ABSTRACT
Parameter design using design of experiments methods is the best way to achieve high reliability at low cost. Since probably over 80 percent of product reliability is determined before production, designers apply much of their expertise to setting parameter values. They use methods such as Failure mode and effects analysis and Fault tree analysis, followed by confirmation experiments for suspected problem areas. The established methods address reliability of manufacturing processes, not technical reliability at the front end. More and more engineers worldwide are using his parameter design to investigate reliability during the design phase. Parameter design can use theoretical or calculated data. Analysis using the signal-to-noise ratio is more comprehensive and objective than conventional variance analysis, and more manageable as well. Compound factors can keep parameter design manageable, even with many control factors and large matrixes. The parameter design establishes the target value for each parameter in as few test runs as possible.
