Prediction of Surface Roughness and Feed Force in Milling for Some Materials at High Speeds

Omar Monir Koura, Tamer Hassan Sayed

  Open Access OPEN ACCESS  Peer Reviewed PEER-REVIEWED

Prediction of Surface Roughness and Feed Force in Milling for Some Materials at High Speeds

Omar Monir Koura1,, Tamer Hassan Sayed2

1Mechanical Department, Faculty of Engineering, Modern University for Technology & Information, Egypt

2Design & Prod. Eng. Department, Faculty of Engineering, Ain Shams University, Egypt

Abstract

Machining at relatively high speed perform differently than when traditionally cutting speeds are used. High speed machining affects to a great extent the quality of the manufactured products. The effects differ from material to another. The aim of the present paper is to compare the quality of the machined parts at different cutting speed ranges. The study covers several engineering materials. Neural Network techniques was applied in the prediction of both the resulted surface roughness and the developed feed forces.

At a glance: Figures

Cite this article:

  • Koura, Omar Monir, and Tamer Hassan Sayed. "Prediction of Surface Roughness and Feed Force in Milling for Some Materials at High Speeds." American Journal of Mechanical Engineering 3.1 (2015): 1-6.
  • Koura, O. M. , & Sayed, T. H. (2015). Prediction of Surface Roughness and Feed Force in Milling for Some Materials at High Speeds. American Journal of Mechanical Engineering, 3(1), 1-6.
  • Koura, Omar Monir, and Tamer Hassan Sayed. "Prediction of Surface Roughness and Feed Force in Milling for Some Materials at High Speeds." American Journal of Mechanical Engineering 3, no. 1 (2015): 1-6.

Import into BibTeX Import into EndNote Import into RefMan Import into RefWorks

1. Introduction

Milling is currently the most effective and productive manufacturing method for roughing and semi-finishing large surfaces of metallic parts. High speed milling is sometime necessary to mill materials with special characteristics. Milling performance, accuracy and surface texture are tied up with the operating cutting conditions.

Reference [1], built a model to predict surface roughness of milling surface based on cutting speed, feed and depth of cut of end milling operations. The model is based on Genetic expression programming (GEP) which is a solution method that makes a global function search for the problem, developed as a resultant genetic algorithm (GA) and genetic programming (GP) algorithms. The tests were carried out on Aluminum 6061 T8 using a 10 mm diameter HSS end mill. The range of speed used was 23 to 47 m/min, range of feed 135 to 650 mm/min and range of depth of cut 0.25 to 1.27 mm. The model gave the relation between cutting parameters and surface roughness with accuracy of about 91%.

Reference [2], proposed a method for determination of the best cutting parameters leading to minimum surface roughness in end milling mold surfaces of an ortez part used in biomedical applications by coupling neural network and genetic algorithm. A series of cutting experiments for mold surfaces in one component of ortez part are conducted to obtain surface roughness values. The tests were carried out on Aluminum 7075 T6 using a 10 mm diameter Sandvik end mill. The range of speed used was 100 to 300 m/min, range of feed 0.32 to 0.52 mm/rev, range of axial depth of cut 0.30 to 0.7 mm and range of radial depth of cut 1 to 2 mm. A feed forward neural network model is developed exploiting experimental measurements from the surfaces in the mold cavity. Genetic algorithm coupled with neural network is employed to find optimum cutting parameters leading to minimum surface roughness without any constraint.

Reference [3], developed a mathematical model for determining the optimal machining conditions, so as to obtain a surface with specified properties, taking account of the technological constraints on the following parameters: the residual stress; the roughness and micro-hardness (cold working) of the machined surface; the structural–phase composition of the surface layer (the temperature), the tool life; and the standard machine-tool data. It is found that the surface roughness declines with increase in cutting speed and decrease in the feed and depth. Also, the cutting speed and depth have the greatest influence on the surface micro-hardness. Increasing the cutting speed made the surface properties become more uniform in high-speed end milling.

Reference [4], presented an artificial neural network (ANN) model for predicting the surface roughness performance measure in the machining process. Matlab ANN toolbox was used for the modelling purpose. The tests were carried out on Titanium Alloy (Ti-6A1-4V) using un-coated, TiAIN coated and SNTR tools. The range of speed used was 124 to 167 m/min and range of feed 0.025 to 0.083 mm/tooth. The study concluded that the model for surface roughness in the milling process could be improved by modifying the number of layers and nodes in the hidden layers of the ANN network structure, particularly for predicting the value of the surface roughness performance measure. As a result of the prediction, the recommended combination of cutting conditions to obtain the best surface roughness value is a high speed with a low feed rate and radial rake angle.

Reference [5], developed a statistical model for surface roughness estimation in a high-speed flat end milling process under wet cutting conditions, using machining variables such as spindle speed, feed rate, and depth of cut. First- and second order models were developed using experimental results, and assessed by means of various statistical tests.

Reference [6], proposed a method of determining the optimal cutting conditions in the high-speed milling of titanium alloys. The proposed method based on the thermo-physical data regarding cutting and tool wear. This research identified the optimal cutting speed as a function of the mechanical properties of the machined material and the tool material, the mill diameter, the number of cutting teeth, the cutting depth, and the specified tool life.

2. Experimental Work

Two sets of experiments are planned. One is for the study of the effect of cutting conditions on the surface roughness (Ra) in high speed end milling and the second is for the study of the cutting conditions on the feed force. Four types of materials are used, namely Aluminum alloy, Brass, Phosphorus bronze and Steel. Also tests for checking the rate of wear is given in Table 1. The experimental conditions are given in Table 2. It contains the cutting conditions and the various parameters to be measured in each case.

Table 1. Experimental set up for tool wear

3. Results and Discussion

3.1. Effect of Speed on Tool Wear
Figure 1. Tool wear when cutting phosphorus bronze

Figure 1 shows the rate of tool wear when cutting phosphorus bronze at speed of 1257 m/min. The wear increased sharply during the first 5 seconds then increases steadily at a low rate. When cutting steal with speeds 943 m/min or 1257 m/min, the tool wear remains, almost, the same at 0.01 mm after 1 sec. even with different depth of cuts. It reached around 0.1 mm after 5 sec. The better results for steel may be due to the less friction on the rake face of the cutter with the faster flow of the chip.

3.2. Effect of Speed and Feed Rate on Feed Force

Figure 2 & Figure 3 show the effect of speed on the feed force when cutting phosphorus bronze and steel. No effect was noticed when varying the feed from 145 to 220 mm/min. A minimum feed force was obtained at cutting speed 1100 m/min. The increase in the order of magnitude when cutting steel is mainly due to the higher resistivity to cut and the increase in the depth of cut.

Figure 4 shows the effect of feed on the feed force when cutting aluminum alloy. Feed force increases with increasing the feed rate. At the same time little increase resulted as speed changed from 943m/min to 1257 m/min. But comparing materials such as phosphorus bronze and aluminum alloy with steel a greater increase in the feed force resulted. Further increase resulted when cutting speed decreased from 1257 m/min to 1257 m/min. This later decrease may be to the lesser friction between the material and the cutter.

3.3. Effect of Speed on Roughness

Figure 5 shows the change of roughness with the increase of cutting speed at two different feed rates when cutting Aluminum alloy. Roughness varied between 0.35 to 1.3 µm when cutting at feed rate 365 mm/min. the variation was limited between 0.65 to 1.15 µm when cutting at feed rate 550 mm/min. So, it may be stated that increasing the feed rate improves the resulted surface roughness.

Figure 5. A & B Effect of speed on surface roughness for Aluminum Alloy
Figure 6. Effect of speed on surface roughness for Brass, a=0.5 mm

Figure 6 & Figure 7 show the same parameters but when cutting brass and phosphorus bronze. Same results may be noticed. But cutting steel, Figure 8, showed a reverse conclusion, that for harder material surface roughness increases with increasing both feed and speed.

Figure 7. Effect of speed on surface roughness for Phosphorus bronze, a= 0.5 mm
Figure 8. Effect of feed rate on surface roughness for steel, a=0.5mm

4. Artificial Neural Networks (ANNs)

4.1. ANN1 for Prediction of Surface Roughness

In correlating all the variables, neural networks were utilized. The networks were trained to reach the required goal for modeling and predicting the surface roughness (Ra) by adjusting the values of the connections (weights) between elements. The best ANN structure found was that shown in Figure 9. It consisted of 3 layers. The first layer is the input layer which has 4 neurons for the 4 inputs of the network, the second layer is the hidden layer and it consists of 8 neurons, and the third layer is the output layer with 1 neuron for the predicted surface roughness.

MATLAB 2011b is used to develop the ANN. The processing function for the hidden layer is logsig, and for the output layer is purelin. Feed-forward back propagation ANN used, Leven berg-Marquardt back propagation (TRAINLM) algorithm is used for network training and mean square error (MSE) is used as performance function.

The results show that the correlation coefficient between the measured values and predicted values is 0.94867, as shown in Figure 10.

The correlated equation for the roughness (Ra) is:

(1)

Where:

The values h1, h2, …… h8 are given by:

The values of y1, y2, y3 & y4 are given by:

Where: Aluminum alloy=1, Brass=2, Phosphorus bronze=3, steel=4

The GUI used to create the user interface for ANN1 is shown in Figure 11.

4.2. ANN2 for Prediction of Feed Force

Same number of hidden layer as in network ANN1, but the number of neurons was found to 2 neurons to get higher correlation. The structure is shown in Figure 12.

The correlated equation for the roughness (Ra) is:

(2)

Where: H = w9h1 + w10h2 + b3.

The values h1 & h2 are given by:

The values of y1, y2, y3 & y4 are given by:

Where: Aluminum alloy=1, Phosphorus bronze=2, steel=3

Table 4. ANN2 weights and bias for feed force

The results show that the correlation coefficient between the measured values and predicted values is 0.9894, as shown in Figure 13.

The user interface for the feed force model is shown in Figure 14.

5. Conclusions

1. For materials (aluminum alloy, brass and phosphorus bronze) increasing the cutting speed improves the quality of the surface roughness and v.v. in case of steel.

2. For material such as phosphorus bronze the tool wears much faster than in case of steel.

3. Within the present range there was no change in the feed force as feed rate increased, but it has an optimum when cutting speed was at 1100 m/min.

4. Artificial neural network developed for the prediction of surface roughness has proved efficient. The regression equation showed correlation between the cutting conditions and predicted surface roughness of 94.867%. Equation (1) representing this relation.

5. Artificial neural network developed for the prediction of feed force has good correlation between the cutting conditions and predicted feed force of 98.947%. Equation (2) representing this relation.

References

[1]  Oguz Colak, Cahit Kurbanoglu and M. Cengiz Kayacan, Milling surface roughness prediction using evolutionary programming methods, Materials and Design 28 (2007) 657-666.
In article      CrossRef
 
[2]  Hasan Oktem, Tuncay Erzurumlu and Fehmi Erzincanli, Prediction of minimum surface roughness in end milling mold parts using neural network and genetic algorithm, Materials and Design 27 (2006) 735-744.
In article      CrossRef
 
[3]  I. E. Kiryushin, D. E. Kiryushin and T. G. Nasad, Ensuring Surface Quality of Parts in High-Speed End Milling of Quenched Steel, Russian Engineering Research, 2008, Vol. 28, No. 12, pp. 1248-1251.
In article      
 
[4]  Azlan Mohd Zain, Habibollah Haron and Safian Sharif, Prediction of surface roughness in the end milling machining using Artificial Neural Network, Expert Systems with Applications 37 (2010) 1755-1768.
In article      CrossRef
 
[5]  Babur Ozcelik and Mahmut Bayramoglu, The statistical modeling of surface roughness in high-speed flat end milling, International Journal of Machine Tools & Manufacture 46 (2006) 1395-1402.
In article      CrossRef
 
[6]  D. E. Kiryushin, I. E. Kiryushin and T. G. Nasad, High-Speed End Milling of Titanium Alloys, Russian Engineering Research, 2008, Vol. 28, No. 10, pp. 1022-1025.
In article      
 
  • CiteULikeCiteULike
  • MendeleyMendeley
  • StumbleUponStumbleUpon
  • Add to DeliciousDelicious
  • FacebookFacebook
  • TwitterTwitter
  • LinkedInLinkedIn