Researchers from the University of Technology in Ho Chi Minh City have developed a new data-driven method for simulating turbulent flow that could revolutionize fields like aircraft and ship design, industrial process optimization, and weather prediction. The team, led by Duc Minh Nguyen, has combined the strengths of two existing technologies—Video Prediction Transformer (VPTR) and Multigrid Architecture—to create a more accurate and efficient approach to turbulent flow simulation.
Turbulent flow simulation is a complex challenge that has traditionally required significant computational resources. The researchers’ new method, named MGxTransformer, leverages the ability of VPTR to capture complex spatiotemporal dependencies and handle large input data. At the same time, it utilizes the Multigrid Architecture’s multiple grids with different resolutions to capture the multiscale nature of turbulent flows. This combination allows for more accurate and efficient simulations, which can be crucial in various applications.
The team’s experiments have shown that MGxTransformer can accurately predict velocity, temperature, and turbulence intensity for incompressible turbulent flows across various geometries and flow conditions. The results demonstrate superior accuracy compared to other baseline methods while maintaining computational efficiency. This breakthrough could lead to more precise and efficient designs in industries that rely on turbulent flow simulations, such as aerospace and maritime engineering.
The researchers have made their implementation of MGxTransformer available publicly on GitHub, encouraging further collaboration and development in this field. This open-source approach could accelerate advancements in turbulent flow simulation and its practical applications. The team’s work represents a significant step forward in the field of computational fluid dynamics, offering new possibilities for innovation and improvement in various industries. Read the original research paper here.