Introduction and Background Information

Polyethylene (PE) is the primary material used for gas distribution pipeline applications because it has several advantages over traditional materials; for instance light weight, flexibility, ease of joining and long-term durability, along with lower cost and lack of corrosion. Different grades of PE have been used as pipe materials, however the high density polyethylene (HDPE) is the most preferred material for natural gas pipe manufacturing because it has high strength and a high modulus. For the last several years, the HDPE is being successfully used in geotechnical and civil engineering applications [1–4].

In order to join PE materials various fusion joining techniques are being employed; because of its high welding capability. There are lots of developed joining methods for HDPE pipes; however butt fusion welding, electrofusion welding and extrusion welding are the most widely used methods [5]. Among these welding methods, hot plate butt fusion welding has gained immense popularity due to low cost and relatively simple procedure and therefore is more commonly employed for welding of PE pipes in the industry. The butt fusion welding process has been previously studied and described in some detail [3].

Although the welding of HDPE pipes for natural gas transmission are generally carried out by well trained welders in accordance with the prescribed rules and utmost care is usually taken during inspection to ensure defect-free joint, even then cracking and leakage has been frequently observed in welded plastic pipes during service. It is therefore necessary to identify root causes of such unexpected failures and increase the understanding about the welding characteristics and resulting properties and performance of the welded PE pipes.

This article describes the observations, results and conclusions of a failure investigation carried out on HDPE liner which was acting as a non-loading bearing member of a composite compresses natural gas (CNG) cylinder (Type IV). The liner was manufactured by joining the two thermoplastics moldings (pipe section with the dome section) through heated tool butt welding. The pipe section was made by blow molding whereas the dome section was made by injection molding techniques and the two sections were welded afterwards. The fusion butt welding parameters are not known to the authors. While conducting hydrostatic (proof pressure) testing for qualification of composite CNG cylinder leakage was observed after about 2000 cycles at 250 bars. It is necessary to identify root causes of such unexpected failure; therefore detailed failure investigation was initiated to reveal the cause of the leakage.

Visual Observations

Visual examination was carried out on-site and the leakage of water was observed from the surface cracks in the fused joint region. The cracking in cylinder was mainly circumferential around the junction between the cylinder section and dome section, as shown in Fig. 1a. Circumferential cracking was also found at the interface of dome section and metal adapter for water inlet, as shown in Fig. 1b. The presence of circumferential cracks in the joint indicates the existence of longitudinal or bending stresses.

Fig. 1
figure 1

Pictures of composite CNG cylinder showing leakage from HDPE liner at (a) dome-cylinder circumferential weld joint and (b) metal adapter–liner weld interface

Full size image

Macro-Examination

Macro-examination was carried out using a stereomicroscope (Meiji Techno) at a magnification of ×5 and ×25. Samples from the region of leakage containing cracks were sectioned from the joint for macro-examination. Figure 2a shows the photograph of fracture surface of dome section. The crack initiated from the inner surface (root of the joint) of the dome wall and then propagated outwards to the outer surface which ultimately led to break-through in the wall and leakage occurred (Fig. 2b). Lack of fusion (also called cold weld) was also observed in the joint region (Fig. 2c). This defect is very difficult to detect from the outside of the intact weld, because the defect may not result in any actual gap or separation in the weld zone but its failure typically appears brittle in nature [6]. The fracture surface did not exhibit any sign of plastic deformation which shows that the fracture mode is primarily brittle. Moreover, striated structure was prominent in the joint region as shown in Fig. 2b. These beach marks (river-like lines) were found on the edges of both blow molded cylinder section and injection molded dome section, as shown in Fig. 3a and b respectively. The striation lines are one of the fatigue fracture characteristics and are formed by fatigue crack propagation during the loading–unloading cycle of proof pressure testing. Such failures frequently occur due to the fundamental stress cracking susceptibility of semi-crystalline HDPE [1, 7].

Fig. 2
figure 2

Macro photographs of failed HDPE liner showing (a) leaked surface, (b) direction of crack and fatigue striations indicated the crack arrest lines, and (c) lack of fusion

Full size image
Fig. 3
figure 3

Photographs taken by stereomicroscope showing beach marks (striations shown with arrow) on fracture surface of (a) cylinder section and (b) dome section

Full size image

Inspection of the weld beads provides a useful quality check on the weld. Another observation made on dome section was the presence of a rough, bubbly weld bead having abnormal appearance, as shown with arrow in Fig. 3b. It seems that either the dome section (injection molded) was heated to higher temperatures during joining with cylinder section (blow molded) or an uncontrolled melt flow occurred. The poor quality of weld bead is obvious in Fig. 3b. Whereas the weld bead of cylinder section was found uniform without such sponge-like structure.

Dimensions of the weld zone and bead formed on both sides of the butt joint were also noted as it can give qualitative information about the weld, as shown Fig. 4a (schematic of weld bead). It was clearly seen that the beads differ in sizes and appearance. Weld bead of cylinder section was larger than the dome section, which reflects the deficiency in the weld parameter (i.e., pressure, heating time or temperature). Usually, this defect occurs due to misalignment, defective heating tool, heating temperature not within specified range, incomplete facing, or component slip in clamps. Weld bead width is related to wall thickness and it should be 9.52 mm (max.) for wall thickness of about 8 mm. But in this case the weld bead width was measured to be about 10.2 mm, as shown in Fig. 4a. As a rule of thumb, the size differential between two single beads shall not exceed 10% of the actual bead width. Here, it was calculated to be 7.8%, which is within the acceptable range.

Fig. 4
figure 4

Showing (a) schematic of weld bead sizes of fusion butt joint and (b) macrograph of heat affected zone around weld bead

Full size image

Macro-structure was also examined to reveal heat affected zone (HAZ) size; width of the HAZ is a good indicator of the weld quality. Heat affected zone is the region in which the structure is affected by the applied heat. Specimen was taken from the weld joint and ground on 240–400 grit papers and then etched in hot xylene solution. The macro-structure of HAZ is shown in Fig. 4b.

It was seen in Fig. 4b that the width of the HAZ was not sufficient. Presence of thin HAZ implied that there has been excessive melt movement and that the bulk of the molten material has been pushed out from the joint interface which resulted in a high degree of molecular orientation at the interface, resulting in poor ductility of the weld. The most likely cause for this was the use of excessive pressure at the pipe ends during fusion which resulted in most of the weld “melt” being forced from the joint. This caused thinner than normal bonded area. Conversely a wide HAZ would indicate lower than the desired pipe interface pressure during fusion and in the worst case could give rise to void formation as cooling takes place. Therefore, during fusion welding of PE the designer has to select an optimum pressure to ensure thorough mixing of the material across the interface.

SEM Fractography

Fracture surface was also examined under scanning electron microscope (SEM) (JEOL JSM 6380) at higher magnifications to reveal fine detail of the failure and identify elementary process involved in the crack initiation and propagation. Fracture surfaces were mounted onto a copper stub and coated by ion sputtering with gold prior to examination. Figure 5a shows the striations typical of fatigue failure in SEM photograph. The striations were characteristic of stepwise crack propagation, and indicated the crack arrest lines. The fractography confirmed that the crack propagated in a stepwise manner through a fibrous craze. It is known from literature that the crack jump length (i.e. the distance between the fatigue striations or steps) and damage zone size is related to the frequency of cyclic loading, and mean stress and length systematically increased as the frequency decreased [8].

Fig. 5
figure 5

SEM photographs of fractured surface showing (a) striations (beach marking) typical of fatigue failure, (b) fibrils morphology, and (c) drawn fiber sheets

Full size image

In addition to fatigue markings the damage zone (or craze) shows the distinct fibril-like morphology (Fig. 5b) at the ends. Most of the fracture surface was covered by the fibril structure indicating that cracks propagated via a slow crack growth (SCG) mechanism. The fibrils were highly drawn and oriented towards the crack propagation direction as seen in Fig. 5c. These fibrils were intertwist to form thicker fibrils and tightly packed. The ends of the fibrils were slightly pulled down from the material matrix. The small size of these fibers suggests that the applied stress is relatively low during crack propagation. Sheets of fibrils can be seen that were compounded upon each other (see Fig. 5c). On loading the crystallites deformed to form fibrillar crystallites, but the small size and less volume of fibrils indicated that very little deformation was occurred. The fracture surface does not exhibit large deformation; smooth fracture face morphology was evident in Fig. 2. No appreciable pull-out or thinning down of material adjacent to the crack was observed.

The absence of large apparent deformation and relatively smooth fracture face suggested that the failure is actually Stress Cracking (often called “Slow Crack Growth”), which is a well-recognized phenomenon in HDPE. The SCG mechanism in HDPE has been studied by many researchers [4, 9, 10]. It is essentially a macro brittle cracking phenomenon that occurs at a stress significantly less than the yield or break stress of the material. It is invariably initiated at an internal or external defect and imperfections in the material such as an inclusion, void or scratch, etc. The lack of fusion (i.e. cold weld) and voids are evident in the joint (Fig. 5a). It appears that the cracks were most likely start from these defects at the inner surface and then propagate through the wall and then in a circumferential direction. The circumferential cracking indicates that the driving force for crack growth was oriented longitudinally.

Density, Melting Point and Melt Flow Index

The quality of butt-fused joints of HDPE depends largely on material properties and joining parameters. Among physical properties density, melting point and melt flow index (MFI) are most important and largely affects their weldabilities [11]. During the welding process, heating time and temperature can influence crystallization kinetics and the building of interlaminar links, and the flow rate of the melt influences the deformation kinetics. To confirm the material properties of both sections samples from cylinder and dome parts were sliced for density, melting point and MFI measurements. The density of polyethylene indicates the amount of crystallinity in the material. Higher density indicates larger crystallinity in the material. Density of both cylinder and dome section samples was determined as per ASTM standard B311-93 and listed in Table 1.

Table 1 Physical properties of cylinder and dome section HDPE samples
Full size table

Melting point is an important parameter and should be considered whenever different grades of PE need to be joined satisfactorily by fusion welding. Melting point measurement and calorimetric studies of both samples from cylinder section (blow molded) and dome section (injection molded) were performed on differential scanning calorimeter (Thermal Analysis Instruments Q-Series DSC). Few grams of each sample (in a form of granule) was taken using paper cutter, and encapsulated in an aluminum pan and heated at a rate of 10 °C/min from 40 to 200 °C. The mass crystallinity was calculated using an enthalpy of fusion for a 100% crystalline PE of 296 J/g [12]. The degree of crystallinity is an important variable which affects the fracture properties of the HDPE. Degree of crystallinity of HDPE was correlated with mechanical properties by Mandel et al. [3]. Figure 6 shows the DSC thermograms of both blow molded cylinder section and injection molded dome section. Melting points and enthalpy of fusion are listed in Table 1. Blow molded sample exhibited 81% crystallinity. This higher degree of crystallinity in the blow molded sample is responsible for high tensile strength at yield. Presence of double crystallization peak in injection molded dome sample (Fig. 6b) also represented the recrystallization phenomena during the DSC heating process. First melting endotherm peak on lower temperature was due to the fusion of original crystals, this step was immediately followed by the recrystallization phenomenon. The position of first peak is dictated by the simultaneous occurring melting and recrystallization phenomenon. Second melting peak on higher temperature represented the fusion of crystals generated by the recrystallization process during scan [13]. This reflects the need of controlled heating and cooling during injection molding process to produce homogenized microstructure.

Fig. 6
figure 6

DSC thermograms of specimens taken from (a) cylinder section and (b) dome section

Full size image

Melt flow index is a measure of melt viscosity—that is, how big the molecules of resin are, which affects flow. The bigger, longer or more branched molecules get tangled up more, and therefore they flow (slide over one another) with more difficulty. A lower MFI value translates to higher molecular weight. MFI is also a measure of product strength (lower MFI is better) and stress cracking resistance (lower MFI is better). Flow rate of the melt influences the deformation kinetics [14]. MFI of both samples was measured as per ASTM standard D1238-99 at a temperature of 190 °C and load of 2.16 kg (4.76 lb). Flow rates (in g/10 min) are given in Table 1. The strength of welded region between two different grades of HDPE is largely dependent on the molecular structure of the low MFI polymer, which directly influences the processing factors like time, temperate and pressure. Weld time should be directly related to the longest molecular relaxation time in polymer melt i.e. time require for the polymer chain to completely loose its original configuration. As described before, in present study a very thin HAZ (in Fig. 4b) marks the incomplete diffusion of blow molded molecule having low MFI with high MFI injection molded dome molecules.

Hardness, Tensile Testing and Flexural Testing

Shore D hardness was measured on both cylinder and dome section samples using Durometer. The average of five hardness readings is reported here in Table 2.

Table 2 Mechanical properties of cylinder section, dome section and welded specimen of HDPE
Full size table

Dumbbell-shape tensile test specimens were cut along the length of the cylinder, dome, and across the weld, such that the welded region located in the middle of the specimen. Type IV specimen (25 mm gage length) was selected due to limited material and dimensions were set in conformance with ASTM standard D-638-99. Thickness of about 3 mm was taken to produce fracture within 5 min. Three specimens were taken from each section; cylinder, dome and weld joint region to evaluate the tensile properties of blow molded part, injection molded part and weld joint respectively. Testing was carried out on 150 kN UTM (Tinius Olsen) at a crosshead speed of 50 mm/min at ambient temperature. Figure 7 shows the stress–strain curves of cylinder section, dome section and weld joint. Typically following four different failure types [15] has been observed for tensile testing as shown in Fig. 7d.

Fig. 7
figure 7

Stress-strain curves of HDPE specimens from (a) cylinder section, (b) dome section, (c) weld joint, and (d) typical failure types in HDPE

Full size image
  1. (i)

    Failure mode 1: Sample fails before the start of necking with no or little strain. Failure is brittle in nature and if occurred in welded zone, then regarded as bad joint. No such sample failure was observed in present study.

  2. (ii)

    Failure modes 2: Sample pass the yield point and enter into necking zone but fail before the end of necking zone (natural draw ratio). If occurred in welded zone, then regarded as poor joint. All samples in this study containing welded section (Fig. 7c) fall into this type.

  3. (iii)

    Failure mode 3: Sample was completely necked and strains can vary up to 350%. If the weld joints would fail in this mode, they are considered to be “good” joints. The entire samples from the dome section fail in this pattern (Fig. 7b).

  4. (iv)

    Failure mode 4: Sample was completely necked and traverse into the strain hardening region. The maximum strain can reach up to 500%. If the weld joints would fail in this mode, they are considered to be “excellent” joints. Sample from the blow molded cylinder section largely fail in this fashion (Fig. 7a).

Flexural properties were also determined through three-point flexural testing as per ASTM standard D790-99 on specimens of rectangular cross-section made from cylinder section and welded region to judge the quality of the weld. Flat rectangular specimens of about 3.2 mm thickness, 25 mm width and 90 mm length were cut from the cylinder and weld joint (such that the joint comes in center). A support span-to-depth ratio of 16:1 and strain rate of 0.1 mm/mm/min was used. The specimen was deflected until a maximum strain of 5.0% was reached. Three specimens were tested for each condition and average values of flexural stress are given in Table 2.

It was seen from Table 2 that the hardness, yield strength (YS) and percent elongation (%E) of cylinder section specimen (made from blow molding) was higher than dome section and welded specimen. The tensile properties of dome section and weld joint were inferior to cylinder section. Stress and strain at yield of welded specimens were found to be lower than their base ones as shown in Fig. 7. For a weld to be considered satisfactory, tensile specimens should show a large percentage elongation to failure (high degree of ductility). But in the present study, the welded specimen always fractured at welds in a brittle manner with low percentage elongation, as shown in Fig. 8a and b. The weld factor ‘f’ (i.e., strength of the joint/strength of the base material) was calculated to be 0.53. A weld factor of 1 or greater indicates that the joint is as strong as or stronger than the base material. However, in this case it was seen that the weld joint is of about half the strength of the base material. Bowman has reported a weld factor for hot plate welds on PE as high as 0.61 [16]. Tensile test results of welded specimen illustrated that the weld strength was almost equal to the strength of the dome section (injection molded) material. Schmachtenberg has discussed two theories (structural change theory and aging theory) for lower strength of weld joint than their base materials [17].

Fig. 8
figure 8

Photographs of (a) fractured tensile specimens and (b) fracture surface of welded specimen showing brittle failure due to poor quality weld

Full size image

The lower weld strength could also likely results from the unfused region (Fig. 8b) because of lower welding pressure on the weld root. Insufficient welding pressure on the weld root leads to unfused regions and reduce wettability at the interface; since the flow of a molten polymer to the weld root in a short time is difficult owing to the high viscosity of polymer melts [18, 19].

Fatigue Testing

It was anticipated from the fractographs that the cyclic loading played a major role in the leakage of the HDPE liner therefore the fatigue property of the HDPE materials was also evaluated.

During hydrostatic testing, CNG cylinder was pressurized to about 250 bar (i.e. 25 MPa) and total axial extension of about 5 mm (i.e. 10% strain—less than yield point strain) was measured through linear variable differential transducers (LVDT). Upon de-pressurizing, the CNG cylinder retained its original size/volume, which means that the material remains in elastic range. The fatigue test specimens were taken from cylinder (blow molded) section and weld region (with joint in center). Dimensions of the fatigue specimens were in accordance with ISO Standard 527 (Type 1BA). Specimens were machined by milling process with special care to avoid excessive heating. Fatigue testing was carried out on servo-hydraulic fatigue testing machine in strain-controlled mode by applying sinusoidal tension–tension loading at a frequency of 1 Hz and at varying load ratio (R). Having the maximum %strain of about 10% (from the hydrostatic test), strain ranging from 6 to 12% (within elastic region for cylinder section specimen) was selected for the fatigue test. The applied cyclic stresses were lower than the ultimate tensile stresses for cylinder section. Maximum number of cycles about 50000 was selected for fatigue testing since 40000 cycles was the prescribed qualifying criteria for CNG cylinder during hydrostatic testing.

Low frequency of 1 Hz was employed for fatigue testing to simulate the situation analogous to pressurizing–depressurizing cycle of hydrostatic testing. Moreover, low frequency also ensures that hysterical heating of HDPE was low. Fatigue testing was done on at least 2 samples for each condition and the results are listed in Table 3.

Table 3 Fatigue testing results of cylinder section and welded specimens of HDPE
Full size table

From Table 3, it was observed that the cylinder section (blow molded) specimen successfully sustained 50000 cycles at all strain ranges. In contrast, welded specimen failed from weld joint after only 19 cycles at 8% strain; although the applied stress was well below the yield strength. At higher strains, failure occurred after only few cycles. The results suggested that the fatigue life of welded specimen was far shorter than the cylinder section specimens.

Figure 9a and b depicted the variance in strain during flexural testing for blow molded and welded specimen. At higher strain (>10%) welded specimen is more suspected to plastic deformation (necking) compared to blow molded specimen. In cyclic loading mismatch in strains produce higher bending stresses at the dome cylinder juncture and make this section more prone to crack initiation and growth which lead to rapid failure at welded section. In addition to this the geometry of the joint itself form notch between the pipe and the weld bead. These notches could be regarded as potent regions for crack initiation and crack growth under cyclic loading. The ductility and strain compatibility (excellent weld) between welded zone and base material (cylinder & dome) can restrain these failure circumstances.

Fig. 9
figure 9

Load vs. percent strain graph for three-point flexural testing of (a) cylinder specimen and (b) welded specimen

Full size image

Optical Microscopy of the Weld Joint

Microstructure of weld zone and heat affected zone (HAZ) plays an important role in determining the ultimate quality and mechanical properties of the weld. It has been shown that the heating and cooling process at the weld produces microstructure that differs from the base material [20]. Barber and Atkinson have divided the weld microstructure into 5 distinct regions [20], namely:

  1. (i)

    Skin Remnant

  2. (ii)

    Spherulitic, slightly elongated

  3. (iii)

    Columnar

  4. (iv)

    Boundary nucleation

  5. (v)

    Spherulitic

Schmachtenberg and Tüchert [17] reported that the HAZ of hot plate welded PP sheets can be divided into well-distinguishable six regions as travel from the weld centerline to the base material. Some authors have classified the HAZ of HDPE into three regions in addition to the weld zone (i.e. weld flash): stressless recrystallization zone, columnar zone and deformed zone [21, 22].

Thin sections were cut across the butt fusion weld for examination of microstructure under an optical microscope using polarized light. Transverse face of the weld specimens were carefully ground, starting from 240 silicon carbide grit paper finally to 1200 grit paper. Special care was taken to avoid excessive heating and deformation during grinding. Polishing was carried out on nylon paper using 3 μm diamond paste followed by 1 μm. Final polishing was done by 0.05 μm alumina suspension followed by washing with distilled water and alcohol. Specimens were then etched by immersing in hot xylene for few seconds. Change in microstructures from weld center to base material on both sides of the bead was assessed and studied under optical microscope (Olympus G51) at a magnification of ×500 as shown in Figs. 10–12.

Fig. 10
figure 10

Optical micrographs of the (a) weld bead or molten zone, (b) weld zone structure, (c) cylinder section base material—spherulitic structure, and (d) dome section base material—spherulitic structure

Full size image
Fig. 11
figure 11

Optical micrographs of heat affected zone of cylinder section showing three distinct regions (a) fine spherulites in recrystallized zone, (b) columnar (lamellar) zone, and (c) deformed zone with coarse spherulitic structure

Full size image
Fig. 12
figure 12

Optical micrographs of heat affected zone of dome section showing (a) recrystallized structure, (b) coarse spherulitic structure, and (c) columnar (lamellar) zone

Full size image

Figure 10a shows weld zone or molten zone (often called melt thickness) of approximately 100 μm which was not adequate for firmly holding the joint together. The thickness of the melt layer or molten zone affects the mechanical properties and interfacial strength of the joint [23]. Thin molten zone results in brittle weld of lower strength whereas the weld strength increases with increasing the molten zone thickness and width of HAZ [22]. Figure 10b shows the structure of the weld zone taken at ×1000 magnification. It was observed that etching the microstructure with xylene resulted in microscopic relief. It is generally accepted that the non-homogeneity of flow and temperature gradients during welding induces residual stresses (both longitudinal and circumferential) in the pipe and welded HDPE structure [24]. One indication of these stresses is the microscopic surface relief upon etching. The structure of the molten zone (center of weld) was transcrystalline constituted of linear spherulites (Fig. 10b). Coarse spherulitic microstructure was seen in the base material (which was not melted) of both cylinder and dome section (Fig. 10c and d, respectively). A single melting endotherm at 129.37 °C in the DSC thermogram of cylinder (blow molded) specimen (Fig. 6a) also confirmed the presence of mainly spherulitic morphology. The spherulites are the basic morphology of semi-crystalline polymers like PE, PP, etc. They are ordered regions with spheral textures and well defined boundaries, as observed in Fig. 10c and d.

Microstructural changes in the HAZ (from weld center to the base material) of cylinder section are represented in micrographs in Fig. 11. Adjacent to the weld or molten zone was the thin, fine spherulitic microstructure formed due to rapid cooling (Fig. 11a). The size of the spherulites depends upon the cooling conditions and may vary from few dozen to several hundred microns. Fine spherulitic zone was followed by the columnar zone (lamellar type structure) as seen in Fig. 11b. The lamellar spherulites are generally formed at high temperatures. Next to the columnar zone existed the deformed zone (often called shear melt zone) having relatively coarse spherulites, as seen in Fig. 11c. The streamlines in the sheared molten zone were the results of the combined effects of rapid cooling and shear stresses. Brittle failure generally follows the boundaries of the coarse spherulites.

Figure 12 exhibits the evolution of microstructures in the HAZ of dome section. The structure change was almost similar to that of cylinder section HAZ.

Conclusions

The results of the investigation carried out on failed composite CNG cylinder liner made of HDPE and joined through butt fusion welding led to the following conclusions:

  1. (1)

    Visual examination revealed that the leakage was resulted from the failure of circumferential weld joint due to the existence of bending stresses.

  2. (2)

    Optical and SEM fractography showed striations, characteristic feature of fatigue failure, on the fracture surface taken from leaked area.

  3. (3)

    Density and melting point was higher for cylinder section (blow molded) than dome section (injection molded), where as melt flow index was lower for cylinder section. This indicates that the dome section was made out of low grade HDPE. Large difference between the MFI of cylinder and dome section renders both materials very difficult to weld; weld joint of poor quality was observed.

  4. (4)

    Hardness, yield strength, percent elongation and flexural strength of cylinder section were also found higher than that of dome section. Moreover, tensile strength of welded specimen was about half the strength of cylinder section specimen. Percent elongation at fracture of welded specimen was also found drastically lower (about 27%) and all the welded specimens failed in brittle manner from weld joint.

  5. (5)

    Blow molded HDPE (cylinder section) qualifies the specified 50000 cycles at 10% strain (max.) during fatigue testing without failure. Contrary to this, welded specimen failed from joint after sustaining 11 cycles at 10% strain during fatigue loading. Upon cyclic loading during hydrostatic test the fatigue phenomenon was occurred which resulted in brittle fracture from joint. The cracking was attributed to stress cracking via slow crack growth mechanism.

  6. (6)

    Optical microscopy of the weld joint revealed three distinct zones in heat affected zone. Weld zone size was measured to be approximately 100 μm which was not of satisfactory thickness to strongly hold both sections.

  7. (7)

    Results suggested that the root cause of failure was the use of low grade injection molded HDPE in dome section (having lower strength and large MFI difference) combined with improper joining conditions which inhibited proper fusion.