Parallel Equation-Oriented Forecasting for LNG Plant Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
LNG processing facilities face inefficiencies due to complex operations involving hundreds of components, making it challenging to optimize production and profitability with existing predictive modeling techniques that are either sequential or limited to specific operating conditions.
Innovation Solution
The development of systems and methods using equation-oriented models that generate cloned models for parallel multi-period forecasting, allowing for real-time optimization and adjustment based on varying operating conditions, such as ambient temperature and feedstock changes, by executing a script to create multiple models and applying different input sets to solve for optimized solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If sequential predictive modeling techniques are used, then the model structure is simple, but the forecasting speed and efficiency are slow
Solution Approach 1:
The patent divides the forecasting system into multiple independent cloned equation-oriented models that can execute in parallel. Each cloned model handles a specific time period or scenario, allowing simultaneous computation across multiple forecasting tasks. This segmentation transforms a single sequential modeling process into multiple parallel processes, directly improving forecasting speed while maintaining manageable model structures through replication.
Solution Approach 2:
The patent creates multiple cloned copies of the equation-oriented model, where each clone is an independent instance that can be executed simultaneously. These cloned models are identical in structure but operate independently with different input parameters or time periods. This copying approach enables parallel processing of multiple forecasting scenarios, significantly accelerating the overall forecasting process without increasing the complexity of individual model structures.
2Adaptability or versatility
If existing predictive modeling techniques are used, then the operating conditions are limited, but the model development is simpler
Solution Approach 1:
The patent develops a universal equation-oriented model framework that can handle multiple operating conditions and scenarios through a single unified structure. This universal model can be configured with different input parameters, disturbance profiles, and operating conditions without requiring separate specialized models for each scenario. The multi-functionality is achieved through parameterization and flexible input handling, allowing the same model structure to adapt to various operating conditions while maintaining relatively simple model development.
Solution Approach 2:
The patent utilizes parameter changes to adapt the equation-oriented model to different operating conditions. By modifying input parameters, disturbance magnitudes, time periods, and operational constraints, the same model structure can simulate a wide range of scenarios. This parameter-driven approach allows versatile adaptation to different operating conditions without increasing structural complexity, as the model framework remains constant while only the input parameters vary.
3Loss of time
If multiple operating scenarios are simulated sequentially, then the model accuracy is maintained, but the computation time increases
Solution Approach 1:
The patent implements periodic action by dividing the forecasting task into discrete time periods or intervals, with each cloned equation-oriented model assigned to a specific period. Instead of processing all scenarios sequentially in one long computation, the system executes multiple shorter computations in parallel, each handling a specific time period or scenario. This periodic decomposition of the overall task enables simultaneous processing of multiple scenarios, dramatically reducing total computation time while maintaining comprehensive scenario analysis coverage.
Data Source
AI summary
Implementations described and claimed herein provide systems and methods for a scripting technique to clone equation-oriented models of a modeled system for parallel simulation of the modeled system. The multiple equation-oriented models may be solved in parallel to quickly create an optimized solution for different operating conditions by providing different input variable sets to the cloned equation-oriented models. The multiple equation-oriented models may provide real-time optimization of the modeled system to provide continuous optimization of all controls or handles of the system to help achieve a target performance of the system. The equation-oriented models may also provide a nomination tool to predict the output of the system over a nomination period with different input variables and performance monitoring capabilities of the system. Offline “what-if” simulations may also be executed on the equation-oriented modeling system to aid operators in predicting performance of the modeled system and troubleshoot potential problems.