Numerical modelling of surface subsidence induced by underground coal gasification

Abstract

Underground coal gasification (UCG) is an alternative method of energy extraction from coal. This method has significant advantages over traditional coal mining. Because the coal is combusted in situ, there is no need for underground labour and the environmentally unfriendly products of burning remain underground. The method can be applied to the coal seam of poor quality and deep underground where traditional mines would not be profitable. Although the method has been known for a century, there are only a few projects that exist on an industrial scale. One of the obstacles of UCG implementation is surface subsidence, which can damage infrastructure, UCG equipment and boreholes. To organize UCG in the most efficient way, the surface subsidence should be predicted. This work shows that none of the popular constitutive models can predict surface subsidence correctly. To demonstrate, investigate, and find the reasons of the incorrect predictions, the surface subsidence after an uncontrolled collapse of the traditional Longwall mine in Naburn, UK is modelled. The surface subsidence is assumed as a plain problem in the commercial finite-difference software FLAC3D by Itasca. The 2D problem is modelled in the 3D software to demonstrate that FLAC3D's results can be improved for two dimensions before extending the model to three dimensions. Before developing the model, the method of deriving elastic stiffness, friction angle, cohesion, and tensile strength from the boreholes description is developed and described in detail. FLAC3D's embedded Mohr-Coulomb, modified Hoek-Brown and strain softening constitutive models are implemented to model behaviour of the rock. The simulations of the collapse of the conventional mine indicates two possible reasons of the performance of the model, i.e. mesh density and constitutive models. It is noticed that the results depend on mesh density. The detailed mesh analysis is carried out to eliminate the first reason of poor model performance and to recommend some optimal mesh for modelling surface subsidence at a UCG station. To eliminate the second reason, the more advanced constitutive model is recommended. Historically, the double-yield model is utilized in the goaf; however, this model fails to predict the behaviour of the goaf. In an attempt to improve predictions, the built-in modified Cam-clay model, which employs the Critical State concept, is implemented. The model results are closer to the expectations. Since the Critical State model improves the result, CASM and the bubble model are programmed. Both of the models can replicate the modified Cam-clay model under certain conditions for their validation . The programming is started from the elastic, isotropic model, then the von Mises, Drucker-Prager, Tresca, and Mohr-Coulomb models are coded. After validation, CASM and the bubble model are implemented to simulate the clay overburden of the Shatsk UCG station in the Moscow basin. The UCG features, such as ash left in the UCG reactor, the complicated geometry of the UCG reactor, thermal stresses are also considered. The simulations show that CASM (Yu, 1998) and the modified Cam-clay predictions coincide. At the same time, the bubble model (Al-Tabbaa and Wood, 1989) results agree much better with the field measurement.

Date of Award	16 Nov 2019
Original language	English
Awarding Institution	Univerisity of Nottingham
Supervisor	James Walker (Supervisor), Dariusz Wanatowski (Supervisor) & Alec Marshall (Supervisor)