< Back to previous page

Publication

Evaluating aerodynamic efficiency in cycling through digital human modelling

Book - Dissertation

It is fair to say that one of the main objectives of all elite athletes is to improve performance in their sport. To improve, performance needs to be defined and quantified so that it can be monitored and recorded. Naturally, improving performance results in victories and victories result in glory for the individual or the team. However, the power and impact of sport extends beyond the athletes. Sport stands for aspirational values, pushes the boundaries of human performance, drives innovation, stimulates interdisciplinary transfer of innovation, creates employment, and more. Most tangibly, advancements in performance levels of elite sport could also result in growing and supporting an economy of hobby practitioners. In this chain of events, the quantification and measurement of performance is a vital first step. In elite cycling teams as well as in product development, aerodynamic drag force (or ‘drag force’ or ‘drag’) is the most studied and scrutinized topic of road cycling. This is because drag force accounts for 50-90% of forces that a cyclist overcomes depending on speed and pose. Most state-of-the-art methods of evaluating drag are expensive for the amateur athlete. In this thesis, we want to devise a method of evaluating aerodynamic properties of a cyclist on a bike which is accessible to the amateur cyclist. Computational Fluid Dynamics (CFD) is a reasonable substitute for wind tunnel testing for the application of assessing aerodynamic efficiency in cycling. To use CFD software for this application, 3D models of cyclists (and bicycles if available) are required. The two main ways in which to obtain 3D shapes of cyclists are: using stock models or using 3D scans. Both approaches have their limitations. We envision a third method: in this, we could arrive at a high-accuracy base 3D model of an individual through an engine that generates 3D models based on the individual’s anthropometric data. Using 3D pose information of the major body joints from a motion capture system, the base model can be re-posed to obtain a 3D shape of any individual in any pose. This can allow the dynamic 3D models to be used for not only aerodynamic analysis in CFD but also the design and/or optimization of products that dynamically interact with the human body. The main hypothesis of the project is that if dynamic 3D shapes of human beings can be estimated with minimal investment (time, data, expertise costs), it can lead to a usable framework for performing aerodynamic assessments of on-bike cyclist pose to optimize for (minimum) air resistance in a user-friendly and speedy manner. This thesis will investigate how the full 3D body shape of a human being can be constructed in any pose from a static 3D model and joint angle measurements for the application of cycling aerodynamics. The overarching aim of this thesis is to develop and test a ‘Virtual Wind Tunnel’ workflow that enables cyclists to perform low-cost aerodynamic assessments of their on-bike poses, preferably not limited to laboratory setting, using limited and easy-to-measure anthropometric data that does not need expert intervention. The design of such a framework is detailed with the best practices in wind engineering for the application of quick aerodynamic assessments, accurate enough for relative pose comparisons for cyclists, which are fast to obtain. The results are promising: accuracy is maintained, thus validating the re-posing methods and Virtual Wind Tunnel Workflow for both reference as well as re-posed scans for two separate on-bike poses. It was shown that with the help of few (< 10), basic, easy to measure anthropometric data, and the use of a smartphone camera, it is possible to envision a ‘home wind tunnel’ which can provide hobby cyclists with metrics such as projected frontal area, drag area, and drag force. Aerodynamic pose optimization, bike fitting, and regular testing is possible at home. In conclusion, the feasibility of indirectly gauging aerodynamic performance by using digital human modelling was approached from a CFD standpoint with favourable results. The results of such a ‘Virtual Wind Tunnel’ are promising. Using the methods used in this thesis, a cyclist can very well perform ‘rough-and-ready’ aerodynamic calculations with very little investment. Future work could focus on more accurate 3D models or engines, faster and more accurate joint angle capture, and using regression methods to further tune or validate the equations formulated.
Number of pages: 131
Publication year:2021
Keywords:Doctoral thesis
Accessibility:Open