Locked lesson.

Upgrade

About this lesson

The DOE results can be used by design teams to make wise design decisions. This lesson will address how to use the DOE results in predicting system performance, designing system controls and establishing tolerances on system control and response factors.

Exercise files

Download this lesson’s related exercise files.

DOE in Design Creation.docx
60.4 KB DOE in Design Creation - Solution.docx
60.5 KB

Quick reference

DOE in Design Creation

The DOE methodology develops a design space equation that models a product, process, or system design in terms of the control factors and the response factor. The availability of an accurate model of a new design or new technology will reduce business risk because the design team can predict and therefore avoid zones of poor operation.

When to use

Once you are at a prototype stage in the development of a new product, process or technology a DOE can be conducted. It should be done before design freeze so that the insights from the analysis can be easily incorporated.

Instructions

A DOE analysis during design creation provides a tremendous benefit to the design team. Early in the development of a new product, process, system or technology, there is often very little practical experiential data. A DOE can create a design space equation that the team can use to model and predict performance. This means the team can run analyses to discover zones of excellent performance and zones of poor or even dangerous performance. The DOE will predict which factors are significant and which are not. All of these can help the designers as they establish the new product, process, or system. 

When characterizing the performance of the new product, process or system, the full factorial DOE is the best option for creating a complete and accurate design space equation. The Plackett-Burman DOE can also be very useful if there are many possible important factors and the design team needs to identify which are significant. Later in the design and development process, the fractional factorial DOE can be used by the developers to optimize the design for a particular response factor performance.

A DOE will also provide data that can assist in the decisions for how best to control the new product, process, or system. A DOE can determine if control factors are related to the response factor through a straight line linear manner, or if the relationship is non-linear. If linear the factor could be an excellent control that can be easily calibrated and understood. Also, the DOE main effects plots will illustrate the type of each factor’s relationship – positive, negative, or nil. Through an analysis of the design space equation, the design team can determine if there are elements of the new product, process or system that are very sensitive or fragile. The designers can then choose a control factor to monitor that aspect of the design that is also sensitive in that region, allowing it to detect and control small changes.

DOE data is also very helpful when setting tolerances. Factors that have little or no effect on the key output parameters (i.e. their main effects plot is horizontal) can have wide tolerances with no performance degradation. Then set those tolerances based upon other business considerations. By the same token, those factors with steep main effects plots that are not chosen as user controls should have very tight tolerances to minimize the uncertainty that they would otherwise cause in product, process or system performance.

Hints & tips

  • Once you are able to build physical prototypes of a new product, process or system, you are able to conduct a DOE.
  • The DOE design space equation can be used for analyzing many “what if …” scenarios that will save time and effort in application testing.
Login to download
  • 00:04 Hello, I'm Ray Sheen.
  • 00:06 Well we've discovered how to set up a DOE and how to analyze a DOE.
  • 00:10 Now let's look at how we would practically apply what we have learned in our
  • 00:15 DOE analysis, in particular in this lesson, I wanna look at using a DOE
  • 00:20 to help us understand how to create a new design, or a new technology.
  • 00:27 A DOE analysis can be an integral part of a system design project.
  • 00:31 DOE creates data, and the analysis of that data will explain how a system performs
  • 00:36 and the factors that afffect that performance.
  • 00:39 Depending upon the DOE type you can identify both primary factors and
  • 00:43 interaction effects.
  • 00:45 When creating a brand new system, the DOE will help to identify critical factors and
  • 00:49 system tolerances and create the expectation for normal behavior of that
  • 00:54 new system, and when creating brand new systems or new technologies, there
  • 00:58 will inevitably be something that goes wrong, or at least something unexpected.
  • 01:03 DOE can be very helpful in the problem solving mode.
  • 01:06 It will uncover factors or
  • 01:08 factor interactions that are causing the poor performance.
  • 01:11 In addition, the DOE analysis can give the development team
  • 01:15 confidence in the solution they've selected and
  • 01:18 help to obtain the needed resources to implement it.
  • 01:21 Finally, if the problem is one of tolerancing,
  • 01:23 DOE can help to identify the danger zones for factors, allowing the designers to
  • 01:28 avoid those areas in the factor settings and tolerances.
  • 01:32 Probably the best thing that a DOE will do with a new system or
  • 01:35 new technology is provide characterization of its performance.
  • 01:40 Let's look at the characterization process.
  • 01:42 DOE can explore and characterize the full envelope of system performance.
  • 01:47 The design space equation provides a mathematical model for
  • 01:50 how things work within the system boundaries.
  • 01:53 You can use a Plackett-Burman DOE to quickly and
  • 01:57 easily understand the impact of literally dozens of factors on system performance.
  • 02:02 A full factorial DOE will paint the edges and
  • 02:05 corners of the system providing insight into what happens in those regions.
  • 02:10 You can use a fractional factorial DOE to refine and
  • 02:13 optimize the factor settings to get the ideal performance from the new system.
  • 02:19 And let's go back to the design space equation again.
  • 02:22 The designers can use that to optimize for different design attributes, and
  • 02:26 that allows the business to position a new system accordingly in the marketplace.
  • 02:31 You may want the lowest cost with decent performance, or
  • 02:34 the best performance under a set of real world business constraints.
  • 02:38 By analyzing the equation, the designers can create a robust design and
  • 02:42 avoid potentially hazardous or poor performance zones.
  • 02:46 Now let's look at the characterization of the controls to be used
  • 02:50 in the new system we are creating.
  • 02:52 As we've seen in the analysis lessons, the DOE will identify which factors
  • 02:56 are significant, but beyond that they show us the factor setting
  • 03:00 is positively correlated with performance, negative correlated performance, or
  • 03:03 maybe no correlation, and by using multilevel factors the analysis can
  • 03:08 consider nonlinear performance relationships with those factors.
  • 03:12 We can look at the slope of the lines and the main effect plots and
  • 03:15 get a very good understanding about how sensitive each factor is.
  • 03:20 With multilevel factors we can even find zones of sensitive response or
  • 03:25 zones of virtually no response, and using the Taguchi DOE,
  • 03:29 we can identify system robustness for our manufacturing process.
  • 03:33 Now, remember, Taguchi is based upon expert process knowledge, so you may have
  • 03:37 to delay running this DOE until after your pilot production is complete, and
  • 03:42 we haven't discussed it much, but DOE can be very helpful when setting tolerances.
  • 03:46 And with brand new systems it's often difficult to guess what tolerances should
  • 03:50 be set on the control factors, so
  • 03:53 they generally get set two type just to be safe, which adds unnecessary cost.
  • 03:58 The DOE analysis will identify under what
  • 04:01 factor settings the performance drops below an acceptable level.
  • 04:05 The targeted tolerances for
  • 04:06 that factor can then be set to be well within the safe or optimal zone.
  • 04:11 Another interesting aspect of DOE comes from using the interaction plots
  • 04:16 when looking at tolerances.
  • 04:18 In some cases, the factors work against each other, and
  • 04:21 the interaction of those factors is a key item of significance.
  • 04:25 Knowing this, the tolerancing for the factors, especially if they are adjustable
  • 04:29 by the operator, can be set to avoid performance degradation.
  • 04:33 And back to that Taguchi DOE, you can use the minimum and maximum manufacturing
  • 04:38 tolerances as noise factors that are analyzed in the outer array.
  • 04:42 That will help a business determine whether those tolerances
  • 04:45 need to be changed.
  • 04:47 The difference types of DOE's can provide useful information
  • 04:50 when working with new technologies or new systems.
  • 04:53 Decide where your biggest risk is, and
  • 04:56 then set up a DOE to help you overcome that issue.

Lesson notes are only available for subscribers.

PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.

Gift this course