Application of Soft Computing Techniques to Estimate Cutter Life Index Using Mechanical Properties of Rocks

Round 1

Reviewer 1 Report

The authors were recommended to revise the manuscript considering the major and minor comments as below.

1. The cutter life index (CLI) values used in this study, were statistically obtained from the CAI. In this study, the statistical analysis was made for predicted independent value (CLI). Why did the authors select the CLI as an independent variable rather than CAI? The reviewer thought this approach is not appropriate way in statistical analysis. Because the relationship between CLI and CAI shown in Eq. 9 was obtained from a certain database set, it is not a general equation for all rock types (or new data set). In this regard, please provide detailed information to derive the relationship between CAI and CLI in the previous study in terms of rock properties, ranges of CAI and CLI, etc. And please discuss on the problem in more detail in the manuscript.   
2. Table 6: The statistics, such as MSE, RMSE, VAF should be explained in the manuscript in terms of definition, importance, etc.). What is the “VAF”? As far as the reviewer knows, VAF has no meaning in linear regression compared to the training or learning-based model.     
3. Section 4.1.2: As reported by previous study (Yagiz, 2009: Ref. 13), the Brittleness index has a dependency on the UCS, BTS, D. Therefore, the database potentially has an over-fitting problem (or multicollinearity) in statistical analysis. Are there any results for compatibility of input variables?
4. “Figs 8-9, Figs. 10-11, Figs 12-13, Figs 1-15” presents inappropriate way. The figures should be separated.  
5. Authors stated that the database from 80 tunnel cases, the input values (UCS, BTS, D, BI, CAI, CLI) were average values for each tunnel site?
6. 23page, lines 524-528: The authors concluded that the models of D and BI is convenient to estimate the cutter life index. However, in practical view, PPT is not a common testing method although it is useful to estimate the brittleness of rock and thrust force of TBM disc cutter. In geological investigation, UCS, BTS and D are more common rock properties, and they are measured with a higher possibility than BI. Please add the related discussion in the manuscript.  
7. 14page, line 338: Where are Appendix B and C?
8. 17page, line 415: Where is Appendix D?
9. In the manuscript, the inappropriate term “thesis” should be changed into paper or study.
10. This manuscript should be proofread by a professional editor.

Author Response

Please find attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

A brief summary:

The aim of the paper was to develop models allowing to estimate disc cutter wear. The proposed models were based on common mechanical rock properties, such as compressive strength, tensile strength, brittleness and density. The authors used a database created based on the analysis of 80 mechanized tunnel cases. The cutter wear was estimated by means of cutter life index. Furthermore, the authors used artificial intelligence techniques (ANN and FL) as well as linear and non-linear regression models.

Broad comments:

A significant achievement of the presented work is the development of the set of models allowing to predict cutter wear. The wear of the tool in tunneling and mining is the key factor in terms of the costs of hard rock excavation. Authors developed models based on real world projects and the obtained results expand the knowledge of these processes and are applicable in industrial practice. In addition, there is a potential to improve the models using more input data. However, fit of most of the developed models was not very good (coefficients of determination on average approx. 0.6). To achieve better fit one needs to use models with more input variables, which proves to be difficult in industrial practice. The methods and results are presented in a clear way, however there are numerous typos and language errors that need to be corrected.

Specific comments:

• In the abstract and in the text authors describe the tool wear using the word “weariness”. That word refers mainly to a person’s state of mind and the better choice would be to use the word “wear”.
• In the Introduction authors introduce the Cerchar abrasivity index (line 45), then introduce it again in line 57 and refer to Figure 1, which presents the Sievers’ miniature drill test. The authors need to correct it.
• Line 107, line 121: the word “rounds” should be changed into “revolutions”.
• Line 454: change the numeration of Figure 2 into Figure 19.
• Line 473: change the numeration of Figure 3 into Figure 20.
• There are many language errors, i.e: line 55 (uses), line 62 (features), line 217 (rocks are both crucial), line 223 (considered), line 233 (all of them are different functions), line 247 (examine), line 253 (was performed), line 316 (models), line 468 (inputs combinations are used) etc. Authors need to check and correct in all the text.
• There are numerous typos, i.e.: line 283 (alternative approaches), line 284 (analysis), line 520 (easy), line 538 (and), lack of words i.e.: line 83 (developed a model to evaluate), too many words i.e.: line 205 (estimate examine) and even in a few instances grammatical errors preventing comprehension of the sentence, i.e.: line 274. Authors need to check and correct in all the text.

Author Response

Pls find attached file

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

No more comments and suggestions for the revised manuscript.