Transcript Nominal the best case analysis using Minitab

Nominal the best case analysis using
Minitab
Slide 1
Example of a Static Taguchi design (Nominal
the best : seal strength case)
Suppose you are an engineer and need to evaluate the factors that
affect the seal strength of plastic bags used to ship your product.
You have identified three controllable factors (Temperature,
Pressure, and Thickness) and two noise conditions (Noise 1 and
Noise 2) that may affect seal strength. You want to ensure that seal
strength meets specifications. If the seal is too weak, it may break,
contaminating the product and resulting in returns. If the seal is too
strong, customers may have difficulty opening the bag. The target
specification is 18.
1
2
3
4
5
Open the worksheet SEAL.MTW. The design and response data
have been saved for you.
Choose Stat > DOE > Taguchi > Analyze Taguchi Design.
In Response data are in, enter Noise1 Noise2.
Click Graphs. Under Generate plots of main effects and selected
interactions for, check Standard deviations. Click OK.
Click Tables. Under Display response tables for, check Standard
deviations. Click OK in each dialog box.
Slide 2
Slide 3
Hasil perhitungan SN Ratio
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Interpreting the Results - Static
Taguchi Design

The response tables show the average of the selected
characteristic for each level of the factors. The response
tables include ranks based on Delta statistics, which compare
the relative magnitude of effects. The Delta statistic is the
highest average for each factor minus the lowest average for
each factor. Ranks are assigned based on Delta values; rank
1 is assigned to the highest Delta value, rank 2 to the second
highest Delta value, and so on. The main effects plot provides
a graph of the averages in the response table.

Look at the response tables and main effects plots for the
signal-to-noise (S/N) ratios to see which factors have the
greatest effect on S/N ratio, which in this example is nominalis-best.
Slide 9

In this example, the factor with the biggest impact on the S/N
ratio is Pressure (Delta = 8.29, Rank = 1). If you look at the
response tables and main effects plot for S/N ratio, you can
see that Pressure 36 and Pressure 40 have almost the same
average S/N ratio (30.2117 and 30.1406).

Here, the response table and main effects plots for mean both
show that the factor with the greatest effect on the mean is
Temperature (Delta = 1.95, Rank = 1). The response table
and main effects plots for standard deviation both show that
the factor with the greatest effect on the standard deviation is
Pressure (Delta = 0.94, Rank = 1). Next, you may want to use
Predict Results to see how different factor settings affect S/N
ratios and response characteristics - see Example of
predicting results.

© All Rights Reserved. 2000 Minitab, Inc.
Slide 10
We will now predict results for the seal strength
experiment. You had identified three controllable
factors that you thought would influence seal
strength: Temperature, Pressure, and Thickness.
Because you first want to maximize the signal-tonoise (S/N) ratio, you chose factor settings that
increase S/N ratios: Temperature 60, Pressure 36,
and Thickness 1.25.
Slide 11
1 The design and response information has been saved
for you.
2 Choose Stat > DOE > Taguchi > Predict Results.
3 Click Levels.
4 Under Method of specifying new factor levels, choose
Select levels from a list.
5 Under Levels, click in the first row and choose the factor
level according to the table below. Then, use the key to
move down the column and choose the remaining factor
levels according to the table below.
Factor
Temperature
Pressure
Thickness
Levels
60
36
1.25
Slide 12
Slide 13
Interpreting the Results Predicting Results


The predicted results for the chosen factor settings
are: S/N ratio of 33.8551, mean of 17.5889, and
standard deviation of 0.439978. Next, you might run
an experiment using these factor settings to test the
accuracy of the model
The predicted values for the standard deviation and
log of the standard deviation use different models of
the data.
Slide 14

Source : Minitab reference Manual
Slide 15