An efficient experiment methodology to investigate product design: an acoustic sounder case study

An innovative experiment methodology is described which enables the investigation, through planned experiment, of component aspects of a manufactured product whose values cannot be economically set to specified values in an experiment. These aspects are called semicontrolled factors and their presence in an investigation means that conventional experiment plans cannot be employed. The proposed method of experimentation integrates the study of the manufacturing system and the product design. It involves obtaining samples of components from production and careful measurement of their relevant features. The new approach is demonstrated through a case study on the design and analysis of an experiment to investigate how seven factors influence the performance of an electro-mechanical product called an acoustic sounder. A computer search is used to select from a batch of available components those sets which, when assembled into products, will enable maximum information to be gained from product testing. To facilitate this extensive search, an exchange algorithm and a genetic algorithm have been employed. These algorithms are described and their performances in searching for a plan are compared for the case study