Simulation of the drilling process in GFRP composites using system dynamics and validation by ANN and RSM

Chia sẻ: Nguyễn Thảo | Ngày: | Loại File: PDF | Số trang:9

Thêm vào BST

Báo xấu

12
lượt xem 2
download

Download Vui lòng tải xuống để xem tài liệu đầy đủ

This paper intends to present the System Dynamics (SD) as a novel method to simulate the thrust force developed during drilling of GFRP composites. Good quality holes are extremely fundamental so as to accomplish equally good joints amid creation of components prepared from composite for better execution.

Chủ đề:

Bình luận(0) Đăng nhập để gửi bình luận!

Lưu

Nội dung Text: Simulation of the drilling process in GFRP composites using system dynamics and validation by ANN and RSM

International Journal of Mechanical Engineering and Technology (IJMET) Volume 10, Issue 03, March 2019, pp. 585-593. Article ID: IJMET_10_03_060 Available online at http://www.iaeme.com/ijmet/issues.asp?JType=IJMET&VType=10&IType=3 ISSN Print: 0976-6340 and ISSN Online: 0976-6359 © IAEME Publication Scopus Indexed SIMULATION OF THE DRILLING PROCESS IN GFRP COMPOSITES USING SYSTEM DYNAMICS AND VALIDATION BY ANN AND RSM Murthy B. R. N and Vijay G. S* Department of Mechanical and Manufacturing Engineering, Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal, Karnataka, India Corresponding author * ABSTRACT This paper intends to present the System Dynamics (SD) as a novel method to simulate the thrust force developed during drilling of GFRP composites. Good quality holes are extremely fundamental so as to accomplish equally good joints amid creation of components prepared from composite for better execution. Since the nature of a drilled hole is subject to material properties and machining conditions, it is important to think about the impacts of these factors on the nature of hole obtained. In the present work, the machining parameters thickness of the material, drill point angle, drill size, drill speed and feed rate are selected to evaluate their effect on the quality of the hole. Past works uncover the fact that the damage caused to the drilled hole is primarily due to the thrust force. Consequently it is fundamental to limit the thrust force so as to accomplish better quality of the drilled hole. The SD simulation model was implemented through a causal loop diagram. A mathematical equation used in the simulation was developed utilizing the Design of Experiments (DOE) technique. VENSIM programming was utilized to create and run the SD model. The SD simulation results were compared with Artificial Neural Networks (ANN) results, Response Surface Methodoly (RSM) results and the experimental results. A decent agreement was seen between SD, ANN and RSM results. Key words: System dynamics, GFRP drilling, Thrust force, Artificial Neural Network. Cite this Article Murthy B. R. N and Vijay G. S, Simulation of the Drilling Process in Gfrp Composites Using System Dynamics And Validation By Ann And Rsm, International Journal of Mechanical Engineering and Technology, 10(3), 2019, pp. 585-593. http://www.iaeme.com/IJMET/issues.asp?JType=IJMET&VType=10&IType=3 http://www.iaeme.com/IJMET/index.asp 585 editor@iaeme.com
Simulation of the Drilling Process in Gfrp Composites Using System Dynamics and Validation by Ann and Rsm 1. INTRODUCTION In the fields like aerospace, transport, biomedical, sport goods, etc, we can find the wide usage of various types of fiber reinforced plastics due to their excellent structural and functional properties, easy of manufacturing, durability and low cost [1]. The drilling process in composites involves various parameters such as material properties, drill material, drill geometry and machining conditions [2]. The execution of these items is chiefly subject to surface quality and dimensional precision of the hole produced. As distinguished by numerous scientists, the nature of the hole drilled is primarily impacted by the thrust force produced during machining. This thrust force is mainly affected by variables, such as drill geometry, drilling speed, feed rate, and so on [3]. Subsequently, numerous researchers have attempted to limit the thrust force, through different methodologies, for example, changing the drill geometry, optimizing parameters, developing simulations, etc. Scientists have utilized various distinctive methodologies, while recreating boring, so as to most likely portray precisely the intricacy just as to compute push powers, torques, temperatures, instrument wear and so on. Three primary bearings have been embraced throughout the years. 1. The systematic numerical methodology, where the drilling device is scientifically described through complicated equations in 3D space and utilized for thorough geometrical calculations of the drilling procedure. In most of the research endeavours, 2D anticipated geometry is utilized rather, so as to minimise the necessary calculations. [3-6]. 2. The experimental one, in which extensive amount of experiments were done and the outcomes are put away in databases so that various parameters can be utilised for experimentally derived equations [7-11]. 3. The numerical approach, where tools like finite element analysis is used, based on the Lagrangian and Eulerian methods [12-18]. By the literature survey, it is evident that huge contribution has already been given by many researchers for the advancement of different simulation techniques. In any case, in this exploration paper we are utilizing System Dynamics (SD) as a simulation tool to build up a simulation for the thrust force generated while drilling the GFRP composite material. 2. EXPERIMENTAL DETAILS: 2.1 Test specimen: For the present research work, GFRP composite sample specimen was prepared by hand layup process. As a reinforcement material E glass chopped strands of density 2590 kg/m3 and modulus of elasticity of 72.5 GPa is utilised. General purpose resin [GP] is the matrix material and the fiber reinforced volume fraction is 44%. Methyl ethyl ketone peroxide is the harder used. 2.2 Drilling process: To conduct the drilling experiments, 3-hub TRIAC CNC vertical machining centre is used. Kistler dynamometer and the charge amplifier were used to measure the thrust force produced while machining. Once the essential data is collected in the dynoware format, the graphs were converted to word format in order to obtain the mean values of the force generated. The complete experimental set is presented in figure 1. http://www.iaeme.com/IJMET/index.asp 586 editor@iaeme.com
Murthy B. R. N and Vijay G. S Figure 1: Experimental set up Various machining parameters are considered for this study. The details of such selected factors and the levels of them are presented in table 1. Taguchi full factorial design is used to decide the number of experiments. Table 1: Factors and levels. Factor Level 1 Level 2 Level 3 A: Tool speed (rpm). 900 1200 1500 B: Feed (mm/min). 75 110 150 C: Drill diameter (mm). 6 8 10 D: Drill point angle (deg). 90 103 118 E: Thickness of the material (mm). 8 10 12 3. RESULTS AND DISCUSSIONS: 3.1 ANOVA: ANOVA results of the thrust force is presented in Table 2. As per the table the factors which are having very significant influence on the thrust force are drill point angle and the spindle speed. So it is evident that changes in parameters have significant effect on the cutting force. The table also indicates the effects of combined design parameters. It is evident from the table that, combinations of Speed* Drill point angle and Speed*Diameter have significant effect of on the generated cutting force. Table 2: ANOVA table for thrust force. http://www.iaeme.com/IJMET/index.asp 587 editor@iaeme.com
Simulation of the Drilling Process in Gfrp Composites Using System Dynamics and Validation by Ann and Rsm 3.2. Optimization of the thrust force using Taguchi method: Table 3 represents the response table for data means of the thrust force generated. According the table, on the basis of smaller is the better concept, the most significant parameter which is affecting the thrust force is the drill angle, and the next significant parameter affecting is the spindle speed. Hence to obtain the minimum thrust force it is necessary to keep these two parameters at lower values. Table 3: Response table for data means From the response table for means, the optimum combination of parameters to achieve low thrust force is: drill angle = 90o, drill diameter = 6 mm, material thickness = 8 mm, drill speed = 1500 rpm and feed rate = 75 mm/min 3.3 Development of system dynamics simulation model: The VENSIM model to simulate the thrust force is developed by following the steps listed below. 1. Drill angle, material thickness, drill speed, drill diameter and feed rate are chosen as input parameters. 2. Thrust force is selected as the response variable. 3. Each input variable is connected to the output variable in the VENSIM model as illustrated in the figure 2. 4. To obtain the change in the response variable with respect to changes in the input variables, a mathematical model (developed using the MINITAB software) which co-relates all the input parameters with the response, is fed in to the output variable box. 5. Once the development of the model is completed, it ready to show the changes in the response value with respect to changes in the input variables. Figure 2: VENSIM model for the thrust force http://www.iaeme.com/IJMET/index.asp 588 editor@iaeme.com
Murthy B. R. N and Vijay G. S 3.4 Simulation of thrust force by system dynamics The thrust force developed for the several combinations of process parameters were found out by changing the input parameter by moving the sliding bar of the individual variable and the results were charted. Lastly with the aid of tabulated results, simulation graphs were developed using MATLAB software. Figures 3-5 shows the simulation graphs of thrust force for various combinations of input parameters. Figure 3: Simulation graph of thrust force for various spindle speeds and drill angles. Figure 4: Simulation graph of thrust force for various spindle speeds and drill diameters Figure 5: Simulation graph of thrust force for various feed rates material thickness. 3.5. Validation of system dynamic as simulation tool In order to validate the thrust force simulated by Systems Dynamics, the simulation results were validated by comparing them with those simulated by two well-known simulation tools, viz., Artificial Neural Network (ANN) and Response Surface Methodology (RSM). http://www.iaeme.com/IJMET/index.asp 589 editor@iaeme.com
Simulation of the Drilling Process in Gfrp Composites Using System Dynamics and Validation by Ann and Rsm 3.5.1. Validation using ANN A single hidden layer perceptron neural network (SHLPNN) was modelled to simulate the drilling process. The input layer consisted of 5 nodes corresponding to the drilling parameters viz., drill angle, drill diameter, material thickness, spindle speed and feed rate. The output layer consisted of one node corresponding to the drilling response, the thrust force. Since for most non-linear applications, one hidden layer is sufficient to successfully model the problem under consideration, only one hidden layer was used in this paper. However, the number of nodes of the hidden layer (h), learning rate () and the momentum rate () were optimized by hybridizing the SHLPNN and Genetic Algorithm (GA). A customised MATLAB code was developed for achieving the same. The fitness function of the GA depended upon the classification accuracy of the SHLPNN on the training data and the test data. The other parameters used in the network were wij: Connection weights between input and hidden layers, wjk: Connection weights between hidden and output layers. The structure of the GA- SHLPNN used in this work is shown in figure 6. In this work, the SHLPNN was set to run for a maximum of 5000 epochs. Out of the 243 input-output pairs, 170 pairs were randomly chosen for training, and the remaining 73 pairs were used for testing. The training goal considered here was the Mean Square Error (MSE) which was set to 110-4. In the GA, a real valued chromosome of length three units corresponding to the three training parameters (h,  and µ) was considered. A population size of ten was used. The GA was set to run for a maximum of 100 generations. The crossover and mutation rates were selected to be 0.6 and 0.2 respectively. The multi-point crossover was adopted. The selection of the parent chromosomes for reproduction was done by the Normalized Geometrical Ranking method. The fitness function used in GA has been given in equation 1. Fitness=  ηTrg +ηTst  (1) 2 where, Trg and Tst are the prediction accuracy on the training and the test data respectively. The expressions for computing the prediction accuracy,  is given in equation 2. nc η= ×100 (2) n where, nc is the number of correctly predicted SHLPNN output values and n is the total number of experiments (data) under consideration. http://www.iaeme.com/IJMET/index.asp 590 editor@iaeme.com
Murthy B. R. N and Vijay G. S Figure 6: Structure of the GA-SHLPNN Figure 7 shows the comparison of the thrust force obtained from experimentation, and that predicted by the system dynamics and ANN technique. A good agreement between the three is observed. Figure 7: Comparisons of experimental, ANN and SD simulation results. 3.5.2. Validation using Response Surface Methodology: RSM is a mathematical and statistical data collection method which is useful in modelling and analyzing problems of engineering. The main goal of the response surface analysis is to determine the global optimization of the process parameters involved in the problem. In this technique, the main goal is to optimize the response surface that is influenced by various machining parameters. Since RSM is a well-known and widely used simulation method, the comparison of the SD and RSM results is carried out and results are illustrated in Figure 8. The RSM was carried out using the MINITAB software. A very close match is observed between the two results which validates the application of SD as a simulation tool in the machining process. http://www.iaeme.com/IJMET/index.asp 591 editor@iaeme.com
Simulation of the Drilling Process in Gfrp Composites Using System Dynamics and Validation by Ann and Rsm Figure 8: Comparisons of SD results with RSM simulation results. 4. CONCLUSIONS This paper investigates the possibility of the System Dynamics to be used as a simulation tool in the drilling process of GFRP composite material. The SD results are compared with those obtained from ANN and RSM. A comparison is also made with the experimental results. From the present study it is evident that, 1) For the machining conditions considered in this paper, the thrust force increases with increase in material thickness, point angle, drill diameter and feed rate. However, it deceases with increase in spindle speed. 2) Machining with the obtained optimum combination of parameters yields the least thrust force and better quality holes. 3) With the help of simulation graphs, one can predict the thrust force at any intermediate values of the machining conditions within the considered range. 4) The System dynamics results shows a good agreement with the experimental and ANN results, whereas it closes matches with the RSM results. 5) Hence system dynamics can be effectively used as a simulation tool in drilling of composite materials. REFERENCES [1] Jain S., Yang D. and C.H., Delamination-free drilling of composite laminates Trans. ASME, Journal for Engg and Industry, (116) 475-481 (1994). [2] A.M. Abrao, P.E. Faria, J.C. Campos Rubio, Drilling of fiber reinforced composites, Journal of material processing technology, 186, 2007, 1-7. [3] C.C.Tsao and H. Hocheng, ―Parametric study on thrust force of core drill‖, Journal of material processing and Technology, 192-193, 37-40, 2007. [4] Armarego EJA. The unified-generalized mechanics of cutting approach—a step towards a house of predictive performance models for machining operations. Machining Science & Technology 2000;4(3):319–62. [5] Kang D, Armarego EJA. Computer-aided geometrical analysis of the fluting operation for the twist drill design and production. I. Forward analysis and generated flute profile. Machining Science & Technology 2003;7(2):221–48. http://www.iaeme.com/IJMET/index.asp 592 editor@iaeme.com
Murthy B. R. N and Vijay G. S [6] Kang D, Armarego EJA. Computer-aided geometrical analysis of the fluting operation for the twist drill design and production. II. Backward analysis, wheel profile and simulation studies. Machining Science & Technology 2003; 7(2):2 49–66. [7] Wang J, Zhang QA. A study of high-performance plane rake faced twist drills. Part II: predictive force models. International Journal of Machine Tools and Manufacture 2008;48:12 86–95. [8] Paul A, Kappor SG, Devor RE. Chisel edge and cutting lip shape optimization for improved twist drill point design. International Journal of Machine Tools and Manufacture 2005;45:421–31. [9] Koehler W. Analysis of the high performance drilling process: influence of shape and profile of cutting edge of twist drills. Journal of Manufacturing Science and Engineering 2008; doi:10.1115/1.2951932. [10] Claudin C, Poulachon G, Lambertin M. Correlation between drill geometry and mechanical forces in MQL conditions. Machining Science & Technology 2008;12:133– 44. [11] Abrao AM, Rubio JCC, Faria PE, Davim JP. The effect of cutting tool geometry on thrust force and delamination when drilling glass fibre reinforced plastic composite. Materials and Design 2008;29:508–13. [12] Hamade RF, Seif CY, Ismail F. Extracting cutting forcé coefficients from frilling experiments. International Journal of Machine Tools & Manufacture 2006;46: 387–96. [13] Singh I, Bhatnagar N, Viswanath P. Drilling of uni-directional glass fiber reinforcement plastics: experimental and finite element study. Materials and Design 2008;29:546–53. [14] Strenkowski JS, Hsieh CC, Shih AJ. An analytical finite element technique for predicting thrust forcé and torque in drilling. International Journal of Machine Tools and Manufacture 2004;44:1413–21. [15] Zitoune R, Collombet F. Numerical prediction of the thrust force responsible of delamination during the drilling of the long fibre composite structures. Composites Part A: Applied Science and Manufacturing 2007;38:858–66. [16] Kadirgama K, Abou-El-Hossein KA, Mohammad B, Habeeb AA, Noor MM. Cutting force prediction model by FEA and RSM when machining Hastelloy C-22HS with 90° holder. Journal of Scientific & Industrial Research 2008;67: 421–7. [17] Bagci E, Ozcelik B. Finite element and experimental investigation of temperature changes on a twist drill in sequential dry drilling. International Journal of Advanced Manufacturing Technology 2006;28:680–7. [18] Vijayaraghavan A, Dornfeld D. Automated drill modeling for drilling process simulation. Journal of Computing and Information Science in Engineering 2007;7:276–83. http://www.iaeme.com/IJMET/index.asp 593 editor@iaeme.com