Improved Simulation of Fractured Reservoirs


Project Summary 

This ITF project is concerned with the development of new methods for improved simulation and characterisation of fractured and faulted reservoirs. The state-of-the-art finite-element code CSP is being optimized for the fast simulation of multi-phase flow in complex sector-scale reservoir models with discrete fracture representations. Novel techniques for generating unstructured grids developed in collaboration with ICEM Technologies Inc. and for solving the governing transport equations using an implicit-for-pressure and implicit-for-saturation approach within CSP are being applied.

For field-scale simulations, a novel method of fractured-reservoir recovery is being researched and implemented as a commercially available streamline-based reservoir simulator: Geologically realistic models of a range of common fault- and fracture geometries will serve to compute appropriate recovery-upscaling parameters. The results of field-scale streamline- and grid-block scale finite-element-finite-volume computations will be compared to conventional simulation results to validate the approach.


Motivation & Goals

Many fields are extensively fractured or contain faults that act as significant barriers or conduits to flow. Primary depletion below the bubble point typically leads to excessive gas production through the fractures with very little production from the matrix and recoveries typically less than 10%. Gas and water injection schemes are problematic, since the injected fluids may simply channel through the fractures without sweeping the matrix. On the other hand, the fracture network could be used to deliver injected fluids rapidly throughout the reservoir, while gravity or capillary forces could induce flow into the matrix, recovering oil quickly and efficiently.

fem-fv FEM-FV model of discrete interconnected fractures with central production well in planview

The success of such projects depends on the well placement, the connectivity of the fracture network, the fluid properties and the rock wettability. The range of possible recoveries is exceptionally broad. This is in contrast to unfractured reservoirs where even without simulation studies, recovery can normally be estimated to within a 5 - 10% window.

Present reservoir simulation studies of fractured and faulted reservoirs suffer from a number of limitations, meaning that there is little confidence in their predictions of oil recovery. Since the range of possible recoveries is so wide, improved oil recovery projects, particularly in mature and small accumulations, are very risky.

This project will develop novel and improved methods for simulating flow in fractured and faulted reservoirs. The result will be better predictions of oil recovery with less uncertainty. This will allow gas and/or water injection projects to be designed with confidence, resulting in better recoveries from fractured reservoirs. The results will be delivered to the industry as a wide variety of deliverables: A pioneered object-oriented software design concept for FEM-FV fractured reservoir simulation documented in the unified modeling language (UML), the knowledge about the suitability of public-domain algorithms for the specific reservoir simulation problem, the novel reservoir discretization tools of ICEM CFD Technologies Inc., the commercial code of Streamsim Technologies Inc., and the rigorously determined mathematical model for recovery upscaling for use in current conventional reservoir simulators.

To develop a reliable and accurate characterisation of flow in fractured and faulted systems, simulations will be performed on geologically realistic discrete fracture systems in a relevant stress state and on models capturing a range of fault geometries and flow properties. These simulations need to account for the complex geometry of the fractures and faults and to incorporate all the relevant physics - in particular the transfer of fluids between fracture and matrix mediated by capillary pressure. This means to face the challlenge of generating discretizations and conducting ultra-high resolution simulations (>10^7 finite elements), solving the transport problem in the presence of extremely finely resolved mesh domains. This will be achieved using two complimentary simulation tools:

The finite-element code CSP and streamline-based simulator. The CSP code will use ICEM-CFD's unstructured hybride FE meshes to represent topologically complex systems with finite-element sizes ranging from millimetres (for the fractures) to several metres for the matrix. These small- to intermediate scale simulations (models <500 m) will incorporate multiphase flow including capillary effects and the compaction and elastic stress state of the rock and fracture system.

The streamline code will handle field-scale corner-point geometry grids and non-neighbour connections, and it will use a new mathematical model for transfer between matrix and fractures derived from the quantitative insights gained from the discrete fracture flow simulations.

After two years, the initial investigation of the relevant grid-block scale processes in fractured reservoirs will be completed and simulation algorithms using the new mathematical formulation for recovery upscaling for field-scale modeling will have been developed and tested.

Streamline-based simulation has the advantage of being very fast and accurate in modelling advective displacement for systems which are in a steady pressure state (Batycky et al. 1997). However, current models cannot so efficiently model capillary pressure effects and have a fixed resolution (=grid block size). Their extension to unstructured grids appears possible (Prevost et al. 2001), but it is unclear how this approach could be applied to compacting reservoirs or transient fluid flows.

The CSP code is not subject of these limitations, but transport processes are more costly to compute than with the streamline approach, because time stepping is limited by a Courant condition. This condition shall be overcome using new transport algorithms which have been developed by the partner C. Pain (Pain et al. 2001), awaiting application to porous media flow in the presence of fractures.

Simulations for a variety of typical field scenarios will be performed to assess the ideal simulation approach for different cases. The results will also be compared with conventional methods already in the commercial domain. This work will be completed by March 2004.

The classic dual porosity model will thus be replaced by a streamline-based simulation strategy using a new physically based mathematical model for recovery upscaling in fractured reservoirs. This technology shall be commercially available by the end of the project. Streamline-based simulation leads naturally to a particularly elegant formulation of the flow equations. The new matrix-fracture transfer formulation will have a completely general form and will be benchmarked using the discrete fracture simulations from the previous stages of the project. In this respect the project will solve the problem of inadequate fracture simulation.

The simulations with the CSP code will further dramatically diminish the problem of comparing results of heuristic, coarse-grained models with measurements on the actual systems. The high-resolution implies that such simulations are specific and a di rect comparison is valid.


Fracture & Fault Representations

We propose to build realistic fie ld-data based three-dimensional numerical models of structurally complex reservoirs. Reser voir flow will be simulated using two novel and complimentary techniques. First, relevant equations and new transport algorithms invented by C. Pain and co-workers and AMG-based matrix inversion algorithms from K. Stueben shall be explored and optimized for the purpose of the simulation of fractured reservoirs using CSP - a finite-element and finite-volu me based C++ API - as a modular object-oriented framework. This high-resolution FEM modelling will depend on state-of-the-art 3D meshing technology from ICEM Technologies making it possible to discretize intersecting complex fractures using poly-type element meshes. Explicit matrix-fracture transfer influenced by capillary pressure differences will be captured by these simulations.

The code will be tested on a range of common geologically-based discrete fracture networks which shall be represented as CAD models by R. Archer. The results will be compared to results from conventional models and shall serve to parameterize the new mathematical model for large scale matrix-fracture transfer and recovery upscaling. The importance of an explicit, detailed characterisation of the fracture network will be assessed.

Second, reservoir-scale streamline-based simulation will be extended to account for the influence of the fractures on reservoir flow and the impact of faults visible in the 3D seismic. This will be achieved using the newly derived matrix-fracture transfer model to compute a hydrocarbon source term for the advective transport equation solved using the streamline approach. The new mathematical model will use the characteristic fracture pattern, saturation, pressure-gradient etc. as input parameters such that it can also be applied in conventional simulation tools. Methods appropriate for both well-interconnected and poorly interconnected fracture sets will be developed. The appropriate large-scale parameters to describe fracture-matrix transfer will be derived directly from the CSP-based simulations of flow in discrete fracture networks.
The different simulation approaches will be compared on the same datasets and it will be determined which speed-enhancing simplifications under what circumstances permitt to realistically model fractured reservoir recovery such that optimal production strategies can be devised.In the final phase, the methods developed under the proposal will be coded into a commercial streamline-based simulator


Fracture-Matrix Transfer

Present simulation models of fractured reservoirs use a dual porosity or permeability formulation: The fractured reservoir is described with empirical functions used to describe the interaction between fracture and matrix. Because appropriate forms for the matrix-fracture transfer for different displacements are not well established, this model is widely considered to be inadequate to describe flow in fracture networks of complex connectivity, or for designing water and gas injection schemes. The uncertainty in the reservoir description and modelling leads to uncertainties in performance prediction and sub-optimal management decisions. Recoveries from fractured reservoirs are often less than 10%. Inaccuracies in modelling water and/or gas injection prevents the proper design of improved recovery strategies that potentially could result in much higher recoveries. - In this project, matrix-fracture transfer for relevant boundary conditions will be examined with an unprecedented degree of realism such that a rigorous mathematical model for recovery upscaling in fractured reservoir can be established. This will facilitate a fundamental improvement of methods for simulating flow in complex faulted and fractured reservoirs. This technology will be made available as a commercial product for oil companies, service companies and consultancies. The software will be no more expensive than conventional reservoir simulators, while offering unrivalled accuracy and speed.


Associated Personnel


Industrial Partners



  • BP
  • ConocoPhillips
  • Total
  • PetroCanada
  • DTI