Machinability investigation of natural fibers reinforced polymer matrix composite under drilling: Leveraging machine learning in bioengineering applications Open Access

The inception of composite materials can be traced back to the utilization of mud and straw by the ancient Egyptians in the production of bricks.¹ However, using composite materials for commercial purposes commenced in the early 1800s, incorporating cellulose fibers as a reinforcement for resins.² The potential of natural fiber-reinforced polymer (FRP) composites as ecologically friendly substitutes for conventional synthetic fiber materials has attracted much interest lately.³ These composites, which contain jute and rattan fibers, are appealing for a range of industrial applications because they provide an appealing blend of mechanical strength, biodegradability,⁴ and affordability.⁵ However, there are special difficulties in machining these composites,⁶ especially when it comes to getting the right surface properties and reducing delamination while drilling.^7,8 The creation of natural fiber-reinforced polymer composites has potential uses in bioengineering, including scaffolds for tissue engineering and biomedical devices.⁹ Although machining is essential for shaping these materials to fulfill certain design specifications, it is still difficult to maximize machining parameters for better results.^10,11

In recent years, there has been a significant increase in the study of natural fiber-reinforced composites (NFRCs) due to their comparable specific properties compared to composites based on synthetic fibers.¹² Polymeric composites reinforced with natural fibers are commonly utilized as structural components in the automotive, aerospace, and construction industries due to their superior mechanical characteristics, biodegradability, high strength-to-weight ratio, cost-effectiveness, and widespread availability.^13,14 Jute fiber, one of the many types of natural fibers available, is one of the reinforcement materials frequently utilized in manufacturing polymer composites for various structural applications.¹⁵ Extensive research has been carried out regarding the characterization of polymeric composites with jute fibers.¹⁶ The results indicate that the incorporation of jute fibers into polymer composites leads to a significant improvement in their overall performance.^17,18 In addition, another natural fiber rattan incorporation approach can enhance the material’s mechanical properties and toughness (Refs. 19 and 20). According to the research findings by Rachchh et al.,²¹ rattan fiber is becoming increasingly popular as a cost-effective and presumably ecologically superior substitute for glass fibers in composites.

The process of drilling holds significant importance in the manufacturing industry, as it is considered one of the fundamental machining procedures during the final assembly of parts.^22,23 On the other hand, the machining of fiber-reinforced polymer composites poses a challenge due to the microstructural anisotropy and inhomogeneity, as well as the abrasive nature of their reinforcement components. During machining, there is a significant and rapid development of cutting tool wear, which can damage the machined parts’ surface integrity.^24,25 Fiber damage, debonding, fiber pull-out, stress concentration, heat damage, spalling, microcracking, delamination, etc. are only some of the negative outcomes of drilling fiber-reinforced composites.^26,27 However, this constraint can be solved through the development of appropriate experimental methodologies, machining settings,²² and the selection of cutting conditions.^28,29

Numerous scholars have researched the process of drilling fiber-reinforced polymer (FRP) composites. The study conducted by Lotfi et al.³⁰ provided a comprehensive analysis of the material properties, manufacturing techniques, and machinability aspects of FRCs. The conventional mechanical drilling method was determined to be economically viable and convenient. Moreover, according to Rajmohan et al.,³¹ enhancing the quality of the hole can be achieved by utilizing a drill tool with a smaller diameter, increasing the speed, and reducing the feed. In addition, Mohibuddin Bukhari et al.³² conducted an empirical study on composites with a fiber-reinforced polymer matrix. The authors examined the impact of various drill bit geometries, including helix angle, lip angle, point angle, and drilling process parameters, on the resultant quality of the drilled hole. The attainment of superior hole quality can be facilitated by optimizing a drill bit’s helix angle, point angle, and drilling conditions. The primary concern in fiber reinforced polymer composites is delamination, which arises from the complex and non-uniform structure of the fibers that are characterized by their abrasive, hard, heterogeneous, and anisotropic nature.³³ Delamination significantly impacts the dimensional precision and surface quality of the aperture, resulting in a rejection rate of 60% during the assembly phase.³⁴ Drilling parameters of the hybrid composite basalt/glass were studied by Patel et al.,³⁵ along with the impact of tool geometry and lamination sequences on peel-up and push-down delamination. Drill bits with a parabolic geometry and a high feed rate have less delamination when they leave the hole. They found that drilling hybrid composites with low spindle speed, low feed, and a drill bit with parabolic geometry yielded the best hole quality. In another study, the drilling behavior of woven jute fiber-reinforced polymer composites was examined by Yallew et al.³⁶ Polypropylene pellets were employed as a matrix, and jute textiles reinforced the composite. They employed twist, Jo, parabolic carbide drills with an angle of 118° and an 8 mm diameter, three feed rates, and spindle speeds to drill the samples. They also found a similar result that the delamination factor increases for the twist drill and Jo geometries and decreases for the parabolic drill as the feed rate increases. In addition, they also added that the value of this factor drops for the twist and Jo geometry and marginally rises for the parabolic boring geometry as the spindle speed rises.

The analysis of delamination tendency and surface roughness during the drilling of Fiber-Reinforced Polymer (FRP) composites can be achieved by utilizing suitable models that establish a correlation between the process parameters and the response. Numerous researchers have employed the response surface methodology (RSM) in their experimental investigations, which is efficacious.^37–39 When the number of process parameters increases, traditional experiments may require more trials to determine the optimal combination of components to achieve the intended experimental circumstances. The RSM technique, on the other hand, reduces the total number of trials by a significant margin. Multiple Attribute Decision Making (MADM) methods also optimized process parameters during machining.^40–42 Ravikumar et al.⁴³ employed an integrated approach of the Gray Taguchi-based Technique for Order of Preference by Similarity to Ideal Solution (GT-TOPSIS) to optimize drilling parameters such as drill bit speed, feed, and cutting point angle. Through the utilization of this algorithm, it was determined that the experiment was executed at the optimal input parameter setting. Before that, Sonkar et al.⁴⁴ also utilized the TOPSIS approach to optimize the machining process parameters while considering multiple performance characteristics. The primary benefit of this approach is that it eliminates the requirement to compute intricate modeling equations or process simulations, which can be time-consuming and resource-intensive in determining the optimal solution. Using precise empirical data to represent the process results in significantly more dependable solutions, as provided by this algorithm.⁴⁵ 

Notwithstanding the various computer-based modeling methodologies, manufacturing engineers have also found machine learning advantageous for predictive analysis.^37,46,47 Nayak et al.⁴⁸ used an SVR method to foretell delamination at drill entry and exit in GFRP composites. Three performance criteria—root mean square error (RMSE), Nash-Sutcliffe efficiency coefficient, and coefficient of determination (R²)—are used to assess the model’s efficacy. Researchers discovered that using the model saved money and time compared to conducting the same analysis on actual GFRP composite drilling operations. Prakasvudhisarn et al.⁴⁹ suggested the utilization of support vector machines (SVMs) to capture the attributes of roughness and its associated factors. The study also revealed that the SVM model exhibited superior predictive capabilities for surface roughness compared to the regression model. In another study, Jennings et al.⁵⁰ conducted a comparative study on machine learning algorithms for smart manufacturing. They used an Artificial Neural Network (ANN) using the feed-forward back propagation method (FFBP) and random forest (RF) to predict tool wear. Empirical evidence has demonstrated that RFs can produce predictions with higher accuracy compared to Feed-Forward Backpropagation Artificial Neural Networks (FFBP ANNs) that consist of a solitary hidden layer. Another great feature of SVMs and RFs is that the training work may be transformed into a linearly constrained quadratic programming problem with a unique solution. This results in a singular, optimal, and universally consistent solution. They have even been developed to address regression issues.

A comprehensive examination of the existing literature has determined that natural fiber-reinforced composites (NFRCs) are increasingly recognized for their benefits. The machinability of composite materials is a determining factor in their overall utility. However, research on machinability analysis in the context of NFRCs is scarce. With that in mind, the present research aimed to explore the machining of a jute and rattan-based hybrid natural fiber-reinforced polymer matrix composite using a traditional drilling machine. The impact of spindle speed and feed rate, along with two distinct types of tool material (HSS and carbide), is examined concerning surface roughness and delamination. The experiment was designed using the RSM, followed by parameter optimization using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). The inspection of hole quality is conducted using a scanning electron microscope (SEM). Finally, the data obtained from the experimentation were analyzed, and two predictive models for surface roughness were developed using SVM and RF algorithms. These models aim to enhance the analysis of machinability. Therefore, the research objectives of the study are as follows:

• Investigate Machinability Factors: To examine and identify key machinability factors influencing the drilling process of jute–rattan fiber-reinforced polymer composites, focusing on surface roughness and delamination. Explore the chemical and mechanical aspects of these factors to enhance understanding.

• Evaluate Material Performance: To assess the performance of jute–rattan fiber-reinforced polymer composites under different machining constraints, considering the specific influence of drill bit materials, spindle speed, and feed rate. Examine the chemical interactions and structural integrity during the drilling process.

• Optimize Machining Parameters: Using RSM and TOPSIS to optimize machining parameters and investigate the optimal combinations of drill bit material, spindle speed, and feed rate for minimizing surface roughness and delamination in the composite material. Compare the machinability of jute–rattan fiber-reinforced polymer composites using different drill bit materials, specifically high-speed steel and carbide.

• Apply Machine Learning (ML) Techniques: Implement ML models, specifically RF and SVM, to predict surface roughness and delamination. Evaluate the accuracy and performance of these models in predicting machining outcomes based on input variables, and validate the accuracy and reliability of the ML models.

The study on the machinability investigation of natural fiber-reinforced polymer matrix composites under drilling introduces a compelling study at the intersection of materials science and chemistry. Epoxy resin is used as the polymer matrix in jute–rattan fiber-reinforced composites, which are the subject of this study. One important chemical ingredient affecting the composite material’s mechanical and chemical characteristics is epoxy resin. The increasing need for eco-friendly and sustainable materials drives interest in developing composites made of natural fibers. The selection of locally obtained jute and rattan fibers that have undergone chemical processing is an excellent example of how chemistry and materials engineering work together to create composites with improved mechanical qualities. In addition, the research explores the effects of machining limitations on drilling-related surface roughness and delamination, addressing real-world issues with using these composites in industrial settings.

The study also investigates the machining process under compressed-air cooling conditions, highlighting the influence of cooling on the chemical interactions and thermal properties of the composite material during drilling. Including different drill bit materials, such as high-speed steel and carbide, introduces chemical composition and property variations, further emphasizing the interdisciplinary nature of the research. Additionally, using RSM and ML techniques, such as RF and SVM, adds a computational chemistry dimension to the study. The optimization of machining parameters through the TOPSIS method reflects a systematic approach, considering the chemical and physical interactions involved in the machining process. This study emphasizes the critical role that chemistry plays in customizing and improving natural fiber-reinforced polymer composites for industrial applications, in addition to investigating the mechanical elements of drilling them.

The organization of the rest of the paper includes Sec. II, “Materials and Methods,” and is divided into Sec. II A, Materials used in the study, with a focus on the fabrication procedure (Sec. II A 1) that explains how these materials were processed and prepared. Section II B outlines the experimental conditions under which the study was conducted, providing essential context for the subsequent analyses and results. Section II C presents the TOPSIS methodology used for optimization, describing its application in the study. Section II D introduces the SVM technique and its use for predictive modeling. The RF algorithm’s application and details about the performance indicators are considered in Sec. II E 1. Section II F addresses the data pre-processing techniques for preparing the dataset for analysis. Section III, “Results and Discussion,” comprises SEM analysis of drilling holes: an examination of drilling hole characteristics using SEM. Section III B, “Optimization using TOPSIS,” is a discussion of the optimization process and its outcomes. Section III C, Prediction of Response Variables by SVM and RF, provides insights into the predictive models generated using SVM and RF, followed by a detailed analysis of the response variables (Sec. III C 1). Section IV explores the practical implications of the research findings, demonstrating their relevance to the drilling industry and related fields. Finally, Sec. V summarizes the key conclusions (Sec. V A) drawn from the study and acknowledges its limitations (Sec. V B), ensuring a comprehensive understanding of the research’s contributions and boundaries.

This investigation used epoxy resin (Lapox Metalam System B) as the matrix, and jute and rattan fibers were used as reinforcement. The composites’ interfacial adhesion and strength were enhanced by adding a hardener. Jute and rattan fibers are sourced regionally and undergo chemical processing before use. The epoxy resin and hardener came from a chemical supply store. The procedure utilized a 2:1 ratio of epoxy resin to hardener.

Various techniques are utilized in industrial settings to produce composites, including compression molding, vacuum molding, and resin transfer molding. Hand lay-up manufacturing is a straightforward and uncomplicated technique for producing composite materials. The composites in this study were fabricated using the hand lay-up process shown in Fig. 1(a). The fibers underwent a weaving process to form mats before being integrated into the composite material. Producing the woven mats involved repeatedly interlacing jute fibers in a parallel orientation. In contrast, the rattan fibers were interlaced at an angle of 45°, as shown in Fig. 1(b).

Initially, a releasing agent and resin were administered to the surface of the mold. Next, a layer of jute or rattan fiber was carefully placed, and a specific volume of liquid epoxy resin was poured onto it. Brushes and hand rollers were employed to eliminate voids within the fiber structure and ensure uniform resin distribution across the fibers. The procedure was iterated until the desired quantity of layers was accumulated. Ultimately, the specimens are subjected to a hydraulic press to eliminate excess air trapped between the fibers and resin. Subsequently, the samples are left undisturbed for several hours to achieve optimal quality. Once the composite material has fully cured, it is removed from the hydraulic press, and any rough edges are carefully trimmed. The composite laminate samples underwent curing through exposure to typical atmospheric conditions. Figures 2(a) and 2(b) depict the final composite as photographed from the bottom and top, respectively.

The machinability investigation was conducted using a radial drilling machine by utilizing two different types of drill bits composed of High-Speed Steel (HSS) and carbide materials. Both drill bits had a diameter of 8 mm and were employed under compressed cooling air conditions to ensure that the mechanical properties of the composite were not compromised. The workpiece is securely clamped in a vice, and a range of process parameters were chosen, as indicated in Table I. The input variables in this study included drill bit material, spindle speed, and feed rate, while the response variables were average surface roughness, entry delamination, and exit delamination.

The TOPSIS method is a commonly employed technique in multi-objective optimization, renowned for its straightforwardness and efficacy in effectively tackling multi-response issues.^51,52 This methodology is predicated on utilizing attribute information furnished by decision-makers in conjunction with numerical data. The primary aim of this approach is to evaluate, prioritize, and choose subjective inputs through the allocation of weights. As a result, several authors have subsequently introduced and elaborated on this methodology.^24,53,54 The elucidation of standard TOPSIS methods can be presented as follows:⁵⁵ 

• Step 1: Normalized decision matrix (D_ij) construction

where J is the set of benefit attributes or criteria (larger is better) and J′ is the set of negative attributes or criteria (smaller is better).

• Step 4: Calculate the separation measures (⁠$Si+andSi−$⁠) for each alternative

A set of alternatives can now be ranked according to the descending order of $Ci*$⁠.

The original SVM for regression was developed by Smola and Schölkopf⁵⁶ in 2004. An SVM is a computational algorithm that constructs a hyperplane or a set of hyperplanes in a high-dimensional or infinite-dimensional space. These hyperplanes are utilized for classification and regression. The SVM is a supervised learning algorithm that is utilized for the prediction of discrete values. SVR is a method that falls within the realm of SVMs.⁵⁷ The primary objective of this technique is to obtain the line of best fit, which is a hyperplane that maximizes the number of points it encompasses.⁵⁸ The framework of SVM regression is depicted in Fig. 3. The process of determining the hyperplane in SVR involves the selection of extreme points or vectors. These extreme points are referred to as support vectors, which aligns with the naming convention of this technique.

The RF is categorized as a supervised learning algorithm. It is an ensemble learning technique for various tasks such as classification, regression, and other similar objectives. Its functioning involves the creation of numerous decision trees during the training phase, and its output is determined by either the class that appears most frequently among the trees or the average prediction generated by the individual trees.^47,61 This model is widely used owing to its simplicity and diversity for regression and classification. The RF model construction is depicted in Fig. 5.

The experimental results for the average surface roughness, entry delamination, and exit delamination of HSS and carbide drill bits were imported into Google Collaborator using the Pandas Python library. Pandas is commonly utilized for data manipulation. Utilizing the one-hot encoding technique facilitated the transformation of categorical data, specifically the drill bit material, into numerical data.

In the field of optimization, the utilization of the design of experiments is of utmost importance as it significantly contributes to enhancing the reliability of outcomes and reducing the need for conducting excessive experiments while maintaining precision. The concept under consideration is that of accuracy. RSM is widely recognized as a robust and effective approach for modeling and analyzing the influence of controllable factors on various responses, including surface roughness.

32 experiments were conducted using the Box-Behnken approach, facilitated by the design expert software. Figure 6 displays the drilled holes within the composite material. The Mitutoyo Talysurf tester was utilized to evaluate the surface roughness of the machined composite specimens. Table II contains a rundown of the findings from the experiment.

Drilled holes were examined using Scanning Electron Microscopes (SEMs). The images presented in Figs. 7 and 8 illustrate occurrences of fiber pull-outs and uncut fibers, which serve as indications of delamination. The initial two images, Fig. 7(a) and 7(b), depict the perspectives of the drilling composite during the entry and exit stages when utilizing a High-Speed Steel (HSS) drill bit. Conversely, the final two images, Fig. 8(a) and 8(b), exhibit these same perspectives but with the utilization of a carbide drill bit. When the feed rate is increased, there is an observed increase in the undeformed chip thickness, which subsequently leads to an increase in the cutting force. An elevated cutting force has the potential to expand the scope of the angle between the orientation of the fibers and the direction of cutting, thereby resulting in a larger occurrence of fiber pull-outs across a wider area, as found by Alizadeh Ashrafi et al.⁶³ while investigating the machining characteristics of fiber reinforced polymer composites.

Optimization under compressed cooling air machining with variable responses, such as surface roughness and delamination, was determined using TOPSIS. The experimental results demonstrate that the inclination confidence for each test combination can be attained by utilizing the specified settings (1–7). One can determine the preference value for each option based on their proximity to the optimal solution. All yield reactions are given equal weighting, as stated in Tables III and IV, because all of the characteristics above are equally significant when the material is machined under compressed cooling air for both the HSS and carbide drill bit. Therefore, the TOPSIS preference values collected with the rank request from each test are adjusted. The relative proximity to the intended arrangement in terms of the optimal level of performance achieves the highest value of preference and the highest rank. It is considered the most effective reward for the demonstration of the measure. In the case of the HSS drill bit, trial run 1 looks to have the best display qualities with the highest preference request; therefore, it is the optimal setting, whereas run 12 is the optimal setting for the carbide drill bit.

Therefore, the ideal setting acquired is as follows: for the HSS drill bit, spindle speed (V_s: 400 rpm) and feed rate (f: 0.25 mm/rev) are required to attain a surface roughness of 0.205 µm, entry delamination of 1.0176, and exit delamination of 1.0040. For the carbide drill bit, spindle speed (V_s: 800 rpm) and feed rate (f: 0.80 mm/rev) are required to attain a surface roughness of 0.233 µm, an entry delamination of 1.0112, and an exit delamination of 1.0059.

The response values from the drilling operation were predicted using two different regression-based machine-learning models. The total number of data points is 32 for model creation and evaluation. 80% of the input data were randomly picked for model construction (training). The model was validated using the remaining 20% of the input data (testing). In predicting surface roughness, entry delamination, and exit delamination by different models, three input variables are used: drill bit material, spindle speed, and feed rate. To minimize the errors that may arise due to the input parameters’ unit differences, training and testing data were scaled using one hot encoding. Cross-validation was performed using Grid Search CV to check the model’s underfitting or overfitting; both training and test errors were used to ensure the best parameter for the generated model.

The surface roughness, entry delamination, and exit delamination values were obtained from 32 experiments for training and testing purposes. Different predicted values by SVM and RF models are presented in Table V. Three performance indicators are used to judge the predicted models’ accuracy. The testing and training errors are shown in Table VI. The significance of the feed rate is pivotal for all the responses. In the case of surface roughness, with the increase in feed rate, it is observed that the surface roughness is also increasing as helicoid generation takes place and becomes deeper and wider. However, with the increase in feed rate, delamination is decreasing for both HSS and carbide drill bits. Khashaba et al.⁶⁴ observed similar findings while investigating the drilling process on woven carbon fiber-reinforced polymer (CFRP) plates. The presence of non-homogeneous structures within composites poses challenges in achieving a consistent surface quality across the entire length of a hole, from its entry point to its exit. The cutting process significantly affects the surface quality of composites, particularly concerning the orientation of the fibers. As per the error matrices, RF gave better results than SVM regression. RF and SVM’s prediction of response variables is plotted and compared with experimental values in Fig. 9. RF predicted values are closer to the experimental value, reducing the error for RF and making it a better-performing model than SVM.

The study’s broad implications include theoretical and practical aspects of manufacturing and machining composite materials. The research advances machining techniques, paving the way for increased production efficiency and effectiveness by identifying the best machining parameters for jute–rattan fiber-reinforced polymer composites. Investigating composites made of natural fibers, such as jute and rattan, offers a sustainable option for synthetic fibers that align with environmental goals and may even lower production costs. This study strongly justifies using jute–rattan composites in industrial applications as companies increasingly seek eco-friendly and economical materials.

The study’s use of machine learning methods, notably the RF model’s remarkable performance, points to a potential direction for AI-driven approaches in machining result prediction and optimization. This accomplishment inspires additional research into machine learning’s potential in many industrial sectors, which has the potential to improve accuracy and overall effectiveness. Additionally, the knowledge gained about how machining limitations affect surface roughness and delamination facilitates improved quality control procedures. With this knowledge, producers may set up more stringent quality control procedures to guarantee constant product excellence and reduce flaws and waste. The study’s general implications go beyond its immediate conclusions and impact manufacturing techniques, sustainability initiatives, and the landscape of machine learning applications in industrial optimization. The information gained from this study has the potential to improve procedures, increase the usage of composites made of natural fibers, and propel breakthroughs in the sector, providing advantages to academics, businesses, and society.

The jute–rattan reinforced composite machining process was performed under the influence of compressed-air cooling conditions. Two distinct types of drill bits were utilized during the process, and subsequent measurements were taken to assess each scenario’s resulting surface roughness and delamination. The research involved various machine learning algorithms to predict the response variables for both HSS and carbide drill bits. This aspect of the study, combined with the development of a jute–rattan reinforced composite, contributes to its novelty. The utilization of machine learning in contemporary society has brought about a revolutionary impact by effectively reducing errors, optimizing resource allocation, and enhancing time efficiency. Several conclusions can be derived from this study, as follows:

• The surface roughness values obtained experimentally for high-speed steel (HSS) drill bits are relatively lower when compared to those of carbide drill bits. However, no specific trend is observed for delamination.

• The TOPSIS method determined the optimal operating conditions for HSS drill bits under compressed air-cooling conditions. The results indicate that a spindle speed of 400 rpm and a feed rate of 0.25 mm/rev are necessary to achieve a surface roughness of 0.205 µm, an entry delamination value of 1.0176, and an exit delamination value of 1.0040.

• A spindle speed of 800 rpm and feed rate of 0.80 mm/rev are recommended by TOPSIS as the optimal configuration for a carbide drill bit to achieve a surface roughness of 0.233 m, entry delamination of 1.0112, and exit delamination of 1.0059.

• This study observed that the RF model exhibited superior performance compared to the SVM regression model. This conclusion was drawn based on the lower errors obtained from the performance metrics when comparing the two models.

• The R² values for the testing errors of the RF model are 0.997, 0.981, and 0.985 for surface roughness, entry delamination, and exit delamination, respectively. Conversely, the R² values for the training errors are 0.972, 0.982, and 0.908 for the three response variables.

Similar to any research endeavor, this study possesses inherent limitations that must be addressed. Given the recent emergence of the developed composite, it is imperative to undertake thorough investigations involving diverse fiber orientations and compositions to establish definitive patterns in the behavior of the composite. Furthermore, machine learning techniques can be employed to explore the prediction of additional response parameters, such as cutting force and cutting power, while incorporating different helix angles of the drill bit. To enhance accuracy, increasing the number of experimental runs to investigate a wider range of responses is possible.

The authors appreciate the support from Researchers Supporting Project Number (RSP2024R33), King Saud University, Riyadh, Saudi Arabia.

The authors have no conflicts to disclose.

All authors listed have significantly contributed to the development and writing of this article.

Md. Rezaul Karim: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Shah Md. Ashiquzzaman Nipu: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Md. Sabbir Hossain Shawon: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Raman Kumar: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Sheak Salman: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Amit Verma: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). El-Sayed M. Sherif: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Saiful Islam: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Muhammad Imam Ammarullah: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article.