Fast nonlinear mechanical features decoupling to identify and predict asphalt-based composites

Abstract

Nonlinear matrices employed in asphalt-based composites exhibit a prominent nonlinear elastic-viscoplastic-viscodamageable mechanical response. The constitutive model for this type of matrices requires a significant experimental effort to identify the material constants. To alleviate the characterization effort without losing reliability in the mechanical prediction, this paper presents a fast procedure decoupling the elasto-viscoplastic response from the viscodamageable one and facilitating the identification of the material constants via efficient optimization. This procedure relies on a combined experimental–numerical also applicable to other rheologically complex materials like polyurea, thermoplastics or elasto-viscoplastic polycrystals. The experimental part is deliberately designed to only conduct cost- and time-effective monotonic loads at different strain rates in tension and compression. These pure monotonic results are used in the constitutive model to predict more complex loading conditions such as multiple load–unload–reload (LUR) cycles. To do this, an elasto-viscoplastic constitutive model suitable to describe creep-like phenomena and large irreversible deformations is proposed. This model incorporates pressure and strain rate sensitivity, which is essential to capture the compression–tension asymmetry in asphalt-based composites. A rate-dependent damage model is proposed to describe the degradation rate of the elasto-viscoplastic part. The stress and the consistent tangent modulus are derived. The model implementation for user-defined finite element subroutines are given for explicit or implicit solvers. The proposed decoupling facilitates an efficient identification of the constant parameters via an in-house Nelder–Mead-based optimization method. To accelerate the identification, all the stress–strain curves, either in tension or compression, are simultaneously used with applying physically-based constrains. The framework is validated with asphalt matrix at room temperature (24 °C) and verified under LUR conditions at different rates. The identified model predicts correctly the experimental observations, proving its applicability to investigate particle-based composites with highly nonlinear matrices.