Hydraulic Fracture Initiation Prediction Using XGBoost
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional analytical solutions for predicting breakdown pressure and fracture initiation in fracking wells fail to accurately account for three-dimensional complex configurations of perforated wellbores and rock damage behaviors, leading to uncertainties and inaccuracies in hydraulic fracturing treatments, particularly in deviated and cased-hole scenarios.
Innovation Solution
A machine learning approach using a computer-implemented method and XGBoost algorithm to predict hydraulic fracture breakdown pressure and fracture initiation, incorporating a wide range of parameters including mechanical properties, geometric parameters, and in-situ stresses, by training a model with a dataset from historical fracking operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional analytical solutions are used to calculate breakdown pressure, then the calculation process is simple and fast, but the accuracy is insufficient and cannot capture 3D complex configurations of perforated wellbores and rock damage behaviors
Solution Approach 1:
The patent replaces conventional analytical mechanical calculation methods with a machine learning-based predictive system. The ML model learns complex non-linear relationships from historical data, substituting traditional elasticity-based analytical solutions with data-driven predictions that capture 3D wellbore configurations and rock damage behaviors without requiring explicit mechanical modeling.
Solution Approach 2:
The patent transforms the prediction approach by changing from fixed analytical formulas to adaptive parameters learned from data. The system uses multiple input parameters including wellbore geometry, rock properties, and treatment conditions, with the ML model automatically determining optimal parameter relationships rather than relying on predetermined analytical expressions.
2Measurement precision
If more parameters are included in the prediction model to account for complex configurations, then the prediction accuracy improves, but the difficulty of determining breakdown pressure increases due to uncertainties in parameter definition
Solution Approach 1:
The machine learning model performs self-service by automatically learning parameter relationships and interactions from historical data without requiring explicit definition of complex parameter dependencies. The system handles parameter uncertainties internally through pattern recognition, reducing the burden on users to precisely define each parameter's relationship to breakdown pressure.
Solution Approach 2:
The system performs preliminary action by pre-processing and integrating multiple parameters during the training phase. Historical data including wellbore geometries, rock properties, and treatment conditions are pre-analyzed to establish parameter relationships before actual prediction, so that during deployment, the model directly uses these pre-established relationships without requiring real-time parameter determination complexity.
3Adaptability or versatility
If conventional analytical solutions are used, then fewer parameters are required, but the ability to account for rock damage behaviors and 3D wellbore configurations is insufficient
Solution Approach 1:
The machine learning model provides universal adaptability by handling multiple wellbore configurations (vertical, deviated, horizontal), different rock types, and various treatment conditions within a single framework. The system processes diverse parameter sets including geometric parameters, mechanical properties, and in-situ stresses, making it versatile across different fracking scenarios without requiring separate analytical solutions for each case.
Data Source
AI summary
Systems and methods include a computer-implemented method for predicting hydraulic fracture initiation. A fracking operations dataset is prepared using historical field information for fracking wells. A set of hyper-parameters is tuned for use in a machine learning algorithm configured to predict fracture initiation for new fracturing wells. The dataset is divided into training and test datasets. A regression algorithm is applied to train the training dataset and to validate with the test dataset. A target variable of a breakdown pressure for a new hydraulic fracturing treatment is determined. A prediction dataset is updated using at least the target variable. The training dataset is trained using a classifier of the machine learning algorithm. A prediction is made using the prediction dataset whether the new hydraulic fracturing treatment can be initiated or not. The breakdown pressure is incrementally adjusted, and the method is repeated until successful hydraulic fracture initiation is predicted.