CFD Noise Map Subtraction for Automotive Aerodynamic Simulations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computational fluid dynamics (CFD) simulations of vehicle bodies face run-to-run variability due to transient solvers, making it difficult to distinguish meaningful design changes from statistical noise, leading to reduced trust in results and potential missed significant aerodynamic improvements.
Innovation Solution
A CFD noise map subtraction method that involves performing baseline runs, creating noise maps to quantify stochastic variation, and comparing change maps to these noise maps to discount or disregard changes within noise regions, ensuring only statistically significant changes are considered meaningful.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If transient CFD solvers are used for aerodynamic simulations, then simulation capability is provided, but run-to-run variability occurs making it difficult to distinguish meaningful changes from noise
Solution Approach 1:
The patent performs preliminary baseline runs with identical geometry and conditions to establish expected noise ranges before evaluating design changes. This preliminary action creates reference data that allows subsequent change detection to distinguish meaningful variations from normal solver noise, resolving the reliability issue while maintaining simulation capability.
Solution Approach 2:
The patent introduces an intermediary noise map that quantifies expected run-to-run variability. This noise map acts as a mediator between the CFD simulation results and the design evaluation process, allowing engineers to filter out false positives by comparing observed changes against the established noise baseline.
2Ease of operation
If engineers use intuition and experience to assess run-to-run differences, then subjective judgment is applied, but confirmation bias and reduced trust in results occur
Solution Approach 1:
The patent replaces the mechanical system of subjective engineer judgment with an automated statistical evaluation system. The noise map provides objective, quantitative criteria for assessing significance, eliminating confirmation bias and enhancing trust in results while maintaining ease of operation through automated processing.
Solution Approach 2:
The patent implements feedback by continuously comparing observed flow differences against the noise baseline. This feedback mechanism automatically identifies when changes exceed expected variability, providing objective validation that enhances trust in CFD results and reduces reliance on subjective judgment.
3Difficulty of detecting and measuring
If all flow differences are considered significant, then design changes are detected, but false positives increase reducing trust in results
Solution Approach 1:
The patent changes the parameter used for significance assessment from binary (significant/not significant) to quantitative (magnitude relative to noise baseline). By establishing a noise-based threshold, the system accurately detects meaningful flow differences while filtering out false positives, improving both detection capability and reliability.
4Reliability
If flow differences are judged insignificant, then noise is filtered out, but meaningful aerodynamic improvements may be missed
Solution Approach 1:
The patent changes the threshold parameter from fixed significance levels to dynamic noise-based criteria. This allows the system to automatically adapt to local noise characteristics in different flow regions, ensuring that meaningful aerodynamic improvements are not filtered out while maintaining reliable noise rejection, thus improving design evaluation efficiency.
Data Source
AI summary
Disclosed is a method for evaluating computational fluid dynamic simulation results. The method includes: performing at least two baseline runs of a simulated area or volume containing a first vehicle body shape using a set of initial conditions, performing a change run using a second vehicle body using the set of initial conditions, creating a noise map based on differences between the second baseline run and the first baseline run, creating a change map based on differences between the change run and a selected baseline run, and comparing the change map to the noise map.