AZTERLAN Metallurgy Research Centre continues to develop real-time simulation tools for the foundry industry to ensure the quality and performance of cast parts. The new model under development is aimed at predicting the microstructure of spheroidal cast iron and is being developed in collaboration with Betsaide and Draxton Atxondo foundries.
The properties and behavior of castings depend on their microstructure, which is formed during the solidification process as a result of the characteristics of the metal and its cooling rate inside the mold. Currently, thermal analysis tools, such as Thermolan®, developed by AZTERLAN, are available on the market to asses metallurgical quality. These tools allow monitoring the solidification curve of the metal and extracting valuable information to predict key parameters for determining the quality of the metal used to manufacture the components (C content, Si content, free Mg, number of nodules, etc.).
Metallurgical quality, combined with part geometry and mold characteristics—aspects that affect the cooling rate of different areas of the parts—makes it possible to predict the microstructure and the risk of shrinkage defects in the parts. Targeting this issue, AZTERLAN has also developed the Kasandra® predictive simulator, primarily aimed at forecasting the risk of micro-shrinkages appearing in parts.
To go a step further in the development of industrially applicable tools aimed at ensuring part quality, AZTERLAN, in collaboration with the foundry companies Betsaide and Draxton Atxondo, is developing a new model to predict the microstructure of spheroidal graphite castings. The new model will be based on the constitutive equations that govern the nucleation and growth processes of both spheroidal graphite and austenite in nodular cast iron alloys.
In the words of AZTERLAN researcher Dr. Ana Fernández, “Our approach to prediction is completely different from what can be done with current tools. We are looking to develop a fundamental model based on the nucleation and growth equations of solidifying phases over the time they release latent heat. The new model will allow us to predict the number of austenite nodules and grains throughout the entire solidification process, as well as simulate, at all times, the solidification heat released and, therefore, its solidification curve.”
For this new development, curves obtained with the Thermolan® thermal analysis system and from thermocouples located castings will be used to calibrate the parameters of the constitutive model. This will make it possible to simulate all types of metal microstructures/qualities. “We aim at modelling all types of microstructures, including bimodal microstructures: coarse graphites formed at the beginning of the solidification and very fine graphites formed at the end of solidification. These last graphites have had little time to grow, but, right at the end of solidification, their formation makes a significant contribution to compensating for the expansion/contraction balance. Graphites created in the last stage of solidificaion are, therefore, very effective in reducing the tendency for micro-shrikages apparition in late solidification zones.”
Artificial Intelligence and real manufacturing data from Betsaide and Draxton Atxondo will contribute to the optimization of the new model
Additionally, with all the historical data available at AZTERLAN and updated real manufacturing data from Draxton Atxondo and Betsaide foundries (users of the Thermolan® thermal analysis system), and by comparing this data with the modeling results, Artificial Intelligence will be used to optimize the new model. “Using AI, we will group the results into as many groups as necessary to establish the optimal correspondence between the experimental results (types of curves) and the model.”
The final model will be calibrated with experimental tests in AZTERLAN on different eutectic and hypoeutectic alloys, using results from parts with different thermal moduli. Validation will be performed industrially by correlating the Thermolan® system curves with the microstructure of the parts manufactured by Draxton Atxondo and Betsaide.
Finally, the new model will be integrated into the Kasandra® predictive simulation tool.
Kasandra® software predicts the risk of apparition of defects in castings in real time, based on the actual metallurgical quality of the production metal. (go to Kasandra® page.)
This development is being carried out within the framework of the MICROFUN project “Advanced modeling of the microstructure and solidification curve, including its integration into the simulation of spheroidal cast parts,” funded by the Provincial Council of Bizkaia.
