Quality Control: Taguchi Method Overview

Taguchi Method

• Statistical method developed by Genichi Taguchi to improve quality of manufactured goods
• Focuses on off-line quality control, designing robust products and processes, and innovations in the design of experiments

Taguchi Loss Function

• Measures financial loss to society resulting from poor quality
• Defines quality as the loss imparted by the product to society from the time the product is shipped
• Considers costs such as:
  • Cost to operate
  • Failure to function
  • Maintenance and repair costs
  • Poor customer satisfaction
  • Poor design

Taguchi's Five Important Points

• Continual quality improvements and cost reduction
• Measure of quality for a product is the total loss generated by the product to society
• Vary factors one at a time during pre-production experimental procedures
• Customer loss due to poor quality is the square of the deviation of performance characteristics from the target (nominal value)
• Achieving conformance to specifications means achieving the target and minimizing variability by analyzing the non-linear effects of factors

Taguchi's Quadratic Quality Loss Function

• Quality loss occurs when a product deviates from the target or nominal value
• As deviation grows, loss increases
• Taguchi's U-shaped loss function curve

Taguchi Loss Function Formula

• Calculated using Quadratic Equation: L (y) = k (y - m)²
  • L = cost incurred as quality deviates from the target
  • y = performance characteristics
  • m = target value
  • k = cost coefficient

Taguchi's Loss Function Types

• Nominal-the-best (NTB):
  • Nominal value is the best and satisfies customer needs
  • Deviation on either side of the target is undesirable
  • Examples: air pressure in vehicle tires, gauge location on instrument panel
• Smaller-the-better (STB):
  • A smaller value is better, higher values are undesirable
  • Examples: vehicle emissions, fuel consumption
• Larger-the-better (LTB):
  • A larger value is better, smaller values are undesirable
  • Example: gas mileage

Robust Design

• Achieved through a three-step decision-making process:
  1. Define the objective
  2. Define feasible options
  3. Select the feasible option that best meets the objective
• Best criterion to measure robust design is the signal-to-noise (S/N) ratio
• Maximum robustness means minimum quality loss and maximum customer satisfaction
• S/N ratio recognizes and measures deviation from the nominal value, integrating it into a single metric
• It is important to define the measure of the quality loss and incorporate it into the design

Signal-to-Noise Ratio Considerations

• Different S/N ratios are required for each performance characteristic
• Nominal value is the best performance characteristic for a given parameter
• NTB should be used whenever possible because this allows for two-step optimization
• S/N ratio measures deviation from the nominal value, allowing for subsequent adjustment

Response and Signal Ratio

• A large S/N ratio means lower standard deviation
• For dynamic signals, a series of dynamic S/N ratios exist
• A signal factor is a control factor chosen that can modify output response linearly
  • Example: Steering wheel turn angle is the signal factor that adjusts vehicle motion radius of curvature
• The objective of achieving a robust design is to have the highest S/N ratio (smallest standard deviation or variation)
• The mean-squared deviation (MSD) for LTB is given in a specific equation
• The quality loss for LTB is given in a specific formula

Taguchi's Loss Function Diagram

• Shows the relationship between customer tolerance, manufacturer tolerance, target, performance characteristic, and loss
• Illustrates that loss occurs due to off-target performance

Taguchi's Loss Function Formulas

• For NTB and STB, the loss function takes a specific form
• For LTB, the quality loss function takes a different form
• Δ0 should be in the denominator instead of the numerator to maintain consistency among the NTB, STB, and LTB methods