Physics-Informed Neural Networks (PINNs) are computationally efficient tools for addressing inverse problems in solid mechanics, but often face accuracy limitations when compared to traditional methods. We introduce a refined PINN approach that rigorously enforces certain physics constraints, improving accuracy while retaining the computational benefits of PINNs. Unlike conventional PINNs, which are trained to approximate (differential) equations, this method incorporates classical techniques, such as stress potentials, to satisfy certain physical laws. The result is a physics-enforced PINN that combines the precision of the Constitutive Equation Gap Method (CEGM) with the automatic differentiation and optimization frameworks characteristic of PINNs. Numerical comparisons reveal that the enforced PINN approach indeed achieves near-CEGM accuracy while preserving the efficiency advantages of PINNs. Validation through real experimental data demonstrates the ability of the method to accurately identify material properties and inclusion geometries in inhomogeneous samples.

Existing interface models are often inaccurate when modeling delamination of Fiber-Reinforced Polymer (FRP) structures, because they do not account for the non-local effects resulting from extrinsic failure mechanisms. Indeed, local traction separation laws are valid only for simple processes with well-defined delamination fronts. In contrast, non-local bridging events, such as fiber or ligament bridging, can significantly modify the dissipation and depend on the global kinematics of the crack front. We have conceived an approach for modeling delamination of FRP, which enriches the cohesive zone model with a delamination tracking technique, allowing for a comprehensive exploration of the diverse non-local features of interlaminar fracture. In our methodology, the evolving delamination front is identified, followed by the quantification of the effects from multiple mechanisms related to the non-local extrinsic dissipation. By simulating two experimental scenarios featuring with evolving configuration of delamination front, we highlight the robustness of this approach for predicting delamination features in the context of extrinsic energy dissipation. Our approach can be viewed as a versatile toolbox allowing to incorporate the influence of various non-local extrinsic mechanisms.

Interfaces play a critical role in modern structures, where integrating multiple materials and components is essential to achieve specific functions. Enhancing the mechanical performance of these interfaces, particularly their resistance to delamination, is essential to enable extremely lightweight designs and improve energy efficiency. Improving toughness (or increasing energy dissipation during delamination) has traditionally involved modifying materials to navigate the well-known strength-toughness trade-off. However, a more effective strategy involves promoting non-local or extrinsic energy dissipation. This approach encompasses complex degradation phenomena that extend beyond the crack tip, such as long-range bridging, crack fragmentation, and ligament formation. This work explores this innovative strategy within the arena of laminated structures, with a particular focus on fiber-reinforced polymers. This review highlights the substantial potential for improvement by presenting various strategies, from basic principles to proof-of-concept applications. This approach represents a significant design direction for integrating materials and structures, especially relevant in the emerging era of additive manufacturing. However, it also comes with new challenges in predictive modeling of such mechanisms at the structural scale, and here the latest development in this direction is highlighted. Through this perspective, greater durability and performance in advanced structural applications can be achieved.