We ran a CO₂ line pipe spec against API 5L 47th
A line pipe specification for a dense-phase CO₂ pipeline, written to API 5L 46th edition, checked against the 47th: 10 findings the new edition adds.
By Leo Ruspini — Nuclear engineer, PhD · Co-founder & CEO, normify.ai
In June 2026 API published the 47th edition of API Specification 5L, the line pipe specification behind most oil and gas pipelines. For the first time, it says how to order pipe for CO₂.
Carbon capture projects buy a lot of line pipe, and most of their purchase specifications were written against the 46th edition, from 2018. So we checked what the new edition changes for one of them.
The test
We wrote a line pipe specification for a fictional onshore CO₂ pipeline, Havørn: 48.5 km of 24-inch pipe, 14.3 mm wall, grade X65, longitudinally welded, carrying CO₂ in the dense phase at up to 120 bar. Like many CO₂ specifications, it orders the pipe for sour service as a precaution, and it sets a fracture toughness so that a running crack stops on its own.
It's written to the 46th edition: the chemistry and strength limits of its sour-service tables, its clause numbers, its marking.
We checked it twice:
- Against the 46th edition, the one it was written to.
- Against the 47th edition.
Keeping it fair
The 46th check raised one critical point: the specification doesn't show that the toughness method it uses is valid for dense-phase CO₂. It also flagged that the 46th edition has been superseded. Neither is new in the 47th, so we don't count them.
What the 47th edition adds
10 findings: 1 critical, 7 warnings, 2 minor. Seven are conflicts, where the specification says something the new edition doesn't accept. Three are undocumented, where the new edition needs something the specification doesn't state.
The headline: how the pipe stops a running crack
A CO₂ pipeline that ruptures can unzip for a long distance: dense-phase CO₂ decompresses slowly, and the pressure behind the crack stays high. So the pipe has to be tough enough to arrest it.
The specification sets that toughness with a calculation method that the 46th edition allows for gas pipelines, corrected for high values, and skips full-scale testing.
The 47th edition now addresses CO₂ directly (Annex G). If the line can operate denser than gas phase, only a full-scale burst test is permitted to set the arrest toughness (G.1.1, G.13). The calculation the specification relies on no longer counts. That changes the test programme, the cost and the schedule of the pipe order, not just a number in a table.
The other conflicts
- Maximum tensile strength in sour service (Table H.2). The specification allows up to 760 MPa for X65, the 46th edition limit. The 47th edition lowers it to 685 MPa.
- Sour class (H.2, H.4.1.3, H.8.1). The 47th edition introduces sour classes A, L and F, which have to be agreed, and the class letter becomes part of the grade and the marking. The specification has neither.
- Record retention (13.1). Three years in the specification, five in the 47th edition.
- Marking (11.1). The specification's paint stencil inserts the API Monogram data into the mandatory marking sequence. The Monogram has moved out of API 5L, into API Q1.
- Clause references (Annex K). The lamination inspection references point to the old clause numbers. In the 47th edition, one of them now points to an exemption for other sour classes.
What the specification doesn't document
- Local hard zones (H.3.3, H.4.4.3, K.6). The 47th edition asks the purchaser to decide whether hard spots in plate rolled by thermomechanical processing are a cracking risk, and to set hardness controls if they are. It follows field failures of large-diameter pipe in sour service. The specification orders exactly that kind of pipe and says nothing about it.
- Magnetic particle inspection of the pipe ends (K.2.1.5). The specification uses ASTM E709. The 47th edition lists ISO 10893-5 or ASTM E3024.
- Hydrostatic test pressure (10.2.6). The specification points to the clause for the test pressure by its 46th-edition number. In the 47th edition, that number is a different clause.
What this means
Nothing in the specification was wrong when it was written. The standard moved under it. And a purchase specification is a contract: whatever it says is what the mill makes, tests and certifies.
The edition change matters most where it touches money and schedule: a full-scale burst test for the fracture control, a lower strength ceiling the mill has to meet, a new sour class to agree. The rest is clause numbers and paperwork, which is exactly what gets missed when someone re-reads a specification by hand.
That's the check normify runs: a document against the edition it now has to meet, clause by clause.
One run per edition, on a fictional specification we wrote. Counts vary between runs; the findings above are from the reports of 29 October 2026.
Sources
- API, API Announces 47th Edition of Foundational Line Pipe Standard, 2 June 2026: api.org
- API Specification 5L, 46th edition (2018) and 47th edition (2026). Licensed API documents; this article refers to clause numbers only.