Experimental Development of Tyre Tread Design Combined with Parametric Search

A method for optimizing the shape, posture and arrangement of sipes based on parametric design factors was developed to improve the grip performance of winter tyres. This is a method that involves repeatedly conducting friction coefficient measurement experiments on a group of 3D-printed models made of materials with physical properties equivalent to tyre rubber, and Bayesian optimization of multiple parameter combinations that are expected to enhance performance. First, the effectiveness of this method was verified by evaluating the step-by-step improvement in grip performance with each iteration of the Bayesian optimization loop. Furthermore, we attempted to improve grip performance on the newly emerged ‘Twist sipe’ design created by expanding the design space while confirming the effectiveness of this method. Finally, the discussion revolved around the type and selection of acquisition functions necessary for acquiring a Gaussian process regression model that will be used in selecting the combination of design factors for the next loop in Bayesian optimization.