Will a buried composite pipeline system fail at its joints under the effects of overburden soil,pipe operating pressurization,and traffic loads?

Abstract

In real world applications, buried pipelines span across great lengths. It is inevitable that certain sections of a buried pipeline experience external loads in addition to top soil overburden, such as weights of aboveground buildings and traffic loads located directly above these sections. The present study investigated the effects of overburden soil, pipe internal pressurization, and traffic loads on fiber-reinforced plastic pipelines at various pipe sections with particular emphasis on pipe joints using finite element method. This study includes realistic modeling of traffic loading on service road running across a buried pipeline system, consisting of straight, bent, and joint sections. Our results also revealed that surcharge loading might not be a predominant factor in pipe failure or leakage issues as compared to the cyclic pipe internal pressurization. Moreover, it was also confirmed in our study that the pipe joint remained as the most critical region for pipe failure or leakage issues.

Keywords

Fiber-reinforced plastic glass-reinforced epoxy composite finite element vehicle load pipe joint leakage

Introduction

In the recent years, there has been a steady rise in utilization of light-weight composite materials in buried pipelines, providing an alternative to the conventional steel pipes. Fiber-reinforced plastic (FRP) pipes, in particular, offer better corrosion resistance of polymer composites, lower thermal expansion factor, high strength-to-weight ratio, and low friction factors, as compared to the conventional steel buried pipes.^1–5 However, there is a growing concern regarding the structural integrity of these flexible light-weight composite pipes, given that the modulus and pressure expansion values of typical FRP products being over 10 times lower and 25 times higher than those of their traditional metallic counterparts, respectively. Moreover, these underground facilities, which serve a wide array of purposes ranging from sewerage and water lines to fuel transportation, are subjected to combinations of complex loading such as internal pipe pressure, vertical earth loads, surface live loads, and movement at pipe bends.^6–10 All these loads need to be accounted for in the pipeline design. As these pipes are buried underground for a long period of time and any leakage or failure incurs costly excavation, it is particularly important to ensure that the buried pipes work properly and efficiently over their designed lifespan.

Finite element (FE) simulations, which serve as a cost effective alternative to large-scale experiments, are used extensively in understanding various three-dimensional (3D) loading effects on deformable buried pipes comprising metallic and composite materials. The first 3D FE analysis was conducted in 1994 by Moore and Brachman¹¹ which investigated the response of shallow buried steel pipes when subjected to vehicle live loads. These simulated results were then validated against the pipe response from field tests featuring a heavily loaded truck producing axle forces at conventional design limits.¹² Moore and his co-workers have been very active in the study of buried steel pipes subjected to vehicle loads, and more recently, Almahakeri et al.¹³ employed FE simulations to compare the flexural behavior of buried steel and FRP composite pipes, and found that FRP pipes demonstrated superior flexibility in longitudinal bending compared to the steel pipe. Elshimi et al.¹⁴ studied the behavior of the buried long-span steel pipes under truck loading for different truck positions and found that the highest moment or thrust occurred when the truck tandem axles were located above the crown of the pipe. In addition, they had studied the effect of compacted soil on five different rigid and flexible buried pipes with different relative flexure stiffness values using 2D FE models.¹⁵ Of late, Toh et al.¹⁶ compared the responses of the FRP composite and steel buried pipeline systems, with the valve pit subjected to soil load. On the other hand, FRP pipe joints have not been ignored, especially when the pipe joints are known to be the most prone to content leakage. Yang and his co-workers utilized both analytical and numerical methods to study FRP pipe joint under various loading modes,^17,18 while Ladhwe et al.¹⁹ explored the effectiveness of FRP joint under tensile and bending load via experiments and FE simulations. Despite the abundant aforementioned works, there is still a paucity of research investigating the structural integrity of these buried composite pipe joints in a large-scale pipeline system under various real-life load conditions. The present study includes detailed modeling of the interference-fitted bell-spigot composite joints in an entire FRP pipeline system and investigates the effects of overburden soil, pipe internal pressurization, and traffic loads on FRP pipelines at various pipe sections with particular emphasis on pipe joints. This study provides a cost-effective way to analyze various loading effects on the entire buried pipeline system in various real-life applications, and hence provides an insight on our understanding of these realistic loading effects on various regions of a buried pipeline system, including the fragile pipe joints

Materials and methods

Model description

The model consisted of a soil block with a 48-m-long pipe network embedded within it and a service road on its surface (Figure 1). The soil block, which has dimensions of 45.5 m × 10.5 m × 12.0 m, was divided into five soil strata with different properties, following soil investigation (SI) test results^20,21 whilst the pipe network had five pipe sections with various lengths, four 12-m-long straight pipe connections and one 45° elbow connection. All the pipe sections (with a nominal diameter of 400 mm and wall thickness of 6 mm) as well as the bell-spigot at each pipe joint connection were modeled according to the manufacturer's catalogue.²² On the surface of the soil, there was a 6.4-m-wide service road inclined at 45 ° to the pipe axis. The road was assumed to be rigid pavement, which consists of an uppermost concrete slab surface course and a granular subbase course, so that the load was distributed over a large area of subgrade soil.²³ The dimensions and thicknesses of various layers for the rigid pavement were taken from the Singapore Land Transport Authority.²⁴ The 48-m-long pipe was buried at 3 m in the soil and above water table. The entire FE soil–pipe–road model was chosen to be meshed with 1,068,958 first-order reduced-integration hexahedral elements after the mesh convergence study (details in Appendix 1). It should be noted that the use of first-order reduced-integration elements is justifiable as long as a reasonably fine mesh is used with the introduction of the default artificial “hourglass stiffness.”

Material properties

The pipes in the current study are Ameron Bondstrand® 7000 FRP pipe with stacking sequence of [55/−55]₆. The anisotropic properties of the FRP pipes were obtained based on the classical laminate theory^25,26 using the properties provided by the manufacturer Ameron (Ameron). For all five strata of soil, the Mohr–Coulomb soil plasticity model was used based on an empirical estimation from standard penetration tests in a SI report of actual site.²¹ These material properties are summarized in Table 1.

Table 1.

Material properties of all components modeled.

	Material properties
Components FRP	Young's modulus, E (MPa)			Shear modulus, G (MPa)			Poisson's ratio, υ			Density, ρ (ton/mm³)
Components FRP	E₁	E₂	E₃	G₁₂	G₁₃	G₂₃	υ₁₂	υ₁₃	υ₂₃	Density, ρ (ton/mm³)	References
	4430	25,200	11,400	2960	1910	12,800	0.04	0.12	0.56	1.90E-9	Ameron²⁷

Soil	Young's modulus, E (MPa)	Friction angle	Poisson's ratio, υ	Density, ρ (kg/mm³)
1st Soil strata (top soil/backfill)	6.56	28	0.15	1.60E-09	V-PILE²¹
2nd Soil strata (loose yellow fine to coarse-grained sand)	9.38	28.4	0.1875	1.68E-09
3rd Soil strata (medium dense yellow grey silty sand)	49.58	34	0.225	1.75E-09
4th Soil strata (dense light grey silty sand)	66.25	37.9	0.2625	1.83E-09
5th Soil strata (dense to very dense light grey silty sand with gravels)	75.21	40	0.3	1.90E-09

Road	Young's modulus, E (MPa)	Poisson's ratio, υ	Density, ρ (ton/mm³)
Surface course	34,500	0.175	3.15E-09	Fwa,²³ Huang,²⁸ and Bye,²⁹
Base/subbase course	355	0.32	2.04E-09	Fwa,²³ Huang,²⁸ Robinson and Thagesen,³⁰ and LTA³¹

FRP: Fiber-reinforced plastic.

Interactions, boundary conditions, and loading conditions

The interactions between the pipes or/and elbows in the quick lock joints were modeled with normal hard contact and tangential sliding behavior using a friction coefficient of 0.2, whereas contacts between the soil–pipe and soil–road interfaces were modeled using tie constraints. The vertical surfaces of the model were assigned zero displacement boundary conditions in the horizontal and normal directions, allowing vertical motion to simulate lateral soil constraint (Figure 2). The base surface of the soil model was assigned fixed boundary condition to assimilate the hard stratum beneath (Figure 2). The effects of truck load on the soil and pipe were simulated using pressure loads on selected areas of the road surface. The pressure loads were obtained based on the weight of a six axle truck with gross combined weight rating of 38.5 tonnes. According to the positions of the axles,³² the load distribution assigned were 5.4, 15.4, and 17.7 tonnes (12, 34, and 39 kips) for the front, middle, and rear axles, respectively. The tyre pressure was taken as 0.62 MPa (90 psi) which is common among medium to heavy trucks.³³ The tyre-to-ground contact areas for each of the front, middle, and rear tyres were calculated to be 0.0432 m², 0.122 m², and 0.140 m², respectively, according to the proposed analytical model of Majumdar et al.³⁴ The load patches were then placed accurately on the rigid pavement road, taking into account the distances between the load patches.

The present study adopted a multiple-steps approach to simulate various loadings, such as pipe joint interference fitting, pipe pressurization, soil loading, and truck loading, using the FE software Abaqus v6.13-1 (SIMULIA, RI, USA). Prior to application of pipe pressurization and truck loads, the composite pipes were subjected to interference fitting to model the bell-spigot connection between the various pipe sections. Following which, a gravity load was applied in a geostatic step to obtain the effects of soil overburden on an unpressurized pipe. Thereafter, a pressure load was applied on the internal surfaces of the pipes to simulate the internal pressure present during operation. Eventually, pressures simulating the weight of a 38.5 tonnes truck were applied to selected areas of the road surface.

Results and discussion

Effects of overburden soil and pipe pressurization

Soil response

In order to assess possible soil yielding or detachment from the pipe surfaces, the soil stresses and strains were monitored in the various load scenarios. The soil Mises stresses due to the pipe pressurization and overburden load are shown in Figure 3. It was indicated that soil stresses increase with depth, with the highest occurring at the bottom strata due to overburden load on top, as well as its slightly higher strata modulus. The road pavement experienced higher stresses as compared to the surrounding ground, due to the different road layer properties with respect to the soil. With the internal pressurization during operation, the pipe expanded which affected the stress distribution of the surrounding soil adjacent to the pipe. Cross sectional views reveal that the soil at the crown and invert of the pipe generally experienced higher stresses than the soil near to the springlines. Moreover, when the pipe is pressurized, maximum soil Mises stress was no longer at the base strata, but near to the pipe bend and connections, due to pipe expansion that affected the adjacent soil. At the depth of pipe, the soil was generally experiencing stresses of around 0.036 MPa. The plastic strain contour plot revealed that a very small region of soil at the bell connection regions has yielded, with a plastic strain of around 0.04. Nevertheless, soils further away from the bells were still entirely within the elastic range.

Pipe responses

Next, the structural integrity of the pipeline was determined by pipe hoop stresses and axial stresses. Figure 4(a) shows the hoop stress of the pipes after internal pipe pressurization and geostatic loads. High tensile hoop stresses occurred at the exterior bell portions and were more than double of the maximum compressive values. The pipe elbow was observed to experience higher hoop stresses than straight connections, and that the inner springline of the elbow had higher stresses than the outer bend portion. For the pipe-body in general, the crown and invert regions attained slightly higher hoop stresses than the springlines which implied that pipe failure or leakage could likely be at these locations as the crown and invert surfaces bend more. Furthermore, the hoop stresses in the pipe internal wall were higher than that on the external wall, implying that cracks might initiate from the pipe inner walls rather than the outer walls for an operating pipe without external loads. The range of hoop stresses was between 97 MPa to −43.5 MPa. Comparing these values with those from the case of non-operating condition (48.8 MPa and −60.0 MPa), the pipe was now more in the tensile state. It is generally recommended that excessive compressive state be avoided so as to prevent pipe wall buckling or wrinkling. The maximum hoop stress, which occurred at the pipe connections, was within the failure threshold of 220 MPa. The average hoop stresses at the pipe connections were around 80 MPa and −35 MPa. Without pressurization, the compressive stresses were about 20% higher than tensile stresses since there was no internal pipe pressure, and the external pressure on the pipe caused by the geostatic load would lead to the higher compressive pipe stresses.

Figure 4(b) shows the axial stress contours on the pipeline where significant changes in the stress states were observed, with maximum axial compressive stress occurring at the inner bend. High axial tensile stresses were found at the inner spigot region of the pipe joint as well as the springlines of the straight pipes connecting to the pipe bend. As compared to the non-operating pipe condition, the pipe axial stress increased from 8.28 MPa to −9.25 MPa to 11.6 MPa to −10 MPa when the pipe was pressurized during operating condition. The stress distribution of the operating case was also different from that of the non-operating case; with pipe pressurization, the pipe body experienced high axial stresses, while the straight connections experienced relatively lower axial stresses, as compared to the non-operating case. Despite that the peak stresses fell within the glass-reinforced epoxy pipe's axial strength (59 MPa), these peak positive and negative values of axial stresses occurring at the pipe bend revealed this location as the critical location for leakage.

The contact outputs, which determine the quality and integrity of the pipe connections, revealed that contact pressures had increased to a range between 1.04 MPa and 4.75 MPa as compared to the non-operating case (0.442 MPa). Highest contact pressure occurred at the inner springline of the pipe joint interface, while the lowest pressure region was located at the pipe joint's outer springline (Figure 5). With geostatic loading and even pipe pressurization, the highest stress concentration occurring at the pipe connections was still much lower than its material strength. The stress concentration was at pipe connections. Furthermore, the geostatic loads were found to induce less pipe stresses than those caused by connection fitting.

Effect of truck load

Soil and road responses

Figure 6 shows the soil Mises stress contour and a translucent view of the applied truck load relative to the joint location. The maximum Mises stress was found to be around 0.375 MPa which occurred at the road pavement where the truck load was applied. Stresses on the first two soil layers directly under the truck load were slightly higher but eventually become uniform deeper into the ground. The soil immediately adjacent to the pipe experienced about 0.1 MPa.

Higher road and soil displacements, due primarily to the truck load, were predominant at the road pavement region. For both cases with and without internal pipe pressurization, the maximum vertical road settlements were 5.035 and 5.047 mm, respectively, and were located at the road where the truck load was applied. The soil settled less with pipe pressurization. It was showed that pipe pressurization and truck loading were unlikely to cause significant changes in regional soil stresses or ground settlement. With pipe pressurization, slightly upward soil displacements (∼0.5 mm) were also found near the pipe bend which was some distance from the truck load. The applied ground load together with the continuum soil caused the pipe to bend downwards at the road but generated a moment to cause soils further away to move upwards, due to the bedsoil support under the pipeline.

Pipe responses

With the truck load, the contours for hoop stress do not display much change as compared to the case of no truck load, as seen in Figure 7(a).

When comparing the hoop stresses before and after application of the truck load, it was observed that the truck load did not cause much change in the hoop stress of the pipeline beneath the road. Table 2 shows a comparison of the peak hoop stresses at the various stages of loading. It can be seen clearly that the effects of pipe pressurization increased the hoop stresses in the pipe more significantly than the truck load.

Table 2.

Comparison of maximum hoop stresses in pipeline under various loading conditions.

Scenarios	No truck load		Truck load applied
Scenarios	Non-operating	Operating	Non-operating	Operating
Compressive hoop stress (MPa)	−60.0	−43.5	−59.5	−44.3
Tensile hoop stress (MPa)	48.8	97.0	51.0	98.8

Further analysis of results shows that the locations of maximum and minimum hoop stresses, remaining at the elbow connection, were unaffected by the truck load. For the cases with and without pressurization, higher hoop stresses at the pipe-body were observed to be located at the inner wall fibers of the crown and invert. This implies that cracks were more likely to propagate at the internal wall. The affected connection experienced around −36 MPa at the spigot and 72 MPa at the bell of the interconnecting section.

Figure 7(b) shows the axial stresses, of the pressurized pipeline, caused by the truck load. The peak axial stresses were 11.6 MPa for the operating (10 bar) case. The locations for the peak values remained unchanged at the elbow connection, particularly at the bell-slope transitions. For the operating case, the highest compressive axial stresses occurred at the “inner” bend and the bell transition of the pipe bend, while the maximum tensile stresses were found at the spigot region in contact with the bell and also at the inner springline of the straight pipes close to the pipe bend. A difference between the operating and non-operating cases was that the axial tensile stresses were higher at the pipe body for the former than at the straight connection, as shown in Figure 7(b) and Table 2. The peak axial tensile stress for the pressurized pipe was about 43% higher than the unpressurized case.

Our result revealed that surcharge loading at the pipeline might not be a predominant factor in raising pipe stresses, as high stresses usually remained at the bend or straight connections. Although the 38.5 tonnes truck load increased the stresses of the buried pipeline, they were still below the failure threshold. For the case with internal pipe pressure (operating case), the maximum hoop stress was about 45% while the axial stress was about 20% of their respective failure strengths. For the case without internal pressure, under finer resolution of stress values, slightly higher positive axial stresses were revealed to be located at the inner fiber of the invert (∼2.83 MPa). The results implied a slightly higher probability of pipe failure at the invert regions, if all else remained the same. This, however, would likely be a manufacturing issue since the high-stress regions were at the pipe inner wall faces.

The resulting pipeline displacement due to the truck load was analyzed next. Figure 8 shows the pipe displacements experienced by the pipeline. The maximum displacement (of about 1.65 mm) was no longer at the pipe bend but occurred at the crown of the pipe, near to the straight connection where the truck load was applied. The downward displacement vectors were prominent at the pipe region below the road while slightly upward and outward vectors were observed at the pipe bend. It was noted that without pressurization the pipe elbow had inward pointing vectors, whereas after pressurization, the pipe internal pressure tended to “open” up the bend to reduce ovaling and alleviated the detrimental displacements at the pipe elbow.

For pipe vertical displacements, the maximum downward displacement of 1.635 mm occurred below the truck load, while an upwards displacement (+0.09 mm) was revealed at the invert of the pipe bend. This observation implied that there was bending in the pipeline caused by the truckload.

In order to detect any pipe leakage at the pipe joints, the contact pressures at the connection interfaces were included in the analysis. Figure 9 shows the contact pressure at the joints for the truck loading case with operating pressure. It can be seen that the peak contact pressures remain located at the pipe bend. The truck load at the road pavement caused only minor changes to the contact pressures. Table 3 shows the pipe joint contact pressures for various loading and operating conditions. For the non-operating pipeline, the contact pressures with and without truck load ranged from around 2 MPa to 0.9 MPa. When the pipe is in operation, the contact pressure at the bell-spigot joint ranged from about 4.7 MPa to 1 MPa. There was a trend towards a decrease in contact pressure range when the truck was running on the service road. The highest and lowest contact pressures were observed to be close to the inner and outer springlines of the pipe elbow, respectively. The difference in values at two opposite springlines showed that some bending moment was set up along the pipeline under truck load. The positive values of all contact pressure at the joints indicated that with the truck load, the entire pipeline was still intact, and the integrity of the pipeline connections was not breached.

Table 3.

Comparison of pipe joint contact pressure ranges under various loading and operating conditions.

Scenarios	No truck load		Truck load applied
Scenarios	Non-operating	Operating	Non-operating	Operating
Maximum contact pressure (MPa)	2.03	4.76	1.97	4.66
Minimum contact pressure (MPa)	0.847	1.04	0.862	1.08

As for the global pipe response, we had adopted a convenient way of analysis through nodal path plots, which can provide pipe response along the length of the pipeline. Path plots were created at the springline, crown, and invert positions. The start and end points, as well as positions of the four paths are illustrated in Figure 10(a).

Figure 10(b) to (e) shows the path plots of hoop and axial stresses along the four paths identified. For both the cases with and without truck load, the stress levels at the pipe joints were approximately two-folds than the pipe bodies, making the pipe joints the most prone region for content leakage or failure. The similar stress magnitudes before and after application of truck load confirmed our earlier observation that the truck load did not cause much change to the pipe stresses. Stresses were generally constant at the pipe bodies, but had a greater variation nearer to the connections. The peak stresses were all concentrated at the connections due to the fitting stresses generated. The third peak represented the region at the pipe bend. As the elbow had two bell connections, there would typically be two peaks at this region. It was also observed that in general, hoop stresses were about 3 to 4 times higher than axial stresses. The disparity between these values amplified at the pipe connections. The difference between hoop and axial stresses was likely due to the anisotropy of pipe properties.

While the hoop stresses were more constant (as shown by flatter lines) at the pipe-body, the trend in axial stresses generally became more “bulgy” or “arched” after pipe pressurization, with a tension of around 8.5 MPa that dropped towards the connections.

At the straight pipe sections very near to the pipe bend (∼3L), the inner springline axial stresses were higher than the outer springline axial stresses, implying some amount of pipe wall shearing since the stress lines did not overlap each other like those observed further away from the pipe bend. In addition, all the path lines had generally about the same hoop stress magnitudes, except for the inner springline near the elbow where much higher values of around 75 MPa were registered, while the outer springline gave comparatively lower hoop stresses.

Figure 11 shows the vertical displacement which is representative of pipeline settlement due to truck load. The pipeline settled at around −1.6 mm downwards with the applied truck load. Negligible plasticity was observed for the soil near the truck load implying that when the truck moved away from the buried pipeline, the soil and pipeline settlement should recover. The crown underwent larger displacement (−1.63 mm) than the invert (−1.35 mm), and the difference in values implied deformation of the pipe walls at the affected region. The pipeline beyond one pipe length (1L) from the truck load appeared to be undisturbed with the contact. The close values of the crown, invert, and springlines beyond this distance implied that the pipe neither underwent much pipe wall deformation nor kinematic movement, unlike the region under the truck load which yielded larger differences between crown and invert values, implying greater pipe deformation. The vertical displacements of both springlines were very close together, indicating insignificant torsion to the pipeline. At the pipe bend, the crown and invert values were very close, while the springline values were the most apart, signifying greater deformation at the springline of the elbow.

Benchmarking study

Additional benchmarking simulation of buried steel pipeline was conducted, and the predicted pipe deflections were compared with the analytical calculations in accordance to the American Lifelines Alliance's guideline³⁵ (details in Appendix 2). The FE-simulated deflection values agree reasonably well with those obtained analytically, with percentage difference of about 15%.

Conclusion

In this study, the effects of overburden top-soil, pipe internal pressurization, and traffic load on various sections of FRP pipeline with particular emphasis on pipe joints were investigated through modeling and simulation. The effect of the traffic load on pipe responses was found to be insignificant (a 2% increase in the pipe hoop stress) as compared to the pipe internal pressurization (approximately 137% increase in pipe hoop stress). Our results revealed that surcharge loading at the pipeline might not be a predominant factor in raising pipe stresses, as high stresses usually remained at the bend or straight connections. It was also found that the pipe hoop stresses were about significantly larger (about 3 to 4 times) than pipe axial stresses due to the anisotropy of FRP pipe properties, and the disparity between these values amplified at the pipe connections. This study does not only provide the pipeline transport industry with a better understanding of various three-dimensional (3D) loading effects on deformable buried pipes, but also offers the industry a cost-effective way to analyze pipe leakage in various real-life applications.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The support of Defence Science and Technology Agency (DSTA), Singapore, under project grant number R-379-000-031-422 for this work is gratefully acknowledged.

ORCID iD

Kwong Ming Tse

Appendix 1: convergence study

References

Jones RM. Mechanics of composite materials. 2nd ed. London, UK: Taylor & Francis, Inc., 1998, p.538.

Armstrong KB, Bevan LG and Cole WF. Care and repair of advanced composites. Pennsylvania, Michigan: SAE International, 2005.

Beckwith

Greenwood

. Don't overlook composite FRP pipe: satisfying hydraulic and fluid are key considerations in the efficient design of FRP piping systems. Chem Eng 2006, pp. 113: 42–48.

Taheri F. 18 – Advanced fiber-reinforced polymer (FRP) composites for the manufacture and rehabilitation of pipes and tanks in the oil and gas industry. In: Bai J (ed.) Advanced fibre-reinforced polymer (FRP) composites for structural applications. Cambridge, UK: Woodhead Publishing, 2013, pp.662–704.

Almahakeri

Fam

Moore

. Longitudinal bending and failure of GFRP pipes buried in dense sand under relative ground movement. J Compos Constr 2013; 17: 702–710.

Schmit K. Fiberglass reinforced plastic (FRP) piping systems designing process/facilities piping systems with FRP: a comparison to traditional metallic materials. Fiberbond, 1998, p.16.

American Lifelines Alliance. Guidelines for the design of buried steel report, 2001, p.76. http://www.americanlifelinesalliance.com/.

Moser

Folkman

. Buried pipe design, New York: McGraw-Hill, 2008, pp. 601.

Alzabeebee S. Investigating the response of rigid pipes with poor haunch support subjected to trafficlive load using numerical modeling. In: The 11th pipeline technology conference. Berlin, Germany, 23–25 May 2016.

10.

Chapman D, Alzabeebee S, Jefferson I, et al. Investigating the maximum soil pressure on a concrete pipe with poor haunch support subjected to traffic live load using numerical modelling. In: The 11th pipeline technology conference. Berlin, Germany, 23–25 May 2016.

11.

Moore

Brachman

. Three-dimensional analysis of flexible circular culverts. J Geotech Eng 1994; 120: 1829–1844.

12.

Moore

Taleb

. Metal culvert response to live loading: performance of three-dimensional analysis. Transport Res Rec 1999; 1656: 37–44.

13.

Almahakeri M, Moore ID and Fam A. The flexural behavior of buried steel and composite pipes pulled relative to dense sand: experimental and numerical investigation. In: Proceedings of the 9th International Pipeline Conference. Volume 1: Upstream Pipelines; Project Management; Design and Construction; Environment; Facilities Integrity Management; Operations and Maintenance; Pipeline Automation and Measurement; Calgary, Alberta, Canada, 24–28 September 2012. pp. 97-105. ASME.

14.

Elshimi

Brachman

RWI

Moore

. Effect of truck position and multiple truck loading on response of long-span metal culverts. Can Geotech J 2014; 51: 196–207.

15.

Elshimi

Moore

. Modeling the effects of backfilling and soil compaction beside shallow buried pipes. J Pipeline Syst Eng Pract 2013; 4: 04013004.

16.

Toh

Tan

Tse

, et al. Numerical evaluation of buried composite and steel pipe structures under the effects of gravity. Steel Compos Struct 2018; 26: 55–66.

17.

Yang

. Design and analysis of composite pipe joints under tensile loading. J Compos Mater 2000; 34: 332–349.

18.

Yang

Huang

Guan

. Stress model of composite pipe joints under bending. J Compos Mater 2002; 36: 1331–1348.

19.

Ladhwe

Kumar

Singh

, et al. Characterization of GFRP butt-joint under tensile and flexural loading. J Compos Mater 2015; 49: 2567–2578.

20.

BRYAN

. The investigations, design and construction of Paya Lebar Airport, Singapore. In: Proc Inst Civil Eng 1956; 5: 379–408.

21.

V-PILE. Soil investigation report. Kaki Bukit Road 4/ Bartley Road East. Singapore, 2011, pp.1–45.

22.

Ameron. Bondstrand 2000, 4000 and 7000 glass fiber reinforced epoxy (GRE) pipe systems for general service. The Netherlands: Author, 2008, pp.1–20.

23.

Fwa TF. The handbook of highway engineering. Florida, USA: CRC Press (Taylor & Francis), 2006, p.888.

24.

LTA. Land Transport Authority: standard details of road elements. Singapore: Author, 2015.

25.

Soden PD, Hinton MJ and Kaddour AS. Lamina properties, lay-up configurations and loading conditions for a range of fibre-reinforced composite laminates. Compos Sci Technol 1998; 58(7): 1011–1022.

26.

Sun XS, Tan VBC, Chen Y, et al. Stress analysis of multi-layered hollow anisotropic composite cylindrical structures using the homogenization method. Acta Mech 2014; 225(6): 1649–1672.

27.

ByeGC. Portland cement: composition, production and properties. 2nd ed. London, UK: Thomas Telford, 1999, p.225.

28.

Huang YH. Pavement analysis and design. 2nd ed. USA: Pearson/Prentice Hall, 2004, p.775.

29.

Bye GC. Grinding and fineness of cement. Portland cement: composition, production and properties. 2nd ed. London, UK: Thomas Telford, 1999, p.235.

30.

Robinson R and Thagesen B. Road engineering for development. 2nd ed. London, UK: Taylor & Francis, 2004, p.523.

31.

LTA. Land Transport Authority: civil design criteria for road and rail transit systems. Singapore: Author, 2010.

32.

Gillespie TD. Effects of heavy-vehicle characteristics on pavement response and performance. Washington, DC: Transportation Research Board, 1993.

33.

Firestone. Load and inflation tables, medium and light truck tire data book, pp. 59–67. http://octopup.org/img/car/tires/Firestone--2011--Load-and-Inflation-Tables--Medium-and-Commercial-Light-Truck-Tires.pdf (2011).

34.

Majumdar

Lesko

Cousins

, et al. Conformable tire patch loading for FRP composite bridge deck. J Compos Constr 2009; 13: 575–581.

35.

American Lifelines Alliance. Guidelines for the design of buried steel report, p.76. http://www.americanlifelinesalliance.com/.