Dr. Genichi Taguchi has contributed a lot in making Design of Experiments (DOE) more popular. Yet it is wrong to think that the Taguchi methods are just another way of doing designed experiments. It is a complete philosophy and a structured way of “Quality Engineering”. The most important ideas are:

• A quality product creates over its entire life span a minimum loss (to be expressed in money!) to the society. The link between loss and a technical characteristic is described by the Loss function.
  
  
  
  To minimize the loss products must be produced as close as possible to target and with minimum variation.
  
  
• Quality needs to be built-in into the product. A lot more attention should be given to Off Line Quality Control to avoid problems from happening.
  
  
• Different types of noise (product variation within tolerance, changes in operating conditions, losses from nearby systems,  …) influence a system and lead to deviations from the ideal condition. To avoid this robust products and processes must be developed, so they will function to target even under varying noise conditions.
  
  
• Looking for parameter settings that increase robustness (parameter design) is an essential part of the Taguchi methods. Robust design is a form of design that does not attempt to solve problems, but rather tries to approach the ideal function of the system as close as possible under various noise conditions.
  
  

QS Consult view and approach

Within the quality world there has been (and still is) a lot of controversy between advocates of Taguchi and classical design of experiments. QS Consult has a totally different view on this. We are convinced of the power of integrating methods. There are several Taguchi ideas that can be combined with the logic of classical DOE.

• Robustness and parameter design can be performed with classical DOE test designs. Besides, the Taguchi orthogonal arrays are highly fractionated designs, very often Plackett-Burmann screening designs.
  
  
• Management is first and foremost interested in the financial benefits of a project. It is important for a quality professional to be able to turn his technical improvements (like lower process variation) into money. The Loss function offers a way to do this. Do not forget that this is the main reason why Six Sigma has become so important.
  
  
• The choice of a good characteristic to optimize a system is very important. Taguchi advises to use the driving forces of a system rather than the resulting performance (like number of defects). By choosing an additive characteristic interactions can be avoided, which is very important when using heavily fractionated designs. Examples of good characteristics are energy and mass and dynamical characteristics that are expressions of the ideal function.

Services offered

• Training Taguchi Quality Engineering – from beginner to specialist

• Consulting when setting up Taguchi Quality Engineering

Additional information

At the 54th Annual Quality Congress (May 2000) in Indianapolis we gave a presentation on the integration of classical DOE with Taguchi.

Make Love, Not War: Combining DOE and Taguchi (Pdf)