Fatigue life assessment of additively manufactured material through a hybrid framework

Descrizione del progetto

Resistenza alla fatica per una produzione sostenibile

La produzione additiva (AM) promette un futuro sostenibile riducendo l’impatto ambientale dei fabbricati industriali. La sua adozione su larga scala è tuttavia limitata dalla scarsa resistenza alla fatica dei materiali AM, dovuta a difetti interni e superficiali. Gli attuali trattamenti di post-lavorazione possono migliorare tale caratteristica ma sono costosi, lunghi e talvolta poco pratici. Il progetto FLAME, sostenuto dal programma di azioni Marie Skłodowska-Curie, intende superare queste sfide sviluppando una struttura ibrida che integra modelli basati sulla fisica con l’apprendimento automatico. L’approccio si propone di aumentare l’accuratezza della previsione della resistenza alla fatica dal 75 % al 90 %, riducendo la necessità di un’estesa post-lavorazione e riducendo il tempo di produzione a meno di un minuto. Le innovazioni di FLAME dovrebbero far progredire la tecnologia AM.

Obiettivo

Additive manufacturing (AM) has been proven as a likely potential alternative to conventional manufacturing methods as it is able to reduce the global environmental impact of many industrial products. One of the main obstacles that prevent AM from expanding its application towards fulfilling the requisites of sustainability and green manufacturing, is poor fatigue resistance of AM material caused by multiple intrinsic surface and internal defects.
Currently, different post-processing treatments are being used to tackle this challenge by enhancing surface regularity and reducing bulk defects. Despite the enhancement of fatigue resistance after the aforementioned treatments, these are additional steps that result in longer production times and higher costs. Moreover, due to the complexity of geometry and size, sometimes AM components cannot be easily post-processed by regular treatments.
Optimizing the AM processes aiming at maximized fatigue life is a plausible solution that significantly minimizes the need for post-processing. However, the main barrier here is the lack of accurate, fast, and economic prediction tools to correlate fatigue life directly to AM process parameters. FLAME will address the aforementioned needs and pave the way towards efficient fatigue life assessment of AM via developing a hybrid framework through the association of physics-based models and machine learning approaches. It will improve the state-of-the-art fatigue life prediction accuracy from 75% to 90%, making them more economical by eliminating experiments and fast with an implementation time of less than 1 minute.
FLAME, through its aforementioned performance, will effectively maximize the fatigue life at the design stage which results in a reduction of post-processing demand and a decrease in production costs. Moreover, it boosts the advancement of AM technology to replace the conventional methods paving the path to sustainable manufacturing and minimizing environmental impacts.