MRST - MATLAB Reservoir Simulation Toolbox

Gallery
These pages highlight features in MRST by showing examples of simulations and visualizations made using the toolbox. Click on the headlines or images to see a more detailed description or a bigger version of each image. If you want to know more about a specific example, please contact its creator. Moreover, please do not hesitate to submit your own contributions to the MRST gallery.

Examples from users

         

Hydrocarbon reserves estimation

Using MRST, we have developed a new graphical user interface  (GUI) to estimate hydrocarbon reserves of any reservoir model using traditional volumetric method. Furthermore, the proposed GUI helps the user to easily import and rapidly visualize geometrical model and petrophysical properties. Calculated hydrocarbon reserves are practically identical to Petrel, and simulation with PVT and scal data show excellent match with Eclipse 100.

By: Abdelhakim BENALI (SONATRACH)

   

Multilateral Well Stimulation

Fenix Consulting Delft has used MRST to develop Simfish, a simplified reservoir simulation model for optimizing the design of award-winning fishbones technology for multilateral well stimulation. Simfish estimates oil rates, PI increase and incremental
oil resulting from use of fishbones compared with open-hole completion. Simfish runs in 1-6 minutes and generates Eclipse compatible wellbore geometry. See presentation of fishbones, pages 24-26.

           

Ongoing activities 

         
 

Compositional simulator

A compositional simulator: CO2 and trace amounts of Methane injected into a Decane-Methane mixture at a low enough pressure that all components are present in both vapor and liquid phases. All PVT properties are automatically calculated using correlations and thermodynamics. You only need to specify five numbers (critical volumes, molar weight, critical temperature, critical pressures, acentric factors) for each component present, and MRST figures out the rest without any external libraries. Validation against AD-GPRS shows excellent match.

By: Olav Møyner (SINTEF)

     

Multisegment wells

 

Multisegment well capabilities is under develoment in MRST. The implementation allows for general well topology (including loops), and is flexible with respect to choices of flow/pressure relations to account for friction, acceleration and flow across various devices. The figure shows a section of a multisegment well coupled to a radially refined near-well reservoir grid.

By: Stein Krogstad, SINTEF

           
           

Time-of-flight distributions

The TOF-distribution for a given flow field is obtained by tracing a unit-pulse from inlet to outlet. The figure shows the TOF distribution for layer 23 of the SPE10 model, with an induced pressure drop from left to right. The red line indicates 1PVI, and the distribution shape and position relative to this line relate to model heterogeneity. We have used TOF-distributions coupled with 1D-displacement profiles for rapid estimation of polymer injection performance.

By: Stein Krogstad, SINTEF

     

Optimized CO2 storage plans based on minimizing leakage and pressure buildup

In formations with very large CO2 trapping capacities, injection rates are likely to be limited due to pressure buildup. We use the Bjarmeland formation, available from the Norwegian Petroleum Directorate's CO2 Storage Atlas, to demonstrate the importance of optimizing the injection rates, subject to maintaining the formation pressure under a predefined limit. More details found here.

By: Rebecca Allen, SINTEF

           
           
 

Categorizing CO2 storage potentials

Optimized injection strategies are obtained for three formations that are part of the Norwegian Petroleum Directorate's CO2 Storage Atlas. The storage potentials of these formations are then categorized based on whether the storage was limited by leakage, pressure buildup, or a mixture of both. More details found here

By: Rebecca Allen, SINTEF

       
           
           

New in MRST 2016

         

VEMMECH module (geomechanics)

Implementation of the Virtual Element Method for linear elasticity, for MRST unstructured grids.

Top: Compaction test for a segment of Norne. The domain has been padded to a rectangular box (the padding cells are not shown) and no vertical displacement  is imposed on the sides while the horizontal  faces at the top and bottom are free. The colors correspond to the vertical displacement.

Bottom: Incremental volumetric loading on an irregular 2D grid (pebi, triangles, quadrilaterals, hanging nodes). The colors correspond to the absolute value of the second stress component (σyy).

By: Halvor Møll Nilsen / Xavier Raynaud, SINTEF

   

EOR module

Module with simulation tools for water-based EOR such as polymer and surfactant. In this example, the injection of a surfactant plug is simulated in a layer of spe10. The last image shows the extent of the region where the oil has been mobilized.

By: Xavier Raynaud, SINTEF

 

 

 

       

New in MRST 2015b

         

CO2 storage in Stø formation

Location and storage capacity of prospects and greater structural closures in the Hammerfest basin aquifer.  Data are taken from the Norwegian Petroleum Directorate's CO2 Storage Atlas of the Barents Sea. The prospects and greater structural closures were identified using MRST's trap analysis algorithm. Once trap volumes were identified, trapping capacities were calculated assuming 100% storage efficiency.

By: Rebecca Allen, SINTEF

   

Black-oil polymer simulation with shear-thinning/thickening

An experimental fully-implicit polymer simulator is developed based on our black-oil simulator, which will be released as part of MRST 2015b. Non-Newtonian flow rheology, such as shear-thinning/thickening effects, is incorporated. The example shows how the shear-thinning effects influence the water injection process.


By: Kai Bao, SINTEF

           

Unstructured grids

Examples of unstructured 2D grids made by MRST. From upper left to bottom right: (i) triangulation imported from DistMesh, (ii) multiblock grid, (iii) Voronoi grid adapted to well path, (iv) pentadecagonal tiling. The grids in the right column have been used along with our virtual-element geomechanics solver, whereas the grids in the left column are intended to be used with our standard flow solvers.

By: Knut-Andreas Lie / Runar Lie Berge (NTNU)

     

Multiscale waterflood simulation

In this example, we have used the multiscale residual-smoothed basis function method (MsRSB) to simulate a waterflooding scenario in the Watt field case to be published by Heriot-Watt University. The original model has 415 000 cells, and even with only 800 coarse blocks in the multiscale solver, we get perfect match of the production curves with a limited number of iterations. The MsRSB method is implemented in the INTERSECT R&P simulator.

By: Olav Møyner (SINTEF)

           

Flow diagnostics: estimating future production

Tracer partitions and time-of-flight are useful quantities when establishing flow paths and volumetric connections in a reservoir. Here these quantities have been used to postprocess a 3D black-oil simulation of the Norne field. By plotting the fluid distribution as a function of travel time from a well, we get a good picture of future production if present well controls remain constant.

By: Stein Krogstad / Knut-Andreas Lie (SINTEF)

   

Flow diagnostics: flux allocation

We have used tracer partitions to establish volumetric connections in a shallow-marine reservoir (SAIGUP). In the middle plot, solid lines indicate well pairs that are in communication, with color given by the producer and the percentage refering to the flux from the injector that can be associated with the various producers (i.e., flux allocation). The pie charts in each of the four inlets show total rate allocation for the well, whereas the graphs show cumulative rate allocation for individual well completions.

By: Knut-Andreas Lie (SINTEF)

           

 Older examples

         
 

Discrete fractures with fully implicit solver

Example of using fully implicit calculations including flow in discrete fractures. The discretization and inclusion of the discrete fracture properties is done in a preposessing step, such that the fully implicit solver code is unchanged. Such effects as aperture changing with pressure are included as simple functions.

By: Halvor Møll Nilsen, SINTEF

     Joule Thomson effect

Thermal simulation with accurate fluid models

Fully coupled multiphase calculation with thermal effects using state-of-the-art advanced fluid properties. This is achieved by combining the fast prototyping facilities of MRST with the fluid property calculation routines from coolprop.org.

By: Halvor Møll Nilsen, SINTEF

 

 

     

 

 

Geochemistry

The local equlibrium of geochemical systems is solved using robust techniques from geometric programming. The code really highlights the advantages of automatic differentiation.

By: Xavier Raynaud, SINTEF

     

Viscous fingering dynamics

To investigate the dynamics of viscous fingering a moving frame solver was implemented. This requires a proper approximate Riemann solver for the hyperbolic equation. In addition, a fully-implicit, higher-order method was used to minimize numerical diffusion, which is essential if one wants to study the front in a fixed global coordinate system.

By: Xavier Raynaud, SINTEF

 

 

     

 

 surfactant

Surfactant modeling

A surfactant model similar to the one used in the commercial simulator Eclipse has been implemented using MRST. The work was entirely done in a master project.

By: Kristian Jørgensen, IPT, NTNU

     non-newtion injectivity effects

Calculation of shear-thinning near well

Calculation of the shear-thinning effect near a well. The flexibility of prototyping in MRST makes implementation simple. In the figure, a radial discretization is used to minimize the degrees of freedom.

By: Xavier Raynaud, SINTEF

 

 

     

 

 simple mechanic example

Geomechanics

Geomechanical calculations based  on virtual element methods. This is is a promising method that is valid and has a generic implementation for general polyhedral grid.

By: Halvor Møll Nilsen, SINTEF

     Comparison of total simulation time for the Norne field black-oil model using a CPR-AGMG and a CPR-multiscale preconditioner

Multiscale CPR preconditioner

Multiscale finite-volume methods can be used as CPR preconditioners for fully-implicit simulation of black-oil models. For the Norne field model, using a multiscale method is slightly faster than using an aggregation-based algebraic multigrid method.

By: Olav Møyner SINTEF

 

 

     

 

 

Black-oil simulator

The fully-implicit black-oil simulator in MRST has been validated against a commercial simulator for 20 years of production data on the Voador field.

By: Xavier Raynaud, SINTEF

     

Flow diagnostics

Flow diagnostics can be used for interactive post-processing of simulation results, here for the Norne model (well configuration does not
portray that of the real field). [paper]

By: Knut-Andreas Lie, SINTEF

           

Multiscale finite-volume method on complex grids

Approximate pressure solution computed by a multiscale two-point flux-approximation method on a model of the Gullfaks field. The fine model has 216 334 cells and the coarse partition 605 blocks. [paper]

By: Olav Møyner SINTEF

   

Production optimization

Net-present value optimized for the Norne bechmark case using flow diagnostics as a fast simulation proxy to suggest optimal strategies that can subsequently be verified by a computationally expensive commercial simulator. [paper]

By: Stein Krogstad, SINTEF

           
 

Validation: the Delft EGG model

Researchers from TU Delft have validated MRST against Eclipse, GPRS, and MoReS using a version of their Egg model. The results obtained with the four simulators are almost identical. Read more.

By: Jan Dirk Jansen, TU Delft

     

Structural trapping of CO2

Structural trapping for the Sandnes formation computed by the numerical CO2 module: the plots show all structural traps and the corresponding accumulation regions, as well as a spill path from an injecion well.

By: Knut-Andreas Lie, SINTEF

           
 

Flexible gridding

Short introduction to grid generation in MRST. The picture to the left shows an extruded grid consisting of three parts: a Cartesian grid at the outer boundary, a hexahedral grid in the interior, and a radial grid with exponential radial refinement around the two wells.

By: Jostein R. Natvig, SINTEF

   

Visualisation

Overview of basic script-based visualisation techniques and examples of how to visualise cell and face data in MRST. In addition, recent versions of MRST include tools for interactive visualization.

By: Bård Skaflestad, SINTEF

           

Faults and multipliers

Brief discussion of a new method for consistent inclusion of fault multipliers in multipoint discretisations that has been implemented in MRST.

By: Halvor Møll Nilsen, SINTEF

     

Rate optimisation

The figure shows optimisation of production rates using an adjoint MsMFE method for a synthetic setup based on the Norne field.

By: Stein Krogstad, SINTEF

Published November 17, 2009