Avoiding Asymmetric Flows Caused by Off-Centre Pouring into the Tundish: A Novel Impact Pot Design

The practical experiments were carried out using a model representing a twin strand slab casting tundish without any additional flow modifiers except an impact pot. The water model was operated on a scale of 1:3, fulfilling Froude similarity. According to the literature [7], this approach is most likely to simulate flow phenomena of the corresponding full-scale system accurately. The Froude number, which is the ratio between inertial and gravitational forces, is defined by equation 7:

(7)

Where u is the flow velocity, g is the gravitational acceleration, and l is the characteristic length of the system [1]. Table I provides information on the operating conditions considered in the model and values of the real application.

To characterise the flow performance of the given tundish setup, the stimulus response technique, in which the concentration of an injected tracer fluid (e.g., dye) is recorded at the outlets, was applied. Analysis of the measured curve provides key information on the flow characteristics in the tundish such as the minimum residence time and other parameters. Besides the dye concentration measurements, a video was recorded synchronously to get an impression of the general flow characteristics [1].

The first step in the development of a solution to avoid asymmetric flows into the tundish was to understand in detail how the flow develops in the impact zone when an off-centre jet situation occurs. A two-strand slab tundish was chosen as a reference for this study. Figure 2 shows the tundish geometry and positioning of the off-centre jet in the impact pot.

After the incoming jet enters the impact pot, it circulates inside the box and then flows upward towards the slag layer. This process dissipates some of the kinetic energy of the jet, reducing the average velocities of the flow downstream and is one of the reasons impact pots are widely used in the industry.

Figure 3 shows the velocity contours on the bottom surface of the impact pot and Figure 4 depicts the contours of the upward flow velocities as a cross section of the impact pot’s top surface, with both figures comparing centred (Figures 3a and 4a) and off-centre jet conditions (Figures 3b and 4b).

The results illustrate that for the centred jet case, the flow is distributed evenly across the cross section, which would result in a symmetric flow into the tundish, as the flow is distributed equally in all directions. However, for the
off-centre case, the upward flow is strongly concentrated on the opposite side of the incoming jet, as that is the path of least resistance. Such behaviour would very likely result in an asymmetric tundish flow, as there is a clear preferred flow direction. Figure 5 shows contours of the flow on the top surface of the molten steel bath, downstream of the contours shown in Figures 3 and 4, comparing the centred and off-centre conditions. An asymmetric tundish flow can be clearly observed for the off-centre case, in all directions (Figure 5b). Figure 6 shows dye dispersion pictures taken from the front of the tundish at the same flow time in the experiment (i.e., 10 seconds) for both configurations in the water modelling study. The asymmetry of the flow field in the off-centre jet configuration can also be clearly seen, with the incoming jet off-centre to the left side and the downstream flow asymmetric towards the strand on the right side.

If it occurs in a real caster, an asymmetric tundish flow is likely to cause several issues, including accelerated and uneven refractory wear in the region of the preferred flow path, temperature and steel chemical composition differences between strands, shorter residence times, nonmetallic inclusions flowing into the moulds, and a higher tendency of vortex formation due to the rotational flow. Such a scenario created the strong motivation to develop an alternative impact pot solution that would be able to handle off-centre jet conditions, which unfortunately are common in real casters.

Another comparison between the new development and designs with exclusively horizontal barriers or exclusively vertical barriers can be seen in Figure 9, where a lateral view is depicted, and the incoming jet impinges off-centre to the right. When there are only horizontal deflectors, the concentrated jet on the left side is clearly visible and an angled and strong upward flow from right to left occurs (Figure 9a). For the configuration with exclusively vertical barriers, even though it avoids the concentrated or angled upward jets, it shows the disadvantages of lacking mechanisms to dissipate kinetic energy, as this is typically achieved by collisions of the upward flow with horizontal barriers. Thus, stronger upward velocities are seen in this case (Figure 9b). The absence of horizontal barriers also makes it not possible to homogenise the flow between adjacent vertical channels, as achieved by the new development, which is a crucial aspect of avoiding asymmetric flows into the tundish.

Figure 10 shows a comparison of the three different designs performed using water modelling experiments. The design with horizontal barriers (Figure 10a) displays a strong bias towards the right strand already at t = 10 seconds, whereas the design with exclusively vertical barriers has a bias towards the left strand at t = 20 seconds (Figure 10b). The best results in terms of mitigating asymmetric flow were obtained for the new development, with no bias towards any of the strands (Figure 10c).

Figure 11 shows the RTD curves obtained for the three compared configurations. The designs with exclusively horizontal barriers or exclusively vertical barriers show visible differences between the curves for each of the strands, which is a quantification of asymmetry in the tundish flow. In contrast, the new development shows an almost perfect match between the curves for each strand, even though the incoming jet was asymmetric. These results confirm the effectiveness of this design as a solution for avoiding asymmetric flows into the tundish.

Once the effectiveness of the solution developed for avoiding asymmetric flows into the tundish had been validated through water modelling experiments, the next step was to understand which design variables were most influential on the desired outcome. Such a study was performed through a design of experiments (DOE) in which each sample was the CFD result for a different design. About 250 different designs were simulated and the corresponding minimum residence times for each sample were plotted in a chart (Figure 12). For each strand, the minimum residence times were normalised according to the average minimum residence time obtained among all the samples. If the chart is divided in four quadrants, with the axes located at the normalised average value for all samples, the results shown in the top-right quadrant will be the samples with an above average minimum residence time for both strands. From a practical perspective, this is the most desirable outcome, as higher minimum residence times are associated with several benefits for the process and ideally should be obtained for both strands. Analysing the plot, it can be concluded that, from all the samples simulated, only a fraction of those are desirable designs. Therefore, by studying the design dimensions of the desirable designs compared to the full design range evaluated, some insights about the influence of each design parameter can be obtained.

For example, this method can be applied to evaluate the influence of the height of the vertical barrier, as shown in Figure 13. Even though an equal number of designs were simulated for each range of this design parameter, it was only above a certain threshold value that the designs were in the desirable quadrant. This means that not only the height of the vertical barrier is an important design parameter, but also that it should not be too low, otherwise its effect on the flow will not be noticed and the design will fail to work as it should. Such information is very relevant to design effective impact pots for several different steel plants that might each require a tailor-made solution. The same method can be applied to other design parameters of interest, such as protrusion length of the vertical and horizontal barriers, angles of the barriers, and distance between the barriers.

Figure 14 shows a representation of a specific design that performed very well in the simulations and experiments described in this paper, following the concept of the novel development. The next steps planned are to perform plant trials with this new development and measure the improvements obtained in the caster, particularly when
off-centre or angled jet conditions are present.