TAGUCHI METHOD FOR IMPROVING POWDER COATING PROCESS – A CASE STUDY

Abstract

Powder coating is widely used in industry to give a product protection from the atmosphere, improve the aesthetics and the surface finish. Electrostatic powder coating process (EPCP) using Corona gun is extensively studied in this work with the perspective of determining operating variables which govern the quality of coat. The parts which were studied were found prone to orange peel defect and were having high dry film thickness (DFT). It was revealed that three controllable process variables namely the applied voltage, powder flow rate and blow air flow rate were responsible for quality of the film coated. Further the Taguchi method was used to investigate the effects of process parameters on the response characteristics. The effect of each parameter on both response characteristics namely the orange peel defect and DFT was studied using orthogonal array, S/N ratio and ANOVA. Optimal combination of parameters was found for optimum response characteristics. The optimum combination was verified experimentally and it was confirmed that Taguchi method successfully improved the quality of powder coating.

Full text

TAGUCHI METHOD FOR IMPROVING POWDER COATING PROCESS – A CASE STUDY
M. M. BANUBAKODE
1
, A. C. GANGAL
2
& A. D. SHIRBHATE
3
1
Department of Mechanical Engineering, Brhcetkarav, Vangani, Thane, Maharashtra, India
2
Department of Mechanical Engineering, VPM’S MPCOE, Velaneshwar, Ratnagiri, Maharashtra, India
3
Department of Mechanical Engineering, PRMITR, Amravati, Maharashtra, India
ABSTRACT
Powder coating is widely used in industry to give a product protection from the atmosphere, improve the
aesthetics and the surface finish. Electrostatic powder coating process (EPCP) using Corona gun is extensively studied in
this work with the perspective of determining operating variables which govern the quality of coat. The parts which were
studied were found prone to orange peel defect and were having high dry film thickness (DFT). It was revealed that three
controllable process variables namely the applied voltage, powder flow rate and blow air flow rate were responsible for
quality of the film coated.
Further the Taguchi method was used to investigate the effects of process parameters on the response
characteristics. The effect of each parameter on both response characteristics namely the orange peel defect and DFT was
studied using orthogonal array, S/N ratio and ANOVA. Optimal combination of parameters was found for optimum
response characteristics. The optimum combination was verified experimentally and it was confirmed that Taguchi method
successfully improved the quality of powder coating.
KEYWORDS
: Electrostatic Powder Coating Process (EPCP), Dry Film Thickness (DFT), Orange Peel (OP), Design of
Experiments (DOE), Taguchi Method, Signal to Noise (S/N) Ratio
INTRODUCTION
Powder Coating
Powder coating has today established itself as the most effective means of metal surface finishing replacing many
other commercial finishing techniques like liquid painting and electroplating due to its numerous advantages viz. substantial
improvement to chemical corrosion and high mechanical resistance. In powder coating overspray powder can be recycled and
thus it is possible to achieve nearly 100% use of the coating powder.
It is a faster process giving better productivity as compared to its competitors. Also the process is environment
friendly. For the parts to be used in interiors, the main purpose of powder coating is to enhance the aesthetic of finished
parts. This forces the manufacture to produce defect free parts with uniform coating thickness. Powder coated parts are
prone to various defects like High or low dry film thickness, Craters, Pinholes, Blistering, Poor Adhesion, Orange peel,
Gloss too high, Gloss too low, etc. which are collective output of improper setting of various control parameters.
In present study, a semi automatic system with manual operated corona charging electrostatic spraying gun
(Figure 1) is invested for a coating process and it is referred to as an Electrostatic Powder Coating Process (EPCP).
The Corona charging method makes use of a high voltage generator to bring an electrostatic charge (mostly negative) onto
the powder particles and blow it on to the grounded part with help of booster air.
BEST: International Journal of Management, Information
Technology and Engineering (BEST: IJMITE)
Vol. 1, Issue 3, Dec 2013, 167-180
© BEST Journals

168
M. M. Banubakode, A. C. Gangal & A. D. Shirbhate
Figure 1: Schematic View of Corona Type Powder Coating System
Taguchi Method
Taguchi method has established itself as an important tool for the robust design in obtaining the high quality
process and product which are least sensitive to noise with minimal manufacturing costs. It involves various steps of
planning, conducting and evaluating the results of specially designed tables called ‘orthogonal array’ experiments to study
entire parameter space with minimum number of trials to determine the optimum levels of control parameters [1].
Taguchi method recommends the use of loss function which is transformed into the S/N ratio to measure the
performance characteristic deviating from the desired value and then S/N ratio for each level of process parameters is
evaluated based on average S/N ratio response analysis. Greater S/N ratio is corresponding to better quality characteristic
irrespective of category. To find which process parameters are statistically significant, Analysis of Variance (ANOVA) to
be performed. Finally, a confirmation test is conducted to verify the optimal process parameters obtained from the
parameter design [1, 2]. The Taguchi experimental design and analysis to find the optimal combinations of process
parameters broadly includes following steps.
Scope of the Work
Figure 2: Steps in Taguchi Method

Taguchi Method for Improving Powder Coating Process – A Case Study
169
Experimental Procedure
Experimental Set Up
The bed top made of hot rolled steel sheet and pipe, fabricated in a shape shown in figure 3 is selected for study.
The half part of the bed top where perforated sheet is welded having size 108 cm x 78 cm is the area considered for
experiment. The substrate is first passes through pre-treatment stage prier being powder coated which is then cured in
oven. The defects were measured at the five different locations.
Figure 3: The Component (Bed Top)
The part’s inspection reports prior conducting the experiments showed that High DFT and Orange peel has the
maximum contribution in rejection. By brainstorming with the concerned authorities and experts three control parameters
were shortlisted. These parameters with their suggested range are given in Table 1.
Table 1: Process Parameters and Their Ranges
Sr. Process Parameter Range Units
1 Applied voltage 80 – 100 kV
2 Powder Flow Rate 2 - 4 gm / min
3 Booster air flow rate 1.5 - 2.5 gm / cm
3
To add the solidity to the experiment powders from two different supplier’s viz. Asian and Nerolac were used.
Taguchi Experimental Design
To conduct the experiment it is desirable to have three minimum levels of process parameters to reflect the true
behaviour of output parameters of study. Thus three levels have been assigned to the control parameters and two types of
powders became two levels as shown in Table 2.
Table 2: Process Parameters and Their Levels
Factor Process
Parameter Levels
L
1
L
2
L
3
A Powder Asian Nerolac ---
B Applied voltage 80 90 100
C Powder Flow Rate 2 3 4
D Booster air flow
rate 1.5 2 2.5
As per Taguchi experimental design philosophy considering all factors, the interaction between powder and
applied voltage and overall mean the total degrees of freedom come out to be 10 [2]. Also factor A is of two level, so the
nearest mixed level orthogonal array available satisfying the criterion of selecting the OA is L
18
. For each trial in the L
18

170
M. M. Banubakode, A. C. Gangal & A. D. Shirbhate
array, the levels of the process parameters are indicated in Table 3.
Table 3: L
18
(2
1
X 3
3
) Orthogonal Array with Interaction
Trials A B A x B C D
1 1 1 1 1 1
2 1 1 1 2 2
3 1 1 1 3 3
4 1 2 2 1 1
5 1 2 2 2 2
6 1 2 2 3 3
7 1 3 3 1 2
8 1 3 3 2 3
9 1 3 3 3 1
10 2 1 2 1 3
11 2 1 2 2 1
12 2 1 2 3 2
13 2 2 3 1 2
14 2 2 3 2 3
15 2 2 3 3 1
16 2 3 1 1 3
17 2 3 1 2 1
18 2 3 1 3 2
By putting the values of various levels of process parameters from the Table 2 in the Table 3 we get an actual
experimental L
18
(2
1
x 3
3
) orthogonal array as given in Table 4
Table 4: L
18
(2
1
X 3
3
) Orthogonal Array with Actual Experimental Levels of Process Parameters
Trials A B C D
Powder Voltage Powder Flow Rate Booster Air Flow Rate
1 Asian 80 2 1.5
2 Asian 80 3 2
3 Asian 80 4 2.5
4 Asian 90 2 1.5
5 Asian 90 3 2
6 Asian 90 4 2.5
7 Asian 100 2 2
8 Asian 100 3 2.5
9 Asian 100 4 1.5
10 Nerolac 80 2 2.5
11 Nerolac 80 3 1.5
12 Nerolac 80 4 2
13 Nerolac 90 2 2
14 Nerolac 90 3 2.5
15 Nerolac 90 4 1.5
16 Nerolac 100 2 2.5
17 Nerolac 100 3 1.5
18 Nerolac 100 4 2
Experimental Results
The powder coating experiments were conducted for each set of parameters from trial 1 to trial 18. For each
setting of control parameters ten jobs were coated and readings for dry film thickness (DFT) and orange peel were
measured at five different locations. The experimental results for average DFT are given in Table 5 and total orange peel
are given in Table 6.

Taguchi Method for Improving Powder Coating Process – A Case Study
171
Table 5: Experimental Results for Dry Film Thickness
Trials Dry Film Thickness (Micron) S/N Ratio
R1 R2 R3 R4 R5 R6 R7 R8 R9 R10
1 60.4 62.2 62.7 61 61.7 60.3 60.4 63.2 59.5 62 34.1699
2 63.4 64.2 64.2 63.5 63.9 64.8 63.6 62.9 65.4 63.8 38.94
3 61.8 62.8 62.9 62.9 64.1 66.1 62 65.2 64.2 65.5 32.5731
4 64 62.9 63.3 63.4 63.7 60.5 63.8 63 64.5 63.5 35.368
5 65.7 63.6 64.8 64.5 64.4 63.5 65 64.5 64.4 64.1 39.9924
6 67.8 68.4 68.4 68.2 68.2 67.2 67.1 67.7 67.5 68.2 42.9517
7 66.4 67 66.7 66.4 65.6 66.5 65.8 66.8 66.4 67.1 42.8542
8 67.1 67.1 67 66.8 67.6 66.6 66.7 66.8 67.1 67.1 47.4353
9 66.6 68.1 65.6 65.4 68.7 68.3 66.5 67.6 67.7 67.2 35.5471
10 62.1 63.4 62.6 62.9 63.3 62.6 63 62.8 62.9 62.7 44.568
11 63.7 64.9 64.4 65.2 63.8 64.2 64.2 63.6 64.1 63.9 41.891
12 65.2 65.3 63.7 65 64.2 64.7 64.1 64.8 65.3 64.2 41.1223
13 61 60.4 61.3 60.9 60.7 62.1 62.3 61.1 60.7 60.9 40.0004
14 65.2 65.3 65.5 65.7 65.8 65.1 65.1 65.1 63.6 63.9 39.0828
15 69.1 70.9 68.1 70.7 70.6 70.5 70.8 66.4 67.5 68.6 32.646
16 66.8 66.7 66.4 66.4 67.2 65.5 67.6 66.6 65.8 67.1 40.4635
17 68.7 70.4 68.5 68.4 67.9 68.3 71 68.5 69 68.1 36.6307
18 72 70.6 71 70.2 70.3 70.9 69.8 71.2 69.3 70.7 39.392
Table 6: Experimental Results for Orange Peel
Trials Orange Peel (m
2
) S/N Ratio
R1 R2 R3 R4 R5 R6 R7 R8 R9 R10
1 0.047 0.014 0.017 0.045 0.053 0.050 0.065 0.000 0.105 0.036 25.7554
2 0.022 0.054 0.000 0.027 0.000 0.000 0.080 0.000 0.071 0.016 28.0063
3 0.105 0.000 0.022 0.038 0.000 0.000 0.087 0.000 0.000 0.038 26.5825
4 0.104 0.003 0.009 0.008 0.000 0.137 0.018 0.012 0.031 0.000 25.0629
5 0.021 0.000 0.000 0.000 0.008 0.000 0.000 0.000 0.000 0.000 42.9671
6 0.178 0.220 0.159 0.368 0.143 0.048 0.160 0.111 0.072 0.161 14.7812
7 0.015 0.039 0.036 0.022 0.010 0.029 0.000 0.049 0.000 0.012 31.5416
8 0.207 0.023 0.010 0.016 0.084 0.000 0.000 0.010 0.024 0.021 22.8477
9 0.000 0.149 0.000 0.044 0.037 0.064 0.045 0.037 0.101 0.102 22.7083
10 0.072 0.043 0.160 0.165 0.000 0.073 0.045 0.068 0.056 0.089 20.8148
11 0.107 0.031 0.000 0.000 0.063 0.008 0.000 0.000 0.063 0.000 26.9011
12 0.165 0.072 0.000 0.018 0.032 0.015 0.000 0.014 0.188 0.000 21.5788
13 0.057 0.011 0.041 0.078 0.087 0.102 0.052 0.121 0.218 0.187 18.8956
14 0.000 0.000 0.004 0.007 0.014 0.000 0.000 0.000 0.014 0.000 43.4008
15 0.144 0.316 0.099 0.421 0.151 0.348 0.245 0.018 0.000 0.076 12.8595
16 0.020 0.008 0.011 0.004 0.059 0.000 0.039 0.000 0.000 0.014 32.3665
17 0.249 0.394 0.165 0.178 0.053 0.143 0.446 0.203 0.148 0.083 12.4541
18 0.549 0.446 0.408 0.357 0.339 0.445 0.325 0.500 0.239 0.322 7.89519
ANALYSIS AND DISCUSSIONS OF RESULTS
To study the individual effect of each parameter on dry film thickness and orange peel the average value and S/N
ratio of the response characteristics for each variable at different levels were calculated from experimental data. The main
effects of process variables both for raw data and S/N data were plotted. The response curves (main effects) are used for
examining the parametric effects on the response characteristics. The analysis of variance (ANOVA) of raw data and S/N
data was carried out to identify the significant variables and to quantify their effects on the response characteristics.
The most favourable values i.e. the optimal settings of process variables in terms of mean response characteristics were
established by analysing the response curves and the ANOVA tables.

172
M. M. Banubakode, A. C. Gangal & A. D. Shirbhate
Dry Film Thickness
The experimental data collected and tabulated in Table 5 is used and the average values of dry film thickness for
each parameter at different levels for raw data and S/N data are plotted in Figures 4 and 5 respectively. Figure 4 shows that
dry film thickness increases with increase applied voltage and powder flow rate. This is because increasing voltage
increases the quantity of charged powder particles and powder flow rate increases the quantity of powder coming out of the
coating gun. Both ensure more powder particles sticking to the substrate.
Dry film thickness decreases slightly with Increasing blow air flow rate but it increases again, but its effect is not
much significant. The effect of both the powder looks very much same on dry film thickness. Figure 5 shows that S/N ratio
increases with increase in applied voltage and blow air flow rate. For powder flow rate it increases first and then decreases.
It is clear from the Figures 6 and 7 that there is moderate interaction between powder and applied voltage in affecting the
dry film thickness since the responses at different levels of these process parameters are non parallel.
NEROLACASIAN
68
67
66
65
64
1009080
432
68
67
66
65
64
2.52. 01.5
POW DER
Mean of Means
VOLT AGE
PFR BAFR
Main Effects Plot for Means
Data Means
Figure 4: Effects of Process Parameters on Dry Film Thickness (Raw Data)
NEROLACASIA N
40.8
39.6
38.4
37.2
36.0
1009080
432
40.8
39.6
38.4
37.2
36.0
2.52. 01.5
POWDER
Mean of SN ratios
VOLTAGE
PFR BAFR
Main Effects Plot for SN ratios
Data Means
Signal-to-noise: Nominal is best (10*Log10(Ybar* *2/s**2))
Figure 5: Effects of Process Parameters on Dry Film Thickness (S/N Data)
1009080
69.0
67.5
66.0
64.5
63.0
NEROLACASIAN
69.0
67.5
66.0
64.5
63.0
POWDER
VOLTAGE
ASIA N
NEROLA C
POW DER
80
90
100
VOLTA GE
Interaction Plot for Means
Data Means
Figure 6: Effects of Process Parameters Interactions on Dry Film Thickness (Raw Data)

Taguchi Method for Improving Powder Coating Process – A Case Study
173
1009080
42.5
40.0
37.5
35.0
NEROLACASIAN
42.5
40.0
37.5
35.0
POWDER
VOLTAGE
ASI AN
NEROLA C
POW DER
80
90
100
VOLTA GE
Interaction Plot for SN ratios
Data Means
Signal-to-noise: Nominal is best (10*Log10(Ybar** 2/s**2))
Figure 7: Effects of Process Parameters Interactions on Dry Film Thickness (S/N Data)
Residual plots are used to evaluate the data for the problems like non normality, non random variation, non
constant variance, higher-order relationships, and outliers. It can be seen from Figures 8 and 9 that the residuals follow an
approximately straight line in normal probability plot and approximate symmetric nature of histogram indicates that the
residuals are normally distributed. Residuals possess constant variance as they are scattered randomly around zero in
residuals versus the fitted values. Since residuals exhibit no clear pattern, there is no error due to time or data collection
order. In order to study the significance of the process variables towards dry film thickness, analysis of variance (ANOVA)
was performed. The ANOVA of the S/N data and the raw data for dry film thickness are given in Tables 7 and 8
respectively.
210-1-2
99
90
50
10
1
Residual
Percent
7068666462
2
1
0
-1
-2
Fitted Value
Residual
210-1-2
4.8
3.6
2.4
1.2
0.0
Residual
Frequency
18161412108642
2
1
0
-1
-2
Observation Order
Residual
Normal Probability Plot Versus Fits
Histogram Versus Order
Residual Plots for Means
Figure 8: Residual Plots for Dry Film Thickness (Raw Data)
420-2- 4
99
90
50
10
1
Residual
Percent
454035
4
2
0
-2
Fitted Value
Residual
3210- 1-2-3
4.8
3.6
2.4
1.2
0.0
Residual
Frequency
18161412108642
4
2
0
-2
Observation Order
Residual
Normal Probability Plot Versus Fits
Histogram Versus Order
Residual Plots for SN ratios
Figure 9: Residual Plots for Dry Film Thickness (S/N Data)

174
M. M. Banubakode, A. C. Gangal & A. D. Shirbhate
Table 7: Analysis of Variance for Dry Film Thickness (S/N Data)
Source Df Seq SS Adj SS Adj MS F P
POWDER 1 1.977 1.977 1.977 0.36 0.556
VOLTAGE 2 13.515 13.515 6.758 1.23 0.343
PFR 2 33.698 33.698 16.849 3.06 0.103
BAFR 2 91.739 91.739 45.869 8.32 0.011
POWDER*VOLTAGE 2 99.740 99.740 49.870 9.04 0.009
Residual Error 8 44.117 44.117 5.515
Total 17 284.787
DF - degrees of freedom, SS - sum of squares, MS - mean squares(Variance), F-ratio of
variance of a source to variance of error, P < 0.10 - determines significance of a factor at
95% confidence level
Table 8: Analysis of Variance for Dry Film Thickness (Raw Data)
Source Df Seq SS Adj SS Adj MS F P
POWDER 1 3.547 3.5467 3.5467 1.48 0.258
VOLTAGE 2 57.003 57.0329 28.5164 11.90 0.004
PFR 2 39.391 39.3915 19.6957 8.22 0.011
BAFR 2 0.710 0.7100 0.3550 0.15 0.865
POWDER*VOLTAGE 2 2.567 2.5673 1.2837 0.54 0.605
Residual Error 8 19.166 19.1660 2.3958
Total 17 122.414
DF - degrees of freedom, SS - sum of squares, MS - mean squares(Variance), F-ratio of
variance of a source to variance of error, P < 0.10 - determines significance of a factor at
95% confidence level
From the tables it is clear that voltage, powder flow rate and blow air flow significantly affect both the mean and
the variation in the dry film thickness values which is indicated by the bigger values of sum of squares and p-values.
The response tables given in Tables 9 and 10 respectively for S/N ratios, means show the average of each response
characteristic for each level of each factor. The tables include ranks based on Delta statistics, which compare the relative
magnitude of effects. The Delta statistic is the highest minus the lowest average for each factor. Minitab assigns ranks
based on Delta values; rank 1 to the highest Delta value, rank 2 to the second highest, and so on.
The ranks and the delta values show that blow air flow rate have the greatest effect on dry film thickness and is
followed by powder flow rate, voltage and powder in that order. As dry film thickness is ‘nominal is best’ type quality
characteristic, it can be seen from Figure 4 that the second level of voltage (B
2
) and powder flow rate (C
2
) provides the
nominal value of dry film thickness, while blow air flow rate and powder have all the levels pretty close to the nominal
value. By the S/N data analysis (Figure 5) it is seen that B
3
, C
2
and D
3
has highest S/N ratio while powder seem to be a non
contributing parameter. Thus it suggests (A
1
or A
2
, B
3
, C
2
and D
3
) as the best levels for optimum (nominal) dry film
thickness in powder coating process.
Table 9: Response Table for Dry Film Thickness (S/N Data)
Level POWDER VOLTAGE PFR BAFR
1 38.87 38.88 39.57 36.04
2 39.53 38.34 40.66 40.38
3 40.39 37.37 41.18
Delta 0.66 2.05 3.29 5.14
Rank 4 3 2 1

Taguchi Method for Improving Powder Coating Process – A Case Study
175
Table 10: Response Table for Dry Film Thickness (Raw Data)
Level POWDER VOLTAGE PFR BAFR
1 65.03 63.46 63.61 65.70
2 65.92 65.18 65.59 65.21
3 67.79 67.23 65.51
Delta 0.89 4.33 3.62 0.48
Rank 3 1 2 4
ORANGE PEEL
Using the experimental data collected and tabulated in Table 6 the average values of orange peel for each
parameter at different levels for raw data and S/N data are plotted in Figures 10 and 11 respectively. It is clearly seen that
orange peel increases with increase in voltage and powder flow rate, but it increase slightly and the decreases with increase
in blow air flow rate. Also Nerolac powder seems to be more prone to orange peel compared to Asian powders. Figure 11
shows that initially with increase in voltage and powder flow rate S/N ratio increases, with further increase the S/N ratio
declines. While S/N ratio increases continuously with increase in blow air flow rate. Asian powder has higher S/N ratio
than Nerolac powders. It is clear from the Figures 12 and 13 that there is moderate interaction between powder and applied
voltage in affecting the dry film thickness since the responses at different levels of these process parameters are non
parallel. Residual plots do not show any problem in the distribution of the data and model assumptions (Figures 14 and 15)
NEROLACASIAN
0.150
0.125
0.100
0.075
0.050
1009080
432
0.150
0.125
0.100
0.075
0.050
2.52.01.5
POWDER
Mean of Means
VOLTA GE
PFR BAFR
Main Effects Plot for Means
Data Means
Figure 10: Effects of Process Parameters on Orange Peel (Raw Data)
NEROLACAS IAN
30
25
20
1009080
432
30
25
20
2.52. 01.5
POWDER
Mean of SN ratios
VOLTA GE
PFR BAFR
Main Effects Plot for SN ratios
Data Means
Signal-to-noise: Sm aller is better
Figure 11: Effects of Process Parameters on Orange Peel (S/N Data)
1009080
0.20
0.15
0.10
0.05
NEROLACASIAN
0.20
0.15
0.10
0.05
POWDER
VOLTAGE
ASIAN
NEROLA C
POWD ER
80
90
100
VOLTAG E
Interaction Plot for Means
Data Means
Figure 12: Effects of Process Parameters Interactions on Orange Peel (Raw Data)

176
M. M. Banubakode, A. C. Gangal & A. D. Shirbhate
1009080
28
24
20
NEROLACASIAN
28
24
20
POWDER
VOLTAGE
ASIA N
NEROLA C
POW DER
80
90
100
VOLTAG E
Interaction Plot for SN ratios
Data Means
Signal-to-noise: Sma ller is better
Figure 13: Effects of Process Parameters Interactions on Orange Peel (S/N Data)
0.10.0-0.1
99
90
50
10
1
Residual
Percent
0.30.20.10.0
0.10
0.05
0.00
-0.05
-0.10
Fitted Value
Residual
0.100.050.00-0.05-0.10
4
3
2
1
0
Residual
Frequency
18161412108642
0.10
0.05
0.00
-0.05
-0.10
Observation Order
Residual
Normal Probability Plot Versus Fits
Histogram Versus Order
Residual Plots for Means
Figure 14: Residual Plots for Orange Peel (Raw Data)
In order to study the significance of the process variables towards orange peel, analysis of variance (ANOVA)
was performed. The ANOVA of the S/N data and the raw data for orange peel are given in Tables 11 and 12 respectively.
From these tables, it is observed that powder flow rate and voltage significantly affect while blow air flow rate and powder
moderately affect both the mean and the variation in the orange peel values. The response tables (Tables 13 and 14) show
the average of each response characteristic (S/N data and raw data) for each level of each factor.
20100-10-20
99
90
50
10
1
Residual
Percent
3025201510
10
5
0
-5
-10
Fitted Value
Residual
1050-5- 10
3
2
1
0
Residual
Frequency
18161412108642
10
5
0
-5
-10
Observation Order
Residual
Normal Probability Plot Versus Fits
Histogram Versus Order
Residual Plots for SN r atios
Figure 15: Residual Plots for Orange Peel (S/N Data)
The Tables include ranks based on delta statistics, which compare the relative magnitude of effects. The delta
statistic is the highest minus the lowest average for each factor. Minitab assigns ranks based on delta values; rank 1 to the
highest delta value, rank 2 to the second highest, and so on. The ranks indicate the relative importance of each factor to the
response. The ranks and the delta values for various parameters show that powder flow rate has the greatest effect on
orange peel and is followed by blow air flow rate, powder and voltage in that order. As orange peel is the ‘lower the better’
type quality characteristic, from Figure 10, it can be seen that the first level of powder, i.e. Asian powder (A
1
), first level of
voltage (B
1
), first or second level of powder flow rate (C
1
or C
2
) and third level of blow air flow rate (D
3
) result in
minimum value of orange peel. The S/N ratio analysis (Figure 11) also suggests same levels of the variables except for

Taguchi Method for Improving Powder Coating Process – A Case Study
177
voltage, but considering the ultimate aim of maximizing S/N ratio, the levels of parameter (A
1
, B
2
, C
2
, D
3
) will be the best
levels to minimize orange peel.
Table 11: Analysis of Variance for Orange Peel (S/N Data)
Source DF Seq SS Adj SS Adj MS F P
POWDER 1 103.14 103.14 103.14 1.02 0.342
VOLTAGE 2 69.73 69.73 34.86 0.35 0.718
PFR 2 428.96 428.96 214.48 2.13 0.182
BAFR 2 108.83 108.83 54.42 0.54 0.603
POWDER*VOLTAGE 2 26.06 26.06 13.03 0.13 0.881
Residual Error 8 806.85 806.85 100.86
Total 17 1543.56
DF - degrees of freedom, SS - sum of squares, MS - mean squares(Variance), F-
ratio of variance of a source to variance of error, P < 0.10 - determines significance
of a factor at 95% confidence level
Table 12: Analysis of Variance for Orange Peel (Raw Data)
Source DF Seq SS Adj SS Adj MS F P
POWDER 1 0.022447 0.022447 0.022447 3.01 0.121
VOLTAGE 2 0.019178 0.019178 0.009589 1.29 0.328
PFR 2 0.037216 0.037216 0.018608 2.50 0.144
BAFR 2 0.006662 0.006662 0.003331 0.45 0.655
POWDER*VOLTAGE 2 0.020263 0.020263 0.010132 1.36 0.310
Residual Error 8 0.059622 0.059622 0.007453
Total 17 0.165389
DF - degrees of freedom, SS - sum of squares, MS - mean squares(Variance), F-ratio
of variance of a source to variance of error, P < 0.10 - determines significance of a
factor at 95% confidence level
Table 13: Response Table for Orange Peel (S/N Data)
Level POWDER VOLTAGE PFR BAFR
1 26.69 24.94 25.74 20.96
2 21.91 26.33 29.43 25.15
3 21.64 17.73 26.80
Delta 4.79 4.69 11.70 5.84
Rank 3 4 1 2
Table 14: Response Table for Orange Peel (Raw Data)
Level POWDER VOLTAGE PFR BAFR
1 0.04610 0.04232 0.04744 0.09142
2 0.11673 0.07971 0.05112 0.09832
3 0.12222 0.1568 0.05450
Delta 0.07063 0.07990 0.09824 0.04382
Rank 3 2 1 4
VALIDATION OF EXPERIMENT
In order to validate the results obtained, ten confirmation experiments were conducted for each of the response
characteristics i.e. dry film thickness and orange peel at optimal levels of the process parameters. The average values of the
characteristics were obtained and compared with the predicted values. Minitab software was used to predict the values of
S/N ratios and mean. The results are given in Table 15. It is to be pointed out that these optimal values are within the
specified range of process variables.

178
M. M. Banubakode, A. C. Gangal & A. D. Shirbhate
Table 15: Predicted Optimal Values and Results of Confirmation Experiments
Performance Measures Optimum Set of Parameters Predicted Optimal Value Actual Value*
Dry Film Thickness A
1
, B
3
, C
2
, D
3
(Asian, 100, 3,
2.5) 67.03 micron 68.48 micron
Orange Peel A
1
, B
2
, C
2
, D
3
(Asian, 90, 3,
2.5) 0.0085 m
2
0.02 m
2
*Average reading of ten conformation Experiments
CONCLUSIONS
Based on the experimental results, S/N and ANOVA analysis performed, the following conclusions have been
arrived to obtain optimal process parameters to achieve better surface finish characteristics for powder coating process
improvement:
• Based on the illustrations, it is clear that factorial design experiments and Taguchi methods can effectively be
used for control of variables that affect the quality of powder coated parts.
• The effects of the process parameters viz. applied voltage, powder flow rate, blow air flow rate and powder, on
response characteristics viz. dry film thickness and orange peel, were studied.
• All the process parameters considered for analysis has significant impact on the response characteristics.
• The optimal sets of process parameters were obtained for various performance measures using Taguchi’s design
of experiment methodology. The optimal values of process parameters brought the dry film thickness within the
desired range and considerably reduced the orange peel.
• It is noticed that, there was a good agreement between estimated and actual values obtained in respect of dry film
thickness and orange peel within the preferred significant level.
REFERENCES
1. Ranjit K. Roy, ‘A primer on Taguchi method’, Van Nostrand Reinhold, New York (1990).
2. Madhav S. Phadke, ‘Quality engineering using robust design’, Prentice-Hall, N.J., USA (1989).
3. Patricia C. Irwin, ‘Powder Coating Process – Improvements using Statistical and DOE Methodologies’, Electrical
Insulation Conference and Electrical Manufacturing & Coil Winding Conference, Proceedings, IEEE Conference
Publications, (1999) , 577 - 581
4. Pongchanun Luangpaiboon, ‘Improving an Electrostatic Powder Coating Process via Signal to Noise Response
Surface’ American Journal of Applied Sciences (2010) 7 (11): 1521-1527
5. Mohammed Yunus, Dr.J.Fazlur Rahman, ‘Optimization of process parameters of wear and hardness
characterization of industrial ceramic coatings using Taguchi design approach’, International Journal of Advanced
Engineering Science and Technologies Vol. No. 9, Issue No. 2, 193 – 198
6. Resit Unal, Edwin B. Dean, ‘Taguchi Approach to Design Optimization for Quality and Cost: An Overview’,
Presented at the 1991 Annual Conference of the International Society of Parametric Analysts.
7. Sunil A. Misal, R.W. Gaikwad, Dr. Dhirendra and Dr. Vivek S. Sathe, ‘Optimization of Control Parameters Using
Improved Relay Tuning and Taguchi Method’, International Journal of ChemTech Research CODEN (USA):
IJCRGG ISSN : 0974-4290 Vol. 3, No.3, pp 1081-1087, July-Sept 2011

Taguchi Method for Improving Powder Coating Process – A Case Study
179
8. Chao-Lieh Yang, ‘Optimizing the Glass Fiber Cutting Process Using the Taguchi Methods and Grey Relational
Analysis’, New Journal of Glass and Ceramics, 2011, 1, 13-19 doi:10.4236/njgc.2011.11003 Published Online
April 2011
9. M. Aghakhani, E. Mehrdad, and E. Hayati, ‘Parametric Optimization of Gas Metal Arc Welding Process by
Taguchi Method on Weld Dilution’, International Journal of Modeling and Optimization, Vol. 1, No. 3, August
2011