An Accelerated Degradation Durability Evaluation Model for the Turbine Impeller of a Turbine Based on a Genetic Algorithms Back-Propagation Neural Network
Round 1
Reviewer 1 Report
A model for evaluating the durability of the turbine impeller is proposed. From a physical point of view, the model ratios proposed for the turbine impeller are beyond doubt.
It should be noted that the volume of the article is clearly overstated. The content of tables 3.3-3.4 is recommended to be significantly reduced. Obviously, the authors believe that the large amount of data presented in these tables demonstrates good results of verification of the proposed model, but this is a controversial thesis.
The article also claims that the model offers a feasible research idea for solving durability assessment problems. Obviously, this proposal needs more detailed elaboration and extensive verification. In the proposed version of the article, this is not fulfilled.
Author Response
Question 1: It should be noted that the volume of the article is clearly overstated. The content of tables 3.3-3.4 is recommended to be significantly reduced. Obviously, the authors believe that the large amount of data presented in these tables demonstrates good results of verification of the proposed model, but this is a controversial thesis.
Answer to question 1: We have eliminated some useless data in tables 3.3-3.4 and modified in the paper.
Question 2: The article also claims that the model offers a feasible research idea for solving durability assessment problems. Obviously, this proposal needs more detailed elaboration and extensive verification. In the proposed version of the article, this is not fulfilled.
Answer to question 2: We have elaborated the conclusion in more detail based on the research content and results of the paper.
Reviewer 2 Report
The present work is done well and includes valuable results and I think that the revision manuscript can be published and the authors should consider the following point:
1- the novelty of the present work should be bolded in the end part of the introduction section.
2- related to figure 2.1,please describe about the type of analysis, is it static ? In addition, please explain the finite element simulation in details such as type of element and etc.
3- figure 2.2 is not clear and it is better to use a large scale images.
4- page 5 and line 142, it is better to say: It includes three stages.
5- tables 3.1 and 3.2, it is better to report temperatures in Kelvin and delete other unit.
6- related to the contours 3.3 & 3.4, which stress or strain is more effective on the fatigue life of the component (I.e., maximum principal stress or equivalent von misses stress), it is strongly suggested to describe more about your reason and which one is close to reality?
7- it is stated the structure 2-4-1 is used for Ann, is it the optimal one or not check ?
8- related to table 3.5, it is better to report other checking parameters such as RSME and ..., also present the used formula to obtain the result reported in this table. To do this, the authors can checked the following publication:
https://doi.org/10.3390/buildings12040438
9- what is the number of used sample in ANN ? It is important and effective on the accuracy of the model
10- the used data for the present model is based on the finite element simulation that should be validated by experimental data and it is very important.
11- the conclusion should be rewritten with more details including the most achievements of the current research.
Author Response
Question 1: The novelty of the present work should be bolded in the end part of the introduction section.
Answer to question 1: We have modified in the paper.
Question 2: Related to figure 2.1, please describe about the type of analysis, is it static? In addition, please explain the finite element simulation in details such as type of element and etc.
Answer to question 2: We have explained the finite element simulation process in detail in Figure 2.1 of the paper.
Question 3: Figure 2.2 is not clear and it is better to use a large scale images.
Answer to question 3: We have modified in the paper.
Question 4: Page 5 and line 142, it is better to say: It includes three stages.
Answer to question 4: We have modified in the paper.
Question 5: Tables 3.1 and 3.2, it is better to report temperatures in Kelvin and delete other unit.
Answer to question 5: We have modified in the paper.
Question 6: Related to the contours 3.3 & 3.4, which stress or strain is more effective on the fatigue life of the component (I.e., maximum principal stress or equivalent von misses stress), it is strongly suggested to describe more about your reason and which one is close to reality?
Answer to question 6: The maximum principal stress criterion is usually chosen as a criterion when studying the crack direction. In this paper, when fracture occurs in a plastic material like K418 alloy, it is more appropriate to use the energy perspective as a criterion, that is, the equivalent stress criterion. The contents of Tables 3.3 and 3.4, which integrate the comments of Reviewer #1, have eliminated some useless data.
Question 7: It is stated the structure 2-4-1 is used for Ann, is it the optimal one or not check?
Answer to question 7: The selection of the number of nodes in the hidden layer is related to the overall structure of the network, the number of samples, etc. Too few nodes in the hidden layer will reduce the accuracy of the model, and too many nodes will cause overfitting. In this paper, the number of nodes in the hidden layer is determined based on an empirical formula, which has been supplemented in the paper.
Question 8: Related to table 3.5, it is better to report other checking parameters such as RSME and ..., also present the used formula to obtain the result reported in this table. To do this, the authors can checked the following publication: https://doi.org/10.3390/buildings12040438.
Answer to question 8: We have modified in the paper.
Question 9: What is the number of used sample in ANN? It is important and effective on the accuracy of the model.
Answer to question 9: The sample number in ANN is 3492. This is shown in Section 3.1.
Question 10: The used data for the present model is based on the finite element simulation that should be validated by experimental data and it is very important.
Answer to question 10: The finite element simulation in the paper is used to determine the failure mode and failure site of the turbine blade and to determine the amount of degradation as well as the sensitive parameters based on the simulation results. For the model this paper is based on the data expansion of the endurance limit data for the turbine impeller material K418 alloy in Table 2.2. The non-linear relationship between the endurance limit, accumulated load and temperature was obtained by combining these data with GA-BP. The data in Table 2.2 are validated by extensive tests and have a certain accuracy and validity.
Question 11: The conclusion should be rewritten with more details including the most achievements of the current research.
Answer to question 11: We have modified in the paper.
Round 2
Reviewer 1 Report
The manuscript has been sufficiently improved to warrant publication in the journal Applied Sciences.
Reviewer 2 Report
The authors tried to provide revision manuscript based on the reviewers' comments and also response comments one bye one. I believe that the authors do the best to edit the initial manuscript and now, it can be published as the present form.
