Reservoir Characterization Model: Definition and Data Inputs

What Is a Reservoir Characterization Model?

A reservoir characterization model is a three-dimensional static representation of a subsurface reservoir that integrates structural geometry, lithological facies, porosity, and permeability distributions derived from seismic interpretation, well logs, core data, and biostratigraphy, providing the geologic framework that reservoir engineers populate with fluid and pressure data for dynamic simulation.

How Reservoir Characterization Modeling Works

The reservoir characterization workflow begins with structural modeling: the geophysicist interprets fault surfaces and horizon picks on the 3D seismic volume, generates isochore maps for each stratigraphic interval, and assembles these into a faulted structural framework. Modern industry software (Petrel by SLB/Schlumberger, IRAP RMS by Roxar/Emerson, Paradigm SKUA-GOCAD by Emerson, and OpenWorks by Halliburton) builds this framework as a cornerstone grid, a three-dimensional mesh of hexahedral cells whose geometry conforms to the fault and horizon surfaces. Cell dimensions are chosen to balance geologic detail against computational cost; full-field models typically use cells of 50 to 100 m (165 to 330 ft) laterally and 0.5 to 2 m (1.6 to 6.5 ft) vertically, while near-well detail models may use 5 to 10 m cells.

Once the structural grid exists, the geologist populates it with facies and petrophysical properties. Facies modeling assigns a rock type (e.g., channel sand, crevasse splay, overbank shale) to every cell using stochastic simulation methods that honor the proportions and geometries established by the depositional model. Object-based simulation explicitly places channel bodies, lobe deposits, or mound-shaped reef objects of specified dimensions and orientations drawn from outcrop analogs or seismic geomorphology; it is best suited to systems with distinct, geometrically recognizable facies. Variogram-based simulation (Sequential Indicator Simulation, SIS) models facies using spatial correlation statistics derived from well data; it works in any depositional setting but tends to produce overly layered geometries in low-well-density areas. Multi-point statistics (MPS) methods, which use a training image derived from outcrop photographs or process-based forward stratigraphic models as a statistical template, reproduce complex, curvilinear geometries such as meandering channels, deltas, or karst networks that variogram methods cannot capture. After facies are assigned, petrophysical properties (porosity, water saturation, net-to-gross) are distributed within each facies class using Sequential Gaussian Simulation co-kriged with seismic-derived acoustic impedance or amplitude attributes to constrain the interpolation between wells.

Permeability assignment follows the porosity model using poro-perm transforms calibrated to core plug measurements. These transforms are facies-specific because a channel-fill sand and a bioturbated heterolithic sand at the same porosity may differ in permeability by one to two orders of magnitude due to differences in grain size, sorting, and clay content. Permeability anisotropy, expressed as the ratio of vertical to horizontal permeability (kv/kh), is a critical parameter: a kv/kh of 0.01 in a shale-baffled reservoir versus 0.1 in a clean sandstone can change the predicted oil recovery factor by 10 percentage points or more. Core-derived vertical permeability measurements, often taken from whole-core CT-scanned plugs analyzed using digital rock physics, provide the basis for kv/kh assignment. The completed static model is then upscaled from the fine geological grid to a coarser dynamic simulation grid (typically 3 to 10 times coarser in all dimensions) using flow-based upscaling algorithms that preserve effective transmissibility rather than simply averaging cell values.

Data Inputs to the Reservoir Characterization Model

No single data type provides sufficient information to characterize a reservoir alone; the value of a reservoir characterization model lies in its integration of multiple independent and complementary measurements.

3D seismic: Provides the only continuous spatial sampling of the subsurface between wells. Structural interpretation defines the trap geometry; amplitude and acoustic impedance inversion constrain facies and porosity distribution between well control points. Spectral decomposition, coherence (similarity), and curvature attributes identify channels, faults, and fracture zones invisible to well control. Fault seal analysis, conducted within the characterization workflow, determines which faults are likely to act as flow barriers versus conduits during production. The seismic bandwidth limits vertical resolution to approximately one-quarter wavelength (typically 10 to 20 m / 33 to 66 ft at reservoir depths), so thin beds below this limit are invisible to seismic and must be inferred from well data and depositional models.

Well logs: The gamma-ray log identifies sand-shale boundaries and clay content; the density and neutron-porosity logs determine total and effective porosity; the resistivity log identifies the hydrocarbon-water contact and water saturation; the sonic log provides velocity data for seismic-to-well tie and for rock physics modeling; the photoelectric (PE) log and spontaneous potential (SP) log contribute to lithology discrimination. LWD (Logging While Drilling) tools provide real-time versions of these measurements while drilling, enabling geostopping decisions in thin reservoirs. MWD measurements (inclination, azimuth) define the well trajectory precisely so that log depth correlates accurately to the structural model.

Core data: Conventional core plugs (38 to 63 mm / 1.5 to 2.5 in diameter) provide direct measurements of porosity, absolute permeability (both horizontal and vertical), relative permeability curves (oil-water and gas-oil), capillary pressure, grain density, and wettability. Special core analysis (SCAL) measurements (mercury injection capillary pressure, MICP; centrifuge and porous-plate relative permeability; nuclear magnetic resonance NMR pore size distribution) are essential inputs to both the static model and the reservoir simulation. In tight and unconventional reservoirs (Montney, Duvernay, Marcellus), nanopore analysis using scanning electron microscopy (SEM), FIB-SEM (focused ion beam), and microCT scanning reveals pore throat sizes below 100 nm (0.1 micron) that are invisible to conventional mercury injection and control gas adsorption and transport in organic-rich shale matrix.

Production data: Well test pressure transient analysis (PTA) provides effective reservoir permeability (often dramatically different from core plug values at centimeter scale) and average reservoir pressure. Production logs (production logging tools) profile inflow along horizontal wellbores, identifying which intervals contribute flow and which are non-productive. Interference tests between wells constrain interwell connectivity, helping validate whether faults in the static model act as barriers or baffles. History matching, in which the dynamic simulation is tuned until its predicted production history matches observed well rates and pressures, provides a posterior constraint on static model parameters, particularly permeability multipliers and fault transmissibility.