Abstract

The deformation processes of hydrate sediments during methane gas production are complex due to the coupled thermo-hydro-mechanical (THM) phenomena of deformation, fluid flow, heat flow and phase changes. From a geotechnical perspective, several potential geohazards may result from gas production: wellbore instability; damages in well casing or supporting well infrastructure; submarine landslide; seafloor instability; and seafloor subsidence. In order to investigate the mechanisms of these geohazards occurring, a cost effective method is to have numerical simulations of the hydrate gas production. Hence, it is essential to have accurate numerical modelling of the geomechanical behaviour of the hydrate-bearing sediments.

The Cambridge methane hydrates geomechanics simulator (CMHGC) is a 2D thermo-hydro-mechanical coupled finite difference code developed at Cambridge using FLAC2D . Within this code, a methane hydrate critical state (MHCS) model was developed in order to capture the mechanical behavior of hydrate-bearing sediments based on the Modified Cam-Clay model. The developed MHCS model assumes the soil materials within one element are homogenous and present an isotropic response. For real site data, we need to consider the site turbidite soil formation as well as the hydrate saturation heterogeneity within each element. Due to the limitations of the MHCS model, an anisotropic MHCS model was developed and implemented into ABAQUS UMAT . This developed Anisotropic-MHCS model can better describe the anisotropic feature of the turbidite formation, which includes the fabric evolving process during the hydrate dissociation. This evolving process represents the material changing from anisotropic heterogeneous system into an isotropic homogeneous system.

By applying the AMHCS model, an established parameter calibration method was also developed to determine the upscaled model parameters. To assess the performance of this upscaeld model, the geomechnical behaviors of hydrate bearing sediments from upscaled model simulation were investigated and compared to the results of simulations with previous model.