Rate of Penetration Prediction Method for Ultra-Deep Wells Based on LSTM–FNN

Round 1

Reviewer 1 Report

The paper “Rate of Penetration Prediction Method for Ultra Deep Wells Based on LSTM-FNN” presents a study of interest for specialists working in fields such as petroleum exploration and production and well drilling. The use of artificial intelligence in these areas of interest is very important in the current context of the energy "crisis" and the energy transition.

General remarks:

• The topic of this paper corresponds to the topic of the Special Issue (Artificial Intelligence Applications in Petroleum Exploration and Production) of Applied Sciences journal.
• I recommend including a Discussion chapter before the Conclusions. In this chapter, the results obtained must be compared with the results obtained in similar studies for various oil fields around the world. The purpose of this chapter is to show the importance of using this model compared to other existing models.
• The resolution of some figures can be improved.
• The parameters used in this paper together with notations/abbreviations, can be included in a special section of the paper, before the References.

Specific remarks:

• Line 19. Replace “R2” with “R²”.
• Line 20. In the abstract the authors discuss about "data of well 1 and adjacent wells" but in the content of the paper only well1 and well2 are discussed.
• I recommend mentioning the abbreviations in the text at the first appearance. For example, Line 30: Replace "ROP" with "Rate of penetration (ROP)"
• I recommend that the Introduction chapter include a short presentation (with examples) of the current state of implementation of ANN in the evaluation and prediction of ROP but also the main drilling parameters used in this regard.
• Line 44.  Replace “Ma-chine” with “Machine”.
• Line 95.  Replace “Relu” with “ReLU”.
• Lines 100-102.  Hochreiter and Schmidhuber (1997) and “Felix Gers, Fred Cummins, and others” do not appear in the References chapter
• Figures 2-6. Are the logic diagrams made for this paper or are taken over / modified from another source? In the second case it is necessary to cite the source.
• Figure 2 caption. It is necessary to describe the elements presented in the figure.
• Line 110: In text appears (C^t) and (h^t) but in Figure 2 appears (c_t) and (h_t).
• Eq. 4: Replace sign “⊙” with “·”.
• I recommend merging figures 3, 4, 5 and 6 into one and named a, b, c and d. Also, the new figure can be moved after its mention in the text (at the end of chapter 2.2).
• Line 191: It is necessary to mention the figure referred to.
• Figure 7: The order of the parameters on the x-axis is reversed. Please review the figure.
• Line 195: Delete “0”.
• Figure 8. It is necessary to refer in text to this figure.
• Line 207: Replace “T time” with “t time”.
• Line 219: It is necessary to mention the figure referred to.
• Line 241: Replace “(R2)” with “(R²)”
• Lines 250-251: The information is redundant.
• Line 259: Replace “As shown in the Figure 6” with “As shown in Table 4”.
• Figures 10 and 11: Images quality is poor and “Real value” cannot be clearly distinguished from “Predictive value”. I recommend increasing the resolution of the figures and describing the 3 situations (a, b and c) in the figures caption.
• For figures 12 and 13 no reference is made in the text. Figure 12 caption is incomplete. You need to add MRE at the end of description. I recommend grouping the two figures into one and numbered with a and b.
• Line 281: Replace “figure 11” with “figure 13”.
• Line 283: Replace “the other three models” with “the other two models”.
• The Conclusions chapter can be improved after the introduction of a discussion section.

Author Response

Point 1: I recommend including a Discussion chapter before the Conclusions. In this chapter, the results obtained must be compared with the results obtained in similar studies for various oil fields around the world. The purpose of this chapter is to show the importance of using this model compared to other existing models. The resolution of some figures can be improved. The parameters used in this paper together with notations/abbreviations, can be included in a special section of the paper, before the References. The Conclusions chapter can be improved after the introduction of a discussion section.

 

Response 1: Dear review experts, because the data and input parameters used in different papers are different, we cannot directly compare the accuracy in different papers. Generally speaking, the relative error of ROP accuracy predicted by ROP equation is about 40%. The relative error of ROP prediction model established in this paper is about 15%, which has been greatly improved.

Point 2: Replace “R2” with “R2”. In the abstract the authors discuss about "data of well 1 and adjacent wells" but in the content of the paper only well1 and well2 are discussed. I recommend mentioning the abbreviations in the text at the first appearance. For example, Line 30: Replace "ROP" with "Rate of penetration (ROP)". Replace “Ma-chine” with “Machine”. Replace “Relu” with “ReLU”. Hochreiter and Schmidhuber (1997) and “Felix Gers, Fred Cummins, and others” do not appear in the References chapter. Figures 2-6. Are the logic diagrams made for this paper or are taken over / modified from another source? In the second case it is necessary to cite the source. Figure 2 caption. It is necessary to describe the elements presented in the figure. In text appears (Ct) and (ht) but in Figure 2 appears (ct) and (ht). Replace sign “⊙” with “·”. I recommend merging figures 3, 4, 5 and 6 into one and named a, b, c and d. Also, the new figure can be moved after its mention in the text (at the end of chapter 2.2). It is necessary to mention the figure referred to. Figure 7: The order of the parameters on the x-axis is reversed. Please review the figure. Delete “0”. Figure 8. It is necessary to refer in text to this figure. Replace “T time” with “t time”. Lines 250-251: The information is redundant. Replace “As shown in the Figure 6” with “As shown in Table 4”. Replace “figure 11” with “figure 13”. Replace “the other three models” with “the other two models”.

 

Response 2: According to your suggestions, the errors in the text have been corrected.

 

Point 3: I recommend that the Introduction chapter include a short presentation (with examples) of the current state of implementation of ANN in the evaluation and prediction of ROP but also the main drilling parameters used in this regard.

 

Response 3: According to your suggestion, I added an example of ROP prediction and the input parameters of the model.

Author Response File: Author Response.docx

Reviewer 2 Report

The summary needs to be shorter and clearer, with emphasis on the findings that result from the research conducted.

After line 149, the figures are not cited in the text. 

The statment (lines 165, 166) states that the differences in the physical properties of the rock are small. What does this mean?

Table 1 lists the drilling parameters. In the Engineering group, mention the inlet and outlet density. Density of what? Drilling fluid? Cuttings?

 

Author Response

Point 1: The summary needs to be shorter and clearer, with emphasis on the findings that result from the research conducted.

 

Response 1: According to your suggestions, the summary has been optimized. Please check it.

 

Point 2:. After line 149, the figures are not cited in the text.

 

Response 2: A reference has been added according to your suggestion

 

Point 3:. The statment (lines 165, 166) states that the differences in the physical properties of the rock are small. What does this mean?

 

Response 3: Rocks also have certain properties, such as lithology, porosity, compressive strength, density, etc., which means that the physical properties of rocks have little difference.

 

Point 4:. Table 1 lists the drilling parameters. In the Engineering group, mention the inlet and outlet density. Density of what? Drilling fluid? Cuttings?

 

Response 4: Outlet density refers to the density of drilling fluid. According to your suggestion, it has been supplemented in the text.

Reviewer 3 Report

Dear Authors,

            I carefully read this very interesting manuscript related to the rate of penetration prediction method based on the hybrid artificial intelligent technic. In my humble opinion, if the manuscript is thoroughly revised, it can make a good publication. To help improve the quality of this manuscript, I have added more comments as follows: 

            General Comments:

            * The authors should show an overview of other research about the application of the LSTM-FNN hybrid technique in the petroleum industry, especially in the prediction of the rate of penetration.

            * The authors use the data from only two wells in the Tarim oilfield with the LSTM-FNN technique. To improve the certainty of this technique in ROP prediction, more data from other wells should be used.       

            Detailed comments:

            * Lines 61-> 66: give the citation

            * Section 2: there is no citation, the author should give a citation of different articles/references about artificial intelligence algorithms in this section

            * Equation 1: give the meaning of φ_active

            * Equation 2: give the meaning of x

            * Equation 4: give the meaning of b

            * Lines 151->157: give the citation

            * Line 165: As seen in Figures 10,11, the well depth of well 1 and well 2 is 5200m and 5600m, respectively. These values are smaller than 6000m

            * Lines 170->189: there is no citation in this section

            * Lines: 190 -> 197:  should explain in more detail the way to obtain the values: 0.08; 0.81

            * Line 195: verify the value ‘0’

            * Line 259: Figure 12 instead of Figure 6

            * Line 265: 4.16%  à 4.17%

            * Line 280: Figure 11 à Figure 13

 

Kind regards,

 

Reviewer

Author Response

Point 1: The authors use the data from only two wells in the Tarim oilfield with the LSTM-FNN technique. To improve the certainty of this technique in ROP prediction, more data from other wells should be used.

 

Response 1: Due to the confidentiality of the data, it is difficult to obtain the drilling data, and the quality is uneven, so this paper only uses the data of two wells, but compared with the research of other scholars, the amount of data has been large, and some scholars' data samples are hundreds or even dozens. Later, we will continue to supplement and improve the data to improve the reliability of the model.

 

Point 2: Lines 61-> 66: give the citation. Section 2: there is no citation, the author should give a citation of different articles/references about artificial intelligence algorithms in this section. Equation 1: give the meaning of φactive. Equation 2: give the meaning of x. Equation 4: give the meaning of b. Lines 151->157: give the citation,Lines 170->189: there is no citation in this section. Lines: 190 -> 197:  should explain in more detail the way to obtain the values: 0.08; 0.81. Line 195: verify the value ‘0’. Line 259: Figure 12 instead of Figure 6. Line 265: 4.16%  à 4.17%. Line 280: Figure 11 à Figure 13.

 

Response 2: According to your suggestions, the errors in the text have been corrected.

 

Point 3: Line 165: As seen in Figures 10,11, the well depth of well 1 and well 2 is 5200m and 5600m, respectively. These values are smaller than 6000m

 

Response 3: When mapping the whole well prediction data, the picture will be too long and unclear, so the data of some well sections are selected for display. It has been indicated in the text according to your suggestions.

Round 2

Reviewer 1 Report

Dear Authors,

Most of my observations from the first review have been resolved. 

There are some small mistakes that need to be corrected like:

L. 19: Replace "wells 2" with "well 2".

L. 96: Replace "relu" with "ReLU".

L. 108: " ...Felix Gers, Fred Cummins..." Need citation in text and References.

There are also some questions from the first review left to answer:

1) Figures 2-6 (Now - Figures 2-3). Are the logic diagrams made for this paper or are taken over / modified from another source? In the second case it is necessary to cite the source.

2) Figure 7 (Now - Figure 4): The order of the parameters on the x-axis is not reversed? Please review the figure.

Author Response

Point 1: L. 19: Replace "wells 2" with "well 2".L. 96: Replace "relu" with "ReLU".L. 108: " ...Felix Gers, Fred Cummins..." Need citation in text and References.1) Figures 2-6 (Now - Figures 2-3). Are the logic diagrams made for this paper or are taken over / modified from another source? In the second case it is necessary to cite the source. Figure 7 (Now - Figure 4): The order of the parameters on the x-axis is not reversed? Please review the figure

 

Response 1: Dear review experts, according to your suggestions, the errors in the text have been corrected.

Reviewer 3 Report

Dear Authors,

I have carefully read the new version of the manuscript. Most of the comments have been adopted by the author. However, the first suggestion about analyzing an overview of other research about the application of the LSTM-FNN hybrid technique in the prediction of the rate of penetration has not been mentioned by the authors

 

Kind regards

Author Response

Point 1: Analyzing an overview of other research about the application of the LSTM-FNN hybrid technique in the prediction of the rate of penetration has not been mentioned by the authors

 

Response 1: Dear reviewer, according to your suggestion, the application research of network overlay has been added. Please check it.