Back to results
US 10,851,636 B1Granted
Systems and methods for analyzing resource production
AssigneeDataInfoCom USA, Inc.
InventorsAtanu Basu, Daniel Mohan, Chun Wang, Frederick Venter
Life of this patent
Estimating the term…
Abstract
A method of drilling a well includes receiving depth specific drill parameters associated with drilling wells within a field; receiving drilling report entries associated with drilling the wells within the field; applying a report analysis engine to the drilling report entries to determine a drilling event; determining a field-specific drilling model utilizing the depth specific drill parameters and drilling event; prescribing a drilling recipe using a drilling analysis engine applied to the field-specific drilling model; and drilling the well in accordance with the drilling recipe.
Claims20
A method of drilling a well, the method comprising:
- receiving one or more unstructured drilling reports associated with one or more previously drilled wells within a field, the field having field characteristics including geological, geophysical, petrophysical, lithological, and mineralogical characteristics;
- analyzing, using a report analysis engine, the one or more unstructured drilling reports to detect one or more drilling events associated with the one or more previously drilled wells within the field, the detected drilling event being in the form of structured data;
- receiving drilling parameters associated with the one or more previously drilled wells within the field;
- based on the detected drilling event and the drilling parameters, building a plurality of field-specific models, each of the plurality of field-specific models being specific to the field characteristics of the field;
- determining a test set of information associated with a further one or more previously drilled wells within the field, and validating each of the plurality of field-specific models using the test set of information;
- based on results of the validation, selecting one of the plurality of field-specific models;
- training, using a machine learning technique, a drilling analysis engine based on the selected field-specific model, the drilling analysis engine comprising at least one of the following:
- a regression model, a classification model, a Markov chain, a time series model, a state space model, a Bayesian model, an artificial neural network model, an autoregressive integrated moving average (ARIMA) model, a seasonal ARMIA (SARIMA) model, an autoregressive conditional heteroskedasticity (ARCH) model, a general autoregressive conditional heteroskedasticity (GARCH) model, a genetic algorithm, a polynomial fit, a random forest model, a K-neighbors model, and a Bayesian ridge model;
- receiving information associated with drilling or designing a new well in the field;
- prescribing, using the drilling analysis engine and based on the information associated with drilling the new well, a drilling recipe for the new well, wherein the drilling recipe includes a number of segments and associated depths;
- and
- drilling the new well in accordance with the drilling recipe.
9 more elements
A method for drilling a well, the method comprising:
- receiving one or more unstructured drilling reports associated with one or more previously drilled wells within a field, the field having field characteristics including geological, geophysical, petrophysical, lithological, and mineralogical characteristics;
- analyzing, using a report analysis engine, the one or more unstructured drilling reports to detect one or more drilling events associated with the one or more previously drilled wells within the field;
- based on the detected drilling event, building a plurality of field-specific models, each of the plurality of field-specific models being specific to the field characteristics of the field;
- selecting one of the plurality of field-specific models;
- training, using a machine learning technique, a drilling analysis engine based on the selected field-specific model, the drilling analysis engine comprising at least one of the following:
- a regression model, a classification model, a Markov chain, a time series model, a state space model, a Bayesian model, an artificial neural network model, an autoregressive integrated moving average (ARIMA) model, a seasonal ARMIA (SARIMA) model, an autoregressive conditional heteroskedasticity (ARCH) model, a general autoregressive conditional heteroskedasticity (GARCH) model, a genetic algorithm, a polynomial fit, a random forest model, a K-neighbors model, and a Bayesian ridge model;
- prescribing, using the drilling analysis engine and based on the selected field-specific model, a drilling recipe for a new well, wherein the drilling recipe includes a number of segments and associated depths;
- and
- drilling the new well in accordance with the drilling recipe.
6 more elements
Dependencies are inferred from the claim text.