Reliability, Maintenance and Asset Management

Introduction and Context

Physical assets—machines, buildings, vehicles, utilities, information systems and infrastructure—create value only when they deliver their required functions. An asset that is technically available but produces defective output, consumes excessive energy or creates unacceptable risk is not performing effectively. Reliability, maintenance and asset management therefore form an integrated discipline concerned with preserving function, controlling risk and optimising lifecycle value.

Reliability asks whether an asset will perform its required function for a specified period under stated conditions. Maintenance concerns the actions used to restore or preserve that function. Asset management extends the perspective across the entire lifecycle: need identification, specification, acquisition, commissioning, operation, maintenance, renewal and disposal. The managerial challenge is not simply to minimise maintenance expenditure. It is to achieve the required performance, safety, quality and sustainability at an acceptable total cost.

In operations management, poor reliability causes downtime, capacity loss, schedule disruption, quality failures, emergency procurement and safety incidents. Conversely, excessive maintenance can waste labour, introduce installation errors and replace components that still have useful life. Effective policy therefore depends on failure consequences, failure behaviour, data quality, technology and organisational capability.

Learning Objectives

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

1. Distinguish reliability, availability, maintainability and asset-management performance.
2. Calculate and interpret common reliability and maintenance measures.
3. Explain failure modes, failure distributions and failure-analysis methods.
4. Compare corrective, preventive, predictive and prescriptive maintenance.
5. Design a risk-based maintenance and condition-monitoring programme.
6. Explain total productive maintenance and the role of operators.
7. Evaluate replacement decisions using lifecycle economics.
8. Integrate maintenance decisions into strategic asset management and governance.

1. Core Concepts and Performance Measures

1.1 Reliability and failure

Reliability is the probability that an item performs its required function without failure for time t under stated operating conditions. If T is the random variable representing time to failure, the reliability function is:

Here, F(t) is the cumulative distribution function of failure time. The equation applies to a defined population, mission and operating environment. Reliability is not an intrinsic permanent property: changing load, temperature, maintenance quality or operating procedure changes the relevant probability.

The probability density of failure time is f(t), and the hazard rate is:

The hazard rate represents the instantaneous failure tendency conditional on survival to time t. It is not the same as the proportion of assets that fail at exactly t.

A common special case is the constant-hazard exponential model:

where λ is a constant failure rate, measured in failures per unit time. Its mean time to failure is:

This model is useful for random failures in a stable operating period, but it is inappropriate when wear-out or early-life defects dominate.

1.2 Availability and maintainability

Availability is the probability that an asset is in a state to perform when required. For a repairable asset under stable conditions, a long-run approximation is:

MTBF is mean time between failures and MTTR is mean time to repair. Both must use the same time unit. This expression assumes a repeating cycle of operating time followed by repair, and it normally excludes planned downtime unless a broader availability definition is specified.

Operational availability may instead be expressed as:

where downtime includes the actual effects of logistics, waiting, diagnosis, repair and restart. In practice, managers must define whether preventive-maintenance time, administrative delays and supply delays are included.

Maintainability is the probability that a failed item can be restored within a specified time under stated conditions. If repair time M is exponentially distributed with repair rate μ:

The mean repair time is 1 ÷ μ. Maintainability is influenced by accessibility, modular design, diagnostics, skills, documentation, tools and spare-parts availability.

1.3 Reliability of systems

For independent components in series, system success requires every component to succeed:

For two independent, identical components arranged in parallel, the system succeeds if at least one succeeds:

These formulas assume statistical independence and clearly defined system success. Common-cause failures, shared utilities, human error and dependency can make these simple structures misleading. A standby system also requires assumptions about switching, standby deterioration and common support systems.

Worked Example 1: Availability improvement

A pump has MTBF of 800 hours and MTTR of 8 hours. Its steady-state availability is:

Thus availability is approximately 99.01%. If improved diagnostics reduce MTTR to 4 hours while MTBF remains unchanged:

Availability rises to approximately 99.50%. The example demonstrates that reducing restoration time can produce a substantial operational benefit even without changing failure frequency.

2. Failure Behaviour and Analysis

2.1 The bathtub curve

Many equipment populations are described conceptually by a bathtub curve with three regions:

1. Early-life failures: declining hazard, often associated with manufacturing defects, installation errors or commissioning problems.
2. Useful life: approximately stable hazard, where random failures dominate.
3. Wear-out: increasing hazard caused by fatigue, corrosion, erosion, ageing or insulation breakdown.

The curve is a population-level model, not a universal law. Modern electronic components may not show classical wear-out, whereas bearings, cutting tools and aircraft structures may exhibit strong age-related deterioration. The curve should not be used to justify time-based replacement without evidence of age-related failure.

2.2 Failure modes and effects analysis

Failure Modes and Effects Analysis (FMEA) examines how an asset can fail, the effects of each failure, its causes and existing controls. A team commonly records severity, occurrence and detectability on ordinal scales. A traditional risk-priority number is:

where S is severity, O is occurrence and D is detectability. RPN is a prioritisation aid, not a probability of loss. Multiplication can hide important differences—for example, a very severe but rare failure may receive the same score as a frequent minor failure. Modern practice supplements RPN with explicit consequence categories, criticality thresholds and scenario analysis.

2.3 Fault-tree and root-cause analysis

Fault Tree Analysis (FTA) begins with a top event, such as “line fails to deliver product,” and logically decomposes it using AND and OR relationships. Under independence assumptions, an OR combination of rare events has approximate probability equal to the sum of individual probabilities; an AND combination has probability equal to their product. These approximations should not be used where common causes or dependencies are material.

Root-Cause Analysis (RCA) investigates why a failure occurred and why organisational controls did not prevent or detect it. Useful techniques include the five-whys method, causal trees, barrier analysis, Pareto analysis and change analysis. A technically correct RCA must avoid stopping at “operator error”; it should examine design, training, workload, procedures, incentives and system conditions.

2.4 Criticality analysis

Criticality ranks assets according to consequences of failure. A practical model considers safety, environmental, regulatory, quality, production, customer and financial consequences, together with likelihood and detectability. Criticality should guide inspection intensity, spare strategy, redundancy, cybersecurity controls and emergency preparedness. It should not automatically determine maintenance frequency: an asset may be critical but condition-monitoring may be more effective than frequent intrusive maintenance.

3. Maintenance Strategies

3.1 Corrective maintenance

Corrective maintenance restores an asset after a detected failure. It may be appropriate for non-critical, inexpensive and easily replaceable items where failure has limited consequences. Unplanned corrective maintenance produces disruption and secondary damage; planned corrective maintenance, sometimes called run-to-failure, deliberately allows failure while preparing labour, tools and spares.

3.2 Preventive maintenance

Preventive maintenance is performed before failure to reduce the probability or consequence of failure. It includes scheduled inspection, lubrication, adjustment, cleaning, calibration, servicing and age-based replacement. Time-based replacement is justified only when failure probability or consequence increases with age and the replacement interval is economically and technically defensible.

Over-maintenance creates risks: unnecessary disassembly, induced defects, wasted resources and exposure to maintenance hazards. Maintenance intervals should therefore be reviewed using failure history, manufacturer information, operating context and condition evidence.

3.3 Predictive and condition-based maintenance

Predictive maintenance uses condition data to estimate deterioration and schedule intervention before functional failure. Typical technologies include vibration analysis, thermography, oil analysis, ultrasonic inspection, motor-current analysis, acoustic emission, pressure trends and electrical testing.

A condition-monitoring programme requires:

• a failure mode that produces a detectable physical symptom;
• a sufficiently long and reliable P–F interval, the time between potential failure and functional failure;
• a measurement method with adequate sensitivity and specificity;
• a decision rule and responsible owner;
• work-order capacity to act on alarms.

A sensor without a response process is not a maintenance system. False alarms create distrust, while missed alarms create risk. Thresholds should be based on engineering limits, baseline behaviour, statistical evidence and consequence—not on arbitrary dashboard colours.

3.4 Prescriptive maintenance and analytics

Prescriptive maintenance combines condition information, failure models, operating context and decision rules to recommend action. Machine learning can detect anomalous patterns or estimate remaining useful life, but it requires representative data, stable labels and controls against model drift. A prediction is not automatically a decision: the organisation must still consider safety, production schedules, spare availability and the cost of intervention.

Comparison of maintenance approaches

4. Reliability-Centred Maintenance and Total Productive Maintenance

4.1 Reliability-Centred Maintenance

Reliability-Centred Maintenance (RCM) is a structured process that asks:

1. What functions must the asset perform?
2. In what ways can it fail to perform those functions?
3. What causes each functional failure?
4. What happens when the failure occurs?
5. Why does the failure matter?
6. What proactive task or default action is appropriate?
7. What should be done if no proactive task is effective?

RCM distinguishes hidden failures, safety or environmental consequences, operational consequences and non-operational consequences. It selects condition-directed, scheduled restoration, scheduled discard, failure-finding or run-to-failure tasks according to failure characteristics and consequences. Its discipline prevents the common error of assigning a generic monthly inspection to every asset.

4.2 Total Productive Maintenance

Total Productive Maintenance (TPM) treats equipment effectiveness as an organisation-wide responsibility. Operators perform autonomous maintenance such as cleaning, inspection, lubrication and early abnormality detection, while specialists handle complex technical work. TPM commonly addresses eight pillars: autonomous maintenance, planned maintenance, focused improvement, quality maintenance, early equipment management, training and education, safety-health-environment, and administrative TPM.

Overall Equipment Effectiveness (OEE) is often calculated as:

Availability is usually planned production time minus downtime, divided by planned production time. Performance compares actual output rate with ideal output rate, and quality is good units divided by total units. The definitions must be stated because organisations sometimes use different planned-time conventions. OEE is a diagnostic measure, not a universal productivity ranking; maximising it can encourage local optimisation or production of unwanted inventory.

5. Maintenance Planning, Scheduling and Work Management

A mature maintenance process moves through identification, screening, planning, scheduling, execution, close-out and learning. A work request should describe the symptom and asset, not prescribe an unverified solution. Planning identifies scope, risk assessment, labour, permits, isolation requirements, tools, drawings and spares. Scheduling matches prepared work to available resources and production windows.

The work order should capture failure mode, cause, remedy, delay codes, labour, materials, readings and follow-up actions. Close-out quality is essential: incomplete records undermine reliability analysis and future planning. Key performance indicators should balance cost, reliability and work quality. Examples include emergency-work percentage, schedule compliance, planned-work percentage, wrench time, mean time to repair, repeat failures, preventive-maintenance compliance, backlog age and maintenance cost per unit of output.

A backlog is not automatically a problem. It becomes a risk when overdue work exceeds the organisation’s ability to control exposure, when critical tasks are deferred or when planners cannot distinguish ready work from technically incomplete requests.

6. Lifecycle Asset Management

Asset management aligns asset decisions with organisational objectives. The lifecycle includes concept and business case, specification, design, procurement, installation, commissioning, operation, maintenance, renewal and disposal. Decisions made during design strongly influence later maintainability, energy use, spare requirements, safety and data availability. The purchase price is therefore only one component of value.

A lifecycle cost model may be expressed as:

C₀ is initial acquisition and installation cost, C_t is cost incurred in year t, r is the discount rate per period, n is the analysis horizon, and S_n is residual or salvage value at the end. All monetary values must use consistent price bases and the same time period. The model may include energy, labour, spares, downtime, compliance, risk and disposal costs. Uncertainty should be tested using sensitivity analysis or scenario analysis rather than hidden in a single precise estimate.

Replacement decisions

A replacement decision compares the future cost and risk of retaining an asset with the equivalent cost of alternatives. Important factors include escalating maintenance cost, declining efficiency, obsolescence, safety exposure, unavailable spares, capacity constraints and technology compatibility. A low purchase price may be inferior if the alternative has greater downtime or energy cost.

For mutually exclusive alternatives with unequal lives, analysts can compare present values over a common period or calculate equivalent annual cost. For a constant annual equivalent and end-of-life horizon n:

PV is the present value of the alternative, r is the periodic discount rate and n is the number of periods. This conversion assumes comparable service and a constant discount rate. It should not replace strategic judgement where capacity, risk or technological flexibility differs.

7. Governance, Safety and Digital Transformation

Maintenance decisions can affect workers, the public, the environment and service continuity. Permit-to-work, lockout/tagout, confined-space controls, lifting plans, electrical safety, contractor management and management of change are governance mechanisms, not administrative inconveniences. Deferring maintenance may be economically attractive in the short term but unacceptable when it defeats a safety barrier or regulatory requirement.

Asset data also creates cybersecurity and privacy concerns. Connected sensors, remote access and cloud platforms expand the attack surface of operational technology. Governance should define ownership, access control, patching, network segmentation, backup, incident response and vendor responsibilities. Digital twins can support simulation and diagnosis, but their usefulness depends on model validity and data quality.

Artificial intelligence should be introduced through a controlled use case: define the decision, establish a baseline, validate against independent data, monitor false positives and negatives, document human accountability, and periodically review model drift. Explainability is especially important when a recommendation could affect safety or regulatory compliance.

8. Illustrative Case: Reliability Improvement in a Food-Processing Plant

Background

A fictional food-processing plant experiences frequent stoppages on a packaging line. The line’s filler, conveyor and sealer are interconnected; a short failure at one station stops downstream production. Maintenance records contain many emergency work orders but inconsistent failure descriptions.

Problem

The plant reports 96% line availability, but customer service observes missed shipments. Investigation shows that the figure excludes micro-stops and changeover losses. Repeated sealer-temperature faults account for 35% of recorded downtime, while operators sometimes bypass alarms to maintain output.

Analysis

A cross-functional team maps the functional requirements and performs an FMEA. The critical sealer failure mode is unstable temperature caused by heater degradation, loose electrical connections and contaminated sensors. Thermography and resistance testing can detect deterioration before functional failure. The team also finds that the same alarm code represents several different causes, making trend analysis unreliable.

The team establishes a common failure taxonomy, separates planned changeovers from unplanned downtime, and calculates availability using agreed definitions. It introduces a daily operator inspection, weekly sensor verification, monthly thermographic inspection and a planned replacement criterion based on condition rather than calendar age. The maintenance planner creates a standard job plan with isolation steps, test limits, parts and restart checks.

Approach and solution

Operators receive training in cleaning, visual inspection and abnormality tagging. A technician validates sensor readings and records actual failure mode and cause. A small buffer is added only after a capacity analysis confirms that it will not conceal recurring problems. The reliability engineer reviews repeat failures weekly using Pareto analysis and RCA.

Outcome

After three months, emergency work on the sealer falls by 50%, MTTR declines because the correct spare assemblies are kitted, and the proportion of planned work increases. Reported availability initially falls because the new definition includes previously hidden micro-stops; however, effective output and on-time shipment improve. This illustrates why measurement integrity matters more than a superficially favourable KPI.

Learning points

1. Reliability improvement begins with functional definitions and trustworthy data.
2. Condition monitoring is valuable only when linked to an executable work process.
3. Operator involvement can detect deterioration early without transferring specialist responsibility.
4. A KPI can worsen while operational performance improves if measurement becomes more complete.
5. Root-cause elimination and maintainability improvement should accompany preventive tasks.

9. Challenges and Best Practices

Common implementation challenges include unreliable asset registers, inconsistent failure coding, poor spare-part identification, weak planner capability, production-maintenance conflict, excessive preventive tasks and lack of executive sponsorship. Organisations may also adopt sensors before defining decisions, or purchase analytical software without data governance.

Best practice is to begin with business-critical functions and a small number of high-consequence failure modes. Define measures before collecting data; standardise asset hierarchies and failure taxonomies; combine engineering judgement with evidence; review maintenance plans after significant failures; involve operations, maintenance, engineering, finance, safety and supply chain; and treat design-for-maintainability as a procurement requirement. Governance should make risk acceptance explicit and ensure that cost reductions do not silently transfer risk to workers, customers or the environment.

Chapter Summary

Reliability is the probability of performing a required function without failure for a specified time and condition. Availability combines reliability and restoration capability, while maintainability concerns the probability of restoring function within a specified period. Reliability measures require clear definitions, appropriate assumptions and consistent data.

Failure analysis uses tools such as FMEA, FTA, RCA and criticality analysis to connect failure modes with consequences and controls. Corrective, preventive, predictive, prescriptive and proactive maintenance are not competitors in every situation; an effective programme selects among them according to failure behaviour, consequence, detectability and cost. RCM provides a disciplined task-selection method, while TPM extends responsibility for basic equipment care across the organisation.

Lifecycle asset management integrates performance, risk, cost, safety and sustainability from design through disposal. Replacement decisions should include discounted lifecycle costs, uncertainty and service equivalence rather than acquisition price alone. Digital monitoring and AI can improve decisions but require valid data, cybersecurity, human accountability and action processes. The central managerial principle is to optimise asset value and risk—not merely maintenance expenditure.

Key Terms

Reliability: Probability that an asset performs its required function for a specified time under stated conditions.

Failure: Loss of ability to perform a required function.

Availability: Probability that an asset is ready to perform when required.

Maintainability: Probability of restoring an asset within a specified time.

MTBF: Mean time between failures for a repairable asset.

MTTF: Mean time to failure for a non-repairable item.

MTTR: Mean time to repair or restore function.

Hazard rate: Instantaneous failure tendency conditional on survival to a given time.

Failure mode: Specific manner in which an asset fails.

FMEA: Structured analysis of failure modes, causes, effects and controls.

Fault Tree Analysis: Top-down logical analysis of combinations causing a defined top event.

Root-Cause Analysis: Investigation of technical and organisational causes of failure.

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

Preventive maintenance: Maintenance performed before failure according to time, usage or condition.

Predictive maintenance: Maintenance planned using measured evidence of asset condition.

P–F interval: Time between detectable potential failure and functional failure.

Reliability-Centred Maintenance: Function- and consequence-based method for selecting maintenance policy.

Total Productive Maintenance: Organisation-wide approach to maximising equipment effectiveness.

Lifecycle cost: Total discounted cost of owning, operating, maintaining and disposing of an asset.

Overall Equipment Effectiveness: Product of availability, performance and quality factors.

Review Questions

Short Answer Questions

1. Distinguish reliability from availability and maintainability.
2. What assumptions underlie the exponential reliability model?
3. Explain why time-based replacement is not suitable for every failure mode.
4. What is the purpose of an FMEA?
5. Define the P–F interval and explain its maintenance significance.

Long Answer Questions

1. Compare corrective, preventive, predictive and prescriptive maintenance, including appropriate applications and limitations.
2. Explain how RCM and TPM contribute differently to maintenance performance.
3. Discuss the role of lifecycle costing in asset acquisition and replacement decisions.
4. Evaluate the strengths and weaknesses of OEE as an operations-management measure.
5. Design a governance framework for using AI-based condition monitoring in a safety-critical operation.

Analytical / Application Questions

1. An asset has MTBF of 1,200 hours and MTTR of 12 hours. Calculate its steady-state availability and explain two practical ways to improve it.
2. A plant’s failure data show frequent low-consequence motor failures and rare failures of a safety interlock. Propose a criticality-based maintenance policy and justify the difference in treatment.
3. A manager proposes installing sensors on every asset. Develop a decision framework that determines which assets should receive condition monitoring and which should remain under corrective or preventive maintenance.

Multiple Choice Questions with Answers

Question 1

Which statement best defines reliability?

A. The percentage of planned time an asset is scheduled to operate

B. The probability that an asset performs its required function for a specified time under stated conditions

C. The average time required to repair an asset

D. The cost of maintaining an asset over its life

Correct Answer: B. The probability that an asset performs its required function for a specified time under stated conditions

Question 2

A repairable asset has an MTBF of 800 hours and an MTTR of 8 hours. Its steady-state availability is approximately:

A. 80.0%

B. 98.01%

C. 99.01%

D. 99.90%

Correct Answer: C. 99.01%

Question 3

Which maintenance approach is most directly based on measured deterioration?

A. Run-to-failure corrective maintenance

B. Calendar-based preventive maintenance

C. Predictive condition-based maintenance

D. Unplanned emergency maintenance

Correct Answer: C. Predictive condition-based maintenance

Question 4

In an FMEA, the traditional RPN is calculated as:

A. Severity + occurrence + detectability

B. Severity × occurrence × detectability

C. Occurrence ÷ severity × detectability

D. Detectability − severity − occurrence

Correct Answer: B. Severity × occurrence × detectability

Question 5

For two independent identical components operating in parallel, system success occurs when:

A. Both components fail

B. Exactly one component fails

C. At least one component succeeds

D. The components have identical MTTR values

Correct Answer: C. At least one component succeeds

Question 6

Which statement about the P–F interval is correct?

A. It is the time between acquisition and disposal

B. It is the time between potential failure detection and functional failure

C. It is always equal to MTBF

D. It measures the cost of a preventive task

Correct Answer: B. It is the time between potential failure detection and functional failure

Question 7

The principal purpose of Reliability-Centred Maintenance is to:

A. Assign the same maintenance interval to every asset

B. Select maintenance policies based on functions, failure modes and consequences

C. Eliminate the need for operator involvement

D. Maximise the number of preventive work orders

Correct Answer: B. Select maintenance policies based on functions, failure modes and consequences

Question 8

OEE is conventionally calculated as:

A. Availability × performance × quality

B. MTBF × MTTR × quality

C. Downtime ÷ output × cost

D. Reliability + maintainability + safety

Correct Answer: A. Availability × performance × quality

Question 9

Which item is normally included in a lifecycle-cost analysis?

A. Only the initial purchase price

B. Only annual maintenance labour

C. Acquisition, operating, maintenance, risk-related and disposal costs, appropriately discounted

D. Only the salvage value

Correct Answer: C. Acquisition, operating, maintenance, risk-related and disposal costs, appropriately discounted

Question 10

A key governance requirement for AI-based maintenance recommendations is:

A. Removing all human review

B. Using the largest possible number of sensors regardless of purpose

C. Validating performance, monitoring model drift and assigning human accountability

D. Treating every anomaly as proof of imminent failure

Correct Answer: C. Validating performance, monitoring model drift and assigning human accountability