Accueil > EGW 2018 : 6th European Geothermal Workshop > Abstracts > Session 2 : Reservoir Engineering > Session 2 : Oral Presentations

Session 2 : Oral Presentations

KEYNOTE

Flexible parallel implicit modelling of coupled thermal–hydraulic–mechanical processes in fractured rocks

Antoine B. Jacquey, Mauro Cacace, Guido Blöcher, Ernst Huenges, Magdalena Scheck-Wenderoth

PDF - 374.9 ko
Jacquey et al.

In this study, we present an overview of recent and ongoing efforts to develop a robust, yet efficient multi-physics and multi-component porous media modelling framework applicable to reservoir applications. We rely on numerical approaches to characterize interactions among thermal, hydraulic, mechanical, and ultimately chemical processes across relevant time and length scales of interest to applications including extraction of geothermal heat and fossil energy as well as storage of water, carbon dioxide, nuclear waste and thermal energy.

Based on the MOOSE (Multiphysics Object Oriented Simulation Environment, Gaston et al., 2009), we have developed a numerical simulator called GOLEM (Cacace and Jacquey, 2017, Jacquey and Cacace, 2017) to solve, in a tightly implicit manner the governing equations for groundwater flow, heat and non-reactive mass transport by including poro- and thermo-elastic as well as inelastic (viscous and plastic) deformation processes and their non-linear feedbacks. Equations of State (EOS) for relevant fluid properties (density and viscosity) as based on the latest IAPWS (International Association for the Properties of Water and Steam, 2008) release as well as structure-property (poro-perm) relations are also considered to close the systems of equations. The simulator is interfaced to an in-house developed geometric/meshing software (MeshIt, Cacace and Blöcher, 2015) which enables the integration of details of the three-dimensional geological architecture of real case reservoirs into a consistent Finite Element/Volume representation. Interface FEM elements are used to represent all components, that is, 3D reservoirs units, 2D faults and fractures, 1D boreholes and 0D localized sources. The resulting equations are homogenized relying on the concept of effective (hydro- mechanical) aperture for the lower dimensional elements (i.e. fractures and wells) which therefore ensure local mass and energy conservations as well as continuity of the problem variables (pore pressure, temperature and matrix displacements) across all element interfaces (Jacquey et al., 2017).

The flexibility of the software for the (up)scale of reservoirs is discussed by presenting results obtained for the Groß Schönebeck geothermal facility, Germany (see Figure 1). In a first study case, THM simulations of the reservoir behaviour during injection and production are discussed. The aim of this study is to quantify how existing fault zones and induced fractures affect the overall productivity and long-term sustainability of the system. The second study case investigates the impacts of a hydraulic stimulation treatment and production of geothermal fluid on the stability of the fault systems present at depth within the reservoir. Finally, we will present the impacts of a stimulation treatment onto the far field hydraulic of the reservoir by means of poroelastic coupling.

The second part opens a discussion on ongoing studies, in which we aim at improving the understanding of the processes responsible for micro-cracking and strain localisation. The model formulation will be briefly presented. It considers a damage rheology to account for the organization of micro defects and appropriate elasto-viscoplastic constitutive equations to simulate the hardening and softening of the elastic stiffness of the porous rock. Localization of the deformation is modelled by introducing a viscoplastic component, non-linearly dependent on the rate and amount of damage accumulated throughout the loading history of the rock. Hydro-mechanical coupling is also integrated via appropriate porosity and permeability dependencies as constrained by laboratory triaxial tests under controlled p-T conditions combined with microstructural analysis of the post-mortem samples. We will end the contribution by addressing planned activities to extent the current model formulation to non-isothermal loading conditions.

Systematically changes in MT signal during deep drilling operations

Nadine Haaf, Eva Schill, Ragna Karlsdottir, Knutur Arnason

PDF - 411.8 ko
Haaf et al.

The Horizon 2020 project “Deployment of deep enhanced geothermal systems for sustainable energy business (DEEPEGS)” aims at demonstrating advanced engineering technologies in geothermal reservoirs under different geological conditions in Iceland and France. The concept of developing a deep EGS well at Reykjanes comprises injection of fluid underneath the conventional geothermal field to support production. Therefore, the 2,500 m deep RN-15 production well was deepened to 4,659 m depth in the framework of the Icelandic Deep Drilling Program (IDDP-2) from August 2016 until January 2017. The drilling progress was accompanied by partial and up to total circulation loss. Below 3,200 m total circulation loss indicates a highly permeable zone and led to total lack of cuttings. The Iceland Deep Drilling Project 4.5 km deep well, IDDP-2, in the seawater-recharged Reykjanes geothermal field in SW Iceland has successfully reached its supercritical target (Friðleifsson et al, 2017).

Contemporaneously to the drilling, continuous magnetotelluric (MT) monitoring was carried out in order to reveal information on the directional development of the reservoir and the evolution of preferential hydraulic connectivity at comparably high flow rates.

Two continuous running MT stations, GUN and RAH, were installed on the Reykjanes peninsula. RAH and GUN are located about 6 and 1 km away from IDDP-2. Both MT stations are equipped with two electric dipoles in N-S and E-W direction, as well as three magnetic sensors oriented in N, E and vertical direction. Magnetotelluric monitoring was carried out between December 2016 and July 2017 with a sampling frequency of 512 Hz. The processing of the first data from the late drilling phase revealed the bad data quality of RAH hence it was stopped in May 2017. Consequently, MT data were processed using single site method with the code Bounded Influence Remote Reference Processing (Chave and Thomson, 2003). Due to a temporally noise signal in the time series they were down filtered to get the lower bands and hence to clean the time series.

First results from the late drilling period reveal changes in the resistivity distribution over time. Prominent drops in electric resistivity are observed in a time period of about 24-48 hours before "major" seismic events (ML>1.0 or frequency >10/d) occurred. They occur at frequencies of 0.25-5 Hz. A second frequency range (0.2-0.125 Hz) of transient decrease in electric resistivity correlates with time periods of large losses of circulation fluids. In general, most of the changes are observable in the YX-component of the transfer functions suggesting a directional development.

However, for final interpretation, different possible sources of the signal changes will be investigated such as comparison with the variation of the Earth’s magnetic field and ongoing operations at the drilling site.

Using the Pressure While Drilling Data to inform drilling decisions in the Kawerau and Rotokawa field, NZ

Morgane Le Brun, Lutfhie Azwar

PDF - 526.9 ko
Le Brun and Azwar

During the 2016-2017 Mercury Ltd drilling campaign, one injection well and three production wells were completed at both Kawerau (KA55 and KA56) and Rotokawa (RK35 and RK36) geothermal fields in New-Zealand. This drilling campaign started with a depth target and an injection/production capacity target for each well. A regular update on the capacity of the well once drilling in the reservoir section was one key parameter to decide when to stop drilling and complete the well (i.e. decision to TD the well).

To improve drilling efficiency, the number of stage tests was reduced and a methodology was defined to calculate the wells capacity using the data from the Pressure While Drilling tool (PWD). This tool measures annular and pipe pressures using sensors within the Bottom Hole Assembly (BHA), these data give information in real time about the permeability and stability of the wellbore. The annular pressure data, combined with the fluid losses calculated while drilling the reservoir interval, were used to estimate the capacity of the well in real-time. This information about the capacity of the well enabled the TD decision to be based on more information and with less waiting time on data.

The PWD data were analyzed in a two-steps process to estimate the well capacity while drilling. The first step was to estimate the injectivity index (II) of the well with the PWD data and the fluid losses in the wellbore, calculated using the flow in and out of the well recorded every 30 seconds while drilling. One method used the PWD data and reservoir pressure scenarios to calculate the II versus depth while drilling. The figure below shows the PWD data in blue, the II in orange calculated with the cold reservoir pressure (dotted line).

The other method didn’t rely on the assumption on the reservoir pressure, it used the change in pressure recorded by the PWD tool at a particular depth when the injection rate in the well was changed for drilling purposes (e.g wiper trips, trip out). The comparison of the values obtained with these two methods helped better constrain the II of the well and reduce the need for stage tests. For the production wells, this II was used to estimate the Productivity Index (PI) by using a factor representing the expected decrease in permeability when producing the well, due to the reheating of the formation and change of fluid phase.

The second step was to translate the II or PI data into well capacity using an in-house wellbore modelling tool. This modelling enabled the simulation of possible scenarios of well capacity under operational conditions. For the injection wells, the II in t/h/bar obtained with the PWD data was changed into a formation II in m3 to take into account the difference in viscosity between the cold river water injected while drilling and the hotter fluid that will be injected during the operation of the well. This II along with the well completion were used in the wellbore model to generate an injection curve and thus the expected capacity of the well. For the production well, the PI along with different reservoir pressure and fluid enthalpy scenarios were used in the wellbore model to obtain optimistic and pessimistic output curves and thus get a range of production capacity under operational conditions. This well capacity range was very valuable for determining well success and decision making during drilling (e.g., TD or side-track).

The use of the PWD data for well capacity estimation during this campaign highlighted two main areas of improvement for the future make-up wells. One area is the underestimation of the II with the PWD data compared to the completion test II. This underestimation lead to a conservative evaluation of the well capacity, understanding the source of this difference would improve the quality of information provided during the drilling of the well. The other area of improvement is the calculation of losses for the II calculation under partial losses conditions. In this drilling campaign, a paddle sensor was used to estimate the flow out of the well while drilling. The calibration of this paddle to obtain the flowrate was not always accurate, which increased the uncertainty on the II calculation. The flow out was checked with data from the mudloggers but the resolution of the losses was not as precise as the resolution of the PWD data. Partial losses conditions were generally linked to II value that were lower than the targeted II so changing the paddle sensor for a more accurate measurement with a flowmeter was not expected to change the decisions made during drilling.

This paper describes the results from the analysis and modelling from the PWD data collected in the different wells, and the benefits and limitations in the usage of the current PWD technology to inform the continuation of drilling during the drilling operation.

Experimental observations and constitutive description of the effect of compaction banding on the hydraulic properties of porous rock

Julia Leuthold, Theodoros Triantafyllidis

PDF - 302.2 ko
Leuthold et al.

The aim of the present work is the investigation, both from an experimental and a constitutive point of view, of the effect of localization of deformation on the hydraulic properties of rock, with special focus placed on the formation of compaction bands. The localization of deformation into thin bands of a given orientation is not only a source of instability, but also a source of anisotropy as far as both the mechanical and the hydraulic properties of the rock are concerned. The available evidence, both in situ and in the laboratory, is limited and, to some extent, contradictory. Baxevanis et al. [1] performed laboratory tests on Tuffeau de Maastricht calcarenite and observed permeability reactions that, though not in agreement with the Kozeny – Carman model [2], [3], were smaller than an order of magnitude. Holcomb and Olsson [4] on the other hand registered permeability variations of up to two orders of magnitude as a result of the compaction of Castlegate sandstone with an initial porosity of 28%. Ballas et al. [5] measured the permeability of naturally occurring compactive shear bands in porous sandstone in Provence. Cataclastic shear zones were found to possess a permeability that was three to five orders of magnitude smaller than that of the host rock.

From a practical point of view, such behavior is of particular interest when considering reservoirs in soft, porous rocks. The reduction in pore pressure, which is linked to the production, leads to the possibility of compaction in the vicinity of the borehole. One obvious effect is the risk of the loss of stability or of increased sand production. Another is the reduction of the permeability locally. The probability of such occurrences and the magnitude of such effects is currently under debate.

Triaxial and oedometric tests on a soft calcareous sandstone were performed for the needs of the present work, both under dry and under saturated conditions. In the second case, the permeability and its variation as a function of the axial strain was also measured. On the basis of these results and results from the literature, it is examined whether the Kozeny - Carman formulation may be applied in this case. Its limits of validity are discussed and an alternative is suggested, taking into account the volumetric collapse and phenomena taking place in the microscopic scale, such as grain crushing.

From a numerical point of view, a nonlocal model is suggested to simulate the formation of compaction bands and validated against the experimental results. The effects on the evaluated permeability are numerically calculated and the resulting anisotropy of the hydraulic properties is assessed.

Finally, conclusions are drawn and implications on reservoir compartmentalization are discussed.

Modelling of thermal shock induced fracture growth

Ivar Stefansson, Adriana Paluszny, Inga Berre, Eirik Keilegavlen

PDF - 441.3 ko
Stefansson et al.

The properties and characteristics of geothermal reservoirs are highly influenced by the presence of fractures at various scales. The fractures may be pre-existing or, as in the case of enhanced geothermal systems, induced by various energy production related processes, such as pressure elevation and cooling due to fluid injection.

Modelling and numerical simulation of fracture nucleation and propagation offers a valuable supplement to experimental studies and field observations, both in predicting behaviour and in furthering the understanding of the physical processes involved. This is a highly challenging task, not least due to the constantly changing, possibly highly complex fracture network geometry. This aspect requires deliberation in the model design, and trade-offs between accuracy and computational cost are inevitable. In this work, the finite element (FE) method is used, with re-meshing whenever the fracture geometry changes. This approach allows capturing complex geometries at moderate computational costs compared to alternatives such as discrete element methods, and the extended FE method.

If the mechanical deformation and fracturing is influenced by other processes such as fluid pressure or thermal non-equilibrium inducing thermal stresses, the complexity of the modelling is further increased by the need to couple the models involved. Thermo-mechanical (TM) phenomena have been extensively investigated experimentally, e.g. through thermal shock testing of ceramics. One classical experiment considers a pre-heated thin ceramic plate which is rapidly cooled by water immersion. If the thermal shock is sufficiently large, fractures form at the specimen’s interface with the coolant. Typically, the fractures are dyed to allow detection of their geometry and length and spacing statistics.

The thermal shock experiment may lead to extensive and complex fracture formation, while the set-up and material parameters may be tightly controlled. This makes it a favourable case for validation studies on numerical simulation tools, as has been done in 2d using FEM, e.g. by Li et al. (2013), and more recently also in 3d for DEM methods based on peridynamics theory, see Bourdin et al. (2014) and D’Antuono and Morandini (2017). Following this tradition, we here validate the TM part of a FE based model in which the fractures are represented by 2d surfaces in the surrounding 3d rock matrix. Furthermore, we investigate applicability to geothermal processes by considering relevant temperatures and rock properties. We investigate which ranges of heat shock magnitude caused by fluid injection may cause fracturing in the near-well region.

The implementation is part of the Imperial College Geomechanics toolkit (Paluszny and Zimmerman, 2011), and more details on the implementation of the TM model are found in Salimzadeh et al. (2018).

Multi-fracture propagation in porous medium under shear stimulation of fluid flow

Hau Trung Dang, Inga Berre, Eirik Keilegavlen

PDF - 433.6 ko
Trung Dang et al.

We are interested in low-pressure stimulation of naturally fractured porous media, as applied e.g. in enhanced geothermal systems. Naturally occurring fractures are considered joint surfaces that can withstand tectonic stresses due to friction by asperities in the fracture walls. Hydraulic stimulation by the injection of fluids in the fracture network can overcome the frictional resistance, leading to shear deformation along the fracture surfaces. The shearing is associated with a dilation of the fracture aperture and increased permeability.

Here we study how the shear deformation will alter the local stress fields, and can potentially trigger fracture propagation. The porous medium and fluid flow are modelled by the discrete fracture-matrices model that the fractures are treated as interfaces and allows the fluid transport from high-permeable conductive fractures to the rock matrix and vice versa. We treat the rock matrix as a linearly elastic medium, while the nonlinear shear behavior of fractures is represented by the Barton–Bandis joint model. The governing equations are discretized by a combination of finite element and finite volume methods. We discuss the numerical modeling of fracture propagation in this setting, with an emphasis on representation of the new fracture in the computational grid. Numerical examples shows the interaction between fluid pressure and mechanical forces.

3 octobre 2018