Modeling the Dynamic Behavior of Recycled Concrete Aggregate-Virgin Aggregates Blend Using Artificial Neural Network
, Yi Huang 3, Tingyu Li 3, Pin Deng 1, Yuliang Chen 3, Yuanjie Xiao 2,4,* and Joseph Ali 2
Round 1
Reviewer 1 Report
This study addresses the rise in construction waste, focusing on anisotropic behavior of blended construction and demolition waste (CDW) aggregates. Repeated load triaxial tests reveal shear stress ratios' impact on permanent deformation. An advanced neural network model predicts deformations and highlights key factors, aiding pavement construction planning. However, before further consideration of the manuscript, the authors must “fully” address the comments listed below:
1- In terms of the model's applicability and generalizability, could you elucidate the methodology utilized to assess its performance with external datasets? How did the model's predictions align with the actual outcomes, further attesting to its efficacy in capturing the nuanced complexities of permanent deformation in blended CDW materials?
2- Could you provide a more detailed account of the training and validation process employed to refine the ANN model? How did the experimental data from the repeated load triaxial tests synergize with the network architecture to attain a high degree of convergence, as evidenced by the coefficient of determination of 0.99? How does this convergence bolster the model's predictive capabilities?
3- Could you delve into the architectural intricacies of the artificial neural network (ANN) model, expounding upon its layered organization and the interplay between processing elements? How does this model encapsulate the complexities of the blended CDW materials' behavior, and what facets of its design enable it to effectively learn from the experimental data and make accurate deformation predictions?
4- To traverse the domain of permanent deformation under varying moisture conditions, could you elaborate on the empirical manifestations of moisture content's impact on the permanent deformation of blended construction and demolition waste material? How does the interplay between aggregate inter-particle interactions, friction reduction, and shear strength engender varying deformation responses for specimens situated above and below the optimum moisture content (OMC)?
5- The authors mentioned that “The performance criteria of the developed ANN model indicate its capability to 440 predict the permanent deformation of the blended proportions with high precision and 441 accuracy”. However, this statement (that is choosing the best model to improve the model accuracy) is partially true because the dataset might be highly complex such that even a better/more optimized network may not necessarily improve the model accuracy. In machine learning, this can refer to “Kolmogorov complexity” denoting the length of the shortest computer program that produces the output. Therefore, write a paragraph in your paper arguing that reducing the complexity of your dataset can potentially improve the accuracy of the deep learning model (you may read and references the two journal papers that efficiently leveraged decreased dataset complexity to rapidly improve their model accuracy: paper 1: https://doi.org/10.1007/s10462-019-09750-3, and paper 2: https://doi.org/10.1038/s41598-023-28763-1)
6- Could you provide a more intricate elucidation of the artificial neural network (ANN) deformation prediction model's conceptual foundation? How does this model effectively encapsulate the nuanced complexities stemming from the amalgamation of blended CDW and natural aggregates, thereby transcending the limitations of conventional prediction models in accounting for these intricate interactions?
The English is fine in the current format, minor revisions are required.
Author Response
This study addresses the rise in construction waste, focusing on anisotropic behavior of blended construction and demolition waste (CDW) aggregates. Repeated load triaxial tests reveal shear stress ratios' impact on permanent deformation. An advanced neural network model predicts deformations and highlights key factors, aiding pavement construction planning. However, before further consideration of the manuscript, the authors must “fully” address the comments listed below:
Point 1: In terms of the model's applicability and generalizability, could you elucidate the methodology utilized to assess its performance with external datasets? How did the model's predictions align with the actual outcomes, further attesting to its efficacy in capturing the nuanced complexities of permanent deformation in blended CDW materials?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. In the follow-up study, we will consider additional experimental parameters and integrate them with field experimental test data to enhance the model. By combining the laboratory experimental data with the field measurement data, we aim to improve its applicability in engineering.
Point 2: Could you provide a more detailed account of the training and validation process employed to refine the ANN model? How did the experimental data from the repeated load triaxial tests synergize with the network architecture to attain a high degree of convergence, as evidenced by the coefficient of determination of 0.99? How does this convergence bolster the model's predictive capabilities?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. According to your suggestion, we have added more information about the training and validation process employed to refine the ANN model, as quoted below for your review.
Page 13, lines 367 t 386: “ The process of training a neural network involves adjusting the weights and bias values of the network to optimize the network performance. In this paper, the Levenberg-Marquardt (LM) training algorithm and the mean square error (MSE) performance function were used. The MSE performance function is the mean squared error between the network output a and the target outputs t, as defined in Equation (X). The LM algorithm updates the network weights and biases to approach second-order training speed without having to compute the Hessian matrix. The iteration of this algorithm can be written in the form shown in Equation (X).
|
, |
(12) |
|
, |
(13) |
where xk is the current weights and biases, J is the Jacobian matrix that contains first derivatives of the network errors with respect to the weights and biases, e is the vector of current network errors, and μ is an adaptive value.
During training, Equation (12) will keep iterating while the performance value of the model will keep decreasing. As the training reaches a minimum of the performance, the gradient will become very small. If the magnitude of the gradient is less than XXX or the number of iterations is greater than XXX, the training will stop.
After training, test data is fed into the trained ANN to output the predicted values. Performance metrics such as R2 are calculated based on the predicted and actual values. If the R2 calculated on the test set is less than or equal to the value calculated on the training set, it means that the network can replicate the experimental results well and at the same time has a certain generalization ability (i.e., no overfitting occurs).”
Point 3: Could you delve into the architectural intricacies of the artificial neural network (ANN) model, expounding upon its layered organization and the interplay between processing elements? How does this model encapsulate the complexities of the blended CDW materials' behavior, and what facets of its design enable it to effectively learn from the experimental data and make accurate deformation predictions?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. A multi-layer neural network contains alternating combinations of multiple linear and nonlinear layers. The linear layers can extract low-level features, while the nonlinear layers can gradually extract more complex features by means of multilevel nonlinear transformations, which is why this paper ultimately used a neural network with two hidden layers. As shown in Equation (5), during the computation of a neuron, each input is multiplied by the corresponding weight, then a linear summation operation is performed with a bias term, and finally a nonlinear transformation is performed by means of an activation function. By using different activation functions, the neuron can realize different ways of feature extraction. As shown in Table 5, this paper systematically compared the effects of activation function type, number of hidden layers and number of neurons in each hidden layer on the prediction accuracy of the ANN model. More number of hidden layers and number of neurons in hidden layers means more connections and the network is able to learn more complex functions. It was found that XXX (specify which activation function is better and give an example of the fact that more hidden neurons dose not mean better prediction accuracy).
Point 4: To traverse the domain of permanent deformation under varying moisture conditions, could you elaborate on the empirical manifestations of moisture content's impact on the permanent deformation of blended construction and demolition waste material? How does the interplay between aggregate inter-particle interactions, friction reduction, and shear strength engender varying deformation responses for specimens situated above and below the optimum moisture content (OMC)?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for missing detailed discussion in the original manuscript. In this paper, the effect of moisture content on the accumulated plastic deformation was investigated from a macroscopic perspective using laboratory repeated load triaxial tests. The interplay between inter-particle interactions, friction reduction, and shear strength, which involves meso-scale explanation, is not the focus of this paper and deserves to be further explored by numerical modeling such as discrete element simulation.
Point 5: The authors mentioned that “The performance criteria of the developed ANN model indicate its capability to predict the permanent deformation of the blended proportions with high precision and accuracy”. However, this statement (that is choosing the best model to improve the model accuracy) is partially true because the dataset might be highly complex such that even a better/more optimized network may not necessarily improve the model accuracy. In machine learning, this can refer to “Kolmogorov complexity” denoting the length of the shortest computer program that produces the output. Therefore, write a paragraph in your paper arguing that reducing the complexity of your dataset can potentially improve the accuracy of the deep learning model (you may read and references the two journal papers that efficiently leveraged decreased dataset complexity to rapidly improve their model accuracy: paper 1: https://doi.org/10.1007/s10462-019-09750-3, and paper 2: https://doi.org/10.1038/s41598-023-28763-1)
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the confusion due to the lack of clarification and details in the original manuscript. According to your suggestion, we have added a paragraph in the revised manuscript, as quoted below for your reference.
Reducing the complexity of the dataset is not in conflict with increasing the complexity of the network (e.g. number of layers and neurons). The recommended papers are not really very relevant.
In fact, we had developed neural networks with a single hidden layer and different hidden neurons, and the two neural networks with the best prediction performance were listed in Table 5. As you said, "a more optimized network may not necessarily improve the model accuracy", and a neural network with two hidden layers may not necessarily improve the model accuracy. As you said, “a more optimized network may not necessarily improve the model accuracy”, and neural networks with two hidden layers may not have better prediction accuracy than those with a single hidden layer. For example:
The selection of input variables in this paper was based on the following considerations.
(1) Permanent deformation of the material is highly correlated with the loading history (i.e., number of loading cycles);
(2) At least two parameters are required to characterize the stress conditions;
(3) The most important characteristic of the studied material is the content of CDW;
(4) The material was found to be moisture sensitive through repeated load triaxial tests.
In addition, statistical correlations between the input variables were calculated in this study. (To do)
In addition to minimizing the complexity of the dataset (the redundant input variable σd has been eliminated), the size of the dataset has been simplified from XXX (N = 1, 2, ..., 50,000) to XXX (N = 10, 20, ..., 50,000) in order to reduce the training time of the neural network.
On the other hand, to predict the permanent deformation of the material as accurately as possible for a small number of loading cycles, it has been decided to use the dataset of the above size as well as the neuron network with two hidden layers. However, further research is still needed to balance the size of the dataset and the prediction accuracy of the neural network.
Point 6: Could you provide a more intricate elucidation of the artificial neural network (ANN) deformation prediction model's conceptual foundation? How does this model effectively encapsulate the nuanced complexities stemming from the amalgamation of blended CDW and natural aggregates, thereby transcending the limitations of conventional prediction models in accounting for these intricate interactions?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. The ANN model, a standard back-propagation network, can approximate any measurable function with a desired accuracy by employing a sufficient number of hidden neurons. The most direct reason is that traditional prediction models do not take CDW content into account at all. Neural network models, however, can fit any complex relationship by utilizing the nonlinear nature of the transfer function. We have add more information of ANN model conceptual foundation in the manuscript, as quoted below for your review.
Page 10, Lines 254 to 265: “The ANN draws inspiration from the biological nervous system, i.e., the brain's information processing mechanism. The ANN comprises numerous artificial neurons, referred to as processing elements. These elements are extensively interconnected, essential for producing an efficient output [48-49]. The neural network can learn from experience and information to enhance its performance. In multilayer perceptions, processing elements are organized in layers, including input, output, and one or more intermediate layers, known as hidden layers. These hidden layers contain neurons that determine patterns and the relationships between the output and the input [50]. Which can approximate any measurable function with a desired accuracy by employing a sufficient number of hidden neurons [51]. In each hidden layer, the input variables are multiplied by the weights and summed, and a bias value is added to the system [21], as shown in Equation (5).”
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper titled "Modeling the dynamic behavior of recycled concrete aggregate-virgin aggregates blend using Artificial Neural Network" presents an intriguing approach to understanding the dynamic behavior of a blend between recycled concrete aggregates (RCAs) and virgin aggregates (VAs) through the utilization of Artificial Neural Networks (ANNs). While the topic is of significant interest due to its implications for sustainable construction practices, there are several key points that require substantial revision before the paper can be considered for publication:
The paper lacks a clear delineation of the specific objectives and scope of the research. The introduction should precisely state what the study aims to achieve, emphasizing the gap or problem being addressed. This would help readers understand the significance of the research from the outset.
The current literature review is insufficient in providing a comprehensive understanding of previous works related to modeling the dynamic behavior of blended aggregates. The authors should consider expanding this section to include a broader range of studies, highlighting their methodologies and findings. This will provide context and establish the paper's contribution to the field.
The methodology section introducing the ANN model needs substantial refinement. The choice of input variables, the rationale behind their selection, and their relevance to the dynamic behavior of the blend must be thoroughly justified. Additionally, the architecture and parameters of the ANN should be elaborated in more detail, considering the intricacies of the problem being addressed.
The paper lacks transparency regarding the dataset used for training and validation of the ANN model. Details such as the size of the dataset, data preprocessing techniques, and any inherent biases in the data should be explicitly mentioned. Furthermore, the experimental setup and conditions under which dynamic behavior testing was conducted need to be clearly stated.
The results section presents limited analysis of the model's performance. It is crucial to include quantitative measures of the model's accuracy, such as Mean Absolute Error (MAE) or Root Mean Squared Error (RMSE), and compare the predicted values with actual experimental results. Moreover, the discussion should delve into the implications of the findings, addressing any discrepancies between model predictions and actual behavior.
The conclusion should succinctly summarize the key findings and their implications for the field. Additionally, a section on future work should outline potential directions for further research, building upon the current study's limitations and gaps.
The paper requires significant language and presentation improvements. The writing should be concise, grammatically accurate, and avoid unnecessary jargon. Graphs, figures, and tables should be clearly labeled and adequately explained in the text.
Example: see-Table-3 (Specimen volume cm3), Table-6
Addressing these major points of concern will greatly enhance the quality and impact of the paper. The authors are encouraged to undertake a thorough revision, taking into consideration the constructive feedback provided, to ensure that the paper meets the standards of rigorous scientific inquiry and contributes meaningfully to the field.
Author Response
Point 1: The paper titled "Modeling the dynamic behavior of recycled concrete aggregate-virgin aggregates blend using Artificial Neural Network" presents an intriguing approach to understanding the dynamic behavior of a blend between recycled concrete aggregates (RCAs) and virgin aggregates (VAs) through the utilization of Artificial Neural Networks (ANNs). While the topic is of significant interest due to its implications for sustainable construction practices, there are several key points that require substantial revision before the paper can be considered for publication:
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. Your positive compliment and support are greatly appreciated and mean a lot to us. As you may find, all your review comments have been addressed one by one with itemized responses. It is our hope that the revised manuscript would be up to your satisfaction.
Point 2: The paper lacks a clear delineation of the specific objectives and scope of the research. The introduction should precisely state what the study aims to achieve, emphasizing the gap or problem being addressed. This would help readers understand the significance of the research from the outset.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for missing detailed explanations about the research gap and objective. According to your suggestion, we have already added additional paragraphs in the introduction section, as quoted below for your reference.
Page 3, Lines 119 to 132: “The main improvements made in this study compared to previous studies are as follows: (a) The maximum number of cyclic loading in the repeated load triaxial tests conducted in this study reached 50,000 cycles, which exceeds the 10,000 cycles used in most of the literature. (b) The concept of shear stress ratio, defined as the applied stress level under moving wheel loads in relation to the material’s shear strength, was employed in this study to properly characterize and predict the permanent deformation behavior of blended CDW. (c) Considering the complex characteristics of CDW, the ANN model developed in this paper integrates the effects of material properties (i.e., CDW content, viscosity and friction angle of the blended materials), physical conditions (i.e., moisture content) and stress levels. (d) Based on the geometric pyramid rule, ANN models with different architectures were developed in this study. The prediction accuracies of these models were systematically compared, and the reliability of the optimal model was validated. (e) Based on the results of the sensitivity analysis, implications for engineering practices were also presented.”
Point 3: The current literature review is insufficient in providing a comprehensive understanding of previous works related to modeling the dynamic behavior of blended aggregates. The authors should consider expanding this section to include a broader range of studies, highlighting their methodologies and findings. This will provide context and establish the paper's contribution to the field.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the confusion due to the lack of reference citation in the original manuscript. According to your suggestion, we have expanded the introduction section to include more comprehensive and in-depth literature review, as quoted below for your reference.
Page 2, Lines 88 to 93: “There were mainly two types of methods used for establishing the prediction model of permanent deformation in unbound granular materials under cyclic loading in the previous studies. The first method involved developing an empirical model based on test data obtained from laboratory experimental tests [22-23]. The second method involved establishing a numerical simulation model for predicting the permanent deformation in unbound granular materials [24-25].”
Point 4: The methodology section introducing the ANN model needs substantial refinement. The choice of input variables, the rationale behind their selection, and their relevance to the dynamic behavior of the blend must be thoroughly justified. Additionally, the architecture and parameters of the ANN should be elaborated in more detail, considering the intricacies of the problem being addressed.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for missing the explanations about the choice of input variables and the architecture and parameters of the ANN. According to your suggestion, we have added more details in the corresponding sections, as quoted below for your reference.
Page 10, Lines 287 to 299: “While the cumulative axial strain was the only output of the ANN model. Table X lists the statistical parameters of the input and output variables and Figure X shows the correlation coefficients between the input variables. The selection of the above input was based on the following considerations. As described in the Section 3, the permanent deformation of the blended CDW was highly correlated with the loading history (i.e., the number of loading cycles). Stress conditions can significantly affect the deformation of the material and at least two parameters are require to characterize the stress level in triaxial compression tests, and thus SSR and confining pressure were also selected as inputs. As demonstrated in the Section 3.3, the blended CDW was found to be moisture sensitive through repeated load triaxial tests. As the primary properties of the studied materials, the CDW content was taken as an input together with the moisture content, reflecting the influence of internal factors on the permanent deformation.”
Page 10, Lines 290 to 293: “As shown in Table 5, ANN models with different architectures were developed in this study to investigate the effects of the activation function, the number of hidden layers, and the number of neurons in each hidden layer on the prediction accuracy.”
Page 19, Lines 484 to 489: “It is generally believed that the weight value represents the degree of one neuron’s influence on another. The weight and bias values between the neurons in various layers of the optimal ANN model are shown in Tables 7, 8, and 9, which can be accessed by readers to use and further improve the developed ANN.”
Point 5: The paper lacks transparency regarding the dataset used for training and validation of the ANN model. Details such as the size of the dataset, data preprocessing techniques, and any inherent biases in the data should be explicitly mentioned. Furthermore, the experimental setup and conditions under which dynamic behavior testing was conducted need to be clearly stated.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the confusion due to the lack of clarification and details in the original manuscript. According to your suggestion, we have added more details about the dataset used for training and validation of the ANN, the data preprocessing flow, and the test equipment and conditions in the corresponding section, as quoted below for your reference.
Page 11 to 12, Lines 318 to 324: “In addition to minimizing the complexity of the dataset (e.g., the redundant input variable σd was not included in the inputs), the size of the dataset was simplified from N = 1, 2, ..., 50,000 to N = 10, 20, ..., 50,000 in order to reduce the training time of the neural network. On the other hand, to predict the permanent deformation of the material as accurately as possible for a small number of loading cycles, it was decided to use the dataset of the above size. However, further research is still needed to balance the size of the dataset and the prediction accuracy of the neural network.”
Point 6: The results section presents limited analysis of the model's performance. It is crucial to include quantitative measures of the model's accuracy, such as Mean Absolute Error (MAE) or Root Mean Squared Error (RMSE), and compare the predicted values with actual experimental results. Moreover, the discussion should delve into the implications of the findings, addressing any discrepancies between model predictions and actual behavior.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the confusion due to the lack of clarification and details in the original manuscript. According to your suggestion, we have added more detailed information about the measurement of the model’s accuracy and the implications of the findings in the corresponding section, as quoted below for your reference.
Page 12, Lines 388 to 396: “Table 5 demonstrates the impact of the number of hidden neurons and the transfer function within the hidden layers on the ANN models' performance. MSE, RMSEs, and the R2 were utilized to evaluate the models' performance. As per Table 5, the model with a log-sigmoid transfer function and 24 and 12 neurons in the first and second layers, respectively, exhibits the most precise prediction and demonstrates satisfactory performance with the lowest values of MSE and RMSE and a high R2 value in comparison to the other models examined.”
Table 5. Comparison of different ANN model architecture.
|
Model |
Hidden Layer 1 |
Hidden Layer 2 |
MSE |
RMSE |
R2 Value |
||
|
Transfer Function |
Number of Neurons |
Transfer Function |
Number of Neurons |
||||
|
1 |
Tansig |
22 |
- |
- |
0.005200 |
0.0724 |
0.9990 |
|
2 |
Logsig |
24 |
- |
- |
0.005100 |
0.0711 |
0.9989 |
|
3 |
Tansig |
16 |
Tansig |
4 |
0.008600 |
0.0929 |
0.9981 |
|
4 |
Tansig |
16 |
Tansig |
8 |
0.039600 |
0.1990 |
0.9915 |
|
5 |
Tansig |
20 |
Tansig |
4 |
0.000850 |
0.9998 |
0.9998 |
|
6 |
Tansig |
20 |
Tansig |
8 |
0.000150 |
1.0000 |
0.9999 |
|
7 |
Tansig |
20 |
Tansig |
12 |
0.000069 |
0.0083 |
0.9999 |
|
8 |
Tansig |
24 |
Tansig |
4 |
0.000150 |
0.0122 |
0.9999 |
|
9 |
Tansig |
24 |
Tansig |
8 |
0.000360 |
0.0189 |
0.9999 |
|
10 |
Tansig |
24 |
Tansig |
12 |
0.000210 |
0.0143 |
1.0000 |
|
11 |
Logsig |
20 |
Logsig |
4 |
0.029900 |
0.1728 |
0.9936 |
|
12 |
Logsig |
20 |
Logsig |
8 |
0.000103 |
0.0102 |
0.9999 |
|
13 |
Logsig |
20 |
Logsig |
12 |
0.000056 |
0.0075 |
0.9999 |
|
14 |
Logsig |
24 |
Logsig |
4 |
0.000750 |
0.0273 |
0.9998 |
|
15 |
Logsig |
24 |
Logsig |
8 |
0.000076 |
0.0087 |
0.9999 |
|
16 |
Logsig |
24 |
Logsig |
12 |
0.000022 |
0.0047 |
0.9999 |
Page 15, Lines 414 to 428: “Figure 11 illustrates the ANN model performance against the measured permanent deformation for the composite construction and demolition waste unbound granular aggregate. When the data aligns along the line of equity (line drawn at 45 degrees), the predicted values equal the actual values, suggesting perfect prediction. The goodness of fit between the predicted and the measured values is evaluated using R2, which ranges between 0 and 1. When the R2 is close to 1, perfect prediction is achieved. The figure above shows that the value of R2 for both training and testing almostly equals 1, indicating a perfect fit and an excellent performance of the ANN model with a strong correlation between the measured and predicted data. In addition, the MAE and the RMSE are within 0.0024 and 0.0047 for both training and testing datasets, indicating that the ANN model underpredicts the permanent deformations at an average of 0.0024 and the deviation between the experimental and predicted values is a significantly small 0.0047. A higher R2 value and lower RMSE and MAE values denote better model performance. Therefore, the result indicates that the ANN model can replicate the laboratory-measured permanent deformation with high accuracy and precision.”
Figure 11. Lab-measured permanent deformation vs. ANN-predicted permanent deformation.
Page 15, Lines 446 to 458: “Figure 12 exhibits the experimental results, the UIUC model’s predicted values, and the ANN model's predicted values of the blended CDW aggregate materials. The robustness of the ANN model and the accuracy of the UIUC regression-based model proposed by Chow [42] are compared. It demonstrates that the ANN-predicted values are closer to the experimental values, indicating that the developed model successfully predicted the permanent strain of the blends considering different influencing factors. The result shows that the ANN model outperforms the regression-based model, demonstrating that the ANN model can describe the complex relationship between the input and output variables using neurons in the hidden layers.”
|
|
|
|
(a) |
(b) |
|
|
|
|
(c) |
(d) |
Figure 12. Comparison between the predicted ANN Model, the predicted UIUC Model, and the experimental result: (a) 0% CDW, σ3= 50 kPa; (b) 65% CDW, σ3= 50 kPa; (c) 85% CDW, σ3= 50 kPa; (d) 100% CDW, σ3= 50 kPa.
Page 19, Lines 489 to 497: “Figure 13 displays the influence of input parameters on the model for permanent deformation prediction. The results indicate that all the input parameters are crucial for predicting permanent deformation. The number of loading cycles with a 31% influence strongly impacts the output variable, followed by confining pressure and the SSR with 25% and 21% influence, respectively. The impact of the construction and demolition waste content was more substantial than the moisture content. The findings indicate that testing conditions and materials play a significant role in deformation, guiding policymakers and practitioners in regulating the use of these materials and ensuring proper pavement maintenance. Specifically, care should be taken to control the content of CDW to ensure that the blended materials has a relatively high shear strength. Extra effort should also be made to control traffic loads not to exceed the design limits and to prevent uncontrolled development of permanent deformation by regular maintenance.”
Point 7: The conclusion should succinctly summarize the key findings and their implications for the field. Additionally, a section on future work should outline potential directions for further research, building upon the current study's limitations and gaps.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for missing detailed discussion in the original manuscript. According to your suggestion, we have rewritten the related section in the revised version, as quoted below for your reference.
Page 20, Lines 503 to 543: “This study focuses on developing an ANN-based model to predict the permanent deformation behavior of blended CDW. Experimental data from laboratory repeated load triaxial tests are employed to train further and validate the ANN model. The drawn conclusions from the obtained results can be summarized as follows:
1.The repeated load triaxial tests indicated that the newly developed matrix of blended aggregates could serve adequately as base or subbase materials under traffic loading. The SSR and deviator stress significantly influence the deformation behavior of the CDW materials. The effects of deviator stress, confining pressure, and moisture content are more noticeable at a high SSR (e.g., SSR = 0.7) compared to a lower SSR (e.g., SSR = 0.5). The blended materials showed sensitivity to moisture variation, with moisture content above the optimum producing considerable permanent strain.
2.The ANN model was proposed to predict the permanent deformation of blended CDW and VA. Due to the material’s complexity and the dataset's enormous size, the optimum neural network determined in this study was a four-layered network with two hidden layers. The complexity associated with the CDW was fully incorporated into the model. The ANN model demonstrated a high degree of accuracy, with an average coefficient of determination of 0.99.
3.The performance criteria of the developed ANN model indicate its capability to predict the permanent deformation of the blended proportions with high precision and accuracy. Additional performance criteria were utilized to determine the model's applicability to predict its performance when applied to an external dataset. The results confirmed the model's adequacy in predicting the permanent deformation of external data.
4.The comparison between the ANN-based and regression-based models revealed that the ANN model outperforms conventional regression-based models. The ANN model, accommodating various combinations of input parameters at any given loading application, can predict the accumulative permanent strain more efficiently than the regression model, with improved computation time. Most regression-based models rely on specified equations that involve tedious linear and nonlinear calculations.
5.The ANN model surpasses the regression-based model by determining each input parameter's contribution through network weights via sensitivity analysis. The result indicated that all the selected input parameters influence the accuracy of the permanent deformation model.
6.Sensitivity analysis revealed that all input parameters are essential for predicting permanent deformation. Testing parameters such as the number of loading cycles, confining pressure, and SSR significantly control the permanent deformation behavior of the materials. With a total sensitivity coefficients close to 80%, the loading condition greatly affects the permanent deformation behavior of the blended CDW. The content of construction and demolition waste and moisture content also play roles in determining the material's response.”
Point 8: The paper requires significant language and presentation improvements. The writing should be concise, grammatically accurate, and avoid unnecessary jargon. Graphs, figures, and tables should be clearly labeled and adequately explained in the text.
Example: see-Table-3 (Specimen volume cm3), Table-6
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We apologize for the poor language and presentation of the original manuscript. We have thoroughly revised the grammar and format and hope that the revised manuscript would be up to your satisfaction.
Page 4, Lines 156:
Table 3. Laboratory compaction test parameters according to Chinese standard (JTG 3430-2020).
|
Test Method |
Specimen Height (cm) |
Specimen volume (cm3) |
Sub-layers |
Blows per sub-layer |
Hammer weight (kg) |
Falling Height (mm) |
Maximum particle size (mm) |
|
Heavy II-2 |
12 |
2177 |
3 |
98 |
4.5 |
450 |
40 |
Page 16, Lines 440:
Table 6. Statistical criteria for external validation of ANN model.
|
Criterion |
Limit |
Result Obtained |
|
It should be close to R2 |
1.000 |
|
|
It should be close to R2 |
1.000 |
|
|
-8.0614e-06 |
||
|
-8.0614e-06 |
||
|
0.997 |
||
|
1.000 |
||
|
1.000 |
Point 9: Addressing these major points of concern will greatly enhance the quality and impact of the paper. The authors are encouraged to undertake a thorough revision, taking into consideration the constructive feedback provided, to ensure that the paper meets the standards of rigorous scientific inquiry and contributes meaningfully to the field.
[Authors’ response]: Thank you again for reviewing this manuscript and sharing your valuable comments. As you may find, all your review comments have been addressed one by one with itemized responses. It is our hope that the revised manuscript would be up to your satisfaction.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper's primary objective is to formulate an artificial neural network model for forecasting the lasting deformation characteristics of blended construction and demolition waste. To achieve this, laboratory-based repeated load triaxial test outcomes are utilized for the model's training and validation. However, the quality of the paper need to be improved, especially for some Tables and Figures. Additionally, the following issues should be addressed:
1. Table 1 and Table 3 were not described in the manuscript, and they should be explained.
2. In line 150, the Figure 3 should be Figure 4; and in line 153, the Figure 2 should be Figure 3. There are several similar errors throughout the entire paper. Please carefully check and make the necessary corrections.
3. In Figure 5b, why is there no curve for SSR = 0.9?
4. In Figure 6b and Figure 7b, please provide a detailed clarification regarding the differences in the curves presented for the confining pressures of 100 kPa and 150 kPa compared to the other curves.
5. In Line 227, where is equation 21? Is it equation 5？Is this equation a result of the author's research? If not, please provide the appropriate reference to the corresponding literature.
6. In Lines 228 to 231, it seems like there are missing letters, and the expression is incomplete.
7. In Line 260, “As proposed by previous researchers, the minimum ratio of the dataset over the number of input parameters for model acceptability is three.” Could the author clarify which “previous researchers”?
8. In Line 280, the equation 3 should be equation 6.
9. In Lines 398 to 405, these equations have been misnumbered and need to be reorganized.
10. The analysis of the data in Tables 6, 7, and 8 lacks depth. The author should delve into the patterns of the data and examine the reasons for any errors and deviations.
The language quality of the paper need to be improved.
Author Response
The paper's primary objective is to formulate an artificial neural network model for forecasting the lasting deformation characteristics of blended construction and demolition waste. To achieve this, laboratory-based repeated load triaxial test outcomes are utilized for the model's training and validation. However, the quality of the paper need to be improved, especially for some Tables and Figures. Additionally, the following issues should be addressed:
Point 1: Table 1 and Table 3 were not described in the manuscript, and they should be explained.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. According to your suggestion, we have added more information to described the Table 1 and Table 3, as quoted below for your review.
Page 3, Lines 141 to 142: “the main composition ratio of CDW used in this study are shown in Table 1.”
Page 3, Lines 149 to 150: “The compaction test parameters are listed in Table 3.”
Point 2: In line 150, the Figure 3 should be Figure 4; and in line 153, the Figure 2 should be Figure 3. There are several similar errors throughout the entire paper. Please carefully check and make the necessary corrections.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for such careless mistakes in the original manuscript. According to your suggestion, we have revised these mistakes, as quoted below for your review.
Page 5, Lines 176 to 180: “The Mohr-Coulomb criterion, depicted in Figure 4, defines the SSR as the ratio of shear stress to shear strength, as shown in Equations (1) through (4). The time-history curve of axial stress and the loading waveform of half-sine with a frequency of 5 Hz is depicted in Figure 3.”
Point 3: In Figure 5b, why is there no curve for SSR = 0.9?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the confusion in the original manuscript. We have redrawn Fig. 5b, as quoted below for your review.
Page 7, Lines 210:
Point 4: In Figure 6b and Figure 7b, please provide a detailed clarification regarding the differences in the curves presented for the confining pressures of 100 kPa and 150 kPa compared to the other curves.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the confusion due to the lack of clarification and details. According to your suggestion, we have added in the revised manuscript the explanations for the differences between the curves presented for the confining pressures of 100 kPa and 150 kPa and the others in Figure 6(b) and Figure 7(b), as quoted below for your review.
Page 9, Lines 238 to 247: “The discrete nature of the components in the CDW should be noted, as it leads to significant fluctuations in the strain curves of certain test samples. For example, in Fig. 6(b), Fig 7(b), and Fig. 8(b), the strain curve exhibits fluctuations. As shown in Fig. 6(b), the cumulative axial strain of the specimen during the pre-test period at σ3 = 100 kPa is actually higher than that of the specimen tested at σ3 = 150 kPa. This phenomenon can be attributed to the complex composition of the CDW, where the components have varying mechanical properties. The presence of impurities with poor mechanical properties in the specimen makes it more susceptible to deformation, thus reducing its overall resistance to deformation.”
Point 5: In Line 227, where is equation 21? Is it equation 5? Is this equation a result of the author's research? If not, please provide the appropriate reference to the corresponding literature.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for such a careless mistake in writing. There should be equation (5) and reference [21]. We have corrected the mistake in the revised manuscript, as quoted below for your review.
Page 10, Lines 265 to 265: “In each hidden layer, the input variables are multiplied by the weights and summed, and a bias value is added to the system [21], as shown in Equation (5).”
Point 6: In Lines 228 to 231, it seems like there are missing letters, and the expression is incomplete.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. The equation (5) references the literature [21]. According to your suggestion, we have added the reference, as quoted below for your review.
Page 10, Lines 263 to 265: “In each hidden layer, the input variables are multiplied by the weights and summed, and a bias value is added to the system [21], as shown in Equation (5).”
Point 7: In Line 260, “As proposed by previous researchers, the minimum ratio of the dataset over the number of input parameters for model acceptability is three.” Could the author clarify which “previous researchers”?
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for missing the important references in the original manuscript. We have added the literature in the revised manuscript, as quoted below for you review.
Page 11, Lines 314 to 315: “As proposed by previous researchers, the minimum ratio of the dataset over the number of input parameters for model acceptability is three [57].”
Point 8: In Line 280, the equation 3 should be equation 6.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the clerical error. We have revised the mistake, as quoted below for your review.
Page 12, Lines 342 to 343: “The log-sigmoid transfer function, as described in Equation (6), is considered appropriate for this study.”
Point 9: In Lines 398 to 405, these equations have been misnumbered and need to be reorganized.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We apologize for the mistakes in writing. According to your suggestion, the equation numbers have be revised, as quoted below for your review.
Page 18 to 19, Lines 477 to 483:
Normalization of the collected weight from successive layers:
|
, |
(12) |
Multiplication of the normalized weight matrices:
|
, |
(13) |
where w1 represents the normalized weight matrix between the input layer and the output layer. The multiplication is carried out till the final.
|
, |
(14) |
The relative importance percentage can be obtained from the equation below.
|
, |
(15) |
Point 10: The analysis of the data in Tables 6, 7, and 8 lacks depth. The author should delve into the patterns of the data and examine the reasons for any errors and deviations.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We apologize for missing detailed discussion on the model’s parameters and performance. According to your suggestion, we have added more details in the revised manuscript, as quoted below for your review.
Page 18, Lines 484 to 489: “It is generally believed that the weight value represents the degree of one neuron’s influence on another. The weight and bias values between the neurons in various layers of the optimal ANN model are shown in Tables 7, 8, and 9, which can be accessed by readers to use and further improve the developed ANN. The relative significance of the input variables to the predicted permanent deformation can be deduced from these values using Equation (14)-(17).”
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments are addressed
Author Response
Point 1: Comments are addressed.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments.
Author Response File: Author Response.pdf
Reviewer 2 Report
The author has thoroughly responded to all the comments given by the reviewer and has made improvements in the manuscript, hence now the research paper is accepted from my side.
Author Response
Point 1: The author has thoroughly responded to all the comments given by the reviewer and has made improvements in the manuscript, hence now the research paper is accepted from my side.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments.
Author Response File: Author Response.pdf
Reviewer 3 Report
The author has addressed the previous suggestions and comments. However, it is advised to continue checking for minor grammatical errors in the manuscript, such as on Page 18, Line 474, where 'equation' should be 'equations', and the "equations" should be simplified as "Eqs.", and so on.
It is advised to continue checking for minor grammatical errors in the whole manuscript.
Author Response
Point 1: The author has addressed the previous suggestions and comments. However, it is advised to continue checking for minor grammatical errors in the manuscript, such as on Page 18, Line 474, where 'equation' should be 'equations', and the "equations" should be simplified as "Eqs.", and so on.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We apologize for the mistakes in writing. According to your suggestion, we have revised such minor grammatical errors, as quoted below for your review. In addition, we have proof checked the entire manuscript thoroughly and corrected similar mistakes wherever necessary, as you may find from the revised version.
Page 5, Lines 176 to 178: “The Mohr-Coulomb criterion, depicted in Figure 4, defines the SSR as the ratio of shear stress to shear strength, as shown in Eqs. (1–4).”
Page 12, Lines 341 to 342: “The log-sigmoid transfer function, as described in Eqs. (6–7), is considered appropriate for this study.”
Page 12, Lines 360 to 361: “These statistical indexes aid in determining the best model architecture. These parameters are defined below Eqs. (8–11):”
Page 18, Lines 474 to 475: “The steps are summarized in the Eqs. (14–17) below.”
Page 19, Lines 484 to 488: “The weight and bias values between the neurons in various layers of the optimal ANN model are shown in Tables (7–9), which can be accessed by readers to use and further improve the developed ANN. The relative significance of the input variables to the predicted permanent deformation can be deduced from these values using Eqs. (14–17).”
Point 2: It is advised to continue checking for minor grammatical errors in the whole manuscript.
[Authors’ response]: Thank you for reviewing this manuscript and sharing your valuable comments. We sincerely apologize for the careless quality deficiency in English language and writing. According to your suggestion, we have proof checked the entire manuscript thoroughly and corrected mistakes wherever necessary, as you may find from the revised version.
Author Response File: Author Response.pdf
