Modelling geological subsurfaces

Contents
This section demonstrates Siscat applied to geological modelling. The illustrations
on this page are all produced by Roxar Software Solutions using the RMS system.
More detail on the underlying structures can be found by going there. Illustrations
in this demo covers:
 A complete reservoir
 Composed horizons
 Isocores

Roxar Software Solutions

Roxar Software Solutions is a software and consulting company with proven
expertise in reservoir software, seabed mapping software, consultancy and
services. The IRAP systems developed by Roxar Software Solutions are widely
used for geological modelling by an international client base. The new IRAP
Reservoir Modelling System, introduced in 1994, sets a new standard for user
friendly mapping and reservoir modelling. In RMS Roxar Software Solutions has
integrated Siscat as a numerical core for surface construction, and has greatly
influenced the development of Siscat towards this application area. Roxar Software
Solutions is now SINTEF Applied Mathematic's main collaborator in the field of
geological modelling.

A complete reservoir

This figure shows a complete model of reservoir, including the top and bottom
horizons, fault surfaces and indication of the oil wells. Here, the fault and horizon
surfaced are constructed by Siscat, while the topology and rendering functionality
is part of RMS. The surfaces are constructed by curvature minimization techniques
for the horizons and triangulations for the fault surfaces. The horizons are
constructed from typical geological data, such as seismic tracks and fault polygons.
If you click on the figure you will load a larger version (50k).

The pictures on the left illustrates the reservoir when the top horizon layer is
removed. Click to see. The top horizon layer is discontinuous across the fault area
as illustrated by this figure. In order to model these discontinuities in the surface
the Mask Method in Siscat is used. This method constructs a regularly gridded
surface by minimizing curvature, through solving a particular set of differential
equations. In the formulation of the differential equations, the faults are modelled
by imposing certain internal boundary conditions.

The illustrations below shows the top layer with and without the intersecting fault
surfaces. Click them to get larger versions.
Composed horizons

It is also possible to construct a simple faulted surface, i.e. by keeping the surface
description as one surface model. This is obtained by using the composed surface
type implemented in RMS by simultaneous use of regular grids and triangles. The
key idea is to construct triangulations towards the faults to obtain sharp edges. An
illustration of this is given on the right. Click it to get a better view. Notice the sharp
edges and nice contour curve behavior. Also here the basic surface construction
schemes are functionality from Siscat, such as the Mask Method for the gridding
and the functionality for constrained triangulation towards the faults. To get more
info about this technique take a look at this.

Isocores

Isocores are surfaces describing the volume between two horizons. These are
used for volume calculations and estimation of the amount of oil contained in the
reservoirs. Such predictions are crucial for planning oil recovery at the quality of
the volume calculations are therefore of vital importance. The methods used are
based on data from the wells which leads to the interpolation of a limited number of
points. The methodology is based on a Siscat method described here. The four
figures below shows isocores in different presentations. Click them to take a closer
look.
Page 2. Modelling geological subsurfaces

Previous page.

Some characteristics of geological data

Figure 1. Various scattered data sets in the plane. a) Seismic lines from 3D
seismics. b) Seismic lines from 2D seismics. c) Well data. d) Fault polygons.
One source of information from which to gather geological data is seismic sections.
These result from seismic exploration surveys and subsequent data processing
done by geophysicists and geologists. The seismic sections are interpreted
manually by extracting data from interfaces between geological layers. Figure 1 a)
and b) show examples from so-called 2D and 3D seismic surveys.
Faults represent discontinuities in geological layers caused by severe movements
of the earth's crust. Correct representation of surface models in faulted regions is
important in connection with connectivity analysis, generation of simulation grids
and extraction of high quality isolines and cross sections. Fault polygons are
boundaries of faulted regions given as either 2D- or 3D piecewise linear curves.
The geometry of the fault polygons may take very complex forms as can be
observed in Figure 1 d).
Well logs, as shown in Figure 1 c), provide another source of information about
geological layers. Data resulting from well traces are accurate compared to
interpreted seismic data, but the access to such data is limited. Well data can be
used for correcting surface models computed from other seismic data.

Surface reconstruction from large data sets with geological fault data

As a result of the inability of regular grids to model faults properly, Roxar Software
Solutions have implemented a composite grid structure in their surface modules.
This format is a decomposition of the domain into rectangular grid cells and
triangles. The format is tailored for representing surfaces with discontinuities along
specified polygons, for example fault polygons. The underlying regular grid is
created using the Mask Method in Siscat. Faulted regions are triangulated and fault
polygons are inserted as constrained edges in the triangulations.
Chapter 17 of the book "Numerical Methods and Software Tools in Industrial
Mathematics" describes the composite grid concept in detail.
Figure 2. Example of employing the composite grid functionality in IRAP to data
from a geological layer in the North Sea.
a) Input data.
b) Contour plot from traditional regular grid approximation.
c) Composite grid structure corresponding to the input data.
d) Contour plot from the composite grid.

Figure 3. 3D view of a composite surface with constrained triangulations of faulted

regions and the boundary.
Figure 4. Contour map from the IRAP system

Surface reconstruction from well logs and user specified trends

Well logs provide accurate data for subsurface modelling compared to interpreted
seismic data, but the access to such data is of course limited. However, there
might be additional sources of information about the structure of the surface such
as lineaments and features, and the geologist might have a mental image of the
geometry of the surface.
Siscat provides a variational method based on cosine expansion that is designed
to meet such requirements. The method has many optional parameters that
permits the user to control interpolation accuracy and smoothness of the surface.
Also, the user may control the shape of the surface such as the "flatness" and
preferred direction of smoothness. Thus, one is allowed to vary the shape of the
surface considerably, and still being able to interpolate the input points.
Figure 5. Surface reconstruction from well logs.
Figure 5 shows two interpolating surfaces to the same data set using a method
based on cosine expansion in the explicit surface library in Siscat. The uppermost
surface interpolant is created by giving parameters such as to minimize the overall
curvature of the surface. The other surface interpolant is created by adding a user
specified dominating direction of smoothness 45 degrees from the horizontal line.