EXPERIMENTAL DESIGN AND RELIABILITY
One must try by doing the thing; for though you think you know it, you have no certainty until you try
Imagine a biscuit manufacturing process; simplifying it can be divided into two basic phases: dough preparation and baking. Each involves a large number of variables under our control, from the proportion of each mass component, to temperature, time and moisture during cooking. Design of experiments is a methodology, rather than a tool, aimed at learning how each of these factors affects the cookies’ characteristics of interest: color, hardness, degree of crispness, etc.
We are therefore faced with a learning method. It is often surprising how little is known about the behaviour of industrial processes and products. For questions such as: What happens to the hardness of the biscuits if we increase the temperature of the oven? Will it affect some other characteristic? Frequently there are as many answers as technicians consulted. Naturally, the situation gets worse when the question gets more complicated: What happens to the hardness when the oven temperature is increased and the proportion of butter and the cooking time are decreased?
By asking yourself similar questions about processes you know about, you will probably realize that your knowledge about these, in theory, well-known processes is weak. In most cases the answers to the questions will be unknown or vague and not quantified. Rarely will the answer be like, ‘if ...