Predictive accurate machine learning models for aerodynamics

Purpose and goal:

Machine learning is transforming aerodynamics research and product development by enabling rapid design iterations and precise optimizations. By analyzing airflows and identifying effective design changes, ML reduces the need for wind tunnels and CFD simulations, saving time and resources. The goal is to create more aerodynamically efficient, high-performance structures. This technology drives innovation and competitiveness, even in other areas where performance and efficiency are critical.

Expected results and effects:

Machine learning (ML) is expected to revolutionize aerodynamics and vehicle design by enabling more accurate simulations and optimizations. The results include more efficient shapes that reduce drag and improve performance, leading to increased fuel efficiency and higher speeds. By replacing time-consuming testing with data-driven analysis, development cycles can be shortened, saving resources and driving innovation.

Approach and implementation:

The project aims to improve truck aerodynamics using CFD simulations, machine learning (ML) and optimization. Key steps include identifying critical geometric factors, conducting CFD simulations when necessary, training and validating ML models to improve predictions, and using parametric optimization to refine the design. Integration into PredictiveIQ´s platform and training of Scania´s engineers ensures efficient implementation and improvement.