In complex oil reservoir a detailed characterization of lateral variations of lithologies and reservoir properties is essential to optimize field development plans and maximize effective recovery of in-place hydrocarbons.

This paper illustrates how a flexible and multidisciplinary methodology can integrate 3D seismic data with geological and petrophysical information in order to enhance the identification of lithologies, quantify bed thickness, perform vertical zonation, establish the continuity and extend of reservoir zones and predict inter-well properties, 3D Seismic data (amplitudes, interval travel times between events, frequency variations, etc.) are correlated with porosity, fluid type, lithologies and net-pay thickness via the use of geostatistical models. A methodology of stochastic simulation is proposed in order to integrate 3D seismic data (amplitudes) and well data (rock-types) through a multisteps conditioning approach. This approach is as follows:

  • In each grid node, local conditional probability distribution functions are derived from the global calibrated regressions between seismic amplitudes (RMS-Response Phase Map) and well petrophysical properties. A probability field simulation of primary variables is performed, based on these cumulative probability density functions (cpdf).

  • In a second step the simulation is conditioned to the well data petrophysical values.


Account for all the information available to reproduce the reservoir internal architecture is onne of the main goals of the geoscientists, reservoir engineers and geostaticians. The scarcity of data compared with the total reservoir volume and his heterogeneities is the most usual problem found in this type of studies. The so called "hard data" corresponds to the well data locations, and even providing a high degree of confidence in the measurements, could not be linearly extrapolated to provide information in the inter-wells area, specially for formations deposited in a complex sedimentary environments. Interpolation and simulation are usually required to build the reservoir architecture displaying the different properties and heterogeneities of the oil reservoir, always under geological guidance and interpretation1.

Contrary to the often sparse and irregularly spaced well data (hard data), 3D seismic surveys provide a dense and regular volume of data. Conventional methods make use of seismic amplitudes by converting them into acoustic impedances which may be related to either rock properties or physical conditions of the reservoir2,3,4,5. Empirical or regression formulas may be obtained by cross-plotting impedances against porosity measured at the wells.

An alternative way of using seismic information in the reservoir modeling is the integration of 3D seismic data with well-logs and core information, through a geostatistical model. There are several approaches to integrate the 3D seismic data. The co-kriging methodology6 was the first one to be developed and more sophisticated methodologies exist today, like for instance, a "Markov-Bayes" approach, proposed by Zhou & Journel7.

This paper is a contribution to these developments, with an example of a new method proposed by Soares A.8. The main features of this method is the incorporation of the soft data combined with the hard data to condition the simulation of the reservoir proprieties in the inter-well areas, i.e., at any point in the grid to be simulated the importance of the soft data depends of the relative distance to hard data where well measurements were recorded. In the final maps there is a major influence of the hard data in the areas surrounding the measured points while in other areas, the influence of the soft data prevails.

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