Magnetic particle testing on a structural fabrication job is not certified by the tester itself; it is certified by the chain of documents that proves the method, the acceptance level, the NDT percentage, and the inspector's qualification all match the execution class on the drawing [S1][S3].
EN 1090-2 places MT in a specific slot: 100% visual per EN ISO 17637 on every weld, plus 10% MT or PT on critical fillet welds as a sample check under Table 24, with acceptance to EN ISO 5817 Quality Level C by default [S3]. Getting the percentage wrong, or pairing MT with a non-ferromagnetic weld, is the fastest way for a notified-body audit to fail the file.
MT physics and material limits that drive the spec
MT works by inducing a magnetic flux in the part, then watching ferromagnetic particles migrate to the leakage field above a surface or near-surface crack that breaks the flux path [S1]. Reliable detection depth is limited to about 3 mm below the surface, so MT is a surface/NDT method, not a volumetric one, and it must never be specified to "verify" internal soundness that only PAUT or radiography can see [S1].
The material constraint is hard: MT applies only to ferromagnetic materials (carbon steel, low-alloy steel, martensitic and ferritic stainless), and is physically impossible on austenitic stainless, aluminium, titanium, and nickel alloys [S1]. A fabricator that runs MT on a 304/316 stainless weld either has a methodology error or a documentation error; either way the report is invalid for EN 1090, AWS D1.1, ASME VIII, API 650, API 1104, and EN 15085 audits [S1].
Directional sensitivity is the second physics constraint: a crack parallel to the flux lines produces almost no leakage field and will be missed, so the induced field must be perpendicular to the expected crack plane, with at least two orthogonal shots per examination zone, and that requirement belongs in the WPS, not in the inspector's discretion [S1].
EN ISO 17638 method requirements you must see on the procedure
EN ISO 17638:2016 sets the procedure rules that every MT report on a structural job must reference: a minimum tangential field strength of 2 kA/m at the examination surface, declared particle type (dry powder, wet colour contrast, or wet fluorescent), surface preparation with a documented maximum roughness, and the lighting/irradiance numbers for the visual read [S1].
For colour contrast, the standard asks for at least 500 lux of white light at the surface; for fluorescent MT, it asks for UV-A irradiance of at least 10 W/m² measured at 300 to 400 mm working distance [S1]. If the inspector cannot show a calibrated lux meter or UV-A radiometer reading on the report, the procedure is not auditable, regardless of how clean the indication photos look.
Magnetisation technique must be named in the procedure: AC or DC electromagnetic yoke for portable field work on butt and fillet welds, prod contacts for large plate areas and localised HAZ checks, or coil/through-current for fixed-shop inspection of smaller components [S1]. The choice is driven by part geometry and accessibility, and changing technique mid-job without revising the WPS is a common audit finding [S1].
EN ISO 23278 acceptance levels: pick one in the ITP, not on site
EN ISO 23278:2015 defines three graded acceptance levels for MT indications, and the level has to be fixed in the contract, the Inspection and Test Plan (ITP), or the welding quality plan before the first weld is tested [S1]. Level 1 is the most stringent and is typical for critical pressure-bearing welds, seismic joints, and EN 15085 CL1; Level 2 is the intermediate default for general structural welds including EN 1090 EXC3 and API pipework; Level 3 is the least stringent and applies to non-structural secondary welds at EXC1 [S1].
For a standard EN 1090 EXC2 structural job, the ITP normally cites EN ISO 23278 Level 2 for the sample MT on critical fillet welds, with EN ISO 5817 Quality Level C as the dimensional/visual baseline, and any tighter acceptance is only triggered if the project specification (EN 1993 Eurocode, client spec) explicitly demands it [S1][S3]. Leaving the level blank and asking the inspector to "use engineering judgment" is one of the four classic audit findings that the AQC inspection checklist flags in its MPI procedure guidance [S2].
Matching MT to the EN 1090 execution class
EN 1090-2 defines four execution classes (EXC1 to EXC4), with NDT percentage, acceptance stringency, and welding-coordinator level scaling up with class [S3]. EXC2 is the commercial mainstream, covering commercial buildings up to 15 m, warehouses, industrial halls, standard multi-storey commercial buildings, station canopies, and crushing/screening plant structures, and it requires sample NDT per EN 1090-2 Table 24: typically 10% UT or RT on complete-joint-penetration tension welds and 10% MT or PT on critical fillet welds, on top of 100% visual per EN ISO 17637 [S3].
EXC3 and EXC4 step the NDT percentages up, add stricter acceptance (often EN ISO 5817 Quality Level B), and may require a higher welding coordinator credential (IWT or IWE) rather than IWS; EXC1 drops MT/PT sampling for most fillet categories and is acceptable for agricultural and simple secondary structures [S3]. For a fabricator verifying a sub-supplier, the right first question is not "are you EN 1090 certified" but "which execution class is on the drawing, and can your ITP show the matching MT percentage and acceptance level in writing".
What the audit actually checks on the MT file
An EN 1090 notified-body audit reads the MT file as one linked chain, not as a single test sheet, and the AQC checklist for auditable MPI procedures breaks that chain into seven documents that must reconcile [S2]. They are: the WPS (qualified per EN ISO 15609-1, backed by a PQR per EN ISO 15614-1), the welder qualification per EN ISO 9606-1 with a current 3-year certificate, the MT procedure sheet referencing EN ISO 17638, the MT report with technique, particle type, field strength, lighting, and acceptance level, the visual report per EN ISO 17637, the dimensional/inspection report, and the EN 10204 3.1 mill certificates that prove the base material is actually ferromagnetic and in scope for MT [S2][S3].
If any one of those is missing or refers to a different revision than the ITP, the whole chain breaks, and the typical audit finding is a request for re-test or re-issue, not a soft observation [S2]. A fabricator's MT report template should therefore carry fixed fields for EN ISO 17638 reference, EN ISO 23278 level, EN ISO 17637 visual reference, parent-material grade and thickness range, magnetisation technique, field strength, particle type, lighting/UV-A value, and the signature block for the Level 2 (minimum) or Level 3 inspector with employer statement per EN ISO 9712 or ASNT SNT-TC-1A [S1][S2].
Common failure modes and how to pre-empt them
Four failure modes account for most rejected MT submissions on structural work, and each maps to a checklist line you can verify before the auditor sees the file [S1][S2]. First, MT specified on an austenitic stainless weld: the WPS parent-material field should be checked against the procedure qualification range, and any "stainless" entry on a drawing should auto-trigger PT, not MT. Second, single-direction magnetisation: the procedure should call out at least two orthogonal shots, with a sketch or photo of pole placement on the report for each examination zone. Third, missing or wrong acceptance level: the ITP, the WPS, and the MT report must all carry the same EN ISO 23278 level number, and any deviation must be raised as a non-conformance before the weld is accepted. Fourth, lighting or field-strength numbers missing: the lux and UV-A readings, plus a 2 kA/m field verification (for example, a Hall-effect gaussmeter or a pie-gauge), belong on the report, not just in the inspector's pocket notebook [S1].
A pre-empt checklist that mirrors the AQC "clear, auditable MPI procedures" guidance is a one-page form with those four lines plus the inspector's certificate number, expiry, and the employer's written statement of qualification, and it travels with every MT report so the reviewer does not have to chase the supporting documents after the fact [S2].
Selecting and verifying the magnetic particle tester (equipment side)
Beyond the paperwork, the equipment itself has to be in scope: an AC or DC electromagnetic yoke, a prod gun with calibrated current/timer, a wet bench with circulating bath and stirrer for fluorescent particles, and a UV-A lamp with a 365 nm peak output and a radiometer for the 10 W/m² verification at 300 to 400 mm [S1]. For a structural job that lives mostly on site rather than in a workshop, a portable AC yoke is the default, with the prod method kept in reserve for plate areas where the yoke leg length cannot straddle the weld.
Calibration evidence to keep on file: yoke lift-weight check at the start of each shift (the standard 4.5 kg or 18 kg dead-weight test for AC and DC yokes respectively, per the equipment manufacturer's procedure, commonly cited against ASTM E709 and EN ISO 17638), prod contact pressure and amperage setting, particle batch and certificate of conformity, and lux/UV-A meter calibration with traceability to a national standard [S1]. For deeper reading on how a related process instrument family is specified in similar documentation chains, see the magnetic sensor selection reference and the safety certification reference. For the equipment-class terminology and the difference between a portable tester and a fixed bench, the magnetic particle tester page is the entry point.
EN 1090-2 EXC2 procurement checklist for the buyer
For a buyer placing an EN 1090 EXC2 structural package (the ~70% by tonnage slice of EU steelwork), the verification list before releasing a PO is short and mechanical, and the Indian-fabricator guide lays it out in the same order a notified body would [S3]. Demand the EN 1090-1 factory production control certificate with notified body number and expiry, the welding coordinator's EN ISO 14731 credential (IWS minimum for EXC2), WPS list matched to the joint types on the drawing with PQR references, welder qualification certificates per EN ISO 9606-1 dated within the last 3 years, the ITP naming EN ISO 17637 100% visual plus 10% MT/PT on critical fillets to EN ISO 23278 Level 2, EN 10204 3.1 mill certificates for the plate grades, and the Declaration of Performance template per EN 1090-1 Annex ZA [S3].
If a fabricator is still in the notified-body audit window and cannot yet issue a DoP, the buyer can either accept CPR-scope work shipped under the buyer's own certified entity, or postpone the order; running EXC2 plate work without an EN 1090-1 FPC certificate is the single highest-risk path in EU procurement, regardless of how strong the in-house ISO 9001 system is [S3]. For a broader view of how acceptance levels in one NDT method relate to instrument-grade selection in adjacent process equipment, the magnetic material property reference is a useful cross-check on the parent-material side.
Track the next update by watching two signals: any revision of EN ISO 17638 or EN ISO 23278 published by CEN after 2026-08, and any change in EN 1090-2 Table 24 NDT percentages for EXC2 that would shift the default 10% MT/PT figure; the magnetic drive pump material compatibility reference and the dust particle meter spec reference are not direct crossovers but are useful for QA teams standardising calibration and particulate documentation language across NDT shops.
For related coverage, see GWR Level Meter Compatibility with Switching Repeatability: Spec Map 2026.