Weld Maps & Weld Registers

When a pressure vessel, pipe spool or steel structure fails, the first questions are always the same: which weld, who made it, to which procedure, and what did the NDT say? Weld traceability exists so those questions can be answered years after handover, from the documentation alone. The codes formalise it, but the driver is practical: without a numbering system, "the weld near the nozzle" is not an auditable statement.

The industry solved this with a pair of documents that mirror each other: a marked-up drawing (the weld map) and a table (the weld register). Everything else in the welding chapter of a data book — the WPS, PQR and WPQ records, the NDT reports, the filler material certificates — hangs off the weld numbers these two documents define.

The weld map is usually a copy of the general arrangement drawing, detail drawing or isometric, with a numbered balloon at every weld joint. The numbering scheme is agreed before fabrication starts: sequential per drawing, per line number, or per assembly — what matters is that every physical joint has exactly one identifier, and that repairs and re-welds get traceable suffixes (W-014-R1) rather than new numbers.

Shop welds and field welds are distinguished (typically SW/FW prefixes), and on piping work the map often lives on the isometric itself. The map answers the spatial question a table cannot: where is weld 14, and what does it join?

The weld register — also called weld log, weld summary or weld history sheet — is the database view. One row per weld number. It is a living document during fabrication: rows are created when joints are fit up, welder stamps are recorded as beads go down, and NDT columns fill in as reports come back. At handover it freezes into the version that enters the MRB.

Because it accumulates gradually, the register is where traceability usually breaks. A welder ID recorded nowhere, an NDT report number never filled in, a repair executed but not logged — each is invisible in the shop and glaring at review. Keeping the register current, not reconstructing it afterwards, is the single best predictor of a clean data book review.

Every column is a pointer into another chapter of the data book. That is the register's real function: it is the index of the welding chapter, the place where a reviewer starts every spot check.

Read one register row from left to right and you walk the full chain of custody for that joint: the base materials trace back via heat numbers to EN 10204 material certificates; the WPS traces to its PQR; the welder stamp traces to a WPQ valid on the welding date; the filler batch traces to a consumable certificate; and the NDT report numbers trace to signed reports with matching acceptance criteria. Five independent document trails, all converging on one weld number.

This is why reviewers love spot checks: verifying one weld end-to-end tests the integrity of the entire documentation system. If weld 14 checks out — and weld 41, picked at random, does too — confidence in the remaining hundreds rises sharply.

EN 1090-2 ties weld traceability to execution class: EXC3 and EXC4 require records identifying welder, procedure and location per weld; many project specs pull the same requirement down to EXC2. ASME BPVC (VIII, B31.3) requires the manufacturer to maintain welder and procedure traceability per joint — the map/register pair is the industry's standard way to demonstrate it. NORSOK and most oil & gas operator specifications require weld maps explicitly, and PED 2014/68/EU conformity assessment leans on the same records to prove qualified personnel and procedures were used.

The practical rule: assume the map and register are required, agree their format at kick-off (ideally in the ITP), and never let them lag fabrication.

• Map ↔ register mismatches — numbers on the drawing missing from the table, or rows without a balloon.
• Broken NDT cross-references — report numbers that don't exist, or reports covering weld numbers the register doesn't mention.
• Welder not qualified on the date — a stamp whose WPQ expired or didn't cover the process/position.
• Unlogged repairs — an RT report showing a repair excavation with no R1 entry in the register.
• Illegible scans — a hand-kept register photocopied to the point of uselessness.

All five are cross-reference failures, not welding failures. The welds may be perfect; the book still bounces. See the complete MRB checklist for how the welding chapter sits among the others.

In the assembled Manufacturing Data Book, the welding chapter conventionally opens with the weld map, followed by the weld register, then the WPSs, PQRs and WPQs it references. NDT reports usually live in their own chapter, reachable via the register's report numbers. Compiled digitally, those references can become actual links — a reviewer clicks from register row to NDT report instead of leafing through binders — which is precisely the kind of cross-referencing a purpose-built platform automates.