Real-Time Pore Pressure Estimation Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurate estimation of pore pressure during well drilling is crucial to prevent non-productive drilling time and serious incidents like blowouts, but existing methods lack real-time monitoring and dynamic adjustment capabilities.
Innovation Solution
A method using machine learning to determine a real-time pore pressure log by analyzing existing data logs of surface drilling parameters, Logging While Drilling (LWD) data, and mud gas, which dynamically adjusts drill bit weight and mud properties to stabilize hydrostatic pressure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pore pressure estimation methods are used, then the drilling process can proceed with basic pressure monitoring, but real-time accurate pore pressure estimation is not achieved leading to delayed response to pressure changes
Solution Approach 1:
The system performs preliminary actions by pre-processing drilling data (weight on bit, rotary speed, mud flow rate, gas cuttings) and maintaining a database of formation characteristics before drilling encounters pressure changes. This allows the machine learning model to rapidly estimate pore pressure without delay when real-time data becomes available.
Solution Approach 2:
The system implements continuous feedback by monitoring drilling parameters in real-time, comparing actual measurements with expected values based on formation models, and dynamically adjusting pore pressure estimates. The system feeds back pressure estimates to drilling control systems for immediate mud weight adjustments.
2Reliability
If real-time monitoring and dynamic adjustment of drill bit weight and mud properties is implemented, then drilling safety and efficiency are improved, but the complexity of the drilling system increases
Solution Approach 1:
The system achieves multi-functionality by using a single integrated machine learning model that processes multiple drilling parameters (weight on bit, rotary speed, mud flow rate, gas cuttings) simultaneously to estimate pore pressure. This single system performs data acquisition, processing, analysis, and control functions that would otherwise require separate systems.
Solution Approach 2:
The system implements self-service through autonomous machine learning algorithms that automatically analyze drilling data, detect pressure changes, and recommend control adjustments without requiring constant human intervention. The system self-calibrates using historical drilling data and formation models.
3Measurement precision
If machine learning algorithms are used to analyze multiple data logs simultaneously, then pore pressure estimation accuracy is enhanced, but the computational requirements and data processing complexity increase
Solution Approach 1:
The system segments the complex data processing task by dividing drilling data into distinct parameter streams (weight on bit, rotary speed, mud flow rate, gas cuttings) that are processed separately through dedicated sensors and preprocessing modules before being integrated by the machine learning model. This modular approach reduces computational complexity.
Solution Approach 2:
The system introduces intermediary components including data acquisition modules that standardize sensor inputs, preprocessing algorithms that filter and normalize data streams, and feature extraction layers that convert raw data into meaningful parameters for the machine learning model. These intermediaries simplify the computational burden on the core estimation algorithm.
Data Source
AI summary
A method for determining a real-time pore pressure log of a well in a reservoir, including the steps: storing existing data logs of surface drilling parameters, logging while drilling (LWD), and mud gas of existing wells in a database, storing existing pore pressure logs of the existing wells in the database, wherein the existing pore pressure logs correspond to the existing data logs, determining a relationship between the existing data logs and the existing pore pressure logs, drilling a new well into the reservoir, determining new data logs of surface drilling parameters, LWD, and mud gas of the new well while drilling the new well, inputting the new data logs of the new well into the relationship while drilling the new well, determining a real-time pore pressure log of the new well by outputting an estimated pore pressure at a certain depth by the relationship while drilling the new well.