Visual inspection report
A VT sheet to EN ISO 17637 with the acceptance table already worked out: give the quality level, weld type and thickness, and the generator fills in the ISO 5817 limit for every imperfection and leaves columns for the measured value and the assessment. It downloads as a table.
Short answer
Every weld goes through visual testing — the other methods are applied on a sample basis. ISO 17637 sets the conditions: at least 350 lx at the surface, the eye no further than 600 mm, the angle no less than 30°. The limits come from ISO 5817 at the level the documentation calls for. The generator below builds the report with the table of permitted values for your thickness.
What visual testing sees, and what it does not
The eye finds only what reaches the surface: undercut, excess weld metal, root concavity, linear misalignment, pores, cracks in the cap and in the root. Lack of fusion and lack of penetration inside the weld will not be seen — those need radiography, ultrasonics or another volumetric method. They still appear in the report table, and that is not a mistake: the table lists the acceptance criteria to ISO 5817, not the list of what the eye can make out. On a single-sided weld with access to the root, lack of penetration is in fact visible.
The quality level comes from the design documentation, not from the examiner’s judgement. Under EN 1090-2 it follows from the execution class: EXC1 — D, EXC2 — C, EXC3 — B. Where the documents name no class, EXC2 is what load-bearing steelwork usually implies, but confirming that is the job of the person responsible for welding, not of the person holding the magnifier.
What to record so the report is accepted
The empty rows in the sheet are not a formality. The weld number and the drawing reference tie the report to the item, the WPS reference ties it to the procedure, and the welder’s stamp ties it to a person. Without that chain an entry reading “acceptable” refers to nothing in particular and settles nothing if questions are asked later.
The test conditions are recorded too: illuminance, distance and angle are requirements of the standard, and if they were not met the result cannot be claimed as obtained to ISO 17637.
How to carry out visual testing, step by step
- Establish the level and the documents. The quality level (B, C or D) comes from the drawing or specification, the weld number from the weld schedule, the procedure from the WPS. Without those three the report has nothing to refer to.
- Prepare the surface. The weld must be free of slag and spatter. Under slag you see neither undercut nor surface pores.
- Check the conditions. Lux meter: at least 350 lx. Eye no further than 600 mm, viewing angle no less than 30°. Where you cannot get close — a mirror or an endoscope.
- Walk the weld and measure. First the whole length by eye, then the suspect spots with the gauge: the deepest undercut, the highest reinforcement, the largest misalignment. The report takes the worst value, not the average.
- Compare with the table and judge. Reading ≤ permitted value — accepted. Above it, or an imperfection marked “not permitted” — rejected: record it and pass it to the person responsible for welding.
Worked example: 10 mm butt weld, b = 14 mm, three levels side by side
Data as in the generator defaults: butt weld, thickness t = 10 mm, weld width b = 14 mm, root width 6 mm. The table shows what the generator puts in the “permitted” column at each level (values from the same rules as the ISO 5817 calculator):
| Imperfection (ISO 6520-1) | Formula for C | B | C | D |
|---|---|---|---|---|
| 5011 undercut | 0.1·t, max 0.5 | 0.50 mm | 0.50 mm | 1.00 mm |
| 502 excess weld metal (butt) | 1 + 0.15·b, max 7 | 2.40 mm | 3.10 mm | 4.50 mm |
| 504 excess penetration | 1 + 0.6·b, max 4 | 2.20 mm | 4.00 mm | 5.00 mm |
| 511 incompletely filled groove | 0.1·t, max 1 | not permitted | 1.00 mm | 2.00 mm |
| 515 root concavity | 0.1·t, max 1 | 0.50 mm | 1.00 mm | 2.00 mm |
| 507 linear misalignment | 0.15·t, max 4 | 1.00 mm | 1.50 mm | 2.50 mm |
| 2017 surface pore | 0.2·t, max 2 | not permitted | 2.00 mm | 3.00 mm |
| 402 lack of penetration | — | not permitted | not permitted | 2.00 mm |
| 401 lack of fusion, 100 crack | — | not permitted at any level | ||
What this means in practice. A 0.7 mm undercut on this plate passes at D but not at C — and EXC2 is exactly C. Reinforcement of 3.5 mm is too high for C (3.10 mm) and acceptable for D. Misalignment of 1.2 mm: accepted at C, rejected at B. That is why the level goes into the report before measuring, not chosen afterwards.
How to read the acceptance table in the report
Formula and cap. Each value is a formula of thickness t or width b, cut off by a fixed maximum. On thin plate the formula decides, on thick plate the cap: undercut for C is 0.1·t but no more than 0.5 mm, so from t = 5 mm up the value no longer grows.
Which width. For butt reinforcement b is the weld width. For excess penetration the standard works from the width of the root on the back, and the generator has a separate field for it. In the example the root is 6 mm: at level B the limit is 1 + 0.2·6 = 2.2 mm, whereas the 14 mm face width would give 3 mm — looser than the standard. Leave the field empty and the generator writes the formula into the report; the figure goes in once the root has been measured.
“Not applicable” and “not permitted”. The first means the line does not apply to this weld type (fillet convexity on a butt weld). The second means the imperfection is prohibited and any trace of it gives a reject. Causes and prevention for each imperfection are in the weld defects catalogue, and the rules of the standard itself on the ISO 5817 page.
Common mistakes in a VT report
- “OK” instead of a number. Without a measured value nobody can check that the verdict was right.
- Inspecting before cleaning. Slag hides undercut and pores; “accepted” on an uncleaned weld means nothing.
- No inspection conditions. If the report has no lighting and distance, nobody knows whether ISO 17637 was met.
- A level picked to suit the result. The level comes from the documents, not from what the measurement showed.
- A repair without a record. A repaired weld is inspected again and recorded separately — same number, marked as repaired.
Frequently asked questions
Which quality level should I pick?
None — the level is not picked. It stands in the design documentation, and under EN 1090-2 it follows from the execution class: EXC1 gives D, EXC2 gives C, EXC3 gives B. The examiner applies the level rather than setting it.
Why are lack of fusion and lack of penetration in the table if VT cannot see them?
Because it is a list of acceptance criteria to ISO 5817, not a list of what the eye can see. Some of them genuinely need volumetric testing; some are visible — on a single-sided weld with access to the root, lack of penetration shows up in the root itself.
Does the examiner need a certificate?
ISO 17637 requires trained personnel, and a particular level to EN ISO 9712 is required by the client or by the standard being applied. For structures to EN 1090 the personnel requirements are set by that standard together with the order specification.
The measured value came out above the limit — what now?
Record the non-conformity and stop. The decision to repair or reject belongs to the person responsible for welding, under an approved repair procedure. The examiner measures and records but does not decide what happens to the weld.
How many lux do you need for visual inspection?
ISO 17637 requires at least 350 lx at the surface examined and recommends 500 lx. Ordinary shop lighting often gives less, especially inside a structure or under a beam flange — so measure with a lux meter at the point of inspection and add a torch. The reading is worth recording in the report.
What do you measure undercut, excess weld metal and misalignment with?
A weld gauge: a multi-purpose one has scales for undercut depth, reinforcement height, fillet leg and misalignment. Add a magnifier, a torch and a mirror for places you cannot reach. The reading goes into the “measured” column next to the permitted value, not just “OK”.
Is visual testing done only after welding?
No. ISO 17637 covers inspection before welding (edge preparation, gap, cleanliness), during (root, interpass cleaning) and after. The generator report is for the final stage, but a preparation error found before welding costs minutes to fix, and after welding costs a repair.
Updated: