Array Selector Image Map

Background

    Dr. Genichi Taguchi's approach to finding which factors effect a product in a Design of Experiments can dramatically reduce the number of trails required to gather necessary data.  An orthogonal array selector can assist in determining how many trials are necessary, and the factor levels for each parameter in each trial.  A parameter is an independent variable that may influence the final product, whereas a level is a distinction within that parameter.  To use this array selector, make sure the number of levels is the same for each parameter.  Note that this testing method does not compensate for correlations between parameters, but assumes each parameter is independent.
 

Example of how to use Orthogonal Arrays and the Array Selector

Step 1 -- Problem definition and separation of parameters

    Parameters effecting the shelf life for a food product may include baking temperature, percent by volume of additive X, and ambient air humidity.  Levels within the baking temperature parameter could be 320, 325, 330, and 335 degrees F.  Levels for the additive percentage could be 0.25%, 0.30%, 0.35%, and 0.40%.  And the four levels of the ambient air humidity could be 30%, 35%, 40%, and 45%.  Notice four levels exist for each of the three parameters.

Step 2 -- Using the array selector to find necessary testing sequence

    Go to the Taguchi Orthogonal Array Selector and click on the area that corresponds to four levels and three parameters (P=3, L=4).  The "L'16 Orthogonal Array" table that now appears lists your testing sequence.  The sixteen rows on the chart correspond to the sixteen required trials for your experiment.  Notice that this chart has five columns representing five distinct parameters.  Since our experiment has only three parameters, we will ignore the right-most two columns.

Step 3 -- Perform testing according to array from Step 2

    Set the parameter that is the hardest to change on the far left column; the parameters easier to change should be on the right.  For our example, humidity is not easy to change, so since that parameter is in the far left column, we can test a number of these without changing this parameter.  In each row, under each parameter, is a number signifying which level the parameter will be tested.  For example, row seven lists the numbers 2, 3, 4 for humidity, additive %, and temperature, respectively, so we will perform the test at 35% humidity, 0.35% additive X, an 335 degrees F.  After all data points are taken, statistical tools can be used to analyze the data to determine which level is optimal for each parameter.