1. RME-085
Total Quality Management
By:
Dr. Vinod Kumar Yadav
Department of Mechanical Engineering
G. L. Bajaj Institute of Technology and Management
Greater Noida
Email: vinod.yadav@glbitm.org
Topic: Taguchi’s Quality Engineering Philosophy
2. Taguchi’s Quality Engineering Philosophy (Dr. Genichi Taguchi – Electrical Engineer)
The loss a customer sustains can take many forms, but it is
generally a los of product function or purpose
[1].
Taguchi’s Quality loss: Loss imparted to society from the
time a product is shipped “No amount of inspection will ever
improve the quality of production” .
Quality must be “Engineered in” since it cannot be
inspected out[1].
Taguchi’s contribution:
1. Quality Engineering philosophy: Target and loss function.
2. Methodology: System, parameters, tool design.
3. Experimental design: Use of Orthogonal Arrays (OA).
4. Analysis: Signal to Noise ratio (S/N).
Taguchi’s Concept of quality (Robustness)
Robust: The product performance must be Immune to
the noise variable.
Plan
Robust Design: Product development process[1]
Dr. Genichi Taguchi
Concept
System
level
design
Detailed
design
Testing
Production
Robust design concept
Quality improvement
steps (too late)
Robust parameter and tool
design
Product development process:
Design phase- Less cost
Service phase- More costly
Post delivery phase- Expensive
3. Taguchi’s Quality Engineering Philosophy contd.
Note: The contents used in this slide is being used for academic purposes only, and is intended only for
students registered in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not
intended for wider circulation.
- Quality sciences developed in UK - Design of
Experiments (DOE)[2].
- Quality sciences developed in US - SQC.
Taguchi method
- Robust design method - Statistical method. (Step-1:
System design Step-2: Parameter design Step-3: Tolerance
design).
- Improve the quality of manufactured goods.
- Applied to Engineering, biotech, marketing and
advertising.
- Taguchi’s Three principal contributions to
statistics: (i) Specific loss function (ii) Off-line quality
control (iii) Innovations in design of experiments
(DOE).
Loss Function (Taguchi) – Deviation from targeted value
(results into loss) – either lower than target or higher than target.
- Loss imparted to society from the time a product is
shipped[2].
- Societal losses: Failure to meet customer requirements,
failure to meet ideal performance and harmful side
effects.
Observation: The design and specifications of TV’s
manufactured by Sony-Japan and Sony-USA were identical,
still U.S. customers preferred the color density of shipped TV
sets produced by Sony–Japan over those produced by Sony–
USA.
Reason: Significant difference in frequency distributions.
Fig. 1 – TV color distribution [2]
Customers preferred the quality as meeting the target (Sony-Japan) rather than just meeting
the specifications (Sony-USA).
Worst
Best
4. Taguchi’s Quality Engineering Philosophy (Nominal-the-Best)
Note: The contents used in this slide is being used for academic purposes only, and is intended only for students registered
in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not intended for wider circulation.
Quality loss: Common measure - out of USL or LSL (But
not suitable for all areas).
Nominal-the-Best: Taguchi developed more than 68 loss
functions, many situations are approximated by the
quadratic function that is called “Nominal-the-best” type.
Fig. 2 – Step and Quadratic loss function [2]
Within spec: loss = $0
Quadratic loss function (Taguchi’s concept)
L = k (y-τ)2
Where,
L = Cost incurred as quality deviates from the target
y = performance characteristic
τ = target
k = quality loss coefficient
k is determined by setting Δ = (y - τ), the deviation from the target.
When Δ is at the USL or LSL, the loss to the customer of repairing or
discarding the product is $ A .
Thus, k = A / (y - τ)2 = A / Δ2
Target value
A quadratic loss function penalizes a product for being “off target”
deviation
5. Taguchi’s Quality Engineering Philosophy (Average loss)
Note: The contents used in this slide is being used for academic purposes only, and is intended only for students registered in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not intended
for wider circulation.
Average loss (AL): Assumes that the quality characteristic is static.
- In actual practice, one cannot always hit the target (τ).
- AL is varying due to noise (like virus), and the loss function must reflect
the variation of many pieces rather than just one piece.
- Noise factors classification - External (environment where product is to be
used) and Internal (component spec. deviation from set target).
- Internal noise: (i) Unit-to-unit (e.g. seal loose) (ii) Deterioration: Due to
design problem (e.g. wear and tear of specific part).
- Noise must be minimized.
- Noise factors cause deviation from the target, thereby causing loss to the
society.
- The loss can be minimized by reducing the variation, s, and by bringing
the average, y, to bring it on target (τ).
6. Taguchi’s Quality Engineering Philosophy (Other loss functions)
Note: The contents used in this slide is being used for academic purposes only, and is intended only for students registered in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not intended
for wider circulation.
Other loss functions: (i) Smaller-the-better (ii) Larger-the-better
Fig. 3 – Other loss functions [2]
e.g. Defects e.g. Warranty
7. Taguchi’s Quality Engineering Philosophy (Other loss functions)
Note: The contents used in this slide is being used for academic purposes only, and is intended only for students registered in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not intended
for wider circulation.
Other loss functions: (i) Smaller-the-better (ii) Larger-the-better
Smaller-the-better (STB)
- The target value is 0 (zero).
- No negative values for the performance characteristic.
- Examples: By pass from cooling coils, radiation leak from an
oven, cost, time, material, start time of laptop, pollutant
emission from car etc.
Larger-the-better (LTB)
- The target value is infinity (∞), which gives a zero loss.
- No negative values (Worst case is at y = 0).
- Opposite to “Smaller-the-better” concept.
- Achieving ∞ is typical. Hence, reciprocal of LTB is good
option .
- Examples: Adhesive bond strength, weld strength, warranty
of product, larger mileage, service life of product etc.
Table-1 Relationship of the loss function to the
mean squared deviation (MSD)[2]
8. Taguchi’s Quality Engineering Factors and their Classification[1]
Note: The contents used in this slide is being used for academic purposes only, and is intended only for students registered in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not intended
for wider circulation.
Control factors: (Optimum design factors ) – Part specifications, material, roughness etc.
Noise Factors: Expected noise while production or during usage (e.g. dimensional inaccuracy, operating
temperature etc.)
Adjustment factors: Affects the mean (target) but not the variance (e.g. bonding time of adhesives)
Signal factors: Set by the users to meet customer’s requirements (e.g. location of welding torch).
Stages of product
development
Sources of Noise
Environmental
variables
Product deterioration Manufacturing variations
Product design stage Yes Yes Yes
Process design stage No No Yes
Manufacturing stage No No Yes
Table-1: Noise reduction measures[1]
9. Taguchi’s Quality Engineering (Robust Design Method[1])
Note: The contents used in this slide is being used for academic purposes only, and is intended only for students registered in B.Tech Mechanical Engineering at AKTU Lucknow in VIII semester 2019-20, and is not intended
for wider circulation.
Taguchi proposed 3 step off-line QC technique for product design
1. System Design: concept, synthesis, innovation etc. (Materials, processes,
values, parameters).
2. Parameter design: (Robust product with less noise) : Signal-to-noise ratio
(S/N).
Signal: Square of mean value of quality characteristics - Close
to tolerance (desirable).
Noise: Measure of variability (variance) – Undesirable
Large S/N is expected for robust design
3. Tolerance design: (Optimized tolerance to minimize cost)