Taguchi methods: parameter design, tolerance design, thequality loss function, design of experiments.

7 points (magic 7) to Taguchi's quality philosophy

• 1. An important dimension of the quality of a manufacturedproduct is the total loss generated by that product to society.
• 2. In a competitive economy, continuous quality improvementand cost reduction are necessary for staying in business.
• 3. A continuous quality improvement program includedincessant reduction in the variation of product performance characteristicsabout their target values.
• 4. The customer's loss due to a product's performancevariation is often approximately proportional to the square of the deviationof the performance characteristic from its target value.
• 5. The final quality and cost of a manufactured productare determined to a large extent by the engineering designs of the productand its manufacturing process.
• 6. A product's (or process') performance variation canbe reduced by exploiting the nonlinear effects of the product (or process)parameters on the performance characteristics.
• 7. Statistically planned experiments can be used to identifythe settings of product (and process) parameters that reduce performancevariation.

  ---

1. An important dimension of the quality of a manufacturedproduct is the total loss generated by that product to society.

• Expanded definition of quality
• Eight aspects of quality: performance, features, reliability, conformance,durability, serviceability, aesthetics, perceived quality.
• -exact definition changes with context with which it is used
• -Taguchi definition: quality is the loss imparted to the society asa whole from the time a product is shipped.
• examples of society losses -failure to meet customer needs for fitnessfor use -failure to meet the ideal performance -harmful side effects causedby the product
• -aim of quality control is to reduce the total society cost
• investments look more attractive when looked at from a long term pointof view of helping society

2. In a competitive economy, continuous quality improvementand cost reduction are necessary for staying in business.

• -best way to increase market share is to provide high quality productsat a low price.
• -continually strive to increase quality, thereby reduce cost, as partof corporate strategy.

3. A continuous quality improvement program includedincessant reduction in the variation of product performance characteristicsabout their target values.

• -must be able to measure quality characteristics in order to be ableto improve them
• -must know ideal values for quality characteristics, called the targetvalue
• -must identify and focus on performance characteristics-those whichdetermine the product's performance in meeting customer demand
• -try to use continuous scales for measurement whereever possible. Intervalscales not good enough.
• -in continuous data is not available, use a categorical scale, ratherthan simply good or bad

4. The customer's loss due to a product's performancevariation is often approximately proportional to the square of the deviationof the performance characteristic from its target value.

• -larger deviation from the target value, the greater the loss to thecustomer. Loss not linear. Quadratic function a good approximatation.
• -the average loss due to performance variation is proportional to themean squared error of a performance characteristic around the target value.
• -still have loss even if the product conforms to be specifications,but misses the target value. Or, the larger the mean squre error, the largerthe loss.

5. The final quality and cost of a manufactured productare determined to a large extent by the engineering designs of the productand its manufacturing process.

• -Product development cycle consists of three stages: product design,process design, and manufacturing.
• Increased product complexity means we must place more attention tothe first two stages.
• Problems caused by environmental variable, product deterioration andmanufacturing variations. Classical qc highlites manufacturing variation.We can take countermeasures against environmental variation and productdeterioration only in the product design phase.

6. A product's (or process') performance variationcan be reduced by exploiting the nonlinear effects of the product (or process)parameters on the performance characteristics.

• -need for off-line quality control activities; technical aids for qualityand cost control in product and process design.
• -examples: sensitivity tests, prototype tests, accelerated life tests,reliability tests.
• -go beyond tolerance limits. Need additional parameters. Taguchi callsthem system design, parameter design, and tolerance design.

System design goes beyond functional prototype design. Need understandingof both customer's and manufacturing needs.

Parameter design is the process of identifying the settings of product(or process) parameters that reduce the sensitivity of engineering designsto the sources of variation. Identify products which can tolerate largevariations in environmental conditions.

Tolerance design is the process of determining tolerances around thenominal (target) sttings identified by parameter design. Are a tradeoffbetween the customer's loss due to performance variation and increase inmanufacturing cost.

7. Statistically planned experiments can be used toidentify the settings of product (and process) parameters that reduce performancevariation.

• 1. Identify all variables that affect performance characteristics
• 2. Classify them into two categories-design parameters and sourcesof noise Design parameter: product parameters whose nominal settings canbe chosen by design engineer Sources of Noise: those variables that causethe performance characteristics to deviate from their target values
• 3. Conduct experiment: objective is to identify the settings of thedesign parameters at which the effect of noise factors on the performancecharacteristics is minimum.
• 4. How: a) systematically vary the settings of the design parametersin the experiment b) compare the effect of the noise

Uses of statistically planned industrial experiments

• 1. to identify the settings of design parameters at which the effectof the sources of noise on the performance characteristic is minimum
• 2. to identify the settings of design parameters that reduce cost withouthurting quality
• 3. to identify design parameters that have a large influence on themean value of the performance characteristic but have no effect on itsvariation. Such parameters can be used to adjust the mean value.
• 4. to identify those design parameters that have no detectable influenceon the performance characteristics. Tolerances on such parameters can berelaxed.

Go to week7 of syllabus

This document created by David Lewis 6/20/96

Last modified