SOME INVESTIGATIONS ON THE USE OF FLOW MODIFIERS IN CONTINUOUS CASTING TUNDISH

Abstract

Continuous casting process is most widely used process for production of slabs and billets by steelmakers in today's time of global energy crisis. The set up consists of three components namely ladle, tundish and mold. The ladle receives the molten material from furnace and delivers it into the tundish from where it is fed into the water cooled copper mold. Initially the main functions of the tundish was considered to be as a buffer vessel which maintains an even, relatively uniform flow as the casting ladle is being emptied and replaced by a new teeming ladle. Over a period of time, however, it was realized that the functions of tundish go beyond being a buffer between the ladle and mold. In fact, being the last reservoir for the molten metal before its entry into the mold, tundish plays an important role in deciding the quality of continuous cast products. A proper flow pattern inside the tundish directly affects the performance of the tundish in terms of metallurgical phenomena like mixing, grade change and inclusion removal processes inside the tundish. In other words, if the flow of metal in the tundish is not properly controlled, it may even deteriorate the 'Quality' of steel produced in the ladle. Thus tundishes, in terms of their shape and use of flow modifiers (dams, turbo stop etc.) are designed to provide optimum flow characteristics. One of the few important objectives with which a tundish is effectively utilized in the formation of intermixed grade steels during ladle change-over in continuous casting process. During this, when the (:)Id ladle (representing old grade of steel) is emptied, a new ladle (representing new grade of steel) replaces the old one resulting into continuous supply of molten steel to the mould. Mixing of the old and the new grade steel in tundish as well as in the strand produce intermixed grade steel, the composition of which falls between the composition of the old and the new grade steel and is generally considered to be of lesser demand as compared to the old or the new grade. Although industries use few methods to avoid formation of intermixed amount, the sequential casting using ladle change over still remains the only practical method to get the slabs/billets through continuous casting route by large steelmakers. Since the ladle change-over method results into formation of large intermixed amount which is very difficult to be avoided completely, efforts can be made to minimize it. Minimizing the intermixed amount has direct impact on the saving of cost associated with low demand intermixed steels formed during ladle change-over. iv A thorough review of literature shows that the tundish design may be improved by several techniques such as change in position of outlets, use of flow modifiers (Dam, Weir, and Advanced Pouring Box etc), use of shroud immersion depth etc. The present investigation is an endeavor to study the improvement in tundish design with the use of flow modifiers in tundish keeping in mind the objective of minimizing the intermixed amount in tundish during ladle change over. The investigation has been carried out in a twin strand slab caster tundish which is boat shaped and a six-strand billet caster tundish of delta shape. The present research endeavor primarily consists of the investigation on two aspects. The first aspect deals with experimental and numerical investigation of the effect of flow modifiers on intermixed amount formation in slab caster and billet caster. The second aspect mentions about Ike ANN model preparation for predicting the effet of flow modifiers on the intermix amount formation in tundish. Experimental investigation was conducted on a 1/3.5th scale model tundish (slab cster) and l/Lith scale billet caster tundish. The tundishes were fabricated with Perspex.sheet. Water was taken as the old grade fluid representing old grade steel and salted water was used representing new grade of steel. Step input tracer concentration study was carried out after attainment of steady state velocity field and intermixed amount was calculated from different outlets. The experimental results of the intermixing process was used to validate with the numerically obtained results where the concentration of new grade steel is monitored at the outlet of the tundish in case of old ladle carrying old grade steel being replaced by new ladle with new grade steel. A close match was found between the experimental and numerically obtained results. The study was further extended numerically to study the effect of flow modifiers like dam, weir, turbo-stop and combination of dam and turbo stop on the intermixed amount in tundish. It was found that while placement of dam led to reduction in intermixed amount, no improvement is observed in terms of reduction in intermixed amount when weir was used as a flow modifier. Change in height and position of dam do not have much significant effect on intermixed amount. Use of turbo-stop also does not result into decrease in intermixed amount. However, when turbo-stop and dam both are used, it results into decrease in intermixed amount. Investigation of intermixed amount formation in billet caster tundish has been carried out with and without the use of various flow modifiers like dam (before outlet/outlets), turbo-stop and combination of dam and turbo-stop. Experimental investigation was carried out for bare tundish and tundish with dam before near outlet for validation purpose and a reasonably good match was found between numerically and experimentally obtained 17-curve. Numerical investigation was carried out using a dam before either of the near, middle or the far outlet. Eight different positions were considered for the dam placement at the bottom surface of the tundish. Three different dam height, dimensionless heights of 0.3, 0.4 and 0.5 were used to study the effect of dam height on intermixed amount. It was found that when small dam (dimensionless height 0.3) was placed in tundish before near outlet, reduction in intermixed amount was observed through each of the outlet (near, middle and far) as compared to bare tundish. Movement of dam past near outlet i.e. in between the near and middle outlet resulted into short circuiting through near outlet similar to that observed for bare tundish. If the dam is further moved between medium and far outlet, it did not see improvement in results in terms of reduction in intermixed amount. It is seen that as the dam height is increased, intermixed amount through the near outlet decreases. The minimum average intermixing is observed for tundish with medium/ large height dam at position b i.e. placement of dam closer to near outlet. About 12% reduction in average intermixed amount was found for 20:80: grade specifications as compared to bare tundish. Dams were also placed before each of the outlet to see its effect on intermixing. The dams were kept at fixed position and heights of dams were varied (three different heights). The average intermixed amount was found to be more when smaller height of dam was,placed before near outlet and placement of larger dam before near outlet was found to be effective in terms of reduction in average intermixed amount. The minimum average intermixed amount is formed when height of the near, middle and far dam were taken in decreasing order. About 26% reduction in intermixed amount was found for 20:80 grade specifications as compared to bare tundish. Flow modifiers like turbo-stop (alone) and turbo-stop along with dam were also used in tundish. The use of turbo-stop along with three dams, each before near, middle and far outlet gave minimum intermixed amount. Attempt was made to analyze the change in shape (slope) of the F-curve with time by dividing the. F-curve into three time zones, time zone 1(0 to 500 sec), time zone 11(501 to 1000 sec) and time zone 111(1001 to 1600 sec) to find which of the time zone plays important role in controlling the intermixed amount from a particular outlet. It was found that when small height dam was used, intermixed amount through near outlet was controlled by Intermixing vi during Time zone II and III whereas intermixed amount through middle and far outlet were controlled by Time Zone II and Time Zone III respectively. Increasing the dam height to 0.4 (medium height) noticed that intermixed amount through near outlet is controlled by time zone I whereas through middle and far outlet, time zone II was the controlling factor. The effect of time zone on intermixed amount when dam of large height was used was seen to be similar to that for the medium height of the dam. The slope of F-curves (obtained trough placement of dam before each of the outlets) has also been analyzed. It was found that time zone I governs the intermixed amount at near outlet. Time zone II and III control the intermixing at middle outlet and zone 1 at far outlet when dams of equal heights were used in tundish. Further analysis reveals that when dam of varying height is used, any particular trend is not observed. Artificial Neural Network based modeling approach has been adopted to find the effect of dam position and height in multi-strand tundish with an aim to obtain global minimum value of intermixed amount. As there can be a large number of combinations of dam height and positions which can affect the intermixed amount formation, it will be an arduous task to find the intermixed amount for all these cases computationally or experimentally. The intermixed amount for tundish with placement of dam at different positions was been found numerically and using these results, back propagation neural network model was trained for predicting the intermixed amount for more number of combinations of dam height and positions. It was found that when the dimensionless height of dam was increased from 0.45 to 0.5, there was no change in the intermixed amount with the use of dam for most of the positions. Average intermixed amount is observed to be minimum when dam of dimensionless height 0.45/0.5 is placed closer to near outlet. The ANN based modeling approach has also been applied when dam is placed before each of the outlet. Overall intermixed amount was found to be minimum when all dam of larger height is placed at dimensionless position of 0.151, 0.390, and 0.451. About 30 % reduction in average intermixed amount is observed as compared to bare tundish. "Artificial neural network" was used to study the effect of change in position and height of dam used in the billet caster tundish on the intermixed amount. Because there can be very large combinations of dam height and position, and numerical computation requires too large computational time to analyze all those cases, ANN (Artificial Neural Network) has been used to predict intermixed amount for varying geometrical parameters like height and position of vii the dam. The intermixed amount for tundish with placement of dam of different height at different positions has been found for near, middle and far outlet respectively with the Numerical simulation. These results were further used to train the neural network through Back propagation Neural Network Learning Algorithm. Successfully trained networks are further used for predicting the intermixed amount for all possible combinations of dam height and position so that a suitable combination of dam and height may be suggested to steelmakers for decreasing the intermixed grade amount. The ANN based study has been divided into two phases, in first phase predictive model has been developed for use of one dam in tundish. Two hidden layers with nine neurons in each were adopted in order to keep MSE (mean squared error) to be minimum. Use of one dam predicted that near outlet produces the minimum intermixed amount when dam is placed between near and middle outlet, but in case of far and middle outlet the same is true when dam is placed before near outlet. It was found that dam of dimensionless height 0.45 and 0.5 predicted the same trend at most of the positions. Minimum overall intermixed amount is found when dam is placed before the near outlet for dimensionless height 0.45 and 0.5. Second phase predictive modeling is developed when dam is used before each of the outlet. Two hidden layers with 12 neurons in each were considered to keep MSE minimum. When dam was used before each of the outlets predictive modeling suggested that placement of dams closer to the outlets, gave better results in terms of reduction in intermixed amount from near and far outlets, however no significant reduction in intermixed amount is observed through middle outlet. If the height of dam is reduced or increased by 0.05 margins it did not see any improvement in results. Overall intermixed amount was found minimal when all dam of larger height is placed at dimensionless position of 0.151, 0.390, and 0.451. Though several other combinations of dam's positions and heights are possible, and trail and error procedure has also been adopted for other combinations but due to no betterment of results no further prediction was carried out.