FabTime Cycle Time Management Newsletter Abstracts

FabTime Cycle Time Management Newsletter Abstracts - Volume 12 (6 issues)


Variability Metrics for Fabs: Part 1 (Issue 12.06)

In this issue, we have an announcement about a change to our tip of the month email list (a separate subscription from the newsletter, for customers). Our FabTime tip of the month is about identifying top causes of equipment downtime. In our subscriber discussion forum we have two responses to last month’s question about capacity planning for cascading tools.

In our main article this month, we focus on metrics for fab variability. This article was inspired by informal discussions with several people at the November Fab Owner’s Association meeting in Austin, Texas. These discussions encouraged us to consider whether we are providing the best toolkit that we can in FabTime in terms of fab variability metrics. We review the sources of variability in fabs, and our current approach for tracking fab variability, and propose a brief variability sources snapshot report. We seek our subscribers’ feedback regarding other metrics that should be added to this fab variability toolkit.



Using OEE to Enhance Factory Performance (Issue 12.05)

In this issue, we have community announcements about the upcoming Fab Owners Association meeting at Spansion, and a call for editors for the International Journal of Production Research. Our FabTime software tip of the month is about setting default filters for charts. This month’s subscriber discussion forum includes several responses sparked by the main topic of the last issue, PM Scheduling. We also have a new question about capacity analysis for cascading tools.

Our main article this month is about using OEE to enhance fab performance. Recently, in response to a suggestion from one of our customer sites, FabTime changed the method by we calculate OEE (Overall Equipment Effectiveness) Loss Factors. Several of our customers were interested in the details of not only the equations used; but also the methodology of using OEE to improve operations. In this article we discuss the definition and calculation of OEE, introduce FabTime’s current methodology for calculating OEE Loss Metrics, and review how to properly use the information provided by OEE to continuously improve an organization’s manufacturing capacity.



PM Scheduling and Cycle Time (Issue 12.04)

We have a community announcement about two new FabTime employees in this issue. Our FabTime user tip of the month is about setting a default home page tab for login. In our subscriber discussion forum we have two responses to last month’s article about queueing models for wafer fabs, as well as a new question about measuring coefficient of variation for effective process times.

Our main article this month is about PM scheduling. Equipment downtime in general is one of the top contributors to fab cycle time. Scheduled downtime, and more specifically preventive maintenance, contributes to fab variability, but is somewhat controllable. It’s possible to take the cycle time impact into account when deciding whether or not to group maintenance events, and thus minimize the impact of the scheduled maintenance. In this article, we discuss ways to do that.



Queueing Models for Wafer Fabs (Issue 12.03)

In this issue, we begin with a call for papers for the ISMI Symposium on Manufacturing Effectiveness. Our FabTime user tip of the month is about using a PowerPoint add-in to display live FabTime charts (mixed with other content) on monitors. In our subscriber discussion forum we have inputs on analyzing staffing productivity, embracing the downturn, and scheduling in the lithography area.

In our main article this month we discuss the application of queueing models to wafer fabs. We begin by outlining the benefits and drawbacks of queueing models (as compared with static models and with simulation). We then discuss toolgroup-level models, as implemented in FabTime's operating curve spreadsheet, as well as different approaches for constructing fab-level models. We conclude by discussing the simplified approach of using aggregated fab-level inputs in a simple G/G/c queueing model, and where this approach might, and might not, be useful. If any readers would care to share their experiences in applying queueing models to fab planning or operations, we will post those in a followup article. We welcome your feedback.


Ten Fab Management Discussion Topics (Issue 12.02)

In this issue we also have two calls for papers for conferences. Our FabTime user tip of the month is about ways to export full chart datasets to Excel. In our subscriber discussion forum we have two responses to last month’s question about managing in the presence of multiple constraints, and a follow-up from Bob Kotcher to last month’s main article about confidence intervals vs. prediction intervals.

In our main article this month, we have provided a forum for re-introducing a number of previously raised subscriber discussion topics. Our hope is that some of you will find that you have something to say on one or more of these topics, so that we can all learn from one another as a community. We welcome your feedback.


Prediction Intervals vs. Confidence Intervals (Issue 12.01)

In this issue we have three announcements, one about a survey from WWK, another with a call for papers, and the third about staying in touch with FabTime via my LinkedIn profile. Our software tip of the month is about using the new lot line yield charts in FabTime. We only have one subscriber discussion question, but it is quite detailed (about fab management in a multi-constraint environment).

In our main article this month, we address the difference between confidence intervals and prediction intervals. Both can be applied to simulated or actual recorded data, anything where you have repeated, variable observations (cycle times, WIP, etc.). Confidence intervals are used to estimate an underlying value that can’t be directly observed, like the “true” mean cycle time for a product line. Prediction intervals, instead, are used to establish a range in which it is likely that a future observation will occur, given a series of past observations. So, for example, you might use a prediction interval to predict the upper and lower bound of expected fab throughput next week. We hope that you’ll find this discussion useful.