Acoustic Emission Based on Cluster and Sentry Function to Monitor Tensile Progressive Damage of Carbon Fiber Woven Composites

Round  1

Reviewer 1 Report

Authors made substantial changes to the manuscript. Few minor things

Fig. 1 title is not correct (this is not a schematic; authors may keep both schematic and original sample image)

Fig. 7 title is confusing. it can be written as 

Fig. 7. (a)The image of the whole fractured sample. (b)-(c) SEM micro-photographs of the fractured of composite: delamination and matrix crack; fiber breakage and fiber pull out. 

I am recommending the manuscript for publication. However, very careful proofreading is necessary before publication.

Author Response

Response to Reviewer 1 Comments

Point 1: English language and style are fine/minor spell check required.

Response 1: Thanks! The English has been checked by a well-versed colleague. Obvious wrong syntaxes have been revised.

Point 2: Fig. 1 title is not correct (this is not a schematic; authors may keep both schematic and original sample image).

Response 2: Thanks for this good suggestion! Figure 1 and the title have been revised.

Fig. 1. (a) The schematic illustration of weave composite. (b) Images of composite specimens.

Point 3: Fig. 7 title is confusing. it can be written as

Fig. 7. (a)The image of the whole fractured sample. (b)-(c) SEM micro-photographs of the fractured of composite: delamination and matrix crack; fiber breakage and fiber pull out.

Response 3: Thanks for this kind suggestion! The title has been changed to “Fig. 7. (a)The image of the whole fractured sample. (b)-(c) SEM micro-photographs of the fractured of composite: delamination and matrix crack; fiber breakage and fiber pull out.”

Author Response File: Author Response.pdf

Reviewer 2 Report

According to the response of the authors to my first review of their submitted paper, some of my queries and suggestions have been addressed satisfactorily. However, some of them need to be clarified better, while some new corrections have to be made. Specifically:

The captions of Figures 1 and 2 need to be corrected.

Lines 194-198: The addition about the linear source location filter is not quite explanatory. Please be more precise and give appropriate references.

Figure 5: The presented recorded signal may correspond either to 2 overlapping waveforms which are not separated as distinct events due to high value of HDT and/or low threshold, or may be the waveform of a single AE event. In the first case, amplitudes and durations have not been determined correctly. My suggestion was to present 2 distinct waveforms that correspond to 2 events (points) from Figure 5b, one of low duration and high amp (bottom right side on the graph) and the other of low amp and high duration (upper left side on the graph). Also, the vertical scale is Voltage not Amplitude.

Figure 8: My comment is not answered satisfactorily. Is there any relation of the inflection points to damage mechanisms, as it is mentioned by the authors in Lines 267-269?

Figure 9: The authors in their response claim that energy and hit rate can distinguish deformation, cracks propagation and instability fracture in laminated specimens, but they not provide any evidence for that. So, I insist on my initial suggestion to authors to incorporate this info in Figure 8.

Lines 306-307 (274 of the 1st version): Outside the operation frequency range (100-900 kHz) the spectral sensitivity of the sensors drop dramatically, so I am not convinced that frequencies below 50 kHz should be safely related to a specific damage mechanism in the present case. How did you calculate the frequency distribution in Figure 10(b)? Are these the mean values at frequency intervals of 20 kHz?  

Lines 367-369: Use italics for all symbols. The symbol most often used for density is ρ, i.e. the lower case Greek letter rho, not p.

I also draw attention to the authors for the correct use of the English language throughout the manuscript.

Author Response

Response to Reviewer 2 Comments

Point 1: Extensive editing of English language and style required.

Response 1: Thanks for this kind suggestion! The English has been checked by a well-versed colleague. Obvious wrong syntaxes have been revised.

Point 2: The captions of Figures 1 and 2 need to be corrected.

Response 2: Thanks for this kind suggestion! New Figure 1 and the title have been revised.

Fig. 1.(a) The schematic illustration of weave composite. (b) Images of composite specimens.

The caption of Figures 2 has been changed to “Fig. 2. Experimental system of tensile tests and AE monitoring”.

Point 3: Lines 194-198: The addition about the linear source location filter is not quite explanatory. Please be more precise and give appropriate references.

Response 3: Thanks for this good suggestion! “In the present research, two filtering procedures are focus on how to reduce the effects of noises. First, based on a linear source location filter, preliminary screening limits the mechanical noises and ensure the most AE signals originating from the material damage. The operation of a linear source location filter is a convolution operation that can be implemented using fast Fourier transforms and other fast algorithms. However, sparse false signals still remain in the datasets after this procedure.” has been changed to “Maillet et al. [39] proposed an energy-based approach by using two sensors, which can effectively obtain the accurate location and selection of AE signals originated from the damage of the composites. In order to ensure that most AE signals were generated in the gage section, the AE signals were monitored with two identical sensors being 55 mm apart. From the times of arrival at top (t1) and bottom (t2) sensors, the time difference is Δt (< Δtmax). It could be concluded that most of AE signals were generated in the gage section during the tensile process.” in the paragraph 2, section 3.1.

Point 4: Figure 5: The presented recorded signal may correspond either to 2 overlapping waveforms which are not separated as distinct events due to high value of HDT and/or low threshold, or may be the waveform of a single AE event. In the first case, amplitudes and durations have not been determined correctly. My suggestion was to present 2 distinct waveforms that correspond to 2 events (points) from Figure 5b, one of low duration and high amp (bottom right side on the graph) and the other of low amp and high duration (upper left side on the graph). Also, the vertical scale is Voltage not Amplitude.

Response 4: Thanks for this kind suggestion! “Second, the high correlated between amplitude and duration (i.e., low-amplitude emission corresponding to short-duration and high-amplitude emission corresponding to longer-duration) should results in concentrate in a banded region [39]. The typical AE signal waveform is shown in Fig. 5 (a). V1 represents a high amplitude signal greater than 1V (80dB) for a shorter time. In contrast, the signal represented by V2 has lower amplitude but longer time. On basis of relationship, the amplitude-duration filtering was used in separating outliers from amplitude-duration scatter plot. Chai et al. [40] and Yu et al. [41] used amplitude-duration filtering to remove outliers in amplitude-duration scatter plot in the crack growth behavior of steel members. In this step, to mitigate the effect of noise, statistical pattern recognition technique of the Maxmin Distance algorithm was employed to locate outliers representing ‘fake’ AE signals [42]. The Maxmin Distance algorithm is a tentative-based algorithm in the field of pattern recognition. The purpose is to select objects that are far apart from each other as the clustering center, avoiding the situation that the clustering seeds is too close when the initial value of the k-means algorithm is selected. This algorithm not only intelligently determines the number of initial cluster seeds, but also improves the efficiency of dividing the initial data set. The outliers are abandoned from the data set, and then the typical plots of duration versus amplitude after eliminating outliers are shown in Fig. 5 (b).” has been changed to “In addition, Babu et al. [40] concluded that there is a good correlation between amplitude and duration of AE signals. Similarly, Chai et al. [41] and Yu et al. [42] used amplitude-duration filtering to remove outliers of AE signals. Therefore, a statistical pattern recognition technique of the Maxmin Distance algorithm was employed to locate outliers [43]. The Maxmin Distance algorithm is a tentative-based algorithm in the field of pattern recognition. The purpose is to select objects that are far apart from each other as the clustering centre, avoiding the situation that the clustering seeds are too close when the initial value of the k-means algorithm is selected. The typical AE waveforms are shown in Fig. 5(a). The high amplitude and low duration (in the range of 0.00008s) waveform at the top of Fig. 5(a) corresponds to AE events (points) in the bottom right side on the Fig 5(b). The low-amplitude (near the threshold) and high duration (reaches up to 0.0003s) waveform at the bottom of Fig. 5(a) corresponds to AE events (points) in the upper left side on the Fig 5(b).” in the paragraph 2, section 3.1.

Figure 5 and the title have been revised.

Fig. 5. (a) Typical AE waveforms. (b) The relationship between amplitude and duration of AE signals

Point 5: Figure 8: My comment is not answered satisfactorily. Is there any relation of the inflection points to damage mechanisms, as it is mentioned by the authors in Lines 267-269?

Response 5: Thanks for this good suggestion! “Meanwhile, it also shows that the damage crack is generated and continues to expand, and the internal damage of the laminate specimen is weak but the range is extensive.” has been changed to “At the first inflection point, AE energy increases sharply, and the energy release rate and AE counts are also relatively high, which attributes to miro-damage and matrix cracking in composites. Then, AE energy remains at a high level and AE counts increase gradually, which are related to the propagation of miro-damage and interfacial failure. As reaching the second inflection point, AE energy increases, and AE counts and count rate increase sharply. It can be concluded that the fiber/matrix debonding and more serious damage are induced in composites.” in the paragraph 1, section 3.3.

Point 6: Figure 9: The authors in their response claim that energy and hit rate can distinguish deformation, cracks propagation and instability fracture in laminated specimens, but they not provide any evidence for that. So, I insist on my initial suggestion to authors to incorporate this info in Figure 8.

Response 6: Thanks for this kind suggestion! The original Figure 9 has been deleted, and information on the AE energy release rate and the AE event count rate has been added into Figure 8.

Fig. 8. Tensile load, accumulative energy, counts and energy release rate, AE event count rate vs. time for the composite specimen

Point 7: Lines 306-307 (274 of the 1st version): Outside the operation frequency range (100-900 kHz) the spectral sensitivity of the sensors drop dramatically, so I am not convinced that frequencies below 50 kHz should be safely related to a specific damage mechanism in the present case. How did you calculate the frequency distribution in Figure 10(b)? Are these the mean values at frequency intervals of 20 kHz? 

Response 7: Thanks for this kind suggestion! “However, the progressive peak frequency distribution at different percentage of loads is varied. From Fig. 10(b), there is no dominant AE event at 30% of load while the dominant peak frequency distribution lies from 0 to 50 kHz at the 60% of the failure load. With the tensile load increase up to 90% of load, the main range from 200 to 300 kHz is steep increase in the no. of event. More importantly, the load reached its maximum safety limit accompany with that the appearance of high peak frequency signals exceeding 300 kHz, indicating severe damage to the material structure.” has been changed to “Furthermore, the peak frequency distribution of AE signals at different load level is shown in Fig. 9(b), the mean values of frequency intervals is 50 kHz. There is no dominant AE event at 30% of failure load. Outside the operation frequency range (100-900 kHz) the spectral sensitivity of the sensors drop dramatically, the frequencies below 50 kHz at the 60% of the failure load are not well related to a specific damage mechanism. As the tensile load increases up to 90% of failure load, the main frequencies range from 200 to 300 kHz. More importantly, when the load reaches up to failure load, AE signals with high peak frequency over 300 kHz are generated, indicating severe damage in composites.” in the paragraph 3, section 3.3.

Point 8: Lines 367-369: Use italics for all symbols. The symbol most often used for density is ρ, i.e. the lower case Greek letter rho, not p.

Response 8: Thanks for this good suggestion! All symbols have been revised. The symbol of density has been revised to ρ in the paragraph 5, section 3.4.

Point 9: I also draw attention to the authors for the correct use of the English language throughout the manuscript.

Response 9: Thanks for this kind suggestion! The English has been checked by a well-versed colleague. Obvious wrong syntaxes have been revised.

Author Response File: Author Response.pdf