Fracture tests on welded polyethylene gas pipe

TWI Bulletin, November/December 1988

The susceptibility of welded polyethylene gas pipe to brittle fracture as a result of high strain rates, weld defects and incorrect welding practice has been assessed using Charpy and dropweight testing. Sharp notches and contamination of the welding hotplate are found to be important factors affecting the short term integrity of pipelines.

Polyethylene pipe is widely used for low pressure gas and water distribution systems. Typically pipes are 150mm diameter, 14mm wall thickness. The seamless pipe lengths are joined on site by hotplate butt fusion welds. In this process, the pipe ends are held in clamps which press them on to a plate heated to about 200°C. When sufficient time has elapsed for a semi-molten layer to form at the pipe end (called the heat soak time or bead up time), the pipes are pulled back, the tool is withdrawn manually, and the pipe ends pressed together to form a weld. Molten material is squeezed out around the joint to produce an external and internal bead. Finally, pressure is maintained while the weld cools. A small portable welding machine like that shown in Fig.1 is used on site.

Leaks or complete fractures in these pipe systems could have consequences ranging from purely economic in the case of water supply pipes, to extremely dangerous, in the case of gas pipes. Polyethylene is a tough material under normal conditions, and is thought to be tolerant to variations in welding procedure, but the parent material is known to be sensitive to the presence of sharp notches and high strain rates, both of which can cause brittle fractures. The most likely causes of sharp notches are mechanical damage and weld defects.

The short investigation described here had three main objectives:

1. To assess whether high strain rates could cause fracture at welds in polyethylene gas pipe;
2. To examine the effect of artificially induced defects on weld fracture resistance;
3. To determine whether incorrect welding practice could contribute to the occurrence of brittle fracture.

Experimental details

Material

The pipes used were gas pipe grade 90 SDR11 (90mm OD, 9mm wall thickness) from both the sources of supply available in the UK, namely BP Rigidex and Dupont Aldyl 'A'.

Welding conditions

The welding machine shown in Fig.1 was used for this work and the welding conditions are summarised in Table 1. The optimum conditions for both pipes involve a hot plate temperature of 205°C, but slightly different heating times and forces are needed for the two materials. The main procedural abuses investigated were a low hotplate temperature, an incorrect heat soak time and contamination of the hotplate, (in this case, deliberate contamination with engineer's chalk).

Table 1 Welding conditions for pipe samples

*Clamping time 5min for all samples

Testing

In the preliminary tests, the welds were cut into cross weld strips approximately the size of Charpy specimens (10mm cross section, 50-60mm long), some with the bead on, some with the bead removed (as is current practice in parts of the UK), and some with sharp (razor cut) notches.

These specimens were tested in a small Zwick impact machine with a striking speed of approximately 4 m/sec and in The Welding Institute's drop weight tower, with a striking speed of 13.4 m/sec (30 mph).

Similar specimens from a previous investigation [1] on 180mm OD 10.5mm wall thickness pipes were also tested on a larger Charpy machine with a striking speed of about 5 m/sec. One set of specimens from the 90mm OD 9mm wall thickness welds was also tested on this machine to give a cross reference.

All of these small specimens were struck on the side corresponding to the pipe's inside surface, thus putting the pipe's outer surface in tension.

The final set of tests involved a set of welds made at optimum and non-optimum conditions which were tested at full size in the drop weight tower. These were to be tested with and without artificially introduced notches. Preliminary tests in a Charpy machine on 10x10x55mm strips of both parent materials showed that a saw cut notch approximately 2mm deep had no effect on specimen behaviour. Saw notched and unnotched specimens bent on impact. In both materials however, specimens with scalpel cut notches fractured with no visible deformation. It was decided that scalpel cut notches would be used in the full size tests.

The full size tests were carried out in the following manner. The specimens consisted of an 800mm length of pipe with a weld at the midpoint. The specimen was placed in a three point bend jig under the dropweight tower. The lower anvils of the jig were 400mm apart and 200mm high, so that the pipe would be bent to an angle of 90° by the impact. The upper anvil was attached to the falling carriage of the dropweight tower, which weighed about 24kg. This weight ensured that the pipe experienced deformation to 90° at an approximately constant deflection rate. Little deceleration of the carriage occurred during bending, and the upper anvil always completely flattened the circular cross section of the weld against the base of the jig. A pipe specimen is shown in Fig.2, resting in the lower anvils of this test rig.

For consistency, all full size specimens were tested with the bead removed. This is understood to be normal industrial practice. Some of the specimens were notched for 180° around the tension half of the weld circumference, using a scalpel to produce a notch about 1-2mm deep on the weld line.

Results and discussion

The test results are summarised in Table 2 which indicates whether the specimens bent or fractured upon impact. Specimens which fractured did so with little prior deformation. Those which did not fracture simply bent, as shown in Fig.3. The absorbed energy was recorded when Charpy machines were used for testing, but is not given in the table as it depends on the specimen cross section, a variable which was not closely controlled in these tests. Energy could not be recorded for tests in the dropweight tower.

Table 2 Impact test behaviour

Small specimens were tested in pairs. Both results are given only of the two showed different behaviour.
NF- not fractured, F- fractured
*Samples taken from welds in larger diameter pipe for previous work.

In fact, Charpy size specimens which fractured usually absorbed only 2 or 3J, whereas those which bent usually absorbed about 20J.

The results of the 10 x 10mm square bar specimens showed a clear pattern of behaviour. At all impact speeds between 4 and 13 m/sec, neither material exhibited brittle behaviour when welded at optimum conditions, nor when the welding conditions were varied from optimum by a considerable margin ( e.g. low hotplate temperature or short weld heating time). This was true whether the weld beads were removed or not.

In both materials, brittle behaviour was only observed when welds were made with a contaminated hotplate, and the weld beads were removed before testing.

The tests made on similar sized specimens from welds made for a previous investigation ^[1] showed slightly different behaviour, as specimens in the Dupont material tended to split even if they were welded at the optimum conditions, whereas those in BP material did not. In addition, specimens in Dupont material welded with a low hot plate temperature and a long heating time, fractured. It is not clear whether this apparent difference between 90 and 180mm diameter pipes arises because of a batch to batch variation in materials, or because of a difference in welding response for the different diameters.

In both materials, welds made at optimum conditions only fractured in the full size tests if the welds were notched. Unnotched welds made with contaminated hotplates showed some splitting in full size tests, but the only sign of a difference in toughness between materials was found when testing welds made with short heating times. With notches present, specimens in both materials fractured, but when unnotched specimens were tested, the weld in the Dupont material cracked, over about 100° beneath the anvil impact point. This difference may be partly explained by the fact that the reduced heating time used for the Dupont weld was short (8sec) compared to that for the BP weld (20sec). This reflects the large difference in recommended heating times for the two materials (Dupont 20sec, BP 90sec). The brittle appearance of the fractures induced by sharp notches is shown in Fig.4.

Summary

In general, the small scale tests showed that there is no effect of strain rate on fracture behaviour up to the strain rate generated by striking a Charpy sized specimen at 13.4 m/sec. Large scale tests showed that strain rates up to that caused by three point bending of a pipe at 13.4 m/sec did not cause fracture, if pipes were welded at the recommended conditions.

Large and small scale tests showed that sharp notches have a serious effect on fracture performance of polyethylene parent material and welds over the whole range of strain rates tested, whatever the welding conditions.

The only abuse of welding procedure which caused a large deterioration in fracture performance was the use of a contaminated hotplate. This appears to be capable of causing complete brittle fracture of the weld under impact loading conditions.

There was some evidence that the Dupont material was more sensitive to changes in welding condition than the BP material, giving rise to cracking in full scale impact testing.

From a practical point of view, the results of this work imply that the most important factors with regard to short term integrity of pipe butt welds are the presence of sharp notches, and contamination of the welding hotplate. Work on the fracture behaviour of plastics welding is in progress at The Welding Institute, and member companies with an interest in this area are invited to contact the author at Abington.

Reference