Mala soldadura frente a buena soldadura: Cómo identificar la calidad y evitar los defectos

A structural fabrication shop delivers a batch of 40 equipment frames on schedule and within budget. Visual inspection on the shop floor passes every unit. Three months into service, two frames develop cracks at the same location — the weld toe on a primary load-bearing connection. Closer examination of the remaining frames finds undercut at that joint averaging 0.4 mm depth across most of the batch. The undercut was present all along, visible to an inspector who knew what to look for, but missed by incoming inspection because nobody measured it systematically. The undercut reduced the section at the stress concentration point — exactly where fatigue cracks initiate — by just enough to put the joint below its fatigue life target at operating loads. Rework of the remaining 38 frames costs more than the original fabrication contract. In our shop floor experience, the gap between good welding and bad welding is often narrower and more systematic than it appears. It’s not random welder error — it’s process drift that produces the same subtle defect across an entire batch.

Distinguishing good welding from bad welding is an engineering discipline, not a subjective judgment. Good welds have measurable characteristics — bead geometry, surface condition, dimensional conformance, internal integrity — that can be evaluated against defined acceptance criteria. Bad welds have identifiable causes and predictable failure modes. This guide covers what good welds look like and why each characteristic matters structurally, the common defects that indicate bad welding along with their causes and risk levels, how parameters drive defect formation, which inspection methods detect which defects, applicable acceptance standards, and practical prevention strategies for production environments.

What Does a Good Weld Look Like? The Engineering Perspective

A good weld is not simply a visually clean weld — it’s a weld whose geometry, surface condition, and internal integrity collectively indicate that the process was stable, the parameters were correct, and the joint will perform as designed. Visual characteristics are the observable evidence of process control.

Characteristics of a Quality Weld

Uniform bead profile: A consistent ripple pattern along the full weld length indicates stable heat input and consistent travel speed. Width variation exceeding approximately 15–20% over the weld length suggests parameter instability — fluctuating current, inconsistent torch speed, or arc length variation. These fluctuations also vary penetration, which means the mechanical performance of the weld varies along its length.

Smooth transition at the weld toe: The junction between the weld face and the base metal (the weld toe) is the most fatigue-critical location in most welded joints. Stress concentrations at notches and sharp transitions are where fatigue cracks initiate under cyclic loading. A good weld blends gradually into the base metal. A sharp geometric notch at the toe — even without visible undercut — increases the stress concentration factor and reduces fatigue life. AWS D1.1 and similar standards specify maximum weld profile angles for this reason.

Appropriate reinforcement height: Excessive weld reinforcement (overfill above the base metal surface) is not simply cosmetic. A high, peaked weld bead creates a stress concentration at the toe where the steep weld flank meets the plate — functionally similar to undercut in its effect on fatigue performance. Too little reinforcement (underfill) reduces the effective throat thickness. Both deviations indicate heat input or deposition rate problems.

Absence of visible defects: Surface porosity, cracks, excessive spatter, incomplete fill, and visible slag all indicate process problems. Many surface defects are accompanied by internal defects — porosity on the surface is often accompanied by internal porosity clusters; surface cracks may be the visible end of a crack that extends into the weld.

Continuity: The weld should be continuous along its intended length without gaps, arc strikes outside the joint, or visible restart problems. Arc restart locations — where a welder stopped and restarted — are frequent defect locations because the cold start condition often produces lack-of-fusion if not managed deliberately.

Visual Quality Assessment Sequence

When evaluating a weld visually, a systematic sequence prevents missed defects:

1. Overall bead uniformity: Step back and view the full weld length — width, height, and ripple consistency should be uniform
2. Weld toe condition: Close examination of both toes for undercut, sharp transitions, and surface discontinuities
3. Surface integrity: Systematic scan for porosity, cracks, and spatter distribution
4. Geometry compliance: Width and height measurements against drawing callout or applicable standard
5. Coverage: Confirm weld extends to full called-out length with no missed sections

Good vs Bad Weld: Feature Comparison

Welding Defects: Causes, Detection, and Risk Level

Porosity

What it is: Gas voids — spherical or irregular — trapped in the solidified weld metal. Surface porosity appears as small holes; internal porosity is only detectable by RT or UT.

Root causes: Surface contamination (oil, moisture, rust, mill scale) introducing hydrogen and hydrocarbon gases into the weld pool; insufficient shielding gas coverage allowing atmospheric nitrogen and oxygen ingress; excessive travel speed preventing gases from escaping before solidification; moisture in electrode coatings or flux.

Structural consequence: Porosity reduces the effective cross-sectional area of the weld. Distributed fine porosity within ISO 5817 Level C limits has modest structural impact. Clustered porosity or large voids create stress concentration sites that reduce fatigue strength, particularly in dynamically loaded structures.

Risk level: Medium, conditional on size and distribution. ISO 5817 Level B permits individual pore diameter up to 0.5 mm with maximum 2% of weld area; Level D permits up to 2.4 mm diameter with 4% area coverage. Cracks always require repair; porosity within limits is acceptable.

Cracks

What they are: Fractures in the weld metal, heat-affected zone (HAZ), or base metal. Classified by location (weld metal, HAZ, base metal) and timing (hot cracks during solidification; cold cracks forming hours or days after welding).

Root causes for hot cracks: High sulfur or phosphorus content in base metal or filler; solidification shrinkage stress in highly restrained joints; filler metal incompatibility.

Root causes for cold cracks (hydrogen-induced): Hydrogen absorbed from moisture, organic contamination, or humid electrodes enters the HAZ; residual tensile stress in the hardened HAZ; susceptible martensitic microstructure in higher-carbon or alloy steels. Carbon equivalent above approximately 0.40–0.45 significantly increases cold cracking risk, and preheat becomes necessary.

Structural consequence: Any crack is potentially catastrophic. Cracks provide initiation sites for fatigue crack propagation under cyclic loading and can propagate to fracture under static overload. No crack is within acceptance criteria under AWS D1.1 for structural welds — any crack requires repair without exception.

Risk level: Critical. Automatic rejection. AWS D1.1 categorically prohibits any crack in structural welds regardless of size or location.

Undercut

What it is: A groove melted into the base metal at the weld toe that is not filled with weld metal. Visible as a longitudinal depression along one or both weld edges.

Root causes: Excessive welding current melts more base metal than the weld pool can fill; high travel speed prevents pool from flowing back to fill the groove; incorrect torch angle directing the arc away from the weld center toward the base metal at the toe.

Structural consequence: Undercut creates a geometric notch at the weld toe — exactly the location where fatigue cracks initiate. Even shallow undercut (0.2–0.3 mm) measurably reduces fatigue life in dynamically loaded structures by increasing the local stress concentration factor.

Risk level: Medium to high depending on depth. ISO 5817 Level B limits undercut to 0.05 mm depth (essentially none visible); Level C allows up to 0.1 mm; Level D up to 0.2 mm. The weld described in the introduction — 0.4 mm average undercut — would fail all three levels.

Lack of Fusion

What it is: Absence of complete metallurgical bonding at the interface between weld metal and base metal, between weld metal and the joint sidewall, or between weld passes. The weld metal is present but is not bonded to an adjacent surface.

Root causes: Insufficient heat input — current too low for the joint configuration and material thickness; incorrect torch angle directing heat into previously deposited weld rather than unwelded surfaces; inadequate joint preparation with contaminated or oxidized fusion surfaces; rapid travel speed not allowing temperature at the fusion interface to reach the melting point.

Structural consequence: Lack of fusion is a planar defect that effectively reduces the load-carrying cross-section of the weld to only the bonded area. Under tensile or bending load, the unbonded region acts as a pre-existing crack. Under cyclic loading, it propagates to fatigue failure. This is considered the most structurally dangerous common weld defect because it’s internal, planar, and reduces load capacity directly.

Risk level: Critical. AWS D1.1 prohibits any lack of fusion in structural welds. It is not visible in visual inspection and requires UT for reliable detection — which is why it appears in service failures that passed visual inspection.

The important subtlety: Lack of fusion is frequently undertreated by operators who reduce current to avoid burn-through. The correct response to burn-through is faster travel speed, not lower current. Reducing current below the minimum for the joint configuration is the fastest way to produce lack of fusion.

Slag Inclusion

What it is: Non-metallic material — flux residue or oxide inclusions — trapped within the weld metal. Appears as irregular dark areas on radiographic films.

Root causes: Incomplete removal of slag between passes in multi-pass welds (the most common cause); incorrect electrode angle allowing slag to run ahead of the arc; undercut or surface irregularities on previous passes that trap slag during the next pass; too-rapid solidification that traps slag before it floats to the surface.

Risk level: Medium to high. Slag inclusions reduce effective weld cross-section and can serve as crack initiation sites. Elongated inclusions aligned with the stress direction are particularly detrimental. ISO 5817 specifies maximum permissible dimensions; AWS D1.1 limits apply based on individual inclusion size and aggregate length.

Burn-Through

What it is: The weld pool penetrates completely through the base material, leaving a hole or partial hole.

Root causes: Excessive heat input for the material thickness — current too high, travel speed too slow; root opening too large without backing support; thin material without adequate support.

Risk level: High. Burn-through is a physical void in the joint — automatic rejection. Repair requires grinding out the affected area and re-welding, often with backing.

Distortion

What it is: Geometric deformation of the welded assembly — bowing, warping, angular change — from non-uniform thermal expansion and contraction.

Root causes: Concentrated or unbalanced heat input; unsequenced welding that deposits all passes on one side of a symmetric structure; insufficient fixturing; thin section material with high sensitivity to thermal effects.

Risk level: Medium for structure; high for dimensional conformance. Distortion doesn’t reduce weld strength directly, but it causes assembly misalignment, dimensional non-conformance that affects fit-up of downstream components, and may introduce residual stresses that reduce fatigue life.

Defect Severity Reference

How Welding Parameters Drive Defect Formation

Defects are almost never random — they’re the predictable output of specific parameter conditions. Understanding the parameter-to-defect mapping allows operators to adjust correctly rather than experimenting.

The critical interaction: Parameters work through heat input. Welding current and voltage increase heat input; travel speed reduces it. The same defect can result from too-high current at correct speed or from too-slow speed at correct current. Diagnosing correctly requires measuring the actual parameter combination, not just one variable.

The operator trap: The most common incorrect operator response to burn-through is to reduce current. This reduces heat input but also reduces the energy available for fusion. If travel speed was the actual problem (too slow), reducing current produces lack of fusion instead of burn-through. The WPS should specify parameter ranges — operators adjusting outside those ranges introduce new defects while eliminating the original one.

Inspection Methods and When to Use Each

The visual inspection limitation: Visual inspection is fast and cost-effective, but it cannot detect internal defects. Lack of fusion — the most structurally dangerous common defect — produces no visible surface indication. A weld that passes 100% visual inspection can contain significant lack of fusion. For structural or safety-critical applications, supplementary UT or RT is essential, not optional.

The inspection sequence: VT first on all welds, followed by appropriate NDT based on application criticality and standard requirements. VT that finds surface cracks or significant undercut should trigger more extensive inspection of the surrounding area.

Applicable Standards: What Acceptance Means in Practice

The Core Principle

A weld is acceptable if its imperfections fall within the defined limits for the applicable standard and quality level. A weld does not need to be defect-free — it needs to be within limits. This distinction matters because over-inspection and unnecessary rejection is as costly to production as missing actual defects.

ISO 5817 Quality Levels

ISO 5817 defines three quality levels with specific limits for each imperfection type:

Level B (Stringent): The highest quality level. Extremely tight limits — undercut maximum 0.05 mm, porosity maximum 1 mm diameter. Used for pressure vessels, aerospace structures, and safety-critical primary structures where weld quality is directly linked to system safety.

Level C (Intermediate): Standard industrial quality. Moderate limits — undercut up to 0.1 mm, porosity up to 1.5 mm diameter. Used for general engineering structures, mechanical equipment, and industrial fabrication where standard structural performance is required.

Level D (Moderate): The least stringent level. Used for non-critical applications where appearance and gross structural adequacy are the requirements and fatigue life is not a design driver.

AWS D1.1 in Practice

AWS D1.1 provides the structural welding framework most widely used in North America for steel structures. It categorizes welds by loading type (statically loaded versus cyclically/fatigue loaded) with more stringent requirements for cyclic loading. In cyclically loaded structures, undercut is limited to 0.25 mm maximum in tension zones — recognizing that undercut at tension-side weld toes is where fatigue cracks initiate.

For procurement teams: Specifying “ISO 5817 Level C” or “per AWS D1.1” in purchase orders provides a defined quality baseline that suppliers must document through WPS records, qualified welders, and inspection reports. Without a standard reference, “acceptable quality” is undefined and unenforceable.

Preventing Bad Welds: The Process Control Approach

Bad welds are almost always the result of system failures, not individual mistakes. A single welder making a random error produces one bad weld. A process control failure produces the same defect across an entire batch — as in the introduction case where undercut appeared consistently across 40 frames.

Process Parameter Stability

The foundation of consistent weld quality is operating within a validated parameter window. The Welding Procedure Specification (WPS) defines this window: minimum and maximum values for current, voltage, travel speed, heat input, interpass temperature, preheat, and shielding gas flow. Welders operating within the WPS produce consistent results. Operators adjusting outside the WPS introduce uncontrolled variation.

Production practice: Monitor and record actual parameters for each welding operation, not just the setup values. Current measured at the power source may differ from current at the arc due to cable resistance. In-process monitoring that records actual arc parameters catches parameter drift before it produces a full batch of defective welds.

Material Preparation

A significant fraction of porosity defects trace back to material condition before welding. The prevention protocol is straightforward: clean all joint surfaces to bright metal within the weld zone, verify no oil, grease, moisture, rust, mill scale, or coating is present in the weld area, and store consumables (electrodes, wire, flux) per manufacturer requirements.

The welding sequence matters: joint preparation on Monday should not be followed by welding on Thursday if the joint has been exposed to humid conditions in the interim. Mill scale that wasn’t an issue at ambient humidity becomes a porosity source after condensation. Surface preparation immediately before welding, or protection of prepared surfaces from re-contamination, is part of the quality system.

Operator Qualification and Standardization

Welder certification (AWS, EN 9606, ISO 9606) establishes that the welder can produce acceptable welds in specific positions, processes, and materials. Certification is necessary but not sufficient — it demonstrates capability under test conditions, not consistent production performance.

Standardization — defined torch angles, travel speed references, restart protocols, and interpass cleaning procedures — is what converts individual capability into consistent production output. Experienced welders who develop personal technique that deviates from the WPS are a common source of batch variation, even when their personal welds look good.

Equipment and Environmental Factors

Power supply stability, cable condition, contact tip wear, nozzle fouling, and shielding gas delivery all affect arc stability and therefore weld quality. Equipment maintenance schedules that keep these variables within acceptable ranges prevent the slow drift that produces gradual quality deterioration.

Environmental factors — ambient wind above approximately 3 km/h disrupts GMAW and GTAW shielding; ambient humidity above 80% increases hydrogen risk for hydrogen-sensitive materials; temperature below 5°C requires preheat for many steels — need to be managed as process conditions, not treated as uncontrollable variables.

Evaluating Welding Supplier Quality

For procurement teams sourcing welded fabrications, supplier qualification should assess process capability, not just sample appearance. A supplier who produces attractive samples on demand but lacks systematic process control will produce inconsistent quality in production.

Supplier Qualification Framework

Process documentation: Does the supplier have WPS documents for the processes and materials in the scope of work? Are welders qualified to those WPS documents? Can the supplier produce PQR (Procedure Qualification Record) evidence that the WPS parameters were validated through mechanical testing?

Weld consistency across production: Request samples from multiple production batches, not just freshly made demonstration welds. Measure bead width, height, and toe condition across the sample set. Consistent parameters produce consistent measurements; inconsistent results indicate process instability.

Inspection capability: Does the supplier perform VT on 100% of welds? What NDT capability is in-house versus outsourced? Can they provide inspection reports with traceability to individual weld joints?

Quality management: ISO 9001 certification establishes that a quality management system exists and is followed. It doesn’t guarantee weld quality directly — a poorly managed welding operation can be ISO 9001 certified — but it does establish the documentation and corrective action infrastructure that enables systematic quality improvement.

Red flags in supplier evaluation:

• Attractive samples but no WPS documentation
• Welders who cannot name the applicable procedure for their current work
• No in-house NDT capability for applications requiring internal defect detection
• Unable to provide inspection reports from current production
• No systematic tool or equipment maintenance records

Cost of Poor Welding Quality

The cost of bad welds compounds across the production and service lifecycle. Direct rework costs are visible; indirect costs are often larger.

Direct rework: Grinding out a defective weld zone, re-welding, and re-inspecting typically adds 25–60% to the original weld labor cost for the affected joint. For a large structural weld that originally took 4 hours to produce, repair takes 5–8 hours including grinding, re-welding, and re-inspection.

Scrap: When defects make repair impractical — typically when cracks or lack of fusion have propagated through a critical section, or when distortion has taken the assembly outside correctable limits — the part scraps. High-value fabricated assemblies that scrap represent total loss of all labor and material invested to that point.

Schedule impact: Defect discovery late in the production sequence causes cascade delays. A weld defect found at final inspection delays delivery; the same defect found at intermediate inspection delays all downstream operations while repair is completed. Early detection through in-process VT is always lower cost than late detection.

Service failures: The highest-cost scenario — field failure of a welded joint. The direct costs (replacement, repair) are typically 10–50× the cost of the original weld. Indirect costs (downtime, liability, investigation) can be larger. The preventability of most weld failures makes this cost category particularly frustrating — the frame failures in the introduction were entirely preventable through routine weld inspection with a fillet weld gauge.

Conclusión

Good welding and bad welding are not defined by appearance alone — they’re defined by process stability, measurable geometric conformance, and compliance with established acceptance criteria. A good weld is evidence of a controlled process: consistent heat input, adequate fusion, proper shielding, and appropriate parameters for the material and joint configuration. A bad weld is evidence of process failure — one or more variables outside the range required to produce acceptable fusion, geometry, and surface condition.

For engineers specifying weld quality, the practical framework is selecting the appropriate standard and quality level (ISO 5817 Level B or C for most structural applications, AWS D1.1 for structural steel), specifying the inspection method required to verify internal as well as surface quality, and recognizing that visual inspection alone cannot verify structural welds in applications where lack of fusion or internal cracking would be safety-critical. For production teams, the prevention approach is process-first: stable parameters within a validated WPS, clean material preparation, qualified welders, and systematic in-process inspection that catches defects as they occur rather than at final inspection. For procurement teams, supplier evaluation based on process documentation, production sample consistency, and inspection capability identifies partners who produce consistent quality rather than impressive demonstrations. When all three — engineering specification, production process control, and supplier qualification — are aligned, welding quality stops being a reactive problem and becomes a predictable output.

Preguntas más frecuentes

What does a good weld look like?

A good weld shows consistent bead width and ripple pattern along its full length, smooth gradual transitions at both weld toes without notches or undercut, appropriate reinforcement height within specified limits, and a clean surface free of visible cracks, porosity, and excessive spatter. Most importantly, it meets the measurable acceptance criteria of the applicable standard — ISO 5817 or AWS D1.1 — not just visual expectations. Consistent geometry across the full weld length is the most reliable indicator of a controlled, repeatable process.

What are the signs of bad welding?

The most significant bad weld indicators are cracks (fractures in the weld or HAZ — always unacceptable), undercut (groove at the weld toe reducing section area), lack of fusion (incomplete bonding — internal, requires NDT to detect), and porosity (gas voids affecting cross-section and fatigue performance). Lesser indicators include uneven bead profile, excessive spatter, and incomplete fill. Cracks and lack of fusion are structural defects requiring repair; minor porosity and slight surface irregularity may be within acceptable limits depending on the applicable standard and quality level.

How is weld quality inspected?

Visual testing (VT) is the first inspection method applied to all welds — it detects surface defects including cracks, undercut, bead irregularity, and burn-through. Ultrasonic testing (UT) detects internal defects, particularly lack of fusion and internal cracks, and is required for structural and safety-critical applications. Radiographic testing (RT) provides imaging of internal porosity, slag, and inclusions with a permanent record. Dye penetrant (PT) and magnetic particle (MT) testing detect fine surface-breaking cracks on non-ferromagnetic and ferromagnetic materials respectively. No single method detects all defect types — the inspection plan must match methods to the defects of concern.

What welding defects are automatically unacceptable?

Under AWS D1.1 for structural steel, any crack is automatically rejected regardless of size or location. Any lack of fusion is also prohibited. Burn-through (hole in the base material) is automatically rejected. Other defects — undercut, porosity, slag inclusion — have defined size limits above which they require repair. The principle: defects that provide crack initiation sites under cyclic loading, or that reduce structural cross-section below design requirements, are categorically unacceptable.

How can bad welds be prevented in production?

Systematic prevention requires four integrated elements: parameter control within a validated WPS (defining current, voltage, travel speed, heat input, and shielding gas ranges that produce acceptable welds), clean material preparation (removing all contamination from joint surfaces before welding), operator qualification and standardization (welders certified to applicable procedures, following defined technique), and in-process inspection (VT during and after welding to catch defects at source before they compound into larger quality problems). The goal is process stability across an entire production batch — not perfection on individual welds followed by rework on others.