Integration of Six Sigma and the Design of Experiments Process

The Six Sigma methodology has been implemented in nearly every imaginable industry. It has generated stunning results for companies in the service, hospitality, and banking industries as well as the manufacturing industry. By isolating existing business processes, companies can use the strategy to identify areas in which inefficiencies exist. The results include lower costs to bring a product to the end customer, higher quality, and improved customer satisfaction. However, the methodology has inherent limitations.

If a business process includes only a few inputs, it is relatively simple for a Six Sigma team to analyze for improvement. On the other hand, if a stream contains 100, 20, or even 10 inputs, it is far more complex to review using a traditional approach. The Design of Experiments (DoE) system was developed to help address this issue.

In this article, I'll explain the traditional Six Sigma approach as it is currently deployed throughout organizations in every industry. Then, I'll describe Design For Six Sigma (DFSS) and explain how it relates to the original methodology. Lastly, we'll explore the factors and responses that DoE was created to address.

The Traditional Approach

As we've covered in past articles, the original system follows a DMAIC (Design, Measure, Analyze, Improve, and Control) system of steps. Each step has been categorized according to its goals. The Design step is dedicated to identifying an organization's objectives and deliverables as they relate to the end customer (who might be within or outside the company).

The Measure step is devoted to understanding the current performance levels of the business process under review. It involves determining the key metrics and gathering data regarding the outputs of the process. It is during this step that performance benchmarks are established. The Improve step involves the use of statistical analysis to further understand the influence of key inputs on the performance benchmarks.

The next two steps (Improve and Control) are focused on isolating problem areas that are generating defects or inefficiencies, creating solutions, and implementing those solutions. Once they have been implemented, the Green Belts and Black Belts monitor their impact, and design a plan for handing off the monitoring responsibilities to others.

What Is DFSS?

DFSS stands for Design For Six Sigma. Though the methodology is related to the original DMAIC approach, there are important differences. While Six Sigma and DMAIC involve the analysis and improvement of existing processes, DFSS emphasizes the creation of an entirely new process. One of the specialized tools that DFSS relies upon is DoE. In a way, DFSS is an amalgamation of DMAIC and DoE.

While the last two steps of DMAIC are devoted to improving and controlling process outputs, DFSS uses two alternate steps: Design and Verify. Both focus on optimizing and measuring the experimental runs that will eventually become business processes.

Factors And Responses Of DoE

We have arrived at factors and responses (in the context of DoE) through the back door. You already know that DMAIC identifies key inputs that impact quality. You also know that the methodology monitors outputs in order to measure the success of the implemented solutions. Factors and responses are simply inputs and outputs. However, they involve a bit more complexity than the inputs and outputs used in a traditional Six Sigma deployment.

The factors are analyzed through factorial experiments. While DMAIC is designed to isolate and study one variable at a time, factorial experiments are conducted to measure the results of multiple inputs simultaneously. In the process, DoE can monitor multiple responses that result from the inputs. This is necessary in order to create an entirely new business process. Otherwise, doing so would take far too much time.

Six Sigma and Design of Experiments (and in some cases, other DFSS tools) should be integrated when multiple variables need to be analyzed at the same time. It is a relatively new combination of methodologies. But, it will likely grow in popularity as business processes become more complex.