A fundamental investigation and ultrasonic characterisation of coal effective stress behaviour

Coal seam gas is an important energy resource worldwide. The methane extraction from coal seam is often accompanied by CO2 injection to enhance the gas recovery and reduce the greenhouse gas emission footprint. Amongst all processes active in coal seam gas extraction, understanding and characterizing the coal effective stress and its evolution with time need special attention. The reservoir characteristics of coal seam that control its effective stress evolution, however, differ from that of other hydrocarbon resources in granular sediments. In conventional reservoirs, it is the pore volume determining the gas storage, but in coal seams, it is the pore surface area that defines the capacity through adsorption. As naturally fractured reservoirs, coal seams often contain an extensive fracture network called cleats thus, the coupled relationship between the mechanical behaviour and sorption effect can significantly influence the effective stress and in turn permeability. The main aim of this dissertation is thus to investigate and model the interaction of effective stress evolution and adsorption/desorption processes using advanced and innovative laboratory experiments and numerical simulations. This dissertation includes four parts consisting of the modelling and fundamental studies employing novel experimental and numerical techniques. The first part of the dissertation seeks to understand the characteristics of sorption-hydromechanical behaviour through the microscale and non-destructive investigation using micro-computed tomography (μCT), ultrasonic and their combined measurements. These fundamental investigations shed light on the coupled physical processes in different coal samples extracted from the Sydney Basin Australia through tracking 3D internal geometry. Using an X-ray transparent triaxial system, a range of stress-pore pressure boundary conditions are applied on different coals to obtain the 3D internal structure. The different coal components and their fracture patterns are analysed with respect to the bulk, matrix, and fracture compressibilities. In a further step, the coupled stress and swelling strain responses of coal when exposed to carbon dioxide (CO2) and helium (He) are studied by imbedding ultrasonic sensors in the X-ray transparent triaxial system. With the real-time visualisation of fracture porosity due to different adsorption levels, the effects of CO2 adsorption and involved processes on ultrasonic responses are investigated. The combined physical and numerical methods determine the main factor influencing wave propagation in coal to i) assist developing the representative constitutive model and ii) be used in the acoustic-driven parameterisation in the final part of the dissertation. The micro-scale investigation highlights the importance of internal structure affecting the mechanical properties in coal and the acoustic wave velocity allows evaluating the changes in facture characteristics during CO2 adsorption. Using detailed understanding of the physical processes involved in coal multiphysics, a constitutive model based on the continuum mechanics and non-equilibrium thermodynamics is developed in the second part of the dissertation. The model considers changes in gas content in the tight coal matrix through sorption and diffusion processes along with gas leakage from the matrix into fractures where Darcy type flow takes place. Also, the time dependency of coupling processes is accounted for especially the volumetric strains induced by gas sorption and their overall effects on changes in the fracture aperture, hence in the bulk flow conductivity. The novelty of the proposed model specially lies in the derivation of the thermodynamically consistent formulation of time-dependent effective stress law. The next part of the dissertation seeks to validate the developed effective stress law experimentally and to investigate i) the performance of commonly used theoretical models for swelling stress estimation, ii) the validity of thermodynamics coupling coefficient defining the swelling stress and iii) the effect of external stress on coal volumetric strain response. Especially designed experiments on two coal samples are conducted, including time-dependent diffusion and volumetric strain experiments under various stress and CO2 pressure conditions. A new experimental method is proposed to characterize the key input of the model which is the swelling coupling coefficient. Results of these series of experiments also show that the stress induced compression has minor effect on gas desorption. Since the wave velocity and porosity are interrelated, in the last part of the dissertation, some key parameters involved in the set of developed hydromechanical relationships are measured/modelled using acoustic measurement and finite element simulation. First, the effective stress coefficient is predicted using the percolation theory and hydromechanical and ultrasonic laboratory measurements on coal samples. The swelling coefficient representing adsorption induced volumetric strain development is next studied using acoustic simulation. As a newly proposed coupling coefficient in the model development, the relationship between the coefficient and wave velocity are correlated in three pore pressure conditions and its response to each condition is collected and analysed. Finally, the fracture permeability in coal seams is estimated using a novel physics-informed neural network (PINN) technique. In the training of PINN model, a synthetic dataset is built from several ultrasonic measurements and numerical simulations, with input variables of wave velocity and density. This model is successfully applied in a field case study where downhole geophysical logging data is available. In general, the acoustic-driven technique provides a strong and useful pathway to predict model parameters using geophysical logging data in a field setting, where sonic logs are available