You are on page 1of 22

176

CHAPTER 9

TAGUCHI / ANOVA ANALYSIS FOR OPTIMIZATION ON DIFFERENT

PROPERTIES

9.1 INTRODUCTION TO TAGUCHI METHOD

The Taguchi approach is form of design of experiments (DOE) with


special application principles for the most experiments carried out in the industry
the difference between the DOE and the Taguchi approach is in the method of
application. Taguchi method is important tool for the robust design in obtaining
the process and product conditions which are least sensitive to noise to produce
high quality products with low manufacturing costs (Roy R.K. 1990 and 2001). 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 (Noordin, M. Y., Venkatesh, V. C., Sharif, S., Elting, S.
and Abdullah, A. 2004) . A quality loss function is then designed to evaluate the
deviation between experimental value and desired value. Taguchi method
combine experiment design theory and quality loss function.

Taguchi method recommends the use of loss function which is then


transformed into a 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 and greater S/N ratio is
corresponding to better quality characteristic irrespective of category and quality
is evaluated based on average S/N ratio response analysis and greater S/N ratio is
corresponding to better performance characteristic regardless of category and
quality. To find which process parameters are statistically significant, Analysis of
177

Variance (ANOVA) to be performed (kamaruddin, S., Zahid Khan, A. and


Foong, S.H. 2010). Finally, to verify the optimal process parameters obtained
from the parameter design, confirmation test to be conducted. To find the optimal
combinations the following step by step procedure is followed for the DOE.

Since this method recommends the use of s/n ratio to measure the
quality characteristics deviating from the desired values. There are several S/N
ratios available depending on the type of characteristics: smaller is better,
nominal is best and larger is better.
(I) Smaller the better
N = -10 Log10 [mean of sum of squares of measured data] (9.1)
This is usually the chosen S/N ratio for all undesirable characterstics like
“defects” etc. for which the ideal value is zero. Also, when an ideal value is finite
and is maximum or minimum value is defined, then the difference between
measured data and ideal value is expected to be as small as possible The generic
form of S/N ratio then becomes,
N = - 10 Log 10 [mean of sum of squares of {measured – ideal}] (9.2)
(II) Larger -the -better
N = - Log 10 [mean of sum squares of reciprocal of measured data] (9.3)
This case has been converted to SMALLER –THE-BETTER by taking the
reciprocal of measured data and then taking the S/N ratio as in the smaller the
better case.
(III) Nominal –the –best

(9.4)

This case arises when a specified value is MOST desired, meaning that neither a
smaller nor a larger value is desirable.
178

Fig.9.1. Taguchi’s methodology


9.2 ANOVA METHOD

Analysis of variance (ANOVA) is performed to evaluate the


significate process factor which considered affecting the product quality the
percentage contribution of variance of factors can be calculated by determining
the quantities such as the degree of freedom, f sum of squares, ss variance, v and
the percentage contribution for each factor. Optimization of process parameters
of wear and hardness characterization of industrial ceramic coatings using
Taguchi design approach.
a) Total degree of freedom , f
(9.5)
Where N is the total number of results
For Factor A, f fA = kA -1 (9.6)
Where kA is the number of level of factor A
For error, fe = fT - (fa +fb+fc+fd+fe) (9.7)
b) Sum of squares, S
Total sum of squares, ST

=( (9.8)

For factor A,

= +….+ - (9.9)
179

For Error,

= -( + + + + ) (9.10)

Variance, V

For Factor A, = (9.11)

For Error, = = =0 (9.12)

(d) F- ratio, F

For Factor A, = x 100 (9.13)

9.3 CONFIRMATION TEST

The confirmation test is used to verify the estimated result with the
experiment results. If the optimal combination of parameters and their levels
coincidently match with one of the experiments in the orthogonal array (OA),
then the confirmatory much with one of the experiments in the OA, then the
confirmatory test is not required.

9.4 DESIGN OF EXPERIMENTS

9.4.1 Taguchi Experimental Design for Wear Test

In wear characterization of ceramic coatings, the design of experiment


using the results of weight loss were carried out with four process parameters
(Ferit Ficici, Murat Kapsiz and Mesut Durat, 2011) each at three levels and the
feasible space to assess quality of output and values for the four parameters of
wear study were defined by varying the applied pressure, sliding velocity, sliding
distance and type of coating for a range are tabulated in table1. The fractional
factorial design layout reduces to L27 type and it requires twenty seven trials are
180

shown in table 9.2. The interaction between A and C process parameters were
taken to evaluate the effect of PV (product of applied pressure and sliding
velocity) factor (Fazlur Rahman, J. and Mohammed Yunus, 2009).
Table 9.1. Process parameters and levels of the experimental design
Parameters Labels Level 1 Level 2 Level 3
Applied Pressure (MPa) A 0.1 0.2 0.3
Sliding Distance ( m) B 4 6 8
Sliding velocity (m/sec) C 2.5 7.5 12.5
Type of Coating D PSZ AT Alumina

Table 9.2. L27 Orthogonal Array used in Taguchi method


A A B C AXC D
1 1 1 1 1 1
2 1 1 2 2 2
3 1 1 3 3 3
4 1 2 1 1 2
5 1 2 2 2 3
6 1 2 3 3 1
7 1 3 1 1 3
8 1 3 2 2 1
9 1 3 3 3 2
10 2 1 1 2 2
11 2 1 2 3 3
12 2 1 3 1 1
13 2 2 1 2 3
14 2 2 2 3 1
15 2 2 3 1 2
16 2 3 1 2 1
17 2 3 2 3 2
18 2 3 3 1 3
19 3 1 1 3 3
20 3 1 2 1 1
21 3 1 3 2 2
22 3 2 1 3 1
23 3 2 2 1 2
24 3 2 3 2 3
25 3 3 1 3 2
26 3 3 2 1 3
27 3 3 3 2 1
181

Table 9.3. Experimental layout and results of summary for Wear test
Experiment trials A B C D Weight loss S/N ratio
(Recipes) in mg in dB
1 0.1 4 2.5 PSZ 3.0 9.54
2 0.1 4 7.5 AT 1.8 5.11
3 0.1 4 12.5 A 5.25 14.40
4 0.1 6 2.5 AT 3.8 11.59
5 0.1 6 7.5 A 4.75 13.53
6 0.1 6 12.5 PSZ 8 18.06
7 0.1 8 2.5 A 5.6 14.96
8 0.1 8 7.5 PSZ 8 18.06
9 0.1 8 12.5 AT 8 18.06
10 0.2 4 2.5 AT 3.5 10.88
11 0.2 4 7.5 A 3.6 11.13
12 0.2 4 12.5 PSZ 14 22.9
13 0.2 6 2.5 A 10 20
14 0.2 6 7.5 PSZ 10 20
15 0.2 6 12.5 AT 20 26.02
16 0.2 8 2.5 PSZ 8 18.06
17 0.2 8 7.5 AT 11 20.83
18 0.2 8 12.5 A 25 27.96
19 0.3 4 2.5 A 6.5 16.25
20 0.3 4 7.5 PSZ 12 21.58
21 0.3 4 12.5 AT 16 24.08
22 0.3 6 2.5 PSZ 9 19.08
23 0.3 6 7.5 AT 16 24.08
24 0.3 6 12.5 A 25 27.96
25 0.3 8 2.5 AT 12 21.58
26 0.3 8 7.5 A 16 24.08
27 0.3 8 12.5 PSZ 24 27.6

9.4.2 Taguchi Experimental Design for Hardness Test

In hardness characterization of ceramic coatings, the following design


of experiment has been used.
182

Table 9.4.Control factors and levels of the experimental design for Hardness
Factors Labels Level 1 Level 2 Level 3
Applied pressure (MPa) A 0.1 0.2 0.3
Sliding Distance ( m) B 4 6 8
Sliding velocity (m/sec) C 2.5 7.5 12.5
Partial Stabilized Alumina-
Type of Coating D Alumina (A)
Zirconia (PSZ) Titania (AT)

Table 9.5. Experimental layout and summary of results for Hardness test

S.No. A B C D Rockwell hardness number S/N ratio in dB


1 16 Super-Z 100 100 110 40.83
2 16 ZTA 110 150 109 40.75
3 16 PSZ 120 200 90 39.08
4 16 AT 140 300 115 41.21
5 25 Super-Z 110 200 116 41.29
6 25 ZTA 100 300 118 41.43
7 25 PSZ 140 100 85 38.59
8 25 AT 120 150 120 41.58
9 30 Super-Z 120 300 120 41.58
10 30 ZTA 140 200 112 40.98
11 30 PSZ 100 150 83 38.38
12 30 AT 110 100 106 40.51
13 40 Super-Z 140 150 125 41.94
14 40 ZTA 120 100 115 41.21
15 40 PSZ 110 300 86 38.69
16 40 AT 100 200 120 41.58
183

9.4.3 Taguchi Experimental Design for Thermal Barrier / Drop Test

Experiment design using Taguchi method was carried out firstly using
the results of thermal barrier / Thermal drop and secondly using the data of
Thermal cycling test. The experiments were carried out with three control factors
each at three levels are tabulated in table 9.6.

Table 9.6. Control factors and levels of experimental design of Thermal Barrier
test
Factors Labels Level 1 Level 2 Level 3
Temperature on coating side (0C) A 700 900 1000
Type of Coating B PSZ Alumina-Titania Alumina
Thickness of Coating (µm) C 150 225 300

Table 9.7. Experimental layout and summary of results for Thermal Barrier test
Trials A B C Thermal Barrier/ S/N ratio in dB
( Recipes) drop 0C
1. 700 PSZ 150 190 45.58
2. 700 AT 225 165 44.35
3. 700 A 300 135 42.61
4. 900 PSZ 225 200 46.02
5. 900 AT 300 170 44.61
6. 900 A 150 125 41.94
7. 1000 PSZ 300 215 46.65
8. 1000 AT 150 160 44.08
9. 1000 A 225 130 42.28

The feasible values for the three control factors of above study were
defined by varying the Temperature on coating side, type of coating and
184

thickness of coating in the temperature range of 700 – 1000 0C, PSZ – AT – A,


and 150 – 300 microns respectively. The fractional factorial design reduces to
orthogonal array of L9 type and the design requires nine trials are shown in
table2. The interactions between the control factors were neglected.

9.4.4 Taguchi Experimental Design for Thermal Cycling Resistance Test

In high temperature characterization of ceramic coatings, the


following design of experiment has been used.

Table 9.8. Experimental layout and summary of results for thermal cycling test

Trials Temperature on Type of Thickness of No. of cycles S/N ratio


coating side in 0C coating coating in µm withstood in dB
1. 700 PSZ 150 320 50.10
2. 700 AT 225 315 49.97
3. 700 A 300 260 48.30
4. 900 PSZ 225 370 51.36
5. 900 AT 300 340 50.63
6. 900 A 150 255 48.13
7. 1000 PSZ 300 400 52.04
8. 1000 AT 150 320 50.10
9. 1000 A 225 265 48.47

9.5 RESULTS AND DISCUSSION

9.5.1 Results of Wear Test Optimization of Parameters

There are twenty seven different tests were conducted using the
control factor combinations in the specified orthogonal array table value. Nine
specimens were prepared for each set of parameters to prepare complete response
185

table. Taguchi method uses the S/N (signal-to-noise) ratio. S/N ratio is used to
determine the most significant factor. There are three types of S/N ratio criteria
for optimization; smaller the best, larger the better and nominal the best. To get
the better performance of results, smaller the weight loss is desired and hence
smaller the best criteria has been selected and following expression was used for
analysis.
(9.14)

Where y represents the observed data and n number of tests in one trial. The
mean responses of S/N ratio for weight loss are calculated for all factors and are
tabulated. A sample calculation for one of factor (A) is shown below.
9.54+5.11+14.4+11.59 +13.53+18.06+14.96+18.06+18.06
Level1= ----------------------------------------------------------------------------=13.7 dB
9
10.88+11.13+22.9+20+20+26.02+18.06+20.83+27.96
Level2= ----------------------------------------------------------------------- = 19.75 dB
9
6.25+21.58+24.08+19.08+24.08+27.96+21.58+24.08+27.6
Level3= ---------------------------------------------------------------------------=22.92dB
9
Difference, (max-min) = 22.92-13.7
= 9.22 dB
Table 9.9. Mean S/N Ratio Response Table for Wear test

Symbol Factors Level1 Level2 Level3 Rank


A Pressure -13.7 -19.75 -22.92 9.22 1
B Sliding Distance -15.09 -20.035 -21.24 6.15 3
C Sliding Velocity -15.77 -17.6 -23.004 7.234 2
D Type of coating -19.43 -18.025 -18.92 1.405 5
AXC Interaction effect(PV factor) 20.3011 18.392 17.682 2.62 4
186

Fig.9.2. Average Response S/N ratio Graph for Wear test

Additionally, it is seen that most significant factors can be determined


by the larger difference of S/N ratio from figure2 and table 911 gives the results
of ANOVA, it is found that mostly significant factor with contribution ratio, that
decreases the weight loss is Applied pressure followed by the Sliding velocity,
sliding distance and the type of coating.

Table 9.10. The optimal set of factors for wear test

Symbols Parameters Optimum setting


A. Pressure 0.3 MPa
B. Sliding Distance 8 Km
C. Sliding Velocity 25 m/sec
D. Type of coating PSZ
187

ANOVA Results of Wear Test

ANOVA is used to determine the most significant factor


affecting the optimum combination of process parameters on output quality and
characteristic using the quantities such as degree of freedom (f), sum of squares
(SS), variance (V), percent contribution of each factor (F-ratio) and then
contribution ratio or contribution of variance were determined. From the table
9.5, it is observed that highest contribution ratio is for type of coating (B) while
Temperature on coating side (A) is the lowest and thickness of Coating (C) has
moderate value between the other two. Tables 9.9 illustrate the average S/N ratio
values for weight loss. Tables 9.11 show the results of ANOVA. It is observed
that the Applied Pressure (42.13%) is most significantly influences the material
or weight loss followed by sliding velocity (27.16%), sliding distance (20.36%)
and type of coating (0.9714%) and (3.52%) reveals that the interaction effect of
the process parameters is also acceptable.

Table 9.11. Analysis of Variance (ANOVA) results for Wear test


Factors SS Dof variance F-ratio % Contribution ratio

A 395.018 2 197.509 50.35 42.13


B 190.897 2 95.4493 24.331 20.36
C 254.622 2 127.3109 32.452 27.16
D 9.10801 2 4.554 1.161 0.9714
AXC 33.02 4 8.255 2.104 03.52
Pooled error 54.968 14 3.923 1 05.862
Total 937.634 26 100

Confirmation Test

Furthermore, the confirmation test is conducted to verify the


improvement of results and to predict the optimum performance at the selected
188

levels (since all factors have a confident level more than 90%) of significant
parameters such as A3, C3, B3 and D1. The most optimal set of combination of
parameter is found out. The predicted mean (M) of the response characteristic of
TBC can be expressed as,
M = (A3-T) + (B3-T) + (C3-T) + (D1-T) + T, (9.15)
where T= current grand average of S/N ratio

Table 9.12. The comparison between actual and predicted results of Wear test
Optimum Level
Estimation Experimental Difference
Level A3B3C3D1 A3B3C3D1
Wear in mg 25.7 24 1.7
S/N Ratio in dB 30.22 27.6 0.6

= (22.92-18.792) + (21.24-18.792) + (23.004-18.792) + (19.43-18.792) +


18.792 = 30.22dB (9.16)
A confidence interval (C.I) has to be evaluated on the for the M using
the following expression [9-10]

=4.9 (9.17)

Where Nf= and R= number of trials to

run confirmation test. Using ANOVA table ve= 3.923, fe=14 and F (1, 14) =4.6
at 95% confidence level and tabulated.

At the 95% confidence level, the predicted mean of the weight


loss (WL) was found to be in the range of 25.32 dB < WL > 35.12 dB.
189

9.5.2 Results of Hardness Test Optimization of Parameters

For the above test, Twenty seven different hardness tests were
conducted using the process parameters combinations in the specified orthogonal
array L16 as shown in table 9.8. To get the better performance, larger the hardness
number is desired and in this case, the larger the best criteria has been selected
for S/N ratio in the analysis.

Table 9.13. Mean response S/N ratio of the hardness test


Levels Rank
Factor Label
I II III IV
Torch power (KW) A 40.47 40.72 40.37 40.86 0.49 3
Type of Coating B 41.41 41.09 38.69 41.22 2.72 1
Standoff distance (mm) C 40.56 40.31 40.86 40.68 0.55 2
Coating thickness 40.73 0.44 4
D 40.29 40.66 40.73
( micron)

Table 9.14. ANOVA results of hardness test


F-ratio Contribution
Factors Label SS DOF Variance
ratio (%)
Torch power 6.56 2.84
A 3 0.20531
(KW) 0.61592
Type of Coating B 19.8011 3 6.6 211 91.23
SOD (mm) C 0.64567 3 0.21522 6.88 2.975
Coating 5.83 2.524
D 3 0.18257
thickness in µm 0.54772
Pooled error 0.094 3 0.0313
Total 21.704 15
190

Confirmation Test

The predicted mean (M) of the optimal set of results and confidence
interval (C.I.) were obtained by using the following equations,
= (40.855-40.602)+(41.41-40.602) + (40.86-40.602) + (40.733-40.602)
+40.602=42.052dB (9.18)
C.I. = [F (1, fe) Ve {1/neff +1/R}]^1/2=0.76 (9.19)
The predicted mean of the response S/N ratio for the hardness (HRC)
lies in the range of 41.294dB < HRC > 42.81dB at the confidence level of 95%
( =0.05).

Table 9.15. Comparison of results between actual and estimated performance of


hardness test
Optimum Level
Estimation Experimental Difference
Level A4B1C3 C4 A4B1C3 C4
Rockwell hardness number 116 122 06
S/N Ratio in dB 42.052 41.72 2.15

Most significant factor is Super-Z alloy (B1) followed by stndoff


distance 120mm, Power of 40KW and least significant factor is when the
thickness of coating is 200micron.

From ANOVA (table 9.14) results, it is inferred that maximum


contribution ratio is for coating (91.23%) followed by standoff distance
(2.975%), Power (2.84%) and least significant factor is the thickness of coating
(2.524%).
191

Figure 9.3. Average Response S/N ratio Graph for Hardness test

9.5.3. Results of Thermal Barrier Test Optimization of Parameters

There are nine different tests were conducted using the usage
parameter combinations in the specified orthogonal array table to measure the
quality and characteristics of output from desired value. Three specimens were
prepared for each set of parameters to prepare complete response table. Taguchi
method uses the S/N (signal-to-noise) ratio. S/N ratio is used to determine the
most significant parameter. There are three types of S/N ratio criteria for
optimization; smaller the best, larger the best and nominal the best. To get the
better performance, larger the thermal barrier and thermal cycling resistance are
desired. Hence larger the best criteria has been selected and following expression
was used for analysis.

(9.20)
Where yi represents the observed data and n number of tests in one trial. The
average responses of S/N ratio for thermal barrier are calculated for all the
parameters and are tabulated in table3. A sample calculation for one parameter is
shown below.
192

Figure 9.4. Average response S/N graph for Thermal Barrier test.

Table 9.16. The average response of S/N ratio for thermal barrier test
Symbols Process Parameters Level Level Level Rank
1 2 3
A Temperature on coating side in 0C 44.18 44.19 44.34 0.16 3

B Type of coating 46.08 44.35 42.28 3.8 1


C Thickness of coating 43.87 44.22 44.63 0.76 2

Table 9.17. The optimal set of parameters for thermal barrier test
Symbols Parameter Optimum setting
A. Temperature on coating side (0C) 1000
B. Type of Coating PSZ
C. Thickness of Coating ( microns) 300

From the above table 9.16 and figure 9.4, it is seen that most
significant parameters can be determined by the larger difference of S/N ratio.
193

The experimental results have been compared with optimal parameters obtained
by Taguchi technique, it is found that most significant parameter that increases
thermal barrier is type of coating followed by thickness of coating and
temperature on coating side.

ANOVA is used to determine the most significant parameter affecting


the output quality and characteristic using the quantities such as degree of
freedom (f), sum of squares (SS), variance (V) , percent contribution of each
parameter (F-ratio) and then contribution ratio or contribution of variance is
determined. From the table 9.18, it is observed that highest contribution ratio is
for type of coating (B) while Temperature on coating side (A) is the lowest and
thickness of Coating (C) has moderate value between the other two.

Table 9.18. ANOVA results for thermal barrier test


Parameters SS DOF Variance (V) F-ratio Contribution Ratio (%)
A 0.0462 2 197.5089148 19.43 0.203
B 21.79162 2 95.4493 18.025 95.83
C 0.860422 2 127.3109 18.92 3.784
Error 0.038 2 0.019 1 0.17
Total 22.74 8 100

The confirmation test is conducted to verify the results and to predict


the optimum condition at the selected levels of significant parameters such as
A3, B1 and C3. The most optimal set of combination of parameter is found out.
The predicted average (M) of the response characteristic of TBC can be
expressed as.
M = (A3-T) + (B1-T) + (C3-T) +T, where T=overall average of S/N ratio. (9.21)
= (44.34-44.24) + (46.08-44.24) + (44.63-44.24) + 44.24=46.57dB.
194

A confidence interval (C.I) has to be evaluated on the for the M using


the following expression
= 0.31 (9.22)

Where Nf= and R= number of trials to

run confirmation test. Using ANOVA table, v e= 0.019 and fe=2 and F (1, 2)
=18.51 at 95% confidence level and tabulated.

At the 95% confidence level the estimated / predicted average of the


thermal barrier (TB) found to be in the range of 46.26 dB < TB > 46.88dB.

Table 9.19. The comparison of experimental and predicted results of TB test


Optimum Level
Particulars Estimation Experimental Difference
Level A3B1C3 A3B1C3
Thermal barrier/ 205 215 10
Temperature drop in 0C
S/N Ratio in db 46.57 46.65 0.29

9.5.4. Results of Thermal Cycling Test Optimization of Parameters

For the above test, nine different thermal cycling tests were conducted
using the control factor combinations in the specified orthogonal array shown in
table 9.20. To get the better performance, larger the thermal cycling resistance is
desired and in this case also once again the larger the best criteria has been
selected for S/N ratio in the analysis.
195

Table 9.20. Experimental layout and summary of results for thermal cycling test

Trials Temperature on Type of Thickness of No. of cycles S/N ratio


coating side in 0C coating coating in µm withstood in dB
1. 700 PSZ 150 320 50.10
2. 700 AT 225 315 49.97
3. 700 A 300 260 48.30
4. 900 PSZ 225 370 51.36
5. 900 AT 300 340 50.63
6. 900 A 150 255 48.13
7. 1000 PSZ 300 400 52.04
8. 1000 AT 150 320 50.10
9. 1000 A 225 265 48.47

Table shows the L9 array for the thermal cycling test. The
experimental results have been compared with optimal parameters obtained by
Taguchi Technique and it is found that most significant factor is type of coating
followed by thickness of coating and temperature on coating side which is
similar as found in thermal barrier test as shown in table 9.21 and figure 9.5.

From the ANOVA, it is noticed that type of coating (B) indicates the
most significant factor and the highest contribution ratio is thermal cycling
resistance having 85.13%, while Temperature on coating side (A) has the lowest
about 7.744% and Thickness of Coating (C) has moderate value between the
other two 6.122% as shown in table 9.22.
196

Table 9.21. The S/N ratio for Number of cycles withstood by Different TBC
Symbols Process Parameters Level1 Level2 Level3 Delta Rank

A Temperature on coating 49.46 50.04 50.20 0.74 3


side
B Type of coating 51.17 50.23 48.3 2.87 1
C Thickness 49.44 49.93 50.32 0.88 2

Fig.9.5. Average response S/N graph for Thermal Cycling test

Table 9.22. ANOVA Results for Thermal Cycling test


Factors Sum of Degree of Variance F-ratio Contribution
squares (SS) freedom(f) ratio (%)
A 0.924467 2 0.4622335 9.24467 6.122
B 12.82667 2 6.413 128.26 85.13
C 1.16667 2 0.583335 11.6667 07.744
Pooled error 0.148993 2 0.074496 1 0.989
Total 15.0668 2 100
197

In order to verify the results once again in the thermal cycling resistance
test, confirmation test has been conducted using predicted average (M) and
Confidence interval (C.I.) at 95% confidence level.

= (50.20-49.9)+(51.17-49.9)+(50.32-49.9)+49.9 = 51.89dB
in the range of 46.26dB < TCT > 46.88dB.

=1.23. (9.18)

At t he 95% confidence level the estimated / predicted average (mean)


of the thermal cycling test (TCT) found to be in the range of 46.26dB < TCT >
46.88dB.

Table 9.23. The optimal set of factors for Thermal Cycling test

Optimum Level
Estimation Experimental Difference
Level A3B1C3 A3B1C3
Number of cycles withstood 342 400 58
S/N Ratio in dB 50.66 52.04 1.38

You might also like