Various industries use experimentation to improve products and processes. By manipulating different input factors, a company can come up with better quality products while reducing production costs and time.
Design of experiment (DOE) or experimental design is one strategy that a number of sectors employ. The approach is a viable alternative to other experimentation methods like expert trial-and-error and one factor at a time. DOE tends to be more rigorous than these two options.
It is meant to demonstrate the relationships of the different variables in a system to simplify every stage of the problem-solving process. Experimental design can look easy, but it requires a deep understanding. In most instances where the model doesn’t have the desired effects, implementation is usually missing a few steps.
What is Design of Experiment?
Design of experiment (DOE) is a systematic methodology that identifies the connection between elements affecting a process and the output of the process.
DOE is a branch of applied statistics that evaluates the factors controlling the value of a parameter through conducting, planning, analysing and interpreting controlled tests.
It is a collection and powerful analysis tool used in Six Sigma that is used to manipulate input factors to see their impact on the desired output. Through this approach, you can identify the critical interactions in an experiment. The procedure is conducted under controlled conditions to see how the input and output elements relate to each other.
Sir Ronald A. Fisher developed the method in the 1930s for use in a myriad of experimental situations.
Experimental design example
A simple example of experimental design is an interior design project for a new house. For the place to turn out as desired you have to combine the right wall colours, floors, placement of the furniture, window treatments, light fixtures and appliances, among others. All these are in different shapes and sizes. Every aspect is designed to achieve a specific look.
Now DOE shows that changing one of these factors will affect the whole style, but so will an alteration of a combination of multiple elements at the same time. Therefore, you can study which of the factors have the biggest impact on the outcome.
Benefits of Experimental Design
What value can a company expect from conducting DOEs? For one, it can minimise the time spent on the design and development of new products. By testing the variables in a production process, you can tell what works and what doesn’t. The production can then focus on the most crucial factors that generate the desired output, thus saving time.
Design Of Experiments boosts product reliability and performance. Through experimentation, you can find out how each of the factors relates to one another. This knowledge allows you to balance combination levels to achieve the best version of a particular good.
Experimental design lets you consider design alternatives, system tolerances and materials, among others. The approach provides a deeper understanding of cause-effect relationships. Because you are aware of what you are doing and what you intend to achieve, you can easily tell what is missing. So, you are open to looking at other options besides the standard design parameters.
Service organisations can use DOE to solve excessive variation challenges. Through experimentation, service companies can understand critical process variables.
Steps in the Design of Experiments
When conducting DOE, you have to follow certain procedures sequentially to ensure the success of your analysis. The stages are:
- Define the problems affecting the process
- Establish the objectives
- Brainstorm / Thought Showers
- Design experiments
- Conduct the experiments and gather data
- Evaluate the data
- Interpret the results
- Verify the predicted outcomes
The manufacturing industry has been using the design of experiments for a long time to improve the quality of goods. Any sector that finds the need for DOE can adopt the approach to suit specific circumstances. For successful implementation, users must have some knowledge of statistical methods. The tool also demands effective communication between all the involved parties that understand how it works.
DOE methodology Terms
Blocking: Avoid any unnecessary variations in the process.
Hypothesis testing: Process of drawing conclusions on testing.