Statistics for Innovation : Statistical Design of "Continuous" Product Innovation Based on Real and Simulated Experiments

4. 1. 1 ImportanceofComputerSimulation The importance of experimenting for quality improvement and innovation of pr- ucts and processes is now very well known: “experimenting” means to implement signi?cant and intentional changes with the aim of obtaining useful information. In particular, the majority of industrial experiments have two goals: • To quantify the dependence of one or more observable response variables on a group of input factors in the design or the manufacturing of a product, in order to forecast the behavior of the system in a reliable way. • To identify the level settings for the inputs (design parameters) that are capable of optimizing the response. The set of rules that govern experiments for technological improvement in a ph- ical set-up are now comprehensively labeled “DoE. ” In recent years, the use of - perimentation in engineering design has received renewed momentum through the utilization of computer experiments (see Sacks et al. 1989, Santner et al. 2003), which has been steadily growing in the last two decades. These experimentsare run on a computer code implementing a simulation model of a physical system of int- est. This enables us to explore the complex relationships between input and output variables. Themain advantageofthis is that thesystem becomesmore“observable,” since computer runs are generally easier and cheaper than measurements taken in a physical set-up, and the exploration can be carried out more thoroughly. This is particularly attractive in industrial design applications where the goal is system - timization. 4. 1.

The objective of this book is to illustrate statistical methodologies that incorporate physical and numerical experiments and that allow one to schedule and plan technological innovation, similar to any other productive activity. This methodology should be implemented through a structured procedure aimed at reducing the high rate of commercial failure characterizing actual innovation processes. In fact, it is well known that:

i) The rate of commercial failure of a innovative idea is very high (90–94 out of 100 proposals for innovation undergo substantial failure in the EU and in the USA).

ii) Low reliability in the long run and sensitivity to usage conditions are the factors that determine the failure of the innovation.

The definition of an iterative design activity is an objective that can be reached by subdividing the complex innovation process into "short" steps in experimental statistics research. The approach adopted to analyze customer needs and the tools used to reduce unwanted variability form the framework for the statistical design of "continuous" product innovation.

Starting from the observation that product innovation is achieved when a "quality" that is able to satisfy a new customer need is conferred on the product and survives over real operating conditions and time, this book illustrates the operative steps required to perform the whole innovation process iteratively.