Simulation-Based Analytics for Quality Control Decision Support: Pipe Welding Case Study

Abstract

This research aims to enhance industrial pipe welding quality management decision support processes from both an operational and tactical management level by introducing a quantitatively driven analytics approach that allows simulation models to be adjusted by real-time data and measurements. The approach sources and extracts useful information from multirelational data located in both quality management and engineering design systems. The approach implements a Bayesian statistics–based fraction nonconforming estimation to recalibrate and realign models with real-time data, which are generated by actual quality control systems. The approach also develops descriptive and predictive analytical metrics, namely operator quality performance measurements and project quality performance forecasts, for supporting and improving decision-making processes. For practical purposes, a C#-based prototype is deployed to facilitate implementation at an industrial company in Edmonton, Canada. The prototype system was shown to generate accurate and reliable decision metrics in a real time manner and to reduce the data interpretation load of practitioners.