When can we apply DOE (Design of Experiments)?

P

Priss80

When can we apply DOE? During process optimization at initial set up, process improvement or when there is quality issue and we do have solution and need to run DOE to stream down the possibility? :thanx:
 
M

mwohlg

In my experience, DOE is best run at initial set up. If you wait until you have a quality issue, a) you will already have a quality issue on your hands, and all it's associated costs and problems:bonk: and b) depending on your customer it may be too late to change the process from that which was already tested/validated/approved (i.e. your customer may not want to spend $ and time retesting a modified process)

Optimize your process from the beginning where it makes the most sense.
 

Bev D

Heretical Statistician
Leader
Super Moderator
DoE can be appropriate at all of the stated opportunities and more.
 
P

Priss80

I heard the term process optimization. Does the process optimization similar to DOE? What is the method to conduct the process optimization and DOE?:thanx:
 

Bev D

Heretical Statistician
Leader
Super Moderator
DoE is a term used to describe a large set of statistically valid experimental structures or 'designs'.

Process optimization typically involves a class of designs known as screening designs that are used to determine the few critical inputs that control the output(s) of the process. Once the critical inputs are determined the inputs effect on the output(s) are modeled using various design structures collectively referred to as response surface methods.

As a starting point I recommend an article by Davis Balestracci entitled "Using DoE as a Process Road Map" from the February 2006 issue of Quality Progress. (Clicking on the title will take you directly to the article...)

Three great primers on DoE and its various uses are:
"Quality Improvement Through Planned Experimentation" by Moen, Nolan and Provost

"Understanding Industrial Experimentation" by Dr. Donald Wheeler

"Statistical Engineering" by Steiner and MacKay

There is also a decent manual on Response Surface Methods: "RSM Simplified" by Anderson and Whitcomb
 
G

guruprasadb

In general, use DOE when you want toIdentify and quantify the impact of the vital few Xs on your process outputDescribe the relationship between Xs and a Y with a mathematical modelDetermine the best configuration
 

reynald

Quite Involved in Discussions
What I teach and practice is that DoE can be used with the following objectives:
1. Screening - what factors have significant effect?
2. Characterization - what is the relative contribution of each factor, and which factors interact?
3. Searching and Optimization - On which direction should I change the factor to be able to realize the changes that I want to get.

Screening can be used during the set-up stage
Characterization can be used during the early up to mass-pro stage
Searching and Optimization can be used for matured/stable stages.

Bottom Line: You can use DoE for any stages but the objective (and thus the design of the experiment) will change.

Regards,
Reynald
 
Y

yprasanna

Hi,

How to find the most effective input parameter on system response. Thanks in advance.
 

Miner

Forum Moderator
Leader
Admin
It is always helpful to provide more context around a question. Your question is very general, and could be answered by the post by reynald immediately prior to yours.
 
Top Bottom