NDT Inspection Methods for HDPE Welds: Ultrasonic Testing of Butt and Electrofusion Joints
Nov 09, 2026
Leave a message
Non-destructive testing enters a polyethylene project at a different point than it does on steel. On a steel line you radiograph the weld because the alternative is cutting a coupon out of a joint that is already in the trench. On a PE line the same commercial logic applies, but the physics do not transfer: polyethylene is low density, low acoustic impedance, and it scatters and attenuates a high-frequency beam far more than steel does. The method has to be adapted, not borrowed.
That distinction is worth money. On a gas PE100 line, the joint that fails in service is almost never the joint that failed a pressure test - it is the one that passed it. Ultrasonic examination is the only routine method that looks at the fusion plane of the specific joint you are about to bury. Everything else either stops at the surface or describes a different joint from the one in the ground.
This page covers what each method can prove, how the scan is actually set up on a butt fusion and on an electrofusion joint, what the common defects look like on the display, and where the technique stops being useful. If you are still deciding whether a suspect joint is a cold weld at all, start with detecting a cold weld and come back here when the question becomes "prove it".
Why polyethylene does not behave like steel under a probe
Four properties decide the whole method, and each one changes a setting an operator would otherwise carry over from steel work.
- Sound travels at roughly 2,400 m/s in HDPE, against 5,900 m/s in steel. At the same frequency the wavelength is more than twice as long, so the smallest resolvable flaw is more than twice as large. A defect that a 5 MHz probe finds easily in a 10 mm steel wall needs a wider aperture and a lower frequency in a 10 mm PE wall.
- Attenuation is high and rises steeply with frequency. Most PE procedures work between 2 and 5 MHz. A 10 MHz probe that would be routine on thin steel produces almost nothing through 20 mm of polyethylene.
- The defect is a plane, not a volume. A cold weld has essentially zero thickness. There is no void to fill with echo - there are two surfaces pressed together. The reflection depends on contact pressure, couplant and surface finish, which is why gain settings and a reference block are not optional.
- The material creeps. Probe pressure, wedge contact and even the weight of the cable change the coupling. Scanning technique matters more than it does on steel, not less.
The four methods, and the question each one answers
The most common mistake in PE inspection is treating these as a hierarchy, where the expensive one replaces the cheap ones. They answer different questions. A project usually needs two of them, in a defined order.
| Method | What it can prove | What it cannot prove | Where it is defined |
|---|---|---|---|
| Visual examination and bead gauge | Bead geometry, symmetry, surface condition, ovality, visible contamination | Anything below the surface. A cold weld produces a normal-looking bead. | EN 13100-1; imperfection classes in DVS 2202-1 |
| Destructive coupon - bend, tensile, peel | That the procedure, the machine setting and this crew can make a sound joint | Nothing about any joint other than the one that was cut out | EN 12814-1; ISO 13953; ISO 13954 / 13955 for peel |
| Conventional ultrasonic testing, single crystal | Large voids, thick-wall laminations, gross unbonded areas | Thin walls, small couplers, and anything needing lateral resolution | EN 13100-3 |
| Phased array ultrasonic testing (PAUT) | The fusion plane of an installed joint: planar lack of fusion, voids, root defects, gap in an electrofusion body | Thermal degradation, molecular weight, anything that leaves no acoustic discontinuity | EN 13100-3, assessment against DVS 2202-1 classes |
The coupon test is the one people misread. It is a test of the process, not of the pipeline. Cut a ring out of joint number 214 and you have learned something important about joint 214 and something statistical about the rest. On a water main that statistics is usually enough, because a leak is contained and repairable. On a gas main it is not, which is exactly why the fourth method exists.
Setting up a scan on a butt fusion joint
The fusion plane is vertical and runs the full wall thickness directly beneath the bead. The beam has to cross it, which means an angled incidence - typically 45° to 70° - and it means the beam must not run parallel to the plane, or the reflection goes back along the same path and tells you nothing about bonding.
Three practical consequences follow, and they are where most field scans go wrong:
- One pass from each side of the bead. The far half of the fusion plane sits in the acoustic shadow of the near half. A single-sided scan can pass a joint that is unbonded on the side away from the probe.
- The bead is an obstacle, not a landmark. The collar changes the standoff and the coupling. Procedures either scan in front of the collar with a standoff wedge or work through it with a contoured shoe - but the choice is written into the procedure, not improvised at the trench.
- A reference block of the same grade and wall section. Polyethylene attenuation varies with grade, filler and wall thickness. Without a block made from the same material, the gain setting is a number someone liked, and an indication at 40% screen height means nothing in particular.
Where the joint has already been made and the pipe is still clamped, scanning is straightforward. Where the pipe has been lowered and backfilled, it is often impossible without excavation, which is why the inspection point has to be planned before the first joint rather than after the first failure.
What changes on an electrofusion joint
An electrofusion coupler is harder to inspect than a butt fusion joint, and for one reason: the heating wire is inside the component, at a known depth, and it is an excellent reflector. The coil is not noise - a competent operator uses it as a built-in depth reference - but it does mean the interesting volume sits between the wire and the pipe surface, in a gap of a few tenths of a millimetre.
The indications that matter are different too. In electrofusion the failure mode is usually not a single planar defect but a band of unbonded material running alongside the coil, caused by an oversized gap, an unscraped oxide layer or contamination. The scan is looking for continuity of contact along the insertion length, and for a gap that closes at one end and opens at the other.
Size limits are real. Below roughly dn63 the wall and the fitting body are too thin for a reliable angled beam, and most procedures exclude those sizes. If the specification demands 100% inspection of small couplers, that requirement is not satisfiable with ultrasound and should be resolved before tendering, not on site.
Reading the display: what the common defects look like
Two shapes carry most of the diagnostic value. A line across the display at a constant depth is a planar reflector - lack of fusion, a cold weld, an oxide layer. A scatter of dots is volumetric - porosity, contamination, moisture. The distinction matters because the two have different causes and different consequences: a planar defect runs the full bead and will open under load, while scattered porosity reduces the bonded area but may still carry the design pressure.
| Indication | Most likely cause | What to do with the joint |
|---|---|---|
| Continuous line at mid-wall, full bead length | Cold weld - insufficient heat soak, long changeover, low pressure | Cut out. Not repairable by re-heating. |
| Scattered point echoes through the wall | Moisture, dirt or oil at the interface; wet trench work | Assess against the acceptance class. Usually reject on gas, assess on water. |
| Echo merged with the inner surface, backwall lost | Lack of fusion at the root - typically facing or alignment related | Cut out. The root is the highest-stress region of the joint. |
| Thin echo immediately under the outer surface | Oxidation layer - scraping skipped or done too shallow | Cut out. Re-welding over an oxide layer reproduces the defect. |
| Entry echo and backwall only, nothing between | Sound joint, as far as ultrasound can tell | Accept, and still record the parameters that produced it. |
Which document actually decides acceptance
Three documents do three different jobs, and confusing them is how projects end up arguing at handover. ISO 21307 or DVS 2207-1 sets the welding procedure - temperatures, pressures, times. EN 13100-3 sets how the ultrasonic examination is performed. DVS 2202-1 classifies the imperfections that the examination finds.
None of the three tells you whether a particular indication is acceptable on your pipeline. That is the project specification, and it should reference an imperfection class rather than describe a feeling. The useful test of a specification is whether two different inspectors, given the same display and the same document, reach the same verdict. If the answer depends on who is holding the probe, the document is not finished.
For the process side of the same argument - what the parameters should be and why a pressure table does not exist - see how to calculate butt fusion pressure.
Where the technique stops
Ultrasonic testing finds discontinuities. It does not measure molecular weight, and it cannot see thermal degradation. A joint that was overheated has a continuous, well-bonded interface and a shortened chain length; it will pass every scan and fail in slow crack growth three years later. That defect is only caught by parameter logging, by bead appearance, and by not letting the plate run hot in the first place.
The other limits are practical rather than physical: wall thickness below about 6 mm, couplers below dn63, joints that are already backfilled, and access blocked by a coating or a clamp. Each of these is a reason to plan the inspection window, not a reason to skip the inspection.
Operator qualification deserves its own mention. A phased array set produces a picture that looks authoritative whether or not the operator understands what produced it. Reference blocks, written procedures and a second-reader review of any reject are the cheap controls. The expensive control is discovering at commissioning that nobody ever read the scans.
Planning the inspection before the first joint
The plan should be written while the joints are still countable, not after a leak. Three decisions carry most of the value:
- Rank joints by consequence, not by count. A road crossing, a river bore, a tie-in between two spreads and a deep main under a carriageway are not equivalent to 400 m of straight run in a field. Scan the ones that cannot be reopened.
- Fix the inspection point in the sequence. Scan after cooling and before lowering. A joint inspected after backfill has not been inspected; it has been excavated.
- Record the parameters with the scan. A pass with no parameter record tells you the joint was bonded and nothing about why. When the next joint fails, the log is the only thing that explains the difference.
That record is also what makes a construction sequence and quality control system auditable rather than decorative, and it is the same evidence a utility will ask for at handover.
Frequently asked questions
Can ultrasonic testing replace destructive sampling?
No, and it should not be sold that way. Destructive sampling proves the procedure and the crew; ultrasonic examination proves an individual joint. A project needs the first to know that the process is capable and the second to know that a specific buried joint is sound. Dropping the coupon test leaves you with no evidence that the process was ever right; dropping the scan leaves you with no evidence about the joint you cannot reopen.
How long after welding can a joint be scanned?
Once the joint has cooled to ambient and the clamps are released, the interface is stable and can be examined. Scanning a joint that is still cooling is not useful, because the acoustic properties of the material are still changing and the coupling is unreliable. In practice the inspection follows the cooling time, which is also the point at which the joint can be safely handled.
Does a scan damage the joint?
No. Ultrasonic examination is non-destructive by definition - the probe is coupled to the outer surface with gel or water and never penetrates the material. The only requirement is that the couplant is cleaned off afterwards, since a residue on a PE surface attracts dirt and can affect subsequent marking or coating.
Is phased array necessary on a water main?
Usually not for every joint, and that is a commercial decision rather than a technical one. The argument for scanning a water main is the same as for gas - a buried joint cannot be inspected later - but the consequence of failure is lower and contained, so sampling plus a pressure test is often proportionate. The joints that justify a scan on a water main are the ones that cannot be excavated: crossings, bores and connections to existing mains.
What does a failed scan actually require?
Cut out and re-weld. There is no accepted field repair for a planar lack of fusion in a PE joint, and re-heating a suspect joint does not restore the interface. The cut-out has to clear the heat-affected material on both sides, using the same logic as a cold weld cut-out.
Related reading
- Cold weld in HDPE butt fusion - how to prove one exists and how to cut it out.
- HDPE pipe fitting defects: finding the cause before you re-weld - the symptom-to-parameter map this page sits under.
- Electrofusion welding problems: 12 failures, the signs and the fix - the failures the scan is looking for.
- Why electrofusion joints fail: the three root causes - contamination, geometry and early release.
- HDPE butt fusion fittings: construction sequence and quality control - where inspection sits in the joint cycle.
- PE pipe fusion welding: butt fusion and electrofusion compared - which method suits which site condition.
- Cold fusion and false fusion in HDPE joints - the nine causes behind the defect this test is looking for.
Send Inquiry














