Maintenance Economics and Performance Management

Introduction

Maintenance is not merely a technical activity concerned with repairing equipment. It is an economic management system that converts labour, materials, technology and engineering knowledge into asset availability, production capability, safety and quality. At postgraduate level, maintenance decisions must therefore be evaluated over the entire asset life rather than by the immediate repair bill alone.

A low-cost maintenance department may produce poor business results if it creates excessive downtime, quality losses, unsafe conditions or premature replacement. Conversely, a department with relatively high maintenance expenditure may create superior value if it protects throughput, extends asset life and reduces operational risk. The central managerial question is not “How much does maintenance cost?” but “What level and pattern of maintenance maximizes lifecycle value subject to safety, quality, environmental and operational constraints?”

This chapter develops methods for evaluating lifecycle cost, total cost of ownership, downtime losses, replacement alternatives and maintenance investments. It then develops a performance-management system using OEE, availability, schedule compliance, planned-work ratio and cost measures. Particular attention is given to definitions, measurement boundaries, data quality and the dangers of optimizing one indicator at the expense of system performance.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Distinguish maintenance cost, lifecycle cost and total cost of ownership.
2. Estimate downtime losses and incorporate them into maintenance decisions.
3. Compare repair, refurbishment and replacement alternatives using discounted cash flow.
4. Evaluate preventive and predictive-maintenance investments under uncertainty.
5. Calculate and interpret availability, OEE, schedule compliance, planned-work ratio and cost indicators.
6. Design a balanced maintenance KPI system aligned with production, reliability, safety and financial objectives.
7. Critically assess data quality, gaming risks, ethical issues and the limitations of maintenance metrics.

1. The Economic Logic of Maintenance

Maintenance creates value through several mechanisms:

• preserving the productive function of an asset;
• reducing the probability and consequence of failure;
• protecting product quality and process capability;
• reducing safety and environmental exposure;
• extending useful life;
• improving energy and material efficiency; and
• providing information for operating and investment decisions.

Its economic effects may be classified as follows:

A financially sound maintenance policy balances these categories. Minimizing direct maintenance expenditure alone is a local optimization and may reduce total value.

1.1 Cost classifications

Fixed costs do not change substantially with short-term workload, such as core supervisory salaries or a maintenance-management system. Variable costs change with work volume, such as consumables, contractor hours and replacement parts. Avoidable costs disappear if a maintenance policy or asset is discontinued; unavoidable costs remain. This distinction is important in replacement analysis because sunk costs—costs already incurred—should not influence a future choice except where they have tax or salvage implications.

Failure cost is the cost associated with an unplanned loss of function. Preventive cost is incurred to reduce the likelihood of failure, while predictive or condition-monitoring cost is incurred to identify deterioration and improve intervention timing. Corrective work may be planned or unplanned; “corrective” does not automatically mean “poor practice.” A condition-based replacement carried out before functional failure is planned corrective work.

2. Lifecycle Cost and Total Cost of Ownership

Lifecycle cost (LCC) is the total cost of an asset from conception through disposal. A useful general expression is:

Here, C₀ is initial acquisition and installation cost; C_o,t is operating cost in period t; C_m,t is maintenance cost; C_d,t is downtime-related cost; C_r,t is risk, regulatory or failure consequence cost; C_e,t is end-of-life or disposal cost; S_n is salvage value at the end of period n; and i is the discount rate per period. All monetary values must be expressed in the same currency and time basis.

The equation assumes that expected costs can be estimated and discounted. Where failures are uncertain, each period’s consequence should be an expected value:

where p_j,t is the probability of failure mode or consequence j in period t, and C_j,t is its cost. Probabilities should represent the relevant operating context, not merely historical averages when the asset, load or environment has changed.

Total cost of ownership (TCO) is often used more narrowly for the complete cost borne by the owner, including purchase, logistics, operation, support, downtime and disposal. In practice, TCO and LCC overlap. The distinction is one of scope: LCC emphasizes the asset’s complete life, while TCO emphasizes the owner’s total economic burden, which may include supply-chain and contractual costs.

2.1 Discounting and equivalent annual cost

A future cost is worth less than an identical present cost because capital has an opportunity cost and future outcomes are uncertain. The present value of a future amount F_t is:

For alternatives with unequal lives, equivalent annual cost (EAC) can make annual economic burdens comparable. If PV is the present value of all relevant costs, n is life in years and i is the annual discount rate, then:

This capital-recovery expression assumes a constant discount rate and an equivalent annual cash-flow interpretation. It should not be used mechanically when utilization, inflation, taxation or residual value differs materially between alternatives; those effects must first be included in the present-value calculation.

3. Downtime Economics

Downtime is economically important because an unavailable asset may prevent production, disrupt connected operations or create quality and delivery consequences. The loss per hour should not automatically be equated with sales revenue. A more defensible estimate begins with contribution margin:

L_d is downtime loss per hour; Q_h is saleable output that could have been produced per hour; P^c is contribution price per unit; V is variable cost per unit avoided when production is lost; C_x is additional recovery or expediting cost; C_q is expected quality-related cost; and C_s is other consequential cost. The formula assumes that lost output is not fully recovered elsewhere and that the selected contribution measure is appropriate.

If only a proportion r of lost production is genuinely unrecoverable, then the first term should be multiplied by r. If production can be recovered later through overtime, the net loss may be lower, although overtime, inventory and delivery effects must be included.

Expected annual downtime cost can be estimated as:

where λ is expected failure frequency per year, T^f is average downtime per failure, T^p is planned outage time, and L_d^p is the loss rate during planned downtime. Planned downtime may have a lower loss rate if it is scheduled during low-demand periods.

Worked example. A bottleneck machine produces 40 saleable units per hour. Contribution margin is $18 per unit. A failure causes 6 hours of lost production and $1,200 of expediting and quality cost. The production loss is:

40 × 18 × 6 = $4,320.

Total event consequence is therefore $5,520, assuming all lost output is unrecoverable. If a condition-monitoring programme costs $14,000 annually and is expected to prevent four such events, its gross avoided consequence is 4 × 5,520 = $22,080. Before approving it, management should subtract monitoring cost and include false alarms, implementation cost, residual failures and the time value of money. The simple first-year net benefit is $8,080, not $22,080.

4. Replacement and Maintenance-Investment Decisions

Replacement decisions compare the future economics of keeping an existing asset with those of acquiring an alternative. The correct comparison uses relevant incremental cash flows: future costs and benefits that differ between options. Original purchase price is normally sunk. Important variables include:

• remaining life and reliability of the current asset;
• major overhaul cost and expected life extension;
• operating, energy and labour costs;
• expected downtime and quality consequences;
• acquisition, installation and commissioning cost;
• training, integration and disposal cost;
• salvage value and tax effects; and
• strategic capacity, safety and obsolescence effects.

For an existing asset, the economic cost of retaining it in the next period is not its accounting book value. It is the expected future cost of operation, maintenance, downtime and eventual replacement. An overhaul should be accepted only when its incremental benefits exceed its incremental costs, not because the asset has already received substantial investment.

4.1 Illustrative replacement comparison

The following is a fictional illustrative example. A plant is comparing one-year continuation of an old pump with immediate replacement by a new pump. The discount rate is ignored because the comparison concerns one year. The old pump requires $24,000 maintenance, $18,000 energy, and expected downtime cost of $30,000. It can be sold for $2,000 at year-end. The new pump requires $10,000 maintenance, $12,000 energy, $8,000 expected downtime cost, $85,000 acquisition and installation, and has a year-end salvage value of $55,000. Disposal of the old pump costs $3,000.

Old-pump one-year net cost:

24,000 + 18,000 + 30,000 − 2,000 = $70,000.

New-pump one-year net cost:

85,000 + 10,000 + 12,000 + 8,000 + 3,000 − 55,000 = $63,000.

The replacement has an estimated one-year advantage of $7,000. This result is conditional: it depends on the reliability estimates, the treatment of disposal, the availability of capital and the chosen horizon. A multi-year discounted analysis is preferable for assets with long lives.

4.2 Investment appraisal

For an investment, net present value is:

I₀ is initial investment, B_t benefits in period t, C_t operating and maintenance costs, S_n residual value, and i the discount rate. A positive NPV indicates value creation relative to the assumed discount rate, provided estimates are credible and constraints such as safety and regulation are satisfied.

The payback period is easy to communicate but ignores benefits after payback and, in its undiscounted form, ignores the time value of money. It is therefore a screening measure rather than a complete decision rule. Sensitivity analysis should vary failure frequency, downtime cost, benefit realization, discount rate and useful life. For high-consequence assets, scenario analysis or real-options reasoning may be more appropriate than a single-point estimate.

5. Maintenance Performance Measurement

A KPI is useful only when its definition, data source, owner, review frequency and decision consequence are explicit. Maintenance measures should connect asset behaviour to organizational outcomes through a hierarchy:

1. Outcome measures: safety, quality, delivery, cost and customer performance.
2. Asset measures: availability, reliability, failure frequency and maintainability.
3. Work-process measures: schedule compliance, planned-work ratio, backlog age and wrench time.
4. Capability measures: skills, data quality, spare-parts effectiveness and planning maturity.

No single KPI represents maintenance excellence. A balanced system prevents improvement in one dimension from damaging another.

5.1 Availability

Availability has multiple valid definitions. Inherent availability focuses on corrective maintenance and excludes logistics and preventive delays:

MTBF is mean time between failures and MTTR is mean time to repair. The formula assumes statistically stable operation and that the relevant cycle consists of operating time plus corrective repair time.

Operational availability includes preventive maintenance and, depending on the organization’s convention, administrative or logistical delays:

The organization must define whether waiting for parts, permits and technicians is downtime. Comparisons are invalid when sites use different boundaries.

5.2 OEE

Overall equipment effectiveness decomposes productive performance into availability, performance and quality:

For a common manufacturing definition:

In unambiguous written form, this is:

PPT is planned production time, ST is stop time during that period, operating time is PPT − ST, and counts refer to the same production interval. Units must be consistent: ideal cycle time per unit multiplied by total count gives time. OEE is not the same as utilization, and it should not be used to compare dissimilar processes without standardized definitions.

Worked example. Planned production time is 450 minutes, stop time is 50 minutes, ideal cycle time is 0.5 minute per unit, total count is 720 units and good count is 684 units.

A = (450 − 50)/450 = 0.8889.

P = (0.5 × 720)/400 = 0.9000.

Q = 684/720 = 0.9500.

Therefore:

OEE = 0.8889 × 0.9000 × 0.9500 ≈ 0.7600, or approximately 76.0%.

The loss tree is more actionable than the OEE number alone: 50 minutes of stopping, 10% speed loss and 5% defective output require different interventions.

5.3 Schedule compliance and planned-work ratio

Schedule compliance measures execution of approved work within the defined period:

The numerator and denominator must use the same unit—work orders, labour hours or equivalent work content. A work-order count can be distorted if many trivial jobs are used to offset missed major jobs. Work content or criticality-weighted compliance may be more meaningful.

Planned-work ratio measures the share of maintenance effort that was planned before execution:

“Planned” must be defined, for example, as work with a defined scope, parts, labour estimate and schedule established before execution. Planning is not identical to scheduling: a job may be planned but not yet assigned to a date. Emergency work should not be hidden by retrospective planning after a failure.

5.4 Cost measures

Useful cost measures include maintenance cost per unit, maintenance cost as a percentage of replacement asset value, cost by asset criticality and budget variance. A simple unit-cost measure is:

Cost per unit can improve because production increases, even while absolute maintenance cost and risk deteriorate. It should therefore be paired with asset health, downtime, safety and quality indicators. Budget variance is a control measure, not a value measure: underspending may indicate efficiency or deferred work.

6. Designing a Balanced KPI System

A robust KPI architecture follows five design principles.

First, define the decision. A KPI should answer a management question, such as whether planning capacity is adequate or whether a bottleneck needs reliability investment.

Second, define the boundary. Specify asset population, time period, exclusions, downtime categories, labour basis and data source.

Third, establish causal links. For example, improved failure coding and planning may increase reported planned work initially while revealing previously hidden emergency work. Interpretation must recognize this transition.

Fourth, balance leading and lagging indicators. OEE and cost are lagging outcomes; schedule compliance, condition-monitoring coverage and overdue critical inspections are leading indicators.

Fifth, review interactions. A target of 100% schedule compliance can encourage low-value work or the removal of legitimate emergent work. A target to reduce cost can encourage postponement. KPI reviews should examine trade-offs rather than reward isolated maxima.

A practical maintenance dashboard may contain the following compact set:

7. Governance, Data and Ethical Issues

Maintenance economics depends on trustworthy data. Common weaknesses include inconsistent failure codes, missing start and finish times, unreliable production counts, unrecorded minor stoppages and mixing calendar time with planned production time. A KPI dictionary should define every measure, formula, inclusion rule, data owner and revision history.

There are also ethical risks. Managers may classify failures as planned, close work orders without effective completion, suppress bad news, or defer safety-critical tasks to meet cost targets. Such behaviour makes the dashboard look better while increasing operational risk. Governance should therefore require auditable records, separation of safety indicators from cost incentives, escalation of critical overdue work and periodic independent review.

Regulatory and contractual requirements may impose minimum inspection intervals, competence requirements, statutory records and environmental controls. Economic optimization cannot justify non-compliance. Where failure consequences involve injury, pollution or public safety, expected monetary value alone is inadequate; risk acceptance criteria and the precautionary principle may dominate.

8. Technology, Analytics and AI

Computerized maintenance-management systems provide work-order, asset, labour, inventory and cost data. Industrial sensors and historians can support condition-based maintenance. Machine-learning models may estimate failure probability or remaining useful life, but the economic value of a model depends on decisions, not prediction accuracy alone.

A model should be evaluated using false-positive and false-negative costs, intervention lead time, data drift, explainability and the consequences of automation. If a false alarm causes unnecessary shutdown while a missed alarm causes catastrophic failure, the decision threshold should reflect those asymmetric costs. AI should support—not replace—engineering judgment, inspection standards, worker knowledge and accountability. Cybersecurity is also part of maintenance economics because compromised control or monitoring systems can create downtime and safety consequences.

9. Integrated Illustrative Case: Packaging-Line Reliability Investment

Background. A fictional food-packaging plant operates a bottleneck line for 4,000 planned hours per year. The line has frequent unplanned stops caused by conveyor bearings and sealing-temperature faults. Production management reports declining delivery performance, while maintenance reports that its annual budget is already high.

Problem. The plant must decide whether to fund a condition-monitoring and planned-maintenance programme. Historical records show 32 significant failures per year, each causing an average 3.5 hours of lost production. Contribution margin is $22 per saleable unit, output is 30 units per hour, and average recovery and quality cost is $600 per failure. The proposed programme costs $28,000 per year and is expected to reduce significant failures by 35%, with a further $5,000 annual software and training cost included in the programme estimate.

Analysis. Lost contribution per failure is:

3.5 × 30 × 22 = $2,310.

Total consequence per failure is 2,310 + 600 = $2,910. Annual avoidable consequence is:

32 × 0.35 × 2,910 = $32,592.

The estimated first-year net benefit is 32,592 − 28,000 = $4,592. This is positive but modest. Sensitivity analysis is essential. If the reduction is only 20%, avoided consequence is $18,624 and the programme loses $9,376 before considering intangible benefits. If the actual consequence is higher because missed deliveries cause penalties, the investment becomes more attractive.

The engineering review also finds that 40% of failures arise from inadequate lubrication and misalignment. A low-cost standardization programme may therefore have a higher return than sensors alone. The recommended approach is a staged investment: correct basic conditions, improve failure coding, pilot monitoring on the most critical bearings, and release further funding only when leading and outcome measures confirm benefit.

Outcome. After six months, the plant should compare a controlled baseline with pilot results using failure frequency, downtime hours, repeat failures, planned-work ratio, quality losses and total cost—not sensor alarms alone. A governance review should verify that production recovery has not merely shifted losses to overtime or inventory.

Learning points. Economic evaluation must use consequences rather than repair cost alone; uncertainty requires sensitivity analysis; basic maintenance conditions may dominate advanced technology; and a KPI system must test whether apparent improvements are real and sustainable.

10. Limitations and Best Practices

Economic models are approximations. Failure distributions may change with load, age and maintenance quality. Downtime costs may be nonlinear when a bottleneck affects an entire network. Benefits such as safety, knowledge retention and reputation are difficult to monetize. Discount rates can bias long-lived safety investments against future benefits. These limitations do not invalidate analysis, but they require transparent assumptions and scenario ranges.

Best practice is to:

• use lifecycle rather than purchase-price thinking;
• separate sunk, fixed, variable and incremental costs;
• value downtime using contribution and actual recovery constraints;
• compare unequal-life alternatives using discounted cash flow;
• define KPI boundaries before collecting results;
• combine outcome, leading and risk indicators;
• investigate loss categories behind OEE;
• audit data and prevent retrospective reclassification;
• use criticality and consequence to prioritize work; and
• review KPI effects on safety, quality, people and long-term asset health.

Chapter Summary

Maintenance economics evaluates the value created by preserving asset function, not simply the money spent on repairs. Lifecycle cost and total cost of ownership include acquisition, operation, maintenance, downtime, risk and disposal, with future cash flows discounted when appropriate. Downtime should be valued using lost contribution and consequential costs, while recognizing recovery, bottleneck and quality effects.

Replacement decisions should use relevant incremental future cash flows and should not be distorted by sunk costs. NPV, equivalent annual cost and sensitivity analysis support investment decisions, but safety and regulatory constraints remain decisive. Performance management requires clearly defined and balanced indicators. Availability describes time readiness; OEE combines availability, performance and quality; schedule compliance measures execution of due work; planned-work ratio indicates the degree of proactive control; and cost measures show financial burden but not value in isolation.

Effective systems align KPIs with decisions, establish consistent data boundaries, combine leading and lagging measures, expose trade-offs and resist gaming. Digital and AI tools can improve prediction and prioritization, but their value depends on economically sound decisions, data governance, cybersecurity and professional accountability.

Key Terms

Lifecycle cost: Total discounted cost of an asset from conception through disposal.

Total cost of ownership: Complete economic burden borne by the owner over the asset’s useful life.

Downtime cost: Economic consequence of lost or degraded productive capability.

Contribution margin: Revenue per unit minus relevant variable cost per unit.

Sunk cost: A past cost that cannot be changed by a current decision.

Incremental cost: Future cost that differs between decision alternatives.

Net present value: Discounted value of future investment benefits less costs.

Equivalent annual cost: Annualized representation of a present-value cost.

Inherent availability: Availability considering operating time and corrective repair time.

Operational availability: Availability including defined operational delays and maintenance losses.

OEE: Product of availability, performance and quality.

Availability: Proportion of defined time during which an asset is capable of performing its function.

Performance: Ratio comparing actual productive speed with ideal speed.

Quality: Proportion of output that meets requirements.

Schedule compliance: Percentage of scheduled work completed within the defined period.

Planned-work ratio: Planned maintenance labour hours as a proportion of total maintenance labour hours.

Failure consequence: Cost or effect resulting from loss of function.

Criticality: Relative importance of an asset based on consequence and likelihood.

KPI governance: Rules defining measurement, ownership, review and accountability for indicators.

Review Questions

Short Answer Questions

1. Distinguish lifecycle cost from direct maintenance cost.
2. Why should contribution margin usually be used instead of sales revenue when estimating downtime loss?
3. Define planned-work ratio and identify one possible measurement distortion.
4. What are the three components of OEE?
5. Why are sunk costs normally excluded from replacement decisions?

Long Answer Questions

1. Explain how a maintenance department can increase total business value while increasing its direct expenditure.
2. Describe the assumptions and limitations of using Aᵢ = MTBF/(MTBF + MTTR) as an availability measure.
3. Develop a balanced maintenance KPI system for a bottleneck production line and justify each indicator.
4. Compare NPV, payback period and equivalent annual cost as investment-appraisal methods.
5. Discuss the governance and ethical risks associated with maintenance KPI targets.

Analytical / Application Questions

1. A machine produces 25 units per hour with a contribution margin of $16 per unit. Each failure causes 4 hours of downtime and $750 of additional cost. If a programme prevents 10 failures annually and costs $12,000, calculate its simple annual net benefit and state two assumptions that should be tested.
2. Planned production time is 480 minutes, stop time is 60 minutes, ideal cycle time is 0.4 minute per unit, total count is 900 and good count is 810. Calculate availability, performance, quality and OEE.
3. A plant’s schedule compliance rises from 75% to 95% while emergency work, repeat failures and safety-critical overdue tasks also rise. Analyze why this may occur and recommend a revised measurement and governance approach.

Multiple Choice Questions with Answers

Question 1

Which cost is normally excluded from a forward-looking replacement decision because it cannot be changed by the decision?

A. Expected future downtime cost

B. Future energy cost

C. Original purchase price already paid

D. Future disposal cost

Correct Answer: C. Original purchase price already paid

Question 2

A machine produces 40 units per hour with a contribution margin of $18 per unit. What is the lost contribution for 3 hours of unrecovered downtime?

A. $216

B. $1,440

C. $2,160

D. $2,400

Correct Answer: C. $2,160

Question 3

If availability is 0.90, performance is 0.80 and quality is 0.95, what is OEE?

A. 0.684

B. 0.703

C. 0.765

D. 0.855

Correct Answer: B. 0.703

Question 4

Which formula best represents a common planned-work ratio?

A. Planned labour hours ÷ total maintenance labour hours

B. Total failures ÷ operating hours

C. Good count ÷ total count

D. Operating time ÷ calendar time

Correct Answer: A. Planned labour hours ÷ total maintenance labour hours

Question 5

What is the main purpose of discounting future cash flows in lifecycle analysis?

A. To remove all uncertainty from forecasts

B. To recognize the time value of money

C. To increase the recorded salvage value

D. To convert maintenance hours into production hours

Correct Answer: B. To recognize the time value of money

Question 6

A line has planned production time of 400 minutes and stop time of 40 minutes. Its availability under the stated OEE definition is:

A. 80%

B. 85%

C. 90%

D. 95%

Correct Answer: C. 90%

Question 7

Which statement about OEE is correct?

A. It measures maintenance cost only

B. It is the sum of availability, performance and quality

C. It combines availability, performance and quality multiplicatively

D. It is identical to equipment utilization in every context

Correct Answer: C. It combines availability, performance and quality multiplicatively

Question 8

Why is sensitivity analysis important in maintenance investment appraisal?

A. It guarantees a positive NPV

B. It tests how conclusions change when assumptions vary

C. It eliminates the need for engineering judgment

D. It converts all qualitative benefits into cash

Correct Answer: B. It tests how conclusions change when assumptions vary

Question 9

Which practice most directly creates a risk of KPI gaming?

A. Defining data boundaries before measurement

B. Auditing failure codes

C. Reclassifying emergency work as planned after completion

D. Reviewing safety-critical overdue tasks

Correct Answer: C. Reclassifying emergency work as planned after completion

Question 10

A programme costs $20,000 and avoids expected consequences of $27,500 in its first year. Ignoring discounting, what is its net benefit?

A. $7,500

B. $20,000

C. $27,500

D. $47,500

Correct Answer: A. $7,500