Implementing a world-class OEE management system
06/01/2003
Overview
Overall equipment effectiveness has become an industry maxim closely associated with successful production. Yet, proper determination of this metric can be done at too high a level, failing to uncover real-life root causes. Achieving comprehensive data requires a four-step closed-loop process that starts with a reporting system, allows ongoing measurement and data mining, and provides for periodic studies.
In recent years, the semiconductor industry has largely adopted the overall equipment effectiveness (OEE) metric as a comprehensive measure for analyzing and reporting equipment performance and assets utilization. Most fabs use the Semi E10 definition of OEE as a key floor shop management metric, while equipment vendors use the indicator to benchmark tool performance and define expected effectiveness level.
As with many other shop-floor indicators, data availability and integrity, along with the question of how to leverage data for corrective and preventive actions (CAPA), are the primary challenges of OEE-based performance management. Most existing information systems and tool monitors capture only high-level OEE losses and do not provide visibility into many of the labor-related OEE inefficiencies, such as "Tool idle," "Tool requires assist," and "Tool waiting for operator" events. The failure to accurately capture this information results in a misrepresentation of true fab performance. Furthermore, even when data are captured accurately, they are often not distilled into actionable information, preventing management from meaningful root-cause analysis, solutions identification, and proactive improvements implementation.
Below we detail a proven methodology for semiconductor facilities to implement a cost-effective OEE management system. This system integrates accurate data with actionable information, providing a true continuous improvement mechanism.
Effective OEE
An effective OEE management system has two main elements: a software solution and a continuous improvement business process. Together, these elements are implemented in a four-step, closed-loop improvement process (Fig. 1) — similar to the six-sigma DMAIC approach of define, measure, analyze, improve, and control — that integrates organizational and operational aspects with user-friendly and easy-to-implement software tools, such as an OEE reporting system and OEE sampling software. This approach has already been effectively adopted by leading semiconductor companies.
 Figure 1. OEE management system's four-step closed-loop process. |
null
OEE reporting system
The first two steps — "OEE measure and report" and "OEE losses data mining" — require the implementation of a friendly, easy-to-use, OEE reporting system. The system, which is Semi E10 compliant, extracts and integrates the required information for determining a tool's OEE directly from a fab's manufacturing execution system (MES), computerized maintenance management system (CMMS), and industrial engineering or capacity management databases.
The OEE reporting system calculates OEE, OEE components, and work in process (WIP) levels, linking tool performance to its required volume of processing. It then presents that information about equipment, performance, and utilization in a clear and concise manner so that step three "Prioritize OEE improvements" can be conducted effectively.
The OEE reporting system gives information on equipment performance and utilization metrics:
- Equipment OEE is measured by actual tool throughput divided by its theoretical maximum capacity (assuming continuous running with no down-time, no WIP shortage, under some product mix assumptions).
- Equipment throughput is actual throughput, as measured by wafer moves, aligns, or other quantities. This includes all production wafers, rework wafers, and engineering wafers. The calculation excludes monitors and test runs. The measured time period could be a shift, day, week, or more.
- Speed efficiency is equipment throughput, divided by the max. theoretical tool capacity, under the constraints of equipment up-time and product mix during the measured period.
- WIP available is an average available WIP for the measured tool (several snap shots) as retrieved from the MES or other WIP management system (if in place).
- Equipment availability is a daily average up-time retrieved from the MES or a maintenance management system (if in place).
The main challenge with most existing OEE reporting systems is that they only provide visibility about OEE losses at the type categorization level, typically following the "Six big OEE losses definition" — quality loss, speed loss, defects in process, equipment failure, idling and assists, and setup and adjustment. They do not provide data and visibility at the OEE losses root-cause categorization level.
Root-cause analysis is essential to drive OEE improvements. For example, when equipment is reported as being in the "Wait for operator" state, the root cause could be attributable to: an insufficient staffing level, requiring operators to tend to other tools at the same time; an insufficient quantity of WIP — the equipment dedication scheme may not support the current product mix; insufficient procedures for work cell coverage or insufficiently designed work cells that prevent an operator from being in the bay at the time; or the tool alarm not being configured well or the micro-layout of the bay not allowing for eye contact from all directions, resulting in an operator not noticing that the equipment required an assist. Without an analysis of root causes behind OEE losses, a focused improvement effort cannot succeed.
OEE study and sampling software
The most effective method of root-cause identification is to supplement existing OEE reporting capabilities with periodic on-the-floor OEE studies, focusing on critical tools that exhibit lower than expected OEE performance. Such a study allows a fab to enhance its OEE visibility from basic losses-by-type information to more advanced losses-by-root-cause knowledge, yielding a clear prioritization of OEE improvement actions.
 Figure 2. An example of analyzed tool wait status root-cause analysis, specifically for a stepper and track system. |
null
The basis of OEE study methodology is a multi-observation study (MOS). During a MOS, the studied tool-set is subjected to continuous observation for an extended period of time, typically a full week. Using a software package installed on hand-held terminals, the project team samples and records the status of the equipment at 15-min intervals. The software allows the user to define equipment sets and expected OEE losses by type and by root cause. At the completion of the on-the-floor data collection phase, the data is uploaded to a PC, where the software allows advanced statistical analysis and graphical presentations of the findings (Fig. 2).
The software reports include an OEE breakdown of studied equipment and quantification of all equipment OEE loss situations and associated causes. The software reports OEE comparisons for shifts, weekends vs. weekdays, and day vs. night, as well as OEE losses during breaks, lunches, and pass downs. Additional customized analysis can be conducted using simple report engines. In general, the software provides a methodical and thorough look at a fab's real-life OEE performance, including prioritization of OEE loss causes, while giving results that are easy to understand and apply so that opportunities for improvement can be identified.
Improvements implementation
Once root causes for OEE losses have been identified and prioritized through the OEE study, the studied fab is then ready to design and implement any necessary improvement efforts. These can involve cross-functional teams capable of focusing on a well-defined improvement opportunity. A commonly used manufacturing improvement technique, Kaizen events, can be applied for OEE improvements. Kaizen events are short-term activities that focus on redesigning a particular process or portion of work for the purpose of improving equipment uptime, reducing setup time, improving labor productivity, quality, on time delivery, etc.
 Figure 3. Tool performance during shift change; production hourly performance focusing on "Wait for load." |
null
Improvement goals associated with a Kaizen event meet the SMART criteria: specific, measurable, attainable, results-oriented, and time-based. If an OEE study shows that a cause for losses is related to shift changes (Fig. 3), a Kaizen event may be designed to tackle this. Activities may include evaluations of shift-change handshakes, end-of-shift checklists, adherence to start and end of shift schedules, operators' coverage, loading, supervisors' responsibilities definition, etc.
After a company gains experience with and confidence in the value of Kaizen events, they can be expanded to a larger number of OEE loss root cause categories, including operator error, batch sizing, maintenance practices, spare parts management, operators' coverage, test wafers, engineering runs, equipment setup, and hot lots.
Overall value
As cost of capital equipment rises and complexity of manufacturing grows, it is important for semiconductor companies to optimize operations by proactively reducing inefficiencies and effectively preventing losses and errors. This requires comprehensive, world-class OEE management tools that involve the integration of a software solution with a continuous improvement business process.
Continuous OEE improvement teams, using Kaizen techniques and six-sigma tools, are recommended to identify and implement improvement opportunities. The closed-loop cycle is maintained through on-going OEE tracking and performance monitoring.
Amir London, Danny Segev, Tefen Ltd., New York, New York
For more information, contact Amir London at Tefen Ltd., 805 Third Ave., New York, NY 10022; ph 212/317-9600, fax 212/317-0604, [email protected].