Research Background

Over the years, I have worked on process analysis and optimization across various manufacturing technologies and materials, particularly in metal forming, thermoplastic injection molding, and composite material forming. My research extends beyond conventional concurrent engineering, integrating in-depth analyses of manufacturing processes, component geometry, material properties, performance, life cycle, and more. The following summary outlines my research expertise and concludes with my current interests. In terms of manufacturing processes, as summarized in the following image, starting from my Ph.D. years, I have been involved in the analysis of metal forming processes by finite element method (FEM) analysis, numerical and analytical modeling. In this regard, metal forming is one of the cornerstones of a huge portion of the manufacturing compartment and for this reason it is also one of my current research areas, though the investigation tools shifted from analytical and numerical modeling to machine learning.

Another key process of interest is represented by reinforced plastic, which refers to both thermoplastic and thermoset polymers. As per the former, considering my prior working experience as quality engineer and afterwards as quality assurance manager at a thermoplastic injection molding company in Italy from 2006 to 2012 (across my bachelor year up to the end of the first year of the master’s degree), I have always had a keen interest in this process. To this end, I gained expertise in finite volume method (FVM) simulation development for the thermoplastic injection molding process, integrating it with FEM structural analysis to study how fiber orientation, influenced by injection locations, affects mechanical performance, also leveraging machine learning optimization tools. As per the latter, thermoset polymers serve as a bridge between these categories, forming the core of both multi-material structures—featuring lightweight, stiff woven prepreg composites—and SLA additive manufacturing, where the resin is a thermoset photopolymer. Across these areas, while the manufacturing process remains central—analyzed in terms of constraints and parameters—my multidisciplinary approach connects it to target-oriented optimum design (e.g., strength or stiffness optimization) and material science (e.g., anisotropic constitutive modeling for composites or strain-strain rate-temperature flow stress modeling for hot-forged metals). Besides, in multi-material structures and additive manufacturing, I have explored tailoring materials through layup or reinforcement definition, optimizing both material properties and processes to achieve near net-shape manufacturing.

Current Research Interests

Considering this introduction over my background in manufacturing engineering, starting from 2022 my main research effort is devoted to the development of highly customizable machine learning models accounting for process parameters, target component geometry, and material properties for all-encompassing optimization tools. In this regard, the current research effort can be summarized as follows.

1) Gradient Enhanced Neural Network (GENN) for manufacturing process optimization (left image): machine learning is well established, and common algorithm can be easily found in online repository and applied to the desired scenario. However, issues arise for the definition of the training dataset and prediction accuracy outside of its boundary. This research focuses on defining a new class of solutions capable of embedding prior engineering knowledge into the training, to limit the need for training data while extending the accuracy outside of the training (latent) space. The modeling approach has been successfully applied for the prediction of the forming limit in Titanium sheet forming and is being tested on force prediction for the ring-rolling process.

2) Adaptable Learning by Domain Adaptation (ALDA) (center image): this second research direction shares some features with the GENN modeling but also answers to a critical aspect shared by various engineering disciplines, including manufacturing engineering, id est how good is the developed model (FEM, analytical, machine learning, etc.) to describe the phenomenon at hand. In this regard, ALDA is envisioned as a module operating across different domains, capable of “adapting” a trained machine learning model to a new domain not included in the original dataset. In a similar way, the adaptation can also be used to target the limitations of the original training dataset by including few experimental results, to reduce the simulation-to-reality (Sim2Real) gap. ALDA was developed for the ring rolling process and is currently being tested on the injection molding process.

3) Composite Additive Manufacturing (right image): SLA additive manufacturing is well-known and employed is small-scale and highly customized scenarios, but the manufactured parts lacks engineering materials’ performance. To this end, the addition of particles/fibers reinforcements, together with the analysis of the influence of the process parameters on the resulting properties distribution, is essential to promote a better understanding and intertwining relationship between geometry, process, and material, to be employed in a topology optimization scheme including all the above and not only the geometry of the component. Considering its multidisciplinary nature, this research direction includes aspects of material science (ex. constitutive modeling, fracture mechanics, etc.), manufacturing process analysis, and optimum design, in a concurrent analysis scheme.