Taguchi DOE Analysis - What S/N ratio (Nominal is best or smaller is better) to use?

[tahirawan11]

Hi,

What S/N ratio (Nominal is best or smaller is better) I need to use if I want to reduce the variation in the response variable caused by the noise factors in Taguchi design experiment.

thanks

 

[Miner]

Re: Taguchi DOE Analysis - What S/N ratio (Nominal is best or smaller is better) to u

It depends on what you want to do with the mean value. If you want to move the average to a target, use Nominal the Best. If you want to reduce the average use Smaller is Better. Both include minimizing the variation, but they do it differently depending on how you want to handle the mean.

 

[tahirawan11]

Re: Taguchi DOE Analysis - What S/N ratio (Nominal is best or smaller is better) to u

Thanks Miner and in my case I need to reduce the average value so i will use 'Smaller the better'. And when I am doing the analysis; first I will look at the S/N ratios to identify the factors which have the highest effect on S/N ratios and then I will tune these factors to levels (low or high) which gives me the lowest Mean value. In this way I will reduce my mean value while minimizing the variation. Am i correct ?

 

[Miner]

Re: Taguchi DOE Analysis - What S/N ratio (Nominal is best or smaller is better) to u

That is correct.

Personally, I prefer to analyze the mean and standard deviation separately and use a tool such as Minitab's Response Optimizer to jointly optimize the mean and standard deviation. You have a better understanding of the process than using S/N ratios, but the Taguchi approach does work.

 

[tahirawan11]

Re: Taguchi DOE Analysis - What S/N ratio (Nominal is best or smaller is better) to u

Thanks but i guess I cannot use the Minitab's Response Optimizer when analysing the exeriment as Taguchi experiment. I need to setup a factorial DOE to use the Response Optimizer or is there a way to use the Response optimizer when i have ran the experiment as Taguchi L8 design?

 

[Miner]

Re: Taguchi DOE Analysis - What S/N ratio (Nominal is best or smaller is better) to u

If your Taguchi orthogonal array has 2 levels for all factors, you should be able to use the Define Custom Factorial Design feature to convert it into a 2-level factorial design whereupon you can use Response Optimizer.