Pipelines that are used for the transport of energy and services are very important lifelines to modem society. Though pipelines are generally buried in unsaturated soils, the design guidelines are based on the assumption that the soil is either dry or fully saturated. For certain geotechnical problems, this assumption may not be acceptable because the water meniscus formed between soil particles creates an additional normal force between them by suction, which in turn forms temporary bonds. A recent series of large-scale physical model experiments at the Pipeline Engineering Research Laboratory (PERL) of Tokyo Gas, Japan show a higher peak load under unsaturated conditions compared to dry conditions. In contrast, recent experiments performed at Cornell University (CU) show that the soil-load due to lateral pipeline movement in dry and unsaturated sands are virtually the same. Thus, the effect of partial saturation on soil loading to pipeline may be different depending on soil type, moisture content and density. The current study investigates this problem through triaxial testing and constitutive modelling of the unsaturated soils used for the experiments and finite element simulations of the experiments. The mechanical behaviour of the sands used in the physical model experiments has been investigated by conducting a series of laboratory experiments. When compacted to the same energy level, Tokyo Gas sand exhibits larger strength in unsaturated conditions than in dry conditions at low confining stress levels mainly due to the suction-induced apparent cohesion generated by the fine particles present in the sand. In contrast, for coarser Cornell sand, the suction effect is found to be small even at low confining stress level, and hence the strength in unsaturated conditions is similar to that in dry ( or fully saturated) conditions. To capture the observed behaviour of dry as well as unsaturated soils, advanced constitutive soil models were developed. For dry (or fully saturated) soils, the modified Mohr-Coulomb and Original Nor-Sand (Cheong, 2006) models were able to simulate the general behaviour including the strain softening effect. To cater for the behaviour of unsaturated soils, the saturated versions of the NorSand and the modified Mohr-Coulomb models were modified in conjunction with the generalised effective stress framework. By simulating the triaxial experimental data, it is demonstrated that the developed models can predict the realistic soil behaviour of unsaturated soils. Using the developed models, the large scale physical model experiments of pipelines subjected to lateral soil movements at PERL and CU were simulated by the explicit finite element method. Good agreement was found between the numerical models and the experiments. Further FE analyses were conducted to investigate the pipeline behaviour under lateral soil movement at conditions of different HID's, moisture contents, and relative densities. The results were synthesized to produce new normalised pipe load charts. Three dimensional finite element analysis was performed to simulate the soil-pipeline interaction under strike-slip fault movements. The finite element model was first validated by comparing the computed results to the data produced from a full scale experiment carried out at CU. The analysis was then further extended by varying the initial conditions of the sand (sand type, density, moisture content, etc.), pipe material, pipe burial depth, and pipeline-fault rupture inclination. It was found in all cases that the peak lateral loads on the pipelines subjected to strike-slip fault movements are less than or equal to the peak loads computed by the 2-D lateral movement simulations.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:609026 |
| Date | January 2010 |
| Creators | Robert, Dilan Jeyachandran |
| Publisher | University of Cambridge |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://www.repository.cam.ac.uk/handle/1810/265508 |
Page generated in 0.0022 seconds