ADR special provision 674
Over-moulded cylinders: an alternative periodic inspection method using destructive sample testing, traceability and statistical evaluation.
Sets an alternative periodic inspection and test for over-moulded cylinders (defined in 1.2.1) under 6.2.3.5.3.1: the test in 6.2.1.6.1 (d) is replaced by destructive tests on samples, plus extra tests specific to the over-moulding. It covers serially produced cylinders on welded steel shells to EN 1442:2017, EN 14140:2014 + AC:2015 or Directive 84/527/EEC, with self-closing valves. Cylinders are grouped into a basic population per year, over-moulder and shell maker, split into sub-groups per owner, and individually traced via an electronic ID and the owner's central database. Random samples undergo burst, adhesion and peeling/corrosion tests after 3 years in service and every 5 years after, evaluated statistically on a Sample Performance Chart. A failing sub-group is segregated and not filled pending root-cause work with the competent authority or Xa-body. Filling centres must meet P200 (7) and (11), read the electronic IDs, access and update the database, and run a certified ISO 9000-type quality system.
| Value | Applies when |
|---|---|
| 3 years | first destructive sample tests after entry into service |
| 5 years | interval of subsequent destructive sample tests |
| 0.5 N/mm² | minimum adhesion value of polyurethane |
| Ri2 corrosion grade | maximum corrosion grade, peeling and corrosion test |
| 95 % / 99.9999 % confidence level / population fraction | lower tolerance limit for burst results |
Conditions
- (a) General: serial over-moulded cylinders on welded steel shells per EN 1442:2017, EN 14140:2014 + AC:2015 or Directive 84/527/EEC; water kept off the steel shell; self-closing valves.
- (b) Basic population: one over-moulder, new shells from one maker, one calendar year, same design type, materials and processes.
- (c) Sub-groups: one per owner within the basic population.
- (d) Traceability: shell marks repeated on the over-moulding, electronic ID per cylinder, owner database kept for the life of the sub-group.
- (e) Random sampling within each sub-group, sizes per the (g) table.
- (f) Tests: burst test replaces 6.2.1.6.1 (d); adhesion and peeling/corrosion tests added; after 3 years then every 5 years.
- (g) Statistical evaluation via a Sample Performance Chart, normal or Weibull k3 factors, no burst result below test pressure.
- (h) Failures: sub-group segregated, root cause investigated with the competent authority or Xa-body; scope widened if not confined.
- (i) Filling centres: P200 (7) and (11) compliance, ID readers, database access and update, certified quality system.
Note. Spans pages 657–661. The sampling table and k3 tables on pages 659–661 are heavily scrambled in the extracted text (sample sizes and factors run in separate columns); sample sizes and k3 values are not reproduced here — read them from the printed table.
Refers to: 1.2.1; 6.2.3.5.3.1; 6.2.1.6.1; 6.2.3.9; P200; 4.1.4.1.
Carried by 5 Table A entries
UN 1011 · UN 1075 · UN 1965 · UN 1969 · UN 1978
Source: ADR 2025 (ECE/TRANS/352), Volume I, section 3.3.1, special provision 674, page 657.
Paraphrase in FreightUtils' own words, not the official text. Not yet reviewed.
ADR 2027 amendments (accepted, applying from 1 January 2027) are not yet reflected on this page.