A Comparative Study of Reliability Screening Predictions in Multifactorial DOE Industrial Trials

This paper examines the reliability of a product in terms of its lifetime, based on the investigation of certain factors within the framework of maintaining the principles of the circular economy and the sustainable design of the experiment. The study aims to investigate the critical factors that could affect the product’s lifespan and long-term efficiency. Initially, a statistical analysis of interactions was achieved through an interaction diagram using the Minitab software and for trials scheduled using the Plackett–Burman planner. Subsequently, the most significant interactions were selected and further analysis was conducted using the screening model, a combination of Pareto chart and the Lenth method to filter out weaker performing controlling factors and their associated two-way interactions. This was followed by a life-data regression analysis, and the results identified statistically strong factors and interactions based on various commonly used reliability distributions. Such findings are important in improving the life extent of a product, which in the specific case was a manufactured industrial-level thermostat, while reducing the risk for early product failure. In conclusion, the study demonstrates the variability and multiplicity in prediction accuracy of the screened effects when considering different reliability distributions in research which employs statistical reliability tools. Such tools are regularly encountered in product design and improvement efforts, aiming to successfully prevent early failures while ameliorating the environmental footprint in production.