Dempsey, DArcher, RRiffault, Jeremy2019-06-202019http://hdl.handle.net/2292/47264This thesis discusses the construction of numerical models to gain insight on Enhanced Geothermal System (EGS) stimulation operations. These operations consist of high pressure injection of cold water in low permeability formations at depth of 2-4 kilometres, promoting the opening of existing fractures and creating an artificial geothermal reservoir. Typically a large seismicity cloud is generated. The physical mechanisms involved in permeability enhancement and seismicity triggering are numerous, and their relative importance is not always agreed on, which makes designing such operations challenging. The extent of the stimulated volume, that underwent permeability enhancement is the principal metric of the success of an EGS stimulation operation, and is not measurable, or estimated only through the size of the seismically active volume. We address those two issues using a novel modelling approach. While a simple physics-based permeability model is explored, most of our efforts are directed to the development of an empirical permeability inversion method. We demonstrate that seismicity density can be used as a proxy for pressure increase. The pressure distribution itself is a function of the hydrological properties evolution, permeability and porosity, with space and time in the course of the stimulation. We create an inverse modelling approach in which permeability enhancement candidate scenarios are proposed until the modelled seismicity density distribution matches the observations. These permeability scenarios have no physical motivation, and thus do not embed any assumptions or bias on the physical mechanisms responsible for stimulation. Stimulation parameters, injection rate and pressure, are also considered in the inversion. Special attention is given to the impact of parameter and observation uncertainty, as well as structural error, on the robustness of our conclusions. We consider two Australian Enhanced Geothermal System projects, Habanero and Paralana, where stimulations were conducted in 2003 and 2011, respectively. Large planar clouds of micro-seismicity were recorded in both cases, with 10,436 and 4,753 events recorded above magnitude of completeness for Habanero and Paralana respectively. A one-dimensional radial geometry version of our inversion method is applied to the Paralana dataset. The recovered permeability regime implies that the bulk of permeability enhancement, up to 30x its initial values, occurred within a 60 m radius of the wellbore and was not coincident with the bulk of the seismicity, which extends more than 400 m away from the injection point. Thus, permeability enhancement and induced seismicity are decoupled in this case, contrary to the traditional assumption in EGS modelling. Hydroshearing seemed to have not been responsible for much permeability increase. The two-dimensional version of the inversion is applied to the Habanero dataset. The solution range is less constrained than for the radial inversion, but still provides some insight on the stimulation process. Large scale permeability enhancement (> x10) is limited to volumes in the vicinity of the wellbore, representing approximately 5 % of the seismically active volume. This method, which already provides valuable insight on stimulation success, could be improved and be of great help to the future of EGS projects.Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated. Previously published items are made available in accordance with the copyright policy of the publisher.https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htmhttp://creativecommons.org/licenses/by-nc-sa/3.0/nz/ThesisCopyright: The authorhttp://purl.org/eprint/accessRights/OpenAccessQ112552657