PetroReal-EVAL utilizes proprietary Deep Machine Learning and Artificial Intelligence to mine large datasets to identify trends and anomalies reflecting changes in reservoir conditions or field operations. The information included in these “big data” volumes are primarily sourced from regional databases consisting of historical data associated with drilling, completion, and production operations that may be augmented with various technical studies. Artificial Intelligence and Deep Machine Learning technology utilize a proprietary time-dependent conditional quantile regression and predictive dynamic zonation (clustering) of individual attributes within a basin or region and non-parametric decline curve analysis.
Unlike conventional forecasting methods that generally use a few fixed parameters from a limited number of analog wells, our analytics incorporate an unlimited number of reservoir and production performance attributes which enables PetroReal-EVAL to reliably recognize and forecast subtle and abrupt changes in production patterns owing to variations in subsurface conditions, completion methods and potentially production operations as well.
AI Production Forecasts are derived using time dependent conditional quantile regression and predictive dynamic zonation of individual attributes along with a non-parametric decline curve analysis. Production curves are calculated for P10 through P90 probability cases.
Input attributes are reported for all PDP wells and AI predicted attributes are provided for any possible PUD location. Each selected well is located on a distribution of all occurrences for each attribute within the ML database. Knowing the position of a well within the distribution of key attributes provides Investors with insight on why wells perform like they do, and how they differ from producing wells elsewhere in a basin or trend.
AI Production Forecasts are derived using time dependent conditional quantile regression and predictive dynamic zonation of individual attributes along with a non-parametric decline curve analysis. Production curves are calculated for P10 through P90 probability cases.
Input attributes are reported for all PDP wells and AI predicted attributes are provided for any possible PUD location. Each selected well is located on a distribution of all occurrences for each attribute within the ML database. Knowing the position of a well within the distribution of key attributes provides Investors with insight on why wells perform like they do, and how they differ from producing wells elsewhere in a basin or trend.
Dynamic ”What-if” Modeling allows for the adjusting of individual attribute values to determine their impact on production. The dynamic modeling allows for rapidly evaluating the impact of key well and completion design attributes and how production may be optimized in new and existing wells. As shown in the example to the left, the horizontal well length of a PUD well was increased from 4448’ to 6500’ to match the mean horizontal length for the highest producing wells.
Dynamic ”What-if” Modeling allows for the adjusting of individual attribute values to determine their impact on production. The dynamic modeling allows for rapidly evaluating the impact of key well and completion design attributes and how production may be optimized in new and existing wells. As shown in the example to the left, the horizontal well length of a PUD well was increased from 4448’ to 6500’ to match the mean horizontal length for the highest producing wells.
Cluster Analysis provides the distribution of production forecasts across an entire basin or trend. AI analytics allow for not only determining remaining production volumes at any location but also evaluating the performance of individual operators and various drilling and completion attributes.
Such analysis also provides investors insight as to how their operating partners are performing as compared to their peer group and that best practices are being utilized.