Workflow variability modelling in semiconductor manufacturing

In the context of Industry 4.0 and the More than Moore’s paradigm, delivery precision and short cycle times are essential to the competitiveness of High Mix Low Volume semiconductor manufacturing and future industries in general. So called “variability” however creates uncontrolled and unpredictable “traffic-jams” in manufacturing systems, increasing cycle times and decreasing the systems’ tractability. This research, a CIFRE PhD between the GSCOP laboratory and STMicroelectronics, addresses this issue of variability in complex manufacturing environment. We first conducted, in the first part of the manuscript, an in-depth study of “variability”: we approached the notion through its consequences in manufacturing systems, clarified that the variability was about the workflow, introducing the notion of workflow variability and measures that come with it, and identified the main sources of variability through a literature review and real-world examples. We focused in the second part of this manuscript on the integration of workflow variability in production management tools: we showed how integrating the stable consequences of workflow variability can improve WIP projections in complex systems and increase the control on such systems, proposed a new tool (the concurrent WIP) to better measure the performances of systems subject to high workflow variability, and showed that complex “dependency” mechanisms play a key role in workflow variability yet are not integrated in any model. Finally, the third and last part of the manuscript organized perspectives for variability reduction: based on the work of this manuscript, we showed a framework for variability reduction on the short term, and proposed a direction for medium and long term research.