Equipment Losses, OEE and Performance Measurement

Introduction

Total Productive Maintenance (TPM) requires more than maintaining equipment until it fails. It requires an evidence-based understanding of how equipment time and capability are consumed. A machine may appear busy while producing defective units, operating below its designed rate, waiting for material, or suffering repeated short stops. Consequently, production performance must be measured through a system that distinguishes productive time from losses.

Overall Equipment Effectiveness (OEE) is the principal TPM measure for examining the productive use of equipment. It decomposes equipment effectiveness into availability, performance and quality. Used appropriately, OEE helps teams identify losses and prioritize improvement. Used mechanically, however, it can encourage target manipulation, inappropriate comparisons and local optimization.

This chapter examines the Six Big Losses, OEE, TEEP, MTBF, MTTR and related measures. It also addresses data definitions, hidden losses, benchmarking, governance and the managerial interpretation of performance indicators.

Learning Objectives

By the end of this chapter, the learner should be able to:

1. Explain the Six Big Losses and relate them to TPM activities.
2. Define planned production time, run time, net run time and fully productive time.
3. Calculate availability, performance, quality, OEE and TEEP.
4. Calculate and interpret MTBF, MTTR, failure rate and related indicators.
5. Distinguish genuine improvement from changes in measurement definitions.
6. Diagnose hidden losses and design a reliable performance-measurement system.
7. Evaluate the managerial and operational limitations of OEE benchmarking.

1. The Logic of Equipment Loss Measurement

Equipment effectiveness is concerned with the proportion of a relevant time window in which equipment produces good output at the intended rate. The word “relevant” is important. A machine cannot be criticized for being unavailable when production was not scheduled, nor should an unplanned shutdown be hidden by excluding all inconvenient periods.

A useful hierarchy is:

• Calendar time: all available time, normally 24 hours per day and 365 days per year.
• Planned production time: calendar time less periods deliberately excluded from production, such as approved shutdowns, holidays or planned non-production intervals.
• Run time: planned production time less unplanned stops and other availability losses.
• Net run time: run time adjusted for speed losses.
• Fully productive time: net run time less time associated with defective output.

This hierarchy should be agreed before measurement begins. Different organizations may classify changeovers, meetings, preventive maintenance and material shortages differently. The resulting percentages are not comparable unless the definitions are comparable.

2. The Six Big Losses

The Six Big Losses classify the principal ways in which equipment effectiveness is reduced. They are commonly grouped under availability, performance and quality.

2.1 Equipment Failure

Equipment failure is an unplanned loss that prevents production or makes production impossible at the required rate. It includes failures of mechanical, electrical, hydraulic, pneumatic, software and control components. The loss begins when production capability is interrupted and ends when the equipment is restored to an accepted operating condition.

A sound measurement system records the event, affected asset, start and end times, failure mode, temporary action, permanent corrective action and output consequence. Merely recording “machine down” is insufficient for learning.

2.2 Setup and Adjustment

Setup and adjustment losses occur when equipment is unavailable for productive output during product changeover, tooling replacement, cleaning, calibration, recipe loading or process adjustment. Some organizations classify planned changeovers outside planned production time; others include them in planned production time. Either policy can be valid, but it must be explicit and stable.

The improvement objective is not simply to reduce the average changeover. It is to reduce total loss while preserving safety, quality and process capability. Single-Minute Exchange of Die (SMED) principles distinguish internal tasks, which require equipment stoppage, from external tasks, which can be performed while the equipment runs.

2.3 Idling and Minor Stops

Minor stops are brief interruptions that may not receive a maintenance work order: a part fails to arrive, a sensor is blocked, an operator clears a jam, or a machine waits for a downstream operation. Individually they may last seconds or minutes; collectively they can consume a large proportion of scheduled time.

A threshold must be defined, such as events shorter than five minutes, but the threshold is not universal. If automatic data collection records every micro-stop while manual reporting records only major stops, the comparison is misleading. Minor stops are often hidden because operators restore operation quickly and because many systems assign the time to “speed loss” or “normal variation.”

2.4 Reduced Speed

Reduced-speed loss occurs when the actual cycle rate is slower than the ideal or standard cycle rate. Causes include worn tooling, material variation, inadequate lubrication, operator caution, unstable process settings and inaccurate standards.

The ideal cycle must be technically defensible. It should represent the fastest demonstrated sustainable rate that satisfies safety, quality and equipment constraints, or a validated engineering standard. A deliberately conservative ideal cycle artificially inflates performance; an unrealistic theoretical rate makes performance appear poor and encourages unsafe operation.

2.5 Process Defects and Rework

A process defect is output that does not meet specification at the first inspection point. Rework may eventually produce an acceptable unit, but it still consumes capacity, labor and time. Therefore, reworked units should not be counted as good first-pass output when calculating quality.

Quality losses include scrap, rework, downgraded product and, where relevant, customer returns attributable to the process. The measurement boundary should be defined: line-level OEE may use defects detected at the line, while plant-level effectiveness may include defects detected later in the process.

2.6 Startup Losses

Startup losses occur after a planned stop or changeover while the process is stabilizing. Examples include incorrect temperature, dimensional drift, unstable pressure, incomplete curing or initial material waste. Startup losses are sometimes concealed by classifying them as adjustment or by excluding the warm-up interval from the denominator. That practice prevents systematic improvement.

3. OEE: Definitions and Equations

OEE is conventionally calculated as the product of availability, performance and quality:

The factors are proportions between 0 and 1, or percentages between 0% and 100%. When percentages are used, divide by 100 before multiplication. OEE is dimensionless.

3.1 Availability

Availability measures the proportion of planned production time during which the equipment was available to run.

Under the common loss classification:

This formulation assumes that equipment failure and setup/adjustment are the selected availability losses. If other scheduled availability losses are included, they must be added consistently.

3.2 Performance

Performance compares the theoretical time required to produce the total output with the actual run time.

Ideal cycle time and run time must use the same time unit, such as seconds. Total count includes good and defective units produced during run time. Performance should not exceed 100% under a correctly defined ideal cycle. A value above 100% usually indicates an incorrect standard, timestamp problem, double counting or inconsistent units.

3.3 Quality

Quality measures the proportion of total output that is good at the first relevant inspection point.

Equivalently, if defect count is defined as all non-good units:

Rework is normally excluded from good count until it has successfully passed the defined inspection. The organization must specify whether quality is measured at the machine, process, line or customer boundary.

3.4 Alternative Time-Based Expression

OEE can also be expressed through fully productive time:

where fully productive time is the time that would have been required to produce good output at the ideal cycle rate. This expression is equivalent to the factorized formula when definitions are mutually consistent and counts and times are complete.

4. Worked OEE Example

An assembly line has an eight-hour shift. A 30-minute meal break is excluded from production. During the resulting planned production time, the line experiences 40 minutes of equipment failure and 20 minutes of setup. It produces 900 units. The ideal cycle time is 24 seconds per unit, and 36 units are defective.

Step 1: Planned production time

Eight hours equals 480 minutes. Therefore:

Planned production time = 480 − 30 = 450 minutes.

Step 2: Run time

Availability losses equal 40 + 20 = 60 minutes.

Run time = 450 − 60 = 390 minutes.

Step 3: Availability

Step 4: Performance

Ideal time for 900 units is 900 × 24 seconds = 21,600 seconds = 360 minutes.

Step 5: Quality

Good count = 900 − 36 = 864 units.

Step 6: OEE

The result indicates that approximately 76.8% of planned production time was converted into good output at the ideal cycle rate. It does not, by itself, identify the best improvement. The loss tree shows that availability, performance and quality all contribute materially.

5. TEEP and the Denominator Problem

Total Effective Equipment Performance (TEEP) relates productive output to total calendar time rather than only planned production time.

where:

Thus:

TEEP is useful for capacity strategy, asset investment and make-or-buy decisions. For example, if OEE is 76.8% and planned production time is 450 minutes in a 480-minute shift, utilization is 93.75%, and TEEP is:

OEE and TEEP answer different questions. OEE asks how effectively scheduled time was used. TEEP asks how effectively total available calendar time was used. A low TEEP may result from low OEE, low scheduling utilization, or both.

6. Reliability and Maintainability Indicators

OEE describes productive effectiveness; reliability and maintainability measures explain equipment behavior.

6.1 MTBF

Mean Time Between Failures (MTBF) is the average operating time between relevant failures for a repairable asset.

Operating time should normally exclude downtime and should be defined consistently. MTBF is meaningful only when failure events have a clear definition and the observation period contains adequate operating exposure. It is not automatically the same as mean life, which is generally used for non-repairable items.

If a machine operates for 1,200 hours and experiences 8 relevant failures:

6.2 MTTR

Mean Time To Repair (MTTR) is the average active repair time required to restore equipment after a failure.

The organization must decide whether diagnosis, waiting for parts, approval and testing are included. A narrow “hands-on wrench time” MTTR is not equivalent to a restoration-time MTTR. For production impact, elapsed downtime is often more useful than hands-on time.

If eight repairs require a total of 32 hours of active repair:

6.3 Failure Rate and Inherent Availability

A simple empirical failure rate is:

Under restrictive assumptions of a constant failure rate and exponential time to failure, the relationship between failure rate and MTBF is:

where λ is the failure rate per unit of operating time. This relationship should not be applied automatically to aging equipment, infant-mortality failures or systems with changing operating conditions.

A commonly used inherent availability approximation is:

It assumes corrective repair is the relevant restoration activity and excludes logistics, preventive maintenance and administrative delays. If downtime includes those delays, operational availability is more appropriate:

The terms must be defined before the indicator is calculated.

7. Data Architecture and Measurement Governance

Reliable metrics begin with reliable event data. A practical equipment-data model should include:

• asset and component identifier;
• event start and end timestamps;
• operating state and reason code;
• product, order and shift;
• planned or unplanned classification;
• quantity produced, good quantity and defect category;
• ideal cycle or validated standard rate;
• operator, maintenance and material context; and
• evidence of correction or approval for amended records.

A data dictionary should define every term. For example, “breakdown” may mean loss of required function, not merely an alarm. “Setup” may begin at the last good unit of the previous product and end at the first good unit of the next product. “Good count” may mean first-pass good count rather than eventually accepted count.

Time synchronization is essential. Machine controllers, Manufacturing Execution Systems, quality systems and maintenance software should use a common clock or a documented reconciliation procedure. Otherwise, the same event may appear to start before it ends, or output may be assigned to the wrong shift.

Data governance also requires version control for standards. When the ideal cycle changes, the reason, approval date and effective period should be retained. Retroactively changing the standard can create an apparent improvement without any physical improvement.

8. Hidden Losses and Metric Distortion

The Six Big Losses are a classification framework, not a guarantee that all losses will be visible. Common hidden losses include:

1. Micro-stops below the recording threshold. Hundreds of short interruptions may be treated as normal operation.
2. Starved and blocked conditions. A machine may be available but unable to produce because upstream or downstream equipment is not synchronized.
3. Waiting for people or materials. Delays may be coded as production time if no machine fault is present.
4. Conservative standards. An inflated ideal cycle makes performance appear strong.
5. Unreported speed reduction. Operators may slow a machine to prevent defects without recording the reason.
6. Rework counted as good output. This inflates quality and may also inflate performance.
7. Unplanned overtime. OEE may remain high while labor and energy costs increase.
8. Planned downtime manipulation. Recurring failures may be moved into planned maintenance or excluded from the denominator.
9. Parallel constraints. A machine may have high local OEE while the line is constrained elsewhere.
10. Quality losses detected downstream. The machine’s quality appears high until defects are linked to their point of origin.

A loss tree should reconcile the time and quantity data. If a shift contains 450 minutes of planned time, all minutes should be assigned to mutually exclusive states whose sum is 450. Similarly, total count should equal good count plus non-good count under the stated definitions. Reconciliation checks expose missing or double-counted events.

9. Benchmarking and Interpretation

Benchmarking OEE requires caution. Comparisons are valid only when the assets have comparable technology, product mix, operating regime, standards, denominator rules and quality boundaries. A high-volume, stable product line should not be compared uncritically with a low-volume line performing frequent changeovers and engineering trials.

Useful benchmarking levels include:

• Internal trend benchmarking: the same asset over time using stable definitions.
• Peer benchmarking: similar assets under comparable conditions.
• Best demonstrated performance: the asset’s validated historical capability.
• External benchmarking: carefully normalized comparison with other organizations.

A universal OEE target is inappropriate. “World-class” percentages sometimes quoted in training materials are not a substitute for a loss-based improvement plan. A plant may improve from 60% to 75% and still have major losses; another may have lower OEE because it deliberately accepts difficult products that create strategic value.

OEE should be used with safety, cost, delivery, energy, labor productivity and customer quality measures. Maximizing OEE can be harmful if it causes overproduction, unsafe operation, excessive inventory or neglect of needed preventive maintenance.

10. Case Study: Illustrative Packaging-Line Improvement

Background

A fictional food-packaging plant operated one filling and sealing line across two shifts. Reported monthly OEE was 82%, apparently close to management’s target. However, delivery performance was deteriorating and overtime was increasing.

Problem

The reported OEE excluded changeovers from planned production time, counted reworked packs as good output after final inspection, and recorded only stops longer than ten minutes. Operators also slowed the filler during material changes without recording a speed-loss reason.

Analysis

A cross-functional team created a common data dictionary and installed automatic state capture for short stops. It redefined good count as first-pass accepted quantity and established a validated ideal cycle for each product family. The team also reconciled line-level counts with quality records.

The revised first-month results were:

The fall in OEE did not represent deterioration. It revealed previously hidden losses. The loss tree showed that short jams, changeover adjustment and speed reduction accounted for most of the gap.

Approach and Solution

The plant used external setup preparation, standardized changeover instructions, sensor-cleaning standards, a centerline-setting sheet and a daily review of the ten largest loss categories. Maintenance analyzed recurring jam modes using cause-and-effect analysis rather than simply resetting the equipment.

Outcome

After three months, governed OEE increased to 78%, changeover time fell by 35%, and overtime declined. More importantly, the plant could explain the improvement through traceable reductions in specific losses. The initial reduction in reported OEE was therefore a measurement improvement, followed by a real operational improvement.

Learning Points

1. A higher reported metric is not necessarily higher performance.
2. Definition governance must precede target setting.
3. Short stops and speed losses can dominate a line’s capacity loss.
4. Quality must be measured at first pass.
5. OEE should support decisions, not replace engineering judgment.

11. Managerial Use, Ethics and Technology

Managers should use OEE to prioritize loss elimination, not to rank operators. Many losses are systemic: poor materials, unstable specifications, inadequate tooling, weak scheduling or delayed maintenance resources. Holding individuals responsible for a metric they cannot control undermines reporting integrity.

Ethically, employees must be able to report downtime and defects without fear of punishment. Incentive schemes based solely on OEE can motivate unsafe bypassing of interlocks, suppression of defect records or reclassification of downtime. Governance should therefore include audit trails, escalation rules and balanced measures for safety, quality and delivery.

Modern systems use programmable logic controllers, sensors, edge computing and Manufacturing Execution Systems to capture state changes automatically. Artificial intelligence can classify alarm patterns, detect anomalous vibration, predict failure and identify combinations of conditions associated with micro-stops. Nevertheless, AI does not resolve poor definitions or biased training data. A predictive model trained on incomplete failure records may appear accurate while missing unreported degradation. Human validation, cybersecurity, access control and explainability remain necessary.

12. Best-Practice Implementation Sequence

A practical implementation sequence is:

1. Define the measurement boundary and purpose.
2. Agree on planned time, run time, ideal cycle, good count and defect rules.
3. Create a controlled reason-code hierarchy.
4. Synchronize clocks and reconcile data sources.
5. Establish a baseline without changing definitions during the baseline period.
6. Build a loss tree by asset, product, shift and failure mode.
7. Attack the largest chronic and sporadic losses using structured problem solving.
8. Verify that improvements are physical, repeatable and sustained.
9. Review standards when products, tooling or process capability changes.
10. Combine OEE with safety, cost, delivery, quality and energy indicators.

The purpose of measurement is learning. If a metric cannot lead to a decision, it may be a reporting burden rather than a management tool.

Chapter Summary

The Six Big Losses provide a practical classification of equipment losses: equipment failure, setup and adjustment, idling and minor stops, reduced speed, process defects and startup losses. Availability captures scheduled time lost to equipment and setup events; performance captures operation below the ideal cycle; and quality captures defective or non-first-pass output.

OEE is the product of availability, performance and quality, or equivalently fully productive time divided by planned production time. TEEP extends the denominator to calendar time and therefore incorporates utilization. MTBF measures average operating time between relevant failures, while MTTR measures average repair time under a defined repair boundary. These indicators are useful only when their assumptions, units and event definitions are controlled.

The principal professional challenge is not arithmetic but measurement governance. Hidden micro-stops, inconsistent standards, rework counting, denominator manipulation and downstream quality detection can distort results. OEE should therefore be used as a loss-analysis instrument, supported by reliable data, balanced with safety and business measures, and interpreted through comparable operating contexts.

Key Terms

Availability: Proportion of planned production time during which equipment is available to run.

Calendar time: Total time available in the selected observation period.

Defect: Output that does not meet the specified acceptance criteria at the defined inspection point.

Equipment failure: Unplanned loss of required equipment function.

Fully productive time: Time theoretically required to produce good output at the ideal cycle rate.

Good count: Units accepted as good at the defined first-pass inspection boundary.

Ideal cycle time: Validated minimum sustainable time required to produce one unit under specified conditions.

Idling and minor stops: Short interruptions that prevent production but may not generate formal downtime records.

MTBF: Mean operating time between relevant failures of a repairable asset.

MTTR: Mean active repair time required to restore equipment after failure.

OEE: Overall Equipment Effectiveness, calculated from availability, performance and quality.

Performance: Ratio of ideal production time for total output to actual run time.

Planned production time: Calendar time designated for production after approved exclusions.

Quality: Ratio of good first-pass output to total output.

Run time: Planned production time minus defined availability losses.

Setup and adjustment: Time lost during changeover, tooling, cleaning, calibration or process adjustment.

Six Big Losses: TPM classification of the principal availability, performance and quality losses.

TEEP: Total Effective Equipment Performance relative to calendar time.

Utilization: Ratio of planned production time to calendar time.

Review Questions

Short Answer Questions

1. What are the Six Big Losses, and how are they grouped under the three OEE factors?
2. Distinguish planned production time from run time.
3. Why should reworked units normally be excluded from good count in OEE quality calculations?
4. Define MTBF and state one condition necessary for interpreting it correctly.
5. What is the principal difference between OEE and TEEP?

Long Answer Questions

1. Explain the conceptual structure of OEE and discuss how denominator definitions affect reported performance.
2. Describe the causes and appropriate TPM responses for minor stops and reduced-speed losses.
3. Compare MTBF, MTTR, inherent availability and operational availability, including their assumptions.
4. Discuss why external OEE benchmarking can be misleading and propose a robust benchmarking approach.
5. Explain how data governance, ethical incentives and audit trails support trustworthy equipment measurement.

Analytical / Application Questions

1. A machine has 600 minutes of planned production time, 72 minutes of failure and setup losses, produces 1,500 units at an ideal cycle of 18 seconds, and records 45 defective units. Calculate availability, performance, quality and OEE.
2. A plant reports OEE of 85% but excludes all changeovers and counts rework as good output. Identify at least four distortions and recommend revised definitions.
3. An asset operates for 2,400 hours, experiences 12 failures and has 54 hours of corrective repair time. Calculate MTBF and MTTR, then explain why these values alone do not establish operational availability.

Multiple Choice Questions with Answers

Question 1

Which loss belongs primarily to the availability category of OEE?

A. Reduced speed

B. Equipment failure

C. Process defect

D. Startup rejection

Correct Answer: B. Equipment failure

Question 2

A line has 480 minutes of planned production time and 60 minutes of availability losses. What is its availability?

A. 75.0%

B. 80.0%

C. 87.5%

D. 92.5%

Correct Answer: C. 87.5%

Question 3

A machine produces 800 units, of which 32 are defective. What is its quality rate?

A. 4.0%

B. 92.0%

C. 94.0%

D. 96.0%

Correct Answer: D. 96.0%

Question 4

Which expression correctly calculates performance?

A. Run Time ÷ Planned Production Time

B. Good Count ÷ Total Count

C. Ideal Cycle Time × Total Count ÷ Run Time

D. Total Downtime ÷ Calendar Time

Correct Answer: C. Ideal Cycle Time × Total Count ÷ Run Time

Question 5

If OEE is 70% and utilization is 80%, what is TEEP?

A. 56%

B. 70%

C. 80%

D. 150%

Correct Answer: A. 56%

Question 6

An asset operates for 900 hours and experiences 6 relevant failures. What is its MTBF?

A. ห 150 hours

B. 906 hours

C. 894 hours

D. 6,900 hours

Correct Answer: A. ห 150 hours

Question 7

Which condition most commonly indicates a problem with the ideal cycle-time definition?

A. Quality is below 100%

B. Performance consistently exceeds 100%

C. Availability is below 100%

D. There are planned breaks

Correct Answer: B. Performance consistently exceeds 100%

Question 8

Why should rework generally not be counted as good output in first-pass quality?

A. Rework always increases cycle speed

B. Rework consumes resources and was not acceptable initially

C. Rework is never recorded

D. Rework reduces planned production time

Correct Answer: B. Rework consumes resources and was not acceptable initially

Question 9

MTTR is best described as:

A. Average operating time between failures

B. Average time between product changeovers

C. Average repair time after relevant failures

D. Total calendar time divided by output

Correct Answer: C. Average repair time after relevant failures

Question 10

What is the most appropriate first step before comparing OEE across two plants?

A. Select the plant with the lower value

B. Verify that definitions, boundaries and operating contexts are comparable

C. Multiply both values by utilization

D. Exclude all quality losses

Correct Answer: B. Verify that definitions, boundaries and operating contexts are comparable