GeoSoftware software

GeoAI encompasses a novel methodology for seismic reservoir characterization with limited well control, speeding up reservoir property predictions with a rock physics driven machine learning technique. Rock Physics theory and statistical simulations generate synthetic data for various geological scenarios. A simplified machine learning approach employs Convolutional Neural Networks (CNN) estimating multiple rock property volumes in a greatly simplified workflow.

In standard supervised machine learning approaches, the seismic-to-rock property relationship is learned using available data. These methods, particularly deep learning, depend on having enough labeled data to adequately train the neural network. WellGen overcomes this challenge by generating synthetic data, simulating many pseudo-wells based on existing well statistics and rock physics modeling.

RockTrace® quantitatively integrates well log elastic rock properties and AVA seismic to produce calibrated, quantitative 3D volumes of rock properties. This module determines key reservoir properties, P-impedance, S-impedance, density and Vp/Vs, through the inversion of partial offset or angle stacks. RockTrace AVO inversion integrates sparse spike inversion technology with sophisticated low-frequency modeling techniques to produce the most advanced deterministic estimates of reservoir properties in the industry.

StatMod® performs single stack geostatistical reservoir characterization to produce multiple predictive fine-scale reservoir models that are consistent across all pertinent geoscience domains. This degree of cross-domain consistency ensures that reservoir models are realistic and maximizes the value of all measured data and inferred information. In addition, these reservoir models remain accurate away from the available well control where reliability of conventional stochastic models typically suffers. StatMod is relevant where single parameter like P-Impedance is sufficient to distinguish the facies of interest.

Save time and resources with Machine Learning. PowerLog can connect to any open-source distribution of Python and perform in-house advanced processing to save your company time and money. You can successfully apply machine learning and deep learning techniques with this tool including: Facies classification, supervised and unsupervised, Unbalanced sampling algorithms to eliminate bias, Data loading, analysis, and migration, Automated log editing for outlier detection, synthetic curve generation, and automated curve substitution