Plastics welds - testing techniques and the effect of defects

Maria Girardi joined the Plastics Joining Department in September 1989, after completing an MPhil at Bath University on matrix modification of glass fibre reinforced composites. She obtained her first degree, in Materials Technology, at Coventry (Lanchester) Polytechnic. Since she came to TWI she has been widely involved in welding gas and water pipes, ultrasonic welding including development of a magnetostrictive transducer, and adhesive bonding.

Recently, she transferred to the Adhesives Section in the Engineering Department to help in the expanding adhesives activities at TWI.

She has also been appointed a technical co-ordinator of the medical industry team to help promote new activities in this area.

Plastics pipes are now in common use and butt fusion welding parameters for pipe sizes up to 250mm diameter are well established, whilst for larger sizes the parameters are based on broad extrapolation from the smaller sizes. But the significance of defects in the welds has not been defined. Maria Girardi reports on a study which has looked at the influence of defects on the mechanical properties of butt fusion welds in medium density polyethylene.

Plastics pipes are used increasingly for gas and water distribution, sewage and effluent handling and in industrial process plant. Of the joining techniques used, the simplest is butt fusion, and it is the only process that can be used to join large diameters.

Butt fusion welding parameters for PE80 grade pipe in sizes up to 250mm diameter are well established. For larger sizes, however, there is less practical experience. The correct process parameters are less well defined and are based on broad extrapolation from smaller sizes. A further concern is that the significance of defects in welds in plastics pipes has not been defined and that current non-destructive techniques for detecting defects are inadequate.

This is of particular importance as plastics pipes are beginning to be used in more critical applications or where the cost of a failure during installation or use would be high. In addition, care is needed with fabricated fittings, such as mitred bends. Stresses on these welds can be much higher than in straight butt joints, so poor quality joints will be more likely to fail.

Currently, the only way to detect large defects or contamination is by removal of the bead and small defects are very difficult to detect. This article is about a study of the influence of defects on the mechanical properties of butt fusion welds in medium density polyethylene (MDPE).

Testing

The significance of defects can only be assessed in the laboratory by mechanical testing of joints. To find the most suitable test method sensitive to defects such as those which may occur in practice, a number of test geometries were assessed. The influence of defect size was then evaluated using the established test geometry.

Most of the experimental work was conducted on MDPE blue grade Rigidex (PEX) pipe material of various wall thicknesses. A small number of welds were made on MDPE yellow grade pipe to compare behaviour between the two grades.

All welds in up to 250mm diameter pipe were made using a manually controlled hydraulically powered machine supplied by Haxey Engineering. Welds made in 400mm pipe were carried out using a large hot-plate machine, designed and built in-house. The control system was microprocessor-based so that there was close control of all welding parameters. All welds, except cold welds, were made at industrially recommended conditions for each pipe size ( Table 1).

Table 1 Welding conditions for a range of pipe diameters

Initially, butt fusion welds with known defects, chalk dust and cold welds (40sec plate removal time), were made using 125mm diameter pipe. Tensile samples were cut from each weld using a British Gas standard ^[1] testpiece geometry. In addition, alternative specimen geometries were cut from 125mm diameter pipes: a parallel-sided sample, a waisted dumb-bell-shaped testpiece, a dumb-bell-shaped testpiece used by Neste Chemicals ( Fig.1a) for tensile stress testing ^[2] and a modified British Gas Standard test geometry with PTFE ( Fig.1b).

In the latter case, the PTFE insert was used as a starter crack to initiate failure at the interface. Tensile testing was carried out at room temperature (20°C) using an Avery Denison universal testing machine at a crosshead displacement rate of 5 mm/min and the fracture surfaces were examined.

Impact tests using single-edge notched bend specimens were conducted on butt fusion welds in 125 and 400mm SDR11 pipe with differing welding parameters; a standard weld and welds produced with long plate removal times (40sec with the 125mm and 90sec with the 400mm diameter pipe). Alternative testpieces were also extracted from the 125mm diameter pipe, with inserted PTFE tape to act as a notch from which a crack can initiate. This was compared with the established test method of introducing a notch by machining. The tests were conducted on a Rosand instrumented falling weight impact tester (IFWIT).

For further comparison, a number of the welds with known defects were debeaded externally and stress crack resistance tested. ^[1] The samples were pressurised through stainless steel test ends at 9.2bar and 80°C for 1000hr. This pressure was maintained by a constant 'top-up' system using pressure reducing valves.

To assess the effect of defects, a number of welds with a range of aluminium foil discs (1-8mm diameter) implanted into the joint interface at the mid-point of the wall thickness were made in pipe with a range of wall thicknesses (approximately 9-40mm, external diameter 90-400mm). Test specimens using the British Gas test geometry were cut from the pipe and tensile tested, each testpiece being cut so as to contain one intentional defect at the centre of the cross section. Test conditions were as above and again the fracture surfaces were examined.

Results

Evaluation of test geometries

Tensile tests

Weld specimens with a PTFE insert, together with chalk dust and the 40sec cold weld, all failed by a planar interface fracture. The 40sec cold weld revealed some stress whitening of the fracture surface, which seemed to indicate that fusion had begun to take place. Welds with PTFE insert and no other known defect gave an initial slow crack growth before fast crack propagation.

Impact tests

Table 2 summarises these results. In 125mm diameter pipe, locating the crack tip at the interface with mechanical notching proved to be difficult and hence parent material values were obtained in the tests. However, the PTFE notched samples appeared to discriminate between welds of different quality and showed that welds made at standard conditions had a slightly lower fracture toughness (mean value of 2.15MPa √ m) than parent material (2.4MPa √m) and that the cold welds had an even lower fracture toughness (mean value of 1.59MPa √m) indicative of poor fusion.

Table 2 Dynamic fracture toughness results for butt fusion welds in 125 and 400mm diameter pipe.

In the 400mm diameter joints the wider weld zone made positioning of a machined notch within the weld zone (but not necessarily at the interface) possible. With the cold weld the fracture path deviated to the weaker interface, and gave a lower fracture toughness value (mean value of 2.13MPa √m), whereas with the weld made at standard conditions the fracture path did not deviate and parent material values were obtained.

Stress crack resistance tests

None of the butt fusion welds with known defects which were subjected to the stress crack resistance test at 80°C for 1000hr failed.

Defects introduced during welding

A progression from ductile to brittle behaviour as defect size increased was observed for different wall thicknesses. Changes to the strength values obtained were insignificant although a reduction in elongation occurred. As a guideline, a weld was considered to be acceptable only if the fracture surface exhibited greater than 50% ductile tearing. No difference in fracture behaviour was observed between yellow and blue grade pipe.

Effect of wall thickness

Figure 2 summarises the fracture behaviour in tensile tests of butt fusion welds in PEX of different wall thickness, showing the influence of defects of different sizes. Small defects were found to become more critical as the wall thickness increased. This could be an effect of plane strain conditions because the test sample thickness was the same as the pipe wall thickness, so further investigations were carried out.

To assess the effect of different specimen thickness, a 12mm thick tensile testpiece, with a 2mm aluminium foil defect, was machined from the centre of a 40mm wall thickness pipe and compared with a tensile specimen from a 12mm wall pipe with the same defect size.

The fracture behaviour exhibited by a 12mm wall specimen was ductile, whereas the 12mm thickness specimen ( Fig.3) machined from a 40mm wall pipe was similar to that of a 40mm thickness specimen; a relatively flat fracture with slow crack growth around the defect.

In addition, finite element analysis was conducted to investigate the effect of defect size with varying wall thickness, by calculating the stress intensity factor for each defect. The results showed the stress intensity factor to remain the same ( Table 3) with the same defect size independent of wall thickness over the range evaluated. For example, the stress intensity factor for a 6mm diameter defect was 1.50 and 1.49 N/mm ^3/2 for 8 and 40mm wall thickness specimens respectively. The fracture tests and the results of the FEA show that the variation in fracture behaviour is not caused by geometry effects but presumably, therefore, by a real variation in joint properties.

Table 3 Stress intensity factors calculated for a range of defect size and wall thickness.

Practical significance

Testing butt fusion welds

Reliable, short term mechanical tests are very important for ensuring quality in practice and for research and development. The British Gas standard tensile test was found to be as good as other tensile tests investigated. However, fracture is not necessarily localised to the joint line, so the numerical values obtained must be treated with caution. Similarly, sophisticated tests such as instrumented impact on notched samples need to be used with great caution because of the difficulty in positioning the notch accurately.

Stress rupture test results were not similarly dependent on the presence of defects; all tests gave excellent results. This is perhaps not surprising as the axial stress is half the hoop stress in a pressurised cylinder. In practice, high axial loads at joints can be applied during installation, or during service if the pipe is bent, or in fabricated fittings, so the tensile properties are significant.

For research and development, use of a crack starter such as PTFE tape inserted in the joint seems to give more informative data in both tensile and impact tests. More investigation of this test is necessary.

Defects in butt fusion welds

The results summarised earlier clearly demonstrate that planar defects (such as a small area of contamination or poor adhesion) at the interface of butt fusion welds greatly influence joint properties. However, the size of the defect needed to cause brittle fracture depends on the wall thickness of the pipe. In Rigidex material, for example, wall thicknesses of less than 12mm appear to present little problem because large ~8mm) defects are necessary to cause brittle fracture. Defects of this size are unlikely to occur undetected.

However, as pipe wall thickness increases the critical defect size decreases, so that at wall thicknesses greater than 25mm, defects < 1mm diameter can cause brittle behaviour. These are much harder to detect in practice and, indeed, service experience indicates that brittle fracture only occurs in wall thicknesses greater than about 15mm (although it is, of course, rare).

Finite element analysis and comparison of test samples of a range of thicknesses has led to an intriguing suggestion - that behaviour with different wall thicknesses is not a geometrical effect but represents a real difference in joint properties. This implies that to make joints less sensitive to defects, joint properties should be improved in some way, e.g. by optimising welding conditions. The scope of this is not known since the cause of the poor properties has not been determined, but further investigation is warranted.

In this work only PE80 material was studied, but it seems likely that the influence of defects on the joint properties of thick walled pipes in higher strength materials, such as PE100, will be similar or more severe.

Conclusions

Butt fusion welding trials have shown that:

• In general, the British Gas tensile test is an acceptable 'go/no-go' test. For research and development, use of a PTFE tape inserted at the weld interface as a crack starter was found to be more informative, and should be used for laboratory investigations.
• Changing behaviour with wall thickness was found to result not from a geometrical effect but a real difference in joint properties. This has indicated that there is a need to optimise welding procedures for large diameter pipes and care should be exercised, particularly if high axial stresses are likely. This information is crucial when considering the use of alternative materials, e.g. PE 100, which are beginning to replace PE80 in some applications and it is recommended that PE100 is used with care until more data are available.

References