Overall Equipment Effectiveness (OEE)

Measures the percentage of scheduled production time that equipment is actually running. This factor captures downtime losses—both planned (changeovers, maintenance) and unplanned (breakdowns, material shortages).

• Formula: Availability = (Run Time / Planned Production Time) × 100
• Run Time = Planned Production Time − Stop Time
• A score of 90% means the asset was down for 10% of its scheduled window
• Edge AI systems detect micro-stops and slow cycles that traditional PLCs miss, feeding precise downtime data directly into OEE calculations

Quantifies how fast equipment runs compared to its designed maximum speed. This factor exposes speed losses—slow cycles, minor stoppages, and reduced throughput that erode capacity without triggering downtime events.

• Formula: Performance = (Ideal Cycle Time × Total Parts Produced) / Run Time × 100
• Ideal Cycle Time is the theoretical fastest time to produce one unit, derived from equipment specifications
• A score of 95% indicates the line is running 5% slower than its nameplate capacity
• Computer vision systems at the edge track individual cycle times in milliseconds, detecting speed degradation before it impacts shift output

Represents the proportion of total output that meets specifications without rework. This factor isolates defect losses—scrap, rework, and startup rejects that consume resources without producing sellable goods.

• Formula: Quality = (Good Parts Produced / Total Parts Produced) × 100
• Good Parts exclude all units requiring rework or scrapped entirely
• A score of 99.9% reflects a Six Sigma-level defect rate
• Edge-deployed neural networks perform inline quality inspection at line speed, classifying defects in real-time and updating the quality rate continuously rather than relying on batch sampling

OEE is the product of its three components, providing a single metric that reflects the compound effect of all manufacturing losses. A perfect score of 100% represents manufacturing only good parts, as fast as possible, with zero downtime.

• Formula: OEE = Availability × Performance × Quality
• Example: 90% Availability × 95% Performance × 99.9% Quality = 85.4% OEE
• World-class manufacturing targets 85% OEE; typical factories operate between 40% and 60%
• The multiplicative nature means weakness in any single factor dramatically reduces the overall score—a 70% Availability rate caps maximum OEE at 70%, regardless of perfect Performance and Quality
• Edge AI platforms compute OEE in real-time by ingesting sensor streams, PLC tags, and vision system outputs, surfacing the metric on dashboards updated every second rather than at shift-end

OEE's diagnostic power comes from mapping each component to specific loss categories. The Six Big Losses framework, originating from Seiichi Nakajima's Total Productive Maintenance methodology, structures root-cause analysis and corrective action.

Availability Losses:

• Equipment Failure: Catastrophic breakdowns requiring repair; measured by Mean Time Between Failures (MTBF) and Mean Time To Repair (MTTR)
• Setup and Adjustment: Changeover time between product variants; targeted by Single-Minute Exchange of Die (SMED) programs

Performance Losses:

• Idling and Minor Stops: Sub-minute stoppages from sensor faults, jammed parts, or misfeeds that operators clear without logging
• Reduced Speed: Operation below design speed due to wear, poor maintenance, or operator caution

Quality Losses:

• Process Defects: Scrap and rework produced during stable operation
• Reduced Yield: Startup rejects from machine warm-up, changeover stabilization, or process tuning

Edge AI systems automatically classify stoppages and defects into these categories using pattern recognition on time-series data, eliminating manual loss accounting.

Traditional OEE tracking relies on manual data collection or batch PLC polling, producing lagging indicators reviewed hours or days after a shift ends. Edge AI transforms OEE into a real-time control parameter.

• Sensor Fusion: Vibration, acoustic, thermal, and vision sensors feed a unified data stream to edge inference engines, detecting micro-stops and speed losses invisible to PLC cycle counters
• Automated Reason Coding: Natural language processing models running on edge hardware parse operator notes and maintenance logs to assign standardized loss categories without manual data entry
• Predictive OEE: Time-series forecasting models project OEE trajectory for the current shift, alerting supervisors to interventions needed to hit production targets before the window closes
• Closed-Loop Optimization: When OEE drops below a threshold, edge agents automatically adjust machine parameters—feed rate, temperature setpoints, or tool offsets—to restore target performance without human intervention
• Federated Benchmarking: Models trained across multiple factory sites share anonymized OEE patterns, enabling a plant to benchmark its performance against a global fleet without exposing proprietary production data

The ratio of actual production time to planned production time. It accounts for downtime losses caused by equipment failures, setups, and adjustments.

• Formula: (Run Time / Planned Production Time) × 100
• Run Time is Planned Production Time minus Stop Time.
• A score of 100% means the process ran continuously during the planned period without any stops.
• Unplanned stops (breakdowns) and planned stops (changeovers) are the primary loss categories.

Measures whether the equipment is operating at its maximum theoretical speed during the time it is available. It captures slow cycles and minor stoppages.

• Formula: (Ideal Cycle Time × Total Count) / Run Time × 100
• Accounts for losses where equipment runs below its nameplate capacity.
• A score below 100% indicates speed losses, often caused by worn components, suboptimal settings, or operator inefficiency.
• Calculated net of availability, isolating pure speed degradation.

The proportion of good units produced relative to the total units started. It exclusively measures production output that meets specifications on the first pass.

• Formula: (Good Count / Total Count) × 100
• Defects and units requiring rework are subtracted from the total count.
• A 100% score indicates zero defects and zero rework.
• Process defects (during steady-state production) and startup rejects (during warm-up) are the two primary loss categories.

A framework for categorizing all sources of equipment inefficiency, directly mapping to the three OEE factors. It provides a structured root-cause analysis tool.

• Availability Losses: 1. Equipment Failure (breakdowns), 2. Setup and Adjustments (changeovers).
• Performance Losses: 3. Idling and Minor Stops (jams), 4. Reduced Speed (wear).
• Quality Losses: 5. Process Defects (scrap), 6. Reduced Yield (startup rejects).
• Addressing these losses systematically is the primary goal of OEE improvement programs.

The shift from manual, shift-end data collection to streaming telemetry that computes OEE continuously. Edge AI systems ingest sensor data to provide sub-second metric updates.

• Requires direct integration with PLCs and sensors via OPC UA or MQTT Sparkplug.
• Enables immediate operator alerts when Availability, Performance, or Quality deviates from targets.
• Transforms OEE from a lagging historical report into a leading operational indicator.
• Edge computing eliminates cloud latency, ensuring deterministic calculation for closed-loop control.

A related metric that measures utilization against all available calendar time (24/7/365), not just planned production time. It exposes the hidden cost of idle capacity.

• Formula: OEE × Utilization, where Utilization = Planned Production Time / All Time.
• While OEE measures how well you run during planned time, TEEP measures how much of your total asset capacity you actually use.
• A high OEE with a low TEEP suggests significant market demand or scheduling constraints, not equipment inefficiency.
• Used for strategic capacity planning and capital expenditure justification.