Assigning a modeling method

With the Assign Method form (model > Facies > Assign Method), you specify a modeling method to populate your 3D grid with facies. You assign a method to each one of your Volumes of Interest (VOIs), i.e. the sub-sets of your 3D grid, one by one.

Volume of Interest (VOI)  A sub-set of your 3D grid. It allows you to populate your 3D grid with facies by modeling each VOI separately.

Defining VOIs  VOIs are determined by selections you make on the Create Model form, more specifically the property you select under 'Volume definition' (optionally in combination with nesting via the option 'Nest classes from').

Going through the workflow on a per-VOI basis  It is recommended to take each VOI one by one through the workflow. You start with assigning a modeling method to one VOI. Only once you have completed defining the model parameters for one VOI, you should proceed with the next VOI. This lets you to focus on each VOI separately, analyze its data and fine-tune its modeling parameters. At the end of the workflow, you can run your model in parallel for all or only certain VOIs you selected.

Workflow steps  Since each modeling method has its own specific set of modeling parameters, the workflow will automatically take you to the next relevant step. Some steps might be skipped depending on the method you selected or the data you have. For example, assigning a training image is only relevant when you are using MPS and assigning a variogram is not necessary for MPS.

Facies model  The container which holds all the associated metadata: upscaled property, volume definition (i.e. VOIs), modeling method and modeling parameters per VOI, etc.

Facies grid property  The resultant grid property, which is a combination of all modeled VOIs. Once you have run your model, it will be available in the JewelExplorer (3D Grids > Your 3D Grid > Properties > Facies > Your Facies Grid Property). It is read-only and has the same name as the facies model.

To assign a modeling method

  1. Open the Assign Method form and select the facies model from the Model drop-down list.
    2. You can hover over the factsheet icon () next to the drop-down list to check the key inputs (e.g. 3D grid, upscaled property, volume definition) to the facies model you have selected.
  2. Select a Volume of Interest (VOI) from the table on the left side of the form.
  3. Under Method, select the modeling method that you want to apply to the selected VOI.
    4. You can also copy-paste the modeling method from one VOI to another VOI using the context menu options Copy - Paste (to paste to one VOI) or Copy - Paste Method Options (to paste to multiple VOIs at once). The last option opens a separate form on which you can check the boxes for all VOIs that you want to paste to. Note that checkbox selection can be done in bulk by holding down SHIFT while selecting a range of VOIs (they turn blue), then checking the box of one of them. All other VOIs in the selected range will get checked simultaneously.

    With the Paste Method Options form, you can paste a copied method to multiple VOIs, i.e. target VOIs, at once. The grayed out boxes at the top of the form display information about the Input VOI from which you have copied the method. Subsequently, check the boxes for the Target VOIs to which you want to paste the method. When you are pasting the method, you have the following options regarding the modeling settings (e.g. all settings that you set during the various steps of the workflow, i.e. variogram and transformation settings, conditioning and settings on the 'Control Method' form).

    • Restore previous settings  This option is applicable to target VOI(s) that previously had the same method as the Input VOI. By using this option, you can restore those previous settings, so that you do not need to re-define them.
    • Clear settings  This option is a fast way of assigning a method with no settings. You can use this option, if you have many VOIs to which you want to assign the same method.

    Paste method options    click to enlarge

  4. Click Apply to apply your selection and keep the form open, or OK to proceed to the next step of the Facies Modeling workflow:

Assign TI if you selected MPS,

Experimental variogram or Variogram model if you selected SIS or TGS,

Control method in all other cases.

Facies modeling methods

This selection assigns a single facies to all the grid cells within the VOI.

Sequential Indicator Simulation (SIS) is a facies modeling technique, which allows assigning different indicator variogram-models to each facies. SIS provides multiple ways to condition the simulation to soft or secondary data, additional to the hard data (wells). Soft or secondary data can be obtained from geological interpretation or geophysical measurements. To run SIS, each facies requires one indicator variogram model to be created with the Variogram Model form. Both the VPC and variogram models are automatically retrieved. SIS allows the user to specify the size of the search ellipsoid, the input proportions (either global or VPC proportions), the method for integrating soft or secondary data, the correlation coefficient between the to-be modeled data and the secondary data, and strength of the cleaning.

More information about SIS can be found in 'A sequential indicator simulation program for categorical variables with point and block data; BlockSIS' (Deutsch, 2006), see http://dx.doi.org/10.1016/j.cageo.2006.03.005.

The Truncated Gaussian Simulation (TGS) is a stochastic facies modeling method, which allows simulating facies successions with control on the facies order. Typically this method is used in depositional environments where an ordered (laterally or concentric) occurrence of facies is expected, e.g. shore face deposits, progradational sequences, carbonate shoals and reefs. TGS operates with one variogram model, derived from the spatial continuity from one facies of interest. Variogram models with large ranges will lead to smoother results and less artifacts in the simulation results.

The name Truncated Gaussian Simulation relates to the underlying simulation processes. Values of a Gaussian field generated with Sequential Gaussian Simulation are truncated according to the facies proportions in a subsequent step.

MPS is one of the most advanced facies modeling techniques delivering the realistic depositional pattern reproduction of object-based modeling with the speed of variogram-based techniques in the presence of many wells. With MPS you can:

  • incorporate geological interpretations and spatial relationships of facies into the modeling result
  • perform facies modeling in unsampled locations
  • capture internal heterogeneity to reproduce more realistic models

A typical example is the modeling of clastic systems with channels, levees and crevasse splays. These architectural elements do not appear randomly in space but follow geological rules. With MPS you can model channels with levees and crevasse splays realistically arranged in space, honoring their spatial relationship.

In order to do this, MPS needs a training image. A training image is a template (in 2D or 3D) which contains the geostatistical information that MPS uses to calculate the spatial probabilities. Any facies pattern that you can draw on a training image can be reproduced by MPS. The MPS facies distribution replicates the facies associations found in the training image, while honoring control points (i.e. the wells).

JewelSuite Subsurface Modeling builds on the MPS technology Impala™ from Ephesia-Consult. Impala™ is an advanced and fast MPS implementation taking advantage of parallelization and further optimizations. Impala™ does not require the training image to be stationary (i.e. the training image may contain a trend), because it is able to incorporate trends via auxiliary trend properties.

  • Training image without a trend   In the simplest scenario, you can run MPS by assigning a training image to a Volume of Interest (VOI). If the facies in the training image are identical to the facies in the upscaled grid property, MPS will honor the well during simulation. You can assign a training image with the workflow step Assigning a Training Image.
  • Training image with a trend  In a more advanced scenario, the training image could contain a trend, e.g. from proximal to distal with changing facies patterns. In this case you need to describe this trend by creating an auxiliary trend property. This trend property needs to be available as a continuous property of the training image grid in the JewelExplorer. Your model grid (the grid that holds your VOI) needs to have a corresponding trend property. You subsequently need to link the trend property of the training image to the trend property of the model grid. Linking the trend can be done with the workflow step Trends and proportions.

While the above described auxiliary trend property describes the link between areas of the training image with areas of the VOI, you might want to constrain your modeling to facies probabilities, which is an entirely different concept. Facies probabilities enable seismically constraint modeling. To guide the spatial distribution of facies during MPS simulation, you have to provide one probability property per facies on the model grid. In practice, MPS only requires such probability property for one less facies, because it calculates the probability of the last facies. As with facies probabilities for SIS, the probabilities have to sum up to 1 for any grid cell.

MPS can also scale the facies patterns in the training image during the simulation. Factors greater than 1 will enlarge the patterns in the training image, any factors less than 1 will shrink the patterns. Reasonable results are possible within a scaling range of 0.5 to 2. If you need to enlarge or shrink your facies patterns beyond these values, it is recommended to build a different training image.

Rotation enables you to align the patterns in the training image with the desired orientation in your VOI. You can either rotate the training image globally, i.e. the same rotation angle everywhere. Or you can work with an azimuth region property. Such property contains local azimuth information at coarse level. For instance, you could rotate the training image in the eastern half of your VOI +30deg and in the western part -30deg. A locally varying rotation angle is also required if the trend on the simulation grid changes in more than one direction and the pattern in the Training Image needs to be rotated so that the patterns follow the trend. To obtain such locally varying azimuth regions consistent with the changes in the trend, use the functionality on the Rotation tab on the Trends and proportions form.

In order to understand the steps required while modeling a MPS facies model, see Example - Modeling facies with MPS and 3D trend property.