A Big Data Approach for Investigating Bridge Deterioration and Maintenance Strategies in Taiwan
Round 1
Reviewer 1 Report
This study uses a knowledge discovery process in a database to investigate big data stored in a bridge management system in Taiwan. Applying machine learning algorithms to inventory and multi-year inspection data from nearly 3,000 bridges, key structures were identified. Bridge managers can use these results to design maintenance strategies. This study could be accepted before revisions.
1. Figure 3 has a low resolution, it is suggested to have a high-resolution one.
2. The introduction part has a deep review of the BMS and some assessment method of distributed bridges. It is suggested to add some network-level bridge assessment and management method with the state-of-the-art techniques, such as, https://doi.org/10.1002/stc.2915, https://doi.org/10.1007/s00158-022-03210-3, https://doi.org/10.1061/(ASCE)BE.1943-5592.0001774, and https://doi.org/10.1061/(ASCE)BE.1943-5592.0001815
3. Data is the key for the analysis. It is suggested to illustrate the number of attributes that existed in this database, and what is the importance ranking of them for the assessment.
4. The main techniques for the assessment is cluster analysis and association analysis. How to choose these two methods for the assessment?
5. Authors mentioned that it could concentrate their limited resources on bridges. This study did not contains the cost evaluation part. How to consider the resources facing with the deteriorations.
6. The suggestion of improvement on this system could be added.
Author Response
- Comment: Figure 3 has a low resolution. It is suggested to have a high-resolution one.
Response: The figure has been updated and included in the revised version of the manuscript.
- Comment: The introduction part has a deep review of the BMS and some assessment method of distributed bridges. It is suggested to add some network-level bridge assessment and management method with the state-of-the-art techniques, such as: https://doi.org/10.1002/stc.2915 https://doi.org/10.1007/s00158-022-03210-3, https://doi.org/10.1061/(ASCE)BE.1943-5592.0001774, https://doi.org/10.1061/(ASCE)BE.1943-5592.0001815
Response: Thank you for suggesting these papers. These have been included in the revised manuscript.
- Comment: Data is the key for the analysis. It is suggested to illustrate the number of attributes that existed in this database, and what is the importance ranking of them for the assessment.
Response: Following screening, 49 non-open fields in the bridge basic data master table can be used in this study's clustering step. To obtain outstanding clustering quality, as indicated in Section 2, Materials and Methods, trial-and-error was used in the cluster stage. The optimal combination, as shown in the results, is that a single span is clustered by 7 attributes, 2-3 spans are clustered by 7 attributes, and the last group is clustered by 6 attributes. Because early research in this study was unable to specify the combination of fundamental data column content that bridges are prone to deterioration, this study must first group bridges in order to acquire significant research results.
- Comment: The main techniques for the assessment is cluster analysis and association analysis. How to choose these two methods for the assessment?
Response: As in the preceding question, because there are several types and types of bridges, it is impossible to determine the combination of basic data fields that predicts bridges' vulnerability to deterioration without prior clustering. To acquire significant research outcomes, this study must therefore cluster the bridges first. In addition, this study seeks to identify the combination of fundamental data fields that are susceptible to degradation. Association analysis is used to determine the correlation between test item degradation and basic data in order to establish a connection between the seemingly unconnected basic data and test data.
- Comment: Authors mentioned that it could concentrate their limited resources on bridges. This study did not contain the cost evaluation part. How to consider the resources facing with the deteriorations.
Response: Financial resources as a parameter or deciding attribute was not included in this study. However, we believe that by identifying the bridge structural components and structural configuration that are most prone to deterioration, bridge maintenance agencies will be able to develop more cost-effective bridge inspection procedures.
- Comment: The suggestion of improvement on this system could be added.
Response: The government established this system. The system's function is highly practical and is constantly being developed. The level of analysis will be broader and deeper if the maintenance and management unit can truly use all of the module features. As a result, it is currently advised that the system development unit incorporate this research finding into the system. It is useful for this study in addition to providing a reference for front liners of the bridge maintenance agencies.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors presented Bridge deterioration and maintenance strategies using machine learning-based clustering and association methods for big data were used. The article is well-written and adequately presented. However, the following are a few concerns about the suitability of this manuscript:
1. Although bridge maintenance using available data is the need of the hour to prevent further deterioration and provide timely repairs, the use of trivial algorithms such as K-means clustering is not acceptable as a scholarly contribution since there are several other superior and mathematically robust methods available. Please provide your opinion on why these two methods were adopted for this article.
2. There are similar papers already published by the authors; what is the new information you are presenting in this submitted manuscript that was not published before? Provide strong reasoning.
3. The introduction section needs to be thoroughly improved; although the authors' focus is not TBMS, it is critical to introduce other literature on bridge deterioration using similar or advanced methodologies. Revise it.
4. Figure 3's quality is inferior; please revise it with an acceptable resolution and size.
5. To improve the article, provide a contrasting introduction and cite the following articles to strengthen its readability.
(a) An, Y., Chatzi, E., Sim, S.-H., Laflamme, S., Blachowski, B., & Ou, J. (2019). Recent progress and future trends on damage identification methods for bridge structures. Struct. Control Health Monit., 26(10), e2416.
(b) Sony, S., Gamage, S., Sadhu, A., & Samarabandu, J. (2022). Multiclass Damage Identification in a Full-Scale Bridge Using Optimally Tuned One-Dimensional Convolutional Neural Network. J. Comput. Civ. Eng., 36(2), 04021035.
(c) Wu, C., Wu, P., Wang, J., Jiang, R., Chen, M., & Wang, X. (2022). A critical review of data-driven decision-making in bridge operation and maintenance. Struct. Infrastruct. Eng., 18(1), 47–70.
(d) Xia, Y., Lei, X., Wang, P., & Sun, L. (2022). A data-driven approach for regional bridge condition assessment using inspection reports. Struct. Control Health Monit., 29(4), e2915.
Author Response
- Comment: Although bridge maintenance using available data is the need of the hour to prevent further deterioration and provide timely repairs, the use of trivial algorithms such as K-means clustering is not acceptable as a scholarly contribution since there are several other superior and mathematically robust methods available. Please provide your opinion on why these two methods were adopted for this article.
Response: This study focuses primarily on the search for the combination of bridge attributes that is prone to deterioration, rather than on the use of good classification and grouping technology to classify bridges more accurately; therefore, the easier to implement and more effective grouping technology is chosen to differentiate between bridge groups. Also, since there are many different attribute combinations used in this study, association analysis can be used to quickly and accurately determine which field combination has the highest level of support and confidence. This research results will provide recommendations for bridge management units to check the safety state of bridges following disasters, with the hope of resolving the problem of bridge maintenance units' manpower shortage.
- Comment: There are similar papers already published by the authors; what is the new information you are presenting in this submitted manuscript that was not published before? Provide strong reasoning.
Response: The authors' first paper simply defines what TBMS is and how much data it contains. The authors list software applications that are commonly used at the end of the study. The second paper used R software to analyze the 24 bridges of the Zhongli Section of the Directorate of Highway Generation. Correlation coefficient analysis was performed on the bridges, and the results revealed that the content of several bridges' fundamental data fields was highly connected with bridge degradation. The previous study utilized the same method (correlation coefficient analysis) to evaluate the same sample bridges, but the findings indicated no correlation (ρ = 0) due to the sample distribution. So the work presented in this manuscript aggregated the samples and and applied association analysis. The findings of this study can only be acquired by assessing the contents of the basic data fields that are susceptible to degradation.
- Comment: The introduction section needs to be thoroughly improved; although the authors' focus is not TBMS, it is critical to introduce other literature on bridge deterioration using similar or advanced methodologies. Revise it.
Response: The Introduction section has been updated to include more recent studies and review papers relevant to the work presented in the manuscript.
- Comment: Figure 3's quality is inferior; please revise it with an acceptable resolution and size.
Response: The figure has been updated and included in the revised version of the manuscript.
5. Comment: To improve the article, provide a contrasting introduction and cite the following articles to strengthen its readability.
-
- An, Y., Chatzi, E., Sim, S.-H., Laflamme, S., Blachowski, B., & Ou, J. (2019). Recent progress and future trends on damage identification methods for bridge structures. Struct. Control Health Monit., 26(10), e2416.
- Sony, S., Gamage, S., Sadhu, A., & Samarabandu, J. (2022). Multiclass Damage Identification in a Full-Scale Bridge Using Optimally Tuned One-Dimensional Convolutional Neural Network. J. Comput. Civ. Eng., 36(2), 04021035.
- Wu, C., Wu, P., Wang, J., Jiang, R., Chen, M., & Wang, X. (2022). A critical review of data-driven decision-making in bridge operation and maintenance. Struct. Infrastruct. Eng., 18(1), 47–70.
- Xia, Y., Lei, X., Wang, P., & Sun, L. (2022). A data-driven approach for regional bridge condition assessment using inspection reports. Struct. Control Health Monit., 29(4), e2915.
Response: Thank you for suggesting these papers. These have been included in the rev
Author Response File: Author Response.pdf
Reviewer 3 Report
The technical contents of the paper are in general interesting. Its findings are useful information for bridge field applications in the future. Anyway, the publication in the “Sustainability, MDPI” is not recommended unless the following suggestions are taken into account:
1) The current state of knowledge relating to the manuscript topic is not covered and clearly presented, and the authors’ contribution and novelty are not emphasized within Abstract, Introduction and Conclusions. In this regard, the authors should make their effort to address these issues, by adding additional comments on the state of the art and the proposed aspects.
2) The deterioration of bridges in Taiwan due to the prestressing losses should be considered within the database. Please, read the following references to understand the related problems:
- https://doi.org/10.1177/13694332211022067
- https://doi.org/10.1016/j.istruc.2020.09.080
3) Introduction. The article could be improved by inserting tables where the literature is discussed to briefly characterize each reference (e.g. authors, year, testing, topic, and findings). This could help to give more strength and significance to the state-of-the-art. And, moreover, by introducing original figures with schemes to explain the driving ideas traced by the literature review.
4) Additional comments should be added in regard to the practical value of the work, and how the industry can profit from this article.
5) The further work should be mentioned at the end of the article.
6) A general check of English grammar, punctuation, spelling, verb usage, sentence structure, conciseness, readability and writing style is also suggested.
Author Response
- Comment: The current state of knowledge relating to the manuscript topic is not covered and clearly presented, and the authors’ contribution and novelty are not emphasized within Abstract, Introduction and Conclusions. In this regard, the authors should make their effort to address these issues, by adding additional comments on the state of the art and the proposed aspects.
Response: Thank you for this comment. We have updated these sections, particularly the Introduction section, in the revised manuscript.
- Comment: The deterioration of bridges in Taiwan due to the prestressing losses should be considered within the database. Please, read the following references to understand the related problems:
https://doi.org/10.1177/13694332211022067
https://doi.org/10.1016/j.istruc.2020.09.080
Response: This study only used the inspection data from the TBMS. However, losses in prestressed concrete bridge members in Taiwan were not taken into account in these inspection reports.
- Comment: The article could be improved by inserting tables where the literature is discussed to briefly characterize each reference (e.g. authors, year, testing, topic, and findings). This could help to give more strength and significance to the state-of-the-art. And, moreover, by introducing original figures with schemes to explain the driving ideas traced by the literature review.
Response: The Introduction section has been updated to include more recent studies and review papers relevant to the work presented in the manuscript. However, we chose to incorporate these studies into the main text so as not to impede the flow of ideas.
- Comment: Additional comments should be added in regard to the practical value of the work, and how the industry can profit from this article.
Response: The practical utility of this study arises from a shortage of manpower and resources allocated to undertake bridge inspections on Taiwan's bridges. Bridge maintenance contractors are unwilling to engage in the bidding process, regardless of whether the central government or a local government is procuring bridge inspection services, due to difficulties with the government's unit price. If a bridge collapses in Taiwan, the first step is to determine whether the bridge inspection contractor conducted erroneous inspections or if there were defects that should have been inspected but were not. After a natural disaster, it is difficult for contractors to conduct a comprehensive special inspection due to a lack of personnel. Contractors can therefore resort to this study to determine the combination of fundamental data column fields susceptible to deterioration for single-span, 2-3 span, and more than 4-span bridges.
- Comment: The further work should be mentioned at the end of the article.
Response: Follow-up research can explore the combination of basic data fields that are prone to serious deterioration, that is, only analyze the inspection records with a D value greater than or equal to 3. This was mentioned in the last paragraph of the Conclusion section of the original version of the manuscript.
- Comment: A general check of English grammar, punctuation, spelling, verb usage, sentence structure, conciseness, readability and writing style is also suggested.
Response: The revised manuscript has been subjected to these standard checks prior to submission.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Thanks for the reply. The comments have been fully concerned.
Author Response
Thanks for your suggestion.
Reviewer 2 Report
The authors have improved the manuscript and considered all the suggestions. I recommend it for publication.
Author Response
Thanks for your suggestion.
Reviewer 3 Report
The deterioration of bridges in Taiwan due to the prestressing losses should be considered in the database within the framework of further research and developments. Please, refer to the following literature to understand the corresponding problems:
- https://doi.org/10.1177/13694332211022067
- https://doi.org/10.1016/j.istruc.2020.09.080
Author Response
Regular bridge inspections in Taiwan are carried out every two years, primarily through visual inspection. If an NDT needs to be carried out to quantify the prestressing losses in these components, the inspectors will mark it in the forms, as NDTs are only employed as needed based on the outcomes of these inspections. The inspection records extracted from the TBMS module contain only reports from these visual inspections.
