I N T E R P R E TAT I O N

Fault modelling:

New techniques improve reservoir

description while reducing risk
Errors in fault interpretation can have significant negative impact on reservoir models, faci-
lity plans, and field economics. By using integrated software applications and workflows
that cross disciplines, asset teams can understand complex reservoirs better and faster.

Drew Wharton, Schlumberger Information ved reservoir characterization and better than using single attributes. Three com-
Solutions, Houston, Texas, USA decision-making. These workflows include monly used attributes are variance, dip
using multi-trace seismic attributes to deviation and chaos. While there are signi-
detect fault edges, using swarm intelligen- ficant differences among these attributes,
sing the latest software tools from ce for automated fault interpretation, and all are derived from the same input data.

U Schlumberger Information Solutions

(SIS), geologists, geophysicists and
reservoir engineers are able to collaborate
using seismic attributes, formation pressu-
re data and reservoir simulation to impro-
ve fault seal analysis
Different attributes can detect different
expressions of a geological feature, and
when combined, they offer a more comple-
like never before using integrated worksta- te fault expression and better fault inter-
tions with shared applications and data. Edge detection pretation.
New workflows created for viewing and Multiple edge-detection attributes can
interpreting data have resulted in impro- highlight fault features much more clearly Automated fault Interpretation
Even with the aid of edge-detection
algorithms, manual interpretation of fault
surfaces remains a painstaking exercise,
and traditional 2D interpretation techni-
ques of picking fault and horizon intersec-
tions or fault polygons can produce signifi-
cant errors in 3D space. A novel approach
for automatically extracting fault surfaces
from attribute data is now available. This
methodology, called Ant Tracking, which
has been implemented in Petrel™, enables:
shorter fault interpretation cycles, geologi-
cally consistent sets of faults based upon
unvarying rules, and repeatable fault inter-
pretation solutions.
The automatic extraction of fault surfa-
ces is nontrivial due to the noisy and repe-

This workflow shows how Ant Tracking, Edge

Detection, Automatic and Manual Fault Interpreta-
tion all fit together to generate the best possible
fault interpretation.

30 GEO ExPro September 2004

 Different seismic attributes emphasize different
aspects of the original input signal. The change in
attribute response spatially allows us to correlate
these attributes to geologic features.

titive nature of seismic attributes. The sur- each ant move along what appears to be a develop a proper field plan. Existing 3D
faces usually appear more like trends than fault surface while emitting "pheromone." reservoir modelling techniques cannot
well-defined, continuous surfaces. Further Ants deployed along a fault should be able incorporate the full complexity of fault sys-
data processing is needed, and this is achi- to trace the fault surface for a while before tems. Typically, faults and their properties
eved using the principle of swarm intelli- being terminated. Surfaces meeting expec- are calculated and collapsed onto 2D fault
gence, a term for the collective behaviour tations will trace faults deployed at diffe- planes, with transmissibility multipliers
that emerges from a group of social rent positions in the volume, and will be used to regulate the flow of hydrocarbons
insects. For example, ants are able to find strongly marked by "pheromone." Surfaces across the fault in the reservoir simulator.
the shortest path between the nest and a unlikely to be faults will be unmarked or Combining seismic attributes designed
food source by communicating via phero- weakly marked.The interpreter uses geolo- to highlight stratigraphic events and geo-
mone, a chemical substance that attracts gical insight to validate and edit the statistical algorithms for estimating rock
other ants. results, rather than spending time extrac- properties from wireline logs, it is possible
By encoding fault property expectations ting fault surfaces manually. to improve the interwell distribution of
(rules) as a behaviour of intelligent softwa- reservoir properties resulting in enhanced
re, it is possible to enhance and extract Fault seal fault characterization.
fault-like responses from the attribute. The With faults identified, the asset team still Fault seal analysis is essential to explain
idea is to distribute a large number of the- needs to know whether or not they are how flow paths have been modified
se electronic "ants" in the volume; and let sealing or conduits to flow in order to through faulting.

The timeslice of the Variance cube (left) contains a

lot of responses in the top right corner because the
slice is cutting through some chaotic texture in the
seismic data. In the corresponding Ant Tracking
results (right), the chaotic responses have not been
extracted because they do not fulfill expected fault
properties. Detailed fault data have been derived
from the surrounding voxels.

GEO ExPro September 2004 31

I N T E R P R E TAT I O N
Calculating transmissibility based on the well, and monitoring the pressure build-up production data and pressure data obtai-
fault throw, shale gouge ratio, and fault over time. Pressure build-up curves are ned from well tests. The reservoir model
permeability values, the higher the trans- analysed to determine formation permea- can be adjusted and fault interpretations
missibility multiplier, the easier oil can flow bility and interwell continuity. updated until the simulation matches the
across the fault plane. actual well behaviour, with continual upda-
Reservoir simulation ting throughout the life of the field.
Well pressure data The static reservoir model with the initi-
Accurate fault interpretation is encum- al fault interpretation should be conside-
bered by the fact that the data necessary red a starting point, not a finished product.
to understand communication across Projections based on the static model may
faults isn’t available until the well is drilled. provide expected production rates and
Today, repeat formation test (RFT) analysis pressures during various operations, but
provides accurate pressure correlation bet- these must be updated as the reservoir
ween wells to confirm reservoir continuity. characteristics change during production.
Fluid densities can be measured and fluid Reservoir pressure measurements from
contacts can be located by analysing pres- well tests, coupled with production history,
sure gradients. Pressure measurements in help improve the reservoir description
development or infill wells can be compa- over time.
red with original formation pressures to The initial static reservoir description is
positively identify drainage discontinuities seldom accurate since reservoirs are rarely Streamline simulation of the tortuous fluid flow
paths through the reservoir validates the fault seal
due to faulting or stratigraphy. In addition, homogeneous and significant changes can
interpretation and indicates where improvements
formation pressures from well tests are occur as fluids are injected or produced.
to the reservoir model should be made. The stream-
often used to evaluate reservoirs by crea- Therefore, construction of a dynamic line colour indicates water saturation, while the
ting a pressure transient by withdrawing model using a reservoir simulator should fault plane color represents the shale gouge ratio
fluid from the reservoir, shutting in the combine the static reservoir model with of the damaged zone

32 GEO ExPro September 2004