PP-R Joint Failure: Proving Whether It Was the Pipe or the Weld

A distributor in Amman gets a call on a Tuesday. Eleven joints on a four-month-old riser have wept at the socket, the contractor has already cut two of them out and thrown them in a skip, and the developer wants the whole floor re-piped at somebody else’s cost. The contractor says the pipe was bad. Your customer, who bought forty pallets from you, says the same thing louder. You have a failed joint in a plastic bag on your desk and about a week before somebody issues a credit note.

The instinct is to cut the joint in half and look at it. That instinct is right, and it is also where most of these disputes go wrong — because a cut face answers a narrower question than the one you are being asked. Identifying the defect tells you what happened at that socket. It does not tell you who pays. Those are two different investigations, and only one of them is settled with a hacksaw.

This article is about the second one: what physical and documentary evidence actually separates a claim against your supplier from a claim against your customer’s installer, which published tests discriminate between them, and what to assemble before the argument hardens. If what you need first is defect identification — which of the six fusion defects you are holding, and how each looks on a cut section — that work lives in our companion article on diagnosing PP-R fusion joints that leak. Read that one to name the defect. Read this one to decide what to do about it.

What the Cut Face Can Prove, and What It Cannot

A sectioned socket joint is genuinely diagnostic for a short list of things. You can see whether the melt zones interpenetrated or merely touched. You can see whether the pipe reached full socket depth, because the unfused annulus at the end of the run is visible. You can see displaced material intruding into the bore. You can see a contamination line as a discontinuity that runs where a bond should be continuous.

What it cannot tell you is anything about the population the joint came from. One failed joint is one data point about one socket made by one person on one day. It carries no information about whether the other four hundred joints on that riser were made the same way, and no information at all about whether the pipe met its standard before anyone picked up a welding iron.

It also cannot tell you the sequence. A joint that was made competently and then abused — a riser dropped into service without allowance for thermal movement, or pressure-tested at the wrong figure, or restrained so it could not expand — can present at the cut face in ways that overlap with an under-fused joint. The defect you can see is the end state, not the history.

There is now a published visual reference for this rather than an argument between two opinions. ASTM F3722, the standard practice for heat fusion joining of polypropylene pipe and fittings, was published in November 2024 and covers butt, socket, sidewall outlet and electrofusion. It carries four quality-assurance appendices with images of acceptable and unacceptable joints of each type, and it applies to PP pipe meeting ASTM F2389, CSA B137.11 or ISO 15874. When a contractor tells you a joint looks fine, that document is a neutral place to disagree.

The Two Defendants, and Why They Need Opposite Evidence

Every PP-R joint failure resolves into one of two commercial positions, and confusing them is what turns a recoverable claim into a write-off.

Position one: the pipe or fitting was never compliant. The compound was wrong, degraded, over-filled with regrind, or the wall was thinner than the class and pressure required. The joint then failed at a stress it should have carried. Your counterparty is your supplier. The evidence is material evidence, and critically, it does not come from the failed joint — it comes from unused stock from the same batch, because that is the only sample that has not been through a welding iron and four months of service.

Position two: the pipe was compliant and the joint was made badly. Wrong iron temperature, wrong dwell, dirty socket, insufficient insertion, summer timings used in a cold stairwell. Your counterparty is the installing contractor. The evidence is process evidence — method statements, welder qualification, iron calibration records, site temperature — plus the joint itself.

The asymmetry matters to a distributor specifically. If it is position two, you are a witness, and your commercial exposure is goodwill and freight. If it is position one, you are in the middle: liable to your customer and claiming up the chain to the factory, and you need evidence good enough to survive both conversations. That is why picking the wrong position early is expensive. Spend three weeks arguing the pipe was bad, fail to prove it, and you have simultaneously annoyed your supplier and lost credibility with your customer.

The Material Side: The Four Numbers a PP-R Pipe Is Judged Against

If you are going to allege that the pipe was non-compliant, you are alleging that it would fail a test that a compliant pipe passes. It is worth knowing exactly what those tests are, because “we sent it to a lab” is not a claim — “it failed the 95 °C / 165 h point” is.

ISO 15874-2:2013 Table 10 sets four hydrostatic acceptance points for PP-R pipe. Each is run on three test pieces, water-in-water, with Type A end caps. The pipe shall not burst.

Note the last row and start it immediately if you are going to run it at all. A 1000 h test is roughly six weeks on the rig before anyone reads a result, which is longer than most commercial disputes stay polite. If your contract with the factory names that point, the practical consequence is that the clock starts the day you decide, not the day you finish arguing internally.

The strongest single test on the material side is not in that table. It is the thermal stability requirement in Table 11: 1,9 MPa hoop stress at 110 °C for 8760 h, water-in-air. That is a full year at a temperature no drinking-water system ever sees, and it is designed to expose stabiliser packages that were skimped. A compound loaded with poorly stabilised regrind fails there long before it fails anything else. It is also, for obvious reasons, a test you cite from an existing certificate rather than commission mid-dispute.

Two much cheaper checks are worth running on retained stock in the first week, because they are fast and they either strengthen your position or stop you wasting money.

Longitudinal reversion must be no more than 2 %, tested by Method B of ISO 2505 with PP-R held at 135 °C — note that is a lower oven temperature than the 150 °C used for PP-H and PP-B, and testing PP-R at the wrong one produces a meaningless number. Exposure is 1 h for walls up to 8 mm, 2 h from 8 to 16 mm, and 4 h above that.

Separately, pipe declared opaque must not transmit more than 0,2 % of visible light when tested to ISO 7686. High reversion points at a rushed extrusion line with frozen-in stress — a manufacturing story, and one a lab can give you in days rather than weeks.

The Joint Side: How ISO 15874-5 Actually Tests an Assembly

Here is the distinction almost every page on this subject misses. ISO 15874-2 tests pipe. Your failure was at a joint, and joints have their own part of the standard.

ISO 15874-5:2013 covers fitness for purpose of the system, and its scope is explicit: it applies to PP pipes, fittings, their joints, and to joints with components of other plastics and non-plastics materials. That last clause is the one to remember when a brass-insert transition or another brand’s fitting is in the failed run — mixed-origin assemblies are inside the standard, not outside it.

The internal pressure test for a joint assembly does not use a fixed pressure. You calculate it:

The design stress values you need for the denominator are published in ISO 15874-2:2013 Table A.2:

Which class applies is set by the design temperatures in ISO 15874-1 Table 1 and is a contractual fact about the project, not something you choose after a failure — our guide to the EN ISO 15874 service classes walks through picking the right one.

Part 5 also brings a set of methods aimed at specific failure modes. Match the method to the story you are actually telling — a joint that wept under thermal movement on a riser is a thermal-cycling argument; a joint that let go where a pipe was pulled is a pull-out argument.

• Internal pressure, to the ISO 1167 series
• Leak tightness of assemblies subjected to bending
• Resistance to pull-out under constant longitudinal force
• Thermal cycling, and separately pressure cycling
• Leak tightness under vacuum, to EN 12294

One detail that quietly dates a claim letter: Amendment 1 to ISO 15874-5, issued in July 2018, replaced the normative pull-out reference EN 712 with ISO 3501. If your correspondence cites EN 712, you are citing the pre-2018 text, and the other side’s technical people will notice. The same freshness problem applies to the base standards — the currently adopted UK editions are BS EN ISO 15874-1:2013+A1:2022 and BS EN ISO 15874-2:2013+A2:2022, both dated 30 April 2024. Citing a bare “ISO 15874-2:2013” is not wrong, but naming the current adopted edition signals you checked.

The Workmanship Side: What a Fusion Argument Is Measured Against

If you are going to say the installer did it wrong, you need a published window to say he fell outside. Socket fusion for PP-R has one, and it is narrower than most site crews behave as if it is.

The window is DVS 2207-11, reproduced here as SIMONA’s 2023 processing manual sets it out. The heated tool runs at 260 ± 10 °C, verified with a fast-indicating surface thermometer or an infrared instrument — not by trusting the dial on the machine, which reads the element and not the face. Before welding, the pipe end takes an external chamfer of roughly 15°, 2 mm wide up to 50 mm diameter and 3 mm above, and an insertion depth mark goes on the pipe before it goes near the iron. The absence of a depth mark on unwelded offcuts at the site is itself an observation worth recording.

The timings are short, and they are per diameter. These figures are for SDR 11 at 20 °C ambient with moderate air movement:

Read the change-over column again. On a 20 mm socket the fitter has four seconds between pulling both parts off the iron and having them fully home. That is the number that gets missed on a busy floor, and it is why a crew that is running fast produces cold joints without ever touching the temperature dial.

The ambient qualifier matters as much as the times. Those figures assume 20 °C and moderate air movement. A stairwell in February is not that, and a crew using summer timings in winter conditions produces under-fused joints while believing it followed the table. If the failures cluster by floor, by season, or by which crew was working, that pattern is worth more than any single cut face.

The neutral reference below shows the PP-R electrofusion process end to end — useful for seeing what a controlled joint looks like when the same system is joined a different way, and for understanding why surface preparation is the step that generates most of the argument.

A Worked Claim File: Turning a Failed Joint Into an Evidenced Position

Here is the sequence, in the order the evidence decays. The first two steps are the ones that get skipped, and skipping them is usually fatal.

1. Stop the disposal, today. Send one message to the site instructing that no further failed joints are cut out and discarded, and that everything already removed is bagged and kept. This is the most valuable thing you will do all week and it costs nothing. The person cutting out the evidence is very often the person whose work is in question — not usually malice, just a crew trying to fix a leak — but the effect is the same.
2. Secure retained stock with batch marking intact. Find unused pipe and fittings from the same delivery, with the print line legible, and photograph the marking before anything is cut. Compliant PP-R carries its class and pressure coding on the pipe wall, and without that marking a lab result cannot be tied to your delivery. If every metre of that batch is installed, say so now rather than discovering it in week three.
3. Record the process facts while people still remember them. Which crew, which machine, what iron temperature was set and whether anyone checked the face, what the ambient was, whether depth marks were used, and whether the system was pressure-tested on completion and at what figure.
4. Section a joint properly, and more than one. Cut longitudinally through the socket axis so the full fused length is visible, photograph against a scale, and do it to at least three joints including one that has not failed from the same run. The unfailed joint is the control, and its absence is why so many teardowns prove nothing: without it you cannot tell whether the defect is characteristic of the installation or a one-off.
5. Run the arithmetic before you commission anything. A specific, checkable instruction beats “please test these pipes” — worked through below.
6. Read the result honestly. Including the outcome where nobody you wanted to blame is at fault.

Step 3 deserves a note. Our PP-R pressure test procedure sets out what a defensible commissioning record contains; if that record exists for this system, it is often the fastest route to an answer, because a system that passed a correct test at handover and failed four months later has a much shorter list of possible causes.

Step 5 is where the standard earns its keep. Suppose the failed riser is a class 1 hot-water system with a declared design pressure of 10 bar. From ISO 15874-2:2013 Table A.2, σ_DP for PP-R at class 1 is 3,02 MPa. To test the joint assembly at the 95 °C / 165 h point, σ_P is 3,8 MPa from Table 10. So p_J = 10 × (3,8 / 3,02) = 12,6 bar, held at 95 °C for 165 h, and the assemblies shall not leak. Change the design pressure to 6 bar and the same arithmetic gives 7,5 bar. That is a number you can put in a letter or hand to a laboratory.

Step 6 has three possible readings, and the third is the one people forget. If retained stock passes the acceptance points and the joint fails a Part 5 assembly test, you have an installation claim and your supplier is not the problem. If retained stock fails an acceptance point, you have a material claim, and the failed joint has become almost irrelevant to it. If retained stock passes and the joint assembly also passes, the failure was neither — look at design, restraint, or service conditions outside the declared class.

Two Methods That Will Not Help You, and One Correction

Some evidence looks authoritative and does not survive contact with a competent opponent. It is worth knowing which, because commissioning the wrong test costs both money and credibility.

The correction first, because it is circulating. ISO 18373 is sometimes cited as a way to measure degree of fusion or processing quality in plastic pipe. It is a differential scanning calorimetry method for rigid PVC — Part 1 measures processing temperature, Part 2 relates to degree of gelation, and both are written for PVC-U. Neither is a PP-R fusion-quality method, and asking a laboratory to run it on polypropylene socket joints will produce either a refusal or a number that means nothing in your dispute. If a consultant proposes it, that is a useful signal about the consultant.

An uncontrolled site re-weld proves nothing. A common suggestion is to have the crew make a few fresh joints and test those. Unless the re-weld is made under recorded, verified parameters by a competent welder, all it demonstrates is that a joint can be made — which nobody disputed. Worse, a successful re-weld will be used against you as evidence the material is fine.

Percentage statistics without a study behind them. Figures of the form “X % of joint failures are caused by Y” circulate widely on supplier blogs with no traceable primary research. Putting one in a claim letter invites the other side to ask for the source, and there usually is not one. Cite the acceptance points instead; they are published, dated and checkable.

One more, less obvious: a hydrostatic test result on the failed pipe section. Material that has been through a heat-fusion cycle and months of service is not representative of the batch as delivered, so a poor result on it does not prove a manufacturing defect. That is precisely why retained stock, not the failure, carries the material claim.

What to Fix on the Supply Side: What We Check, and Where We Stop

Most of what makes these disputes unwinnable is decided months earlier, at purchase. Three habits change the odds, and none of them requires a laboratory.

Ask for the batch paperwork at delivery, not at failure. The document set that decides a claim is the one tied to your specific lot, and it is far easier to obtain when nobody is arguing.

The test of a supplier here is not whether he says he runs QC. Every manufacturer says that, and none of it is admissible. It is whether he will put a method reference next to each check, so you can commission the same test independently and compare. Ask for the lot sampling plan by its standard number, the dimensional check by its tolerance basis, the pressure regime by its test method. A supplier who answers with adjectives instead of standard numbers has told you what his QC report will be worth in a dispute.

What we publish, and where we stop

Our own quality control page names the method behind each gate — ISO 2859-1 for per-lot AQL sampling, DIN 4726 for the oxygen-diffusion limit on barrier pipe. It also states, on every one of those gates, that the per-SKU measured bands, the AQL level and acceptance numbers, and the lot test values go out in the technical data sheet and QC report on request, not on the website.

In an article about evidence we are not going to pretend otherwise: Hitze publishes the methods and releases the measured values against a specific enquiry and lot. Published methods are the part you can act on today. The lot values are the part to get in writing before the container ships, not after a riser weeps.

The third-party position is separate, and worth naming precisely rather than as a logo wall. SKZ testing in Germany covers PP-R pressure pipe and PP-R fittings; DVGW type examination covers PP-R drinking-water pipe; an NSF laboratory BS 6920 report covers hot contact up to 65 °C and cold drinking-water contact.

Note what that list does not say. It does not say every credential covers every SKU, and no certificate number appears on this page, because a number printed without the covered model and its validity window is worth nothing to you. Ask which certificate covers the exact model you bought, ask for the reference, check it at the issuing body. Apply that test to us as readily as to the supplier you are arguing with.

Keep a retention set from every delivery. Half a metre of pipe and a handful of fittings per batch, with the print line intact, stored with the packing list. It is close to free and it is the single item whose absence most often ends a material claim before it starts. Our PP-R sample approval checklist covers what to check on arrival, and the wider question of what you are actually buying is covered in virgin versus recycled PP-R — regrind content is the material story behind a surprising share of these failures.

Be explicit about class and pressure on the purchase order. A PP-R pipe carries a separate verdict for each application class, so “PN20” on an order line is not a specification. Name the class and the design pressure, and check the print line against them when the container lands. Sizing detail sits in our PP-R sizes, SDR and PN guide, and the current PP-R pipe and fittings range runs OD 20–110 mm.

What to require at delivery, in the order of how often its absence sinks a claim:

• A retained sample per batch — half a metre of pipe plus a few fittings, print line intact, stored with the packing list.
• Legible class and pressure marking on the pipe wall, photographed on arrival.
• The lot’s technical data sheet and QC report, tied to your delivery rather than to the catalogue.
• Certificate scope confirmed against the exact model you bought, with the reference so you can verify it independently.
• The class and design pressure written on the purchase order, not just a PN label.

One warning about warranties, because this is where they get waved around. A “50-year warranty” — ours included — is a commercial undertaking whose inclusions, exclusions and claim procedure live in the written terms. It is not the same object as the 50-year design life in the standards’ stress calculations. The two share a number and nothing else.

ISO 15874-2 Table A.2 derives its class 1 design stress of 3,02 MPa from a 50-year extrapolation of material performance. That is an engineering basis for sizing, not a promise by anybody to pay for a re-pipe in year 34. Before you lean on a warranty in a dispute, check whether it survives an installation the manufacturer did not supervise — in a fusion joint failure, that is the exact question at issue.

On price structure, the honest answer is that no per-metre figure is published, and you should be wary of anyone in this category who publishes one: PP-R pricing moves with polypropylene resin cost and is quoted per size and wall series against your mix, not as a single catalogue rate. What is not published, and must come from a written quote against your size mix and destination, is that price, trial-order terms, production lead time and any customisation-specific minimum.

Distributor-level programme terms are set out on the importer and distributor page, and the certificate scope by product line on the certifications and compliance page.

Where This Leaves You on Tuesday

Send the message stopping disposal. Bag what is left. Find unused stock from the batch and photograph the print line. Write down who welded, at what setting, in what weather. Then decide which of the two claims you are making, and commission only the test that supports it — with the pressure calculated, the temperature named and the duration stated.

That is a slower first week than cutting a joint in half and sending a photograph to your supplier. It is also the difference between a position that survives a technical review and an argument about who sounds more confident.

Frequently Asked Questions