Investigating Robustness to Variation for Increased Net Output in Production Flow

Abstract

In modern industrial manufacturing processes, a deep understanding of process variation and the system’s inherent robustness is crucial for maintaining high product quality and at the same time improving the process performance. This master thesis project which has been performed in collaboration with SKF Sverige AB aims to systematically analyze and identify the factors that affect robustness, in one of the company’s production flows for producing rollers. The primary objective of the project is to enhance machine utilization and solve a specific cycle time problem within the existing production line, with a clear requirement that the optimization performed should not result in sub-optimization in other parts of the system. In order to approach the problem systematically and with a data-driven approach, the study is structured around Six Sigma, and its proven DMAIC- framework (Define, Measure, Analyze, Improve, Control). In the project’s early phase, a combination of qualitative and quantitative tools were applied, including Affinity-Interrelationship Method (AIM), detailed process mapping, and Cause & Effect- Matrix. These methods were applied to systematically break down the problem, filter out irrelevant variables, and determine the current state of knowledge. Following this, the reliability of the current measuring equipment was assessed using a Measurement System Analysis (MSA) to evaluate whether the existing system was to actually detect the upcoming variation. To then isolate and analyze how key parameters such as tool wear, cycle time variations, and the choice of machine affect the process’s outcome, Design of Experiments were applied. The experimental design was then complemented by Bayesian optimization to find the optimal process parameters, in an efficient way. The results from the statistical analyses highlight the measurement equipment’s actual ability to detect real product variation. Furthermore, the study identifies which of the tested parameters have a statistically significant impact on the process outcome, and separates them from what can be defined as process noise. Based on these insights, the study serves as the foundation for improvements that will enhance the resource efficiency in the bottleneck machine and increase the net output of the production flow.