Experimental Design and Taguchi
Key words: Experimental Design, Statistically Designed Experiments, DOE, Optimisation, Robustness, Taguchi Method, multivariable testing.
- Optimisation of processes and formulations
- Reduction in the variation of products and processes ie improved robustness
- Investigation of process variables
- Investigation of noise variables
- Reduced costs and improved efficiency
Experimental Design
Many industrial processes in chemistry and engineering are not optimised or suffer from seemingly random variation. Researchers have found it difficult to make necessary improvements due to the number of tests required. Experimental design offers a method of getting the maximum information from the minimum number of tests. In addition, valuable information can be gained on interactions between variables - interactions are often very important, and may be missed when investigations are carried out changing one variable at a time.
Here are some definitions:
Main effects: The effects of those variables you are able to control
easily, such as machine speed, amount of a component in a formulation, and
stirring rate. These variables are often, but not always, reasonably well
understood.
Interactions: Suppose you have a good understanding of how two
variables affect the quality of your product, but when you make changes to
these variables at the same time the effect is much greater (or less) than that
expected. This is an interaction. For instance, in an adhesive both resin and
hardener need to be changed to see a significant effect, since the levels of
these components interact. Interactions are usually not well understood but can
be very important. Technologists often have a feel for which variables need to
be altered at the same time to get a particular result, and it may be worth quantifying
these interactions.
Noise:
Those variables that are difficult or impossible to control that affect the
process. These are often environmental such as ambient temperature and
humidity, but can include batch-to-batch variation of raw materials and the use
of different operators and machines.
Robust: A process that does not change with changing noise. This
is an important part of the Taguchi technique, and is often investigated using
a parameter design.
Taguchi
The Taguchi approach to quality engineering places a great deal of emphasis on minimising variation as the main means of improving quality. The idea is to design products and processes whose performance is not affected by outside conditions and to build this in during the development and design stage through the use of experimental design. The method includes a set of tables that enable main variables and interactions to be investigated in a minimum number of trials. Though the concept is reasonably straightforward, choosing the most appropriate table and interpreting the results can be more difficult.
Robustness
Of particular importance is the reduction in variability of products and processes. In other words to make products and processes more robust and less susceptible to changes due to outside influences such as raw material variation, temperature, and changes to machines and operators. Improved robustness can often be achieved without major capital expenditure through the use of these techniques.
Do You Need Help?
Do you need any assistance in the design of experiments and the improvement of industrial processes? Areas of expertise include: factorial design, fractional factorials, self-directed optimisations, Taguchi method, robustness, analysis of main effects and interactions, and orthogonal arrays.
Initial advice and quick queries are free, so if you need any help please do get in touch.
Contact: stefan@orszulik.free-online.co.uk
Tel: 00 44 (0) 1235 764208