Design of Experiments plays a central role in continual improvement, as well as root cause analysis when it is used to test hypothesis about the underlying cause of a quality-related problem. Its foundation of hypothesis testing is meanwhile applicable to statistical process control (SPC) and hypothesis testing as well as DOE. A comprehensive hands-on understanding of DOE requires one or more college-level courses; the deliverable of this webinar is an understanding of the basics that will equip attendees to explain the business benefits of DOE and work effectively with subject matter experts like Six Sigma Green and Black Belts. It will include an overview of the fundamentals of DOE, including hypothesis testing, factors, interactions, randomization, blocking, and evaluation of experimental results.
key Learning Objectives of Topic:
Attendees will be able, upon completion of the webinar, be able to:
· Explain the value of DOE in terms of obtaining the maximum amount of information for the least possible effort and expense.
· Understand the elements of DOE including the response (measurement of the critical to quality characteristic we seek to understand or improve), factors (considerations that are believed to affect the response, such as materials, methods, machines, and other elements of a cause and effect diagram), levels (choices within factors, such as material A versus material B), and interactions. Interactions are synergistic or antagonistic relationships between factors that make the whole greater or less than the sum of its parts.
· Explain the elements of a statistical hypothesis test.
· Plan experiments to exclude extraneous variation from the results through randomization and blocking.
· Recognize how replication (obtaining more than one result for each set of experimental conditions) can deliver decisive results.
· Recognize how factorial and fractional factorial designs can minimize the amount of experimentation necessary to screen out unimportant factors to allow extensive focus on the important ones.
Topics covered includes:
· Basics of a simple designed experiment; the control versus the experiment.
· Bottom line economic value of DOE as shown by a comparison of a 19th century machining experiment without it (industrial statistics had yet to be invented) and a 20th century experiment to assess the effectiveness of a medical diagnostic test.
· Elements of DOE
o The response variable (what we are trying to model or optimize, such as a critical to quality characteristic)
o Factors (considerations, often from the categories of a cause and effect diagram, that affect the response)
o Levels (choices within each factor such as low versus high, or material A, B, or C)
o Interactions; synergistic or antagonistic relationships between factors that make the whole greater or less than the sum of its parts
· Experimental planning and hypothesis testing; this is the foundation of SPC and acceptance sampling as well.
o Null hypothesis; similar to the presumption of innocence in a criminal trial. We assume that the experiment has no effect, the process is in control, or the production lot's quality is acceptable.
o Alternate hypothesis, which must be proven beyond a (quantified) reasonable doubt: the experiment has an effect, the process is out of control, or the production lot's quality is unacceptable.
o Type I risk or alpha risk; the risk of wrongly rejecting the null hypothesis. This is similar to the risk of convicting an innocent defendant, or the risk of the boy crying wolf when there is no wolf. This is also known as the producer's risk in acceptance sampling, where it is the risk of rejecting a production lot at the acceptable quality level (AQL).
o Type II risk or beta risk; the risk of acquitting a guilty defendant, or the risk of the boy not seeing the wolf. This is known as the consumer's risk in acceptance sampling, where it is the risk of accepting a production lot at the rejectable quality level (RQL).
· Understanding the difference between luck and genuine effects
o p value or significance level (the quantifiable reasonable doubt); the chance that we just got lucky (or unlucky)
· Replication means repeating a set of experimental conditions more than once to increase the chance of detecting genuine effects (or reducing the Type II risk).
· Randomization and blocking exclude extraneous sources of variation from the experiment. The paired comparison test is a simple form of blocking.
· Factorial designs are screening experiments whose purpose is to allow dismissal of unimportant factors from further consideration, to allow focus on significant ones that affect the response variable. The fractional factorial design allows assessment of even more factors and interactions with less effort and expense.
Who will be benefited:
Who (With specific designations and working in specific industries.) would benefit from this topic.
Quality and manufacturing managers, engineers, and technicians
Physical CD-DVD of recorded session will be despatched after 72 hrs on completion of payment
Recorded video session
William A. Levinson, P.E., is the principal of Levinson Productivity Systems, P.C. He is an ASQ Fellow, Certified Quality Engineer, Quality Auditor, Quality Manager, Reliability Engineer, and Six Sigma Black Belt. He is also the author of numerous books on quality, productivity, and management.