Failure Mechanisms and Failure Behaviour

Introduction

Maintenance and reliability engineering begins with a central fact: equipment does not fail for one universal reason. A component may fracture suddenly, lose material gradually through wear, become electrically open-circuit, deform under sustained heat, or stop functioning because an operator or maintainer selected an unsuitable procedure. Effective maintenance therefore requires more than identifying that a failure occurred. Engineers must determine the physical mechanism, the conditions that promoted it, the observable failure behaviour, and the controls that can prevent recurrence.

A failure mechanism is the physical, chemical, thermal, electrical, or human process that causes loss of required function. Failure behaviour describes how failure develops and how its probability changes with time, loading, environment, and operating practice. This distinction is important. Fatigue is a mechanism; an increasing vibration trend before a shaft breaks is failure behaviour. Corrosion is a mechanism; progressive wall-thickness loss followed by leakage is failure behaviour.

This chapter examines the principal mechanisms affecting mechanical and electromechanical systems: fatigue, wear, corrosion, fracture, creep, lubrication failure, thermal damage, electrical failure, and human-induced failure. It then introduces failure modes, degradation patterns, bathtub curves, condition indicators, and practical failure analysis.

Learning objectives

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

1. Distinguish failure modes, mechanisms, causes, and effects.
2. Explain the physical basis of fatigue, wear, corrosion, fracture, creep, and lubrication failure.
3. Recognise thermal, electrical, and human-induced failures in industrial equipment.
4. Interpret degradation patterns and the three regions of a bathtub curve.
5. Select appropriate inspection and maintenance responses for different failure behaviours.
6. Apply a structured analysis to an engineering failure and identify preventive actions.

1. Failure terminology and causal structure

A function is what an asset is required to do, such as transmit 50 kW of power, contain fluid at a specified pressure, or measure temperature within a stated accuracy. A functional failure occurs when the asset cannot perform that function to the required standard. A failure mode is the specific manner in which the function is lost: shaft fracture, bearing seizure, seal leakage, insulation breakdown, or incorrect control signal.

The failure cause is the condition or event that initiates or accelerates the mechanism. Examples include overload, contamination, misalignment, inadequate cooling, incorrect material selection, or an incorrect maintenance procedure. The failure effect is the local or system-level consequence, such as increased vibration, production loss, fire, environmental release, or personnel injury.

A useful causal chain is:

stress or exposure → damage mechanism → detectable symptom → failure mode → system consequence

For example:

shaft misalignment → cyclic bending stress → increasing vibration → fatigue crack and fracture → pump shutdown

The distinction between cause and mechanism prevents superficial corrective action. Replacing a fractured shaft addresses the failure mode, but not necessarily the cause. If misalignment remains, the replacement shaft is likely to fail again.

Stress, strength, and margin

Failure occurs when the applied demand exceeds the capability of a component, or when damage reduces that capability below the applied demand. In a simple deterministic representation:

The margin must be positive for the specified operating condition. Strength and demand are not always constant: temperature may reduce material strength, corrosion may reduce cross-sectional area, and cyclic loading may cause fatigue damage even when every individual load is below the static strength.

In reliability analysis, both strength and demand are often treated as random variables. Failure probability is the probability that demand exceeds strength:

where D is demand, S is strength, and P_f is probability of failure. This relationship is conceptual unless probability distributions and relevant data are available. It highlights why nominal design values alone cannot fully describe reliability.

2. Mechanical failure mechanisms

2.1 Fatigue

Fatigue is progressive, localised damage caused by repeated or fluctuating stress. It can occur at stress levels below the material’s static yield or ultimate strength. Fatigue commonly affects shafts, gears, springs, pressure vessels, welded joints, aircraft structures, and rotating machinery.

Fatigue generally develops through three stages:

1. Crack initiation, often at a notch, surface defect, inclusion, weld toe, keyway, or corrosion pit.
2. Stable crack propagation under repeated loading.
3. Unstable final fracture when the remaining section can no longer carry the load.

Important influencing factors include stress amplitude, mean stress, surface finish, residual stress, geometry, temperature, corrosion, load frequency, and material microstructure. Stress concentrations are especially important. A sharp change in geometry increases local stress and can initiate a crack even when the nominal calculated stress appears acceptable.

For constant-amplitude loading, the stress range and alternating stress are commonly defined as:

where σ_max and σ_min are maximum and minimum stress, Δσ is stress range, and σ_a is stress amplitude. These equations apply to a specified stress cycle and require consistent stress units, such as MPa.

Fatigue life is often represented by an S–N curve, relating stress amplitude to cycles to failure. S–N data are empirical and depend on specimen geometry, surface condition, loading mode, environment, and failure criterion. They should not be transferred uncritically from polished laboratory specimens to corroded or notched industrial components.

Fatigue cracks may be detected by vibration analysis, acoustic emission, magnetic-particle testing, dye penetrant inspection, eddy-current testing, or ultrasonic testing. Prevention includes reducing stress concentration, improving surface finish, controlling alignment and balance, removing corrosive conditions, using suitable residual-stress treatments, and avoiding unnecessary load fluctuations.

2.2 Wear

Wear is progressive material loss or damage caused by relative motion between surfaces. The major forms are:

Wear rate is affected by load, speed, hardness, roughness, temperature, lubrication, contact geometry, and contamination. A commonly used engineering model for adhesive or abrasive wear is the Archard relationship:

where V is wear volume, k is a dimensionless empirical wear coefficient, W is normal load, L is sliding distance, and H is hardness of the softer wearing material. The units are consistent when W is in newtons, L in metres, and H in pascals, producing V in cubic metres. The equation is an approximation, not a universal law; it is most useful for comparative estimates under similar wear regimes.

2.3 Fracture

Fracture is the separation of a material into two or more parts. It may be ductile, involving substantial plastic deformation, or brittle, involving little visible deformation. Ductile fracture often produces a cup-and-cone appearance and absorbs considerable energy. Brittle fracture can propagate rapidly and is promoted by low temperature, high constraint, sharp defects, certain microstructures, and high loading rate.

Fracture mechanics focuses on cracks rather than nominal stress alone. A basic crack-tip parameter for linear elastic conditions is the stress-intensity factor:

where K is stress-intensity factor, Y is a dimensionless geometry factor, σ is nominal applied stress, and a is crack size. This expression assumes predominantly linear-elastic behaviour and a defined crack geometry. Fracture becomes critical when K approaches the material’s fracture toughness under the relevant temperature, loading rate, and environment.

Inspection is essential because a component may retain an apparently adequate gross cross-section while a crack has created a critical local condition. Engineering controls include fracture-tough materials, good fabrication quality, crack-tolerant design, avoidance of sharp notches, and inspection intervals based on crack-growth risk.

2.4 Creep and stress relaxation

Creep is time-dependent permanent deformation under sustained stress, usually significant at elevated temperature. It is important in steam pipes, boilers, gas turbines, furnaces, and high-temperature fasteners. Creep commonly has three stages: primary creep with decreasing rate, secondary creep with approximately steady rate, and tertiary creep with accelerating deformation leading to rupture.

Creep damage depends strongly on temperature, stress, exposure time, grain structure, and environment. A component can fail even when the applied stress is constant and below the short-term yield strength. Consequences include dimensional distortion, loss of clearances, sagging, leakage, and rupture.

Controls include temperature limitation, suitable high-temperature alloys, correct stress levels, thermal monitoring, inspection for deformation and cracking, and life assessment based on operating history. Creep and fatigue may interact in cyclic high-temperature service; this is known as creep–fatigue damage and requires specialised assessment rather than independent consideration of each mechanism.

3. Corrosion and environmental degradation

Corrosion is the deterioration of a material, usually a metal, through chemical or electrochemical reaction with its environment. It may be general and relatively uniform, or highly localised.

Important forms include:

• Uniform corrosion: broadly even thickness reduction.
• Pitting: small, deep cavities caused by local breakdown of a protective film.
• Crevice corrosion: attack in shielded regions where chemistry differs from the bulk environment.
• Galvanic corrosion: accelerated attack when dissimilar metals are electrically connected in an electrolyte.
• Intergranular corrosion: attack along grain boundaries.
• Stress-corrosion cracking: crack growth caused by the combined action of tensile stress and a specific environment.
• Erosion–corrosion: combined fluid-motion damage and chemical attack.

A simplified electrochemical representation is the oxidation of iron:

The released electrons participate in a reduction reaction elsewhere on the surface. Corrosion rate depends on material, electrolyte conductivity, oxygen availability, pH, temperature, flow, surface condition, and electrical potential.

Corrosion prevention may involve material selection, coatings, inhibitors, cathodic protection, drainage, elimination of stagnant crevices, environmental control, and thickness monitoring. Corrosion allowance can provide additional wall thickness, but it is not a substitute for inspection because localised corrosion may penetrate rapidly without consuming the entire allowance uniformly.

4. Lubrication and thermal failure

Lubrication reduces friction, limits wear, removes heat, protects against corrosion, and may seal contaminants. Lubrication failure can result from insufficient quantity, excessive quantity, unsuitable viscosity, contamination, degradation, incorrect relubrication interval, or incompatibility between lubricants.

In fluid-film lubrication, the lubricant film separates surfaces sufficiently to prevent damaging asperity contact. In boundary lubrication, a chemically or physically adsorbed film carries part of the load while some asperity interaction remains. Mixed lubrication lies between these conditions. The appropriate regime depends on speed, viscosity, load, temperature, surface roughness, and geometry.

Common symptoms of lubrication failure include rising temperature, abnormal noise, increased vibration, discoloration, metal particles in oil, rapid wear, and seizure. Overlubrication can be as damaging as underlubrication because churning raises temperature and can force seals open.

Thermal failure occurs when heat generation exceeds heat removal or when temperature exceeds the capability of materials, lubricants, insulation, seals, or electronic components. Thermal expansion can create misalignment or excessive preload. Thermal cycling can produce repeated stresses and fatigue. Thermal protection therefore requires attention to cooling design, ventilation, heat exchangers, operating limits, insulation class, and temperature monitoring.

5. Electrical and human-induced failures

Electrical failures include open circuits, short circuits, insulation breakdown, arcing, contact degradation, overload, phase imbalance, grounding faults, and electronic component failure. Typical initiating conditions are overheating, moisture, contamination, vibration, voltage transients, loose connections, incorrect protection settings, and ageing of insulation.

Electrical degradation is often progressive. Insulation resistance may decline, contact resistance may increase, and motor current or temperature may rise before complete failure. Useful techniques include insulation-resistance testing, polarization or dielectric tests where appropriate, infrared thermography, motor-current signature analysis, power-quality measurement, and visual inspection.

Human-induced failure includes incorrect installation, poor assembly, wrong component selection, inadequate torque, bypassed protection, contamination during maintenance, incorrect software or control settings, and failure to follow an approved procedure. Human error is not limited to operator mistakes; it may be enabled by poor design, inadequate training, unclear interfaces, fatigue, time pressure, weak supervision, or an organisational culture that rewards production over safe maintenance.

Controls should therefore use a systems approach: clear procedures, competency assurance, permit-to-work systems, independent verification of critical steps, error-tolerant design, standardised components, good labelling, and learning from near misses. Blaming an individual without correcting the enabling conditions rarely produces reliable improvement.

6. Failure behaviour and degradation patterns

Failure behaviour describes the way a failure develops over time. Four practical patterns are especially useful:

1. Sudden failure: little or no detectable precursor, as may occur with brittle fracture or an electronic component subjected to a transient.
2. Gradual degradation: a measurable condition declines progressively, as with wear, corrosion, insulation ageing, or bearing damage.
3. Intermittent failure: the function is lost and restored repeatedly, often because of loose connections, thermal expansion, vibration, or marginal electronic components.
4. Functional drift: the asset continues operating but accuracy, efficiency, capacity, or product quality gradually moves outside acceptable limits.

Condition-based maintenance is most effective when degradation produces a measurable potential failure condition before functional failure. Examples include increasing vibration, rising bearing temperature, reduced insulation resistance, increased leakage, oil debris, or declining wall thickness. The interval between detectable potential failure and functional failure is the P–F interval. Inspection intervals must be sufficiently shorter than the P–F interval, with allowance for detection uncertainty, decision time, and corrective action.

A degradation model may be represented conceptually as:

This is not a universal predictive equation; the form of accumulated damage depends on the mechanism. Linear wear, exponential chemical degradation, crack growth, and random shock failure require different models.

7. The bathtub curve and reliability interpretation

The bathtub curve describes a common qualitative pattern for the failure rate of a population of non-repaired items. It has three regions:

1. Infant mortality: initially high failure rate caused by manufacturing defects, installation errors, design weaknesses, or early-life contamination. The rate generally decreases as defective items are removed.
2. Useful life: approximately constant failure rate dominated by random stresses, occasional overloads, and unpredictable events.
3. Wear-out: increasing failure rate caused by ageing, wear, corrosion, fatigue accumulation, insulation deterioration, or other time-dependent damage.

Figure: Qualitative bathtub curve for a population of non-repaired items.

The curve is a model, not a law applying to every asset. Many electronic items show approximately random failures after early defects have been removed and may not exhibit a pronounced wear-out period. Conversely, many mechanical items are strongly age-related. Preventive replacement based solely on calendar age is justified only when evidence shows that age is related to failure risk and that replacement is economically and technically effective.

For a non-repairable item, the reliability function is commonly written:

where T is the random time to failure and R(t) is the probability that the item survives beyond time t. If a continuous lifetime distribution is available, the hazard or instantaneous failure rate is:

where f(t) is the probability density of failure time. This relationship requires a valid lifetime model and applies to a defined population under specified conditions. It should not be interpreted as the probability that a particular item will fail in the next instant without considering its condition and history.

8. Failure analysis and maintenance strategy

A practical failure investigation should proceed in a disciplined sequence:

1. Define the required function and the actual functional failure.
2. Preserve evidence and record operating conditions.
3. Identify the failed component and failure mode.
4. Examine fracture surfaces, wear patterns, deposits, deformation, electrical evidence, and maintenance records.
5. Separate immediate cause, physical mechanism, root cause, and contributing conditions.
6. Test competing hypotheses against evidence.
7. Recommend actions that remove or control the cause.
8. Verify effectiveness through follow-up inspection or performance data.

Tools such as failure mode and effects analysis (FMEA), fault-tree analysis, root-cause analysis, Pareto analysis, oil analysis, vibration analysis, and non-destructive testing support this process. FMEA is particularly useful before failure because it ranks failure modes according to severity, occurrence, and detectability, although numerical rankings should not replace engineering judgement.

Maintenance selection should match failure behaviour:

Worked example: estimating wear volume

A sliding contact carries a normal load of 500 N over a sliding distance of 2,000 m. The softer material has hardness 1.0 × 10⁹ Pa, and an experimentally determined wear coefficient is 2.0 × 10⁻⁴. Using the Archard approximation:

Substitution gives:

The estimated wear volume is therefore 0.20 cm³, because 1 m³ = 10⁶ cm³. This result is only an estimate. It assumes that the wear coefficient remains applicable, the load and sliding distance are correctly represented, and the contact remains within a comparable wear regime. Contamination, changing temperature, lubrication breakdown, or a transition to abrasive wear could make the estimate unreliable.

Applied case: recurring bearing failure in a process pump

Background

An illustrative chemical-processing plant experienced repeated failure of the drive-end bearing on a centrifugal pump. The bearing was replaced approximately every four months, although the supplier’s catalogue life indicated substantially longer service under the stated load.

Problem

The immediate failure mode was bearing spalling followed by overheating and seizure. The plant initially responded by installing a higher-capacity bearing, but the recurrence interval changed very little.

Analysis

Vibration spectra showed a developing bearing defect, while oil samples contained fine metallic particles and water. Inspection found that the bearing housing breather was damaged, allowing humid air and wash water to enter. The grease quantity was also excessive, and shaft alignment was outside the maintenance specification. These observations indicated interacting mechanisms: water contamination promoted corrosion and lubricant degradation; excess grease increased temperature; misalignment produced abnormal contact loading and fatigue wear.

Approach and solution

The investigation team replaced the breather, corrected alignment, established a measured lubrication quantity, improved sealing, and introduced periodic oil or grease-condition checks. Vibration monitoring was used to identify developing defects, and technicians received a standardised installation procedure with independent verification of alignment and lubrication.

Outcome

The bearing operated for more than twice the previous interval, vibration remained within the alert level, and no further water contamination was detected in follow-up samples. The important improvement was not simply a stronger bearing; it was removal of the conditions that accelerated the failure mechanisms.

Learning points

• Repeated replacement without causal analysis is not reliability improvement.
• A failure may involve several interacting mechanisms.
• Lubricant condition can reveal both wear and environmental contamination.
• Installation quality and maintenance procedures are part of equipment reliability.
• Condition monitoring is valuable only when alarms lead to timely decisions.

Challenges, limitations, and professional responsibilities

Failure diagnosis is limited by incomplete records, destroyed evidence, changing operating conditions, and uncertainty in inspection results. A correlation between a symptom and failure does not prove causation. Engineers should distinguish evidence from inference and report uncertainty honestly.

Safety-critical and environmentally hazardous systems require conservative decisions. Regulatory requirements may govern pressure equipment, electrical isolation, lifting equipment, hazardous-area systems, and inspection intervals. Maintenance personnel must use competent inspectors, calibrated instruments, approved materials, and controlled documentation. Bypassing protective devices or accepting undocumented temporary repairs can convert a manageable degradation mechanism into a catastrophic event.

Digital monitoring and artificial intelligence can assist anomaly detection, image-based inspection, and predictive maintenance. However, models depend on representative data and may produce false alarms or miss rare failure modes. Sensor validation, cybersecurity, human review, explainable alarms, and a clear response procedure are essential. Technology should support engineering judgement rather than conceal uncertainty.

Best-practice checklist

• Define function and failure criteria before selecting maintenance tasks.
• Identify the physical mechanism, not merely the failed part.
• Control stress concentration, contamination, temperature, and alignment.
• Match inspection technology to the expected degradation mechanism.
• Use trend data and establish actionable alarm limits.
• Keep accurate records of load, environment, repairs, and observed condition.
• Investigate recurring failures using evidence and competing hypotheses.
• Treat human factors and organisational conditions as engineering variables.
• Verify that corrective actions prevent recurrence.
• Apply risk-based, legally compliant controls to safety- and environment-critical equipment.

Chapter Summary

Mechanical, thermal, electrical, chemical, and human factors can all cause loss of equipment function. Failure analysis is strongest when it separates function, failure mode, mechanism, cause, and effect. Fatigue develops through crack initiation, propagation, and final fracture; wear removes or damages material through interacting surfaces; fracture may be ductile or brittle; creep produces time-dependent deformation at elevated temperature; corrosion causes electrochemical or chemical degradation; and lubrication failure increases friction, heat, and wear. Electrical failures often involve insulation, contacts, overload, moisture, or transients, while human-induced failures are frequently enabled by poor procedures, design, training, or organisational conditions.

Failure behaviour may be sudden, gradual, intermittent, or a drift in performance. The bathtub curve provides a useful population-level model involving infant mortality, useful life, and wear-out, but it must not be applied indiscriminately. Condition-based maintenance is appropriate when degradation is detectable before functional failure; age-based replacement is appropriate only when age-related failure and replacement effectiveness are demonstrated. Effective reliability engineering combines physical understanding, inspection, data, competent maintenance, and disciplined root-cause analysis.

Key Terms

Failure mechanism: Physical, chemical, thermal, electrical, or human process that produces loss of function.

Failure mode: Specific manner in which a function is lost.

Failure cause: Condition or event that initiates or accelerates a failure mechanism.

Failure effect: Consequence of a failure at component, system, or organisational level.

Fatigue: Progressive damage caused by repeated or fluctuating stress.

Wear: Progressive material loss or surface damage caused by relative motion.

Fracture: Separation of a material into two or more parts.

Creep: Time-dependent permanent deformation under sustained stress, usually at high temperature.

Corrosion: Material deterioration through chemical or electrochemical reaction with the environment.

Pitting: Localised corrosion producing small, deep cavities.

Lubrication failure: Loss of effective lubricant protection through shortage, degradation, contamination, or incompatibility.

Stress concentration: Local increase in stress caused by geometry, defects, or discontinuities.

Potential failure: Detectable condition indicating that functional failure is developing.

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

Bathtub curve: Qualitative failure-rate pattern showing infant mortality, useful life, and wear-out.

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

Condition monitoring: Measurement and interpretation of equipment condition indicators.

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

Root cause: Underlying condition whose removal prevents recurrence.

Review Questions

Short Answer Questions

1. Distinguish a failure mode from a failure mechanism.
2. What are the three stages of fatigue failure?
3. Name four major forms of wear.
4. What is meant by the P–F interval?
5. Describe the three regions of a bathtub curve.

Long Answer Questions

1. Explain how stress concentration, surface condition, and corrosion influence fatigue life.
2. Compare ductile and brittle fracture, including their implications for inspection and design.
3. Explain the principal forms of corrosion and appropriate prevention methods.
4. Discuss how lubrication quantity, contamination, viscosity, and temperature influence bearing reliability.
5. Explain why human-induced failure should be analysed as a systems issue rather than simply as individual error.

Analytical / Application Questions

1. A pump bearing shows rising vibration, increased temperature, and metal particles in the lubricant. Identify possible mechanisms, propose an investigation plan, and recommend immediate and long-term actions.
2. A pressure vessel has uniform wall thinning but also several deep pits near a weld. Explain why average thickness alone may be insufficient for assessing integrity and identify suitable inspection methods.
3. An electronic control unit fails randomly during operation, with no consistent age trend. Using failure-behaviour concepts, recommend a maintenance strategy and explain why scheduled replacement may not be effective.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes a failure mechanism from a failure mode?

A. A mechanism is the physical process; a mode is the manner in which function is lost.

B. A mechanism is always an operator error; a mode is always a material defect.

C. A mechanism describes the consequence; a mode describes the maintenance cost.

D. A mechanism is a reliability statistic; a mode is a probability distribution.

Correct Answer: A. A mechanism is the physical process; a mode is the manner in which function is lost.

Question 2

Which condition most directly promotes fatigue crack initiation?

A. A smooth, polished surface under constant temperature

B. A sharp notch producing local stress concentration

C. A component carrying no fluctuating load

D. A lubricant with stable viscosity

Correct Answer: B. A sharp notch producing local stress concentration

Question 3

In the Archard approximation, increasing sliding distance while all other variables remain constant will generally:

A. Decrease wear volume

B. Leave wear volume unchanged

C. Increase wear volume

D. Eliminate abrasive wear

Correct Answer: C. Increase wear volume

Question 4

Which wear mechanism involves hard particles cutting a softer surface?

A. Adhesive wear

B. Abrasive wear

C. Fretting wear

D. Fatigue wear

Correct Answer: B. Abrasive wear

Question 5

What is characteristic of creep?

A. Instantaneous elastic recovery under repeated loading

B. Time-dependent deformation under sustained stress, commonly at elevated temperature

C. Electrochemical dissolution only at room temperature

D. A sudden open circuit caused by a loose wire

Correct Answer: B. Time-dependent deformation under sustained stress, commonly at elevated temperature

Question 6

Which region of the bathtub curve is associated with decreasing early-life failure rate?

A. Infant mortality

B. Useful life

C. Wear-out

D. Post-maintenance region

Correct Answer: A. Infant mortality

Question 7

Which observation is most directly associated with a potential failure condition?

A. A measurable increase in bearing vibration before loss of function

B. A completed fracture with no remaining load path

C. A component removed before installation

D. A machine that has never operated

Correct Answer: A. A measurable increase in bearing vibration before loss of function

Question 8

For the relationship K = Y σ √(πa), which variable represents crack size?

A. Y

B. σ

C. a

D. K

Correct Answer: C. a

Question 9

Which action is most appropriate for a recurring bearing failure caused by water contamination and misalignment?

A. Install a larger bearing without changing conditions

B. Increase grease quantity indefinitely

C. Correct sealing and alignment and control lubrication practices

D. Disable vibration monitoring to avoid false alarms

Correct Answer: C. Correct sealing and alignment and control lubrication practices

Question 10

Why may calendar-based replacement be ineffective for many randomly failing electronic items?

A. Their failure probability always becomes zero with age

B. Their failures may not show a strong age-related wear-out pattern

C. Electronic items cannot be inspected

D. Replacement always increases operating temperature

Correct Answer: B. Their failures may not show a strong age-related wear-out pattern