iSixSigma

Mid points in DOE

Six Sigma – iSixSigma Forums Old Forums General Mid points in DOE

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #30770

    Lee
    Member

    Can someone explain to me why we have midpoints in an DOE?

    0
    #80606

    Anonymous
    Participant

    Because we have mid-points in real life.  Really it is because often the optimal settings you get in you DOE will occur not at the high or low value, but in between.  So you can add in midpoints to see where you get optimal settings.  If your equipment is not capable of being set to the midpoint, then you shouldn’t use a DOE design that includes midpoints.  Remeber, DOE is all about optimization, so testing at multiple locations, not just high and low, is generally better.  You can also think of it as getting more samples.  More accurate results come from bigger samples.

    0
    #80609

    Eileen
    Participant

    No problem. In the most common DOE designs (where each factor has two levels), the response is assumed to be linear. A line is drawn between the low level effect and the high level effect for each factor and interaction. Sometimes, there is a question about whether the response may be non-linear (exponential, for example). If there is any doubt, it is wise to run what is frequently called a “center-point” (or as you said a middle point) design. This provides the ability to actually test for curvature or for a quadratic response.
    Eileen, Quality Disciplines
     

    0
    #80618

    Erik L
    Participant

    BB,
    Eileen provided one of the typical uses for the addition of centerpoints to the factorial design.  Some other uses for center points are:

    It provides a way to estimate the MSE term without the use of full replication of the design space to provide an estimate of experimental error.
    It can provide a similar benefit that blocking could provide.  If you have a series of center points run over multiple days, these could become a diagnostic to see if the process response has shifted/drifted over the course of the design.
    Regards,
    Erik

    0
    #80635

    rams
    Participant

    Great discussion guys.
    Same thought as Eileen, I recommend adding midpoints to test for non-linearity. Should there be evidence that the effect is non-linear, then just add the “axial” points to complete a central composite design.
    rams

    0
Viewing 5 posts - 1 through 5 (of 5 total)

The forum ‘General’ is closed to new topics and replies.