iSixSigma

Blocking in DOE

Six Sigma – iSixSigma Forums Old Forums General Blocking in DOE

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #46748

    Furneaux
    Member

    Hello,
    I have a question about how blocking numbers are assigned in DOE.  For example, suppose I want to do a three factor two level design with three replicates.  I can have up to 12 blocks.  If I select 12 blocks with the software I am using, I get the blocks assigned as shown below.  The blocks are assoicated with AB, AC and BC.  So, you get the blocks by looking at which signs for these three are the same.  But is there a process for assigning certain runs to Block 1, certains ones to Block 2,etc.  Or does it really matter?
     Std. Run         Block   A         B         C         AB      AC      BC
    4          Block 1            1          -1         -1         -1         -1         1
    19        Block 1            -1         1          1          -1         1          1
    13        Block 2            -1         -1         1          1          -1         -1
    10        Block 2            1          1          -1         1          1          -1
    16        Block 3            1          -1         1          -1         -1         -1
    7          Block 3            -1         1          -1         -1         1          -1
    22        Block 4            1          1          1          1          1          1
    1          Block 4            -1         -1         -1         1          -1         1
    5          Block 5            1          -1         -1         -1         -1         1
    20        Block 5            -1         1          1          -1         1          1
    14        Block 6            -1         -1         1          1          -1         -1
    11        Block 6            1          1          -1         1          1          -1
    8          Block 7            -1         1          -1         -1         1          -1
    17        Block 7            1          -1         1          -1         -1         -1
    2          Block 8            -1         -1         -1         1          -1         1
    23        Block 8            1          1          1          1          1          1
    21        Block 9            -1         1          1          -1         1          1
    6          Block 9            1          -1         -1         -1         -1         1
    15        Block 10          -1         -1         1          1          -1         -1
    12        Block 10          1          1          -1         1          1          -1
    18        Block 11          1          -1         1          -1         -1         -1
    9          Block 11          -1         1          -1         -1         1          -1
    24        Block 12          1          1          1          1          1          1
    3          Block 12          -1         -1         -1         1          -1         1

    0
    #154861

    ezweld
    Participant

    I am really bad about the order I run experiments. I always end up letting the operators get their way and run the order that is easiest for them. One day this is going to bite me in the arse. The order that a program like Minitab picks is ideally the order you should run. The reason is its un-bias the factors are arranged to not build up dynamics. Dynamics that may occur after blocking is bias and would not have shown up if the DOE was not blocked. I was taught to set my most important factor as the first variable and only block for the first variable if possible. I am not sure if that was the consultant’s advice or if he had gotten it from some other source.

    0
    #154894

    BTDT
    Participant

    ezweld:I learned to be very careful using the word, “random,” when describing the run order. We had a potential problem with setting the voltage as one factor on an electrostatic precipitator unit for a power plant in Saudi Arabia. This was used to remove water and salt from the crude oil before feeding the turbines.The runs took quite a few hours to run and we could not complete all the planned runs in a single day. The operators did not want to climb up the machine to change the settings on the transformer more than once, so they interpreted the word, “random,” to mean that they could do all the runs in whatever order they pleased.The temperature, water content and salt concentration of the crude oil changed enough from day to day that it would have been impossible to separate the voltage factor from blocking effects.We changed the way the DOE was to be conducted by giving the operators only the settings they were to use on each day. They tended to run them in an order that was more convenient, but they only ran 2-3 per day anyway – not a serious problem.When MINITAB come up with the randomized run order, it make no different which is the first factor or which is the second, etc.Cheers, BTDT

    0
Viewing 3 posts - 1 through 3 (of 3 total)

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