Proxy Model for Hydraulic Fracturing Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wellsite operation management systems, such as those for hydraulic fracturing, face challenges in accurately representing real-time conditions due to complex first principles models that require significant computing resources and are slow to update, leading to delays and inefficiencies in operations.
Innovation Solution
The implementation of proxy models that represent first principles models, allowing for real-time control and optimization of hydraulic fracturing operations by simulating and updating model outputs based on measured data, using techniques like standard model predictive control and fuzzy logic inference to adjust setpoints and optimize operational parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex first principles models are used to represent wellsite operations, then accuracy of real-time representation is improved, but computational speed and update frequency deteriorate
Solution Approach 1:
The patent creates simplified proxy models that copy the essential behavior of complex first principles models. These proxy models are trained offline using historical data and model outputs, then deployed for real-time control decisions. The proxy models replicate the input-output relationships of the complex models without requiring the full computational complexity, thus achieving both accuracy and speed.
Solution Approach 2:
The patent segments the modeling task into two distinct phases: an offline training phase where complex first principles models are used to generate training data, and an online execution phase where simplified proxy models make real-time predictions. This segmentation allows the system to leverage the accuracy of complex models during training while achieving the speed of simple models during operation.
2Reliability
If complex first principles models are updated in real-time, then accuracy is improved, but operational delays increase
Solution Approach 1:
The patent performs preliminary action by training the proxy models offline before real-time operations begin. During offline training, the system processes historical data and complex model outputs to create simplified models that are ready for immediate deployment. This preliminary preparation eliminates the need for time-consuming updates during critical wellsite operations, thus preventing operational delays.
Solution Approach 2:
The proxy models serve as pre-computed copies of the complex first principles models. Instead of updating the complex models in real-time, the system uses the trained proxy models to replicate their behavior instantly. This copying approach maintains accuracy while eliminating the time penalty associated with real-time updates of complex models.
3Productivity
If simplified proxy models are used, then computational burden is reduced, but model accuracy may deteriorate
Solution Approach 1:
The proxy models are trained to copy the input-output behavior of complex first principles models using extensive historical data. By learning from numerous training examples, the simplified models capture the essential relationships and patterns, achieving acceptable accuracy for control decisions while maintaining computational efficiency.
Solution Approach 2:
The patent changes the parameters of the model by transitioning from complex physical-based parameters in first principles models to simplified empirical parameters in proxy models. This parameter transformation allows the system to maintain predictive capability with reduced computational requirements, as the proxy models use fewer and simpler parameters that are optimized during training.
Data Source
AI summary
Techniques for managing a hydraulic fracturing operation include receiving a selection of a proxy model that represents a first principles model of a hydraulic fracturing operation, the proxy model including at least one property of a plurality of properties of the first principles model associated with the hydraulic fracturing operation; simulating the selected proxy model to generate a modeled output based on the property; and determining a value of a control setpoint for hydraulic fracturing operation equipment based on the modeled output.