Reliability, Maintenance and Asset Management

Introduction and Context

Industrial systems create value only when they perform their required functions safely, economically, and consistently. A machine that is unavailable, produces defective output, consumes excessive energy, or creates unacceptable risk is not a productive asset, even if it is technically operational. Reliability engineering studies the probability and causes of successful performance; maintenance engineering restores or preserves that performance; asset management coordinates technical, financial, operational, and risk decisions across the asset life cycle.

These disciplines are closely related but not identical. Reliability is primarily concerned with failure behavior and performance over time. Maintenance is concerned with actions that prevent, detect, or correct deterioration. Asset management is the broader organizational system for realizing value from assets while balancing performance, cost, risk, and opportunity. In machine technology, the combined application of these ideas supports safer plants, higher equipment effectiveness, better spare-parts decisions, and more sustainable production.

Learning Objectives

After studying this chapter, you should be able to:

1. Explain reliability functions, failure distributions, hazard rate, MTBF, MTTR, and availability.
2. Distinguish corrective, preventive, predictive, and reliability-centered maintenance.
3. Apply FMEA and fault-tree analysis to identify and prioritize failure risks.
4. Select replacement policies using economic and risk considerations.
5. Explain total productive maintenance, condition monitoring, and risk-based asset management.
6. Interpret maintenance data and recommend technically and economically defensible actions.

1. Reliability Fundamentals

1.1 Function, failure, and time to failure

A component is reliable only with respect to a specified function, operating environment, performance standard, and time interval. For example, a bearing may be considered failed when vibration exceeds a limit, when temperature becomes unsafe, or when it can no longer support the required load. Failure definitions must therefore be operational and measurable.

Let T be the non-negative random variable representing time to failure. The reliability function is the probability that an item performs its required function for at least time t:

Here, F(t) is the cumulative distribution function of failure time. R(t) is dimensionless and ranges from 0 to 1. The expression assumes a clearly defined population, mission, operating conditions, and failure criterion.

If a probability density function f(t) exists:

The probability of failure within the interval from t₁ to t₂ is F(t₂) − F(t₁), or equivalently R(t₁) − R(t₂).

1.2 Conditional reliability

The probability that an item survives an additional period s, given survival to t, is conditional reliability:

This distinction matters because the future risk of an aged component may differ from that of a new component.

1.3 Hazard rate

The hazard rate, or instantaneous failure rate, describes the conditional tendency to fail at time t, given survival until t:

Its unit is inverse time, such as failures per operating hour. It is not the same as the percentage of a population that fails during a period. The cumulative hazard is:

and therefore:

For a constant hazard rate λ, the exponential model applies:

This model assumes a constant failure rate and is often useful for electronic equipment or the useful-life period of repairable systems. It should not be automatically applied to every machine.

1.4 Common lifetime models

The Weibull distribution is especially useful because its shape parameter can represent different failure behaviors:

where β is the dimensionless shape parameter and η is the characteristic life, in the same time unit as t. For β < 1, early failures dominate and the hazard decreases. For β = 1, the model becomes exponential with constant hazard. For β > 1, wear-out behavior and increasing hazard are indicated.

The Weibull model is valuable for bearings, seals, mechanical components, and life-testing data, but parameter estimates require representative data and suitable statistical methods. A bathtub curve is a useful conceptual model: decreasing hazard during infant mortality, approximately constant hazard during useful life, and increasing hazard during wear-out. Real components do not always follow the complete bathtub pattern.

2. Reliability of Systems

System reliability depends on the logical relationship among components.

For n statistically independent components in series, every component must function:

If all components have identical reliability R(t), then Rₛ(t) = [R(t)]ⁿ. Series systems become less reliable as components are added unless individual reliability is improved.

For n independent, identical components in parallel, the system succeeds if at least one component functions:

For two identical parallel components, Rₚ = 1 − [1 − R]². These formulas assume independence, identical mission conditions, and no common-cause failure. Shared power, environment, software, maintenance error, or a common support system can invalidate the apparent benefit of redundancy.

A k-out-of-n system functions when at least k of n components operate. For independent identical components:

where C(n,j) is the binomial coefficient.

3. Maintainability, MTBF, MTTR, and Availability

Reliability concerns failure-free operation; maintainability concerns the probability that a failed item can be restored within a specified time under stated conditions. If repair time is represented by the random variable M, maintainability is:

For repairable equipment, mean time to repair is the expected active corrective-repair time. In practice, organizations must distinguish active repair time from logistics delay, diagnosis, waiting for permits, and waiting for parts.

Mean time between failures, MTBF, is commonly used for repairable equipment as the average operating time between consecutive inherent failures. Under stable operating conditions and a constant failure rate:

This reciprocal relationship is not universally valid for changing hazard rates, mixed failure modes, or nonstationary systems. For a nonrepairable item, the more precise term is mean time to failure, MTTF.

Mean time to repair is:

where E[M] is the expected repair duration. If a machine operates for U hours and is unavailable for D hours during a period, observed availability is:

For a simple steady-state repairable model with statistically stable alternating up and down periods:

This is inherent or basic availability when only failure and repair times are considered. Operational availability should include delays such as spare-parts acquisition, staffing, setup, and administration. Availability is not the same as utilization: a machine may be available but not scheduled to produce.

Worked Example 1: Availability

A pump has an MTBF of 1,200 operating hours and an MTTR of 8 hours. Assuming stable operating conditions and that MTTR includes the relevant restoration time:

Thus, inherent availability is approximately 99.34%. If logistics add 12 hours to each restoration, the effective downtime becomes 20 hours and operational availability is:

The example shows why reducing waiting time can improve availability even when physical repair time is unchanged.

4. Failure Analysis Methods

4.1 Failure Mode and Effects Analysis

FMEA is a bottom-up method. It begins with component or process functions, identifies possible failure modes, examines their effects and causes, and defines controls and actions. A typical FMEA includes:

Traditional FMEA may rank severity, occurrence, and detection on ordinal scales and calculate a risk priority number, RPN = S × O × D. However, RPN can be misleading because ordinal scores are not true numerical measurements, and different score combinations can produce the same product. Modern practice therefore combines FMEA with explicit risk matrices, criticality analysis, and engineering judgment. High-severity hazards should not be hidden by a low numerical product.

4.2 Fault-tree analysis

Fault-tree analysis is a top-down method. A system-level undesired event is placed at the top, and logical gates connect it to contributing events. An AND gate requires all input events; an OR gate requires at least one input event. For independent events A and B:

and

The independence assumption must be examined carefully. Common-cause failures, shared utilities, and human actions may create dependence. Fault trees are particularly useful for safety-critical systems, protection systems, and identification of minimal cut sets: the smallest combinations of basic events capable of producing the top event.

FMEA asks, “What can fail and what will result?” Fault-tree analysis asks, “What combinations can produce this undesired event?” Used together, they provide complementary perspectives.

5. Maintenance Strategies

Corrective maintenance

Corrective maintenance restores equipment after a detected failure. Run-to-failure can be rational for noncritical, inexpensive, easily replaceable items whose failure has little safety or production consequence. It is not automatically poor practice. For critical assets, however, unplanned corrective work can create secondary damage, overtime, quality losses, and safety exposure.

Preventive maintenance

Preventive maintenance is performed at predetermined intervals or according to prescribed criteria. Time-based replacement is justified when failure probability increases predictably with age and the replacement cost is lower than the consequence of failure. Excessive preventive maintenance can introduce installation errors, disturb healthy components, and waste resources.

Predictive and condition-based maintenance

Predictive maintenance uses measured condition and analysis to determine when intervention is needed. Condition indicators include vibration, temperature, lubrication debris, oil chemistry, motor current, ultrasound, thermography, pressure, and electrical signatures. A condition-monitoring program should specify the failure mode, a measurable indicator, a warning threshold, a critical threshold, the inspection interval, and the required response.

The P–F interval is the time between detectable potential failure P and functional failure F. An inspection interval must be sufficiently shorter than the P–F interval, allowing for measurement uncertainty, planning, parts, and safe intervention. A sensor without a decision rule is not a maintenance strategy.

Reliability-centered maintenance

RCM selects maintenance policies by analyzing functions, functional failures, failure modes, consequences, and technically feasible tasks. Its central question is not “What maintenance can be done?” but “What must be done to preserve the required function?” Outcomes may include condition-based tasks, scheduled restoration, scheduled discard, failure-finding inspections, redesign, or deliberate run-to-failure.

6. Replacement Models

A component may be replaced individually at failure, preventively at age T, or through a block-replacement policy. For an age-replacement policy, a component is replaced at age T if it survives, or immediately at failure if failure occurs first. If Cₚ is planned replacement cost, C_f is failure replacement cost, and c is the cost rate, the long-run cost rate is conceptually:

where X is the random failure time and:

The numerator is expected replacement cost per cycle; the denominator is expected cycle length. This model assumes renewal cycles, consistent costs, and that downtime consequences are either included in C_f or modeled separately. The economically best interval is found by evaluating feasible T values or optimizing the expression.

Replacement decisions must include safety, quality, downtime, environmental effects, labor, spares, and disposal—not only purchase price.

7. Total Productive Maintenance and Overall Equipment Effectiveness

Total productive maintenance treats equipment effectiveness as a shared responsibility of operators, maintainers, engineers, managers, and suppliers. Its practices commonly include autonomous maintenance, planned maintenance, focused improvement, early equipment management, training, quality maintenance, safety and environmental management, and administrative improvement.

Overall equipment effectiveness is often expressed as:

where availability is planned production time minus downtime divided by planned production time; performance compares actual speed with ideal speed; and quality is good units divided by total units. The factors must be defined consistently. OEE is a diagnostic indicator, not a universal measure of business profitability. Increasing OEE on a bottleneck may be valuable, while maximizing OEE on a nonconstraint may increase inventory without improving output.

8. Risk-Based Asset Management

Asset management coordinates asset performance, expenditure, risk, and organizational objectives over the life cycle: need identification, specification, design, procurement, commissioning, operation, maintenance, modification, and disposal. Risk-based prioritization can be expressed conceptually as:

This is a simplified representation. Consequences may include injury, environmental harm, production loss, quality loss, legal exposure, and reputation. Risk registers should state assumptions, uncertainty, existing controls, owners, due dates, and residual risk.

Criticality analysis ranks assets according to consequence and vulnerability. A low-cost valve may be more critical than an expensive machine if its failure can cause a hazardous release. Risk-based maintenance allocates inspection, spares, engineering effort, and improvement funds where risk reduction is greatest.

Governance requires traceable work orders, calibrated instruments, competent personnel, management of change, permit-to-work controls, lockout/tagout, cybersecurity for connected assets, and compliance with applicable safety and environmental regulations. Maintenance records are also evidence of due diligence.

9. Data, Condition Monitoring, and Artificial Intelligence

A useful maintenance data architecture links asset hierarchy, functional location, equipment identity, failure codes, work history, meter readings, condition data, bills of material, and costs. Poor naming conventions and inconsistent failure coding often limit analysis more than the absence of advanced software.

AI and machine learning can support anomaly detection, remaining-useful-life estimation, image inspection, and work-order prioritization. Their use requires representative data, stable sensor quality, explainable decisions where safety is affected, protection against cyberattack, and human review. A model that predicts correlation without identifying a maintainable failure mechanism may generate false alarms or missed failures. AI should augment, not replace, engineering accountability.

10. Illustrative Case: Cooling-Water Pump Reliability Improvement

Background

A food-processing plant operated two centrifugal cooling-water pumps, one duty and one standby. Operators reported repeated bearing failures, fluctuating discharge pressure, and frequent emergency callouts. Production losses were significant because inadequate cooling reduced line speed.

Problem and analysis

The maintenance team first defined the required function: deliver at least 180 m³/h at the specified pressure. FMEA identified bearing lubrication, misalignment, cavitation, seal leakage, and motor electrical faults. A fault tree for “insufficient cooling flow” showed alternative paths involving pump failure, blocked strainers, closed valves, and loss of electrical supply. Investigation found that the standby pump was not routinely tested and that vibration readings were recorded without trend analysis.

The team installed consistent vibration measurement points, corrected alignment, improved lubrication control, added a strainer differential-pressure indicator, and introduced a weekly standby-pump test. Alarm and trip limits were documented, and technicians received training in measurement repeatability.

Outcome

After implementation, early bearing deterioration was detected through a rising vibration trend rather than through functional failure. Planned work replaced the bearing during a scheduled stoppage. Emergency failures declined, standby readiness improved, and maintenance records became comparable across shifts. The improvement did not depend on a single sensor; it combined failure analysis, basic mechanical quality, condition monitoring, operator involvement, and management of spares.

Learning points

1. Reliability improvement begins with a precise functional requirement.
2. Redundancy is ineffective when the standby asset is not maintained or tested.
3. Trends are often more informative than isolated measurements.
4. Root-cause correction is more valuable than repeated component replacement.
5. Availability improvement requires attention to logistics and planning as well as repair skill.

11. Challenges and Best Practices

Common challenges include incomplete failure data, inconsistent definitions of downtime, premature preventive replacement, excessive reliance on MTBF averages, poor spare-parts governance, inadequate planning, common-cause failures, and treating software alarms as diagnoses. Best practice is to define failure and downtime codes, separate active repair from delay, validate instruments, review bad actors, use risk-based task selection, standardize job plans, preserve configuration information, and audit whether completed maintenance actually reduced risk or improved performance.

Maintenance also has ethical dimensions. Workers must not be pressured to bypass safeguards or defer safety-critical work to meet production targets. Asset decisions should consider environmental releases, energy efficiency, lifecycle waste, and the consequences imposed on communities and future users. Professional judgment requires transparent assumptions and escalation when evidence is inadequate.

Chapter Summary

Reliability expresses the probability that an asset performs its required function for a specified time. Reliability functions, failure distributions, hazard rate, and conditional reliability provide the mathematical basis for understanding failure behavior. Series and parallel models show how system configuration affects reliability, while common-cause dependence limits simplistic redundancy claims.

MTBF, MTTF, MTTR, maintainability, and availability describe different aspects of asset performance and must not be interchanged. FMEA provides a bottom-up analysis of failure modes and effects; fault-tree analysis works from an undesired event toward causal combinations. Corrective, preventive, predictive, and reliability-centered maintenance should be selected according to failure behavior, consequence, detectability, and cost.

Replacement models compare lifecycle cost and risk rather than purchase price alone. TPM integrates operators and maintainers, while OEE decomposes equipment effectiveness into availability, performance, and quality. Risk-based asset management aligns lifecycle decisions with safety, environmental, operational, financial, and regulatory objectives. Condition monitoring and AI can improve decisions when supported by sound failure mechanisms, reliable data, competent people, and clear governance.

Key Terms

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

Failure: Loss of ability to perform a specified required function.

Reliability function: Probability of survival beyond time t, R(t).

Hazard rate: Conditional instantaneous tendency to fail at time t.

Cumulative hazard: Integral of hazard rate over time.

Weibull distribution: Flexible life distribution whose shape parameter represents failure behavior.

MTTF: Mean time to failure of a nonrepairable item.

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

MTTR: Mean active time required to restore an item.

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

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

FMEA: Bottom-up analysis of functions, failure modes, effects, causes, and controls.

Fault-tree analysis: Top-down logical analysis of an undesired event.

Condition monitoring: Measurement and interpretation of equipment condition indicators.

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

Reliability-centered maintenance: Process for selecting tasks that preserve required functions.

Total productive maintenance: Organization-wide approach to maximizing equipment effectiveness.

OEE: Product of availability, performance, and quality.

Asset management: Coordinated lifecycle management of asset value, performance, cost, and risk.

Common-cause failure: Failure of multiple redundant elements due to a shared cause.

Review Questions

Short Answer Questions

1. Define the reliability function R(t) and explain what its value means.
2. Distinguish MTBF from MTTR.
3. Why can a parallel configuration fail to provide its expected redundancy benefit?
4. What is the purpose of the P–F interval in condition-based maintenance?
5. State the three factors in OEE.

Long Answer Questions

1. Explain the bathtub concept and compare it with the Weibull hazard model.
2. Compare FMEA and fault-tree analysis, including their direction of reasoning and appropriate applications.
3. Discuss the assumptions and limitations of the availability formula based on MTBF and MTTR.
4. Explain how reliability-centered maintenance selects tasks and avoids unnecessary preventive maintenance.
5. Discuss how asset management integrates technical, financial, safety, environmental, and lifecycle considerations.

Analytical / Application Questions

1. A machine has an MTBF of 900 hours and an MTTR of 15 hours. Calculate its inherent availability and explain one reason operational availability may be lower.
2. A safety function fails if either of two independent sensors fails. If each sensor has reliability 0.98 during the mission, calculate the safety-function reliability and identify the system configuration.
3. A plant repeatedly replaces failed pump bearings but has no vibration trends, alignment records, or lubrication standard. Propose an investigation using FMEA, condition monitoring, and root-cause controls. Justify the sequence of actions.

Multiple Choice Questions with Answers

Question 1

Which expression defines the reliability function for a failure time T?

A. R(t) = P(T ≤ t)

B. R(t) = P(T > t)

C. R(t) = E[T]/t

D. R(t) = h(t) + f(t)

Correct Answer: B. R(t) = P(T > t)

Question 2

For a constant hazard rate λ, which reliability model applies?

A. R(t) = 1 − λt

B. R(t) = λt

C. R(t) = e^(−λt)

D. R(t) = e^(λt)

Correct Answer: C. R(t) = e^(−λt)

Question 3

A repairable pump has MTBF 1,000 h and MTTR 10 h. Its inherent availability is closest to:

A. 90.0%

B. 98.0%

C. 99.0%

D. 99.9%

Correct Answer: C. 99.0%

Question 4

Which method begins with an undesired top event and develops combinations of causes using logical gates?

A. FMEA

B. Fault-tree analysis

C. Autonomous maintenance

D. Age replacement

Correct Answer: B. Fault-tree analysis

Question 5

For two independent components in parallel, the system succeeds when:

A. Both components fail

B. Exactly one specified component operates

C. At least one component operates

D. Neither component requires maintenance

Correct Answer: C. At least one component operates

Question 6

In the Weibull model, a shape parameter β greater than 1 generally indicates:

A. Decreasing hazard and infant mortality

B. Constant hazard only

C. Increasing hazard and wear-out behavior

D. Zero probability of failure

Correct Answer: C. Increasing hazard and wear-out behavior

Question 7

Which maintenance approach uses measured condition and trends to determine intervention timing?

A. Predictive or condition-based maintenance

B. Unplanned corrective maintenance only

C. Fixed-interval disposal without inspection

D. Run-to-failure for every asset

Correct Answer: A. Predictive or condition-based maintenance

Question 8

The P–F interval is the period between:

A. Procurement and commissioning

B. Potential failure detection and functional failure

C. Two consecutive preventive jobs

D. Failure and disposal only

Correct Answer: B. Potential failure detection and functional failure

Question 9

If availability is 0.90, performance is 0.80, and quality is 0.95, OEE is:

A. 0.684

B. 0.720

C. 0.855

D. 2.650

Correct Answer: A. 0.684

Question 10

Which statement best describes risk-based asset management?

A. It prioritizes all assets solely by purchase price

B. It balances asset value, performance, cost, and risk over the lifecycle

C. It eliminates the need for maintenance records

D. It replaces engineering judgment with an unverified AI prediction

Correct Answer: B. It balances asset value, performance, cost, and risk over the lifecycle