Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References




















Abstract
Figures
Introduction
Models
Workflow
Case study
Conclusions
References

Introduction

Tectonic stresses are relevant for hydrocarbon exploration and production for a variety of reasons. Paleo-stress fields were responsible, for example, for fracture formation and reactivation, whereas present-day tectonic stresses are of critical importance for borehole stability, orientation of hydraulically induced fractures, fluid flow anisotropies as well as the slip and dilation tendency of the existing faults and fractures, among others. Thus, a robust – ideally pre-drilling – prognosis of past and recent tectonic stresses in a reservoir is desired to optimize exploration and production, particularly of tight gas and fractured reservoirs.

Any reliable stress prognosis – either prior to the first exploration well or for the interwell space of a reservoir – is impeded by the fact that the orientation and magnitude of the stress field in reservoirs can be highly variable. Particularly near faults, within fault compartments and at lithological boundaries (e.g. salt structures) the local stress orientations, and hence the geometry and hydraulic properties of the fracture network can differ by up to 90° from the regional trend (e.g., Maerten et al., 2002; Yale, 2003). In such cases, inference of reservoir in-situ stress orientations and fracture geometries from regional-scale data compilations would inevitably lead to incorrect predictions. Any robust prognosis has to incorporate the specific reservoir geometry and the specific mechanical properties of the reservoir rocks. Regarding the complexity of real reservoirs, such a prediction can only be provided by a numerical modeling approach.

The goal of the present study is to demonstrate the potential of geomechanical reservoir models to predict tectonic stresses (past and recent) and the geometry of the natural fracture network. Modeling is based on finite element (FE) techniques honoring the structural and lithological complexity of the reservoir as much as possible. At first, the work flow and specific data requirements for a geomechanical analysis are discussed. To assess the practical value of the approach it is then applied to a data set from a hydrocarbon reservoir in the eastern Sirte Basin of Libya and model predictions are compared to actual field observations (stress and fracture orientations). This case study is carried out as a history match to mimic the progressive gain in information during the exploration and appraisal stage.

Static vs. Dynamic Geomechanical Models

Two different modelling approaches have to be distinguished: static and dynamic. Static models are based on the present-day reservoir geometry and the present-day ambient stress field. Modelling results provide a quantitative basis to predict the recent stress distribution within the reservoir, particularly the local perturbations near faults (e.g., Maerten et al., 2002; Henk, 2005). The corresponding stress tensor data can then be used to calculate, for example,  shear and normal stresses relative to the existing fault and fracture surfaces and infer the corresponding slip and dilation tendencies, respectively. In contrast, fracture formation and possible reactivation typically took place under stress conditions which were different from the present-day situation. Therefore, dynamic models have to account for the tectonic evolution of the reservoir and temporal changes in regional stress fields and/or reservoir geometry have to be incorporated. If the reservoir geometry has been modified substantially, the tectonic evolution can be divided into several modelling stages with different model geometries. In such cases, past reservoir geometries can be generated using geometrical balancing techniques, (e.g. Henk and Nemcok, 2008). Ideally, the reservoir evolution should be described by a continuous forward model.

Workflow

The workflow for building a geomechanical reservoir model is schematically depicted in Figure 1. Two different types of data are required: input data (reservoir geometry, material parameters, boundary conditions) and independent calibration data (in situ stress measurements, fractures from cores and logs) to compare observations with model predictions. Such calibration data usually is only available if wells have already been drilled. Further work steps include import of the subsurface geometry into the numerical simulation software, population of the model with lithology-specific material parameters and assignment of boundary conditions. During the subsequent calibration stage input parameters are modified iteratively within reasonable limits until a satisfactory fit between model calculations and field observations is achieved. This validated model can then be used to predict tectonic stresses and fracture characteristics in the undrilled parts of the reservoir.

The modeling approach used in this study utilizes the Finite Element (FE) technique and the commercial FE code ANSYS^® (Ansys Inc., Houston, USA), respectively. The model geometry is based on a boundary representation of faults and lithological horizons which can be constructed using data sets like fault maps and isopleth maps for the various lithological layers considered. The ideal data base utilizing the full power of the modelling approach, however, would be a reservoir model based on 3D seismic and geometrically consistent with all available data, e.g. a depth-converted Petrel^® project. Some manual editing is usually required to honor the specific needs of the FE technique, particularly the representation of the existing major faults. During subsequent discretization (meshing) the subsurface is subdivided into numerous tetrahedral and/or hexahedral elements (triangular and quadrangular in 2D). So-called contact elements are defined at opposing sides of existing faults. Fault friction coefficients can be assigned to the contact elements, which will slip if the shear strength described by the Mohr-Coulomb law is exceeded. Material properties are assigned to the elements representing the various lithologies. The FE models can describe elastic and plastic rock deformation. Mechanical behavior in the elastic domain is described by Hooke’s law, whereas plastic deformation by brittle failure is defined by the Mohr-Coulomb law using lithology-specific values for cohesion and angle of internal friction. If ductile rheologies like salt are involved, their plastic deformation can be approximated by temperature and/or strain rate-dependent creep laws. Finally, boundary conditions representing the regional stress field are assigned to the vertical faces of the model domain. No displacements are allowed along the bottom model boundary, while typically a lithostatic pressure representing the load of the overburden is applied to the top boundary.

Modeling results comprise, among others, the full stress tensor (direction and magnitude of the three principal stresses) for each part of the model, which can also be used to infer the slip and dilation tendency of faults and fractures. In addition, stress and strain information can be combined to predict fracture types and fracture orientations and to provide fracture intensity maps throughout the reservoir. These modeling results can be transferred back to the initial Petrel^® project for planning of well trajectories and frac treatments, for example.

Case Study

In order to assess the practical value of the geomechanical modeling approach outlined above, it is applied to a data set from a reservoir located in the eastern Sirte Basin of Libya. The subsurface geometry, including the main faults and lithological boundaries of top and base reservoir, is imported from a Petrel^® project. The corresponding FE model comprises a block with dimensions of 11.2 (W-E) x 9.3 (N-S) x 0.95 km. It consists of about 76,000 volume elements as well as 16,000 contact elements (Figure 2). Different mechanical material properties were assigned to the Maragh Formation, Upper and Middle Sarir Formation as well as the basement rocks. Boundary conditions for the static model were inferred from Ben- Suleman (2006) suggesting a regional NE-SW orientation of σ_Hmax. Information on the variable paleo-stress orientations since the Lower Cretaceous were derived from Amrose (2000) and large-scale palinspastic reconstructions. Of particular relevance for dynamic modelling and fracture prediction, respectively, is the Upper Cretaceous normal faulting/strike-slip faulting regime for which a NE-SW directed σ₃ is assumed. Model predictions are compared to data from three wells for which stress observations (direction of σ_Hmax) and fracture orientations from FMI logs were available.

Static modelling results indicate a rather uniform stress orientation parallel to the regional σ_Hmax throughout most of the reservoir. This is in accordance with stress observations at two of the calibration wells. Some local perturbations (up to 30°) are predicted for individual fault-controlled compartments and for the immediate vicinity of faults. The latter is actually observed in the third well and can be predicted with considerable accuracy if a corresponding fault model is used.

Dynamic modelling and pre-drilling fracture prediction, respectively, are less robust. This is partly due to uncertainties with respect to the exact tectonic regime in the Upper Cretaceous (normal faulting vs. strike-slip) and hence the orientation of the predicted conjugate shear fracture set differs by about 30°. In addition, the impact of paleo-pore pressures on the formation of tensile fracture is not known. Thus, some calibration wells are required to constrain the appropriate boundary conditions and choose the actual fracture orientations from the various options theoretically possible. In the present case study from the eastern Sirte Basin, this differs between individual fault blocks, thus following calibration a robust prediction may only be limited to the same fault compartment.

Conclusions

The case study shows that the general modelling concept can be applied successfully utilizing the data sets typically available during the exploration and appraisal stage. In particular, it illustrates that geomechanical modelling for stress and fracture prediction can be fully integrated into a corresponding Petrel^® project of the reservoir. The case study indicates that a robust prediction of the stress field, including its local perturbations near faults, can be based primarily on the reservoir geometry with only sparse well control. Fracture prediction is more complex and definitely requires well data for calibration as the model has to use several poorly constrained parameters like the magnitude of the paleo-stresses and the paleo-pore pressures to infer fracture orientations. Once the geomechanical models are calibrated they provide a valuable tool for a variety of tasks like optimizing well trajectories (borehole stability), pre-drilling planning of hydraulic fracture treatments, e.g. multiple fracs in horizontal wells, as well as positioning of wells in zones of enhanced fracture intensity.

References

Ambrose, G., 2000, The geology and hydrocarbon habitat of the Sarir Sandstone, SE Sirte Basin, Libya: Journal of Petroleum Geology, v. 23, p. 165-192.

Ben-Suleman, A., 2006, Active tectonics and related stress fields of northern Libya: Geophysical Research Abstracts, 8, 08954, SRef-ID: 1607-7962/gra/EGU06-A-08954.

Henk, A., 2005, Pre-drilling prediction of the tectonic stress field with geomechanical models: First Break, v. 23, no. 11, p. 53-57.

Henk, A. and M. Nemcok, 2008, Stress and fracture prediction in inverted half-graben structures: Journal of Structural Geology, v. 30, no. 1, p. 81-97.

Maerten, L., P. Gillespie, and D.D. Pollard, 2002, Effects of local stress perturbations on secondary fault development: Journal of Structural Geology, v. 24, no. 1, p. 145-153.

Yale, D.P., 2003, Fault and stress magnitude controls on variations in the orientation of in situ stress, in M. Ameen, (ed.), Fracture and In-Situ Stress Characterization of Hydrocarbon Reservoirs: Geological Society, London, Special Publications 209, p. 55-64.

Return to top.

Copyright © AAPG. Serial rights given by author. For all other rights contact author directly.