Early Detection and Identification of Damage in In-Service Waterworks Pipelines Based on Frequency-Domain Kurtosis and Time-Shift Coherence

Round 1

Reviewer 1 Report

Review of the Manuscript ID: water-2278575

"Early Detection and Identification of Damage in In-service Waterworks Pipelines based on Frequency-Domain Kurtosis and Time-Shift Coherence"

Authors: Sun-Ho Lee, Choon-Su Park, Dong-Jin Yoon

Submitted to section: Urban Water Management

The authors of the manuscript have developed a method that allows in some situations to prevent the emergency destruction of underground water pipelines by recognizing in advance the construction work in dangerous proximity to the pipeline. The method is based on the analysis of wave processes in the fluid flow. The practical significance of the work becomes obvious if we consider the scale of pipeline communications and the relative fraction of anthropogenic impacts in the causes of their destruction.

The authors managed to find unexpected criteria for recognizing shock effects on the pipeline by the wave reaction of the flow. The authors have the right not to reveal the ways that ensured their success. But it would be interesting to know how they managed to come to the discovery of the original method. Their work was preceded by a rather numerous series of papers (see references in the manuscript) using the statistical analysis of random wave processes. It turned out that the waves formed during shock impacts on the ground can be distinguished from the ensemble of wave processes caused by other actions.

I think the material is interesting and worthy of publication.

I would like to know how suitable the developed method is for pipelines of other functional purposes (oil and oil product pipelines, gas pipelines).

Questions and comments on the text of the manuscript.

a)    Line 145.  average radius. It should be explained why averaging is needed, how it is carried out.

b)    Line 146.  It should be explained what is angular speed.

c)    Lines 159 – 161.   «Here, the material properties and specifications of all pipes were in accordance with the American Society of Mechanical Engineers standard ASME B36.19M». It is not clear what is being said. Who conducted the experiments? Why do they need ASME B36.19M compliance information?

d)    Lines 173 – 175. «The coherence function, which measures the similarity between two signals, may not be able accurately to measure this in cases where other signals, such as transient signals, may exist in different time domains [21]».   How to understand it?

e)    Page 5. Figure 3. The functions x(t), y(t) are introduced for the first time without explaining what they mean.

Figure 3 requires more detailed comments.

«It is assumed». It is assumed? What did the authors really want to say?

How did you manage to single out the shock signals represented in red?

Line 186. «The impact signals are represented in red, with a center frequency of 100 Hz and a time difference of arrival of 0.3 seconds». Why?

And further, what is 100 Hz band? If a band, what is its width?

f)       Line 203. Formula (3) cannot be understood without explanation. Kurtosis is distribution kurtosis? What random variable? What is X_stift? If the formula cannot be understood without referring to reference literature, then why should it be given in the text of the manuscript?

g)    Lines 304 – 307. «Therefore, as shown in Figure 13(b), it can be seen that it is difficult to distinguish the impact signal from the noise due to the low signal-to-noise ratio in the raw signal». If the autobahn influences so much, then the question arises about the influence of railway and tram lines, the construction and repair of buildings, roads, etc. What are the limitations in applying the methodology?

h)    Lines 349 – 351.  «The error may be attributed to a variety of factors, but it is primarily thought to result from the use of theoretical speed values, as well as from the excavation activities being conducted within a range of about 10 meters near a pre-established distance of 80 meters». Estimating the error, one should not lose sight of the fact that the arrival time difference (0.085 s) is also unreliable and is determined “by eye”.

i)      Line 370. «Figure 18 (b) presents an excavation signal propagated 80 m, and it can be seen that the ground-borne wave arrives approximately 0.57 seconds later». Is it possible to determine the time with such accuracy (up to hundredths of a second), given that measurements are always made with errors?

j)      Lines 396 – 397.  «However, in Figures 19(e) and (f), it can be observed that the sensor 2 signal arrives earlier because the distance from the impact to sensor 2 is shorter than the distance to sensor 1». On fig. 19(e) I failed to find it.

 

 

 

 

Design notes.

1)    Abbreviations are introduced, which are then not used: ML (line 127), SCOT (line 128).

2)    Poorly designed list of references.

3)    Page 4. On the page 4 first is mentioned the average radius ρ, and then the diameter. 2ρ = diameter?

4)    Line 322. The abbreviation ТС is not deciphered.

5)    Рис. 19. There is no indication of what is shown on the left and what is on the right side of the Figure.

 

Author Response

Thank you for your thorough and detailed review comments. Please find the attached document.

Author Response File: Author Response.docx

Reviewer 2 Report

Overall, a well formulated manuscript, with a useful algorithm for an important application. I only have the following minor comments that I would like addressed prior to being accepted for publication.

1. I would recommend referencing the following work in the introduction of your manuscript: “Chronic leak detection for single and multiphase flow: A critical review on onshore and offshore subsea and arctic conditions” by Behari et. al. The cover a lot of literature that has overlap with the content of your manuscript and could be relevant for readers who may be interested in expanding the literature search to the oil and gas industry with leak detection for arctic and subsea pipelines.

2. I do not see much of a discussion on how the results might vary for different pipe materials and thicknesses. I would recommend providing a discussion on this, and if possible supplementary simulations in order to corroborate the discussion. I believe this would strengthen the contributions presented in the manuscript.

Author Response

Thank you for your thorough and detailed review comments. Please find the attached document.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Revised manuscript review (for editorial)

“Early Detection and Identification of Damage in In-service Waterworks Pipelines based on Frequency-Domain Kurtosis and Time-Shift Coherence” (water-2278575).

The present version of the manuscript clarifies some of the provisions that were unclear in the original version. The manuscript has benefited from corrections. Some points that were misinterpreted by the authors, I indicated in the review for the authors. Although the manuscript, in my opinion, has some weaknesses, a second correction is hardly advisable. I think that the work should be published in a journal Water.

13.02.2023.

For authors Some remarks

Dear Reviewer #1

Thank you for your valuable comments and questions on the manuscript, “Early Detection and Identification of Damage in In-service Waterworks Pipelines based on Frequency-Domain Kurtosis and Time-Shift Coherence” (water-2278575). We revised the paper according to your comments. Followings are answers to your questions and comments.

Your comments & questions:

a)    Line 145.  average radius. It should be explained why averaging is needed, how it is carried out.

b)    Line 146.  It should be explained what is angular speed.

c)    Lines 159 – 161.   «Here, the material properties and specifications of all pipes were in accordance with the American Society of Mechanical Engineers standard ASME B36.19M». It is not clear what is being said. Who conducted the experiments? Why do they need ASME B36.19M compliance information?

d)    Lines 173 – 175. «The coherence function, which measures the similarity between two signals, may not be able accurately to measure this in cases where other signals, such as transient signals, may exist in different time domains [21]».   How to understand it?

e)    Page 5. Figure 3. The functions x(t), y(t) are introduced for the first time without explaining what they mean.

Figure 3 requires more detailed comments.

«It is assumed». It is assumed? What did the authors really want to say?

How did you manage to single out the shock signals represented in red?

Line 186. «The impact signals are represented in red, with a center frequency of 100 Hz and a time difference of arrival of 0.3 seconds». Why?

And further, what is 100 Hz band? If a band, what is its width?

f)       Line 203. Formula (3) cannot be understood without explanation. Kurtosis is distribution kurtosis? What random variable? What is X_stift? If the formula cannot be understood without referring to reference literature, then why should it be given in the text of the manuscript?

g)    Lines 304 – 307. «Therefore, as shown in Figure 13(b), it can be seen that it is difficult to distinguish the impact signal from the noise due to the low signal-to-noise ratio in the raw signal». If the autobahn influences so much, then the question arises about the influence of railway and tram lines, the construction and repair of buildings, roads, etc. What are the limitations in applying the methodology?

h)    Lines 349 – 351.  «The error may be attributed to a variety of factors, but it is primarily thought to result from the use of theoretical speed values, as well as from the excavation activities being conducted within a range of about 10 meters near a pre-established distance of 80 meters». Estimating the error, one should not lose sight of the fact that the arrival time difference (0.085 s) is also unreliable and is determined “by eye”.

i)      Line 370. «Figure 18 (b) presents an excavation signal propagated 80 m, and it can be seen that the ground-borne wave arrives approximately 0.57 seconds later». Is it possible to determine the time with such accuracy (up to hundredths of a second), given that measurements are always made with errors?

j)      Lines 396 – 397.  «However, in Figures 19(e) and (f), it can be observed that the sensor 2 signal arrives earlier because the distance from the impact to sensor 2 is shorter than the distance to sensor 1». On fig. 19(e) I failed to find it.

 

 

The answers are followings; the answer is in blue and revised or added sentences are in red.

Answer a)

        Thank you for your comment. We appreciate the reviewer's comments regarding the terminology used in the manuscript. After reviewing additional references [1,2,3] in the field of pipes, we have determined that "mean pipe radius" is a more appropriate term than "average radius" for pipes. Therefore, we have modified the sentence to use "mean pipe radius" in order to accurately convey the intended meaning of the radius calculated based on the average value of the outer and inner diameters of the pipe.

Уважаемые коллеги. Мой вопрос был не о терминологии. Что такое средний? Полусумма внешнего и внутреннего? Трубопровод может конструктивно состоять из труб разного диаметра. Диаметр меняется из-за отложений на стенке. Осреднение возможно по времени, по секции трубопровода. Я предполагал, что уточнение не будет лишним.

Dear Colleagues. My question was not about terminology. What is an average? Half the sum of external and internal? The pipeline can be structurally composed of pipes of different diameters. The diameter changes due to deposits in the pipe. Averaging is possible over time, over a pipeline section. I assumed that clarification would not be superfluous.

Additional Reference
[1] Deckmyn, G., Evans, S. P., & Randle, T. J. (2006). Refined pipe theory for mechanistic modeling of wood development. Tree Physiology, 26(6), 703-717.
[2] Gross, N., & Ford, H. (1953). The flexibility of short-radius pipe-bends. Proceedings of the Institution of Mechanical Engineers, 167(1b), 480-509.
[3] Xu, J. J., Sun, B. C., & Koplik, B. (2000). Local pressure stresses on lateral pipe-nozzle with various angles of intersection. Nuclear engineering and design, 199(3), 335-340.

(In Sec. 2.1)

where c_f, B_f, a, E, ρ, h, and ω denote the longitudinal wave speed of a fluid in free space, the bulk modulus of the internal fluid, the mean pipe radius, which is calculated based on the outer and inner diameter of a pipe, Young’s modulus of the pipe, the density of the pipe material, the pipe thickness, and the angular speed, respectively.

Answer b)

We appreciate the reviewer's feedback regarding the terminology. The angular speed mentioned in the review was actually a typographical error where we incorrectly wrote angular speed instead of angular frequency. We have made the necessary corrections to the relevant section of the manuscript.

(In Sec. 2.1)

where c_f, B_f, a, E, ρ, h, and ω denote the longitudinal wave speed of a fluid in free space, the bulk modulus of the internal fluid, the average radius, Young’s modulus of the pipe, the density of the pipe material, the pipe thickness, and the angular frequency, respectively.

Answer c)

Thank you for your feedback. We understand that the mention of ASME B36.19M in our manuscript may not have been clear, and we appreciate your interest in understanding its importance. Therefore, we have made some modifications to provide a clearer explanation. The reason for mentioning ASME B36.19M is that different manufacturers may use their own material properties and dimensions when manufacturing pipes. As a result, even for pipes with the same diameter, differences in thickness can lead to variations in the quasi-longitudinal wave propagation speed shown in Figure 2. Therefore, we wanted to clarify the dimensions of the pipes used in our numerical calculations. We hope that this explanation addresses your concerns and appreciate your valuable feedback. Please find the revised manuscript below.

(In Sec. 2.1)
In other words, as the diameter of a fluid-filled pipeline increases, the initial speed of the quasi-longitudinal wave propagating through it decreases, and the dispersion curve becomes steeper. Here, all the specifications for the material and dimension of the pipe used in Figure 2 were based on the American Society of Mechanical Engineers standard ASME B36.19M.

 

Answer d)

In order to clarify the meaning of the current sentence, which is difficult to understand, the manuscript has been amended as follows. Therefore, we would like to make the following semantic changes to the manuscript to clarify the meaning.
 When comparing transient signals that exist at different times, the coherence function may produce inaccurate results. This is because the coherence function measures the degree of similarity between signals based on their phase information, and when transient signals exist at different times, their phase relationship is not stationary and changes with time, leading to an incorrect estimation of coherence.
 As a result, we have modified the manuscript to convey this meaning more accurately.

(In Sec. 2.2)

Due to the characteristics of fluid-filled pipelines, which determine the propagation form of energy, both TPI and leaks generate the same quasi-longitudinal wave. However, it is known that the shapes of the waveforms generated by TPI and by leakages differ significantly. Leak signals are typically continuous signals, while impact signals are typically transient signals. This difference suggests that algorithms designed for leak detection may not be suitable for detecting TPI and other impact damage. The reason for this is that while the leak signal is included over the entire time when observing PSD for the selection of the optimal frequency band, in the case of impact signals, it can only exist in part of the signal, leading to an inaccurate PSD. The coherence function measures the degree of similarity between two signals based on their phase information and assumes that the signals are stationary and have a constant phase relationship over time. However, if transient signals exist at different times, their phase relationship is not stationary and changes with time, which can lead to an incorrect estimation of coherence [21].

 

 

 

Answer e)

To address potential issues with the clarity of Figure 3 (a) and (b), which represent the virtual measurement signal simulations, we have revised the accompanying sentence to improve its clarity and ensure that the simulation situation is easily understandable for our readers

(In Sec. 2.2)

In summary, applying the PSD and coherence, which are known as observation data for selecting the optimal frequency band for leak detection [22], to impact signals, as shown in Figures 3 (a) and (b), may result in the phenomena discussed below. These signals, designed to reflect real scenarios, include white Gaussian noise as background noise, and the damage signal, represented in red, is designed as a damped sine wave with a center frequency of 100 Hz. It is assumed that the damage wave has a propagation speed of 1,000 m/s and occurs at a position 200 m away from a total length of 700 m. Therefore, the virtual measurement signal x(t) shown in Figure 3(a) is composed of white Gaussian noise and an impact signal that occurs around 0.2 seconds, while the virtual measurement signal y(t) shown in Figure 3(b) is designed as an impact signal occurring around 0.5 seconds, along with white Gaussian noise. Accordingly, the time difference between the impact signals present in the two signals is designed to be about 0.3 seconds.

 

. Mismatch between text and picture.

 

Answer f)

In response to your feedback, we have revised the manuscript by adding appropriate references and explanations to enhance the reader's understanding. As a result, we have modified the original equation based on the cited references. We believe that the revised equation, which has been modified based on the notation used in the cited references, will aid the reader's understanding. We have corrected the reference numbers [25, 26, 27, 28] to [26, 27, 28, 29], as new references have been included in the revised manuscript

(In Sec. 2.2)

FDK was proposed by Dwyer to detect transient signals that are generated by ice cracks [23]. Furthermore, it was demonstrated that FDK is more effective when used to detect transient signals than the power spectrum density, which is a representative indicator for observations of signal characteristics [24]. Afterwards, FDK was reintroduced by Ottonello after being reorganized [25]. FDK can be calculated by Equation (3),

 

                       (3)

 

Here, M represents the number of segments into which the signal samples have been divided for the purpose of estimating discrete Fourier transform (DFT), while  refers to the ith segment obtained after dividing the signal samples used in the estimation of DFT.

 

 

Answer g)

As you pointed out, measurement signals in actual field conditions can be affected by various types of noise. For example, noise from railway and tram lines can be considered as a continuous signal, similar to autobahn influences, and can be mitigated by applying the proposed method in this study. Additionally, both continuous and transient signals can be produced during the construction and repair of buildings. For instance, hammering produces a transient signal, and our experimental results have shown that such signals exhibit a higher attenuation rate than quasi-longitudinal waves transmitted through buried pipelines, indicating that the long-range propagation of the wave may not occur. However, if a transient signal is generated from construction and repair of buildings above the buried pipelines, it should be detected as it is one of the detection objectives of our study.

Answer h)

 We carefully read the manuscript again, and revised our manuscript and Figure 17 to address the issues raised. Specifically, the estimated arrival time difference was calculated using the maximum value of cross-correlation, which is a method used to estimate the time delay between two signals. However, upon reviewing Figure 17, we realized that the information was presented in a somewhat confusing manner. As a result, we have made appropriate changes to improve the clarity and readability of both the sentence and the figure to enhance reader comprehension.

(In Sec. 4)

Therefore, by applying the arrival time difference (0.085 s), the propagation speed (820 m/s), and the total distance (255 m) to Equation (1) to express the source location, the damage location can be calculated as 92.65 m. This result shows that there is an error of approximately 12.65 m from actual TPI. The error was calculated using the time delay obtained by the cross-correlation of the two signals, as shown in Figure 17. The error may be attributed to a variety of factors, but it is primarily thought to result from the use of theoretical speed values, as well as from the excavation activities being performed within a range of approximately 10 meters from the designated location. This experimental result has demonstrated that the detection method proposed in this study can provide early warnings before the failure of buried pipelines occurs, and can be effectively utilized under actual field conditions, which suggests that the method is reasonably practical.

 

 

 

Answer i)

We carefully read the manuscript again, and revised our manuscript and Figure 18 to improve reader comprehension. We have modified the sentence to provide a clearer explanation due to a potentially inappropriate choice of words, in order to improve clarity. To clarify, the purpose of Figure 18 is to illustrate that ground-borne waves exhibit a lower speed compared to quasi-longitudinal waves. Therefore, we have revised the relevant sentence in the manuscript and made changes to Figure 18 to prevent any confusion and improve reader understanding. In this case, the time difference between different types of waves was calculated based on the first peak.

(In Sec. 4)

Additionally, when an impact is applied to the surface of buried pipeline area, a ground-borne wave can also propagate [25,26], as shown in Figures 18(a) and (b). The ground-borne wave refers to the wave that is generated when the impact energy is transmitted through the ground, and it is one of the noise sources that can cause errors in source location. In particular, since the ground-borne wave is a type of transient signal that has a different propagation speed from the wave that propagates through the pipeline, it can introduce noise in time-difference-of-arrival-based source location and needs to be filtered. Figure 18 (a) shows the measured signal from the sensor position 60m from the excavation site on the ground. As previously mentioned, quasi-longitudinal waves propagate at a speed of approximately 820 m/s and the ground-borne wave then arrives approximately 0.35 seconds later. Because the propagation distance is 60 m, it can be determined that the propagation speed of the ground-borne wave is approximately 141 m/s. Figure 18 (b) presents an excavation signal propagated 80 m, and it can be seen that the ground-borne wave arrives approximately 0.6 seconds later. Here, the time between the two wave groups was calculated based on the first peak, as shown in Figure 18 (a) and (b). Therefore, it can be estimated that the propagation speed of the ground-borne wave is approximately 114 m/s in this case. The ground-borne waves in the area where this experiment was conducted are known to propagate at approximately 200 m/s or less [27,28].

(a)	(b)

Figure 18. Time-domain signals obtained from each sensor including a ground-borne wave: (a) propagation distance of 60 m, and (b) propagation distance of 80 m

 

 

Answer j)

 We have modified Figure 19 to provide a clearer explanation of the time delay between the two waves. Therefore, we have revised both the figure and the manuscript to improve reader comprehension. We have made several modifications to Figure 19 to improve clarity, particularly in regards to the representation of the time delay between the two waves. We have added a red dashed line to indicate the arrival time of the waves, which is based on the first peak of the waves. Additionally, we have incorporated the feedback from Design note 5) and added sensor information to the figure. Our intention was to use Figure 19 to show how the time delay varies based on the location of the TPI, and we hope these modifications will aid in conveying that information to readers. Thank you for your meticulous review and valuable feedback.

 

(In Sec. 4)

Figures 19 (a~f) present signals obtained through excavation experiments conducted at various points. Here, the red dotted line indicates the location of the first peak of the quasi-longitudinal wave generated by the TPI. Figure 19 (a) shows the results of an excavation test at approximately 60 meters from the sensor 1 position. In this case, the signal that reaches sensor 2 propagated a distance of approximately 195 meters. As previously described, it can be confirmed that at a relatively short distance of 60 meters, the ground-borne wave propagated together with the quasi-longitudinal wave at a speed of approximately 140 m/s. However, this cannot be confirmed at sensor 2, as the ground-borne wave was fully attenuated. Figures 19 (b) through (f) present signals obtained from excavations at distances of 80, 100, 125, 150, and 180 meters from the position of sensor 1, respectively. Especially when observing the excavation signal at a location approximately 75 meters away from sensor 2 in Figure 19 (f), it can be observed that the ground-borne wave was measured concurrently, as shown in Figure 19 (a). As described above, these results suggest that the proposed  method contributes significantly to the selection of the optimal frequency bands for source location. Since the signals were aligned with reference to the sensor 1 signal in Figures 19(a-f), it can be observed that the time difference between the two signals gradually decreases. However, in Figures 19(e) and (f), it can be observed that the sensor 2 signal arrives earlier because the distance from the impact to sensor 2 is shorter than the distance to sensor 1.

 

Mismatch between text and picture.

Reviewer 2 Report

Paper is acceptable in current form.