tailieunhanh - Ebook The management of quality and control: Part 2

(BQ) Part 2 book "The management of quality and control" has contents: Experimental and robust design, strategie issues, producer-supplier relationships and the economies of quality, the control of quality in a temporal setting. | CHAPTER 7 Experimental and robust design Introduction The complexity of business problems, organizations, operational and service systems, the number of variables they involve, as weIl as the often chaotic environment to which they are subjected, make it difficult to use prior knowledge (in the form of mathematical models, for example) to construct and calibrate these systems. In these cases, experimentation is an important approach to generate knowledge which can be used for effective analysis and decision making. When a product is put to use, the number of intervening variables may be too large, some of which mayaiso be uncontrollable. Further, experiments are usually costly; there may be many variables and potentially a great deal of experimental variation and errors, making the experimental results obtained difficult to compare and analyse in a statistically acceptable manner. For such situations, experimental design, when it is properly used, provides a set of consistent procedures and principles for collecting data so that an estimate of relationships between one set of variables, called explanatory variables, and another, called dependent variables, can be performed (even if there are experimental errors). For example, we might seek to build a relationship between supply delay (the dependent variable) and a number of explanatory variables such as the number of transport trucks (which can be controlled), weather conditions and traffic intensity (which cannot be controlled). When variables can be controlled, this can be used to reduce the amount of experimental variation. In other cases, selection of the levels associated with these variables might be desired and valued in terms of some objective function. The selection of variables' levels is a design problem which we will consider at the end ofthis chapter. Both the experimental and design problems are extremely important and useful. For example, to test a production process in a factory, it might be .