3D Fluid Flow Surrogate Modeling for Interactive Aerodynamic Design

Solution Overview

Problem

Computational fluid dynamics (CFD) simulations are computationally expensive and time-consuming, requiring hours to optimize design object shapes due to the complexity of solving equations over many time steps, making iterative design processes inefficient.

Innovation Solution

A method involving parameterization of a design object into a polycube representation, computing a distortion grid, and using neural networks to model surface pressure and velocity fields, enabling real-time visualization of aerodynamic effects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional CFD simulation is used to simulate fluid flow around design objects, then accurate velocity field and surface pressure data can be obtained, but the simulation process becomes computationally expensive and time-consuming (16-17 minutes per simulation)

Engineering Contradiction:

Improveaccuracy of fluid flow simulationVSAvoidsimulation execution time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system pre-computes and stores fluid flow characteristics (velocity fields and surface pressures) for various design object shapes in a database before actual design work begins. When a user modifies a design object, the system quickly retrieves pre-computed data matching the new shape parameters, avoiding the need to run full CFD simulations from scratch. This preliminary preparation resolves the contradiction by trading initial computational effort for rapid subsequent queries.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

Instead of performing expensive CFD simulations for every design iteration, the system creates simplified representations (polycube models) that copy essential geometric features of design objects. These polycube copies can be rapidly processed and matched against pre-computed fluid flow data, providing accurate results without the computational burden of full-scale simulations.

Inventive Principle:

Principle #26Copying

2Manufacturing precision

If iterative shape optimization is performed using CFD simulation, then aerodynamic performance can be improved, but the process requires hours or even more than a day to complete

Engineering Contradiction:

Improveaerodynamic design optimizationVSAvoiddesign iteration speed

Core Design Contradiction:

Manufacturing precisionVSProductivity

Solution Approach 1:

The system pre-computes and stores fluid flow characteristics (velocity fields and surface pressures) for various design object shapes in a database before actual design work begins. When a user modifies a design object, the system quickly retrieves pre-computed data matching the new shape parameters, avoiding the need to run full CFD simulations from scratch. This preliminary preparation resolves the contradiction by trading initial computational effort for rapid subsequent queries.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces polycube models as intermediary representations between the original design objects and the fluid flow simulation data. These polycube intermediaries simplify complex geometries into standardized forms that can be efficiently matched against pre-computed data, enabling rapid iterative optimization without sacrificing aerodynamic accuracy.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Measurement precision

If CFD simulation is run for each shape modification during design, then accurate aerodynamic evaluation is achieved, but the computational complexity and cost increase significantly

Engineering Contradiction:

Improveaerodynamic evaluation accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

Instead of performing expensive CFD simulations for every design iteration, the system creates simplified representations (polycube models) that copy essential geometric features of design objects. These polycube copies can be rapidly processed and matched against pre-computed fluid flow data, providing accurate results without the computational burden of full-scale simulations.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system transforms complex design object geometries into simplified polycube parameterizations with fewer degrees of freedom. By changing the representation parameters from detailed mesh geometries to polycube control point coordinates, the system reduces computational complexity while preserving essential shape characteristics needed for accurate aerodynamic evaluation.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12530507B2Machine learning three-dimensional fluid flows for interactive aerodynamic design

Publication Date: 2026.01.20 AUTODESK INC

AI summary

Embodiments of the invention disclosed herein provide techniques for simulating a three-dimensional fluid flow. A parameterization application parameterizes a first representation of a design object to compute a first polycube representation. The parameterization application computes a first distortion grid based on the first polycube representation. A machine learning application computes, via a first neural network, a surface pressure model based on the first polycube representation. The machine learning application computes, via a second neural network, a velocity field model based on the first polycube representation and the first distortion grid. The machine learning application generates a visualization of the surface pressure model and the velocity field model for display on a display device.