Narrow Results

The increasing customer awareness and global competition have forced manufacturers to capture the entire life cycle issues during product design and development stage. The thorough understanding of product behavior (degradation process) and various uncertainties associated with product performance is paramount to produce reliable and robust design. This paper proposes a multi-objective framework for reliability-based robust design optimization, which captures degradation behavior of quality characteristics to provide optimal design parameters. The objective function of the multi-objective optimization problem is defined as quality loss function considering both desirable and undesirable deviations between target values and the actual results. The degradation behavior is captured by using empirical model to estimate amount of degradation accumulated in time t. The applicability of the proposed methodology is demonstrated by considering a leaf spring design problem.

Tolerance design technique balances the expected quality loss due to variations of the system performance and the cost due to controlling these variations. Measures of quality are discussed and quality loss function is used for tolerance design. The goal is to minimize the total loss that consists of the quality loss to the customer and the cost increase to the producer. The design methodologies are presented for the tolerances of products before shipping to the customer and the tolerances of lower-level characteristics. The approaches to tolerance design for components and subsystems are also demonstrated using the variation transfer function. Examples are given as illustrations of the methodology.

This paper presents a set of related quasiconvex quality loss functions. Characteristics of quasiconvex functions that are desirable for modeling quality loss are noted. Three frequently used univariate quasiconvex quality loss functions are discussed. Bivariate and multivariate quasiconvex quality loss functions are developed. A set of necessary and sufficient conditions is established for the quasiconvexity of multivariate quality loss functions. An industrial product example is used to illustrate the development of a bivariate quadratic quality loss function.

All production processes produce variance around the desired target value of quality characteristic. This variance affects the product quality level. Accordingly variance reduction needs to be done as the main goal of quality improvement programs. However effort to improve quality of each product unit must take into account to improvement costs.

This paper proposes an optimization model for quality improvement in multi-stage processes using a non linear programming model by selecting alternatives process and determining unit of production of each stage to maximize profit as the difference between total income and total relevant cost. Total cost includes manufacturing cost, quality loss cost, rework and scrap cost, and quality improvement implementation cost. This optimization model is implemented in make-to-order manufacturer that produces crimper (a parts of joining plastic packages in packaging machine) which consist of five main stage manufacturing processes. Sensitivity analysis shows that the optimal solution is not sensitive if little changes occur in the constraints scenario. Thus, adding the value constraint on the quality specification, stage capacity, and quality improvement budget will not improve the objective function.

A new method for parameter design, where the cost is influenced remarkably by the parameter central value, had been put forward by me in 1996. In this paper, the quality loss function is improved based on it so that a synthetic method, in which parameter design and tolerance design can be made in progress at the same time, can be constructed. The synthetic method can theoretically achieve the minimum quality loss value and it is verified by a calculating case of mechanical design that the synthetic method can obtain a smaller quality loss value than Taguchi's classic method