Recent advancements in multiphysics and multiscale modelling of complex materials, which are materials endowed with microstructure, detectable at different scale levels (nano, micro, meso, macro), and characterised by a complex material behaviour (plasticity, damage, fracture) are required by novel applications.

Advanced composites (ACs), consist of various components (metal, polymer, ceramic, etc.) with complicated internal architectures, including porosity, and reinforcement with fibres or particles of different properties, shapes, and sizes. Optimal distribution of the (1) reinforcing phase within the matrices or (2) different phases in multiphase materials is the major task in designing complex composites to get the required material response to the various kinds of loads. The AC’s macroscopic properties are subjected to multi-degradation phenomena which are governed by multiphysics processes that occur at one to several scales below the level of observation, suggesting the application of multiscale approaches. A thorough understanding of how these processes influence the reduction of stiffness and strength is of key importance for the analysis of existing, and the design of improved, complex materials. It is widely recognized that important macroscopic material properties, such as stiffness and strength, are governed by processes occurring at one to several scales below the macro-observation. A thorough understanding of how these processes influence gross behaviour is key to the analysis and the design of existing and/or performance-improved composite materials (multiscale analysis).

The recent advancement in applied computer science and artificial intelligence in the multiphysics modelling of materials aims at modelling of multi-damage and failure processes, validated through experimental assessment of local mechanical properties and microstructures. For example, data-driven parametrically-upscaled constitutive models with machine learning and uncertainty quantification are a novel idea which proposes a parametric representation of lower-scale microstructural descriptors expressed as functions of representative aggregate microstructure parameters including data information about microstructural morphology and crystallography. The application of a machine learning tool is utilized for the generation of constitutive descriptors. Moreover, recently proposed non-local data-driven models for green ACs with particular emphasis on the derivation of the formulation of non-classical models for materials continua and the description of the necessary algorithms and procedures adopted to develop the proposed multiscale model.

Furthermore, innovative multiscale modelling strategies applied to the study of ACs under static and fatigue loading to crack initiation on atomistic and microstructural length scales as well as macroscopic final failure using scale-appropriate methods will be also of interest for the course, to compare these approaches with experimental results for many practical cases. The course also covers recent developments in the modelling of complex materials as non-local continua obtained through the adoption of multiscale approaches.