Riser Response Model Predicts Wellhead Fatigue
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for predicting wellhead fatigue in offshore drilling operations are inefficient, requiring costly sensors and lacking real-time data display capabilities, with conservative mathematical models failing to accurately estimate stress and strain on wellheads.
Innovation Solution
A method using machine learning algorithms, specifically neural network regression, to generate a riser response model from environmental, tension, and mud data, predicting wellhead fatigue without the need for sensors, and visualizing the results through a graphical user interface.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If conservative mathematical models are used to estimate wellhead stress and strain, then the prediction process is simple, but the accuracy of stress and strain estimation is insufficient
Solution Approach 1:
The patent replaces traditional conservative mathematical models with machine learning algorithms (neural networks) that process environmental data, tension data, and mud data to predict riser response and wellhead fatigue. This substitution enables more accurate stress and strain estimation while maintaining computational efficiency through trained models.
Solution Approach 2:
The patent transforms the prediction approach by changing from fixed conservative parameters to dynamic parameters derived from multiple data sources (environmental conditions, tension measurements, mud properties). The machine learning model learns optimal parameter relationships from training data, enabling accurate stress and strain prediction under varying operational conditions.
2Loss of information
If sensors are deployed on or near wellheads to measure displacement for stress and strain calculation, then real-time data can be obtained, but the cost and complexity of the system increases
Solution Approach 1:
The patent creates a virtual model (digital twin) of the wellhead system that replicates the physical wellhead's behavior. By training a neural network on historical data and environmental parameters, the model generates accurate predictions of stress and strain without requiring physical sensors on the wellhead, thereby reducing system complexity and cost while maintaining real-time monitoring capabilities.
Solution Approach 2:
The patent introduces an intermediate computational model that processes readily available data (environmental conditions, tension measurements, mud properties) to infer wellhead stress and strain. This intermediary approach avoids the need for direct sensor installation on the wellhead while still providing real-time predictive information through machine learning algorithms.
3Ease of operation
If existing technologies are used for predicting wellhead fatigue, then the implementation is straightforward, but the prediction accuracy and real-time capability are lacking
Solution Approach 1:
The patent implements preliminary action by training the neural network model offline using historical data before deployment. The model learns from past environmental conditions, tension data, and mud properties to establish predictive relationships. Once trained, the model can rapidly predict wellhead fatigue in real-time without complex computations during operation, maintaining ease of use while improving accuracy and real-time capability.
Data Source
AI summary
Methods, systems, and storage media for predicting physical changes to a wellhead coupled to a riser in an aquatic volume of interest are disclosed. Exemplary implementations may: obtain training data; obtain a machine learning algorithm; generate a riser response model by applying a machine learning algorithm to the training data; store the riser response model, obtain target environmental data, target tension data, and target mud data, generate predicted riser response data, transform predicted riser response data, generate a representation, and display the representation.