Analysis of induced seismicity and heat transfer in geothermal reservoirs by coupled simulation

Open Access
Gan, Quan
Graduate Program:
Energy and Mineral Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
July 14, 2015
Committee Members:
  • Derek Elsworth, Dissertation Advisor
  • Jamal Rostami, Committee Member
  • Shimin Liu, Committee Member
  • Tong Qiu, Special Member
  • Geothermal reservoir
  • thermal front
  • induced seismicity
  • permeabiltiy evolution
  • breakdown pressure
Massive water injection during the stimulation and production stages in geothermal reservoirs is able to induce strong coupled response. Fracture properties may be significantly altered due to fluid transport, heat transfer, and chemical reactions. The constitutive relationships governing the transport of mass, heat, deformation, and chemical reaction are defined by various parameters, including permeability, porosity, temperature, and local stress state. One of the purposes of this study is to investigate the principal factors including the thermal, hydraulic, and mechanical effects in inducing the reactivation of pre-existing fractures. The seismicity induced from the reactivation of fractures and faults is influenced and triggered by different processes at different length and timescales. We explore these effects related to the timing of seismicity and the roles or effective stresses and thermal stresses in fractured reservoirs including the use of reactivation to control permeability evolution in a manner beneficial to thermal recovery in EGS reservoirs. This dissertation comprises four chapters and an appendix. In Chapter 1 we explore the propagation of fluid pressures and thermal stresses in a prototypical geothermal reservoir containing a centrally-located critically-stressed fault from a doublet injector and withdrawal well to define the likelihood, timing and magnitude of events triggered by both fluid pressures and thermal stresses. We define two bounding modes of fluid production from the reservoir. For injection at a given temperature, these bounding modes relate to either low- or high-relative flow rates. At low relative dimensionless flow rates the pressure pulse travels slowly, the pressure-driven changes in effective stress are muted, but thermal drawdown propagates through the reservoir as a distinct front. This results in the lowest likelihood of pressure-triggered events but the largest likelihood of late-stage thermally-triggered events. Conversely, at high relative non-dimensional flow rates the propagating pressure pulse is larger and migrates more quickly through the reservoir but the thermal drawdown is uniform across the reservoir and iv without the presence of a distinct thermal front, and less capable of triggering late-stage seismicity. In Chapter 2 we develop a dimensionless model to predict the thermal drawdown response, and quantify the relationship between the timing and magnitude of late stage seismic event and the induced thermal stress from thermal drawdown. We evaluate the uniformity of thermal drawdown as a function of a dimensionless flow rate D Q that scales with fracture spacing s (m), injection rate q (kg/s), and the distance between the injector and the target point L ( 2 / D Q qs L  ). By assuming the dominant heat transfer by heat conduction within the fractured medium, this model is either capable to predict the timing of induced seismicity by the thermal stress by the analytical formula. Due to the significant influence of fracture network geometry in heat transfer and induced seismicity, a discrete fracture network model is developed (Chapter 3) to couple stress and fluid flow in a discontinuous fractured mass represented as a continuum by coupling the continuum simulator TF_FLAC3D with cell-by-cell discontinuum laws for deformation and flow. Both equivalent medium crack and permeability tensor approaches are employed to characterize preexisting discrete fractures. The evolution of fracture permeability accommodates stress-dependent aperture under different stress states, including normal closure, shear dilation, and for fracture walls out of contact under tensile loading. This discrete fracture network model is applied (Chapter 4) in a generic reservoir with an initial permeability in the range of 17 10 to 16 10 m2, fracture density of ~0.09m-1 and fractures oriented such that either none, one, or both sets of fractures are critically stressed. For a given reservoir with a pre-existing fracture network, two parallel manifolds are stimulated that are analogous to horizontal wells that allow a uniform sweep of fluids between the zones. The enhanced connectivity that develops between the production zone and the injection zone significantly v enhances the heat sweep efficiency, while simultaneously increasing the fluid flux rate at the production well. For a 10m deep section of reservoir the resulting electric power production reaches a maximum of 14.5 MWe and is maintained over 10 years yielding cumulative energy recoveries that are a factor of 1.9 higher than for standard stimulation. Sensitivity analyses for varied fracture orientations and stimulation directions reveal that the direction of such manifolds used in the stimulation should be aligned closely with the orientation of the major principal stress, in order to create the maximum connectivity. In the appendix, we explore a new mechanism in determining the breakdown pressure. Experiments on finite-length boreholes indicate that the breakdown pressure is a strong function of fracturing fluid composition and state as well. The reasons for this behavior are explored including the roles of different fluid types and state in controlling the breakdown process. The interfacial tension of the fracturing fluid is shown to control whether fluid invades pore space at the borehole wall and this in turn changes the local stress regime hence breakdown pressure. Interfacial tension is modulated by fluid state, as sub- or supercritical, and thus gas type and state influence the breakdown pressure.