EN 10204 3.1 vs 3.2: certificate types and how to get the data into your ERP

The four EN 10204 certificate types

EN 10204 defines the types of inspection document a manufacturer can issue for metallic products. Type 2.1 is a declaration of compliance with the order. Type 2.2 is a test report based on the manufacturer's own inspection. Type 3.1 and type 3.2 are inspection certificates based on specific inspection, which means the tests were done on the products supplied or on a part of the consignment they belong to.

German buyers use the titles from DIN EN 10204: Werksbescheinigung for 2.1, Werkszeugnis for 2.2, and Abnahmeprüfzeugnis for 3.1 and 3.2. The 2004 edition has these four types. Older certificates can still cite 2.3, 3.1.A, 3.1.B or 3.1.C from the 1991 edition.

EN 10204 3.1 vs 3.2

A 3.1 certificate is validated by the manufacturer's authorised inspection representative, who must be independent of the manufacturing department. A 3.2 certificate is also signed by the purchaser's representative or by an inspector named in official regulations. The standard itself is EN 10204:2004, sold through national standards bodies, and the wording here follows its definitions. If your purchase order says 3.2 and the supplier sends 3.1, that is a finding on its own.

EN 10204 says what kind of document it is. It does not say which values go on it. That comes from the product standard in your order and, for steel products, from the list of information in EN 10168. The layout still differs from mill to mill, and the language often does too.

The fields worth capturing

Most goods-in checks need the same short list. Capture these as separate columns, not as one block of text.

• Certificate number, issue date and certificate type (2.1, 2.2, 3.1, 3.2).
• Manufacturer and the order or delivery reference it answers.
• Product description, dimensions, quantity or weight, and delivery condition.
• Material grade and the product standard with its edition, for example S355J2 to EN 10025-2.
• Heat number (also called cast or melt number), plus the batch or lot if the certificate has one.
• Chemical composition from the ladle analysis, in percent, one column per element.
• Mechanical results: yield strength, tensile strength and elongation, with units.
• Impact test results where the order asks for them, with test temperature.
• Name, stamp or signature of the inspection representative.

One certificate can cover several heats, or one heat can appear on several lines. Decide the row before you extract. In most ERP imports it is one row per heat per delivered item, with the certificate number repeated on each row.

Checks that catch real errors

Extraction is only useful if a wrong value gets noticed. These checks need no judgement and can run on every certificate.

Right grade, wrong heat number: the certificate belongs to someone else's steel.

1. The heat number matches the material. The number on the certificate must match the marking on the product and the one on the packing list. A certificate for the right grade with the wrong heat is not a certificate for your delivery.
2. The type matches the order. Compare the certificate type with the one in the purchase order.
3. The grade and standard match the order. Compare the designation as written, including the edition if the order names one.
4. Values sit inside the product standard. Yield strength for S355J2 must reach at least 355 MPa for thicknesses up to 16 mm. The minimum drops for thicker products, so read the limit for the actual thickness in the standard.
5. Units and decimals are read correctly. MPa and N/mm² are the same unit. A decimal comma in 0,18 percent carbon must not become 18 or 0 and 18.
6. Nothing is missing. An empty element column is different from a zero. If the order asks for impact tests and the certificate has none, flag it.
7. The document is signed or stamped. A 3.1 certificate with no validation does not meet the definition.

A person should look at anything that fails a check. A value that passes still needs a spot check against the page, because a misread digit can land inside the limits.

Why scanned certificates are hard

Certificates from different mills share almost no layout. Some are tables and some are free text, and stamps often sit over the printed values. Chemical results can run across the page in a dozen narrow columns, and a small misread shifts every value after it. Certificates from outside the EU may be in another language with English headings added. Rules that depend on a fixed position on the page break on the second supplier.

What Corteal does with them today

The Read page shows a sample mill certificate row in an ERP import table, with a heat number and a page reference. The demo values are invented, and the percentages beside them are typed in by hand. They measure nothing. A confidence score is not a guarantee explains why.

In a free pilot, our team extracts up to 20 of your hardest documents and checks every value by hand before you see it. You get rows in your ERP's import format, with the page of each value and a flag on anything uncertain. You then confirm the right values on five files, and we score every field against your confirmation. We do not publish an accuracy figure, because we have not measured one on real documents yet.

If you hold a stack of certificates that someone retypes each week, that stack is a good pilot. Start a pilot and send the ones that cost the most time.