2014, MSC Software Student Simulation Contest
Bobsled is a winter sports that race down a ice track with a sled which is also known as "F1 race of winter sports". Recently, many eminent automobile enterprises, such as BWM, McLAREN, and Ferrari, have been participated in researches to develop faster sled. However, there had been few attempts to investigate an optimal driving strategy in bobsled driving that a driver manipulates the steering handle to drive the sled faster and safer. In this research, an optimization of the steering strategy of a bobsled was conducted with a Genetic Algorithm on a simulation environment using multi-body dynamics simulation tool "MSC Adams".
1. Bobsled Modeling
FIG. 1 Blueprint drawings and a CAD model of a bobsled.
For dynamic simulation, I established a 3D CAD model of bobsleigh. There were blueprints of a bobsled at the International Bobsled Rule website and the CAD model was generated based on the blueprints and the international regulations. A bobsled has steering and suspension mechanisms as an automobile so. The front blades can be rotated in yaw, pitch and roll orientation. The front frame and rear frame is connected with a torsional spring. The mechanical joint and linkage was implemented with the simulation tool, MSC Adams. In the steering system of front blades, the roll joint is coupled with frame by leaf springs, which is difficult to model in a rigid body dynamics simulation. Therefore, a stiffness test was conducted for the leaf spring with a FEA simulation and the leaf spring was modeled as an equivalent torsional spring.
2. Dynamics Simulation Setup
FIG 2. Forces exerted on a racing bobsled.
The forces exerted on a bobsled while racing was defined in the simulation environment. A contact force between blades and ice was also defined by referencing a previous research about a contact force between a bobsled blade and an ice track. On the top of that, the most important force was a "Steering Force" of a blade on ice.
FIG 3. Steering force of a single blade of a sled on ice track.
For a single blade, the magnitude and direction of steering force is like the function stated in the figure above.
FIG 4. Simulation result for steering handle for 6 deg in 1 sec for two cases.
Top: Only Hertzian contact force.Bottom: Steering force applied.
The steering of sled occurs because the blade makes a groove on a ice surface. Therefore, the steering force needs to be included additionally on the regular Hertzian contact force.
3. Genetic Algorithm(GA)
FIG 5. Concept of Genetic Algorithm
A genetic algorithm code was established with MATLAB for the optimization.
FIG 6. Fitness function of the optimization
The Fitness function to evaluate each trials is stated above.
4. Result
Optimization for a 90 degree curved track
MOV 1. Simulation Result for a 90 deg curved track
FIG 7. Best scored gene (Top) and best and average score of each generations(Bottom) for a 90 deg curved track case
Optimization for a 180 degree curved track
MOV 2. Simulation Result for a 180 deg curved track
FIG 8. Best scored gene (Top) and best and average score of each generations(Bottom) for a 180 deg curved track case
Optimization for a 270 degree curved track
MOV 3. Simulation Result for a 270 deg curved track
FIG 9. Best scored gene (Top) and best and average score of each generations(Bottom) for a 270 deg curved track case
5. Conclusion
The optimization result successfully showed improvement of the bobsled speed on the three cases by using the genetic algorithm. Genetic Algorithm is very useful optimization method for a complicate system which is difficult to solve with mathematical method. However, there are some potential risks such as a local minimum. In addition, since the simulation and real life so that it is important to be validated with the hardware experiment. Although this project introduced a new optimal driving strategy for a bobsleigh race but it still remains verification to apply the result in the real system.