The PhD thesis investigates the crucial role of melt preparation in achieving high-quality aluminum casting. Casting defects, like cracks, voids, and porosities, are linked to melt quality, emphasizing the importance of meticulous preparation and casting parameter optimization. Current limitations in equipment hinder real-time assessment of melt quality, delaying checks and increasing time and energy investments. However, recent advancements in evaluating molten aluminum before casting offer potential for improved final product quality and operational efficiency. This progress allows engineers more time to refine the melt pre-casting. Key information on iron intermetallic compounds, precipitate concentration, eutectic phase modifications, and grain refinement greatly impacts recycling efforts. Acquiring such insights usually requires specialized equipment and time-intensive procedures, making real-time assessment during melt preparation impractical. Establishing links between these parameters and readily available thermodynamic data is crucial for reducing waste, time, and energy in final product manufacturing. Current predictive models for non-ferrous alloys focus on alloying element modification and chemical composition data. Yet, understanding the relationship between thermal analysis data and mechanical/microstructural properties in solidified non-ferrous alloys remains incomplete. Predictive models could aid in estimating microstructural defects like hot tearing and segregation across different cooling rates, simulating industrial component solidification processes. Data from thermal analysis at various cooling rates can serve as a foundation for machine/deep learning implementation. With fundamental melt composition and cooling regime data, predicting solidified part parameters becomes feasible without direct thermal analysis. This holds promise for process engineers by providing insights into melt quality and aiding in mold design. Leveraging machine/deep learning for solidification parameter correlation could lead to rapid melt quality estimation. The PhD project aims to uncover crucial connections between quick cooling curves and poorly understood solidified part characteristics. Predictive models will simulate processes using machine learning techniques, revolutionizing melt quality control and enhancing final industrial component quality.

Developing of a predictive model for non-ferrous casting alloys melt quality assessment

YAZDANPANAH, ARSHAD
2024

Abstract

The PhD thesis investigates the crucial role of melt preparation in achieving high-quality aluminum casting. Casting defects, like cracks, voids, and porosities, are linked to melt quality, emphasizing the importance of meticulous preparation and casting parameter optimization. Current limitations in equipment hinder real-time assessment of melt quality, delaying checks and increasing time and energy investments. However, recent advancements in evaluating molten aluminum before casting offer potential for improved final product quality and operational efficiency. This progress allows engineers more time to refine the melt pre-casting. Key information on iron intermetallic compounds, precipitate concentration, eutectic phase modifications, and grain refinement greatly impacts recycling efforts. Acquiring such insights usually requires specialized equipment and time-intensive procedures, making real-time assessment during melt preparation impractical. Establishing links between these parameters and readily available thermodynamic data is crucial for reducing waste, time, and energy in final product manufacturing. Current predictive models for non-ferrous alloys focus on alloying element modification and chemical composition data. Yet, understanding the relationship between thermal analysis data and mechanical/microstructural properties in solidified non-ferrous alloys remains incomplete. Predictive models could aid in estimating microstructural defects like hot tearing and segregation across different cooling rates, simulating industrial component solidification processes. Data from thermal analysis at various cooling rates can serve as a foundation for machine/deep learning implementation. With fundamental melt composition and cooling regime data, predicting solidified part parameters becomes feasible without direct thermal analysis. This holds promise for process engineers by providing insights into melt quality and aiding in mold design. Leveraging machine/deep learning for solidification parameter correlation could lead to rapid melt quality estimation. The PhD project aims to uncover crucial connections between quick cooling curves and poorly understood solidified part characteristics. Predictive models will simulate processes using machine learning techniques, revolutionizing melt quality control and enhancing final industrial component quality.
22-mar-2024
Inglese
DABALA', MANUELE
Università degli studi di Padova
File in questo prodotto:
File Dimensione Formato  
Thesis-Arshad-Yazdanpanah.pdf

embargo fino al 22/03/2027

Dimensione 10.46 MB
Formato Adobe PDF
10.46 MB Adobe PDF

I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.14242/97132
Il codice NBN di questa tesi è URN:NBN:IT:UNIPD-97132