Hybrid Process Simulation for Industrial Plant Control Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing optimization methods for industrial processes require complete knowledge of physical laws and are limited by the complexity of chemical reactions, while machine learning models lack reliability and expertise, failing to account for transient situations and manual actions.

Innovation Solution

A computer-implemented method combining rigorous simulation, predictive simulation, and expert simulation modules, using iterative processes to determine optimal control parameters through heuristic rules and machine learning, ensuring convergence and compatibility across overlapping sub-processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If rigorous simulation models are used for all sub-processes, then measurement precision and reliability improve, but device complexity and difficulty of detecting and measuring increase significantly

Engineering Contradiction:
Improveprecision of process modelingVSAvoidcomplexity of simulation system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The production process is divided into multiple sub-processes, each modeled with appropriate simulation rigor. Some sub-processes use rigorous simulation models where physical laws are well-established, while others use machine learning models where data-driven approaches are more effective. This segmentation allows the system to achieve high measurement precision for critical sub-processes without requiring rigorous modeling everywhere, thus reducing overall device complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different levels of modeling quality are applied to different sub-processes based on their specific characteristics. Sub-processes with well-understood physics receive rigorous simulation treatment, while those with changing nature or insufficient equation information receive machine learning modeling. This local differentiation optimizes the balance between measurement precision and device complexity.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If machine learning models are used for sub-processes with changing nature, then adaptability improves, but reliability and measurement precision deteriorate

Engineering Contradiction:
Improveadaptability to changing processesVSAvoidreliability of process modeling
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The system merges rigorous simulation models and machine learning models into a unified hybrid simulation framework. Machine learning models provide adaptability to changing sub-processes, while rigorous models anchored in physical laws provide reliability. The iterative reconciliation process ensures that both model types work together consistently, allowing the system to maintain high reliability even when adapting to process changes.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

An iterative reconciliation process is implemented where simulation results from different models are compared and adjusted. The system repeatedly simulates the production process, compares results between rigorous and machine learning models, and adjusts parameters until convergence is achieved. This feedback mechanism ensures that machine learning models remain reliable by continuously validating them against rigorous physical models.

Inventive Principle:
Principle #23Feedback

3Manufacturing precision

If iterative reconciliation processes are implemented across multiple simulation modules, then manufacturing precision improves, but loss of time increases due to repeated simulations

Engineering Contradiction:
Improveprecision of control parametersVSAvoidtime for convergence iterations
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-processing input data and initializing simulation parameters before the iterative reconciliation process begins. This preparation work reduces the number of iterations needed for convergence, thereby decreasing the time loss while maintaining manufacturing precision. The preliminary setup ensures that the iterative process starts from an optimized state.

Inventive Principle:
Principle #10Preliminary action

4Reliability

If complete rigorous models are used for optimization, then reliability improves, but ease of operation deteriorates due to inability to handle transient situations

Engineering Contradiction:
Improvecompleteness of physical modelingVSAvoidease of handling transient operations
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The system transitions from static rigorous models to a dynamic hybrid modeling approach that can adapt to transient situations. Machine learning models capture the dynamic behavior of sub-processes during transient operations, while rigorous models provide the physical foundation. This dynamic approach maintains reliability through physical consistency while improving ease of operation during transient and manual actions.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP4081870B1Computer-implemented method for determining an optimal operative state of a production process of an industrial plant
Publication Date: 2026.02.25 MOEVE SA
  • EP4081870B1 patent drawingFigure 1
  • EP4081870B1 patent drawingFigure 2
  • EP4081870B1 patent drawing

AI summary

The present invention relates to a computer-implemented method for the purpose of determining an optimal operative state of a production process of an industrial plant, particularly suitable for an industrial chemical plant. The method allows simulating this production process by instantiating suitable simulation modules and solving a maximization problem combined with an iterative method which, after convergence, makes the values of the parameters of all the instantiated modules compatible, and returns, among other values, the set of control parameters which, used in the production process, would take it to the optimal operative state. The method allows operating in simulation mode, determining the optimal conditions in response to a given set point value, and also in control mode where the parameters are applied to the industrial plant, taking the production process to the values provided by the pre-established optimal conditions.