Welding Defects, Testing and Failure Analysis

Introduction

A welded joint is a manufactured component whose performance depends on materials, joint design, thermal history, process control, workmanship, inspection, and service environment. A weld may appear sound yet contain internal discontinuities, residual stresses, metallurgical damage, or geometric deviations that reduce its fitness for service. Conversely, a discontinuity is not automatically a defect: it becomes a defect only when it exceeds an applicable acceptance criterion or makes the component unsuitable for its intended function.

Welding inspection therefore has two related purposes. First, it detects and characterizes discontinuities. Second, it supplies evidence for engineering decisions about acceptance, repair, continued operation, or replacement. Reliable decisions require a planned combination of visual, non-destructive, destructive, and metallurgical examination, supported by knowledge of welding metallurgy, fracture mechanics, standards, and service conditions.

Learning Objectives

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

• distinguish discontinuities, defects, damage, and failure;
• explain the mechanisms and typical causes of major weld defects;
• select appropriate non-destructive and destructive tests;
• interpret basic visual, penetrant, magnetic-particle, ultrasonic, radiographic, and leak-test results;
• use metallography and fracture evidence in root-cause analysis;
• distinguish repairable defects from conditions requiring replacement or engineering assessment; and
• evaluate fitness for service while considering safety, standards, documentation, and ethics.

1. Discontinuities, Defects, and Failure

A discontinuity is an interruption in the expected uniformity of a material. Examples include a pore, lack of fusion, slag inclusion, undercut, crack, or geometric misalignment. A defect is a discontinuity that violates a specified requirement. Acceptance depends on the governing code, material, loading, location, size, orientation, and service consequence.

A useful classification is:

A welding failure occurs when a component can no longer perform its required function. Failure may mean fracture, leakage, excessive deformation, loss of dimensional accuracy, unacceptable corrosion, or progressive damage. The visible fracture is often only the final event. Root-cause analysis must ask why the discontinuity formed, why it was not detected, why it was accepted, and why service conditions allowed it to become critical.

2. Mechanisms and Causes of Common Weld Defects

2.1 Cracks

Cracks are planar separations and are normally treated as highly serious because their sharp tips produce strong stress concentration. Important types include:

• Hot or solidification cracks, formed during final solidification when a weak liquid film or segregated constituent cannot accommodate contraction. High restraint, unsuitable filler composition, excessive dilution, and unfavorable bead shape increase risk.
• Cold cracking, often associated with hydrogen, a susceptible microstructure, and tensile restraint. It may occur after cooling and can be delayed for hours.
• Hydrogen-assisted cracking, promoted by hydrogen-containing consumables or contamination, hard heat-affected zones, and restraint. Preheating, controlled interpass temperature, low-hydrogen practice, and post-weld hydrogen control reduce risk.
• Lamellar tearing, a through-thickness cracking phenomenon in susceptible rolled plate, generally near highly restrained welds. It is associated with through-thickness strain and non-metallic inclusions.
• Crater cracks, formed when an arc is terminated without filling the weld crater.
• Reheat cracks, which may occur in some alloy steels during post-weld heat treatment or elevated-temperature service.

Cracks should not normally be removed merely by grinding and covered with new weld metal. Their full extent must be established, the crack removed, the excavation examined, and the repair procedure qualified and approved.

2.2 Lack of Fusion and Incomplete Penetration

Lack of fusion is failure to achieve coalescence between weld metal and base metal or between weld passes. Causes include low heat input, excessive travel speed, incorrect electrode angle, inadequate cleaning, oxide films, and an unfavorable joint profile.

Incomplete penetration occurs when the weld metal does not extend through the intended joint root. Root gap, root face, alignment, current, electrode manipulation, and access are influential. These planar conditions can be more dangerous than rounded porosity because they may behave like pre-existing cracks.

2.3 Porosity and Inclusions

Porosity consists of gas cavities trapped during solidification. Moisture, oil, paint, rust, inadequate shielding, excessive gas flow, drafts, long arc length, and contaminated consumables are common causes. Excessive shielding-gas flow can itself create turbulence and entrain air.

Slag inclusions are non-metallic particles trapped between weld passes or at the fusion boundary. Poor interpass cleaning, incorrect electrode angle, low current, an unsuitable bead sequence, and narrow grooves can produce them. Tungsten inclusions may occur in gas tungsten arc welding when the electrode contacts the weld pool or is operated at excessive current.

2.4 Geometric Defects

Undercut is a groove melted into the base metal adjacent to the weld and left unfilled. It may reduce load-bearing thickness and act as a fatigue notch. Overlap is weld metal flowing onto the base metal without adequate fusion. Underfill leaves the weld face below the intended contour. Excessive reinforcement may increase stress concentration rather than improve strength. Distortion and angular misalignment result from non-uniform thermal expansion and contraction.

Control measures include correct parameters, qualified procedures, suitable joint preparation, balanced welding sequences, restraint management, tack-weld control, and dimensional inspection before and after welding.

2.5 Metallurgical Damage

The heat-affected zone (HAZ) may undergo grain growth, hardening, softening, sensitization, precipitation, or transformation. In carbon and low-alloy steels, rapid cooling can produce hard microstructures susceptible to hydrogen cracking. In stainless steels, unsuitable thermal exposure can promote sensitization and intergranular corrosion. In aluminium alloys, welding may reduce strength in the HAZ even when the weld appears geometrically satisfactory.

Heat input is often estimated from electrical power and travel speed. For a simplified direct-current arc calculation:

where Q is heat input per unit length, η is arc efficiency, V is voltage in volts, I is current in amperes, and v is travel speed in millimetres per second. The units of Q are joules per millimetre when V·A is joules per second and v is millimetres per second. The equation is an engineering estimate; actual thermal cycles also depend on joint geometry, preheat, plate thickness, interpass temperature, and heat losses. It should not be used as a substitute for a qualified welding procedure.

3. Inspection Planning and Test Selection

Inspection should begin before welding. A practical inspection plan identifies material certificates, joint configuration, welding procedure, welder qualification, consumables, preheat, interpass temperature, inspection hold points, acceptance criteria, repair rules, and records. The test method must match the defect sought, component geometry, material, accessibility, and required sensitivity.

No method detects every condition. Complementary methods are often necessary—for example, VT plus MT for a ferromagnetic fillet weld, or VT plus UT for a thick pressure-vessel butt weld.

4. Non-Destructive Testing

4.1 Visual Testing

Visual testing is the first and most economical examination. It includes examination of joint preparation, fit-up, root condition, weld sequence, interpass cleaning, final profile, arc strikes, spatter, distortion, and dimensions. Adequate lighting, access, calibrated gauges, magnification where appropriate, and documented acceptance criteria are essential.

Visual inspection should occur before welding, during welding, and after welding. A final visual examination cannot reveal a root defect that was inaccessible and covered by subsequent passes.

4.2 Liquid Penetrant Testing

Penetrant testing uses capillary action. After cleaning and drying, a penetrant is applied and allowed to dwell. Excess penetrant is removed from the surface, developer is applied, and indications are examined under visible or ultraviolet illumination depending on the system.

Linear indications are generally more significant than rounded indications, although acceptance depends on the code. PT is effective for surface-breaking cracks, laps, seams, and porosity in metals and other non-porous materials. It cannot detect a flaw sealed below the surface, and rough, porous, dirty, or excessively absorbent surfaces can create misleading indications. Post-cleaning is important because residual chemicals may affect service or later welding.

4.3 Magnetic-Particle Testing

MT magnetizes a ferromagnetic component. A surface-breaking or near-surface discontinuity interrupts magnetic flux, producing leakage that attracts magnetic particles. The strongest response generally occurs when the discontinuity is approximately perpendicular to the magnetic field; therefore, two magnetization directions are often used.

The method is rapid and highly sensitive to cracks, but it is limited to ferromagnetic materials and may require demagnetization and cleaning. Coatings, geometry, lift-off, excessive background magnetism, and inadequate field strength affect results. Interpretation must distinguish relevant indications from non-relevant responses caused by keyways, abrupt geometry, or magnetic writing.

4.4 Ultrasonic Testing

In conventional pulse-echo UT, a transducer introduces high-frequency sound into the component. Reflections from interfaces or discontinuities return to the transducer. If sound velocity is known, the approximate reflector distance along the beam can be related to measured time:

where d is the distance to a reflector along the sound path, c is the material sound velocity, and t is the round-trip travel time. The factor 2 applies to a pulse that travels to the reflector and back. This relation assumes a known, approximately constant velocity and a clearly identified echo; probe delay, refraction, attenuation, geometry, and beam spread must be accounted for in practice.

UT calibration uses reference blocks or reference reflectors to establish range, sensitivity, and response. Angle-beam probes are commonly used for weld examination because they direct sound toward the fusion zone and root. Indication amplitude alone does not uniquely determine flaw size. Position, length, orientation, echo behavior, scanning coverage, and procedure requirements must be considered. Planar defects oriented parallel to the beam may produce weak reflections, while geometry can create false indications.

4.5 Radiographic Testing

RT records differences in transmitted radiation caused by variations in thickness, density, and attenuation. It is particularly useful for volumetric flaws such as porosity, slag, and incomplete penetration. Radiographic sensitivity depends on source energy, geometric unsharpness, detector or film characteristics, exposure, thickness, and image quality indicators.

Planar flaws may be difficult to detect when their plane is nearly parallel to the radiation beam. RT also involves ionizing radiation, requiring controlled areas, trained personnel, dosimetry, regulatory compliance, and verified exposure procedures. Digital radiography can improve workflow and image processing, but it does not remove the need for competent interpretation.

4.6 Leak Testing

Leak testing evaluates whether a pressure boundary permits fluid passage. Methods include hydrostatic or pneumatic pressure testing, bubble testing, vacuum-box testing, tracer-gas testing, and mass-spectrometer methods. A pressure test demonstrates containment under specified conditions; it does not prove the absence of fatigue-sensitive cracks or every internal discontinuity.

For a simple constant-volume ideal-gas pressure decay test at constant temperature, pressure loss may indicate leakage, but temperature change, instrument drift, trapped air dissolution, and component deformation can produce apparent pressure changes. Consequently, acceptance limits and stabilization periods must be specified by the applicable code or procedure. Pneumatic testing stores much more hazardous energy than hydrostatic testing and requires special risk controls.

5. Destructive Testing and Metallography

Destructive tests are used for procedure qualification, material characterization, failure investigation, and validation of repairs. Common tests include tensile testing, bend testing, impact testing, hardness testing, macro-examination, and fracture testing.

A tensile specimen may provide yield strength, ultimate tensile strength, elongation, and reduction of area. Bend tests expose lack of fusion, cracks, and ductility limitations. Impact tests assess notch toughness at specified temperatures, although results are sensitive to specimen orientation, notch geometry, and sampling location. Hardness traverses across weld metal, HAZ, and base metal can reveal hard zones, softening, or unsuitable heat treatment, but hardness alone is not a complete measure of toughness or service performance.

Metallographic Examination

A metallographic section is cut, mounted, ground, polished, and etched to reveal weld-metal solidification, fusion boundaries, HAZ width, grain structure, inclusions, porosity, cracks, and phase transformations. Macro-etching is useful for penetration, bead sequence, fusion, and gross defects. Micro-examination can identify martensitic regions, carbide precipitation, sensitization-related features, non-metallic inclusions, and hydrogen-related fracture morphology when suitable evidence is preserved.

Sampling must be planned carefully. Cutting can destroy crack tips, polishing can smear soft phases, and overheating can alter microstructure. Photographs, orientation marks, fracture surfaces, and chain of custody should be recorded before sectioning.

6. Failure Analysis and Root Cause

Failure analysis should proceed from evidence to hypotheses, not from a preferred explanation to selected evidence. A disciplined sequence is:

1. Define the failure and its safety significance.
2. Secure the component and preserve fracture surfaces.
3. Collect design, material, welding, inspection, and service records.
4. Map the damage and identify initiation, propagation, and final-fracture regions.
5. Perform VT and suitable NDT before destructive sampling.
6. Examine chemistry, hardness, tensile or impact properties, and microstructure as required.
7. Reconstruct loads, temperatures, environment, weld history, and repair history.
8. Test competing hypotheses against physical evidence.
9. Identify immediate, contributing, and systemic causes.
10. Recommend corrective action and verify its effectiveness.

A root cause is deeper than “welder error.” Possible systemic causes include an inadequate joint design, an unqualified procedure, poor consumable storage, missing inspection hold points, an unsuitable acceptance standard, production pressure, inadequate training, or inaccurate service-load assumptions.

Common Failure Mechanisms

• Fatigue: repeated stress cycles initiate cracks at toes, roots, undercut, weld starts, or attachments. Crack growth is often progressive and may show beach marks.
• Brittle fracture: rapid fracture with limited plastic deformation, promoted by low temperature, high restraint, high hardness, notch-like defects, and inadequate toughness.
• Ductile overload: substantial plastic deformation and a rough, fibrous fracture surface before final separation.
• Stress-corrosion cracking: simultaneous action of tensile stress and a specific corrosive environment.
• Creep: time-dependent deformation and damage at elevated temperature, often involving cavitation and grain-boundary cracking.
• Hydrogen-assisted cracking: delayed cracking in susceptible material under tensile stress, frequently associated with hard HAZ microstructures.

7. Fitness for Service and Repair Decisions

Acceptance and fitness for service are related but not identical. A fabrication code may reject a weld because it exceeds a construction limit, while an engineering assessment may determine whether an existing indication can safely remain under defined operating conditions. Such assessments require reliable flaw dimensions, material toughness, stress history, crack-growth behavior, inspection uncertainty, and consequences of failure.

A repair decision should consider:

• defect type, size, orientation, and location;
• static, cyclic, impact, thermal, and pressure loading;
• material toughness and fracture behavior;
• operating temperature and environment;
• accessibility and feasibility of excavation;
• risk introduced by repair welding, including residual stress and distortion;
• post-repair NDT and possible heat treatment; and
• applicable construction, repair, and occupational-safety requirements.

Grinding may be acceptable for shallow, non-planar surface imperfections when minimum thickness and contour are maintained. Excavation and rewelding require an approved repair procedure. Repeated repairs can alter metallurgy and dimensions; therefore, repair history must be recorded.

8. Applied Case: Delayed Cracking in a Structural Bracket

The following is an illustrative case based on common engineering practice.

Background

A low-alloy steel support bracket welded to a pressure-system frame developed a crack near the weld toe two days after fabrication. The bracket carried fluctuating loads and had been welded outdoors during cold weather.

Problem

The crack was visible at the toe but its length and depth were unknown. Production records showed that the electrode holding oven had been out of service and that preheat temperature had not been recorded.

Analysis

VT identified a narrow linear indication. MT showed that it extended farther than the visible portion. UT confirmed a crack-like reflector entering the HAZ. Hardness testing found a high-hardness region adjacent to the fusion boundary. Metallography revealed a hardened HAZ and a crack passing through that region. The evidence supported hydrogen-assisted cracking rather than fatigue as the initiation mechanism. The outdoor temperature, unverified consumable storage, high restraint, and missing preheat control were contributing factors.

Approach and Solution

The bracket was isolated and unloaded. The crack was completely excavated, verified by MT, and repaired under a qualified low-hydrogen procedure. Electrodes were dried and stored according to the consumable manufacturer’s instructions. Preheat and interpass temperatures were measured and recorded. Controlled cooling was used, followed by the specified post-weld treatment. VT, MT, and UT were repeated after repair, and hardness was checked against the procedure requirement.

Outcome

No crack-like indications remained, hardness was within the specified range, and dimensional inspection confirmed acceptable distortion. The organization also introduced consumable-storage audits, mandatory preheat records, and a hold point before final inspection.

Learning Points

The case demonstrates that NDT identifies evidence but does not by itself establish root cause. Delayed cracking requires correlation among timing, hydrogen control, microstructure, restraint, hardness, and fracture location. A technically correct repair must be accompanied by process-control improvements, or recurrence remains likely.

9. Professional, Regulatory, and Contemporary Considerations

Inspection personnel must be competent, impartial, and appropriately qualified under the governing certification system. Reports should identify equipment, calibration status, procedure, coverage, indications, acceptance criteria, examiner, date, and disposition. Altering or omitting an unfavorable result is a serious ethical violation because inspection information affects life safety and public confidence.

Codes and standards differ by product and jurisdiction. The responsible engineer must identify the applicable construction code, material specification, NDT standard, acceptance level, and repair requirements rather than applying a familiar criterion automatically. Safety controls are essential for radiation, pressure testing, chemicals, confined spaces, hot work, and electrical equipment.

Digital radiography, phased-array UT, full-matrix capture, automated scanning, and data analytics can improve coverage and recordability. Artificial intelligence may assist indication detection or image triage, but it should remain subject to validated procedures, human review, traceable data, and defined performance limits. A model trained on one material, thickness, or defect population may not generalize to another. Automation supports professional judgment; it does not transfer legal or ethical responsibility away from the employer, examiner, or responsible engineer.

Best-Practice Workflow

1. Establish design function, service loads, materials, and acceptance criteria.
2. Inspect materials and joint preparation before welding.
3. Control procedure variables, consumables, preheat, interpass temperature, and welder qualification.
4. Examine each stage that will become inaccessible.
5. Select complementary NDT methods based on probable defect mechanisms.
6. Calibrate equipment and document coverage and limitations.
7. Investigate indications before repair or acceptance.
8. Preserve evidence when failure is suspected.
9. Confirm repairs using the original and any additional appropriate tests.
10. Close the feedback loop through corrective action, training, and process improvement.

Chapter Summary

Welding quality is determined by more than surface appearance. Discontinuities arise from welding physics, joint design, thermal cycles, contamination, restraint, consumable control, and workmanship. Cracks and lack of fusion are planar, crack-like conditions that generally demand particular attention; porosity and inclusions are volumetric conditions whose significance depends on size, distribution, and service. Visual, penetrant, magnetic-particle, ultrasonic, radiographic, and leak tests provide different and complementary evidence. Destructive testing and metallography reveal properties and mechanisms that NDT alone cannot establish.

Effective failure analysis preserves evidence, correlates inspection results with metallurgy and service history, distinguishes immediate from systemic causes, and evaluates fitness for service rather than relying on appearance alone. Safe practice requires qualified personnel, calibrated equipment, applicable acceptance criteria, accurate records, ethical reporting, and controlled repair procedures. Emerging digital and AI-assisted methods are valuable only when validated and governed by competent human decision-makers.

Key Terms

Acceptance criterion: Specified limit used to decide whether a condition is acceptable.

Discontinuity: Interruption in the expected uniformity of a material.

Defect: A discontinuity that fails an applicable requirement.

Heat-affected zone (HAZ): Base-metal region whose properties or microstructure were altered by welding heat.

Lack of fusion: Failure of weld metal to fuse with base metal or a previous weld pass.

Incomplete penetration: Failure of weld metal to extend through the intended joint root.

Porosity: Gas cavities trapped in solidified weld metal.

Slag inclusion: Non-metallic material trapped in a weld.

Undercut: Groove melted into base metal beside the weld and left unfilled.

Hydrogen-assisted cracking: Cracking promoted by hydrogen, susceptible microstructure, tensile stress, and often delayed after welding.

Preheat: Heating applied to the base material before and during welding.

Visual testing: Direct or aided examination of surfaces and weld geometry.

Liquid penetrant testing: Capillary-based method for surface-breaking discontinuities.

Magnetic-particle testing: Magnetic-flux-leakage method for ferromagnetic materials.

Ultrasonic testing: Examination using reflected high-frequency sound waves.

Radiographic testing: Examination using transmitted X-rays or gamma rays.

Metallography: Preparation and examination of material structure, commonly using etched sections.

Root cause: Fundamental condition whose removal prevents recurrence.

Fitness for service: Engineering determination that equipment can safely perform its intended function.

Residual stress: Stress remaining in a component after external forces are removed.

Review Questions

Short Answer Questions

1. Distinguish a discontinuity from a defect.
2. State three conditions that promote hydrogen-assisted cracking.
3. Why is visual testing performed before, during, and after welding?
4. Identify one major limitation of radiographic testing.
5. What information does metallography provide that visual testing cannot?

Long Answer Questions

1. Explain the formation, significance, and prevention of hot cracking, cold cracking, lack of fusion, and porosity.
2. Compare PT, MT, UT, and RT in terms of physical principle, suitable materials, detectable defects, and limitations.
3. Describe a complete failure-analysis procedure from evidence preservation to corrective action.
4. Explain why acceptance to a fabrication code is not always identical to a fitness-for-service decision.
5. Discuss how welding procedure control, consumable management, and inspection hold points reduce defect recurrence.

Analytical / Application Questions

1. A steel weld passes visual inspection but develops a delayed crack in the HAZ. Propose an investigation plan and identify the evidence needed to test hydrogen-assisted cracking as the root cause.
2. A thick butt weld contains rounded indications on RT but no indication on a surface examination. Which additional tests would you select, and how would material, flaw orientation, and service loading influence the decision?
3. A pressure boundary shows a small surface-breaking crack during service. Develop a decision path covering isolation, characterization, fitness-for-service assessment, repair, reinspection, and documentation.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes a defect from a discontinuity?

A. A defect is always visible at the surface.

B. A defect is a discontinuity that violates an applicable requirement.

C. A discontinuity is always more serious than a defect.

D. A defect occurs only after the component enters service.

Correct Answer: B. A defect is a discontinuity that violates an applicable requirement.

Question 2

Which combination is most associated with hydrogen-assisted cracking?

A. Low restraint, soft microstructure, and high humidity only

B. Hydrogen, susceptible hard microstructure, and tensile restraint

C. High shielding-gas flow and rounded porosity only

D. Low temperature, ductile metal, and no residual stress

Correct Answer: B. Hydrogen, susceptible hard microstructure, and tensile restraint

Question 3

Which method is limited primarily to ferromagnetic materials?

A. Liquid penetrant testing

B. Radiographic testing

C. Magnetic-particle testing

D. Ultrasonic testing

Correct Answer: C. Magnetic-particle testing

Question 4

Liquid penetrant testing is principally used to detect flaws that are:

A. Deep internal and fully enclosed

B. Open to the surface of a non-porous material

C. Present only in non-metallic inclusions

D. Parallel to every inspection surface

Correct Answer: B. Open to the surface of a non-porous material

Question 5

In the relation d = ct/2, why is the factor 2 used?

A. The sound travels to the reflector and returns

B. The material has two HAZs

C. The probe emits two frequencies

D. The reflector doubles the sound velocity

Correct Answer: A. The sound travels to the reflector and returns

Question 6

Which defect is generally considered planar and crack-like?

A. Lack of fusion

B. Spherical porosity

C. Uniform weld reinforcement

D. Surface spatter

Correct Answer: A. Lack of fusion

Question 7

A major limitation of radiographic testing is that it may detect a planar flaw poorly when the flaw is:

A. Perpendicular to the radiation beam

B. Nearly parallel to the radiation beam

C. Located in a thick section

D. Associated with a density difference

Correct Answer: B. Nearly parallel to the radiation beam

Question 8

Which test is most directly useful for revealing weld penetration and bead sequence in a sectioned specimen?

A. Macro-etch metallography

B. Leak testing

C. Magnetic-particle testing only

D. Pressure decay testing

Correct Answer: A. Macro-etch metallography

Question 9

What should normally happen before repairing a suspected crack?

A. Cover it with a cosmetic weld

B. Establish its full extent and remove it under an approved procedure

C. Ignore it if visual width is small

D. Demagnetize the component without further examination

Correct Answer: B. Establish its full extent and remove it under an approved procedure

Question 10

Which is a systemic root cause rather than merely an immediate operator action?

A. A single incorrect electrode angle

B. One visible pore

C. An inadequate welding procedure and missing inspection hold point

D. A final fracture surface

Correct Answer: C. An inadequate welding procedure and missing inspection hold point