Visual examination finds more rejectable conditions than every other method combined, and it does so with equipment that costs less than a set of tyres. It is also the method most often treated as a formality - a walk past the joint on the way to arranging the real testing. That reversal is expensive, because almost everything the volumetric methods later find could have been prevented by someone looking properly before the arc was struck.
This guide covers what visual inspection covers at each stage of the work, the physical conditions it needs to be valid, the conditions an inspector is scanning for, how to sequence a joint, and where the method genuinely runs out. It describes general practice. The criteria that decide whether anything you see is acceptable come from the code or contract documents governing the work.
Visual inspection is three jobs, not one
The method is usually taught as looking at finished welds, which is the least valuable third of it. Before welding, visual inspection covers material identification, joint preparation, cleanliness, fit-up, alignment, backing, tacks, and the condition of consumables. During welding it covers technique, sequence, interpass cleaning, bead placement, and the conditions the procedure requires. After welding it covers the finished surface, the profile, and the base metal around it.
The order of value runs roughly backwards from the order of attention. A joint with a root opening outside tolerance will produce a root condition no amount of skilled welding fixes. Contaminated preparation will produce porosity. Poor fit-up produces misalignment that is still there after every pass has been laid. Catching these costs minutes; catching them afterwards costs a repair. Visual inspection is best understood as a preventive activity that happens to end with an examination.
- Before welding: materials, preparation, cleanliness, fit-up, alignment, tacks, consumable condition
- During welding: technique, sequence, interpass cleaning and temperature, bead placement
- After welding: surface condition, profile, weld size, base metal around the joint
- Also continuous: equipment condition, the welder's qualification status, and procedure compliance
The conditions the method needs to be valid
Visual examination has physical prerequisites, and an examination performed without them is not really an examination. You need adequate light on the surface, you need to get your eye close enough to resolve what you are looking for, and you need an angle that is not so oblique that the surface becomes a smear.
Lighting is where most of the loss happens. A shop that feels bright overhead can leave a joint in shadow, and a torch aimed straight down a weld flattens everything it illuminates. Raking light - a beam angled low across the surface - is what makes shallow conditions visible, because it converts a depth difference into a shadow the eye can actually see.
Access is the other constraint, and it is a legitimate reason to record a limitation. If a joint cannot be reached, cannot be seen from a workable angle, or cannot be lit, that is a finding in itself. Codes generally require the examination surface to be accessible and adequately illuminated; writing down that it was not is more useful than quietly examining it badly.
What you are actually scanning for
The list of surface conditions worth knowing cold is short. Cracks anywhere. Undercut along the toes. Overlap where the weld metal has rolled onto unfused base metal. Underfill and excessive reinforcement on groove welds. Incomplete penetration or an unfilled root where the root is visible. Porosity breaking the surface. Slag left on or between passes. Arc strikes on base metal. Spatter. Misalignment across the joint. Weld size and length against the call-out.
Two of those deserve specific mention because they look similar and mean opposite things. Undercut is a groove melted into the base metal at the toe and left unfilled - metal removed. Overlap is weld metal that has spilled over the toe without fusing to the base metal beneath - metal added, and unbonded. Both live at the toe, both are surface conditions, and confusing them in a report suggests the observer did not understand what they saw.
Arc strikes are the condition most often shrugged off and most often worth catching. A momentary arc on base metal outside the joint produces a small, rapidly quenched, hardened spot with no filler and no controlled cooling, which is a plausible crack initiation site in exactly the place nobody will inspect later.
A sequence that stops you missing things
Working the same order every time is what separates repeatable inspection from wandering around a joint. The sequence below works on almost any weld and takes very little longer than an unstructured look, because the time saved by not backtracking pays for the discipline. It is also what makes two inspections of the same joint, carried out by two different people, arrive at the same list of conditions.
The important habit inside it is to scan the whole joint before measuring anything. Candidates and new inspectors alike tend to find the first condition, measure it, write it up, and then discover a second one, which means going back over ground already covered. One scanning pass followed by one measuring pass is meaningfully faster.
The second habit is to look at the base metal, not just the weld. Toe cracks, arc strikes, gouges from grinding, lamellar tearing showing at an edge, and heat damage all sit outside the weld metal and are all findings. A visual examination that stops at the weld toes has examined about half the area that matters.
- Confirm you have light on the surface and an angle you can work from
- Scan the full length of the joint before touching a gauge
- Scan the base metal either side of the weld for strikes, gouges and cracking
- Note every candidate condition and its rough location on a sketch
- Return with gauges and measure the governing value for each condition
- Compare each measurement against the applicable acceptance provision
- Record condition, location, measurement, provision and disposition together
The tools that do the work
The kit is small and stays small. A fillet weld gauge set covers leg size and profile. A bridge cam type gauge covers reinforcement height, undercut depth, preparation angle and misalignment depending on how it is set. A hi-lo gauge covers internal alignment on pipe. A steel rule, a low-power magnifier, a mirror for the far side of a joint, and a light source you can angle yourself complete it.
Magnification is worth thinking about carefully. Low power helps resolve a fine surface condition; high power narrows the field so much that scanning becomes impractical and depth of field disappears. A modest magnifier used to confirm something the naked eye already flagged is the productive pattern.
Whatever the tool, seating is what determines whether the reading is real. A gauge resting on spatter, on scale, or on the crown of an adjacent bead reports a dimension the joint does not have. Two seconds spent confirming metal-to-metal contact prevents the far more expensive cycle of taking a reading, doubting it, and taking it twice more.
- Fillet weld gauges for leg size and profile, in both convex and concave styles
- Bridge cam type gauge for reinforcement, undercut, preparation angle and misalignment
- Hi-lo gauge for internal alignment across a pipe joint
- Rule, mirror, low-power magnifier, and a light you can angle across the surface
Where visual inspection genuinely stops
The method sees surfaces. It cannot see a lack of fusion between passes, a slag inclusion buried under a cap, a subsurface crack, or an incomplete root on a joint welded from one side with no access to the far face. Those are the reasons volumetric and surface-enhanced methods exist, and knowing the boundary is part of using the method honestly.
It is also limited by what the surface has been allowed to keep. Grinding a weld flush removes profile information. Painting or coating removes almost everything. Blasting can smear and close fine surface conditions rather than reveal them. Sequencing visual examination before any of that happens is a scheduling matter an inspector should be arguing for, because a condition found visually is found earlier, more cheaply, and with fewer arguments about interpretation than the same condition found later.
Recording it so somebody else can use it
The report is the deliverable, not the walk-around. A record that names the condition, fixes its location against something the next person can find, gives the governing measurement, and identifies the provision applied is usable by a fabricator, a re-inspector and an auditor without any of them speaking to you.
Location is the field most often written uselessly. A depth recorded against a two-metre weld with no reference point cannot be found again, which means the repair is guesswork and the re-inspection is not verifying the same thing. Give a datum and a distance, or mark the steel. A weld map with stationing on it costs an hour once and pays for itself on every finding after that.
Finally, record what you could not examine. Areas with no access, joints already coated, surfaces you could not light - those limitations belong in the report exactly as findings do. An examination record that quietly implies full coverage it did not have is the one document in the file that will not survive scrutiny.