OEE Gap and Unknown Downtime on a Stamping Press
An automotive parts plant tracks OEE in MES across three stamping presses. Press 2 shows OEE at 54% versus 71% for the other presses. The MES downtime log shows 62% of downtime coded as unknown and the rest as operator break. Maintenance records show the press's tonnage monitor calibration is overdue by 45 days, and a die-change PM was skipped last month. The vibration sensor on the press sends data every 30 seconds, but MES stores only 5-minute averages. The production supervisor wants to add a new downtime reason code called material shortage because operators say the coil feeder jams. The maintenance planner wants to prioritize a rebuild during the next shutdown in 3 weeks. What is your analysis and action plan?
1. Analyze the OEE gap and data quality. What can and cannot be concluded?
Read the worked answer
The 54% OEE is real as a calculated result, but its components are not trustworthy. If 62% of downtime is coded unknown, the loss breakdown is dominated by missing data, not by a proven equipment failure. Operator break as the only other code suggests the reason-code list is inadequate or operators are defaulting to the nearest code. The 5-minute vibration averages hide short feeder jams that occur within a minute, so the MES cannot correlate jams with downtime. The overdue tonnage calibration affects the reliability of tonnage data but does not by itself explain OEE. The skipped die-change PM is a real risk. Do not conclude that a rebuild is justified from this OEE alone; also do not dismiss the press. The correct conclusion is that the loss data are not yet diagnostic.
2. What immediate and medium-term actions would you take?
Read the worked answer
Short term, put a temporary coding rule in place: operators must select a defined reason at the time of the stop, and unknown requires a supervisor note. Add material shortage only after reviewing feeder jam events and defining the code precisely; otherwise it becomes a catch-all. Capture downtime at the source by integrating the press PLC stop signal with MES, and store vibration data at the native 30-second interval or as event-based summaries rather than 5-minute averages. Review the feeder jam reports and observe the coil feeder. Restore tonnage calibration and complete the skipped PM. For the rebuild, require condition evidence such as vibration trends, tonnage deviation, and maintenance history. Use the next three weeks to collect clean data; if the press remains the outlier after coding is fixed, prioritize the rebuild. If safety or quality risk appears, escalate sooner.
3. How would you govern downtime reason codes so the new code does not become another unknown?
Read the worked answer
Create a controlled reason-code taxonomy with clear definitions, examples, and a rule that codes must be mutually exclusive where possible. Limit the list to codes operators can identify at the point of loss, and separate cause from symptom. Require a secondary detail field for feeder jams, such as coil splice, sensor fault, or mechanical jam. Train operators and supervisors, and audit coding accuracy by comparing MES entries with line observations. Track unknown-code percentage as a data-quality KPI with a target, and review weekly. Do not allow a new code to be added without a change request, owner, and definition. Use Pareto analysis after two weeks of clean data, and feed the results into maintenance planning and continuous improvement. This governance keeps the reason-code set stable and makes the OEE loss tree usable for decisions.