Operation, Maintenance, and Continuous Improvement
What this domain tests
Operation, Maintenance, and Continuous Improvement tests the running phase: SOPs and training, maintenance strategy, spares, calibration, backup/restore, performance measurement, root-cause analysis, management of change, and incremental improvement.
Core rules and concepts
- Define planned production time before calculating OEE. OEE = Availability x Performance x Quality, where availability uses planned time, performance uses ideal cycle time and actual run time, and quality uses good units over total units.
- Choose maintenance tactics by failure consequence: reactive for non-critical, preventive for age-related, predictive/condition-based for random or hidden failures, and design-out where recurring.
- Maintain critical spares based on lead time, failure rate, and production impact, not on shelf appeal.
- Calibration, backup/restore, and patch management are control-system maintenance, not IT afterthoughts.
- Use management of change (MOC) for any change to software, setpoints, recipes, or hardware. Update drawings, SOPs, training, and spare parts.
- Use 5 Whys, fishbone, and Pareto for root cause; verify the fix with data.
Worked example
Planned production time = 480 min/shift. Downtime = 48 min. Ideal cycle = 30 s/unit. Total count = 800; good count = 760. Run time = 480 - 48 = 432 min. Availability = 432/480 = 90.0%. Performance = (0.5 min/unit x 800 units) / 432 min = 400/432 = 92.6%. Quality = 760/800 = 95.0%. OEE = 0.90 x 0.926 x 0.95 = 0.792 = 79.2%. Reasoning: do not use 480 min as the performance denominator; use actual run time. The biggest loss here is quality (5%) followed by performance (7.4%), so investigate rejects before buying a faster filler.
Traps that cost marks
- Using calendar time in availability and inflating or deflating OEE.
- Counting planned changeover as downtime without agreeing the definition.
- Treating OEE as a target without a loss tree; it is a diagnostic, not a strategy.
- Doing maintenance without MOC, so the backup and drawings no longer match the plant.
- Fixing symptoms without verifying root cause with data.
- Ignoring operator training and SOP updates after a change.