Techniques in Geostatistics are increasingly being used to generate
reservoir models and quantify uncertainty in reservoir properties. This
is achieved through the construction of multiple realizations to capture
the physically significant features in the reservoir. However, only a
limited number of these realizations are required for complex fluid flow
simulation to predict reservoir future performance. Therefore, there is
the need to adequately rank and select a few of the realizations for
detailed flow simulation.
This thesis presents a methodology for building and ranking
equiprobable realizations of the reservoir by both static and dynamic
measures. Sequential Gaussian Simulation was used to build 30
realizations of the reservoir. The volume of oil originally in place,
which is a static measure, was applied in ranking the realizations.
Also, this study utilizes Geometric Average Permeability, Cumulative
Recovery and Average Breakthrough times from streamline simulation as
the dynamic measures to rank the realizations. A couple of realizations
selected from both static and dynamic measures were used to conduct a
successful history match of field water cut in a case study.
CHAPTER ONE – INTRODUCTION
1.1 PROBLEM DEFINITION
In Geostatistical reservoir characterization, it is a common practice
to generate a large number of realizations of the reservoir model to
assess the uncertainty in reservoir descriptions for performance
predictions. However, only a limited fraction of these models can be
considered for comprehensive fluid flow simulations because of the high
computational costs. There is therefore the need to rank these
equiprobable reservoir models based on an appropriate performance
criterion that adequately reflects the interaction between reservoir
heterogeneity and flow mechanisms.
Most techniques used in ranking of realizations are based on static
properties such as highest pore volume, highest average permeability,
and closest reproduction of input statistics. The drawback of these
simple techniques is that they do not account for dynamic flow behavior
which is very essential in predicting future reservoir performance.
This thesis work seeks to build and rank equally probable
representations of the reservoir using petrophysical properties such as
porosity, water saturation, and permeability. The multiple reservoir
descriptions are ranked using both static (Stock tank oil originally in
place) and dynamic (geometric average permeability, connected
hydrocarbon pore volume, average breakthrough times and cumulative
The main purpose of reservoir characterization is to generate a more
representative geologic model of the reservoir properties. The
objectives of this study are as follows.
In building a static representation of what a reservoir is most
likely to be it is necessary to adequately capture the uncertainty
associated with not knowing its exact picture. In the effort of
capturing the uncertainty, this study focuses on generating multiple
equiprobable realizations of the reservoir with each having unique
static and dynamic properties.
Only a few number of the realizations generated could be carried on
for complex flow simulation due to the computational cost. Usually
ranking of the realizations to select a limited number are done by
static means which do not capture the dynamic mechanisms essential to
reservoir performance. Only few dynamic ranking measures are currently
in use. This research work seeks to define new dynamic ranking criteria
and compare them with existing criteria.