Design of Experiments (DOE)
What is DOE?
By definition, DOE stands for Design of Experiments, but is often referred to as Designed Experiments, Experimental Design, or even a systematic approach to investigation of a system or process. Design of Experiments is specifically useful in a process analysis where an evaluation of the process input has a significant impact on the process output. This input can be experimented with and optimized to achieve a desired result that is not theoretical knowledge.
All the information required to collect this information on inputs can be designed as experiments and hence this is called Experimental Design. It is a series of structured tests that are introduced into the inputs of a process or system. An assessment is made to study the effects of these changes on a predefined output.
A design that is well designed is as simple as possible and it gets you the required information in a cost effective and optimal way.
A well designed experiment may very well answer questions such as what are the best input attributes when the process or system will deliver the optimum performance, how can the variation in the output be reduced, what are the main components of a process etc
For any analysis of a Experimental design, there are 3 aspects
- Inputs to the process/system or Factors
- Attributes of each input or Levels
- Output of the experiment or Response
Among all the ways of data collection for analysis purposes (viz. historical data, gather new data, run random & specific tests on the system/process and Experimental design in a structured way), the Experimental design is the best because it increases your overall relevance of the tests.
DOE and One-At-a-Time experiments
OAT (or one-change-at-a-time or one factor experiments) are experiments with the most popular experimental method. Here, all inputs are held constant when changing only one. Measure and analyse the output. Repeat this process for all other inputs. But this is not a thorough method to test all possible cases. The interaction among the factors or inputs has not been taken care of. Generally people tend to stop the testing or experimentation after they feel they have got the required result. But DOE does not stop at one experiment. DOE allows a lot of flexibility in experimenting with output of standalone inputs or a combination of inputs.
Steps in Experimental Design process
- Clearly define the problem
- Identify the objectives
- Collect data by brainstorming
- Design the experiment
- Collect the data by conducting the experiments
- Analyze the data and the results
- Compare with the predicted results
Advantages of DOE
While most of our cell phones, computer accessories, and consumer electronics items have gone through well designed experiments, there are quite a few more advantages to DOE.
- DOE saves you a lot of time, energy and increases your ROI.
- You can solve problems with DOE faster than anybody else
- DOE finds a place in all industry verticals including service, manufacturing, software, research, medicine etc
- You can solve problems that cannot be solved by common methods
- Catch the errors sooner in the earlier stages of the product/.service life cycle and thereby improve the cost of quality
- Produce the best quality products or services