Graded materials have been extensively explored by researchers due to the superior performance compared to their uniform counterpart. Up to now, the exceptional energy absorption of the graded lattice and shell-based structures have been demonstrated. In this work, we experimentally and numerically investigated the effects of different design strategies on the mechanical performance of polymer foams. Our results show that three failure mechanisms dominate the failure of polymer foam, i.e., binder-failure-only, shell-failure-only, and binder-shell-failure. Based on these failure mechanisms, we created three types of graded designs, including graded binder, graded shell thickness, and hybrid graded design. It was found that the specific energy absorption for the hybrid graded polymer foam increases by 125%, 185%, and 34% compared to the uniform poly foam, graded binder foam, and graded thickness foam, respectively. Furthermore, when compared to graded lattice and graded shell-based structures, the specific energy absorption of hybrid graded foam is increased by 141% and 32%, respectively. The findings in our work opens a new avenue to design architected materials with enhanced mechanical properties that can find applications ranging from structural components of defense systems to personal protection equipment.

Malleable crosslinked polymers represent a new class of materials, wherein reversible covalent bonds are employed. These materials can exhibit mechanical properties of typical thermosets under ambient conditions, yet at elevated temperatures or under other stimuli they can be reprocessed and recycled like thermoplastics through a cross-link exchange and rearrangement process, usually with the aid of catalysts. This presentation will focus on the development of a light-weight composite material consisting of a catalyst-free thermoset matrix and low-density nanofillers or carbon fibers (CFs). The resulting composites become malleable upon thermal activation, thus enabling its unique reprocessability, rehealability, and full recyclability while retaining good mechanical properties.

Piezoresistive laser induced graphene are embedded within the interlaminar regions of fiberglass composite materials using a treatment-free and scalable transfer-printing process, all while maintaining mechanical properties. Through passive resistance measurements, the multifunctional material is demonstrated to be capable of in-situ tracking of both monotonic and cyclic strain, in addition to detecting distinct damage events under tensile and flexural loading conditions. Furthermore, The LIG interlayers are also shown to enable three-dimensional damage localization in fiberglass composites, along with the monitoring of structural damage progression and accumulation throughout fatigue life. The information can be then combined with smart prognostic algorithms, such as neural networks, in order to predict the onset of catastrophic structural failure.

The objective of this study was to implement a supervised machine learning method that utilizes the radial basis function neural network for 3D electrical impedance tomography conductivity distribution reconstruction of complex cellular lattice structures. This data-driven algorithm, which was trained by a variety of damaged cases, is significantly faster than conventional EIT while enabling greater accuracy of 3D conductivity distribution reconstruction. Both numerical simulations and experimental results are presented in this work, and the machine learning based EIT results are compared with those obtained using conventional EIT.

Since the late 20th century, the speed of digital data acquisition and processing from multi-element ultrasonic arrays has increased dramatically. We are now in the age of ‘computational ultrasound’ where NDE performance advances are increasingly driven by superior data analysis rather than new physics or sensors. In this talk I will look back over some of the developments in which I have been involved, including guided wave arrays for large area inspection and ultrasonic arrays for localised inspection. I will discuss how computational ultrasound can benefit from deep learning techniques and the crucial role it will play in Industry 4.0.