iSixSigma

Output that become Input

Six Sigma – iSixSigma Forums Old Forums General Output that become Input

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #29361

    Ovidiu Contras
    Participant

    Hi everybody ,
    I have a dilemma concerning the following :
    DoE’s are very helpful in determining the relationship between inputs and outputs .What happens if the output becomes input ? How should this case be treated ?
    Thanks

    0
    #75027

    Jack Stuart
    Participant

    I would like more specifics, but there are some things you should do under all circumstances.  1. Drop the input hiding as an output(IHO) out of the DOE.  The math is impossible.  If the IHO is a factor(x) in your Y=f(x) equation and it could be a Y also, then your equation is too big and you need to break it down into smaller experiments.  It could be that your IHO is a quadratic or higher order factor. That is why we need more detail.  2. Find out if your mislabelling of the IHO is due to not understanding the process or a case of mistaken design.  If it is the first case, walk the process, process map the thing  and talk to the operators about the “Factory Physics”, you are probably too far away from the real process.  If it is the latter case, be more rigorous in your design, “Measure twice and cut once.”

    0
    #75028

    Ovidiu Contras
    Participant

    Thanks ,
    For more clarification : this process generates heat . Heat translates to temperature ,and temperature affects the process …..

    0
    #75032

    Jack Stuart
    Participant

    Then you can go three ways.  1.) Leaving Heat as the output.  Take it out of the input side and let the factors that create heat be the inputs. This will allow you to see what your root causes’ equation looks like.  This is good if Heat is a true output (What your customer will pay for) like in a steam generating plant vs heat as wasted friction byproduct like in a car engine.  2.) Leaving Heat as an input.  This will show your non-linear relationship to your final output (Provided it isn’t THE final output. A simple regression study may allow you to characterize Heat in your process without an expensive DOE.  3.) Do both, layer your DOE designs so you examine Heat as input and output.  Taguchi designs with inner and outer arrays might work here also.

    0
    #75033

    Ovidiu Contras
    Participant

    Thanks Jack ,
    situations like this are not obvious….

    0
    #75268

    Manee
    Participant

    With Taguchi, an indirect signal can work.  But you should contact real Taguchi expert such as Dr. Taguchi himself or Madhav Phadke
    Manee

    0
Viewing 6 posts - 1 through 6 (of 6 total)

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