Home  /  Courses  /  IOE899 Seminar Schedule

IOE 899: Seminar in Industrial and Operations Engineering
Wed Oct 28, 2009, 4:00-5:00pm, 1680 IOE

Michael Hassoun, University of Michigan
"Identification and cause analysis of WIP bubbles in fabs"

Abstract
We study the phenomenon of work in process (WIP) bubbles in a semiconductor fabrication plant (fab). The operational difficulties associated with inflated work-in-process (WIP) along the complex, highly re-entrant process used in the semiconductor industry represent one of the major roadblocks to the smooth operation of fabs. The cycle time of lots going through a congested fab is tightly correlated with WIP levels by Little's law. Thus, a reduction in cycle time is frequently obtained through an improvement in WIP management. Most semiconductor manufacturers agree that one of the most harmful events in WIP management is what is known as a WIP bubble. However, bubbles remain to be studied as phenomena in their own right. After a bubble has emerged at a certain stage of the production process, typically at a bottleneck station, substantial efforts are needed to reduce the WIP. The bubbles are particularly harmful to the WIP flow and cycle time of the fab processing lines because of the re-entrant nature of the semiconductor production process. As a preliminary step, we formalize the concept of WIP bubbles by decomposing them into local events of relatively acute and temporary WIP congestion. The local bubble is empirically identified, and its impact on local waiting time distribution assessed. We then estimate its marginal impact on overall line waiting time, showing that its effect is substantial. In a second phase, we address the sources of bubbles using a non-volatile (flash) memory fab simulation model. The model performance is studied through statistical data mining methods. The tendency of certain process segments in a fab to undergo bubbles, as defined previously, appears to be predictable over time, based on a set of local characteristics. Finally, we present and studied those characteristics that best facilitate bubble prediction, finding that the utilization patterns of the process segments under analysis appear to be the strongest determinants.
 
Bio
Michael Hassoun is a postdoctoral research fellow at the Electrical Engineering and Computer Science at the University of Michigan. His main research topic is reentrant production systems as found in the semiconductor industry. He earned his PhD in the Industrial Engineering and Management Department at Ben Gurion University of the Negev, Israel and his BSc in Mechanical Engineering from the Technion, Israel. Before his PhD, he worked at the Intel fab 8 plant in Jerusalem as a functional area industrial engineer. His PhD research interest was focused on the bubble phenomenon in semiconductor fabrication plants.

back to top

Topic-Based Site-Wide Navigation

back to top

Search

Audience-Based Site-Wide Navigation:

back to top