‫الجمهورية التونسية‬

‫وزارة التعليم العالي والبحث العلمي‬

‫جامعة تونس‬
‫المدرسة الوطنية العليا للمهندسين بتونس‬
Département de Génie Mécanique ‫قسم الهندسة الميكانيكية‬

END OF YEAR PROJECT

OF

1st Mechanical Engineering

DEVELOPPED

BY
AJMI ISRA
&
BALTI CHAYMA

The Taguchi Method

Supervisor: Mr. Ali TRABELSI

College year: 2023-2024

 Acknowledgements

We would like to express our sincere gratitude to our families for their unwavering support throughout
this project. Their encouragement and understanding during challenging moments were invaluable.
We are also grateful to Professor Ali TRABELSI for his guidance and expertise. Mr Ali’s insightful
feedback and willingness to answer our questions significantly enhanced our learning and the final
outcome of this project and helped us navigate the complexities of this project.

We are grateful for the opportunity to collaborate on this project. Working together allowed us to learn
from each other's strengths and perspectives, fostering a successful partnership.

1
Table of contents

I. INTRODUCTION TO THE TAGUCHI METHOD ................. 5

1. Principles of the Taguchi Method ............................................................................................5

1.1. Taguchi Loss Function .....................................................................................................5

1.2. Robust Design Concept ....................................................................................................7

1.3. Signal-to-Noise Ratio (S/N Ratio) ....................................................................................8

1.4. Orthogonal Arrays ......................................................................................................... 10

2. Applications of the Taguchi Method ...................................................................................... 12

3. Case study1 ........................................................................................................................... 12

3.1. Identifying the Design Factors and the Noise Factors ..................................................... 13

3.2. Orthogonal Array ........................................................................................................... 13

3.3. Experimental Details...................................................................................................... 14

4. Case study 2 .......................................................................................................................... 18

4.1. Determining the Primary Function and Any Negative Impacts ....................................... 18

4.2. Identifying the Control Factors and their levels ............................................................... 19

4.3. Orthogonal Array ........................................................................................................... 19

4.4. Executing The Matrix Test ............................................................................................. 20

4.5. Experimental Details ...................................................................................................... 21

II. CONCLUSION .......................................................................... 24

BIBLIOGRAPHY REFERENCES ................................................ 25

2
Figures

Figure 1-Loss Function .......................................................................................................................6

Figure 2-Variation of values according to the loss function. ................................................................7
Figure 3-The Process of the Robust Design .........................................................................................8
Figure 4- The effect of the noise on signal. ..........................................................................................9
Figure 5- Signal+Noise ..................................................................................................................... 10
Figure 6 -Average per Run................................................................................................................. 15
Figure 7- S/N Ratio per Run ............................................................................................................... 16
Figure 8- AVERAGE E ......................................................................................................................... 17
Figure 9- AVERAGE F ......................................................................................................................... 17
Figure 10- AVERAGE G ...................................................................................................................... 18
Figure 11-Charts Showing Parameter Level v/s S/N Ratio .................................................................. 22

3
Tables

Table 1- Levels of process parameters 11

Table 2- Design Factors and their Levels 13
Table 3- Noise Factors and their Levels 13
Table 4- Orthogonal Array 14
Table 5- Tabulated S/N ratios 15
Table 6 -Best levels for design factors 16
Table 7- Tabulated average for each noise factor 16
Table 8- Best levels for noise factors 18
Table 9- Control factors & Noise Factors 19
Table 10- Selected Factors and their Levels 19
Tableau 11- Orthogonal Array L9 20
Table 12- OA with Control Factors 20
Table 13- Measured values of surface roughness 21
Table 14- Tabulated S/N ratios 21
Table 15- Average S/N Ratios for each factor 22
Table 16-Optimum values of factors and their levels 23

4
 I. Introduction to the Taguchi Method
In today's highly competitive industrial landscape, the pursuit of optimal performance,
enhanced quality, and cost efficiency is paramount for organizations across various sectors. To
achieve these objectives, engineers and researchers employ sophisticated methodologies rooted
in statistical principles and experimental design. One such methodology that has garnered
widespread attention and adoption is Taguchi Methods, developed by Dr. Genichi Taguchi, a
renowned Japanese engineer. Taguchi Methods offer a systematic approach to optimizing
processes and products, with a focus on robustness and efficiency.

At the core of Taguchi Methods lies the utilization of Orthogonal Arrays, a structured
experimental design framework that enables researchers to efficiently study the effects of
multiple factors on a system while minimizing the number of experiments required. By
strategically selecting factor levels and utilizing orthogonal arrays, researchers can uncover
valuable insights into process variability, identify significant factors, and optimize system
performance.

The application of Taguchi Methods and Orthogonal Arrays spans various industries, including
manufacturing, automotive, electronics, healthcare, and beyond. From fine-tuning
manufacturing processes to enhancing product design and performance, Taguchi Methods have
proven to be invaluable tools in the pursuit of operational excellence.

1. Principles of the Taguchi Method

The Taguchi method stands as a robust quality engineering approach, highlighting the
significance of crafting products and processes resilient to environmental and manufacturing
fluctuations. Key tenets of this method encompass:

1.1. Taguchi Loss Function

The Quality Cost Function, or Taguchi Loss Function, is a useful tool which assists
firms comprehend how product quality affects their bottom line. Through the
conversion of deviations from a desirable target value into a monetary loss, it goes
beyond a conventional pass/fail inspection. Rework expenses, warranty claims, scrap
materials, and even dissatisfied customers might all be included in this loss.

5
 The Taguchi Loss Function's basic principle is that deviations from the ideal, even those
that fall within allowable limits, have a cost. The greater the deviation from the intended
value, the higher this cost becomes. Slightly off goal products may need to have modest
revisions made, whilst severely off target products may result in major waste or
dissatisfied clients.

Figure 1-Loss Function

The equation for the Taguchi Loss Function is:

L(x) = K * (x - T) ^2.

L(x): is the monetary loss resulting from a particular departure from the target value.

K: the loss coefficient is a constant that establishes how quickly monetary loss rises in
response to departures from the desired value.
x: is the actual value of the quality characteristic under evaluation.
T:is the desired, ideal, or best value for the quality attribute.

The reason for the significant implications of the Taguchi Loss Function is its dichotomous
relationship which is quadratic in nature. Other terms when a product deviates from
specification linearly indicate that losses also increase proportionally. When calculated as
quadratic, the losses when a product moves from two or three standard deviations from
quality can explode exponentially even if the actual production is well within two or five

6
sigmas from performance. It is easy to envision a situation when even a statistically
insignificant discrepancy over a product dimension can elicit more defective units. The
expenses for the required rework, scrapped materials, and potential warranty claims can grow
drastically, while one would not expect it. Therefore, the potential costs can eliminate any
profit from producing the respective product.

For manufacturers, this emphasis on the exponential nature of quality-related expenses is a

potent wake-up call. It emphasizes how crucial it is to reduce inconsistency and produce work
of consistently excellent quality. Manufacturers can prevent potential financial setbacks and
establish a reputation for reliability by meeting client expectations and sticking to tight
tolerances. This will ultimately result in increased customer satisfaction and market success.

Figure 2-Variation of values according to the loss function.

1.2. Robust Design Concept

The Robust Design Concept aims to create products and processes that perform
consistently under varying conditions. Key aspects include identifying critical
parameters, understanding variation, optimizing designs to minimize the effects of
variation, and employing tolerance design. It often involves experimental techniques
like Design of Experiments and incorporates reliability engineering principles. The
process is iterative, focusing on continuous improvement based on feedback. Overall,
robust design minimizes risks, reduces costs, and enhances customer satisfaction by
considering variation upfront in the design and optimization process.

7
 Key aspects of the Robust Design Concept include:

1. Identifying Critical Parameters

2. Understanding Variation
3. Optimization
4. Tolerance Design.
5. Experimental Design
6. Reliability Engineering
7. Iterative Improvement

Figure 3-The Process of the Robust Design

1.3. Signal-to-Noise Ratio (S/N Ratio)

Use of the Signal-to-Noise Ratio (S/N Ratio) in process optimization methods such as
Taguchi is essential. In essence, it evaluates the quality of the process by contrasting the
intended result with undesirable variances. Picture automobiles being painted on a
production line. ‘The signal’ might be the paint layer's intended thickness, which is
determined by variables like paint viscosity. The "noise" could be caused by small
differences in the metal panels or temperature swings in the paint booth.
The signal intensity in relation to the noise level—the undesired, random variation we
wish to reduce—is essentially expressed by the S/N Ratio. In comparison to the noise,
a stronger, more reliable signal is indicated by a greater S/N Ratio. This results in a
product that is consistently of high quality, has fewer flaws, and eventually makes
customers happier.

8
 Figure 4- The effect of the noise on signal.

There are several types of S/N Ratios, including:

1. Larger-the-better S/N Ratio: Used when the goal is to maximize a performance
characteristic.
1 𝑛 1
𝑆/𝑁 = −10 × 𝑙𝑜𝑔10 ( ∑ ( 2 ))
𝑛 𝑖=1 𝑦𝑖

2. Smaller-the-better S/N Ratio: Used when the goal is to minimize a performance

characteristic, such as defects or costs.
1 𝑛
𝑆/𝑁 = −10 × 𝑙𝑜𝑔10 ( ∑ 𝑦𝑖 2 )
𝑛 𝑖=1

3. Nominal-the-best S/N Ratio: Used when the goal is to achieve a target value for a
performance characteristic.

𝑦2
𝑆/𝑁 = −10 × 𝑙𝑜𝑔10 ( 2)
𝜎

- (n): Number of observations

- (yi): Observed value

- (σ²): Variance of the noise

9
 The S/N Ratio is a powerful tool for assessing the quality of a product or process and for
identifying the parameters that most influence that quality. It is widely used in
manufacturing industries, engineering, and other fields where process optimization is
critical.

Figure 5- Signal + Noise

1.4. Orthogonal Arrays

Orthogonal Arrays (OAs) in Taguchi Methods streamline experimentation by systematically

varying factors and levels. They ensure balanced coverage of factor interactions while
minimizing the number of experiments needed, making optimization more efficient.

How Are Orthogonal Array Tests Performed?


Determine Variables and Levels: To start, figure out what variables are present in your
testing event. The various elements or qualities you wish to examine are represented by
variables. Determine the levels of each variable, or the precise values or states that it
can assume. For instance, variables used to test a login form can be the multiple-level
authentication technique, password, and username.

Make an array that is orthogonal: Choose an orthogonal array that is suitable for the

10
 number of variables and their respective levels. The particular orthogonal array you
select will rely on the specifications of the project.

Create Test Cases: To create a collection of test cases, use the orthogonal array that was
chosen. An orthogonal array's rows each correspond to a distinct set of variable levels,
or test case. For example, you will have three test cases covering different combinations
of variables and levels if your orthogonal array consists of three rows.

Example: Manufacturing Quality Control

Let us take an example where a business produces circuit boards for electronic devices. In this
instance, OA can facilitate the most effective testing of the boards' numerous components,
including connectors, capacitors, and resistors. You can choose which particular combinations
of components to test by utilizing an orthogonal array. For example, OAT allows you to
efficiently test a subset of these components, covering different combinations, if you have
varied resistance values, capacitor sizes, and connector kinds.
Let’s consider three variables: resistor value (R1, R2, R3), capacitor size (C1, C2, C3), and
connector type (T1, T2, T3). We can use an L27 orthogonal array, which is a suitable choice
for testing three variables with three levels each.

Table 1- Levels of process parameters

Factors Level 1 Level 2 Level 3

resistor value R1 R2 R3

capacitor size C1 C2 C3

connector type T1 T2 T3

In the L27 orthogonal array, with 13 columns available, a 3-factor, 3-level setup would require
27 experiments. However, despite this, the L27 setup was chosen to accommodate additional
factors for further study with the same material and it is given by 3³ =27.

11
 2. Applications of the Taguchi Method

The Taguchi Method finds application across various industries, including manufacturing,
automotive, electronics, healthcare, and beyond. It is particularly useful in optimizing
processes, improving product quality, and minimizing costs. By systematically varying factors
and levels using techniques like orthogonal arrays, Taguchi Methods help identify optimal
settings that enhance performance and robustness. This approach has been instrumental in
streamlining production processes, reducing defects, and enhancing customer satisfaction in
diverse industrial settings.

Examples of applications in different industries

. Manufacturing Process Optimization

. Electronics Product Design
. Healthcare Process Improvement
. Chemical Process Optimization
. Automotive Manufacturing Quality Improvement

3. Case study 1

In a 1987 Quality Progress report, Genichi Taguchi used a real-world example to illustrate his
methodology. Maximizing the pull-off force—which is essential for automotive applications—
between a connector and a nylon tube was the goal. Taguchi's technique aimed to reduce this
connection's sensitivity to fluctuations in order to increase its dependability. By using intensive
experimentation and statistical analysis, he determined the crucial design characteristics and
refined them to attain resilience. Robust Parameter Design, a methodology that prioritizes
robustness to production variability and environmental influences, changed quality engineering,
especially in the automobile sector.

12
 3.1. Identifying the Design Factors and the Noise Factors

Table 2- Design Factors and their Levels

Design Factors Levels

A Interference Low Medium High

B Connector wall thickness Thin Medium Thick

C Insertion depth Shallow Medium Deep

D Percent adhesive Low Medium High

Table 3- Noise Factors and their Levels

Noise Factors Levels

E Conditioning time 24h 12Oh

F Conditioning temp 72° 150°

G Conditioning humidity 25% 75%

3.2. Orthogonal Array

Given that each parameter has 3 levels, and there are 4 parameters:

Number of arrays is 34=81

So, there would be 81 possible arrays for 4 parameters, each with 3 levels according to

Taguchi's method.

13
Instead, Taguchi used the L8 design to model the noise factors and the L9(34) series of
orthogonal arrays to model the design factors.

Table 4- Orthogonal Array

Outer array(L8) E 1 1 1 1 2 2 2 2
F 1 1 2 2 1 1 2 2
G 1 2 1 2 1 2 1 2
Inner array(L9)
Run A B C D Average
1 1 1 1 1 15.6 9.5 16.9 19.9 19.6 19.6 20.0 19.1 17.5
2 1 2 2 2 15.0 16.2 19.4 19.2 19.7 19.8 24.2 21.9 19.4
3 1 3 3 3 16.3 16.7 19.1 15.6 22.6 18.2 23.3 20.4 19.0
4 2 1 2 3 18.3 17.4 18.9 18.6 21.0 18.9 23.2 24.7 20.1
5 2 2 3 1 19.7 18.6 19.4 25.1 25.6 21.4 27.5 25.3 22.8
6 2 3 1 2 16.2 16.3 20.0 19.8 14.7 19.6 22.5 24.7 19.2
7 3 1 3 2 16.4 19.1 18.4 23.6 16.8 18.6 24.3 21.6 19.8
8 3 2 1 3 14.2 15.6 15.1 16.8 17.8 19.6 23.2 24.2 18.3
9 3 3 2 1 16.1 19.9 19.3 17.3 23.1 22.7 22.6 28.6 21.2

3.3. Experimental Details

The objective was to maximize the pull-off force of a connector to a nylon tube for an
automotive application so SNL.

1 𝑛 1
𝑆/𝑁 = −10 × 𝑙𝑜𝑔10 ( ∑ ( 2 ))
𝑛 𝑖=1 𝑦𝑖

14
 Table 5- Tabulated S/N ratios

Run No. S/N Ratio

1 24.0
2 25.5
3 25.3
4 25.9
5 26.9
6 25.3
7 25.7
8 24.8
9 26.2

Average
25

20

15

10

0
1 2 3 4 5 6 7 8 9

Average

Figure 6 -Average per Run

15
 S/N Ratio
27.5
27
26.5
26
25.5
25
24.5
24
23.5
23
22.5
1 2 3 4 5 6 7 8 9

S/N Ratio

Figure 7- S/N Ratio per Run

As showed the Run 5 gives the maximum results in which the design factors in the next levels:

Table 6 -Best levels for design factors

Design factor Level

A: Interference 2: Medium
B: Connector wall thickness 2: Medium
C: Insertion depth 3: Deep
D: Percent adhesive 1: Low

The following table gives the average for each of the separate noise factors:

Table 7- Tabulated average for each noise factor

16
 E F G
Level Avg Avg Avg

1 18.14 17.50 19.28

2 21.21 19.73 18.50

Average E
21.5
21
20.5
20
19.5
19
18.5
18
17.5
17
16.5
1 2

Average E

Figure 8- AVERAGE E

Average F
20
19.5
19
18.5
18
17.5
17
16.5
16
1 2

Average F

Figure 9- AVERAGE F

17
 Average G
19.4

19.2

19

18.8

18.6

18.4

18.2

18
1 2

Average G

Figure 10- AVERAGE G

As showed in the table 7 the best levels that gives the maximum result for the noise factors are:

Table 8- Best levels for noise factors

Noise factor Optimum Value

E: Conditioning time 2: 12Oh
F: Conditioning temp 2: 150°
G: Conditioning humidity 1: 25%

4. Case study 2

This attempt demonstrates the use of Taguchi's Method to improve the features of surface
finish on affecting components that were machine-processed on a lathe. A product's surface
roughness is a measurement of how smooth the surface is, and it has a significant impact
on production costs. Any product's life is also impacted by surface finish, so it is ideal to
have higher grades of surface finish at the lowest possible cost.

4.1. Determining the Primary Function and Any Negative Impacts

The main function Using a lathe machine, do a facing operation on an MS work piece
Surface finish variation is a side effect.

18
 Table 9- Control Factors & Noise Factors

Control factors Noise Factors

Cutting speed Vibration

Depth of cut Raw material variation

Feed rate Machine Condition

Nose radius Temperature

Coolant Operator Skill

4.2. Identifying the Control Factors and their levels

The factors and levels were chosen based on the opinions of specialists.
Table 10- Selected Factors and their Levels

FACTORS LEVELS

1 2 3

Cutting speed (v, rpm) 960 640 1280

Depth of cut (t, mm) 0.3 0.2 0.4

Feed rate (f, mm/min) 145 130 160

4.3. Orthogonal Array

As seen in Table 4, the L9 array is the most appropriate orthogonal array for experimentation.

19
 Table 11- Orthogonal Array L9

Experiment Control Factors

No. 1 2 3
1 1 1 1
2 1 2 2
3 1 3 3
4 2 1 3
5 2 2 1
6 2 3 2
7 3 1 2
8 3 2 3
9 3 3 1

4.4. Executing The Matrix Test

Table 12- OA with Control Factors

Experim Control Factors

ent V(rpm) t(mm) F(mm/min)
No.
1 960 0.3 145
2 960 0.2 130
3 960 0.4 160
4 640 0.3 160
5 640 0.2 145
6 640 0.4 130
7 1280 0.3 130
8 1280 0.2 160
9 1280 0.4 145

20
 Table 13- Measured values of surface roughness

Experiment Surface Roughness (Ra, μm)

No. 1 2 3 4 5 Mean
1 2.35 2.43 1.94 2.91 2.77 2.48
2 2.5 3.6 2.66 2.98 2.64 2.876
3 2.43 2.82 4.01 2.96 4.1 3.264
4 2.24 3.38 2.45 4.05 4.79 3.382
5 2.54 3.67 2.70 4.25 4.37 3.506
6 4.76 4.25 3.19 3.36 4.35 3.982
7 2.04 2.49 3.84 1.71 3.79 2.834
8 4.4 2.5 3.15 3.24 3.1 3.278
9 3.94 2.19 2.31 2.44 3.30 3.306

4.5. Experimental Details

The S/N ratio was determined using the objective function (Surface Finish), which is a smaller-
is-better type of control function. Every experiment's S/N ratio was computed and tabulated.

1 𝑛
𝑆/𝑁 = −10 × 𝑙𝑜𝑔10 ( ∑ 𝑦𝑖 2 )
𝑛 𝑖=1
Table 14- Tabulated S/N ratios

Experiment No. S/N Ratio (dB)

1 -7.9702
2 -9.2568
3 -10.4539
4 -10.9196
5 -11.0971
6 -12.1010
7 -9.2385
8 -10.4642
9 -9.2941

21
 Table 15- Average S/N Ratios for each factor

Level Speed Feed Depth of Cut

Avg S/N ratio Avg S/N ratio Avg S/N ratio

1 -9.2268 -9.3909 -9.45

2 -11.3722 -10.273 -10.21

3 -9.68 -10.616 -10.65

(a). Speed (b). Feed Rate

(c). Depth of Cut

Figure 11-Charts Showing Parameter Level v/s S/N Ratio

As can be seen, the "smaller-the-better" sort of objective function was employed to calculate
the Surface Roughness.
To optimize the condition, the factor levels that corresponded to the maximum S/N ratio were
selected. These linear graphs make it evident that table 9 has the factors' ideal values and
levels.

22
Table 16-Optimum values of factors and their levels

Parameter Optimum Value

Speed (rpm) 960
Feed Rate (mm/min) 145
Depth of cut (mm) 0.3

23
II. Conclusion
In conclusion, the Taguchi method offers a powerful and efficient approach to process
optimization. By focusing on minimizing variation and optimizing control factors, we
were able to achieve significant improvements in [mention the specific outcome you
achieved, e.g., reducing defect rates, enhancing product quality]. This approach proved
valuable in [mention the context of your project, e.g., streamlining a manufacturing
process, improving customer satisfaction]. The Taguchi method's structured framework
and emphasis on reducing noise factors position it as a valuable tool for future process
improvement initiatives.

24
Bibliography References

https://eng.libretexts.org/Bookshelves/Industrial_and_Systems_Engineerin

g/Chemical_Process_Dynamics_and_Controls_(Woolf)/14%3A_Design_of_

Experiments/14.01%3A_Design_of_Experiments_via_Taguchi_Methods_-

_Orthogonal_Arrays

https://testsigma.com/blog/orthogonal-array-testing/

https://support.minitab.com/en-us/minitab/help-and-how-to/statistical-
modeling/doe/supporting-topics/taguchi-designs/what-is-the-signal-to-noise-
ratio/

https://ieda.ust.hk/dfaculty/ajay/courses/ielm317/lecs/robust/robustdesign1.pdf

25