The research proposal aimed to achieve the following objectives:
- increasing casting reliability by optimising calcium treatment in low sulphur and high sulphur steel grades through predicting casting behaviour in calcium treated steels by means of a mathematical model;
- investigating the influence of nozzle design and casting parameters on nozzle clogging and steel cleanliness by means of mathematical and physical models.
The main findings of the present work as regards inclusion scenario can be thus summarized:
- inclusion composition is strictly related to steel composition; in particular Ca, Al, Mn, S and total O content have been found as the elements most influencing the composition of inclusions; also Nitrogen pick-up during casting has been found influencing the final inclusion composition, particularly as refers to Al2O3 content;
- the composition of inclusions found in slabs is significantly different from inclusions found in blooms, due to the different steel composition, particularly in Al, Si and Mn content;
- also the amount of inclusions depends on cast product: blooms show a lower inclusion amount than slabs; nevertheless, no clear evidence of the influence of inclusion amount on castability has been found; some links between inclusion size and castability have been evidenced;
- a complete modification of alumina into 12CaO.7Al2O3 has been rarely found; CaO rich aluminates tend to react with S, resulting in a CaS phase placed at the borders of the inclusions (this modification depends on Ca amount and time, probably occurs during the late stage of casting and does not affect castability);
- rich Al2O3 phases are often found in the central part of the aluminate inclusions, which confirms a recent kinetic interpretation of the alumina transformation by means of calcium treatment; this seems any way not to affect castability;
- heats with clogging behaviour show a higher average Al2O3 content, a wider Al2O3 rich phase an less CaS formation;
- the presence of MgO in inclusions has always been found, in variable amounts, always associated with Al2O3 (Spinel); some links with castability have been found with trends depending on steel grades.
A mathematical model has been developed to predict inclusions precipitation in nozzles on the basis of chemical composition and process parameters, by selecting literature results and by use of metallurgical thermodynamic data and statistical correlations derived from trial heats data. The model is able to work both in predictive mode (by predicting the casting behaviour for a given steel composition) and in operator-guide mode (by suggesting Ca addition to achieve good casting performance).
The prediction is based on melt chemistry, Castability Index calculated by statistical correlations, Agglomeration Index calculated by the ratio between MID (Mean Inclusion Diameter) and MIPD (Mean Inter-Particle Distance), and Deposition Rate.
The model has been tested with data from industrial heats. Perfomance was found satisfactory in extreme conditions, while some refining work is still needed to improve its sensitivity in medium conditions.
Attempts in use of neural networks for predictive purpose have also been performed with encouraging results, which are quoted in this report. This technique could also be included in the model in order to enhance its reliability. This work will be developed out of the frame of the present project.
Theoretical considerations, physical model trials and numerical simulations were performed to investigate the phenomena of clogging in feeding systems. A simple model was developed which allows identification and quantification of clogging during casting by on-line measuring of the control system movement. Within this model clogging was correlated to the pressure conditions in the feeding system. This correlation was confirmed in physical model trials and numerical computations. With a new numerical model, inclusion deposition has been simulated for a stopper rod controlled feeding
system. Highest deposition rates were found in re-circulation zones in the upper nozzle and around the ports. It has been shown that an increasing stopper rod opening area effects smaller re-circulation areas and therefore less clogging is expected. A gas film at the upper nozzle wall, which can prevent inclusion precipitation and therefore clogging, has been simulated with an annular gas injection in the upper nozzle. This was not achieved by gas feeding at the stopper rod tip. It has been shown that gas injection leads to a reduction of the low pressure in the upper nozzle and therefore less air aspiration is expected.