Taguchi's L8 Orthogonal array

[Jigneshpadia2002]

Can anyone please help me to understand how I can design the Taguchi's DOE (Design of Experiment). I am not aware about the rules.

I came across this equation
Ln(2^K)

Where n= number of experiments
2 = levels per factor (i.e. High & Low )
k = number of factors

But how to find N

I have seen some examples form which I am assuming N = K-1

Can anyone please explain me how to design this experiments and what are the fundamental rules?


I am very thankful for any suggestions and help, Thanks a lot in advance !

Jignesh

 

[MikeHalloran]

See if you can find a copy of Bob Pease's piece,

"What's all this Taguchi stuff, anyhow?"



<humble opinion>

There are a couple of valid ideas at the core,

but..

the _program_ is snake oil.

</opinion>



Mike Halloran
Pembroke Pines, FL, USA

 

[m777182]

The L8(2^7)experimental orthogonal array looks like this:

columns
No. exp 1 2 3 4 5 6 7

1 1 1 1 1 1 1 1
2 1 1 1 2 2 2 2
3 1 2 2 1 1 2 1
4 1 2 2 2 2 1 1
5 2 1 2 1 2 1 2
6 2 1 2 2 1 2 1
7 2 2 1 1 2 2 1
8 2 2 1 2 1 1 2

"1" means a factor at lower level and "2" means a the same factor at higher level, e.g. temperature at 1 =20deg and temp. at 2 =50 deg.
No. experiments are runs you have to perform to measure the efect of factor 1 (column 1).
The design for 7 factors need 56 runs which is fairly considerable compared to a full factorial that would need 2^7=128 separate runs.
m777182

 

[nedian83]

I am looking for an array of The L8(2^6)experiment.
Can anyone help me please.
Thanks

 

[m777182]

I have to correct my answer to Jigneshpadia2002:the design for 7 factors needs 8 runs. Each run is done at levels shown in columns 1..7.
m777182

 

[Jigneshpadia2002]

m777: You mentioned that I need to do 7 X 8 runs = 56 runs ! But I have some other information saying 8 runs will explain all the information and I do not need to perform 56 runs !

I hope I do not require 56 runs. What is your take on it.

Can we use the same format for 6 parameters? This is questioe asked by nedian.

My take will be yes?

What do you say ?

 

[m777182]

I have corrected my answer already: you need 8 runs. For six factors I do not have a design table; however it is neither in my Statistica. I 'll try to construct it over the weekend time.
I think that the efficiency of the Taguchii design ceases with less factors.
m777182