DOE Studies

Quick reference

• 00:04 Hello, I'm Ray Sheen.
• 00:05 There are many different variations of DOE Studies.
• 00:08 However, there are a few common questions and guidelines.
• 00:12 And before we look at the different types of DOE studies,
• 00:15 let's address these common points.
• 00:18 I'll start with some basic ground rules for conducting a study.
• 00:23 First, a DOE is a process, not an event.
• 00:26 It will take time.
• 00:27 It has many steps that must be completed in the correct order.
• 00:31 In fact, one of the reasons we teach this course is to help you to
• 00:35 learn the process, so you can apply it and conduct a successful DOE study.
• 00:39 Because it is a process, it must be planned and
• 00:42 monitored to ensure that the process is followed correctly.
• 00:46 The DOE study results will be based upon a statistical analysis.
• 00:50 If the study process is not managed well, the data points used in the statistics
• 00:55 will not be valid and the statistical analysis will be meaningless.
• 01:00 You will see as we go through the process of different DOE types, that they often
• 01:04 will consist of several studies or sets of experiments, each one narrowing the area
• 01:10 of focus, to improve the understanding and validity of the statistical results.
• 01:14 Again, this process must be managed well so that the results are meaningful.
• 01:19 Another reason we need careful planning is that the study often involves multiple
• 01:24 functions or organizations.
• 01:25 You may need special materials from suppliers, or
• 01:28 specially designed and built test samples.
• 01:31 You may want operations to conduct experiments to remove
• 01:34 one source of variation.
• 01:36 All this will require a coordination and that means project planning and
• 01:40 project management.
• 01:41 One other point is the timing of the study.
• 01:44 If this study is supporting product development of a new product and
• 01:48 not process improvement for an existing product or
• 01:50 process, then the study needs to be done before the design freeze.
• 01:54 The study will determine which parameters or factors are critical to product
• 01:58 performance and the best settings for those parameters.
• 02:01 Learning these after the design work is done and the design is frozen,
• 02:05 will delay the development and create extra costs.
• 02:08 Determining it before the design is frozen, will enhance the ability of
• 02:12 the development team to create a successful new product.
• 02:15 That brings us to a critical aspect of any DOE study and that is the study objective.
• 02:21 A DOE study should have a clear, measurable objective.
• 02:24 This will lead to the selection of the type of DOE, the factors to be studied,
• 02:29 and the communication and application of the study results.
• 02:32 DOE studies can have many different purposes.
• 02:35 They may be used to understand the implication of design trade-offs,
• 02:39 to characterize the design space of new technologies or new products, or
• 02:44 to determine the limits of reliable performance.
• 02:47 The structure of the study will depend upon the objective, so
• 02:50 make sure that that is clear before you start.
• 02:53 Let me give you a few examples of what I mean by an objective.
• 02:56 These are all paraphrases of study objectives I have used in the past:
• 03:01 Improve system accuracy and efficiency by 10%.
• 03:04 Notice that you can have multiple measurements in the objective.
• 03:08 Improve the yield of soldering process by 20%.
• 03:11 Notice, the objective focuses the aspects of the study,
• 03:15 which will be helpful for factor selection.
• 03:18 Achieve 99.5% first pass yield on final acceptance testing.
• 03:23 Notice, the objective sets a target for performance that will be useful for
• 03:27 describing the trade-off among factors, when translating the statistical
• 03:32 results into a practical process and product design and performance parameters.
• 03:37 In each case, the objective is used to help design study, so that it can answer
• 03:42 questions about the impacts of various factors on product or process performance.
• 03:47 And as with any study that requires time and money,
• 03:50 you need to tie your objective to a clear business goal.
• 03:53 That explains why you are expanding your resources.
• 03:55 If there is no business goal that is affected, then there is no reason for
• 03:59 the study.
• 04:00 Let me foreshadow some of the questions that must be answered based upon
• 04:04 the objective.
• 04:05 These questions will impact the study design, which means the time and
• 04:09 resources needed to complete the project.
• 04:11 With a clear objective, these questions are easy to answer.
• 04:14 When the objective is vague or contradictory, the answers are unclear.
• 04:18 First, what is the goal?
• 04:20 This will help to decide the type of DOE study and the factor selection.
• 04:25 Next, how many factors?
• 04:27 There are really two questions here, how many potential factors and
• 04:31 how many can we afford to study in depth.
• 04:33 Again, the answer will determine the type of study and the factor selection.
• 04:37 What is the response factor to be measured?
• 04:40 This is usually the way that the study goal is described.
• 04:43 It must be directly related to the objective, for the study to be successful.
• 04:48 This is the performance measure or measures that are recorded after each
• 04:52 experimental run and used in the statistical analysis.
• 04:55 Now, the remaining questions will get into the study design and
• 04:59 we'll understand the answers to these,
• 05:02 as we explain the details of planning conducting a study in upcoming lessons.
• 05:06 I mention them now because the answers to these will fundamentally change
• 05:11 the structure, planning and monitoring of the study.
• 05:14 The first deals with the type of study.
• 05:17 Full or fractional factorial.
• 05:19 The next is replicates, this means to double or
• 05:22 triple the number of test runs in the study.
• 05:25 Obviously it cost more, but it improves the accuracy of the statistics.
• 05:29 The use of center points.
• 05:30 This is a DOE study calibration technique.
• 05:33 If the study will be conducted over a long time period, the importance of these
• 05:37 points is increased, to ensure that the test conditions are not drifting.
• 05:41 Blocks, again this is a study design question that takes into account the real
• 05:46 world limitations and constraints.
• 05:48 We will discuss this more in detail in another lesson and
• 05:51 provide some ground rules for when to use blocking.
• 05:54 The answers to these last few questions will have a major
• 05:57 effect on the DOE study design and therefore on cost and schedule.
• 06:01 The study objective and real world constraints will dictate the answers.
• 06:06 A foundation for successful DOE study is a clear objective,
• 06:11 up front planning and effective project management of the plan.
• 06:16 These success criteria apply to any DOE study,
• 06:19 regardless of the type of study or breadth of analysis.

Lesson notes are only available for subscribers.