Jason Geoscience Workbench

Jason Geoscience Workbench (JGW) is an extensive toolset for complex seismic reservoir characterization used in the analysis of oil field data for improved production planning and future well placement.
Introduction
The Jason Geoscience Workbench (JGW) integrates geological, geophysical, petrophysical and rock property information into a single consistent model of the subsurface. Seismic, well logs and statistical data are directly input into JGW and used to calculate the geological model.
Geoscientists select the combination of tools they require for each characterization. JGW supports pre- and post-stack deterministic and geostatistical inversion, 3D visualization, wavelet estimation, volume interpretation and body checking, rock physics, petrophysics, geomodeling, and velocity modeling, in single and multi-CPU environments. JGW is used to determine the best locations for in-fill wells, optimize production plans and improve reserves estimates.
Tools within JGW include EPlus, FunctionMod, Largo, EarthModel FT, VelMod, Wavelets, InverTracePlus, RockTrace, StatMod MC, and RockScale. These tools are used in various combinations for a range of applications up to and including complete seismic to simulation studies.
E
This lithologic interpretation environment is designed to properly handle both geologic and geophysical data in a smooth, seamless, efficient manner, allowing better reservoir management decisions.
E provides the following functions:
<ul>
<li>3D Volume View and Body Checking, for rapid interpretation and detailed quantitative and qualitative analysis of large data sets, yielding a map of productive reservoir components including a net pay map for the main pay interval. The Body Checking system is a full 3D volume interpretation and analysis system that relies on deterministic rock property relationships to establish what is and what isn't reservoir quality rock.</li>
<li>Section View, displaying time or depth data, along inlines, cross lines, or any multi-segment arbitrary line, which the user can draw with the mouse in the base map. The section view can display zero-mean seismic and absolute value rock property data simultaneously, and supports fault and horizon interpretation.</li>
<li>Map View, serving both data selection and data viewing functions, and supporting editing functions for horizons, including interpolation, extrapolation, and smoothing with a number of filter operators.</li>
<li>Well Log Editor, integrated with the wavelet estimation process so that log synthetics automatically refresh whenever the selected wavelet changes, and the wavelet is flagged for re-estimation whenever the editing of the log changes.</li>
<li>Cross Plots and Histograms, used to analyze data or relationships between data.
<li>Well Log View, displaying multiple well logs from various wells independent of their X, Y location.</li>
<li>Data Links, including seismic surveys, interpreted seismic horizons, velocity data, well logs and tops from GeoFrame, Landmark, FloGRID, 123DI.</li>
</ul>
FunctionMod
This model manipulation and arithmetic tool allows calculation on JGW data files and is also used as the calculation engine in Largo and the more complicated functions in Map View.
Complex functions can be created, including results that cascade into other computations, and functions that branch to one or more sub-functions. FunctionMod comes with a number of preset functions, including mean, median, root mean square, sum, standard deviation, and over 50 others. Boolean logic and vector operations are supported.
Users can create transforms, which are portable functions that reference a type of file (e.g., a trace file with impedance traces). Files assigned to the transform will have the function applied to them. Complex nested functions can also be defined.
Application examples include:
<ul><li>Calculating lithology logs using Boolean logic such as "If the resistivity is greater than X and the Gamma Ray is less than Y, assign pay sand. If the Resistivity is less than X and the Gamma Ray is less than Y, assign wet sand. Otherwise assign shale."</li>
<li>Calculating derived parameters such as and Lame parameters from P-Impedance and S-Impedance following a RockTrace™ Simultaneous AVA inversion.</li>
<li>Calculating mean and standard deviation volumes following a StatMod MC geostatistical inversion.</li>
<li>Calculating porosity and lithology probability models from StatMod MC results.</li></ul>
Largo
Largo integrates well log data analyses and rock physics modeling with seismic analysis. Largo is the processing and quality control link between petrophysics, rock physics, and seismic inversion.
Largo contains a set of rock physics modeling tools for the calibration and synthesis of Vp, Vs, and density logs for fluid substitution modeling. It supports multiple well processing and well zoning, and can utilize lithology types and generate them from a crossplot analysis of logs. All Largo log plots and crossplots automatically update when the well data changes.
Largo takes a fundamental modeling approach. It first uses theoretical rock models to derive the effective elastic properties from rock and fluid composition and mineral parameters, and then calibrates the model parameters by comparison of the synthetic to the available density and compressional sonic log. A good comparison between the logs allows the parameters thus derived to be applied to shear sonic synthesis. If measured shear sonic is available, then that data can also be used for calibration.
Largo contains the following rock physics algorithms:
<ul>
<li>Xu & White’s model</li>
<li>A faster approximation of Xu & White’s model</li>
<li>A self-consistent model following Berryman</li>
<li>A model for dispersed clay</li>
<li>A grain supported model</li>
<li>A matrix supported model</li>
<li>Critical porosity model</li>
<li>Greenberg & Castagna’s relation</li></ul>
Largo also has a number of simple averaging methods (e.g., Wyllie, Voigt, Reuss, Hashin-Shtrikman). In addition, fluid properties can be estimated based on the formulas from Batzle & Wang.
The rock models are based on the Kuster-Toksöz inclusions model where pores with a given aspect ratio are mixed into a solid matrix. The empty pores are then filled with an effective fluid, assuming low sonic frequencies (Gassmann theory)
. The models compose the solid matrix of two rock members. In case more than two rock members are present, the algorithms can be applied interactively to any desired level of complexity.
EarthModel FT
EarthModel FT builds models that support both geophysical and reservoir modeling and simulation workflows. It supports complex faulting and geology, horizontal wells, and also supports model building in the time domain and use of velocities as trends.
EarthModel FT includes UpdateAbility, which enables it to recalculate all direct and indirect results in the model when parameters change. For example, if nodes of a fault stick are edited, the truncated surfaces are all updated automatically.
Wells are not 'verticalized' in each layer, so that deviated wells, including horizontal wells, can be properly in the EarthModel FT approach. Each well data point is mapped to a cell in the geologic model so that the spatial relationship between the well tracks is correctly calculated.
EarthModel FT uses standard geostatistical algorithms with distance and power parameters to populate the reservoir models from the well logs. Secondary variables can be used on the interpolation process. There is no separate calibration, and no additional calculations needed.
VelMod
VelMod builds 3D velocity models from stacking velocities and calibrates these models to well control. These models can be used for time-depth conversion or low-frequency control during seismic inversion. This is important if there are any strong lateral variations in the velocities away from well control. If unaccounted for, these variations may have a strong impact on the lateral prediction of formation pressure and on the characterization of the reservoir.
The Velocity Conversion module uses the Dix formula to calculate the interval or average velocities from the stacking velocity data. Input velocity data (stacking, interval or average) can be converted into any of the other two velocity data types. To convert the velocities, the time datum of the stacking or average velocities must be entered. The processing datum can be entered as a time horizon.
The Property Conditioning module is used to edit and smooth any data set (including velocity data). Typically this will be the velocity data previously created in the Velocity Conversion module, but it can be any property data set.
Scatter correction and lateral smoothing are applied to ensure the generation of a well-behaved model. The model data can be in time or depth. A 'solid model' (created with EarthModel) is used to segregate the imported data points into layers.
Lateral smoothing is applied to reduce trace to trace variations in the data. The length of the smoothing operator is specified as a number of traces and works on a trace by trace basis. The smoothing operator generates a scale factor for each trace at each layer based on the average values of the trace and the average values of the traces present in the smoothing operator. This scale factor is then applied to all the samples in that layer.
The Time Depth Conversion (TDC) module of VelMod uses the velocity model to convert any horizon data or seismic/property data from time to depth or depth to time.
Wavelets
Accurate wavelet estimation is a key factor in the success of any seismic inversion. The inferred shape of the seismic wavelet may strongly influence the seismic inversion results and therefore subsequent assessments of the reservoir quality.
An inversion project is a two-step process. In the first step, the seismic wavelet is estimated. In the second step, that wavelet is used to estimate the seismic reflectivity. The final acoustic impedance is derived from that estimated seismic reflectivity. Therefore if the wavelet is incorrect, the seismic inversion results will be invalid.
The Wavelets module encompasses several methods that can be applied to one or more wells simultaneously:
<ul>
<li>Model-driven wavelet phase and amplitude spectrum estimation at well control.</li>
<li>Wavelet amplitude spectrum estimation with and without well control. </li>
<li>Wavelet constant-phase spectrum estimation with and without well control. </li>
<li>AVO (AVA) wavelet estimation for input partial stacks. </li></ul>
The wavelet amplitude and phase spectra can be estimated in a statistical manner from the seismic data only or with usage of well control. The availability of one or more wells with sonic and density logs opens an additional option for the process of estimating the wavelet. Statistical techniques are used to obtain an initial wavelet, which is then used to generate an initial well synthetic. When the estimated (constant) phase of the statistical wavelet is consistent with the final result from the model-driven method, the wavelet estimation converges more quickly than when starting with a zero phase assumption. Minor edits and 'stretch and squeeze' can be applied to the well to better align the events, and then the wavelet phase and amplitude spectra are estimated.
Accurate wavelet estimation requires accurate tie of the impedance log to the seismic. Errors in well tie can result in phase or frequency artifacts in the wavelet estimation. Therefore, the Wavelets package is tightly integrated with the EPlus Well Log Editor. If the wavelet is re-estimated or refined, the displays in the well log editor are automatically updated. Similarly, if the well is edited, an information message prompts the user to re-estimate the wavelet with the new well tie.
InverTrace
InverTrace transforms seismic data to an acoustic impedance log at every trace. Acoustic impedance is a property of the rock layer itself, unlike seismic amplitude, which is a property of the interface between two rock layers.
Acoustic impedance is used to produce more accurate and detailed structural and stratigraphic interpretations than can be obtained from seismic (or seismic attribute) interpretation. In many geological environments acoustic impedance has a strong relationship to petrophysical properties such as porosity, lithology, and fluid saturation. Moreover, the acoustic impedance models are more readily understood (versus seismic attributes) by all members of the asset team, and can enable better overall communication within the team.
The Constrained Sparse Spike Inversion (CSSI) algorithm is the centerpiece of InverTrace and is the state of the art implementation of non-linear sparse spike technology. InverTrace produces four high quality acoustic impedance volumes from full or post-stack seismic data: Full Bandwidth Impedance, Band-Limited Impedance, Reflectivity model, and Low frequency component.
RockTrace
RockTrace quantitatively integrates well log elastic rock properties and AVO (AVA) seismic to produce calibrated quantitative 3D volumes of rock properties.
In RockTrace, the objective is to solve for shear impedance (Zs) and density in addition to acoustic impedance, the focus of InverTracePlus, so the constraints are set for all three parameters independently. The parameterization can be in terms of triplets of elastic parameters:
<ul>
<li>Zp, Zs and density </li>
<li>Zp, Vp/Vs and density </li>
<li>P-Sonic, S-Sonic and density </li>
<li>Vp, Vs and density </li></ul>
When applied in global mode a spatial control term is added to the objective function and large subsets of traces are inverted simultaneously. The RockTrace seismic inversion algorithm takes multiple angle-stacked seismic data sets and generates three elastic parameter volumes as output.
The resulting elastic parameters are real rock properties that can be directly related to reservoir properties. The full Knott-Zoeppritz equations are used and there is full allowance for amplitude and phase variations with offset. This is done by deriving unique wavelets for each input partial stack. A near stack wavelet can be used as the starting point for estimating the far angle or offset wavelet.
In addition, the elastic parameters themselves can be directly constrained during the seismic inversion, and rock physics relationships can be applied constraining pairs of elastic parameters to each other. Final elastic parameter models optimally reproduce the input seismic, as this is part of the seismic inversion optimization.
StatMod MC
StatMod MC combines seismic and well data to form a 3D model that has both high vertical and lateral resolution. Because StatMod MC relies on geostatistics in addition to advanced statistical physics, the models it generates include shapes that are geologically plausible and uncertainty that is quantified.
probability density functions (PDF))are used to represent each source data. These PDFs define the overall expected geological texture using variograms, and the likelihood a given value would appear in a given location using histograms. The histograms and variograms come from a combination of analysis, modeling and geological insight.
The various PDFs are merged using Bayesian inference techniques. The resulting posterior PDF represents all known and assumed information. Bias in the process is avoided by allowing the algorithm to assign weights automatically.
A customized Markov Chain Monte Carlo algorithm creates samples from the posterior PDF. These sample are statistically fair and of high detail, accurate and realism.
StatMod MC simultaneously solves for impedance and lithofacies. There are significant synergies in this approach, improving the model and reducing modeling time. The resulting models can be used to jointly cosimulate various rock properties. The process is iterated until a best match to all data is determined.
Uncertainty is estimated by generating a variety of realizations using random seeds and targeting weak points in the data. At the end of this process, uncertainty should be better understood, assisting in development risk assessment.
RockScale
RockScale transfers 3D models from orthogonal (seismic) grids to corner point grids. Its zonal adjustment technique ensures proper mapping of the seismic property samples into the correct stratigraphic layers in the corner point grid (CPG).
Flow geometry is preserved for NTG, Porosity, and other volumes. Seismic derived property models are rescaled directly into the CPG models. The rescaled CPG models can be used directly or as 3D trend models for co-simulation.
Incorporation of seismic data can improve the predictive capability of their reservoir models. The limitations have been in using the full 3D seismic or seismic-derived petrophysical properties to build the model, rather than simply reducing the 3D data to 2D maps of the average attribute in the reservoir layer.
RockScale uses the new stratigraphic model grid (SMG) along with CPGs to manage the transform of the properties (from one grid to the other) and ensure that they are structurally and stratigraphically correct. This includes the management of detailed structural models with many faults and sub para-sequences.
Additionally, CPGs can be rescaled into orthogonal grids to produce seismic synthetics for model QC purposes. This workflow can be used for evaluating 4D seismic studies and their relationship to fluid movement and detection using seismic data.
RockScale handles critical issues in preparing multi-million cell models from seismic and properly preserves geometry (including flow geometries) and volumetrics for any volumes prepared in a seismic gridding system, including net to gross, effective porosity, and impedance (shear and acoustic).
Zonal adjustment is RockScale’s method for properly incorporating 3D properties derived from the seismic data into flow simulation models. It solves the traditional problems of resampling, regridding, and upscaling.
See Also
Seismic to Simulation
Seismic Inversion
 
< Prev   Next >