dinesh national conf paper

7/31/2019 Dinesh National Conf Paper

1/14

1

Development of ANN modeling for coefficient of friction of

electro deposited nickel-graphite composites

S.Dinesh1, G.N.K.RameshBapu

2, K.Ramanathan

3, T.LouieFrango

4

1P.G. scholar, A.C.College of Engineering &Tech, Karaikudi-4,[email protected],9003645530.

2Scientist, Central Electrochemical Research Institute, Karaikudi, 630006,

[email protected], 9486563834.3Assistant Professor of Mechanical Engg, A.C.College of Engineering&Tech,

Karaikudi-4.,[email protected],9994607024.4Assistant Professor of automobile Engg, Shanmuganathan Engineering College,

Pudukkottai 622507,[email protected] ,9443612523

Abstract:

Nickel-Graphite composite coatings are produced by electro deposition

using conventional techniques at various cathode current densities, pH and temperature.

Electro deposition was carried out from a conventional Watts bath. Natural graphite

powder of 20-30 m size was used in this study . 33

full factorial designs of experiments

were designed by adopting the Design of Experiments (DOE) approach with three level of

experiment namely Low, Medium and High. The volume percentage of graphite

deposition in composite coated specimens were measured gravimetrically. The coefficient

of friction of coated specimen was measured using scratch tester. An Artificial Neural

Network (ANN) model was developed using 27 practical data obtained to predict the co

efficient of friction of Ni-Graphite metal matrix. Within the range of input variables for

the present case (pH) = 3 to 5; current density (i) = 3 to 5 A/dm2; temperature (T)= 40 to

600C, the prediction capability of Artificial Neural Network(ANN) is very close to the

experimental measurement of friction of Ni-Graphite metal matrix.

Keywords: Coefficient of Friction, Scratch tester, ANN model, Ni-Graphite composite

coatings, MATLAB

1. INTRODUCTION

Particle-reinforced metal matrix composites generally exhibit wide engineering

applications due to their enhanced hardness, frictional resistance, wear and corrosion resistance

compared to pure metal or alloy. Composite electroplating has been identified to be a

technologically feasible and economically superior technique for the preparation of such kind of
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]


2/14

2

composites. Graphite is a natural material, is known to be the good lubricant material. Because

of their outstanding properties, Nickel-Graphite has been used in EMI gasket and shield

applications.

Because of its excellent mechanical and electrical properties Nickel-Graphite is

of great interest for a number of applications (e.g. bearings, engine parts, electronic gaskets,

etc.). Searching the literature, the impression arises that the applications of Graphite are kept a

bit secret. Most of the relevant references are patents (up to 90% depending on the topic) giving

less exact data about the process. Papers published in journals giving detailed information are

rare.

The volume percent incorporation of Graphite powder in the Ni-Graphite

composite coating measured gravimetrically was earlier demonstrated. The amount of Graphite

deposited in the composite metal matrix is mainly affected by the process parameters such as

current density, pH value, temperature of the path solution and concentration of Graphite

dispersed in the electrolyte. Volume fraction of Graphite influences the Coefficient of Friction

of Ni-Graphite composite coatings and hence it is essential to develop a prediction model for

estimating the Coefficient of Friction of Nickel-Graphite composite using the above parameters.

TABLE 1 Graphite incorporation in nickel

2.0EXPERIMENTAL PROCEDURE

2.1 The electrolyte:

The conventional Watts bath of the following composition was used:

Nickel sulphate - 225 g/l; Nickel chloride- 30 g/l; Boric acid- 40 g/l. The electrolyte was purified

S. NoGraphite

g/lit

Weight of

Graphite

(gm)

Weight of

nickel

(gm)

Vol. of

Graphite

(v1)

Vol. of

Nickel

(V2)

Vol. %

Of

graphite

Vol. %

Of nickel

1 5 0.0157 0.2416 0.0074 0.0271 20.62 79.38

2 10 0.0316 0.2348 0.0142 0.0263 35.07 64.93

3 20 0.061 0.2407 0.0273 0.02704 49.76 50.234 30 0.0094 0.2348 0.0042 0.0263 13.79 86.20

5 40 0.0086 0.2466 0.0038 0.0277 12.20 87.79


3/14

3

in the conventional manner for removal of organic and inorganic impurities [5]. The pH value

of the electrolyte was adjusted electrometrically using dilute H2SO4 or NH3. 0.01-g/l sodium

lauryl sulphate was added to the electrolyte as anti-pitting agent before plating. The temperature

of the electrolyte was maintained using a thermostat.

2.2. Plating procedure:

Deposition was carried out on a 500 ml capacity using conventional technique.

Nickel anodes and mild steel cathodes were used. The cathodes of 7.52.5 cm area were

mechanically polished, degreased, bent to 90, suitably masked to expose an effective plating

area of 12.5 cm2, electro cleaned, first cathodically and then anodically, washed rinsed and then

introduced into the plating electrolyte with the area to be plated in the vertical plane closer to the

bottom of the cell facing the anode. A bagged nickel anode bent similarly was placed above the

area to be coated. Graphite powder (20 to 20 m) was added to the electrolyte in the form of

slurry. The solution was stirred using a magnetic stirrer. Stirring was given initially for 30 s to

bring all the Graphite powder into the suspension and then stopped. The deposition was

continued for 40 minutes to allow the particles to settle on the substrate while the deposition

proceeded. The same process was repeated to obtain various thicknesses.

Figure 1 Electro deposition experimental setup

2.3. Nickel-Graphite deposition:

Natural grade graphite powder of 2030 m sizes was used. Prior to the co-

deposition, the graphite particles were ultrasonically dispersed in the bath for 10 min.


4/14

4

Experiments were conducted at a fixed Graphite concentration of 20 g/l, varying the plating

parameters like temperature, pH, and current density. Ranges of coating parameters in the

coating process are as follows:

Current density, I = 35 A/dm2 ; pH value = 35; Temperature = 40 to 600C .For the prediction of Coefficient of Friction of Graphite under a variation of coating

conditions, a training database with regard to different coating parameters needs to be

established. For the above combination of parameters, twenty seven numbers of Nickel-Graphite

composite coatings were obtained and their Coefficient of Friction was measured from scratch

tester.

2.4 Design of experiment:

Process parameter Units

Levels

Level 1 Level 2 Level 3

pH 3 4 5

Current density A/dm2

3 4 5

Temperature 0C 40 50 60

TABLE 2 Process parameters with different levels

S.NO CURRENT DENSITY (A/dm2)

TEMPERATURE (O

C) pH VALUE

1 2 50 4

2 4 50 4

3 6 50 4

4 2 60 4

5 4 60 4

6 6 60 4

7 2 40 4

8 4 40 4


5/14

5

9 6 40 4

10 2 50 3

11 4 50 3

12 6 50 3

13 2 60 3

14 4 60 3

15 6 60 3

16 2 40 3

17 4 40 3

18 6 40 3

19 2 50 5

20 4 50 5

21 6 50 5

22 2 60 5

23 4 60 5

24 6 60 5

25 2 40 5

26 4 40 5

27 6 40 5

TABLE 3 27 Parameters-Nickel Graphite coatings

2.5 Coefficient of Friction:

Nickel- graphite composites have been tested through the Scratch tester with

constant loadcondition at starting load should be 10 Newtons, loading rate should be zero,

stroke length is 10mm, scratch speed should be 0.20 mm/sec and scratch offset is 0.25mm .Then

we have transfer loading condition enter the file name. Finally we have seen the view file and

how much of Coefficient of friction is obtained.


6/14

6

Figure 2 Coefficient of friction experimental setup

2.5.1 Initial Loading Conditions

Figure 3 initial loading conditions for Coefficient of Friction


7/14

7

2.5.2 Graphs shown Coefficient of Friction:

Figure 4 Coefficient of Friction

2.5.3 Scratch Depth:

Figure 5 scratch is depth measured


8/14

8

3. ARTIFICIAL NEURAL NETWORK (ANN)

ANN is a neural system of imitative biology, and the principle of human brain

operation. Using a large amount of data out of which they build knowledge bases, ANN

establishes analytical model to solve the problem in the estimation, prediction, decision making

and diagnosis. Neural network consist of simple processors, which are called neurons, linked by

weighted connections. Each neuron has inputs and generates an output that can be seen as the

reflection of local information that is stored in connections. The output signal of a neuron is fed

to other neurons as input signals via interconnections. Since the capability of a single neuron is

limited, complex functions can be realized by connecting many neurons. It is widely reported

that structure of neural network, representation of data, normalization of inputs outputs and

appropriate selection of activation functions have strong influence on the effectiveness and

performance of the trained neural network . A Neural network consists of at least three layers

i.e., input layer, hidden layer, and output layer, where inputs are applied at the input layer and

outputs are obtained at the output layer and learning is achieved when the associations between a

specified set of input output pairs as established in Figure.1 .Here Feed forward back

propagation (FFBP) algorithm is used for the prediction of Coefficient of Friction of Nickel-

Graphite in the deposit for the given condition. Figure.vi shows the architecture of a standard

supervised training FFBP ANN and Figure vii shows the perception.

Figure 6 ANN Model with two hidden layers Figure7 a typical processing element (Perception).


9/14

9

Figure. 8 Standard supervised training feed forward neural network

4.0 TRAINING THE ARTIFICIAL NEURAL NETWORK

The neural network has to be first trained and then tested to use for application.

The training was done with MATLAB software using a computer. MATLAB is a software

package used for high performance numerical computations and visualization. It provides an

interactive environment with hundreds of built in functions for technical computations, graphic

and animations. MATLAB stands for matrix lab. Built in functions provides excellent tools for

linear algebra computation data analysis, signal processing, optimization and others scientific

computations. In this work ANN module is utilized for predicting the Coefficient of Friction of

Graphite deposition in Nickel-Graphite composite matrix. The features current density, pH and

temperature are the inputs and the Coefficient of Friction of Nickel-Graphite is the output for

training the neural networks. Weights between input layer & hidden layer and weights between

hidden layer & the output layer are generated randomly for the selected topology of the network.

The number of patterns used for the training of Artificial Neural Network using Feed forward

back propagation algorithm is 27. Training of the ANN was performed without any allowable

error. The patterns are selected for training and testing the ANN. These selected patterns were

normalized so that they lie between 0 and 1. Twenty Seven patterns were selected for training

the ANN. The inputs and outputs are normalized by,m a x

X

XX

i

i Where Xi is the value of a

feature and Xmax is the maximum value of the feature. A 3-6-1 Feed forward back propagation

network was trained and the structure of the network is shown in figure-4.


10/14

10

Figure 9 3-6-1 Feed forward back propagation network.

Figure 10 Performance curve of ANN model for 100 epochs

5. VALIDATING THE ARTIFICIAL NEURAL NETWORK

Once the network was trained such that the maximum error for any of the training

data was less than allowable error, the weights and the threshold values were automatically

saved by the program. As the input values from the validation experiments were given to the

ANN program, the program predicts the required output. To validate the results of the Artificial

Neural Network analysis eight data as shown in Table I were used. Once the pH, current

density and temperature are fed into the trained networks, the Coefficient of Friction

Graphite that can be obtained in the Nickel-Graphite composites could be calculated quickly

using ANN model.


11/14


12/14

12

Figure 9 Percentage error of ANN model in the prediction of Friction

Figure 10Closeness of ANN model in the prediction of friction

7. CONCLUSION

A 3-9-1 Feed forward back propagation Artificial Neural Network (ANN) model

was developed for predicting Coefficient of Friction in Nickel-Graphite composite coated metal

matrix using 27 test data .The developed neural net work was validated with eight data. Values

obtained by the above ANN model were compared with the experimental values of the response

variables to decide about the nearness of the predictions with the experimental values.

-10

-8

-6

-4

-2

0

2

46

8

10

1 2 3 4 5 6%o

fError

No.of Experiments

% of Error ANN model

ANN Model

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1 2 3 4 5 6

FrictionforNi-Graphite

Deposit

ion

No. of experiments

Closeness ANN Model prediction

Actual Friction

Measured Friction


13/14

13

Within the range of input variables for the present case (pH = 3 to 5; current

density (I) = 3 to 5 A/dm2; temperature (T) = 40 to 60

0C), the results showed that Artificial

Neural Network comes in nearness of the predictions to the experimental values of Coefficient

of Friction as the average errors in case of ANN is very less i.e. 0.128484% only.

8.REFERENCES

1 Jack lapinski, derekpletcher, frank c. walsh, The electro deposition of Ni-Graphite

composite layers(2011), Surface and coating technology. Vol. 42, pp.70-73.

2 Haijun Zhao, Lei Liu, Wenbin Hu, Bin ShenFriction and wear behaviour of Nigraphite

composites prepared by electroforming,(2007) Materials& design, Vol. 18, pp.1374-

1378.

3 Didier Floner, Florence Geneste Homogeneous coating of graphite felt by nickel electro

deposition to achieve light nickel felts with high surface area (2007), Electrochemistry

communication, Vol. 9, pp.2271-2275.

4 L.Q. Zhoua,b, J.G. Tanga, Y.P. Lia, Y.C. Zhoub Predicting forming limit of

electrodeposited nickel coating based on practical stressstrain relationship(2008),

Journal of materials processing technology , Vol. 206, pp.431-437.

5 Haijun Zhao , Lei Liu, Jianhua Zhu, Yiping Tang, Wenbin Hu Microstructure and

corrosion behavior of electrodeposited nickel prepared from a sulphamate bath(2007)

Materials Letters ,Vol. 61, pp.1605-1608.

6 Y.S. Jeona, J.Y. Byunb, T.S. Oha Electrodeposition and mechanical properties of Ni

carbon nanotube nanocomposite coatings(2008) Journal of Physics and Chemistry of

Solids ,Vol. 69, pp.1391-1394.

7 K. Subramanian , V.M. Periasamy , Malathy Pushpavanam , K. Ramasamy Predictive

modeling of deposition rate in electro-deposition of coppertin using regression and

artificial neural network (2009) Journal of Electro analytical Chemistry, Vol. 636,

pp.30-35.

8 M. Surender, B. Basu , R. Balasubramaniam Wear characterization of electrodeposited

NiWC composite coatings (2004) Tribology International Vol. 37, pp.743-749.


14/14

14

9 D.P.Weston, P.H. Shipway, S.J. Harris, M.K. Cheng Friction and sliding wear

behaviour of electrodeposited cobalt and cobalttungsten alloy coatings for replacement

of electrodeposited chromium (2009) Wear ,Vol. 267, pp.934-943.

10 J. Lamovec , V. Jovi , D. Randjelovi , R. Aleksi , V. Radojevi Analysis of the

composite and film hardness of electrodeposited nickel coatings on different substrates

(2008)Thin Solid Films, Vol. 516, pp.8646-8654.

11 M. Ghorbani, M. Mazaheri, A. Afshar Wear and friction characteristics of

electrodeposited graphitebronze composite coatings (2005) Surface & Coatings

Technology, Vol. 190, pp.32-38.

dinesh national conf paper

Documents