Degree

Doctor of Philosophy (PhD)

Department

MIE

Document Type

Dissertation

Abstract

The semiconductor industry is one of the most advanced manufacturing sectors, operating at an unparalleled scale and complexity. A new manufacturing facility costs between $500 million and $15 billion, and the industry's investment in automation is unmatched. However, the intricate and cyclical nature of semiconductor manufacturing creates very complex manufacturing facilities where the boundaries between planning and real-time operations often blur.

For the industry to keep improving, there is a need to create a new method that allows comparison among different manufacturing environments to uncover hidden losses that may be obscured by automation.

To better understand these hidden losses, the relationship between soft dedication and preventative maintenance failures to manufacturing performance needs to be investigated. In order to conduct a meaningful analysis of different levels of automation of semiconductor facilities, a new manufacturing performance scale needs to be created.

Operational data from six high-volume semiconductor manufacturing facilities was collected to develop a T-Score, a common denominator. The T-Score provides a means to compare facilities. The research suggests significant correlations between soft dedication, preventative maintenance losses, and T-score, thereby shedding light on crucial aspects of manufacturing performance that are not easily visible.

The research was arranged in three phases. The first phase focused on developing a wholistic manufacturing scoring system to compare facilities with different technology, utilization, and product mix levels. The second phase investigated the effects of soft dedication on manufacturing, and the third phase evaluated preventative maintenance event failures. This research aimed to explore the correlation between the identified losses and manufacturing performance and to alleviate the effects of the two by understanding their influence on production capacity and cycle time to allow facilities to include those in their planning algorithms. The research shows simple ways to capture losses and suggests pragmatic management procedures to incorporate into existing information systems.

The data collected from six US and Asian manufacturing facilities allowed the identification of the losses for soft dedication and preventative maintenance. Comparing and correlating these losses to the facility performance levels was possible using the developed T-Score to normalize these facilities. The T-Score was derived from industry standard throughput, cycle time, and utilization, resulting in a performance metric on an easy-to-understand scale.

The statistical analysis showed that many of the usual industry rules of thumb did not correlate well with manufacturing performance. There was a good correlation between hidden single paths with R2=0.84, ease of flow with R2=0.84, and second-of-a-kind tools with R2=0.54 to Facility performance. On the maintenance side, the study suggests a good correlation between first-time right quals with R2=0.84, short preventative maintenance survival with R2=0.96, and absolute survival relative with R2=0.82 to manufacturing performance.

Date

8-22-2024

Committee Chair

Nahmens Isabelina

Download

Share

COinS