A transport package for radioactive material has one job that matters above all others: whatever is inside stays inside. Not just in normal handling, but after the drop, the stack, the puncture and the fire that the regulations require it to survive.
Almost all of that containment depends on a seal a few millimetres across, sitting in a groove on a lid flange. Testing that seal is what CALT leak testing does.
The double O-ring arrangement
Closure lids on transport packages are not sealed with a single O-ring. They carry two, concentric, in separate grooves — an inner seal and an outer seal — with a small void between them.
That void is the whole point. It is called the interspace, and it is deliberately provided with a test port.
Sealing twice gives redundancy, but redundancy is not the reason for the geometry. The reason is that the interspace can be tested independently of the package. Draw the interspace down and hold it, and you are testing both seals at once — without opening the package, without needing access to its interior, and without introducing anything into the containment volume.
If the inner seal leaks, gas from the contents side bleeds into the interspace. If the outer seal leaks, atmosphere bleeds in. Either way the interspace fails to hold, and the rate at which it fails is the measurement.
What the CALT measures
We use the CALT 9 leakage tester, developed by Croft Associates specifically for this application.
It is not a pass/fail box. It measures the leakage rate and produces a printed record of the test parameters alongside the result — which matters more than it sounds, because on packaging work the evidence is the deliverable. A test that found nothing but cannot demonstrate what it did, or at what sensitivity, is not evidence of containment.
Leakage rates in this field are quoted in reference cubic centimetres per second, and the numbers are small — well below anything a pressure decay test across a whole package volume could resolve. That sensitivity is achievable precisely because the volume under test is the interspace rather than the package.
Why the regulations demand it
Packages for radioactive material are type-approved against containment requirements, and containment has to be demonstrated rather than asserted. The standard governing that demonstration in the UK is BS EN ISO 12807:2021, which covers leakage testing on packages for the safe transport of radioactive material.
The regime distinguishes several kinds of test, answering different questions:
- Design testing — proving the design achieves the containment standard, usually on prototypes, including after the regulatory impact sequence.
- Fabrication testing — proving a particular manufactured package matches the approved design.
- Periodic testing — proving a package still performs after time in service.
- Pre-shipment testing — proving that this closure, made up on this occasion, is sound before the package moves.
That last one is what most people mean by leak testing, and it is the one that has to be quick, repeatable and possible on site — because it happens every time a package is loaded and closed, usually in a working facility against a schedule.
An interspace test satisfies it without disturbing the contents, which is the practical reason the double O-ring geometry became standard.
What actually goes wrong with a seal
The failures are consistent, and knowing them shortens a lot of investigations.
Damage during make-up. An O-ring nipped, twisted or rolled as the lid was seated. The most common cause by a distance, and a handling issue rather than a materials one.
Contamination on the sealing face. Grit, swarf or dried residue on the flange or in the groove. A particle a fraction of a millimetre across holds the seal off its face and gives a leak path.
Surface damage to the flange. A score across the sealing face — a dropped tool, or a lid dragged rather than lifted — creates a channel the O-ring cannot bridge.
Elastomer ageing. O-rings take a compression set with time, temperature and radiation dose. A seal compressed for years no longer springs back, which is why elastomers in radiation service carry a defined life.
Incorrect torque or sequence. Lid fasteners tightened out of sequence, or to the wrong value, load the seal unevenly. The leak is usually opposite whichever side went tight first.
Wrong specification. A replacement O-ring of the right size in the wrong material or hardness. It fits, it looks correct, and it behaves differently.
A failed interspace test does not tell you which of these it is. It tells you the closure is not sound — before the package moves rather than afterwards.
How a test runs
The closure is made up to the operating procedure: correct O-rings, clean and inspected sealing faces, fasteners to the specified torque and sequence. That make-up is part of the test, because a test on a badly made closure only proves the closure was badly made.
The instrument is connected to the interspace test port and the interspace is drawn down. A stabilisation period is observed before measurement begins — measuring too early reads thermal effects and outgassing rather than leakage, which is the classic way to fail a perfectly good seal.
The leakage rate is then measured over the specified period and compared against the acceptance criterion for that package and test category. The CALT prints the parameters and the result.
Sensitivity is verified against a calibrated reference before testing, not after. If the instrument cannot demonstrate it would have detected a leak of the size the criterion turns on, a passing result means nothing — and that verification record is what an assessor asks to see first.
Where the work happens
Our technicians are trained and experienced on the Croft Associates equipment, and the work is almost entirely on client sites. Packages are heavy, licensed and not readily moved, so the instrument travels rather than the package.
That takes the team across the UK and into Europe for nuclear clients including LLW Repository and Sellafield Ltd. It also means the constraints are site constraints: access to the test port, an environment stable enough for the measurement, and a slot in a schedule where the package is closed but not yet moved.
The scheduling point is worth planning around. Pre-shipment leak testing sits at the end of a loading sequence and immediately before a movement, so it is on the critical path by construction. A failure at that point stops a consignment. Building in time to strip, inspect, re-seal and re-test — rather than assuming a pass — is what stops a tired O-ring becoming a missed shipment.
How it fits with the rest
Leak testing proves the closure. It says nothing about the package body, and the two are often confused in a scope.
Where the containment boundary itself is in question — a weld in the vessel wall, a repair, corrosion on a body that has been in service — that is a job for ultrasonic testing or radiography, and on stainless components handled in a fabrication environment, ferroxyl testing for free iron contamination that would otherwise pit the surface in service.
On a periodic examination it is common to need several of these together, and they are worth planning as one visit rather than three. We carry out leak testing alongside the other NDT methods under one quality framework, so a package examination produces one set of records rather than a folder from each supplier.
What a report should contain
The package identity and its closure, the test category, the instrument and its calibration status, the sensitivity verification against the reference leak, the stabilisation and measurement periods, the measured leakage rate, the acceptance criterion applied, and who carried out the test and to what qualification.
A result without the sensitivity verification is the one to be careful of. On paper it looks identical to a good result, and it means considerably less.
If you have packages needing pre-shipment or periodic testing, or a closure that has failed and needs investigating, that is the work this equipment exists for — and it is a conversation worth having before the shipment is booked rather than after the test fails.
TECHNICAL REVIEW — DELETE FROM THE LINE ABOVE, DOWN
Scheduled for 2026-08-20. Not to go live until a Level 3 or the RPA has read it.
Check: REWRITTEN 2026-08-17. My first version was completely wrong — I described chemical absorption leak testing, a different method entirely. CALT is the Croft Associates Leak Tester: interface seal testing between double O-rings on radioactive material transport packages. Corrected against the /calt-leak-testing/ service page and Lee. Because I got this badly wrong once, read it line by line rather than skimming. Specifically confirm: the interspace evacuate-vs-pressurise description, the four test categories as named, reference cm3/s as the unit we quote, and whether naming LLW Repository and Sellafield Ltd is commercially acceptable (it is on the service page, but check).
Image to shoot: The CALT 9 connected to a package interspace test port, instrument display legible, package closure in frame. If a client site cannot be photographed, shoot it on a demonstration flange.
Internal links already in the text: /calt-leak-testing/, /ultrasonic-testing/, /xray-and-gamma-radiography/, /ferroxyl-testing/, /non-destructive-testing-ndt/
