A Novel Approach to Analyze Quality Cost Using Hybrid Bayesian Networks

Abstract

Cost of quality (COQ) is an essential element of total quality management (TQM) system. It generally consists of conformance (preventing poor quality), and non-conformance (failure of product and service) costs. The accurate assessment and analysis of such costs can improve managing the quality of products and services significantly. The prevention–appraisal–failure (PAF) model is one of the most widely used models for analyzing and classifying the cost of quality. This approach identifies the different parameters affecting on COQ, which is generally affected by many parameters. However, the PAF model doesn't take into account the uncertainty (risk) involved in these parameters. Moreover, the casual relationships among these parameters and also environmental and qualitative factors, are not properly addressed. This paper aims to offer a probabilistic model to assess COQ by mapping the PAF model to Bayesian networks (BNs). BNs provide a framework for presenting inherent uncertainties, formal use of experts' judgments and probabilistic inference among a variables set. In this approach, first a qualitative model is developed to prioritize PAF groups. Then, in the group with highest priority (i.e. prevention) a quantitative model is presented. The model captures the affecting parameters in more details and provides a probabilistic analysis for qualitative and quantitative factors. The capabilities of the proposed approach are explained using data collected in a chemical products manufacturing company as a case study.