Tieteestä & Tekniikasta
Application of surrogate modelling to determine stiffness of nanocomposite
Text and Figures: Davide Angelini, Polytechnic University of Turin
Between March and October ’25, M.Sc. Davide Angelini worked at Materials Science and Environmental Engi- neering of Tampere University as Visiting Researcher, supervised by Prof. Mikko Kanerva. Davide’s research is aligned with an Industrial doctoral project in Aerospace Engineering between Polytechnic University of Turin (Italy) and FEV Italy s.r.l. company. Davide’s doctoral dis- sertation is part of the project PNRR-NGEU ( Next Gene- ration EU ). His topic of research is the development of industry-ready simulations of composite materials with nano-sized fillers, that are called nanocomposites here. Nanocomposites offer new design capabilities and show promising applications in aerospace and automo- tive industries due to high specific properties in light- weight designs. For example, silica-doped nanocoating increases fatigue resistance in corrosive environments of additive manufactured metals and potentially improves cleanability of carbon-fiber reinforced composite sur- faces (Cestino, et al. 2024). The main problem is that it is difficult to perform ex ante simulations; the simulations are computationally expensive. Moreover, the uncertain- ties in current materials characterization limit wides- pread application. Davide Angelini, Prof. Mikko Kanerva, as well as Davi- de’s Italian supervisor Prof. Enrico Cestino met in 2024 at the DraF 2024 conference (Ischia Island, Naples, Italy). Later Mikko and Davide also met in the ICAS 2024 confe- rence (Florence, Italy). The new team of these three men identified synergies in research of nanomaterials. When new material is created, there is uncertainty related to its directionality. Some well-known analytical models can predict expected values (Angelini, et al. 2025) but tradi- tional characterization methods are infeasible due to the very low thickness nanocomposites applied as coatings.
Davide Angelini
The hypothesis is that surrogate algorithms coupled with Finite Element Analysis (FEA) can be used to efficiently predict unknown surrogate constants. Surrogate algorithms are a class of stochastic algori- thms where a simpler, computationally cheaper mathe- matical model is constructed to mimic the output of a complex, computationally expensive simulation. This simpler approximation, often called a metamodel , is built or ‘trained’ using data from a limited number of strate- gically chosen runs of the original high-fidelity model, such as the FEA. Davide focused on specific nanocomposite material with graphene (Layek, et al. 2021). A finite element model of nanocomposite sample and an indenter is used to vir-
16 MUOVIPLAST 6/2025
Made with FlippingBook - Online magazine maker