iSixSigma

DOE

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #53523

    Singaram
    Member

    My project has 2 objectives (outputs) that are to be delivered without compromise. Most actions taken to deliver output 1 is resulting in unsatisfactory output 2 and vice versa. I have about 5 factors that have been tested for impact on output 1 and output 2 independently. Can a DOE be run to understand the factor level or combination of factors that deliver output 1 and output 2 satisfactorily? :cheer:

    0
    #190443

    Anawis
    Participant

    Yes, JMP can do this. You can select multiple Y variables with their goals (maximize, minimize, match target) along with their levels and X variables (continuous or categorical) with their levels, select interactions and create models. Using the Profiler you can then select maximize desirability and you’ll get the optimal levels of each X variable, a predicted estimate of your Y variables and a graph of how far off you are from the goal. It optimizes the Y variables in tandem and you can even weight some of the Y variables more than others when you set it up. Other software packages such as Minitab can do this, but I am most familiar with JMP.

    0
    #190444

    Putnam
    Participant

    The simple answer is going to be “yes”.

    The longer answer is going to need more information about your system, factors used, and level selection. If you can provide the entire data set(s) and your desired end result, you’ll probably get the best response from the group.

    0
    #190446

    MBBinWI
    Participant

    It is possible that the design that you have (the X’s and their relationship equation to the 2 Y’s) cannot be solved to provide an acceptable solution for both Y’s. In this case, you will need a new design (relationship of the X’s or new X’s to the Y’s). If you already have the transfer function equations, you can quickly solve these 2 equations and find the Y responses for the X settings.

    0
Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.