Graph-Based Simulation Framework for Multiphysics Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer-aided simulation methods for technical processes are limited by their dependence on specific hardware architectures and mathematical models, restricting their applicability and efficiency, especially in multiphysics simulations where different discretization entities are required.
Innovation Solution
Introducing an additional abstraction plane that combines different mathematical models based on various discretization entities into a graph structure, allowing for generic optimization and parallelization independent of the chosen models and hardware, using generic operators for nodes and edges to propagate evolution problems efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional simulation methods use specific hardware architectures and mathematical models, then the simulation can be optimized for those specific configurations, but the method lacks transferability to other hardware architectures and mathematical models
Solution Approach 1:
The patent creates a universal simulation framework that can operate on any hardware architecture by introducing an abstraction plane between the mathematical models and the hardware. This framework uses generic operators that can be mapped to different hardware platforms, making the simulation method universally applicable while maintaining optimization capabilities.
Solution Approach 2:
The patent introduces an intermediary abstraction plane that mediates between the mathematical models (with their specific discretization entities) and the hardware architectures. This intermediary layer provides generic operators that can be implemented differently depending on the target hardware, enabling transferability without sacrificing performance optimization.
2Productivity
If simulation methods are optimized for special hardware architectures and evolution problems, then performance improves for those specific cases, but the complexity of maintenance and transfer operations increases significantly
Solution Approach 1:
By creating a universal framework with generic operators, the patent reduces maintenance complexity while preserving performance optimization. The same framework can be applied to different hardware architectures and evolution problems without requiring separate optimized versions, thereby simplifying maintenance and transfer operations.
Solution Approach 2:
The patent segments the simulation method into distinct layers: mathematical models, generic operators, and hardware implementation. This segmentation allows each layer to be developed and optimized independently, reducing the complexity of maintaining and transferring the overall system across different platforms.
3Productivity
If monolithic approaches are used for special evolution problems and hardware architectures, then expert knowledge can be applied for optimization, but the approach is not accessible to non-experts and lacks generic applicability
Solution Approach 1:
The universal framework with generic operators makes the simulation method accessible to non-experts by providing a standardized interface that handles the complexity internally. Users can apply the framework to their problems without needing deep expertise in hardware-specific optimizations, while the framework itself can still achieve expert-level performance.
4Measurement precision
If different discretization entities are used for different mathematical models, then each model can be accurately represented, but integrating multiple models becomes precluded from the beginning
Solution Approach 1:
The abstraction plane with generic operators serves as an intermediary that enables integration of different mathematical models with their respective discretization entities. Each model can maintain its specific discretization requirements while the generic operators provide a common interface for coupling and interaction between different models.
Data Source
AI summary
The invention relates to a method for computer-aided simulation of technical processes. Methods of this type are required in particular in evolution simulation models of technical, in particular physical and/or chemical processes, which are intended to be propagated in time (evolution). The method according to the invention for computer-aided simulation of the temporal propagation of technical processes as evolution problem, which is described by at least two different mathematical models as sub-evolution problems which use discretization entities which are different from each other and are propagated in time by means of model-specific algorithms, the simulation being implemented by means of a large number of calculation units, is characterized in that the mathematical models are reproduced on a single coherent graph structure (1) with the discretization entities as nodes (2) and the neighborhood relations thereof as edges (3), edges (3a, 3b) being produced between discretization entities belonging to the same model and also edges (3c) between discretization entities belonging to different models, respectively for the nodes assigned respectively to one mathematical model, for the edges (3a, 3b) between nodes belonging to the same model and also for nodes belonging to different models, respectively specific evolution operators being indicated, and the graph being propagated in time using these evolution operators.