Pin Me

Design of Experiments (DOE) in Six Sigma

written by: Heidi Wiesenfelder • edited by: Jean Scheid • updated: 7/6/2011

Six Sigma is a process management methodology that relies on data for decision making. Design of Experiments (DOE) provides the ability to confirm relationships between input and output variables. It is commonly used in both DMAIC and Design (DFSS) projects.

  • slide 1 of 1

    Outside the Six Sigma framework, business leaders often test beliefs about cause-and-effect relationships either by introducing changes one at a time or by changing multiple factors at once. Both of these approaches have limitations. Testing out one factor at a time can be very time-consuming and does not reveal any interactions that may occur when certain modifications are combined. And modifying more than one variable at a time results in an inability to be sure which variable is responsible for any change in outcome.

    Design of Experiments (DOE) is a systematic method of establishing the effect of altering specific variables on aspects of process performance. It makes use of the same principles as the scientific method, essentially applying them in the business setting.

    Full factorial design for two factors The method actually comprises a set of different techniques or experimental designs for manipulating input variables and measuring the resulting output variables, and Black Belts are trained in these techniques. In some cases he or she chooses a full factorial model, where all individual inputs and all possible combinations of these inputs are assessed for their impact on performance. Other versions such as fractional factorial and response surface models apply specific algorithms to reduce the number of iterations or runs necessary to obtain useful results. Depending on circumstances, the Black Belt assigned to a project selects the appropriate design.

    Not all DMAIC projects require DOE. In some cases project teams rely on existing data and correlation analysis to confirm causal relationships. Yet often existing data is insufficient, so DOE is often used In the Analyze phase of DMAIC to confirm potential root causes. Using DOE, the project team demonstrates whether when the suspected cause is changed, performance is altered in a predictable way. Teams also use DOE in the Improve phase to verify that the planned improvements will yield the expected results.

    In addition, DOE in the Improve phase can yield information about which settings are optimal for each factor, so that the best possible version of the solution is implemented. In the same way, Design for Six Sigma (DFSS) uses DOE to create optimal products and processes.

    Software programs such as Minitab provide a means of establishing the conditions for the set of experiments. The same programs can typically also be used for analyzing the resulting data.

    When the Design of Experiments approach is implemented correctly, it provides valuable information that allows project teams to determine root causes for process performance problems and to identify effective solutions that counter those root causes. It is thus a valuable tool for any Six Sigma process management initiative.