Select Page
[Total: 0    Average: 0/5]
Not widely used in France, the Shainin method is a statistical approach to solving problems, as is the 6 Sigma.

## Introduction

Created by Dorian SHAININ (1914 – 2000), the SHAININ method is a problem-solving methodology.

Trained as an aeronautical engineer, Dorian SHAININ worked for many aeronautical companies: Design Engineer for United Aircraft Corporation, reliability consultant for Grumman Aerospace and Pratt & Whitney. Mentored by Joseph M. Juran, he developed a total of 20 statistical tools to solve problems and improve reliability. In 1984, he created the company Shainin consultants.

In 1996, he was the 15th Honorary member of the ASQ (American Society for America) and the first to win the four major medals. In 2004, ASQ created the Dorian SHAININ Medal for the exceptional use of single or creative application of statistical techniques in problem solving in relation to the quality of products and services.

## The Shainin philosophy

In Shainin philosophy, continuous improvement is a false idea. In reality, it evolves by tier, project after project. Its strategy proposes to choose judiciously the projects on which to work and to use a problem-solving system based on statistical engineering applicable everywhere: design, production, service1… Shainin developed thus 5 principles2.

### Quality is the customer’s enthusiasm

Customer satisfaction is the minimum level to be achieved to be competitive or rather to ” exist “. To be a ” world class “, the challenge is to aim at the customer’s enthusiasm, in other words, to bring him a competitive asset that generates the purchase of the product. The principle is to focus on the characteristic or characteristics that generate this enthusiasm and reduce their variation.

### There is no random variation

For any y-output of a system, there are a number of X input factors that cause the variation. With infinite resources, we could find all the X’s. Nevertheless, looking for all these parameters would be wasteful because according to the principle of Pareto, the discovery of all the small causes and the control of them would be too complex and too costly.

### The Red X

The concept of Shainin’s Red X stems from his collaboration with Joseph M. Juran. In the years 1940, he democratized the Pareto: I observed (as much before me) that quality defects are of uneven frequency, i.e. when a long list of faults is classified in order of frequency, relatively few Defects are the major part of the problems. »3

In the years 1950, Shainin recognized that the principle of Pareto Could be applied to the resolution of the problems of variation. Shainin concludes that among the thousands of variables that could cause a change in the value of an output data, a causal relationship had to be stronger than the others. Shainin called this major cause the ” Red X ” and demonstrated that the cause could exist in the Shape of an interaction between independent variables. The second cause being the ” Pink x “, the third the ” Pale pink x “…

### The principle of combining independent causes

When independent causes are combined, the variance of the resulting distribution is the sum of the variances of the independent causes (principle of additivity of variances). So the only way to get a major improvement in reducing the variation of the process is to find the Red X and control it.

ΣOutput = √ (σredX2 + σpink2 +…)

### Making parts Talk

The main characteristic that distinguished the Shainin method from that of Taguchi was the principle of “ making the parts speak “. In the classical design of experiments or Taguchi, engineers brainstorm to make assumptions about the possible causes of a problem. The method of Shainin refutes this theoretical step, requiring beforehand the diagnosis of the causes through one or more techniques (Ishikawa…) of generating ideas designed to determine, by the empirical investigation of the parts, the First cause, or ” Red X “.

## The SHAININ methodology

More access on the gemba search of the causes and less on the measurement than the method 6 Sigma, the method Shainin follows a fairly close logic and consists of a logical suite of statistical tools.

It is noted that this method is based on non- parametric statistical tools. Certainly less accurate, they have the advantage of working even with small samples and are robust to the non-normal data.

## 1-Focus

In the Shainin process, the first step is to define the problem. We go as in most problem solving methodologies (PDCA, 8D, DMAIC…) Use the 5W2H. This allows to describe the problem and the background of it: Pareto, Default rate…

## 2-Approach

This second phase aims to target the Green Y and validate the means to measure it. The Green Y represents the defect that we want to eliminate and put under control. To ensure that statistically, our study will be ” viable ” and validate our improvements, Shainin proposes from this stage to check whether the measuring system is capable. It relies on a tool that it has developed: the Isoplot.

From this stage we will follow the following process:

## 3-Converge

The Converge phase is the stage where we will actually look for the causes of our defect, using a systematic logic of comparison between good parts and bad4. The Shainin method offers a 2-stroke investigation. A first to enumerate all the assumptions and make a first filter, and a second to identify the actual cause.

The tools of the phase Converge

 Multi Vari Graphics Search for components or products Paired comparison It is a graphical technique for visualizing the variability of parts regarding 3 general factors. By dismantling/reassembling and swapping parts, we will locate the element (or elements) that are causing the defect. Having the same function as the component or product search, this technique is used in case the parts cannot be dismantled or if the dismantling damages them.

### Variable Search

The Finding variables is to tighten a little more the number of factor influencing. The goal is to select a maximum of 4.

### Full factorial Plan

At this point, we have only four factors to the maximum. Via a plan Full factorial Standard, we will validate their influence and quantify it.

## 4-Test

We’ve identified the Red X and solutions. The challenge of the tool B vs C and compare the current situation (current) with the new situation (Better).

## 5 – Understand

The relationship between the Green Y and the Red X is now known. We will optimize it to determine with precision the relationship between the different causes and the Green Y. This will be used to Design Of Experiments for response surface (RSM).

## 6 – Apply

The challenge of this phase is to implement the tools to maintain the improvements achieved.

### Posicontrol

The Posicontrol plan is a Process control plan. To put it in place, we will have to answer the following questions:

• What is the variable to be put under control following the optimizations measured in the previous experiment?
• Who will have to carry out the measurement and monitoring and registration of the control?
• How to measure: measuring system…?
• Where do you measure to have a reliable and representative result?
• When should the measure be done? : first evaluated by experience and then optimize as you are.

### Precontrol

To manage and validate the improvements obtained, we will set up the Precontrol. This is a very similar method to SPC.

### Process Certification

This step is intended to ensure that ancillary causes are not able to disturb the measurement and generate variation: organizational problem, poor maintenance, human error… More often than not, we set up 5S, Poka-Yoké, standards,…

### Operator Certification

It is a step that requires training and a follow-up of education.

## 7-Leverage

We follow the process through the control of the Red X. For this we use the Lot plot.

## Source

1 – D. Shainin (1995) – A Common sense approach to quality management

2 – D. Shainin (1993) – Strategies for technical problem solving

3 – J. Butman (1997) – Juran: A lifetime of influence

B. Maxson, H. Dao (2009) – A convergent approach to problem solving

J. Kosina (2005) – Quality improvement methods for identification and solving of large and complex problem

D. Johnstone (1957) – Statistical quality control

J. H. Ellington (2003) – The optization of General Motor’s warranty, system by reducing mean time to discover failure

S. H. Steiner, R. J. MacKay (2009) – An Overview of the Shainin System for quality improvement