Predictive Model for Drilling Fluid Flow Back Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Drilling operations face challenges in predicting and managing fluid flow back, which can lead to wellbore kicks, a potentially catastrophic event, due to the complexity of fluid dynamics and pressure variations in drilling processes across different geological environments.
Innovation Solution
A predictive model is developed using historical drilling data from various geographic locations to forecast fluid flow back measurements, triggering alarms when deviations exceed predicted values, thereby enabling early intervention and preventing wellbore kicks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fluid is circulated throughout the wellbore annulus during drilling, then drilling operations can proceed, but friction pressure and hydrostatic pressure cause fluid flow back that pushes fluid back to the surface
Solution Approach 1:
The system performs preliminary action by predicting fluid flow back measurements before they occur. The predictive model analyzes historical drilling data and current operational parameters to forecast flow back values, allowing operators to take preventive measures before actual flow back events happen, thus maintaining drilling continuity while mitigating the harmful effect.
Solution Approach 2:
The system implements feedback by continuously monitoring actual fluid flow back measurements and comparing them against predicted values. When deviations exceed thresholds, the system generates alarms that provide feedback to operators, enabling real-time adjustments to drilling parameters to prevent catastrophic events while maintaining operational productivity.
2Reliability
If a predictive model is developed using historical drilling data from various geographic locations, then early identification of wellbore kicks is enabled, but data processing complexity increases
Solution Approach 1:
The predictive model achieves universality by being trained on historical drilling data from multiple geographic locations and different wellbore conditions. This multi-functional training dataset enables the single predictive model to accurately detect wellbore kicks across diverse geological environments, improving reliability without requiring separate models for each location.
Solution Approach 2:
The system introduces an intermediary computational layer that processes raw drilling data through the predictive model to generate flow back predictions. This intermediary model acts as a mediator between complex historical data and simple alarm triggers, managing the complexity internally while presenting a straightforward detection interface to operators.
3Reliability
If real-time fluid flow back measurements are compared to predicted values with alarm triggers, then wellbore kicks can be prevented, but false alarms may occur
Solution Approach 1:
The system applies partial action by implementing multiple alarm trigger conditions rather than a single threshold. It evaluates both predicted flow back values and actual measurements against configured thresholds, generating alarms only when specific deviation criteria are met. This partial application of alarm logic reduces false alarms while maintaining reliable wellbore kick detection.
Data Source
AI summary
A computing device configured to determine when an alarm is triggered for a drilling operation is provided. Measured drilling data that includes a value measured for an input variable during a previous connection event of a drilling operation is received. A predicted value for a fluid flow back measure is determined by executing a predictive model with the measured drilling data as an input. The predictive model is determined using previous drilling data that includes a plurality of values measured for the input variable during a second drilling operation. The second drilling operation is a previous drilling operation at a different geographic wellbore location than the drilling operation. A fluid flow back measurement datum determined from sensor data is compared to the determined predicted value for the fluid flow back measure. An alarm is triggered on the drilling operation based on the comparison.