Introduction to DOE

Design of Experiments (DOE) is a tool used to identify which process inputs have a significant impact on the process output, and what the optimal levels of those inputs should be to achieve a desired result (output). It employs a systematic and statistically rigorous approach to introducing purposeful changes to the inputs of a process, with the goal of answering the research questions of interest as clearly and efficiently as possible.

Information obtained from a DOE analysis is superior to the information from the usual regression analysis because DOE is based on making changes to the process in a prescribed manner while controlling the experimental environment. This allows you to have more confidence in the causal nature of these relationships and make much more informed decisions based on the results. Regression analysis on the other hand, is based on happenstance or observational data which can include the effects of various other factors that are not of interest. Needless to say, DOE analysis is a much more powerful technique (although it is also costlier to run) as compared to Regression analysis.

DOE analysis can be used to answer many different types of questions, such as:

• Compare alternative treatments/machines/processes (e.g. which supplier has the highest quality product for the same terms?)
• Identifying the drivers or significant inputs of a process output (e.g. which among a slew of factors truly affect the rigidity of the plastic material?)
• Optimizing the process output (e.g. which settings of the factors identified as significant will maximize the yield, or minimize the defect rate?)
• Reducing the variability in a process or making a process more robust (e.g. by how much can the input factors to the process be varied without compromising quality?)
• Balancing trade-offs among multiple goals (e.g.: what factor settings will result in the best tasting cake while minimizing the cost of the ingredients?)

Experimental design can be used to reduce design costs by speeding up the design process, reducing late engineering design changes, and reducing product material and labor complexity. Designed Experiments are also powerful tools to achieve manufacturing cost savings by minimizing process variation and reducing rework, scrap, and the need for inspection. In the transactional field DOE is becoming increasingly accepted as a means of characterizing relationships and driving breakthrough improvements in virtually any process with controllable inputs. To name just a few: improving order response time, reducing loan processing time, reducing hospital length of stay.