Maintenance Economics, Analytics and Performance Management

Introduction

Maintenance engineering is not only a technical function. It is an economic decision system concerned with preserving productive capability at an acceptable total cost and risk. A technically effective maintenance task may still be poor management if it consumes excessive labour, causes unnecessary downtime, increases safety exposure, or extends an asset’s life beyond its economic usefulness. Conversely, replacing an apparently repairable component may be economically justified when lost production, energy use, quality losses, and future failure risk are considered.

Maintenance economics connects engineering evidence with financial decision-making. Its central questions include:

• Should an asset be repaired, overhauled, or replaced?
• What maintenance interval minimises total cost and operational risk?
• Which performance indicators accurately describe maintenance effectiveness?
• How should availability, reliability, production, safety, and cost be balanced?
• How can data analytics support decisions without replacing engineering judgement?

This chapter develops life-cycle costing, replacement analysis, cost–benefit analysis, maintenance optimisation, and performance management using commonly applied reliability and financial measures.

Learning objectives

After studying this chapter, a student should be able to:

1. Explain the economic components of an asset’s life-cycle cost.
2. Calculate present value, equivalent annual cost, availability, MTBF, MTTR, and OEE.
3. Apply repair-or-replace and economic replacement analysis.
4. Compare maintenance policies using cost–benefit reasoning.
5. Select and interpret maintenance KPIs without encouraging dysfunctional behaviour.
6. Recognise assumptions, data limitations, uncertainty, and ethical issues in maintenance analytics.

1. The economic role of maintenance

Maintenance creates value by reducing the probability or consequence of functional failure, restoring performance, preserving safety barriers, and controlling degradation. Its cost includes much more than the maintenance department’s budget. A useful economic model distinguishes:

• Direct maintenance cost: labour, materials, contractors, tools, inspection, lubrication, and maintenance management systems.
• Downtime cost: lost contribution margin, lost production, restart losses, overtime, penalties, and customer dissatisfaction.
• Quality cost: scrap, rework, contamination, off-specification product, and inspection.
• Risk cost: expected safety, environmental, legal, and reputational losses.
• Energy and operating cost: inefficient operation caused by wear, fouling, misalignment, leakage, or control degradation.
• Capital cost: purchase, installation, commissioning, modification, and eventual disposal.

Maintenance decisions should therefore be based on total economic consequence rather than maintenance expenditure alone. A reduction in maintenance spending can be harmful if it increases failures and production losses.

A maintenance policy is economically sound when its incremental benefits exceed its incremental costs, subject to safety, legal, environmental, and reliability constraints.

2. Life-cycle costing

Life-cycle costing (LCC) estimates the total cost of owning and operating an asset over a defined life. In its simplest form:

where:

• C₀ = initial acquisition, installation, and commissioning cost;
• Cₘ,ₜ = maintenance cost in year t;
• Cₒ,ₜ = operating cost in year t;
• C_f,ₜ = failure-related cost in year t;
• C_q,ₜ = quality, safety, environmental, or other relevant cost;
• Sₜ = salvage or residual value in year t;
• i = discount rate per period;
• n = analysis horizon in periods.

All monetary quantities must be expressed in the same currency and price basis. If inflation is included in costs, a nominal discount rate should be used. If costs are expressed in constant prices, a real discount rate is appropriate. Mixing nominal and real values produces misleading results.

The formula assumes that costs occur at the end of each period. If a major cost occurs at another time, it should be discounted to that exact time. LCC is a decision model, not a prediction with certainty; uncertain failure frequency, energy prices, production value, and residual life should be represented through scenarios or sensitivity analysis.

Present value and equivalent annual cost

The present value of a future amount Fₜ is:

For a uniform annual cost A over n years, the present value is:

This assumes a constant annual amount and a constant discount rate. To compare alternatives with different lives, convert their present values into an equivalent annual cost (EAC):

EAC includes the annualised capital cost and the annualised present value of operating, maintenance, and failure costs. The alternative with the lower EAC is economically preferable when service, capacity, risk, and time horizon are comparable.

Worked example: life-cycle comparison

Two pumps provide the same duty. Pump A costs $40,000 initially and has an estimated annual operating and maintenance cost of $18,000 for five years. Pump B costs $55,000 initially and costs $14,000 annually for five years. Assume a 10% discount rate, no salvage value, and costs at year-end.

The five-year annuity factor is:

Thus:

• Pump A PV = $40,000 + $18,000 × 3.7908 = $108,234.
• Pump B PV = $55,000 + $14,000 × 3.7908 = $108,071.

Pump B has the slightly lower present cost, despite its higher purchase price. The difference is small, so the decision should also consider reliability, maintainability, safety, energy uncertainty, supplier support, and sensitivity to assumptions. A small numerical advantage should not be treated as a precise truth when input estimates are uncertain.

3. Repair-or-replace and replacement analysis

Repair-or-replace decisions compare the future economics of keeping an existing asset with those of acquiring an alternative. The historical purchase price is a sunk cost and should not influence the future decision. However, the current resale or salvage value of the existing asset is relevant because choosing to retain it sacrifices that value.

Relevant costs of retaining an asset may include:

• immediate repair or overhaul;
• expected future maintenance;
• expected downtime and failure losses;
• energy and quality penalties;
• safety and environmental exposure;
• the opportunity cost of not replacing it;
• residual value at the end of the comparison period.

A basic incremental comparison is:

where NPV means the present value of all relevant future cash flows. The comparison must use the same service level and analysis horizon. If alternatives have unequal lives, use EAC, a common study period, or an equivalent renewal assumption.

Economic life

The economic life is the period that minimises equivalent annual cost, not necessarily the physical life. As an asset ages, capital recovery may fall while maintenance, energy, failure, and quality costs rise. The optimum occurs where the total annualised cost is lowest. It can be shorter than physical life when technology improves rapidly or failure consequences are high.

Replacement analysis is sensitive to:

• discount rate;
• inflation and energy prices;
• expected failure rate;
• production value during downtime;
• technological obsolescence;
• residual value;
• tax and depreciation treatment;
• availability of replacement equipment.

A sensitivity table or scenario analysis should accompany any major replacement recommendation.

4. Cost–benefit analysis and risk-adjusted decisions

Cost–benefit analysis compares the incremental cost of a maintenance intervention with its incremental benefit. Benefits may be reduced failures, increased output, lower energy use, reduced quality loss, improved safety, or avoided environmental damage.

For an intervention with benefits Bₜ and costs Cₜ, the net present value is:

A positive NPV indicates that quantified benefits exceed quantified costs under the stated assumptions. A benefit–cost ratio may also be used:

A BCR greater than 1 indicates economic attractiveness, but NPV is generally preferable when comparing mutually exclusive alternatives because it measures absolute value created.

Failure risk can be represented as expected monetary loss:

where pᵢ is the probability of consequence scenario i and Cᵢ is its monetary consequence. This does not mean that safety can be traded casually for money. Legal requirements, intolerable risks, and ethical duties impose constraints. Economic analysis supports prioritisation among acceptable options; it does not justify violating a safety standard.

5. Optimising maintenance intervals

Maintenance interval optimisation seeks the interval that minimises long-run cost while satisfying reliability and risk requirements. Relevant costs include preventive maintenance cost, corrective maintenance cost, inspection cost, and downtime cost.

A simplified average-cost model for a component replaced preventively every T hours is:

where:

• Cₚ = preventive replacement cost;
• C_f = corrective replacement and consequence cost;
• F(T) = probability of failure before age T;
• C_d(T) = expected downtime-related cost in the cycle;
• T = cycle length.

This is a simplified age-replacement model. It assumes cycles can be represented consistently, costs are stationary, and the failure distribution is known or estimated. Real applications may require renewal-reward models, imperfect maintenance models, condition-based policies, or simulation.

A preventive task is not automatically beneficial. If failures are random and preventive replacement does not reduce failure probability, age-based replacement may add cost without improving reliability. Preventive maintenance is most economically justified when deterioration or age increases failure likelihood, consequences are significant, and the task is effective.

Condition-based maintenance changes the decision from “replace at age T” to “intervene when measured condition indicates unacceptable future risk.” Its economic value depends on detection probability, warning time, false alarms, sensor and analysis cost, and the availability of a planned intervention opportunity.

6. Reliability and maintenance performance measures

Availability

For a repairable item under a simple steady-state approximation:

where MTBF is mean time between failures and MTTR is mean time to repair, both in the same time unit. The expression assumes failures and repairs are represented by stable average rates and that downtime is captured by MTTR. Operational availability may additionally include logistics, waiting, administration, and planned downtime; therefore, the definition must be stated.

A more general operational availability measure is:

Availability is not reliability. A system can have high availability because it is repaired quickly despite frequent failures, or because it fails rarely but takes a long time to repair.

MTBF and failure rate

For N failures over observed operating time Tₒ:

This is an estimate for repairable assets under a defined population and observation period. It should not be interpreted as the guaranteed life of the next item. Under an exponential constant-failure-rate model, the failure rate is:

The exponential assumption is often unsuitable for ageing equipment, infant mortality, or changing operating conditions.

MTTR

The organisation must define whether waiting for parts, permits, access, diagnosis, and testing are included. Narrow “hands-on time” and broader restoration time answer different management questions.

Overall equipment effectiveness

OEE is commonly defined as:

with:

Each fraction is dimensionless. OEE should be calculated using a consistent time hierarchy and agreed definitions. It is not a universal measure of total factory productivity; planned stops, product mix, labour constraints, supply shortages, and demand may lie outside its scope.

Worked example: OEE

A line has 480 minutes of planned production time. It experiences 60 minutes of downtime, produces 800 units, has an ideal cycle time of 0.5 minutes per unit, and produces 760 good units.

• Availability = 420/480 = 0.875.
• Performance = (0.5 × 800)/420 = 0.9524.
• Quality = 760/800 = 0.9500.
• OEE = 0.875 × 0.9524 × 0.9500 = 0.7917, or 79.2%.

The result should lead to loss analysis: downtime, speed loss, and quality loss contribute differently and require different actions.

7. Performance management and KPI governance

A KPI is useful only when its definition, data source, owner, frequency, target, and decision consequence are clear. Common maintenance KPIs include:

A balanced performance system combines leading indicators, such as inspection completion, defect elimination, planning quality, and overdue high-risk work, with lagging indicators, such as failures, downtime, cost, and incidents. Targets should not encourage unsafe deferral, under-reporting, premature component replacement, or manipulation of downtime records.

Performance reviews should ask:

1. What loss occurred?
2. What was its technical and economic consequence?
3. Is the trend real, or caused by changes in coding or measurement?
4. What decision follows from the information?
5. Does the KPI promote the desired behaviour?

8. Applied case: illustrative packaging-line decision

Background

A packaging plant operates a conveyor drive that causes frequent stoppages. The existing drive has a book value of $8,000, a resale value of $5,000, and requires an estimated $12,000 annual maintenance and failure cost. A new drive costs $48,000 installed and is expected to require $4,000 annually. Both alternatives provide the same capacity for four years. The discount rate is 8%, and the existing drive has no residual value after four years. The new drive has a residual value of $10,000.

Problem

Management must decide whether to continue repairing the old drive or replace it. The book value is not relevant to the incremental economic choice, but the $5,000 current resale value is relevant.

Analysis

The four-year annuity factor at 8% is approximately 3.3121.

Retain: future cost = $12,000 × 3.3121 = $39,745. The opportunity cost of retaining the drive is the lost $5,000 resale value, giving relevant PV = $44,745.

Replace: PV = $48,000 + $4,000 × 3.3121 − $10,000/(1.08)⁴.

The discounted residual value is approximately $7,350. Therefore, replacement PV = $48,000 + $13,248 − $7,350 = $53,898.

Approach and solution

On these assumptions, retaining the drive has the lower four-year present cost by approximately $9,153. Replacement is not justified by direct cost alone. However, the analysis must test whether the estimated $12,000 includes production losses and whether the new drive reduces safety risk, energy consumption, quality loss, or future costs beyond year four.

If the old drive causes an additional $4,000 per year of lost contribution margin, its retain PV increases by $13,248 to $57,993, making replacement economically preferable. This illustrates why maintenance economics must include operational consequences rather than maintenance invoices alone.

Outcome and learning points

The case produces a conditional decision, not a universal answer. The correct recommendation is to validate failure and downtime data, include risk and energy effects, run sensitivity analysis, and obtain a decision under approved safety and production constraints. The book value should not drive the decision; relevant future cash flows should.

9. Data analytics, digitalisation, and AI

Computerised maintenance management systems provide work-order history, labour hours, parts usage, failure codes, backlog, and cost information. Sensors and historians can add vibration, temperature, current, pressure, and operating-context data. Analytics can support:

• anomaly detection;
• remaining-useful-life estimation;
• failure-mode classification;
• spare-parts forecasting;
• maintenance-interval optimisation;
• identification of chronic losses;
• scenario and Monte Carlo analysis.

Analytics requires data governance. Asset hierarchies, failure codes, timestamps, operating states, and cost definitions must be consistent. A sophisticated model trained on incomplete or biased records may produce confident but unsafe recommendations. AI should support, not replace, engineering accountability. Models should be validated against known events, monitored for drift, explainable enough for critical decisions, and subject to human approval.

10. Limitations, ethics, and best practice

Economic models contain uncertainty. Failure probabilities may be estimated from too few observations; costs may omit intangible consequences; discount rates may favour short-term decisions; and average KPIs may hide vulnerable subgroups or critical assets. Use ranges, scenarios, confidence limits where appropriate, and explicit assumptions.

Ethically and professionally, maintenance decisions must protect workers, the public, the environment, and product users. Data must not be altered to meet targets. Deferred maintenance should be risk-assessed and authorised. Safety-critical equipment requires compliance with applicable legislation, standards, inspection regimes, and management-of-change processes. Cost reduction is subordinate to mandatory controls and tolerable-risk requirements.

Best practice includes:

• define asset boundaries and KPI formulas;
• separate sunk costs from relevant future costs;
• use common time horizons and discount assumptions;
• include downtime, quality, energy, safety, and environmental consequences;
• validate failure and cost data;
• combine leading and lagging indicators;
• test sensitivity to uncertain assumptions;
• review decisions after actual performance becomes available;
• align incentives with safe, reliable, whole-life performance.

Chapter Summary

Maintenance economics evaluates technical alternatives through their total future consequences. Life-cycle costing includes acquisition, operation, maintenance, failure, risk, and residual-value effects. Present value and equivalent annual cost permit comparison across time and unequal asset lives. Repair-or-replace decisions must exclude sunk costs but include opportunity costs and relevant future cash flows. Cost–benefit and expected-value analysis support intervention choices, while safety and legal constraints remain mandatory.

Maintenance interval optimisation depends on failure behaviour, task effectiveness, downtime consequences, and data quality. Availability, MTBF, MTTR, and OEE are valuable but describe different aspects of performance and require explicit definitions. A balanced KPI system combines technical, economic, operational, and risk indicators. Digital analytics and AI can improve prediction and prioritisation, but only when data governance, validation, explainability, and human accountability are maintained.

Key Terms

Life-cycle cost: Total discounted cost of acquiring, operating, maintaining, and disposing of an asset.

Present value: Current value of a future cash flow after discounting.

Equivalent annual cost: Uniform annual cost economically equivalent to a present value.

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

Opportunity cost: Value of the best alternative forgone.

Repair-or-replace analysis: Comparison of future economics of retaining and replacing an asset.

Cost–benefit analysis: Evaluation of incremental benefits relative to incremental costs.

Expected monetary value: Probability-weighted average consequence.

Economic life: Period producing the minimum equivalent annual cost.

Maintenance interval: Time or usage period between planned interventions.

Availability: Proportion of required time during which an asset is functional.

Reliability: Probability that an item performs its required function for a specified time and condition.

MTBF: Mean operating time between failures of a repairable item.

MTTR: Mean time required to restore an item after failure.

OEE: Product of availability, performance, and quality factors.

Leading indicator: Measure of activity or condition that precedes an outcome.

Lagging indicator: Measure of an outcome that has already occurred.

Data governance: Rules and controls ensuring data quality, consistency, security, and accountability.

Review Questions

Short Answer Questions

1. Why should maintenance decisions be based on life-cycle cost rather than maintenance budget alone?
2. Distinguish a sunk cost from an opportunity cost.
3. Define MTBF and MTTR.
4. What three factors make up OEE?
5. Why must KPI definitions specify the treatment of downtime?

Long Answer Questions

1. Explain the structure and assumptions of a life-cycle costing model.
2. Describe how equivalent annual cost supports comparison of assets with unequal lives.
3. Discuss the economic and technical factors in a repair-or-replace decision.
4. Explain why high availability does not necessarily imply high reliability.
5. Evaluate the benefits and limitations of AI-based maintenance analytics.

Analytical / Application Questions

1. An asset operates for 6,000 hours and experiences 12 failures. Calculate MTBF. If average repair time is 5 hours, estimate steady-state availability using A = MTBF/(MTBF + MTTR).
2. A production line has 450 minutes of planned time, 45 minutes of downtime, an ideal cycle time of 0.4 minutes, 900 total units, and 855 good units. Calculate availability, performance, quality, and OEE.
3. Develop a repair-or-replace decision model for a pump. Identify the relevant cash flows, assumptions, risk factors, sensitivity variables, and KPIs that should be reviewed after implementation.

Multiple Choice Questions with Answers

Question 1

Which cost should generally be excluded from a current repair-or-replace decision?

A. Future corrective maintenance cost

B. Current resale value of the existing asset

C. Historical purchase price already paid

D. Expected production loss during downtime

Correct Answer: C. Historical purchase price already paid

Question 2

An asset has MTBF of 900 hours and MTTR of 9 hours. Using A = MTBF/(MTBF + MTTR), its approximate availability is:

A. 90.0%

B. 98.0%

C. 99.0%

D. 99.9%

Correct Answer: C. 99.0%

Question 3

Which statement best describes equivalent annual cost?

A. The asset’s annual revenue

B. A uniform annual cost equivalent to a present value

C. The annual maintenance budget only

D. The historical depreciation charge

Correct Answer: B. A uniform annual cost equivalent to a present value

Question 4

A line has availability 0.90, performance 0.80, and quality 0.95. Its OEE is:

A. 68.4%

B. 72.0%

C. 76.0%

D. 85.5%

Correct Answer: A. 68.4%

Question 5

Which condition most strongly supports age-based preventive replacement?

A. Failure probability is constant and replacement has no effect

B. Failure probability increases with age and failure consequences are significant

C. The asset has never failed and has no degradation evidence

D. The preventive task is more expensive than the failure consequence

Correct Answer: B. Failure probability increases with age and failure consequences are significant

Question 6

If 20 failures occur during 10,000 operating hours, the estimated MTBF is:

A. 20 hours

B. 200 hours

C. 500 hours

D. 2,000 hours

Correct Answer: B. 200 hours

Question 7

Which is a leading maintenance indicator?

A. Last year’s downtime

B. Number of completed safety-critical inspections before due date

C. Historical scrap cost

D. Previous quarter’s failure count

Correct Answer: B. Number of completed safety-critical inspections before due date

Question 8

A positive NPV for a maintenance intervention indicates that, under the stated assumptions:

A. The intervention has zero risk

B. The intervention creates more quantified present value in benefits than costs

C. The intervention must be legally required

D. The intervention will eliminate all failures

Correct Answer: B. The intervention creates more quantified present value in benefits than costs

Question 9

Which practice most improves the credibility of maintenance analytics?

A. Changing failure codes to meet targets

B. Using a consistent asset hierarchy and validated failure data

C. Ignoring operating context

D. Using only one KPI for all decisions

Correct Answer: B. Using a consistent asset hierarchy and validated failure data

Question 10

Operational availability differs from a narrow uptime measure mainly because it may include:

A. Only purchase price

B. Logistics, waiting, administrative, and planned downtime definitions

C. Only product quality

D. Only operator attendance

Correct Answer: B. Logistics, waiting, administrative, and planned downtime definitions