Aerodynamic predictions of a high-lift aircraft using ADS Code Leo

What is it about?

An area of current focus for the aeronautics community is improving the confidence of computational models for high lift configurations (take-off/landing). The results presented are an investigation of the predicted performance of an open-source high lift aircraft called the NASA Common Research Model - High Lift (NASA CRM-HL). These predictions are part of the 5th High Lift Prediction Workshop (HLPW) where participants from all over the world investigated this case. The paper outlines the results with the GPU-accelerated solver called ADS Code Leo. The results compare predictions across various mesh types and against wind-tunnel data measured by ONERA. The results show how the solver was able to consistently predict the trends of the performance across dramatically different mesh types. Thus allowing various users to have confidence in the results across various setups. Moreover, the speedups gained from the use of the GPU-accelerated ADS Code Leo solver were several magnitudes faster than what would be obtained in an industry standard CPU cluster with hundreds or thousands of cores.

Featured Image

Photo by Pascal Meier on Unsplash

Photo by Pascal Meier on Unsplash

Why is it important?

- The importance of the publication is that it contributes to the engineering community as the challenges and best approaches to modeling high-lift configurations are explored. The contributions in the paper show an extensive comparison across various mesh types. These comparisons showed that ADS Code Leo predicts similar performance across the different mesh types. Thus providing confidence that similar predictions can be obtained across the different best practices of companies or users. - Moreover, the speedups observed on both consumer grade GPUs (gaming laptop) and data-center grade GPUs highlight how ADS Code Leo doesn't only have accurate solutions relative to industry solvers but also is able to reduce running time substantially to allow more design cycles. Even more critically for design teams, the speedups observed have crossed the threshold where the speedup is large enough that the cost per simulation is lower on the GPUs than on the industry standard CPU clusters. Lastly, the ability to execute simulations on both consumer-grade GPUs (gaming laptop) and data-center GPUs (on-site or cloud-based) allows both small enterprises and large enterprises to have access to the revolutionary performance gains of ADS Code Leo.

Perspectives