Generating Bit-Rock Interaction Forces for Drilling Vibration Simulation: An Artificial Neural Network-Based Approach
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors(1) The accurate definition of rock mechanics parameters is an important prerequisite and basis for conducting drilling related simulations, but this study did not provide detailed information on the key mechanical parameters of the rocks used, such as Young's modulus and Poisson's ratio. Therefore, the reviewers believe that the authors need to add a small paragraph in the relevant section of the manuscript to provide key mechanical parameters of the rock materials used for DEM simulation in this manuscript.
(2) The title of the manuscript mentions' fatigue analysis', but the results mainly focus on vibration and force prediction. Therefore, the reviewer suggests that the authors add a brief paragraph to discuss the impact of vibration results on fatigue damage indicators (such as the maximum stress range at critical connections and the number of stress cycles), in order to demonstrate the accuracy of "fatigue analysis".
(3) Drilling simulation is extremely important for maintaining drilling safety and ensuring the smooth progress of drilling operations. The sentence in lines 52–54 of the manuscript require citation of the following references: Numerical insights into factors affecting collapse behavior of horizontal wellbore in clayey silt hydrate-bearing sediments and the accompanying control strategy; -A numerical investigation on kick control with the displacement kill method during a well test in a deep-water gas reservoir: A case study.
(4) The core logic of the research in this article is DEM ->ANN ->drill string dynamics. We need a stronger argument as to why we don't directly use DEM output, but instead use ANN as a transition. Therefore, it is recommended that authors add relevant explanations in the introduction or method discussion, explaining the reasons for this approach and the main advantages of using ANN models (such as increasing computational efficiency by N times compared to directly conducting full-scale DEM simulations), to enhance the logical rationality.
(5) The article emphasizes the "Deviated Well" model, but mainly discusses axial and torsional vibrations. What are the differences in vibration and fatigue between inclined and vertical wells. What are the main differences in settings in the simulation? Therefore, it is recommended that the authors include 4 In Results or 5. Discussion, add a short paragraph to focus on how the wellbore friction or lateral contact force of Deviated Well is handled in the model, and what unique and different effects they have on the final vibration and fatigue results compared to vertical wells.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have written about: “Development of Fully Coupled Deviated Well Drill String Dynamics Simulation Model for Fatigue and Vibration Analysis.”
The topic is interesting, however, the following comments should be noted.
In general:
References cited in the text should be checked according to the author guidelines. For example, lines 141 (Jaime [3]), 143 (Cai et al. [4]), 157 (Dai et al. [5]), … etc. Citations in square brackets are sufficient here; the authors' names do not need to be included beforehand.
Images and tables should be displayed using the same font and font size.
References
Many of the cited references relate to general descriptions such as the cost, construction, and dynamite of the drill, the mechanical behavior of the rock, etc. The authors should focus more on similar literature that also deals with the topic in order to clarify the distinctions and the scientific value of their details. There are many such references, so the following are merely a few examples.
Jiusen Wei, Wei Liu , Deli Gao: Modeling of PDC bit-rock interaction behaviors based on the analysis of dynamic rock-cutting process, https://doi.org/10.1016/j.geoen.2024.212955
Liangjie Mao, Junjie Er, Junlong Zhu, Hai Jia, Lunke Gan: Dynamic characteristic response of PDC bit vibration coupled with drill string dynamics, https://doi.org/10.1016/j.geoen.2023.212524
Heng Zhang et al.: Numerical Simulation and Field Test Research on Vibration Reduction of PDC Cutting of Pebbled Sandstone under Composite Impact Load, https://doi.org/10.3390/pr11030671
Alexis Koulidis, Guang Ooi & Shehab Ahmed: Application of artificial intelligence to predict rock strength and drilling efficiency using in-cutter sensing data and vibration modes, https://doi.org/10.1007/s13202-024-01823-6
…
In this context, related descriptions of simulation methods, ANN, etc., can then be much better understood and assigned.
The authors should omit many very well-known facts that are more in the style of a textbook. Concise, targeted presentations on the topic are much better.
Subchapters:
Subheadings from the subchapters, outlining the topics covered in the authors' study, should be placed as the final paragraph in the "Introduction" chapter after a thorough review of similar literature. This provides the reader with an immediate, focused overview. Otherwise, important details are lost in the main text. For example, line 159: "This research focuses on creating a bit-rock interaction model for PDC bits." Lines 206-209: "This research combines a discrete element model of rock fractures from a single cutter element with an artificial neural network, to move away from overly simple functions and to better approximate the complex excitations one would get from experiments." Line 317-312: "Based on simulation model size and 317 the current literature trends, the authors of this paper have selected DEM to model rock 318 fractures and bit-rock interactions for drilling operations using PDC cutters. Simulation 319 results from PFC2Dmodels will be used to tune an ANN, which will predict the bit-rock 320 force signal for a given operational condition." Lines 403-409 should be in Chapter 3 since the authors write the methodology here, etc.
Chapter 3
What specific data was used to train the ANN? To what extent were experimental studies conducted, and how was the experimental setup designed? What thresholds (safety limits), torques, axial forces, etc., were applied? What input variables (data, parameters, etc.) were used for the ANN? Unfortunately, all of this information is missing, yet it is essential for this study. Without it, a reference is lacking.
In general, the authors should provide more visual material and proceed in a more structured manner.
The authors mention a comparison between PFC2D and PFC3D. A detailed presentation of this comparison would be helpful. It is well known that a 3D simulation requires more computation time than a 2D simulation. However, it is also true that a 3D simulation can achieve more realistic results than a 2D simulation (2D models have numerous simplifications). What is the authors' position on this? In addition to Figure 3, a comparison to the real-world state (see previous comments – Experiment) would be recommended, as well as more precise information on the input parameters and an explanation of the significant simplifications. What boundary conditions, coefficients of friction, meshing, etc., were chosen? Angles are mentioned in lines 469-471, but more detailed information is lacking. To what extent were the material structures (different compositions of the rock minerals taken into account) considered with regard to their dimensional and mechanical properties, which vary considerably? How was the simulation validated? What was the accuracy, and how was it improved? What script was written to avoid – as the authors mention – “human error”? What human errors occur during the simulation? To what extent are 240 simulations sufficient for the ANN model (training)? What limitations are associated with this? Which parameters and variables result from the 240 simulations? Which parameters were predicted by the ANN model and used for reconstruction? Figure 5 is not very informative. The authors should provide more information here. The validation of the ANN model based on Figure 6 is difficult to understand. It would be advisable to present the reference model (which was validated through experiments and then used for further analyses) to the reader and then clearly show the data for comparison.
Only with these clear explanations and presentations can the remaining results be understood.
In general, the authors should reconsider the structure of the paper. They should clearly present the reference model and demonstrate the validation through experimental trials to explain which data they used for the 240 simulations and their validation. All parameters and variables should be clearly presented. The figures should be supplemented with more information.
Comments on the Quality of English LanguageCheck commas and full stops. For example, line 111 is missing a full stop. The tenses should also be checked (past, present) and also some formulations.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors1. In the study, it was mentioned that the use of artificial neural network (ANN) models to predict the interaction force between drill bits and rocks. Please add a brief list or table in the Methods section (such as section 3.2), clearly listing the input features (such as WOB, RPM, cutting depth, etc.) and output forms (such as time series or statistical parameters of axial force and torque) of the ANN model.
2. In the study, the authors used 2D DEM to simulate the force on a single tool and trained ANN with this data to predict the force on the entire PDC drill bit. Why not use 3D models for research? The reviewer suggests that the authors add a sentence in section 3.1 to explain why the simplified model data based on 2D DEM can effectively and reasonably map the macroscopic stress state of the entire 3D drill bit under inclined shaft conditions, in order to enhance the applicability of the method selection.
3. The dynamic model of drill string is very sensitive to the selection of damping ratio. Therefore, it is recommended that authors briefly explain the selection criteria for damping coefficients (such as empirical values, experimental calibration, or based on Rayleigh damping assumptions) in the section describing the dynamic model of the drill string, in order to enhance the rationality of the model settings.
4. Stability and safety in drilling are important prerequisites for ensuring efficient development of oil and gas, especially unconventional oil and gas. Therefore, the first sentence of the introduction needs to be supported by citing the following papers: From CO2 sequestration to hydrogen storage: further utilization of depleted gas reservoirs; -Effects of crosslinking agents and reservoir conditions on the propagation of fractures in coal reservoirs during hydraulic fracturing; -Risk prediction of gas hydrate formation in the wellbore and subsea gathering system of deep-water turbidite reservoirs: case analysis from the south china sea.
5. 2D Discrete Element Method (DEM) simulation is the basis for predicting tool forces, but it does not provide a detailed description of the key parameters of the rock (bonded particle model) used, such as parallel bond strength, Young's modulus, Poisson's ratio, etc. Therefore, the reviewer suggests that the authors add a table or small section of text in section 3.1, listing the key mechanical parameters of rock materials used for DEM simulation and their sources.
6. Some figures lack legends and require authors to supplement them, such as Figure 4.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper can be accepted for publication now. This is because it has been improved heavily by the autors.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have satisfactorily revised the manuscript and answered outstanding questions. Publication can be recommended.
Reviewer 3 Report
Comments and Suggestions for AuthorsAccepted
