Sei sulla pagina 1di 2

Multiple-point Statistics (MPS)

Get realistic images of the internal structure of your reservoir


Isatis offers a powerful implementation of the
Multiple-point Statistics (MPS) algorithm which
integrates the optimized Impala1 v2.0 library by
Ephesia Consult.
Multiple-point Statistics is a facies modeling technique
based on multiple-point statistics instead of the
conventional variogram-based techniques founded on
bi-point statistics. It offers another way to model
complex and heterogeneous geological environments
through the use of a training image which describes
the geometrical characteristics of the facies to model.
The method allows capturing geological elements like
channels, reefs, bars, dikes or differently oriented
facies, while honouring data information.
Isatis,
with
Impala,
powers
MPS
technique
implementation as it allows generating complex yet
realistic geological patterns in an easy and efficient
way.

Training image
depicting a channel

MPS simulation in Isatis using


the previous training image

Acronym for Improved Multiple-point Parallel Algorithm using a List Approach.

Impala algorithm has been optimized in many ways:


It offers a new and efficient strategy for recording
matching configuration. Multi-point statistics are stored
in hybrid tree-list structures rather than search trees,
thus resulting in a substantial reduction in memory
requirements and in a higher calculation speed.
Being fully parallelized, it benefits from multiprocessors, which also drastically improves the
calculation performances.
It uses a multi-grid approach to capture structures
within the training image that are defined at different
scales, which avoids considering a huge pattern.
Unlike classical MPS algorithms which tend to
reproduce the global proportions of the training image,
Isatis Impala-based MPS uses soft local proportions to
simulate non stationary environment.
It may take into account external continuous
conditioning data which can be diagraphy, seismic or
well information, but also a gradient (i.e. distance to
the coast, distance to specific topographic object, etc.).
This allows refining and enhancing the realistic aspect
of the simulations and enables the simulation of non
stationary deposit environment.
Besides, Isatis provides functionalities to create 2D
training images from imported images and considers
object-based or plurigaussian non conditional simulations
as potential 3D training images.
This Impala-based MPS algorithm complements Isatis
Plurigaussian Simulations (PGS) already used to
model complex geology geometries. It enriches the
exclusive range of facies modeling techniques Isatis
already offers (including object based simulations like
Boolean, Dead Leaves or Dilution techniques, variogrambased simulations like SIS, TGS, PGS or training imagebased simulations like the Annealing Simulations).

Simulation of dykes
with Isatis new MPS

GEOVARIANCES offices
France - Avon-Fontainebleau - +33 (0)1 6074 9090
Australia - Wynnum-Brisbane, QLD - +61 (0)7 3348 5333
info@geovariances.com - www.geovariances.com

Isatis MPS Key Points


High performances
Parallel algorithm (CPU performances)
Based on lists (RAM performances) and
trees (speed performances)
Unique features
Multi-grid approach for capturing efficiently
large scale features
Conditioning to hard data (seismic, facies
proportions)
Nonstationarity handled with continuous
auxiliary variables or local proportions
Advantages
Intuitive use of training images (seismic,
geological model, analogs, boolean
simulations)
Full control of parameters
Proper modeling of non stationarity and local
proportions
More flexibility on search template size and
number of facies
Improved efficiency

Potrebbero piacerti anche