The robust design is a concept developed by Taguchi in the 1950s. It is based on the fact that a product must be reliable even unforeseen situations.
Introduction

He will be the father of what’s called the robust design :
“In the next century, the ability to develop robust technologies will be critical to their competitiveness.” G. Taguchi
Definition of robustness
The principle of robustness is to make a product or service less sensitive to uncontrolled factors, called noise factors. The idea is to increase quality by minimizing the effects due to causes that can not be solved. Robustness is defined as follows:
“It is the capacity of a system to maintain its performance, despite changes in the conditions of use or the presence of uncertainty related to its parameters or to its components.” G. Taguchi
For this purpose, Taguchi builds a method of experimentation in 5 steps.
1 – System definition
We will qualify our study system. This is the so-called phase « Screening » where we will prune the different factors to have only those who have a real influence. We will use a plan of Taguchi for Screening.
We will determine 2 types of elements :
- The quality criterion : This is none other than the desired “answer” of our plan of experiences. Preferably, we will take a response directly related to the expectations of the customer”
- Controllable Factors : The set of controllable factors during experimen, but also in production or during use.
2 – Determination of noise parameters
We will realize the diagram of the parameters (P-Diagram), which allows to schematize our system and to identify the noise factors. Taguchi classifies them into 3 families :
- Les bruits externes : outside temperature, altitude…
- Les bruits internes : progressive deterioration of performance due to wear or misuse.
- Noise between Product : fluctuating characteristics of mass-produced parts.
The P-Diagram
The P-diagram is a schematization of the system as a function of all the parameters. It should be noted that the control factors (Signals) affect the mean of the response while the noise factors influence the dispersion around the mean

3 – Realization of the main experience
To carry out this study, we will use the product plan. It consists of crossing the main plane constructed from Signal factors and another plane constructed from noise factors.
4 – Determination of optimal parameters
The determination of the optimal parameters is based on the assumption of additivity of the effects. In other words, if we search for the maximum of the value, then we will take the combination of the effects whose experiments have shown that on this position of the effect, the answer is at the maximum.
5 – Determination of tolerances
Last step of the methodology, the determination of tolerances via the loss function. Taguchi proposes a “monetary” approach to loss in the light of a product’s non-quality. The loss function is used to quantify the loss to the customer due to a dispersion of the functional characteristics. It quantifies a cost that results from a deviation from a target in proportion to that gap.
Source
R. A. Wysk, B. W. Niebel, P. H. Cohen, T. W. Simpson (2000) – Manufacturing processes : integrated product and process design
G. Taguchi, Y. WU (1985) – Introduction to off-line quality control
G. M. Vigier (1988) – Pratique des plans d’expériences. Méthodologie Taguchi
J. C. Chauveau, J. P. Chassaing (1994) – Introduction à la méthode des plans d’expériences par la méthode Taguchi
Alexis et Alexis (1999) – Pratique industrielle des plans d’expériences