Experimental analysis of dependent-variable manufacturing systems

Modern manufacturing systems, because of their complexity, must often by an alysed experimentally. A prerequisite for performing experimental analysis is that the system variables of interest (factors) are independent. Dependi ng upon the nature of the system of concern and the analysis objectives, ho wever, situations may arise where the system variables are not independent. This paper is concerned with the development of a methodology for developi ng and implementing experimental designs for such systems. The methodology involves specifying a system design procedure such that the algorithms used for solving for the variables correspond to the design factors: the experi mental design then involves varying the design approach (i.e. collection of algorithms) rather than the actual system itself. The methodology is illus trated by way of an example. Besides its utility for manufacturing systems analysis, the methodology also offers promise as the foundation of a new, e xperimental approach to developing manufacturing system design heuristics.