An advance of optimization by Simulated Annealing Algorithm and ANFIS to minimize delamination during drilling Chopped Strand Mat GFRP

doi:10.21203/rs.3.rs-671365/v1

Download PDF

Research Article

An advance of optimization by Simulated Annealing Algorithm and ANFIS to minimize delamination during drilling Chopped Strand Mat GFRP

https://doi.org/10.21203/rs.3.rs-671365/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Glass Fibre Reinforced Polymer (GFRP) composites are commonly used in many applications, such as in the aerospace, construction, maritime and automotive sectors, owing to their lightweight, corrosion resistant, strong rigidity and lightweight to strength ratios. This form of material normally struggles on machining with surface delamination, fibre peeling up, fibre pulling away, fibre cracking, fibre de-bonding and fibre breaking. There is a need to optimize the drilling parameters specifically in reducing the type of damage known as delamination which is prevalent during drilling activity on Chopped Strand Mat GFRP material. Regression equations for the experimental results collected from drilling Chopped Strand Mat GFRP material are established for this purpose. Considering the established regression equations as objective functions, under parallel search methodology, Simulated Annealing Algorithm employs innovatively to refine the drilling parameters for reduced delamination. Simulated Annealing Algorithm (SAA) is programmed using C + + and the decrement factor, known as the cooling factor applied during the search process, is varied for the parallel search of optimization. Through this optimization strategy, the tailored drilling parameters for drilling Chopped Strand Mat GFRP material with minimal delamination are substantially enhanced, and the performance of SAA are verified by the Adaptive Neuro Fuzzy Inference System (ANFIS).

Chopped Strand Mat GFRP

Delamination

Simulated Annealing Algorithm

Parallel search

Optimization

ANFIS

Polymer-based matrix composites have broader uses in the structural, aircraft, maritime, automobile, etc. industries owing to their favourable properties such as light weight, corrosion resistant, rigidity and light weight to strength ratio. Artificial fibres are the typical fibre-reinforcing products in the production of these products for such industries. Owing to its biodegradability, recent past natural fibres often act as an alternative to artificial fibres, and the resultant composites are environmentally sensitive. Studies on these composites and their analysis for particular uses are also of great value to society. When the suitability of a commodity for an application is understood, machining should become a standard procedure for the specifications for assembling and finishing. Because of the heterogeneous nature of polymeric composites, challenges arise in the manufacture of such composites for an application as well as in machining. The machining is important for polymeric composites strengthened by fibre due to the physical isolation of constituent materials during the machining process. This leads to fibre pulling out, fibre peeling off, fibre de-bonding, fibre breaks and delamination. Delamination predominates particularly in drilling activity, causing harm to the hole inlet as well as the hole outlet. Again, managing this delamination is another problem, because it is triggered by unsuitable cutting powers, torque and chip shape used throughout the process. It is therefore important to pick correct machining parameters to monitor the ill-effects of response parameters during machining; particularly drilling in which delamination is a major problem for fibre-reinforced polymeric composites and the chosen parameters have a direct influence on the cost of machining, product quality and manufacturing cycle times (Vinayagamoorthy 2018).

Delamination is calculated in many of the research works as a ratio of maximal diameter at the damaged region to nominal diameter of the appropriate hole size (Wang et al. 2017). Abundant test work is seen in the delamination tests, taking into account various control parameters such as drill configuration, drill forms, drill size, drill diameter ratio, drill material, spindle speed, feed rate, material thickness, chisel edge length, point angle, flute length etc. (Arl et al. 2006; Mohan et al. 2005; Panneerselvam and Raghuraman 2021). The researchers find that thrust force and drilling torque have a substantial effect/role on delamination, hole quality, and surface finish in the analysis of drilling related damages (Tan and Azmi 2017). The results are studied/analysed under dry/wet environments with the above controllable parameters. In researching the impact of drill geometry, step drill was found to have developed less delamination relative to twist drill. In another analysis, spindle speed, feed rate, and flute length were greatly influenced by hole impact. In addition, the machinability has been significantly affected by the levels of each parameter, and control of the parameter levels is important to enhance the quality of the hole. Three different types of drills were used in the study of drilling induced delamination on GFRP material; helical flute (HSS) drill, Carbide tipped straight shank (K20), and solid carbide 8 facet drills. The findings show the features of drilled holes (Kumar and Sing 2017).

It is therefore important to choose correct parameters using the optimization techniques employed for its ranges. The single performance factor can be easily optimized from the experimental results using the signal to noise ratio (S/N) and analysis of variance (ANOVA). These optimal conditions will prevent the needless experimentation, time and cost. Many experiments are involved in using statistical methods such as S/N ratio, ANOVA, Taguchi 's gray relational analysis for multi-performance variables, Artificial Neural network, Fuzzy logic, Response Surface Methodology, etc. in this regard (Balaji et al. 2016; Vinayagamoorthy 2016). Experimentally, optimum drilling parameters are identified from the S/N ratio, and from the ANOVA table it is observed that the drill size plays a significant role in determining the delamination value. The optimization of process parameters in GFRP drilling was accomplished through the use of gray relational analysis linked to a fluffy logic system. ANOVA findings indicate that feed rate is the most important parameter which affects the multiple performances of delamination factor and surface roughness (Shunmugesh et al. 2017). These techniques are more adequate for analysing, simulating and finding/predicting the best suitable input factors for delamination-free machining operation. Again, the experimental results could be analysed through regression analysis to generalize and predict the experimental results for untried process parameters in the investigational work. This will pave the way for simulating a broader variety of parameters to evaluate correct parameter selection and mitigate the harm done by drilling activity with the ability to further reduce the degree of delamination and increase the consistency of the hole due to the heterogeneous existence of composite materials (Vinayagamoorthy 2018; Vinayagamoorthy et al. 2016; Prakash 2009).

The drilling parameters such as drill bit material, drill bit diameter, spindle speed and feed rate have been investigated to minimize the delamination for drilling Chopped Strand Mat GFRP (Panneerselvam et al. 2014). Another research work on Chopped Strand Mat Glass Fibre Reinforced Polymer (CSMat GFRP) presented the delamination studies correlated with the effects of drill diameter, drill diameter ratio, spindle speed and feed rate on delamination with the goal of minimizing delamination and optimizing the parameters by analyzing the S/N ratio (Panneerselvam and Raghuraman 2015). The present research in this paper further focuses on CSMat GFRP material with the goal of seeking correct and optimal drilling parameters setting via the Simulated Annealing Algorithm, as several of the literature have been successfully used with Genetic Algorithm, Particle Swarm Optimization, Ants Colony Optimization, Cuckoo Search Algorithm (Lim et al. 2014) and Tabu search as non-traditional optimization techniques. Non-traditional optimization approaches allow researchers to solve complex problems with discrete, continuous-discrete searches due to experimental limitations. Simulated Annealing Algorithm (SAA) is innovatively implemented in this research to exhibit quicker convergence and simpler application in optimizing drilling parameters such as drill diameter, drill diameter ratio, spindle speed and feed rate with a prime objective of specific parameter setting to minimize the delamination for CSMat GFRP material. Subsequently, Adaptive Neuro Fuzzy Inference System (ANFIS) validates the reported SAA results, simulating the experimental values.

2.1 Details for optimization by Simulated Annealing Algorithm

The experimental work was carried out on material made from CSMat GFRP by hand lay-up process. The CSMat GFRP has two constituent materials, isophthalic polyester resin and chopped strand mat glass fibres at the proportionate mixing of 60% and 40% by weight respectively. The E-glass fibres at 10µm in diameter, chopped length of 50 mm have been used in this CSMat GFRP composite fabrication. The experimental work aimed to study the effects on delamination of drill diameter and drill diameter ratio along with spindle speed and feed rate. Standard High Speed Steel (HSS) twist drills of Ø8 mm and Ø15 mm were selected as drill diameters, while 0.8 (Ø8/ Ø10) and 0.53 (Ø8/ Ø15) were the selected drill diameter ratios along with spindle speed and feed rate, each at three different levels: 1000 rpm, 2000 rpm, 3000 rpm and 50 mm/min, 150 mm/min, and 250 mm/min. Here, drill diameter ratio 0.8 was obtained by initially drilling with Ø8 mm drill and subsequently by Ø10 mm drill. Similarly, 0.53 was obtained by an initial drill of Ø8 mm and the next one of Ø15 mm. These parameter effects were studied employing L18 orthogonal array design matrix with replication for two times. The CSMat GFRP material drilling was performed at the MCV 400 machining centre. The drilled specimen was measured using DYNASCAN Profile Projector, Model PT 400 EM to evaluate the delamination (as a ratio of maximum diameter measured at the damaged zone to the nominal drill diameter). The measured delamination values were analysed individually by signal to noise ratio with respect to drill diameter and drill diameter ratio in order to study its main effect from mean values and find the optimal drilling parameters for minimal delamination. The statistical significance was also evaluated via ANOVA, and in both the cases it was found that feed rate has more effect on delamination (Panneerselvam and Raghuraman 2015). This experimental data (Panneerselvam and Raghuraman 2015) is the source for the regression analysis, by which regression equations were developed individually for the effects of drill diameter and drill diameter ratio with spindle speed and feed rate.

Taking into account the experimental limitation on optimizing accurate input parameters from the range of parameters used in the experiment, this paper has attempted to use a non-traditional optimization technique, Simulated Annealing Algorithm (SAA), which accelerates rapid convergence with parallel search approach.

2.2 Regression equations developed for setting objective functions

The regression models developed for the experimental work carried out on material, CSMat GFRP are given in Eq. 1 and Eq. 2 with the intention of minimizing the delamination factor under the impacts of drill diameter and drill diameter ratio. Eq. (1) is concerned with drill diameter effects and indicated as,

F_d1 = 1.0215 + (0.003349*D) – (0.000008*N) + (0.000182*f) (1)

where,

F_d1 – Delamination factor under the effect of drill diameter

D – Drill diameter (mm)

N – Spindle speed (rpm)

f – Feed rate (mm/min)

By Eq. 1, delamination factor was predicted for parameter settings as per the experimental design and compared to the experimental values (Fig. 1) and it is noted that the maximum error percentage was 1.47 percent while the minimum error percentage was 0.28 percent between the model and the experimental values (Fig. 2). This indicates that the regression model developed has good precision in predicting the delamination factor for the ranges of the drilling parameters.

Eq. 2 is concerned with effects of drill diameter ratio and given as,

F_d2 = 1.0677 – (0.0477*D_r) – (0.000008*N) + (0.000238*f) (2)

where,

F_d2 – Delamination factor under the effect of drill diameter ratio

D_r – Drill diameter ratio

N – Spindle speed (rpm)

f – Feed rate (mm/min)

For the drill diameter ratio (Fig. 3), the maximum percentage of error was found to be 0.82 percent while the minimum percentage of error was 0.07 percent (Fig. 4). The regression model once again has demonstrated its accuracy in predicting the delamination factor.

In this work, Eqs. 1 and 2 are considered as objective functions to optimize drilling parameters to minimize delamination with respect to drill diameter and drill diameter ratio. To achieve this objective, the lower and upper limits of each parameter used in the experimental work were considered to be constraints and therefore the objective function given in Eq. (1) shall be subjected to the following constraints:

Ø8 mm ≤ D ≤ Ø15 mm

1000 rpm ≤ N ≤ 3000 rpm;

50 mm/min ≤ f ≤ 250 mm/min

Whereas the objective function given in Eq. (2) shall be subjected to the constraints:

0.53 ≤ D_r ≤ 0.8

1000 rpm ≤ N ≤ 3000 rpm;

50 mm/min ≤ f ≤ 250 mm/min

The explanation behind the two objective functions employed in this paper is that the experimental optimization was initially performed to reduce delamination by observing the effects of drill diameter and subsequently to enhance the hole quality by additional drilling operation on the predrilled holes as a factor of drill diameter ratio. For example, Ø8 mm is reported as an ideal drilling parameter and used for drilling a hole size of Ø8 mm, and the hole is then drilled by Ø10 mm to achieve a drill diameter ratio of 0.8. As the overall reduction in delamination was achieved in two stages, two objective functions were used to determine the most appropriate parameter level for minimum delamination.

2.3 Methodology for Simulated Annealing Algorithm

Simulated Annealing Algorithm (SAA) is one of the non-traditional optimization techniques in the presence of number of solutions to find the globally optimal solution from the local optima. It is an efficient technique for achieving a minimum functional value for an objective and it simulates the natural behaviour of slow cooling of metals/alloys under the process of annealing. The algorithm begins with an initial point and a high enough initial temperature to have large search space. The next point, also known as neighbourhood point is created by Gaussian distribution in the vicinity of current point, and the difference in functional values is determined from these two points, original or current point and neighbourhood point. The point is accepted if the neighbourhood point gives smaller functional value than the current one. Else, it applies Metropolis algorithm to accept the new point. This will complete one iteration of the search procedure and the iterative process will be terminated when the temperature value is reached the absolute minimal difference in functional values as a defined termination criteria (Saravanan et al. 2003).

2.3.1 Simulated Annealing Algorithm

Simulated Annealing Algorithm simulates the method of slow cooling of metals/alloys by annealing in the metallurgical parlance during the quest procedure. Initially, temperature (T) is selected high enough to continue the search process, say 500°C; terminating criterion is selected as 0.001 which means that the iterative procedure is stopped when the difference in function value reaches the termination criterion; decrement factor is used from 0.5 to 0.9; neighbourhood points are generated using Gaussian distribution (Saravanan et al. 2003). Within this algorithm the steps involved are:

1. Initial point, X₁⁽⁰⁾ is assumed from the pool of parameter ranges fixed in the experimental work; T = 500°C; termination criterion = 0.001; decrement factor = 0.5 for the first search, 0.6 for the second search, 0.7 for the third search, 0.8 for the fourth search and 0.9 for the final search; In each search, set iteration, t = 0 and then t = t + 1.

2. Neoghbourhood point is determined from the Gaussian distribution,

X₂^(t+1) = X₁^(t) + σ [∑r_i – n/2]

X₁^(t) is the current point (combination of parameter setting of drill diameter or drill diameter ratio, spindle speed and feed rate)

X₂^(t+1) is the generated new point in the vicinity of current point

σ is the variance, = (maximum parameter value – minimum parameter value)/(2×3)

r_i is the random number set, i = 1 to n and n is the number of random numbers generated which is assumed to be 5.

3. If the change in functional value is, ΔE = E(X^(t+1)) – E(X^(t)) < 0, set t = t + 1 and T = T× decrement factor. Else, Metropolis algorithm is used to accept the new point by checking, e^{− ΔE/T} ≥ r, where r is a random number generated within the range of 0–1 and set t = t + 1 and T = T× decrement factor. Else step 2 is repeated until the acceptance of the neighbourhood point.

4. The optimum value is reported if the termination criterion is met. Otherwise, step 2 is repeated.

5. Similar search procedure is applied for other decrement factor, 0.6, 0.7, 0.8 and 0.9 respectively.

2.4 Adaptive Neuro Fuzzy Inference System

Modeling a system poses several problems and is overcome by certain methods of soft computing. At the same time, modeling results depend on the available experimental data, and improvement in result is achieved with increased number of experimental data. But, in complex systems generation of large experimental data becomes expensive. In such situation, Adaptive Neuro Fuzzy Inference System (ANFIS) may be proposed for modeling process input and output relationship. ANFIS is a neuro fuzzy technique and a hybrid system of fuzzy logic and neural network. The fuzzy logic takes care of the system’s ambiguity to be modelled while the neural network gives it a sense of adaptability. In this approach, fuzzy model is initially constructed on the basis of rules framed for input and output data of the system under study, and then neural network is implemented to fine tune the rules framed for initial fuzzy model.

In the present work, ANFIS is being developed in MATLAB to predict delamination under the effects of drill diameter and drill diameter ratio, and thus ANFIS is being used to validate the findings of SAA as the backbone for the identification of real world experimental system. ANFIS steps involved in modelling, simulating and predicting are classically described and illustrated for the case of delamination effected by drill diameter ratio. Figure 5 shows a Sugeno fuzzy inference system for the three inputs effecting delamination factor and a typical ANFIS structure for training and prediction of delamination factor under the effect of drill diameter ratio is shown in Fig. 6 with three inputs and one output. In the ANFIS structure, first layer is called fuzzification layer, a collection of fuzzy if then rules with triangular membership function for this work and second layer: product layer, third layer: normalized layer, fourth layer: defuzzification layer and fifth layer: output layer. Description of ANFIS implementation can be found in literature (Babajanzade Roshan et al. 2013; Teimouri and Sohrabpoor 2013; Kumar and Hynes 2019; Savkovic et al. 2019). The 18 training data set are known to be input training vectors and the number of epochs for ANFIS training was tried from 3 to 100 and it was found that convergence of results began from 2nd epoch and finally the number of epochs was set to 30. Figure 7 shows the 18 rules for the three inputs and one output parameters and those rules are used to predict the delamination factor as a response.

The program was coded in C++ programming language for Simulated Annealing Algorithm used for two separate objective functions (Eqs. (1) and (2)) with a set of constraints. The program was run for the different decrement factors such as 0.5, 0.6, 0.7, 0.8, 0.9 when determining the minimum delamination using the Simulated Annealing Algorithm to understand the drill diameter effect and drill diameter ratio effect. The optimal values of minimum delamination obtained by running the program with various decrement factors are summarized in Table 1 and Table 2 with respect to the objective functions.

Table 1 Minimum Delamination factor from the effect of drill diameter in SAA

Decrement factor	Minimum Delamination factor	Optimal drilling parameters
Decrement factor	Minimum Delamination factor	Drill diameter (mm)	Spindle speed (rpm)	Feed rate (mm/min)
0.5	1.066	13.7	1724	67.2
0.6	1.087	13.1	2053	207.1
0.7	1.042	9.3	2603	52.2
0.8	1.065	10.8	1273	97.8
0.9	1.037	8.1	2906	66.1

Table 2 Minimum Delamination factor from the effect of drill diameter ratio in SAA

Decrement factor	Minimum Delamination factor	Optimal drilling parameters
Decrement factor	Minimum Delamination factor	Drill diameter ratio	Spindle speed (rpm)	Feed rate (mm/min)
0.5	1.069	0.66	2525	222.8
0.6	1.051	0.78	1744	147.8
0.7	1.034	0.73	1970	70.5
0.8	1.045	0.55	2395	95.1
0.9	1.025	0.75	2520	57.4

Among the various decrement factors for the drill diameter effect, the SAA running at a decrement factor of 0.9 gives the minimum delamination of 1.037 and confirms the optimal parameters at the drill diameter of Ø8.1 mm, spindle speed of 2906 rpm and feed rate of 66.1 mm/rev. The SAA which was run at decrement factor of 0.9 also provides the minimum delamination of 1.025 for the effect of drill diameter ratio and confirms the corresponding optimal parameters at the drill diameter ratio of 0.75 (Ø8.1mm/Ø10.8mm), spindle speed of 2520 rpm and feed rate of 57.4 mm/rev.

In the optimization of drilling parameters, the minimum delamination factor obtained in relation to the decrement factor is also illustrated in Fig. 8 and Fig. 9. By Fig. 8, it is observed that the delamination by the effect of drill diameter is found to be a minimum at 0.9 decrement factor and from Fig. 9, delamination by the effect of drill diameter ratio is also found to be the minimum at a decrement factor of 0.9.

In comparison, it is noted that the delamination value obtained under the drill diameter ratio effect is lower than the value obtained under the delamination effect of drill diameter. The optimal parameter setting was therefore confirmed at Ø8.1 mm drill diameter, 0.75 drill diameter ratio, 2520 rpm spindle speed and 57.4 mm/min feed rate.

The minimum delamination was recorded as 1.021 for the optimal parameter setting by SN ratio analysis (Panneerselvam and Raghuraman 2015). For this experimentation, regression equation was developed to implement SAA and the optimum parameters were obtained for minimum delamination at 1.025. Table 3 lists the final results of these two techniques. Drilling time was also taken into account for comparison and to ensure the validity of the SAA findings. Eq. 3 was used to find the drilling time corresponding to the optimum parameters obtained from each technique.

Drilling time in minutes, T_m= ( L/(f)) (3)

where,

L, length of drill travel = 25.4 mm + 5 mm

25.4 mm is the thickness of CSMat GFRP plate used for drilling operation, 5 mm is assumed as a drill over travel and tool approach is neglected.

f, feed rate in mm/min

Table 3 reveals that the minimum delamination obtained from the study of SN ratio (Panneerselvam and Raghuraman 2015) is 1.021 and from SAA is 1.025. In these delamination values, a marginal change of less than 1% is observed. Again, it is also observed that there is a decrease in drilling time by 12.82% for the parameters setting obtained from SAA compared to SN ratio analysis. This adds benefits to the drilling of CSMat GFRP by conserving the time and cost. Next, it is found that drill diameter ratio and spindle speed used in SAA are lesser in 6.25% and 16% values compared to the corresponding parameters used in SN ratio analysis. During the drilling operation the decreased drill diameter ratio from 0.8 to 0.75 results in an increased amount of material removal. Moreover, the decreased spindle speed of 2520 rpm from 3000 rpm has a benefit of lowering temperature generation during the drilling operation, which in turn reduces delamination and wear of tool. In addition, SAA’s feed rate is 14.8 percent higher than that of the feed rate optimized by SN ratio analysis. This higher feed rate has an advantage of speeding up the drilling process, saving time and cost. From this discussion, one can confirm that SAA has efficacy in finding precise parameter setting for drilling CSMat GFRP material.

Table 3 Comparison: SN ratio Vs SAA

Sl. No.	Optimization Techniques employed	Minimum delamination obtained	Drilling time required (sec)	at Optimal drilling parameters
Sl. No.	Optimization Techniques employed	Minimum delamination obtained	Drilling time required (sec)	Drill diameter ratio	Spindle speed (rpm)	Feed rate (mm/min)
1	SN ratio	1.021	36.48	0.8	3000	50
2	SAA	1.025	31.8	0.75	2520	57.4
Changes		0.39% Increase	12.82% Decrease	6.25% Decrease	16% Decrease	14.8% Increase

3.1 Validating the results of SAA by ANFIS

In this work, ANFIS is employed to validate SAA findings as discussed in the preceding section under the heading of Adaptive Neuro Fuzzy Inference System. Fig. 10 displays the ANFIS model 3D surface plot that correlates drill diameter ratio, spindle speed and feed rate with delamination factor for drilling CSMat GFRP material. Delamination factor increases with decrease of drill diameter ratio (Fig. 10(a) & 10(b)), decrease of spindle speed (Fig. 10(a) & 10(c)) and increase of feed rate (Fig. 10(b) & 10(c)). It is also noted that feed rate ((Fig. 10(b) & 10(c)) has more influence on delamination than drill diameter ratio and spindle speed, and the minimum delamination is observed at 0.8 of drill diameter ratio, 3000 rpm of spindle speed and 50 mm/min of feed rate. Therefore, the 3D surface plot of MATLAB can be used to analyze the interaction effect of input parameters and to obtain optimum parameters for minimal delamination in drilling CSMat GFRP material as per the analysis of SN ratio (Panneerselvam and Raghuraman 2015).

In addition, the experimental data was trained and FIS output was obtained for the number of epochs of 30, and it is seen from Fig. 11 & Fig. 12, FIS output is more in accordance with experimental data at the value of 1.0554e^-06Root Mean Square Error (RMSE) for drill diameter ratio effect on delamination. The FIS output for drill diameter effect on delamination is also found at the value of 1. 0712e^-6 Root Mean Square Error (RMSE).

Table 4 Validating the results (F_d1) of SAA by ANFIS

Decrement factor	Drill diameter (mm)	Spindle speed (rpm)	Feed rate (mm/min)	F_d1by SAA	F_d1 by ANFIS*	Error %
0.5	13.7	1724	67.2	1.066	1.0691	-0.31
0.6	13.1	2053	207.1	1.087	1.0869	0.01
0.7	9.3	2603	52.2	1.042	1.0346	0.74
0.8	10.8	1273	97.8	1.065	1.0638	0.12
0.9	8.1	2906	66.1	1.037	1.0311	0.59

*RMSE = 1.0712 e^-06

Consequently, the ANFIS models developed for both cases of delamination under the effects of drill diameter and drill diameter ratio were used to confirm the obtained SAA results, and the comparative results are shown in Table 4 and Table 5 respectively. The findings of the analysis show that the results of SAA are in strong agreement with ANFIS because the error variation is within +/- 1%.

Table 5 Validating the results (F_d2) of SAA by ANFIS

Decrement factor	Drill diameter ratio	Spindle speed (rpm)	Feed rate (mm/min)	F_d2by SAA	F_d2 by ANFIS*	Error %
0.5	0.66	2525	222.8	1.069	1.0653	0.37
0.6	0.78	1744	147.8	1.051	1.0555	-0.45
0.7	0.73	1970	70.5	1.034	1.0334	0.06
0.8	0.55	2395	95.1	1.045	1.0469	-0.19
0.9	0.75	2520	57.4	1.025	1.0266	-0.16

*RMSE = 1.0554e^-06

In this study, the Simulated Annealing Algorithm was successfully implemented with parallel search using decrement factor to determine the minimum delamination for drilling CSMat GFRP material, and the results of SAA were verified with ANFIS model. The drawn conclusions are:

The parallel search applied in Simulated Annealing Algorithm at program running with 0.9 decrement factor for the effects of drill diameter as well as drill diameter ratio finds the delamination to be minimum.
By considering the impact of drill diameter, the optimum parameters for minimum delamination could be at drill diameter of Ø8.1 mm, spindle speed of 2906 rpm and feed rate of 66.1 mm/min.
A drill diameter ratio of 0.75 (Ø8.1 mm/Ø10.8 mm), a spindle speed of 2520 rpm and a feed rate of 57.4 mm/min may be the optimum setting for minimal delamination by considering the effect of drill diameter ratio.
This paper confirms that the predrill of Ø8.1 mm and the subsequent drill of Ø10.8 mm together with spindle speed of 2520 rpm and 57.4 mm/min produce minimal possible delamination during the drilling of CSMat GFRP material.
ANFIS was found to be the best fit model in predicting and checking the value of delamination in relation to the optimum input parameters obtained by SAA.

Acknowledgement

The authors express their sincere gratitude to The Vice-Chancellor, The Registrar and The Dean, School of Mechanical Engineering, SASTRA Deemed to be University for granting permission and extended support to the successful execution of their research.

Ethical Approval: Not applicable

Consent to Participate: Based on the contribution of Authors and with their consent, their names are mentioned in the manuscript.

Consent to Publish: Authors are having consent to publish their research work in the IJAMT journal.

Authors' contributions: Panneerselvam Thangavel contributed to the study conception, design, and manuscript preparation; Raghuraman Srinivasan commented on previous versions of the manuscript and approved the final manuscript and supervised; Manoj Vasu contributed for data collection, coding and iterating; and Monica Panneerselvam contributed ANFIS data analysis.

Funding: Not applicable

Competing Interests: There are no conflicts of interest among the Authors in this work.

Availability of data and materials: The required data and materials are addressed and discussed appropriately in the manuscript.

Arul S, Vijayaraghavan L, Malhotra SK, Krishnamurthy R (2006) Influence of tool material on dynamics of drilling of GFRP composites. International Journal of Advanced Manufacturing Technology 29:655–662. https://doi.org/10.1007/s00170-005-2581-5
Babajanzade Roshan S, Behboodi Jooibari M, Teimouri R, et al (2013) Optimization of friction stir welding process of AA7075 aluminum alloy to achieve desirable mechanical properties using ANFIS models and simulated annealing algorithm. International Journal of Advanced Manufacturing Technology 69:1803–1818. https://doi.org/10.1007/s00170-013-5131-6
Balaji NS, Jayabal S, Kalyana Sundaram S (2016) A Neural Network Based Prediction Modeling for Machinability Characteristics of Zea Fiber-Polyester Composites. Transactions of the Indian Institute of Metals 69:881–889. https://doi.org/10.1007/s12666-015-0571-3
Kumar D, Sing KK (2017) Experimental analysis of Delamination, Thrust Force and Surface roughness on Drilling of Glass Fibre Reinforced Polymer Composites Material Using Different Drills. Materials Today: Proceedings 4:7618–7627. https://doi.org/10.1016/j.matpr.2017.07.095
Kumar R, Hynes NRJ (2019) Prediction and optimization of surface roughness in thermal drilling using integrated ANFIS and GA approach. Engineering Science and Technology, an International Journal 23:30–41. https://doi.org/10.1016/j.jestch.2019.04.011
Lim WCE, Kanagaraj G, Ponnambalam SG (2014) PCB drill path optimization by combinatorial cuckoo search algorithm. The Scientific World Journal 2014:264518. https://doi.org/10.1155/2014/264518
Mohan NS, Ramachandra A, Kulkarni SM (2005) Influence of process parameters on cutting force and torque during drilling of glass-fiber polyester reinforced composites. Composite Structures 71:407–413. https://doi.org/10.1016/j.compstruct.2005.09.039
Panneerselvam T and Raghuraman S (2012) Experimental Investigations on Delamination to Improve the Hole Quality in Chopped Strand Mat GFRP Material During Drilling Operation. Journal of Applied Sciences 12:1063–1066. https://doi.org/ 10.3923/jas.2012.1063.1066
Panneerselvam T, Raghuraman S, Vidyasundar A (2014) A study to minimise delamination value during drilling chopped strand mat GFRP material. International Journal of Machining and Machinability of Materials 15:136–146 (2014). https://doi.org/10.1504/IJMMM.2014.060545
Panneerselvam T and Raghuraman S (2015) Optimization of Drilling Parameters for Delamination Associated with Pre-drill in Chopped Strand Mat Glass Fibre Reinforced Polymeric Material. Pertanika J Sci & Technol 23:61–72. http://www.pertanika.upm.edu.my/
Prakash S, Palanikumar K, Manoharan N (2009) Optimization of delamination factor in drilling medium-density fiberboards (MDF) using desirability-based approach. International Journal of Advanced Manufacturing Technology 45:370–381. https://doi.org/10.1007/s00170-009-1974-2
Saravanan R, Asokan P, Vijayakumar K (2003) Machining parameters optimisation for turning cylindrical stock into a continuous finished profile using genetic algorithm (GA) and simulated annealing (SA). International Journal of Advanced Manufacturing Technology 21:1–9. https://doi.org/10.1007/s001700300000
Savkovic B, Kovac P, Dudic B, et al (2019) Application of an adaptive “neuro-fuzzy” inference system in modeling cutting temperature during hard turning. Applied Sciences (Switzerland) 9:3739. https://doi.org/10.3390/app9183739
Shunmugesh K, Akhil KT, Aravind S, Pramodkumar M (2017) Optimization of Drilling Characteristics using Grey-Fuzzy logic in Glass Fiber Reinforced Polymer (GFRP). Materials Today: Proceedings 4:8938–8947. https://doi.org/10.1016/j.matpr.2017.07.245
Tan CL, Azmi AI (2017) Analytical study of critical thrust force for on-set delamination damage of drilling hybrid carbon/glass composite. International Journal of Advanced Manufacturing Technology 92:929–941. https://doi.org/10.1007/s00170-017-0152-1
Teimouri R, Sohrabpoor H (2013) Application of adaptive neuro-fuzzy inference system and cuckoo optimization algorithm for analyzing electro chemical machining process. Frontiers of Mechanical Engineering 8:429–442. https://doi.org/10.1007/s11465-013-0277-3
Vinayagamoorthy R, Subramanyam KG, Kumar TN, Reddy YH (2016) Modeling and analysis of drilling induced damages on hybrid composites. Indian Journal of Science and Technology 9(16):87981. https://doi.org/10.17485/ijst/2016/v9i16/87981
Vinayagamoorthy R (2017) Parametric optimization studies on drilling of sandwich composites using the Box–Behnken design. Materials and Manufacturing Processes 32:645–653. https://doi.org/10.1080/10426914.2016.1232811
Vinayagamoorthy R (2018) A review on the machining of fiber-reinforced polymeric laminates. Journal of Reinforced Plastics and Composites 37:49–59. https://doi.org/10.1177/0731684417731530
Wang G dong, Li N, Xiong X hai, et al (2017) 3D level comprehensive evaluation of hole quality in drilling carbon fiber-reinforced plastics. International Journal of Advanced Manufacturing Technology 93:2433–2445. https://doi.org/10.1007/s00170-017-0614-5

VWang G dong, Li N, Xiong X hai, et al (2017) 3D level comprehensive evaluation of hole quality in drilling carbon fiber-reinforced plastics. International Journal of Advanced Manufacturing Technology 93:2433–2445. https://doi.org/10.1007/s00170-017-0614-5
Teimouri R, Sohrabpoor H (2013) Application of adaptive neuro-fuzzy inference system and cuckoo optimization algorithm for analyzing electro chemical machining process. Frontiers of Mechanical Engineering 8:429–442. https://doi.org/10.1007/s11465-013-0277-3

Download PDF

Editorial decision: Minor Revision
27 Mar, 2023
Reviewers agreed at journal
13 Jun, 2022
Reviewers invited by journal
11 Jun, 2022
Editor assigned by journal
30 Jun, 2021
First submitted to journal
29 Jun, 2021

You are reading this latest preprint version

An advance of optimization by Simulated Annealing Algorithm and ANFIS to minimize delamination during drilling Chopped Strand Mat GFRP

Status:

Version 1

Abstract

Figures

1 Introduction

2 Materials And Methods

2.1 Details for optimization by Simulated Annealing Algorithm

2.2 Regression equations developed for setting objective functions

2.3 Methodology for Simulated Annealing Algorithm

2.3.1 Simulated Annealing Algorithm

2.4 Adaptive Neuro Fuzzy Inference System

3 Results And Discussions

4 Conclusions

Declarations

References

Status:

Version 1