Abstract:Acrylic elastomer VHB 4910 is one of the most significant categories of dielectric elastomer and has the promising applications in the field of soft robotics, actuators, energy harvesters and intelligent vibration isolation. But the nonlinear viscoelasticity of elastomer affects the mechanical behavior of the material dramatically. Recently fractional order models have received much successes in modeling complex material. In the work, a 3D tensorial constitutive model of elastomer based on the fractional Kelvin-Voigt model at finite deformation is formulated. And then the constitutive relation of uniaxial stretches is also derived. Subsequently a series of uniaxial experiments of VHB 4910 elastomer under different stretch rates are conducted. Based on the additive structure of the constitutive model, the parameters of the hyperelastic spring and the fractional viscoelastic element are identified, respectively. The material parameters of hyperelastic spring in the model are identified by the Neo-hookean, Mooney-Rivlin and Gent model at first. Then the parameter identifications for both variable order and constant order fractional viscoelastic element are conducted in order to investigate the rate dependency of fractional viscoelastic elastomer. The results show that the Mooney-Rivlin model can give a better fitting result for the hyperleastic spring element. Both of the variable order and constant order fractional viscoelastic model can simulate the rate dependency of the viscoelastic elastomer well. Constraining the order of the fractional model influents the accuracy of fitting results non-significantly. The relationship between viscosity of fractional element and stretch rate is observed to be nonlinear, which indicates the non-Newtonian fluid feature of the elastomer. Then a modified power law is developed to quantitatively describe this nonlinear relationship. The fitting curve of the modified power law is compared with the Cross non-Newtonian fluid model. The result shows that the developed model can lead to a better fitting results than Cross fluid model.