggestions for how various aspects of this study might be taken forward. In addition, there several errors in the English and in some of the equations, that need addressing, see below.
Overall, this could be made clearer and improved in various ways.
Minor errors in English: a proof-reading by a proficient English speaker/writer should capture these (e.g. line 104 based on four (delete the); line 113 following, not wollowing, line 818 exhibited not exibithed)
Equations: some of these don’t display correctly: pi_3 in equation (3) has some overprinted Chinese characters on my version, repeated in some subsequent equations – e.g. (7), and in that equation Re_C lacks the R.
Some equations are incorrect: pi’_8 in equation (5) looks right, but the Shields parameter term in equation (7) has a square root that shouldn’t be there. V*_L in equation (6) and (7) should just have d_50, not the cube of it. In equation (13) you don’t define omega (r). In Table 4, BF14 and BF15 are the same.
In Figure 5 some of the labels and some of the symbol annotations are missing or incomplete.
Author Response
Please find attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
From the obtained result of this manuscript, the equations given by ML models provided reliable predictions of scouring propagation rate. If the results can be added some validation with field cases, it will be much better for the ML ｍｏｄｅｌ ａｐｐｌｉｃａｔｉｏｎ．
Ｓｏｍｅ ｔｙｐｉｎｇ ｏｆ ｆｏｒｍｕｌａｔｉｏｎ ｅｑｕａｔｉｏｎｓ ｃｏｕｌｄ ｂｅ ｃｏｒｒｅｃｔｅｄ ｆｏｒ ｉｍｐｒｏｖｅｍｅｎｔ．
Author Response
Please find attachment
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors have addressed many of the points raised so the paper is considerably improved. Below are a few minor errors to correct, by line number.
94 provided more accurate
104 to provide effective performance
111 deriving empirical [I think you mean?]
129 due to gravity
224 Just two.. [or "Only two.." but not "Just only two.."]
299 To reduce the complexity
599 classification concepts
Author Response
[Q] The authors have addressed many of the points raised so the paper is considerably improved. Below are a few minor errors to correct, by line number.
94 provided more accurate
104 to provide effective performance
111 deriving empirical [I think you mean?]
129 due to gravity
224 Just two.. [or "Only two.." but not "Just only two.."]
299 To reduce the complexity
599 classification concepts
[Reply] All the above-mentioned corrections were made. Please view highlights.
