This paper presents the result of a reservoir study where seismic data have been fully utilized to obtain a better rock-type based reservoir characterization model. The study was conducted for a highly faulted carbonate reservoir in the Middle East. The available 3D Seismic data have been used to determine the structure map, fault and fracture network, and porosity distribution. Additionally, two seismic attributes (Energy Half Time and Max. Peak Amplitude) had also been used to assist the determination of stationarity regions used for geostatistical simulation.
The reservoir characterization strategy for this study was started with the detailed development of a geological reservoir rock type (RRT) scheme. The RRT scheme was developed based on depositional facies, diagenetic overprints and petrophysical properties, including pore throat size distribution, porosity and permeability. The scheme was then used to constrain the property modeling inside a geological framework that was built based on the seismic interpretation.
Prior to the development of 3D porosity model, four porosity maps were generated based on the available interpreted Acoustic Impedance (AI) and well log data. Each map represents the difference of the influence of seismic data, from full seismic influence to no seismic influence. One map has been selected to represent the seismic derived porosity based on the degree of correlation between the AI data and well-log porosity.
The integration of seismic-derived porosity into the 3D simulation model was conducted using Bayessian Update principle inside a conditional simulation technique. In this technique, 3D porosity distribution honors the underlying RRT and at the same time its vertical-column average honors seismic-derived porosity. The comparison of the results between the simulation with seismic and simulation without seismic has clearly indicated the value added by seismic data, namely improving the simulation result in the areas of poor well-coverage. This is important in order to honor the diagenesis effect that is difficult to model due to a limited number of wells.
Simplified models for describing reservoirs with complex geology are not able to adequately represent the reservoir heterogeneities and their impact on connectivity patterns and flow mechanisms. A critical step in the construction of reservoir simulation models is the description of the underlying reservoir geology. This is even more mandatory in the case of carbonate reservoirs, frequently characterized by an intensive diagenetic overprint of the original carbonates. This process frequently includes assessing the complex reservoir internal architecture reflected in the internal discontinuities within the lithological units, vertical and lateral variation of rock properties and similar heterogeneities within the reservoir layers.
The state of the art technology in generating reservoir models is to start with proper characterization of the underlying geology. The result of the characterization should be used in generating the 3-D model. The challenge that needs to be considered from the start is how the model can replicate the reality, i.e., what is defined during the characterization process.
To cope with the geological complexity and assist in understanding the property distribution in highly heterogeneous reservoirs, the concept of "Rock Types" was introduced and has been widely used as a basis for reservoir characterization in recent years.
Typically, once reservoir rock type was determined and distributed in the 3D volume, the next task is to generate properties, namely porosity, permeability, and saturation, based on the well data (log and core). Ideally, these properties should be generated in such away that they are consistent among each other. Depending on the availability of the data, additional data integration may be done to improve the property model. The objective of this study is to provide a consistent property model by integrating geological based rock type, well log/core and seismic data. In this objective, seismic data is used to further improve porosity model, which was originally constrained to rock type and well log data only.