Data Analytics for Applied Sciences

We perform developments in data analysis and artificial intelligence, to solve applied science problems of our clients. Our approches involve:

-Inference of parameters in complex structured models and diverse data types.
-Image processing.
-Convolutional neural networks.
-Artificial intelligence supported by Bayesian networks.
-Hi-performance numerical modeling of natural phenomena.
-Visualization in 3-dimensions and spatial statistics.

Seismic Inversion

Our seismic inversion projects include the following activities:

-Seismic data preparation in the pre-stack domain.
-Well-log analysis and characterization of the relationship between elastic (Vp, Vs, density) and reservoir parameters (Vshale, porosity, saturation).
-Seismic modeling and wavelet-estimation.
-Seismic elastic inversion: we have the most advanced seismic data inversion techniques for the joint estimation of elastic parameters and mass density. Our methods include geostatistical constraints, petrophysical constraints, estimation of the source wavelet spatial variations, multi-component and multi-azimuth inversion if required.
-Our seismic inversion methods can be implemented in time or depth domain, and with variable prior wavelet functions.
-Estimation of reservoir properties and probability indicators for fluid, lithology, porosity and permeability.
-Map generation for the description of sand and hydrocarbon spatial distribution.

Petrophysical and Geostatistical Inversion

From estimated elastic parameters and mass density we have various methods for the estimation of reservoir properties, such as shale volume fraction, total porosity, and water saturation.
The petrophysical inversion includes the following activities:
- Well-log analysis and characterization of the relationship between elastic (Vp, Vs, density) and reservoir parameters (Vshale, porosity, saturation).
- Elaboration of statistical models to relate elastic and reservoir models.
- Elaboration and calibration of petrophysical models (based on fundamental petrophysical relationships) to relate elastic and reservoir properties.
- Estimation of reservoir properties and probability indicators for fluid and lithology.
- Map generation for the description of sand and hydrocarbon spatial distribution.
- Our methods allow for the consecutive or joint seismic and rock-physics inversion, according to the convenience of the area and data.

Geostatistical seismic inversion combines the hi-resolution information of well-logs with the seismic data into a unified inversion method.
This method increases the accuracy and vertical resolution of the seismic inversion:
- The solution assures appropriate matching to all wells used in the inversion jointly with a full explanation of the seismic data.
- Vertical resolution increases in the zone of influence of the available wells according to the characterized covariance ranges.
- Out of the appropriate spatial ranges, the well information does not influence the estimation avoiding the bias of the results.
- Well conditioning could be made from reservoir property and/or elastic reservoir well-logs.
- Joint estimation of mass density, elastic parameters, and reservoir properties, conditioned to wells.

Stochastic Inversion

Stochastic inversion performs a full uncertainty analysis for fluid and lithology based in the pre-stack seismic data and petrophysical models, with a focus on potential prospect areas. This include:
- Well-log analysis and characterization of the relationship between elastic (Vp, Vs, density) and reservoir parameters (Vshale, porosity, saturation).
- Elaboration of statistical models to relate elastic and reservoir parameters.
- Definition and calibration of petrophysical models (based in fundamental petrophysical relationships) relating elastic and reservoir properties.
- Generation of a large number of realizations of the reservoir parameters that honor the seismic data and petrophysical constraints within uncertainties.
- Estimation of various statistics, probability densities, and probability distributions for the reservoir parameters (V-shale, porosity, and saturation).
- Probability densities for net sand and pay.

Time-Depth Conversion

Reflection time after migration and true vertical depth should be accurately related via a 3D time-depth model for and adequate combination of the seismic derived properties with the well-log data and information.
The process of transforming the vertical dimension from time to depth accomplishes this task; the process involves:
- Construction of the prior compressional velocity model in 3D based on well information and interpreted horizons.
- The preliminary transformation from time to depth domain of horizons and seismic derived properties.
- Quantification of residual adjustments to well-tops and well-logs.
- Interpolation of the depth adjustments using a geostatistical method and update of the velocity model.
- Final time to depth transformation applied to the seismic data, seismic derived properties, and interpreted horizons.

Integrated Interpretation

Definition of plays and prospects requires the support of multiple types of information and data regarding the area geology, petroleum systems, fault systems, porosity, lithology and fluid seismic characterization, and hydrocarbon trap identification. We integrate into the interpretation the methods providing the maximum information from the seismic data reflection amplitudes and frequencies, based on the seismic derived reservoir properties.
This involves:
- Seismic interpretation of the relevant horizons and structures.
- Definition of the major plays in the area.
- Elaboration of maps for the expected thickness of the target strata based on the seismic characterization.
- Seismic frequency decomposition for the analysis of geomorphology associated with the sedimentary and transport processes.
- Analysis and delimitation of potential oil and gas prospects.

Resource Estimation

We quantify each prospect and the cumulated resources based on the accepted probability models, in the risked and unrisked scenarios. The success in hydrocarbon discovery is the result of concurrent factors, such as the charge, seal, reservoir, and trap, which are evaluated for the prospects. After randomization of the reservoir parameters involved in the volume estimation with the corresponding probability distributions, the extractable oil and/or gas volumes are modeled and calculated with a Monte Carlo technique.
This involves:
- Success probability modeling.
- Reservoir surface, gross and net thicknesses, gross rock volume, effective porosity, and saturation modeling.
- Extractable volume modeling.
- Prospect ranking according to expectations.

Seismic Processing

Our seismic processing projects involve an integrated approach that takes into account the client's objectives:
- Preservation of the seismic reflection signal, including relative amplitudes, for adequate imaging, seismic inversion, and reservoir property estimation.
- Precise analysis of the terrain elevations and static corrections for vertical seismic data positioning.
- Stacking and migration velocity estimation accounting for both the seismic data and the area geology.
- Approaching the processing sequence with the objectives of the interpretation and reservoir characterization in mind.
- Supported by the use of state-of-the-art technology together with the experience and knowledge of our experts.