As geoscientists and engineers started realizing that more oil and gas could be squeezed out of existing reservoirs, awareness about characterizing the reservoirs grew, gradually at first and rapidly later. Apparently, the three main reasons that led to the development of reservoir characterization can be listed as follows:
- The rapid growth of fast computing that led to innovative visualization packages for interactive interpretation workstations and visionariums.
- The idea of integration of different types of data, which received lukewarm response in the late 1990s, became a reality in the last decade and got firmly rooted in oil and gas companies in the form of asset teams.
- Ever since 3D seismic surveys became routine in oil and gas companies in the early 1990s, there have been significant advancements in data acquisition and processing techniques that have led to more accurate imaging of reservoirs and their internal architecture.
We have now reached the stage where reservoir characterization has come to be known as a discipline of its own. Some universities have started offering a master’s degree program on reservoir characterization so that the required knowledge and skill sets can be imparted directly to the students. We have already seen complete sessions on reservoir characterization at various international conventions, the chief one being the SEG Annual Meeting. In a way, reservoir characterization now occupies center stage.
The term “reservoir characterization” essentially refers to all the pertinent information that is required to describe a reservoir in terms of its ability to store and produce hydrocarbons. This entails knowing the complete reservoir architecture including the internal and external geometry, its model with distribution of reservoir properties (static such as porosity, permeability, heterogeneity, net pay thickness, etc.), and understanding the fluid flow within the reservoir (dynamic). Such information helps improve production rates, rejuvenate oil fields, predict future reservoir performance, minimizes costly expenditure, and helps managements of oil companies to draw up accurate financial models. Regarded as an important phase between the discovery of an oil or gas field and the reservoir management phase, to collate all the information mentioned above, reservoir characterization integrates the technical disciplines of geology, geophysics, reservoir engineering, petrophysics, economics and data management.
The success of the reservoir characterization effort depends on how well the integration of the above disciplines is carried out, an elusive goal in some cases, and the success of each project could vary. Any reservoir characterization exercise usually begins with the available geological information and knowledge about the depositional and facies environment.
Structural interpretation of seismic data as well as a sequence stratigraphy approach is an important input defining the framework of the reservoir model. The contribution of seismic data in populating 3D reservoir models in terms of petrophysical properties comes from many derived seismic attributes as well as their calibration with log data and the relevant rock physics models. This is supplemented with interpretation of available core and log data. 3D reservoir models are often transformed into flow simulation models to validate their ability to predict past and future performance of the field.
In this focus section, we include four diverse papers highlighting the different ways that reservoir characterization is attempted, whether it is in terms of a particular aspect of seismic interpretation, or in terms of using seismic attributes for deriving some property of the reservoir or in terms of describing a newer approach adopted for a rock physics model and its application.
In the first article in this section entitled “Dynamic reservoir characterization of the Lower Cretaceous Paluxy Formation, Delhi Field, Louisiana”, Robinson and Davis discuss the application of time-lapse on seismic data for reservoir characterization and fluid movement of Delhi Field in Louisiana. The timelapse seismic interpretation allowed for CO2 flow paths to be identified within the reservoir as well as led to the identification of permeability baffles in the Paluxy reservoir. Such valuable information would provide more detailed input to the reservoir simulation model studies.
In the next paper entitled “Characterization of a heavy oil reservoir combining multiattribute analysis and spectral decomposition for density prediction, Matu in Sub-basin, Venezuela”, Ruiz and Aldana demonstrate the characterization of a heavy oil reservoir. The rock physics analysis showed that the acoustic impedance of the shales and sandstones was similar and density could be a good lithology indicator. The authors adopt a multiattribute analysis for density estimation using Multi-Layer Feed-Forward Neural Network (MLFN), in preference to Probabilistic Neural Network (PNN), and generate a pseudo density volume. They also found that this density estimation improved with the addition of spectral decomposition attributes. The authors suggest that such an approach could prove to be useful in areas where acoustic impedance between sandstones and shales is not very different.
The next article entitled “Seismic lithology prediction: a Montney shale gas case study”, by Neito et al. describes a workflow where petrofacies information from well log data is integrated with seismic data to predict lithology in the Montney Shale Gas Formation. Prestack elastic inversion on 3D seismic data is used to generate probability volumes via Bayesian classification, for each of the petrofacies determined from log curve data. All this data is used to generate a Most Probable Petrofacies (MPP) volume, which in turn is used to plan the location of wells in the Upper, Middle, and Lower Montney Formations.
In the last article in this section, entitled “New attribute for determination of lithology and brittleness”, Sharma and Chopra propose the use of a new attribute Ερ, to serve as a brittleness indicator, where Ε is the Young’s modulus and ρ is the density. First, using log data the authors demonstrate the higher sensitivity of Ερ over the traditionally used attribute for the purpose, μρ, where μ is the modulus of rigidity. The new attribute shows better separation of clusters in crossplot space (e.g. Ερ versus κρ) corresponding to different lithologies. As well, for a brittle rock, both Young’s modulus and density are expected to be high, and so Ερ attribute does exhibit high values, as has been demonstrated by its application on a data set comprising the Montney Formation.
We hope the readers will find the papers as interesting as I did while putting this section together.