Research On Several Problems In Finite Element Analysis And Optimization Of Pneumatic Tire

Pneumatic tire is an indispensable part of most vehicles, including car,truck and even airplane, and the performances of vehicles are closely relatedto the properties of tires. The investigation of their mechanical properties andthe improvement of their structural designs have attracted much attention allthe time in vehicle engineering, not only tire industry. The mechanicalanalysis of tire involves multiple nonlinearities, including the materialnonlinearity caused by the hyperelasticity of rubber and rubber-cordcomposites, the geometrical nonlinearity related to finite deformation, and theboundary nonlinearity comes from the contact of the tire with ground and rim,which make the analysis, design and optimization highly complex, even withthe help of powerful CAE software. To enhance the efficiency of the analysis,design and optimization, some related key problems are investigated in thisthesis.Digital speckle correlation method is adopted to measure parameters ofMooney-Rivlin model of rubber in the simple tension experiment. The globaldeformation of tire, distribution of equivalent Cauchy stress and distributionof normal contact forces between tire and ground are analyzed.To enhance the efficiency of tire analysis, a new equivalent transverselyisotropic material model of rubber-cord composites is presented. The strainenergy of the equivalent homogenized material is consisted of a Mooney-Rivlin type isotropic hyperelastic part to formulate the properties of rubbermatrix, and an additional transversely isotropic part to formulate thereinforcement of the cord. The total six material parameters in the model areidentified based on numerical experiments of the representative volumeelement divided into refined finite elements. Numerical examples are given toshow the feasibility, efficiency and accuracy of the presented model.To improve the numerical method for the shape optimization of tire tread,a response surface method is adopted. A self-adaptive iteration scheme ofsequential first order and second order local response surfaces is presented toapproximate the original optimization problem. Referred to the ray-stepapproach in full-stress method for the size optimization of truss, a formula toadjust the incremental ray-step is proposed to treat the violation of constraintand bad design-point, in this way the iteration oscillation is suppressed.Numerical example has shown the feasibility of presented approach.To solve the optimization of tire with non-concave objective function andconstraint, a method based on the generalized regression neural network isproposed, which is used to construct the global response surface of objectivefunction and constraint function. The relaxed feasible domain is discretized tofind the coarse global optimum, and then combined with the local search toget the global optimum.