Product development

Design of experiments (DOE)

experimental designExperimental design is about designing your experiments in such a way that you get the information you need with minimal effort. It can be used both when developing new products and when improving existing ones.

My experience is that significant resources can be saved by doing smarter experimental design. Some companies perform far more experiments than necessary because they believe it is the safest thing to do, and because "we always do it this way". As an example, I analysed the results of 1100 trials within an experiment. The same information could have been achieved by running only 150 trials. If you start with a few trials based on your assumptions, you can always increase the number of tests based on your experience as you go. However, there is no need to start with too many tests. Lean Tech can assist you with DOE and help you analyse the result.

Component of Variation (CoV)

experimental designWhen the levels of a factor are random, such as operators, days, batches, it can be better to perform a CoV - Component of variation. While the levels in a DOE is controlled, a CoV estimate how much various factors contribute to the overall variation.

In a CoV the components varies naturally, while in a DOE you control the levels and often use more extreme level settings in order to estimate the effect of a factor. CoV can also be used when developing new product and when improving existing ones; How much do the different process steps contribute to the overall variation in product quality? What factors are critical for product quality? How great is the within batch variation compared to the variation between batches?