The Effects of Weibull Distribution on Supplier Comparison using Lower Process Capability Index: A Case Study

Abstract

In supplier comparison, durability, one of the quality dimensions, is important criterion. Lifetime data is widely used to measure the durability and commonly modeled with Weibull distribution. Many researchers employ process capability indices (PCIs) for comparing suppliers about quality aspect. However, applying these methods bring to some misleading results when lifetime data are considered because the methods are developed under normal distribution. This study attempts to develop the new method for comparing supplier using lower process capability index (Cpl) to apply with the lifetime data. To be guideline for developing the new method, this research aims to study the effects of Weibull distribution on supplier comparison using Cpl via a case study. The effect is studied through 2 performance measures, e.g., producer’s risk and the power of test obtained by Monte Carlo simulation. The simulation results for Weibull distribution indicate that the process’s shape influences on the performance measures obviously. For right-skewed process, the producer’s risks are lower than  = 0.05, and power of tests do not exceed 0.10. Although power of tests are close to 1.00 in case of left-skewed process, the risks are greater than  = 0.05. For symmetric process, the producer’s risks are close to  = 0.05, but the power of tests do not exceed 0.40 in all cases. So, this study points out that the process’s shape is an important factor affecting to the performance of supplier comparison using Cpl leading to the possibly misleading results. To better understand the effects, this paper presents the example of supplier comparison using real data. The example shows that although the popular methods for handling Weibull distribution are applied, the supplier comparison results are not the same. So, manufacturers should be aware in supplier comparison regarding to lifetime data.
 HIGHLIGHTS
 
 Studying the effects of lifetime data modeled with Weibull distribution on the supplier comparison to be guideline for developing the new method
 Presenting the example of supplier comparison via real data
 For Weibull distributed process, the process’s shape is an important factor affecting to the performance of supplier comparison
 The supplier comparison leads to the possibly misleading results although the process is symmetric
 
 GRAPHICAL ABSTRACT