Design of Experiments 

 

 

 

What is Design of Experiments? 

 

Experiment is characterized as the methodical system did under controlled conditions with a specific end goal to find an obscure impact, to test or set up a speculation, or to show a known impact.

At the point when examining a procedure, investigations are regularly used to assess which process inputs have a critical effect on the procedure yield, and what the objective level of those inputs ought to be to accomplish a sought result (yield). Experiments can be composed in a wide range of approaches to gather this data. Design of Experiments (DOE) is likewise alluded to as Designed Experiments or Experimental Design - the greater part of the terms have the same importance. 

 

Benefits of DOE? 

 

- You understand most significant inputs. 

- You understand trivial and vital factors. 

- You find the optimal settings for optimum process outputs. 

- You reduce variability and create a robust process. 

 

 

How to make Design of Experiments? 

 

You first need to have a basical statistical methods and concepts, such as histogram, regression, statistical process control charts, and correlation, sum of squares, degree of freedom, mean square and F test. If you do not know them please review. 

 

Then, you are now ready to understand how to make design of experiments and using sum of squares and distinguishing essential factors and trivial factors. 

 

 

What are components of DOE? 

 

Factors: are the inputs of your output. You can only include controllable factors or converting non-controllable factors to controllable factors into DOE if you have definite time and resources. For instance if one of the factors that you are intending to measure the effect is temperature either you create an ambiance that you can control temperature or wait until the weather is at defined level. 

 

Levels: settings, amount or definite measure of the factors are called levels. For instance, if your factor is temperature and you want to understand the effect of temperature and duration in the oven, on your cake taste: then your factors are temperature and time, and your response is cake taste. The levels are 150 Celcius, 200 Celcius. The levels for time factor might be 40 minutes and 30 minutes. 

 

Response is the output of the experiment. Responses must be defined specifically. For instance, your cake can not be response but you must define it more measurable. Cake taste might be a response but still qualitative response and measurable only subjectively. Cake hardness would be a good response. Responses should be defined by customer directly or indirectly. 

 

 

Recommended Books 

 

- Mark J. Anderson and Patrick J. Whitcomb, DOE Simplified (Productivity, Inc. 2000). ISBN 1-56327-225-3. Recommended 

- George E. P. Box, William G. Hunter and J. Stuart Hunter, Statistics for Experimenters - An Introduction to Design, Data Analysis, and Model Building (John Wiley and Sons, Inc. 1978). ISBN 0-471-09315-7

- Douglas C. Montgomery, Design and Analysis of Experiments (John Wiley & Sons, Inc., 1984) ISBN 0-471-86812-4.

- Genichi Taguchi, Introduction to Quality Engineering - Designing Quality Into Products and Processes (Asian Productivity Organization, 1986). ISBN 92-833-1084-5 

 

 

Recommended Softwares 

- DoE++ : only focusing on DOE

- Minitab: any version is good. 

- SPSS : Any version is good. 

- Statgraphics:   16th version is ok,

                            17th version strongly recommended. 

 

 

 

Recommended Readings 

 

What is design of Experiments?, Keith Bower, ASQ website

 

DOE Tutorial, ASQ