The Fatigue Performance of Electrofusion Tapping Tees Subject To Contamination - Final Version
The Fatigue Performance of Electrofusion Tapping Tees Subject To Contamination - Final Version
The Fatigue Performance of Electrofusion Tapping Tees Subject To Contamination - Final Version
to contamination.
Article:
Tayefi, P., Beck, S.B.M. and Tomlinson, R.A. orcid.org/0000-0003-2786-2439 (2019) The
fatigue performance of electrofusion tapping tees subject to contamination. International
Journal of Pressure Vessels and Piping, 171. pp. 271-277. ISSN 0308-0161
https://doi.org/10.1016/j.ijpvp.2019.03.003
Reuse
This article is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs
(CC BY-NC-ND) licence. This licence only allows you to download this work and share it with others as long
as you credit the authors, but you can’t change the article in any way or use it commercially. More
information and the full terms of the licence here: https://creativecommons.org/licenses/
Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by
emailing eprints@whiterose.ac.uk including the URL of the record and the reason for the withdrawal request.
eprints@whiterose.ac.uk
https://eprints.whiterose.ac.uk/
THE FATIGUE PERFORMANCE OF ELECTROFUSION TAPPING TEES
SUBJECT TO CONTAMINATION
1
THE FATIGUE PERFORMANCE OF ELECTROFUSION TAPPING TEES
SUBJECT TO CONTAMINATION
Abstract
Electrofusion jointing is a common joint process used to weld polyethylene (PE) water pipe
in the UK. For many years, the UK Water Industry has experienced a small number of
premature failures of electrofusion fittings; this inevitably causes leakage. The common
causes of critical failures have been highlighted and as a result of this, a testing programme
was designed and implemented to assess the fatigue performance of electrofusion tapping
tees if they were to be installed by ‘bad practice’. A fine china talc was used to replicate
contamination in the field. The results suggest that failures associated with fatigue are
possible in a relatively short space of time on tapping tees when they are subject to
contamination. The pressure ranges used in the fatigue regime aimed to replicate the potential
2
1 Introduction
The use of plastic pipes within the UK Water Industry dates back to the 1950s. Initially
polyvinyl chloride (PVC) was predominant, but this has gradually been superseded by
polyethylene (PE) which has become a standard material for pressure pipe systems in the
water industry. Current research shows that PE will exceed its expected design life of 50
Buttfusion and electrofusion jointing are the common welding methods used to join PE pipes.
Buttfusion involves the use of a hot-plate to melt the ends of the pipe; the heated pipes are
then forced together to form the joint. Electrofusion jointing uses electricity to heat a filament
wire that is manufactured into the carcass of the fitting; this melts the pipe and fitting locally
and as they cool the bond is formed. Many PE joints (especially electrofusion joints) are
potentially installed in difficult working environments; for example trenches that are open to
the elements. In order to ensure joint integrity, best practice procedures must be followed.
preference, electrofusion welding is somewhat a more ‘manual’ jointing procedure. Here the
operative needs to follow specific preparation procedures that are highlighted in Water
Industry Specifications (WIS) and manufacturers’ guides to best practice. Special care needs
to be taken to ensure that the fusion zone (jointing area) is dry and clean of any contaminants
such as dirt, dust and water. Any contaminants present in the jointing interface could
compromise the integrity of the asset. Research has shown that failures can occur in PE water
distribution networks, specifically at the joint [2]. An independent study of the National
Sewers and Water Mains Failure Database analysed data between 2005 - 2009 and concludes
that there is a failure rate of 8 failures per 100 km per year [3]. At the time of this study, the
database recorded a total length of 65,279 km of PE pipe in the UK. Contrarily, analysis of
1
the same database [2] revealed PE to be the best performing material with regards to failure
rates.
A surge can be defined as a sudden change in pressure [4]. The effects of surge on PE water
distribution networks have been well researched and documented since the late 1990’s [5].
Fatigue can be described as the repetition of such events that can cause a reduction in the
long-term strength of material. Bowman [6] explains that pipe systems may be subject to two
different types of ‘fatigue’: firstly, a diurnal fatigue by which the demand on the network
causes fluctuations in pressure (≤4 bar), ranging between 3.7 x 104 cycles and 9.0 x 104
cycles in a 50 year design life; secondly, the operation of pumps and valves in the distribution
network (>6 bar). It is important to note that during either event, the material will experience
information available on analysis techniques and methods [7]. Hydraulic (pressure) fatigue
tests can be long in duration giving the possibility of long test times required for worthy data
generation. However, it is arguable that pressure testing relates to the long term performance
Electrofusion failures within the UK water industry has been well documented with the three
main causes being: poor scraping, misalignment (including problems associated with ovality)
and contamination [2, 8, 9, 10]. Note that these issues are all associated with on-site practice
(workmanship). Although electrofusion failures are well researched, the problem is far more
Two of the three aforementioned workmanship issues can arguably be overcome by the
an environmental issue that may be harder to overcome. Previous research has shown that
2
applying a talc contaminant to the jointing interface prior to welding has a detrimental
influence in joint performance [11]. It was on this basis that a short term test was developed
to ensure that all electrofusion fittings in the UK carried a certain resistance to contamination
[12]. One test that is incorporated into the WIS known as the ‘resistance to contamination:
short term burst test’ [13, 14] for electrofusion joints encompassing both electrofusion
coupler and saddle (tapping tee) fittings. Here, a talc contaminant is applied to the pipe prior
to the assembly of the fitting. Particle size and distribution with regards to contamination is
extremely important as some particulate contaminants have a greater influence on the fracture
toughness of an electrofusion joint [15, 16]. Furthermore, the WIS test specifies fine china
Specimens prepared to the aforementioned test are attached to a hydraulic rig capable of
increasing the internal pressure at a constant rate of 5 bar/min until failure. For electrofusion
couplers, failure should not be below the nominal pressure (PN) rating of the pipe multiplied
by 2.5 (PN x 2.5); for electrofusion saddle fittings a minimum pressure of 18 bar must be
achieved. A successful joint would indicate that it has some resistance to contamination,
however this does not mean that the joint will not fail in service if best practice principles are
not followed on site. Furthermore, this short term test does not relate to the long term
tapping tees subject to a particulate contaminant; using the WIS short term burst test as the
2 Experimental Procedure
The subjects of the experiment were PE pipe and fittings from the same UK manufacturer.
Pipes were 110 mm diameter with Standard Dimension Ratio (SDR) 17 of grade PE100 with
3
an average density of 950.1 kg/m3. The tapping tees used were injection moulded PE100
grade fittings, although the same product was used consistently through the test programme,
Electrofusion tapping tees are commonly used to connect the water distribution main to the
end user (customer). There are many different designs of tapping tee but the principle of
welding the fittings are relatively similar. A compressive force is applied between the
fitting’s fusion interface and the host pipe. This force is usually achieved by a top-loading G-
clamp or an under-clamping strap and should also ensure that the fitting does not move
The test specimens were prepared in accordance to the WIS short term burst test [13]. The
specification recommends that the contaminant is uniformly applied to the pipe using a
evenly and conservatively brushed onto the pipe using a 25 mm wide soft bristle paint brush.
This method is acceptable for assemblies which fuse with limited pad area [7]. To add an
element of quality control, all specimens were created by the same operative.
With regards to the application of hydraulic pressure, the inlet was attached to the stem of
tapping tee as opposed to the service outlet end recommended by the WIS. This was to
electrofusion end cap was welded to the service pipe outlet leaving less than one diameter of
Sections of the 110 mm diameter pipe were cut perpendicularly to approximately 320 mm in
length allowing approximately one diameter of pipe either side of the fitting. The pipe
product used had a protective skin surrounding the host pipe; this was removed to expose the
pipe’s core. With ‘skinned’ products, no extra preparation is required (i.e. pipe scraping) once
4
the skin is removed. All specimens were welded at an ambient temperature 21 (±1.5) °C –
once welded the specimens were left for a minimum of 24 hours prior to testing. The testing
media used was tap water; left uncovered for a minimum of 12 hours to aid the removal of
air. The specimens were filled with water prior to assembling to the fatigue test machine to
servo-hydraulic fatigue testing machine. The rig was designed to house two tests: (i) the WIS
short term burst test and (ii) a dynamic load (fatigue) test. A calibrated pressure transducer
was used as the primary feedback in the closed-loop control system. It is important to note
that a servo-hydraulic control unit is traditionally used to control the main actuator (load
cell/position control), however, for the purpose of this testing programme both software and
hardware alterations of the control unit needed to take place in order for the pressure
transducer to govern the test. Furthermore, the load cell was not made redundant but simply
switched to be the secondary feedback loop; the pressure transducer being the primary.
Figure 2.1 illustrates a schematic of how the experimental rig and the tapping tees were
5
Figure 2.1 - Schematic showing all components
The aim of this test was to establish an average failure pressure for electrofusion tapping tees
welded to the specification of the ‘talc test’; this was deemed the maximum pressure of the
jointing system ( , ). Once established, this value was used as a benchmark for the
dynamic loading long term test. It is important to note that the ramp rate for both the short
term burst test and the dynamic load test were to be the same, as the aim of the dynamic test
was to establish a link between pressure range variation with respect to fatigue life of
electrofusion tapping tees. The WIS [13] recommends a ramp rate of 5 bar/min but this was
deemed too slow for reasonable test durations in the fatigue tests. Therefore, the ramp rate
was increased from 5 bar/min to 25 bar/min. The short term burst test was carried out four
times per ramp rate. The results from this are shown in Table 2.1. The results show that there
were little difference in the average and median failure pressures of approximately 25 bar
6
third party test house comparing results with those observed with the experimental rig to
A trapezoidal loading pattern was followed for the basis of the fatigue test and the results
plotted as pressure vs. number cycles to failure as per Joseph et al. [17]. The mean pressure
was fixed for all loading patterns and the pressure ranged varied. The ramping rate (pressure
increase/decrease) was fixed at 25 bar/min therefore frequency varied for each loading
pattern. The short term burst test highlighted , of the contaminated tapping tees and
the pressure range of the trapezoidal loading patterns followed percentage decrements from
= %× , (1)
The mean pressure was fixed at 12.5 bar; calculated as half of the average failure pressure
=1 2× , (2)
The resting period at the top and bottom of each cycle was fixed at two-thirds of the ramp
7
=2 3× , (3)
The loading patterns for the fatigue test are shown in Figure 2.2.
Figure 2.2 - Trapezoidal loading patterns for dynamic (fatigue) test showing 1 cycle
; , = 25 , = 12.5 , = %× ,
In addition to the tests on the contaminated tapping tees, two specimens were made to best
practice principles, i.e. with no contamination. These were tested at the highest pressure
best-practice joints.
3 Results
Figure 3.1 shows the results for the fatigue test in a form similar to a stress-life (S-N) curve.
Due to the complex geometry of the tapping tee the ‘stress’ was not calculated as there will
8
inevitably be differential stress concentrations as the geometry changes throughout the fitting.
Therefore, the results are expressed as pressure range vs. number of cycles to failure.
For this testing regime, a failure can be described as the joint not being able to maintain
pressure. This was usually when the delamination of the bonding surface was such to create a
Figure 3.1 - Pressure Range vs. Number of cycles to failure for the dynamic (fatigue) loading
test ( ; = 12.5 )
It is important to note that the fatigue test was stopped if a specimen did not fail after 1000
cycles as this testing programme is classed as low cycle fatigue. A minimum of eight tests
were conducted for each pressure range with the exception of the 90% , pressure
range where three specimens failed to reach one cycle. In contrast, the two tests on “perfect” ,
uncontaminated joints which were conducted at this highest pressure range (circled in Figure
3.1), did not fail after being subject to 1000 cycles. In the 40% , pressure range, five
tests were successful, in that they failed within 1000 cycles, however, four tests did not fail
9
It can be observed that as the pressure range decreases, the distribution of failure increases.
This is becomes evident in the 10 bar range whereby one specimen failed below 50 cycles
and some did not fail after 1000 cycles. It can be argued that this low pressure range pushes
the boundary of the short term fatigue life of the product. Based on the indicative logarithmic
line plotted against the mean number of cycles to failure (Figure 3.1), it can be said that if the
pressure range were to be dropped further, there is a likelihood that most specimens would
successfully reach 1000 cycles without failure. Hence there is an argument that for low
pressure ranges (<10 bar) there may be a fatigue limit in this testing regime.
Using the experimental data in Figure 3.1, a regression analysis was performed and the 95%
confidence limits obtained. Figure 3.2 indicates the average number of cycles to failure with
the 95% confidence limits on pressure range Vs. Log number of cycles to failure. The limits
clearly show two results that lie well below the lower boundary at 15 and 10 bar pressure
10
Figure 3.2 Pressure Range vs. Log Number of cycles to failure
Following the fatigue testing, the leak paths of each joint were observed by using a basic
hand pump to produce a flow of water through the fittings. Each joint was then subject to a
crush decohesion test ( ISO 13955 [18] so that the bonded surface could be observed in order
The crush decohesion test ( ISO 13955 [18] instructs that pipes are inserted into a device that
is able to crush the specimen at a rate of 100 mm/min (1.67 mm/s) until double the wall
thickness of the pipe is achieved. In our case a table vice was used and Figure 4.1 illustrates
the testing apparatus for electrofusion tapping tees; where dp is the distance between the jaws
of the crushing device. If the fitting (the tapping tee) does not yield away from the host pipe
after the crushing exercise, a lever may be used to prise the fitting away from the pipe - no
11
Figure 4.1 Indicative sketch of crushing decohesion test (tapping tees)
With reference to Figure 4.1, as dp decreases during the crush test, the forces present should
induce failure about the jointing interface by producing a crack that propagates as the crush
test is performed.
Figures 4.2 and 4.3 compare the bonded surfaces of a joint that had been subject to the talc
contaminant and a joint that had been made to best practice principles. Figure 4.4 shows
Figure 4.2 shows an aerial view of the pipe and fitting when the crushing device reached its
maximum distance. Once the maximum displacement was reached in the crushing test, the
12
fitting was further removed from the host pipe by using a relatively small amount of force
which was applied by hand, just using the stem of the tapping tee as a lever. As can be seen in
Figure 4.2, there were small black marks on the pipe where it appears a small amount of
bonding had taken place between pipe and fitting. However, indentations on the surface of
the pipe from where the filament wires were embedded into the pipe show that they may have
PE Pipe
Small amount of
bonding Filament wires
Tapping tee
Figure 4.2 Aerial view of contaminated specimen following the crushing decohesion test
Small amounts of ductility are visible about the outer circumference of the fusion zone. Note
that this area was where the filament wires were mostly embedded into the parent pipe. It is
important to note that the failure of the joint when subject to this test should be classed as a
brittle failure. However, the ductility observed is very small and localised about the outer
circumference of the fusion interface; more specifically, either side of the indentations on the
parent pipe created by the filament wires during the welding process.
As a comparison, the crushing decohension test was conducted on a specimen that was
welded to best practice. Figure 4.3 shows the specimen after it has been subject to the full
13
extent of the crushing device. As can be seen, the specimen was just beginning to fail on the
outer part of the fusion zone but still remains fully adhered to the PE pipe when the
maximum crushing distance is reached; suggesting that a ductile failure would occur.
Tapping tee
PE pipe fully
adhered at joint
after crush test
Figure 4.3 ‘Perfect’, uncontaminated specimen following the crushing decohesion test
The assembly was further crushed on the opposing side of the fitting assembly in an attempt
to yield the specimen but with no success. It can also be seen (in Figure 4.3) that there is very
little opportunity to insert a lever to persuade the specimen away from the pipe. ISO 13955
[18] states that no impact forces shall be used to yield the fitting from the pipe, therefore
using only a lever to remove the fitting from the pipe was a next to impossible task. It was
14
therefore concluded that the specimen would fail in a fully ductile manner. This comparison
further enforces the detrimental nature that contamination has on joint integrity.
For illustration purposes, for each pressure range tested, a contaminated specimen was
selected whose number of cycles to failure was the closest to the average number of cycles to
failure. Figure 4.4 figure shows pipe surfaces following the crush decohesion test and also
indicates the approximate leak paths. The leaks were classed as major and minor leaks, where
a major leak can be described as the primary flow of water; whereas a minor leak would be
(a) 90% PMAT, MAX Pressure range (b) 80% PMAT, MAX Pressure range
(c) 70% PMAT, MAX Pressure range (d) 60% PMAT, MAX Pressure range
15
(e) 50% PMAT, MAX Pressure range (f) 40% PMAT, MAX Pressure range
Figure 4.4 Pipe joint surfaces following fatigue test to failure, leak path test and crush decohesion
test. The solid lines indicate major leak paths and dashed lines indicate minor leak paths
Post-failure analysis of the contaminated joints revealed very little bonding between the
electrofusion tapping tee and the host pipe, therefore it can be assumed that all specimens
failed in a brittle manner. In all of the samples tested, the effects that contamination has on
the adhesion of the fitting to the parent pipe are clear, in that, when compared to the “perfect
joint” (Figure 4.3), the fittings were easily fully removed from the pipes using the crush
decohesion test (Figure 4.4). The most logical failure mechanism for brittle failures would be
crack propagation [19] of the jointing surface. However this was proved not to be the case.
As can be seen in Figure 4.2, sections of the electrofusion filament wire were embedded into
the host pipes. This suggests that wire itself may have been offering some degree of structural
integrity if little or no polymer bonding were present. Bonding appears to be the best at the
outer circumference of the fusion zone (See Figures 4.2 and 4.4). Here, only a very small
amount of ductility can be seen and it is believed that this bonding is located either side of the
indentations caused by the filament wires during the welding process. Furthermore, failure
(leak) paths appeared to be random which suggests that the product itself has no obvious
weak points. Minor leak paths seem to become more predominant as the pressure range is
decreased in the fatigue testing regime. For example, the specimen in Figure 4.4(a) (90%
16
PMAT, MAX) has one clear major leak path, whereas in Figure 4.4(f) (40% PMAT, MAX) the
specimen has one major leak path and two minor leak paths. This may suggest that the lower
pressure ranges promote crack growth under dynamic loading; potentially in a circular
manner.
With regards to the two anomaly results that sit well below the 95% confidence limits
(circled in Figure 3.2), there were no obvious reason why these failed prematurely. It can
therefore be assumed that there may have been a variation in the specimen preparation
procedure.
For typical PE pipe products, prior to assembly of an electrofusion fitting, the oxidised layer
on the outside of the PE pipe needs to be removed. This is usually achieved by the use of a
mechanical or hand scraper. Mechanical scrapers are generally preferred where possible as
they leave a more uniform surface for jointing and scraping tends to be more consistent from
operator to operator. As mentioned above, the PE pipe product used for this experiment had a
protective polypropylene skin that needed to be removed prior to assembly of the fitting and
welding. Once removed it revealed the PE pipe with a smooth extruded external surface of
which no additional scraping preparation was required. Therefore, brushing the talc
contamination on the pipe surface is arguably more uniform than if it were to be brushed on
For the fatigue testing programme, a fixed mean pressure of 12.5 bar was used. This value
can be compared to the service pressure of a distribution main and thereby the cycles relate to
the increase/decrease in pressure that may be expected in a water distribution network. The
mean pressure (12.5 bar) would arguably be too ‘high’ for distribution mains of 110 mm
diameter; also, the pipe used was rated with a minimum required strength (MRS) of 10 bar
(SDR 17). It was in the remit of the fatigue experiment to accelerate the time to failure. It is
17
not uncommon in fatigue tests to use an 80°C bath to house specimens to decrease the time to
failure [6]. However, in this experiment it was decided to test at ambient temperature and use
an increased pressure as the failure accelerant. Furthermore, this testing programme has
shown that fatigue failures can occur on contaminated electrofusion joints in relatively short
spaces of time but at higher than expected ‘operating pressures’. From the outcomes of this
experiment, a secondary testing programme has been developed to relate the findings from
this experiment to typical pressures in distribution mains with the aim of predicting the asset
life of a contaminated joint. This will be accomplished by using a fixed pressure range,
As previously mentioned, there was the possibility of a fatigue limit in the current testing
programme as some specimens did not fail after 1000 cycles. Typical fatigue tests on PE pipe
usually indicate a high number of cycles until failure (>105 cycles); this testing programme
has illustrated low cycle fatigue failures at raised pressures using a talc contaminant to
replicate poor workmanship that may be experienced in the field. However, for lower
pressure ranges, it would appear that the prediction of failure becomes more difficult. When
fatigue testing, there will always be an element of scatter in testing results. Bowman [20]
explains with regards to fatigue testing of electrofusion fittings, the fatigue life can vary by a
factor of ten. Furthermore, different grades and batches of PE can give different fatigue
strengths. This may explain the large amount of scatter in the results from this experiment as
the tapping tee fittings, although a single product were tested , were purchased off-the-shelf;
Previous research [11] has shown that the talc contaminant is a ‘worst case scenario’ and
therefore the short term burst test depicted in the WIS is arguably a subjective test in relating
poor workmanship to the long term performance of an electrofusion joint. However, the test
can quickly assess the performance of different electrofusion products compared to long term
18
hydrostatic tests that can prove to be costly. This dynamic loading test programme aimed to
make the short term burst test more realistic and to give an indication of how electrofusion
tapping tees would perform at various pressure ranges under cyclic load. As water
distribution systems do not operate at a constant pressure and fluctuate due to various reasons
such as the start/stop of pumping stations [21]; it can be argued from the testing programme
undertaken that fatigue failures could happen if the asset is installed incorrectly.
To conclude, the experimental data shows that joint failures associated with fatigue are
possible on talc contaminated joints under the proposed testing parameters. Although PE pipe
has good short term resistance to increases in pressure (i.e. surge) in a nominal working
scenario, this research suggests that if a product were to be instsalled incorrectly, the asset
Acknowledgments
The authors would like to thank the Engineering and Physical Science Research Council
(EPSRC), Severn Trent Water and WRc plc for their sponsorship and support of this project.
19
Nomenclature
References
[1] E. Hoang and D. Lowe, “Lifetime predictions of a Blue PE100 Water Pipe,” Polymer
[2] UKWIR, “Leakage from PE pipe systems,” UK Water Industry Research Limited,
London, 2011.
[3] UKWIR, “National Sewers and Water Mains Failures Database. Case Study No. 1:
Polyethylene Pipe Failures on Water Mains,” UK Water Industry Research Ltd, London,
2011.
[4] E. Wylie, V. L. Streeter and L. Suo, Fluid Transients in Systems, Prentice Hall, 1993.
[5] A. L. Headford, D. J. Hill and K. A. Wilson, “20+ Years Experience with PE Pipe for
[6] J. Bowman, “The Fatigue response for Polyvinyl Chloride and Polyethylene Pipe
20
Systems,” ASTM Special Technical Publication Issue, no. 1093, pp. 101-121, 1990.
[7] UKWIR, “Plastic Pipeline Systems - Pipeline Innovation,” UK Water Industry Research
[8] UKWIR, “Pipeline Innovation Final Report,” UK Water Industry Research Ltd, London,
1996.
[9] UKWIR, “Evaluation of welding and jointing of plastic pressure pipes,” UK Water
[10] UKWIR, “Plastic Pipeline System - Analysis of system failures,” UK Water Industry
[11] B. Cosgrove, Thesis: The Toughness Characteristics of Butt Fusion and Electrofusion
Electrofusion Joints in PE pipe systems for the Water Industry,” Progress in Rubber and
[13] WIS 4-32-08, “Specification for fusion jointing of polyethylene pressure pipeline
[14] WIS 4-32-14, “Specification for PE 80 and PE 100 electrofusion fittings for nominal
21
Welding in the UK Water Industry,” in Plastic Pipes IX, Edinburgh, 1995.
short-term tests for electrofusion joints in PE pipe,” in Plastic Pipes XIII, Washington,
2006.
[17] S. H. Joseph and P. S. Leevers, “Failure mechanics of uPVC cyclically pressurized water
[18] ISO 13955, “Plastic pipes and fittings - Crushing decohesion test for polyethylene (PE)
[20] J. Bowman, “The fatigue performance of polyethylene pipe joints made with
electrofusion fittings,” Plastics and Rubber Processing and Applications, vol. 9, pp.
147-153, 1988.
water pipeline systems to surge pressure,” in Plastic Pipes IX, Edinburgh, 1995.
22