Abstract

This paper outlines a multi-step geostatistical methodology for constructing reservoir models using sedimentary data and the notion of Petrophysical Groups.

This method is illustrated through a case study of a mixed carbonate platform oil reservoir (a 20-meter thick sequence). Indeed the technique appears to be particularly suited for reservoirs with great variability of petrophysical properties characterized by a due to the combination of sedimentary and diagenetic effects.

The first stage is the construction of the sedimentary model, with a particular emphasis on the shapes, extents and vertical and areal development of sedimentary bodies.

The second stage consists in constructing Petrophysical Groups from core measurements through statistical clustering techniques. Each Petrophysical Group is a set of reservoir zones having similar porosity, permeability, grain density and capillary pressure curves. Water saturations and relative permeability functions can be assigned to Petrophysical Groups with good confidence.

In the third stage sedimentological and petrophysical data are reconciled in order to map reservoir properties. It appears that the organization of Petrophysical Groups within sedimentary bodies can be explained by depositional and post-depositional factors that may be assessed using well data and regional information. These relationships, once quantified, allow us to build the model of reservoir properties in three steps involving an object-modelling technique, the Sequential Indicator Simulation and Sequential Gaussian Simulation, respectively.

In this complex reservoir the method proves to be a powerful alternative to conventional approaches to the modelling of petrophysical heterogeneities which use lithofacies as a basis for the mapping of reservoir properties.

Introduction

Field development plans and production management have been based on reservoir models for many years. Since most of the newly discovered fields are generally more complex and smaller than in the past (thus having uncertainties which affect key decisions), the richness of these models has been increasing in recent years thanks to the integration of new techniques in geophysics, sedimentology and reservoir engineering.

The first step in model-design generally consists of describing accurately the reservoir structure, both external and internal: field limits, faults and extents of individual flow units (both areal and vertical). This is based on seismic interpretation and on the understanding of the sedimentary and diagenetic processes that have lead to the spatial distribution of rocks in the reservoir.

This phase often results in a set of 2D maps, or, better, a fully 3D model, for the location of the different facies (whatever the definition of facies may be) throughout the reservoir, taking into account the split of the reservoir into stratigraphic sequences. These maps may be generated through a deterministic, geostatistical or hybrid approach. Whatever their origin they usually provide the basis for mapping reservoir rock properties such as porosity and permeability. Indeed the hypothesis that facies are discriminant in terms of petrophysical properties is usually made. Unfortunately this hypothesis proves to be wrong in many cases; if this is so the approach is somewhat discredited and will lead to important uncertainties in the assignment of rock properties to individual gridblocks. This problem especially occurs for many carbonate fields, with strong diagenetic overprint:

P. 927

This content is only available via PDF.
You can access this article if you purchase or spend a download.