Advertisement

International Journal of Metalcasting

, Volume 13, Issue 2, pp 255–272 | Cite as

Effect of Pouring Conditions and Gating System Design on Air Entrainment During Mold Filling

  • Seyyed Hojjat Majidi
  • Christoph BeckermannEmail author
Article

Abstract

Air entrainment during mold filling is a major source of oxide inclusion formation in metal casting. A model was recently developed by the authors to predict the volumetric air entrainment during pouring of metal castings. In the course of validating the model with experimental data for plunging liquid jets, it was shown that the air entrainment rate during mold filling depends fundamentally on the velocity and diameter of the jet formed by the pouring stream. In this study, the effect of more complex pouring conditions and gating system design on air entrainment is examined. Simulations are performed investigating the air entrainment characteristics of castings filled without a gating system, and with bottom-gated and side-gated filling systems. Results indicate that reducing the head height and pouring time, and the addition of a nozzle extension significantly reduces the air entrainment. In addition, using an offset pouring basin with a stopper and pressurizing the gating system further reduces the volume of entrained air. Simulation results also show that the generation of vortex flows inside the filling system is beneficial in reducing free surface turbulence, which results in less air entrainment and oxide inclusion formation during mold filling.

Keywords

oxide inclusions air entrainment model gating system free surface turbulence vortex flow casting simulation 

Introduction

Defects in cast parts affect the material properties and machinability of cast metals, resulting in scrapped castings or their premature failure. Oxide inclusions are among casting defects that are troublesome and damaging to casting performance. Oxygen reacts with the chemical constituents in the liquid metal during mold filling to form oxide inclusions which can be transported into the final cast part. In aluminum alloy castings, as the liquid metal experiences free surface turbulence, the dry side of the oxide covering the melt becomes in contact with the dry side of the other oxide and forms double oxide films or bifilms1 leading to inclusions entrapped in the casting. In carbon and low-alloy steel castings, the reaction between oxygen in the atmosphere and the most reactive elements in deoxidized steel results in reoxidation inclusions.2 In ductile iron casting, the reaction of magnesium oxide and silica during magnesium treatment and further during mold filling is responsible for the formation of dross inclusions.3

Oxide inclusions can form during mold filling as a result of air entrainment. In free surface flows, air is entrained at flow discontinuities when the liquid surface experiences turbulence. Important examples of air entraining flows are: a liquid jet plunging into a pool, a breaking wave, and a hydraulic jump, where a fast moving liquid discharges into a low-velocity atmosphere. Air entrainment manifests itself in the form of bubbles. For a plunging liquid jet, experiments have shown that the amount of air entrainment depends on the velocity, diameter, and turbulence level of the liquid jet.4, 5, 6, 7 During filling of metal castings, the pouring conditions and gating system design affect the air entrainment, and hence the extent of oxide inclusion formation. Examples of air entraining flows in casting gating systems include: liquid metal impingement inside sprues, breaking waves in runners, falling jets inside of the mold cavity, and rising vertical jets or fountains formed in bottom-gated castings. Once air is entrained, the oxygen in the entrained air reacts with liquid metal constituents to form oxide inclusions. Upon formation, these inclusions are transported with the liquid metal and can ultimately end up as nonmetallic inclusions in the solidified casting. Therefore, pouring liquid metals with less air entrainment is a practical and effective method for the production of clean and inclusion free, metal castings. Air entrainment should not be confused with air entrapment. Air entrapment refers to the creation of air pockets when liquid metal fronts converge upon each other or on a mold wall. It can occur due to poorly designed ingates and in addition when the mold is not properly vented or the mold permeability is insufficient. Air entrainment, on the other hand, is always associated with the formation of small bubbles at flow discontinuities.

Several experimental and numerical studies have been conducted to investigate the effects of pouring conditions and filling system design on air/oxide entrainment. Studies have recommended that reducing the velocity of liquid metal at ingates has a significant effect on minimizing air entrainment and oxide defect formation. Several criteria have been proposed for a critical ingate velocity, below which free surface turbulence in the flow entering the casting is avoided. The balance between inertial and surface tension forces acting on the liquid metal has been suggested to predict the critical velocity at the ingate for the onset of entrainment.1,8,9 The critical entrainment onset velocity is calculated to be 0.45 and 0.5 m s−1 for pure liquid aluminum and iron, respectively. The model developed by Lai et al.10 compares the simulated instantaneous free surface area of the melt to the instantaneous free surface area assuming the liquid metal fills the mold quiescently. The difference between these two free surface areas is then used to determine the oxide entrainment magnitude. The results of this study found the instantaneous free surface area increases significantly above ingate velocities of 0.4–0.5 m s−1, and compared favorably with the previous findings.1 Hsu et al.11 proposed using a vortex gate to reduce the free surface turbulence at the entrance into the casting. Creating vortex flows inside the gating systems has been championed in Campbell’s work.1 Campbell argues that vortex wells and gates efficiently absorb the kinetic energy of the liquid metal, and describes how these gating features have been successfully implemented in steel casting.

Several models have been developed to predict entraining events in aluminum casting based on velocity vectors, free surface normals, and liquid volume fractions in free surface cells. As described earlier, oxide entrainment in aluminum castings requires dry sides of oxide films to be in contact with each other. For this to occur, the velocity directions of the two overlapping flow streams must be opposite.1,8 Using this criterion, Yang et al.12 developed a model to predict the entrainment of oxides in aluminum castings. Applying this model, it was shown that a vortex runner is beneficial in reducing the free surface turbulence inside the runner, consequently providing smooth liquid metal flows at ingates. The predicted results were validated through comparison with experimental measurements. The effect of runner design on the formation of oxide inclusions was previously studied experimentally by Latona et al.13 Measurements showed that a pressurized gating system with an area ratio of 4:8:3 (where the lowest melt velocity occurs in the runner) is beneficial in reducing dross inclusions in ductile iron castings. The model developed by Reilly et al.14 uses the velocity vectors, free surface normals, and liquid volume fractions in free surface cells to define a series of Boolean logic criteria for predicting the occurrence of entraining events. The model was applied to several gating systems, and results showed the majority of oxides are entrained at the surface of the pouring basin and at the sprue base. Yue15 modeled three gating orientations: direct pour, vertical bottom gate, and horizontal side gate. Results showed that for a constant head height, the filling system with the plunging jet (direct pour) entrained more air compared to the two other filling systems. Castings produced by the plunging jet flow type (direct pour) were less reliable than castings produced by the rising jet flow (vertical bottom gate) and returning wave flow (horizontal side gate) as demonstrated by bending strength test results for the three casting filling methods.

Few attempts have been made to simulate air entrainment and compare predictions to water modeling experimental results under varying pouring parameters and conditions. Such results can be used to guide the best practices of pouring liquid metals. Experimental water modeling studies have shown that shorter falling heights and filling times reduce the volume of entrained air.16,17 Additionally, it has been shown through water modeling that use of several gating system features significantly reduces air entrainment: an offset pouring basin with a step before the sprue entrance, a nozzle extension submerged into the pouring basin, and a whirl gate.18,19

A model for predicting the entrained air during mold filling allows casting process engineers to evaluate pouring conditions and gating systems. Based on the work by Ma et al.,20 the authors have recently developed a model21 for predicting the local air entrainment rate resulting from disturbances at the free surface of the liquid metal flow. This sub-grid air entrainment model was implemented in a casting filling simulation software to calculate local air entrainment rates during filling. While the authors’ previous work described the model’s development21 and validation by comparison with water modeling experiments,22 the goal of these developments is to apply the model to liquid metals and predict the air entrainment during mold filling. In the present study, the capabilities of the previously developed air entrainment model are demonstrated by simulating liquid metal flow in typical casting filling systems and calculating air entrainment rates. Results presented here provide quantitative comparisons of the effects of various pouring conditions on air entrainment, and hence oxide inclusion formation. The metal casting pouring conditions investigated here are process variables which can be modified to minimize inclusion defects: nozzle diameter, filling time, gating components, and gating ratios.

Air Entrainment Model

The model described here is implemented as part of a standard casting filling simulation23 by performing calculations on flow velocities at the free surface to predict the local air entrainment rate. The casting filling simulation calculates the melt velocity and geometry of the free surface by solving the Navier–Stokes equations with a volume tracking method (VOF method) progressively through time as the casting fills.

Disturbances exist at the surface of free surface flows, for example on the periphery of the liquid jets (Figure 1). These disturbances make the liquid–air interface rough, and air pockets are trapped inside these disturbances as indicated in Figure 1. For a plunging liquid jet, once the liquid jet impinges on the surface of a quiescent pool, the trapped air pockets are entrained into the bulk liquid where they are broken into smaller air bubbles and carried away with the liquid flow. In the model applied here, the local air entrainment rate is a function of the turbulent kinetic energy, \( k \), and the normal derivative of the normal component of the liquid velocity at the free surface interface, \( \partial u_{n} /\partial n \)20:
$$ q = C_{\text{ent}} \frac{k}{g}\frac{{\partial u_{n} }}{\partial n} $$
(1)
where \( q \) is the volumetric air entrainment rate per unit interfacial area, \( C_{\text{ent}} \) is an entrainment coefficient, and \( g \) (m s−2) is the gravitational acceleration. For any free surface flow, air is entrained only if the gradient term in Eqn. 1 is positive, \( \partial u_{n} /\partial n > 0 \). The turbulent kinetic energy in the model is estimated from the sum of the squares of the fluctuating velocity components relative to a spatially averaged mean velocity. Integrating the air entrainment rate per unit interfacial area \( q \) (m s−1) over the interfacial area \( A_{\text{s}} \)(m2), the volumetric air entrainment rate \( Q_{\text{a}} \) (m3 s−1) is calculated as
$$ Q_{\text{a}} = \iint\limits_{{A_{\text{s}} }} {q{\text{d}}A} $$
(2)
Figure 1

Illustration of a plunging liquid jet depicting air bubbles entrained at the impact location. Air pockets trapped inside disturbances along the periphery of the free surface of the jet and important variables are shown.

The entrainment coefficient used in the model was \( C_{\text{ent}} = 0.039 \), and it was determined by calibrating the predicted relative air entrainment rates to experimental measurements reported in Reference 6 for plunging water jets that have variable liquid jet velocities and diameters at low turbulence intensities. A previously published paper21 explains the details of the model including the calculations of the turbulent kinetic energy and normal derivative of the normal component of the liquid velocity at the interface, and the calibration of the entrainment coefficient. Using this air entrainment coefficient value, good agreement between measurements and predictions was achieved for plunging water jets for a range of liquid jet velocities and diameters. An example velocity field for a plunging/free falling liquid jet is shown in Figure 2a with falling height, hj, liquid jet velocity, uj, and jet diameters at the nozzle, dN, and the impact location, dj. Air entrainment results for a liquid jet with constant turbulence intensity are given in Figure 2b, c. The air entrainment is presented throughout this paper using the relative air entrainment, defined as the ratio of volumetric air entrainment rate, \( Q_{\text{a}} \), to the volumetric flow rate of liquid, \( Q_{\text{w}} \). As shown in Figure 2b, with good experimental and model agreement, increasing the liquid jet velocity increases the relative air entrainment rate, since increasing the liquid jet velocity increases the turbulent kinetic energy of the liquid jet. While as shown in Figure 2c, increasing the liquid jet diameter reduces the relative air entrainment rate. The explanation for the effect of jet diameter on air entrainment requires more thought. Increasing the liquid jet diameter increases the perimeter of the liquid jet, which increases the volume of air pockets on the periphery of the liquid jet. This effect increases the volumetric air entrainment rate in proportion to the perimeter of the jet \( Q_{\text{a}} \propto d_{\text{j}} \). However, the relative air entrainment is the volumetric air entrainment rate divided by the volumetric flow rate. Since the flow rate for a circular jet that has a given velocity is proportional to the inverse of the jet area, \( Q_{\text{w}} \propto 1/d_{\text{j}}^{2} \), the relative air entrainment rate is therefore \( Q_{\text{a}} /Q_{\text{w}} \propto 1/d_{\text{j}} \). Therefore, for a given liquid jet velocity, increasing the jet diameter reduces the relative air entrainment as shown for both model and experimental results in Figure 2c.
Figure 2

(a) Velocity field for a free falling liquid jet showing: falling height, hj, liquid jet velocity, uj, and jet diameters at the nozzle, dN, and the impact location, dj, and relative volumetric air entrainment rate as a function of (b) liquid jet velocity and (c) liquid jet diameter at the impact location.

Simulation of Air Entrainment During Mold Filling

The effect of pouring conditions and gating system designs on the air entrainment during casting filling processes was investigated using the model and simulation. For all cases presented here, the cast part is represented by a rectangular block of 406.4 mm (16″) length and width, and 304.8 mm (12″) height. The filling simulations were performed using a commercial casting simulation software,23 and the air entrainment model was implemented based on the software’s velocity and free surface geometry calculations. The material properties for low-alloy steel and furan sand mold were used from the software database. The pouring temperature was 1600 °C (2912 °F) for all cases given here. A uniform mesh with grid spacing of 4.5 mm was used in all cases. Even though the simulations performed here are for low-alloy steel and constant grid spacing, the relative comparisons for air entrainment results between pouring parameters and gating systems should not be affected by the metal type and grid spacing. The readers are advised that the values for relative air entrainment presented here are for relative comparison. As part of ongoing work, experiments are being developed using liquid metals to provide air entrainment measurements to compare with predicted model results.

In order to evaluate three distinct mold filling methods, castings with direct pour (no gating), bottom gating, and side gating were simulated. Geometries and dimensions of the simulation cases without and with gating systems are given in Figure 3. As shown in Figure 3a, for configurations with no gating, the liquid steel is directly poured into the mold cavity. Table 1 lists all cases studied here to predict air entrainment in steel casting. In all of these configurations, the distance between the ladle nozzle exit and the mold inlet was 50.8 mm (2″). The effect of turbulence intensity was neglected in these simulations.
Figure 3

Casting geometry and gating systems used in air entrainment simulations: (a) direct pour (no gating system), (b) bottom-gated system, and (c) side-gated system. Dimensions are in mm.

Table 1

Summary of Air Entrainment Simulation Cases Presented

Trial

Flow rate type

Nozzle extension

Nozzle diameter, dN (mm)

Pouring cup/basin

Vortex component

Gating ratio, Abs:Ar:Ain

Filling time, t (s)

No gating—direct pour

 1

Constant

No

25.4

25.4

 2

Constant

Halfway

25.4

25.4

 3

Constant

All way

25.4

25.4

 4

Constant

No

31.75

25.4

 5

Constant

No

38.1

25.4

 6

Constant

No

50.8

25.4

 7

Constant

No

63.5

12.6

 8

Constant

No

63.5

16.8

 9

Constant

No

63.5

25.4

 10

Constant

Halfway

63.5

25.4

 11

Constant

All way

63.5

25.4

 12

Constant

No

63.5

50.5

 13

Constant

No

76.2

25.4

 14

Variable

No

25.4

25.4

 15

Variable

No

31.75

25.4

 16

Variable

No

38.1

25.4

 17

Variable

No

50.8

25.4

 18

Variable

No

63.5

25.4

 19

Variable

No

76.2

25.4

Bottom gated

 20

Constant

63.5

Cone cup

None

1:1:5.4

30.6

 21

Constant

63.5

Offset basin

None

1:1:5.4

31.3

 22

Constant

63.5

Offset basin with stopper

None

1:1:5.4

31.4

 23

Constant

63.5

Offset basin

Vortex sprue

1:1:5.4

31.2

 24

Constant

63.5

Cone cup

Vortex well

1:1:5.4

32.3

 25

Constant

63.5

Cone cup

Vortex well and gate

1:1:5.4

34.5

Side gated

 26

Constant

63.5

Cone cup

None

1:2:2

30.3

 27

Constant

63.5

Cone cup

None

4:8:3

30.1

 28

Constant

63.5

Cone cup

None

1:1:1

29.1

 29

Constant

63.5

Cone cup

None

1:1:1.7

28.1

Cases are organized by gating system and conditions used

The effects of nozzle diameter, \( d_{\text{N}} \), flow rate type (constant vs variable flow rate), nozzle extension, and fill time (or volumetric flow rate), \( t_{\text{fill}} \), were studied for the no gating (direct pour) configurations. To study the effect of nozzle diameter on the air entrainment, six nozzle diameters were simulated, with smallest and largest nozzle diameters being 25.4 mm and 76.2 mm, respectively. In addition, for each nozzle diameter, a constant and a variable flow rate was examined. For configurations with variable flow rate, the volumetric flow rate is not constant during mold filling. The flow rate profile is kept the same for all nozzle diameters. To calculate the variable flow rate profile, two assumptions were made. First, it is assumed that the maximum velocity of the liquid jet does not exceed \( u_{\text{j}} = 6 \) m s−1. This velocity is reported as a critical transition velocity where the mechanism of air entrainment changes and the disturbances on the periphery of the liquid jet are not solely responsible for all of the air entrainment.4, 5, 6, 7 A maximum volumetric flow rate was selected based on the maximum velocity for the smallest nozzle diameter \( d_{\text{N}} = 25.4 \) mm. Second, an average filling time of \( t_{\text{fill}} = 25.4 \) s is assumed corresponding to a “base” case that has a constant volumetric flow rate of \( Q_{\text{s}} = 2 \) × 10−3 m3 s−1. Moreover, the effect of nozzle extension was studied for two nozzle diameters, \( d_{\text{N}} = 47.6 \) mm and 63.5 mm. For configurations with the nozzle extension, an extension with the same diameter as the nozzle was added to the nozzle exit. Nozzle extension with lengths of 203.2 mm (nozzle extended halfway) and 330.2 mm (nozzle extended all way down to the bottom of the casting) were simulated. Additionally, four filling times (or constant/average flow rates) were studied to examine the effect of the filling time on the air entrainment.

The effects of pouring cup, pouring basin, sprue, and well designs on air entrainment were investigated. Results for simulated cases presented here include: conical pouring cup, offset pouring basins with and without stopper, use of vortex sprue, and use of vortex well. A vortex gate was simulated with the bottom-gated configuration, and for the side-gated configuration, several gating ratios’ were examined. The geometries and dimensions of the pouring cup/offset basin and vortex components are shown in Figures 4 and 5, respectively. In all configurations with gating systems, the nozzle diameter and volumetric flow rate of liquid steel are constant with \( d_{\text{N}} = 63.5 \) mm and \( Q_{\text{s}} = 2 \) × 10−3 m3 s−1, respectively, and the pouring cup/basin height is \( h_{\text{cup}} = 152.4 \) mm. The slight variation in filling times, indicated in Table 1, is due to the difference in gating system volume.
Figure 4

Geometries and dimensions (in mm) for (a) conical pouring cup and (b) offset pouring basin used in air entrainment simulations.

Figure 5

Geometries and dimensions (in mm) of vortex generating gating components: (a) offset basin for creating vortex sprue, (b) vortex gate, and (c) vortex well.

A single case was simulated to study the effect of stopper in offset pouring basins. The stopper is placed at the sprue entrance, sealing the entrance of the sprue. The liquid is poured from the ladle to the basin, and the stopper is removed when the basin is filled to a certain level. Once the stopper is removed, the liquid fills the sprue and the gating system. After removing the stopper, liquid metal is continuously provided from the ladle to maintain a constant falling height from the ladle lip to impact location inside basin. Since moving objects cannot be defined in the software used, a porous filter with large pressure loss coefficient was defined at the sprue entrance to apply the “stopper” effect on the offset basin, and a script file was used to define the time and conditions for removing the stopper. While the flow is blocked at the filter location, liquid steel is allowed to fill the basin for a time of 2 s. Then, the stopper is removed accordingly, and liquid steel is allowed to fill the gating system and casting while continue filling the basin. For all the bottom-gated configurations, the sprue and runner diameters are \( d_{\text{s}} = d_{\text{r}} = 76.2 \) mm, and the ingate diameter is \( d_{\text{in}} = 177.8 \) mm. Details of the gating system components for the side-gated filling systems are shown in Table 2.
Table 2

Overview of the Sprue, Runner, and Ingate Geometry for Side-Gated Filling System Simulation Cases Presented in Paper

Gating ratio, Abs:Ar:Ain

Sprue base diameter, dbs (mm)

Runner width, wr (mm)

Runner height, hr (mm)

Ingate width, win (mm)

Ingate height, hin (mm)

1:2:2

63.5

114.3

55.4

127.0

49.9

4:8:3

63.5

114.3

55.4

93.5

25.4

1:1:1

63.5

88.9

35.6

127.0

24.9

1:1:1.7

47.0

76.2

22.8

127.0

22.8

All cases used a nozzle diameter and volumetric flow rate of \( d_{\text{N}} = 63.5 \) mm and \( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1, respectively

For all the cases, instead of modeling the ladle, the volumetric flow rate of liquid steel was used as the input for the simulations.

Results and Discussion

For each simulation case, results for the velocity field, the relative air entrainment rate variation over time during the pouring event, and the final (total) relative air entrainment volume are presented. To calculate the final relative entrained air volume, \( V_{\text{a}} /V_{\text{s}} \), first, the volume of entrained air, \( V_{\text{a}} \), is determined by integrating the volumetric air entrainment rate over time, and then, this value is divided by the volume of liquid steel poured, \( V_{\text{s}} \). In the results that follow, first simulation results for filling the casting without a gating system (direct pour) are given, and the effects of nozzle diameter, flow rate constancy, nozzle extension, and filling time are shown. In the second results subsection, the effects of gating system components on air entrainment are presented for castings with bottom-gated filling system and compared with the direct pour case. Finally, the results for the side gating configuration are given where several gating ratios are compared with each other and with the direct poured case.

Direct Pour: No Gating

The velocity and local air entrainment contours are shown in Figure 6a at \( t = 8 \) s for the “base” case using direct pour (no gating), and nozzle diameter and fill time of \( d_{\text{N}} = 63.5 \) mm and \( t_{\text{fill}} = 25.4 \) s, respectively. Air bubbles are entrained at the periphery of the liquid jet where it impinges on the surface of the liquid pool (Figure 6b). In Figure 6c, the total relative air entrainment (local air entrainment summed over the system volume) over time is shown. The largest spike in the relative air entrainment plot corresponds to the liquid jet initial impingement on the pool surface. After the initial impact, starting from around 3 s into the filling, the relative air entrainment decreases significantly to a smaller value around 0.1 and it reduces gradually until the end of filling.
Figure 6

Plots of (a) velocity field and (b) local volumetric air entrainment rate, \( Q_{\text{a}} \), at \( t = 8 \) s, and (c) total relative air entrainment rate as a function of time for the simulation case with no gating (direct pour), nozzle diameter of \( d_{\text{N}} = 63.5 \) mm, and filling time of \( t_{\text{fill}} = 25.4 \) s (\( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1) with no nozzle extension.

The effect of nozzle diameter on the air entrainment is shown in Figure 7. Using a constant filling time of \( t_{\text{fill}} = 25.4 \) s, six nozzle diameters were simulated. No nozzle extension was used for this case. For a constant filling time (constant volumetric liquid flow rate), increasing the nozzle diameter reduces the liquid jet velocity at the nozzle exit (Figure 7b), and the liquid jet diameter at impact, and hence the air entrainment. Increasing the nozzle diameter by a factor of three reduces the liquid jet velocity at nozzle exit \( u_{\text{N}} \) by a factor of more than 6 (Figure 7b), while looking at the velocity contours (Figure 7a), the liquid jet at impact, \( u_{\text{j}} \), reduces by a factor of approximately 2. In addition, though the liquid steel pouring stream contracts more for cases with larger nozzle diameters, the liquid jet diameter at impact is still larger for large nozzle diameters (Figure 7a). The reduction in liquid jet velocity at impact, \( u_{\text{j}} \), along with the increase in liquid jet diameter at impact, \( d_{\text{j}} \), significantly reduces the relative air entrainment rate (Figure 2). As shown in Figure 7c, the initial spike is largest for the smallest nozzle diameter configuration \( d_{\text{N}} = 25.4 \) mm, and this case entrains significantly more air than other configurations throughout the pouring event. From Figure 7d, it is shown that for a constant filling time \( t_{\text{fill}} = 25.4 \) s increasing the nozzle diameter by a factor of three reduces the relative entrained air volume approximately 5 times, and more modest increases in \( d_{\text{N}} \) also result in substantially less air entrainment.
Figure 7

Plots showing the effect of the nozzle diameter on variables and the relative air entrainment rate for the no gating case and a constant fill time of \( t_{\text{fill}} = 25.4 \) s (\( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1): (a) the velocity contours at \( t = 8 \) s for four of the nozzles, (b) liquid velocities at the nozzle exit, (c) relative volumetric air entrainment rates for the nozzles in (a), and (d) total relative entrained air volume at the end of filling.

In bottom pour ladles, as the ladle is emptied, the liquid jet velocity at nozzle exit reduces with time; hence, the liquid jet velocity at impact reduces as filling proceeds. To study the effect of this variable flow rate on air entrainment, constant and variable flow rate configurations were simulated for each nozzle diameter. In Figure 8, the relative air entrainment volume is compared for constant and variable flow rates for the nozzle diameters used in Figure 7 where the filling time of \( t_{\text{fill}} = 25.4 \) s. The volumetric steel flow rate used for all nozzle diameters and velocity profiles for three of the nozzle diameters are shown in Figure 8a, b, respectively. Note that since the filling time for variable flow rate is the same as the constant flow rate configuration (\( t_{\text{fill}} = 25.4 \) s), the velocity plots intersect halfway during filling (\( t_{\text{fill}} = 12.7 \) s) in Figure 8b. The velocity (left) and relative air entrainment rate development during filling (right) are compared for the smallest nozzle diameter configuration in Figure 8c. The velocity contours are shown at two times, at 4 and 20 s into the filling. The larger initial velocity at 4 s in Figure 8c is clear for the variable flow rate case. For the constant volumetric flow rate, the liquid velocity at nozzle exit remains constant during filling, while for the variable case, the velocity decreases as filling proceeds. From Figure 8c, initially the relative air entrainment rate is significantly larger for the variable flow rate. At the beginning due to high liquid jet velocity for the variable flow rate, significant free surface turbulence occurs once the liquid steel impinges on the bottom of the mold box. Additionally, the breaking waves resulted from splashing, contributing to more air entrainment. However, after \( t_{\text{fill}} = 12.7 \) s, as the liquid velocity at the nozzle exit reduces below \( u_{\text{N}} = 3.95 \) m s−1 (liquid velocity at the nozzle exit for the constant pouring rate), the air entrainment drops below the air entrainment rate of the constant pouring rate case. Comparison between the constant and variable flow rates (Figure 8d) indicates that increasing the nozzle diameter reduces the difference of the relative entrained air volume between constant and variable flow rates. For smaller nozzle diameters, air entrainment is significantly larger due to the much larger velocity at the beginning of filling. However, as the nozzle diameter is increased, the velocity difference between the constant and variable flow rate configurations is less (Figure 8b), and the air entrainment difference between the constant and variable flow rates becomes negligible (Figure 8d).
Figure 8

Plots showing the effect of variable and constant flow rates on the relative air entrainment rates and key variables for the no gating case using a fill time of \( t_{\text{fill}} = 25.4 \) s (\( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1): (a) variable and constant volumetric flow rates used as a function of time for all nozzle diameters, (b) nozzle velocities for variable and constant flow rates for three of the nozzle diameters simulated versus time, and (c) the velocity fields at times \( t = 4 \) s and \( t = 20 \) s during filling, and the relative volumetric air entrainment rates for the nozzle diameter \( d_{\text{N}} = 25.4 \) mm case, and (d) the final relative entrained air volume for cases having with and constant flow rates for all nozzle diameters simulated.

The effect of using a nozzle extension (or shroud) on air entrainment is shown in Figure 9 for two nozzle diameters and a constant filling time of \( t_{\text{fill}} = 25.4 \) s. The addition of a nozzle extension applies friction to the liquid flow, which reduces the liquid steel velocity exiting the nozzle extension, and therefore reduces air entrainment. Also, once the nozzle extension becomes submerged inside the liquid steel pool, the relative air entrainment rate drops to a small value as shown in Figure 9b. For a submerged nozzle extension, the nozzle extension exit is located at the impact location, and the liquid steel has virtually no interaction with the surrounding air and no air pocket is trapped in the periphery of the liquid jet. Therefore, once the nozzle extension becomes submerged in the liquid, air entrainment decreases significantly.
Figure 9

Plots showing the effect of nozzle extension and nozzle diameter on the velocity field and relative air entrainment for the case with no gating system with nozzle diameters of 25.4 and 63.5 mm and a fill time of \( t_{\text{fill}} = 25.4 \) s (\( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1): (a) the velocity fields at \( t = 8 \) s, (b) relative volumetric air entrainment rate during pouring, and (c) final relative entrained air volume for the two nozzle sizes and three extension conditions.

The effect of fill time (or alternatively the flow rate) on the air entrainment, and oxide inclusion formation, has been debated among foundry engineers. The effect of fill time on the air entrainment is demonstrated by the results in Figure 10 for the no nozzle extension and nozzle diameter of \( d_{\text{N}} = 25.4 \) mm case. Longer filling time implies longer interaction of liquid metal with the air, which implies that more air entrainment occurs. For configurations with short filling time, the liquid steel fills the mold cavity fast, and this results in the free falling height (distance from the nozzle exit to the impact location) and liquid jet velocity at impact reducing over a short period of time as shown in Figure 10a for the four cases at 8 s into the filling process. This is the primary reason that less air is entrained for faster fill times. As indicated in Figure 10b, the relative air entrainment drops drastically for the shortest filling time (high volumetric flow rate), while the reduction in air entrainment is a slower process for the longest filling time case. Clearly, for a pouring process using a constant head height and nozzle diameter, reducing the filling time reduces the air entrainment.
Figure 10

Plots showing the effect of filling time (pouring rate) on the velocity field and relative air entrainment for the no gating system case with nozzle diameter \( d_{\text{N}} = 63.5 \) mm and filling times of 12.7, 16.9, 25.4 and 50.8 s: (a) the velocity fields at \( t = 8 \) s from the start of filling, (b) relative volumetric air entrainment rate during filling, and (c) final relative entrained air volume for each fill time.

Bottom-Gated Castings

The effect of using various gating system components on air entrainment in filling bottom-gated castings was simulated. The effect of pouring cup/basin design and the use of a stopper for the offset basin on the air entrainment are shown in Figure 11. In Figure 11a, the velocity contours at 8 s from start of filling are shown for the five cases considered using a nozzle diameter of \( d_{\text{N}} = 63.5 \) mm and a flow rate of \( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1. The final air entrainment results show that the casting with direct pour (no gating) entrains the least amount of air (Figure 11c). This might be surprising at first glance. However, it must be pointed out that the main reason for this outcome is the difference in falling heights between the direct pour configuration and the bottom-gated configurations (see Figure 3). The initial falling height for the direct pour case is 355.6 mm (Figure 3a), compared to 711.2 mm for bottom-gated case (Figure 3b), and this has resulted in lower air entrainment for the direct pour configuration. As mentioned earlier, reducing the falling height reduces the liquid jet velocity at impact, and therefore the air entrainment.
Figure 11

Plots showing the effect of the pouring cup, basin, vortex sprue and stopper conditions on the velocity field and relative air entrainment rate for nozzle diameter \( d_{\text{N}} = 63.5 \) mm and flow rate of \( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1: (a) the velocity fields at \( t = 8 \) s from start of filling, (b) relative volumetric air entrainment rate during filling, and (c) final relative entrained air volume for the five simulation cases.

The velocity and air entrainment rates for the conical cup and offset basin are also compared in Figure 11. The gating system with conical cup entrains more air than the one with offset basin. Early during filling the offset basin configuration entrains more air, but the conical pouring cup case entrains larger amounts throughout most of the later pouring process. In an offset basin, the liquid steel first impinges on the bottom of the basin surface, and later it plunges to the base of the sprue. For the pouring basin, the flow velocity at the sprue entrance is greatly reduced from the conical pouring cup case due to the presence of the cylindrical step feature (or wier), which results in a smaller liquid velocity at the sprue base. For a conical cup, the liquid steel exiting the ladle nozzle directly impinges on the sprue base and results in significantly larger liquid jet velocities at the sprue base. Also for the offset basin configuration, once the liquid steel passes the circular step/weir of the offset basin occurs inside the sprue (large spike in Figure 11b). However, after approximately \( t = 4 \) s, air entrainment significantly reduces to a small value and continues to drop as the filling proceeds. Overall, the offset basin shows improvement over the conical cup. In addition, the effect of stopper for an offset basin was examined. Once the stopper is removed, the liquid steel fills the sprue in a short period of time, reducing the interaction with surrounding air; therefore, the stopper reduces the effect of initial liquid metal impingement inside the sprue. The use of stopper slightly reduces the air entrainment as seen by the results in Figure 11c.

Another case of interest shown in Figure 11 is the effect on air entrainment for the offset basin with vortex sprue gating system. Creation of a vortex flow inside the sprue requires careful design of the offset basin. The design shown in Figure 5a is patterned after Reference 1. The flow from the ladle impinges on the basin surface, and as it enters the offset part of the basin, it follows the circular path of the design (see rightmost image in Figure 5a). The offset part of the basin is designed to generate a tangential flow to the sprue wall (see Figure 12a top view). The liquid steel swirls down the sprue and does not directly impinge on the bottom of sprue as shown in Figure 12a. In this flow process, the sprue walls generate additional frictional losses in the liquid steel flow and markedly reduce the velocity. Note that this vortex flow reduces the initial spike of the air entrainment to a great extent for the offset basin. These results appear to contradict an often held belief among foundry engineers that such vortex flows draw air into the gating system and increase oxide inclusions.1
Figure 12

Velocity field results for the simulation configuration using the vortex sprue and bottom filled gating system with nozzle diameter \( d_{\text{N}} = 63.5 \) mm and flow rate of \( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1: (a) two views of the flow field development inside the offset basin at early fill times, and (b) views of the velocity flow field at 8 s from the start of filling for the entire model domain (3D view) and on the mid-plane of the model (side view).

The relative effectiveness of the vortex well and vortex gate to reduce air entrainment is given by the results in Figure 13. In the case of a vortex well, after the liquid steel impinges on the well surface and fills the sprue well, the free surface turbulence at the sprue base significantly reduces. Comparison between the velocity contours in Figure 13a indicates that for configurations with the vortex well, most of the sprue fills up early during the filling, which reduces the falling height from the ladle exit into the sprue, and consequently reduces air entrainment. The aim of the vortex gate shown in Figure 13 is to reduce the liquid metal velocity at the ingate and to reduce the free surface turbulence at the entrance to casting cavity, and therefore air entrainment. In a vortex gate, the flow from the runner tangentially enters the gate, and swirls inside the gate and fills the gate and the casting quiescently. The combination of vortex well and vortex gate is shown to reduce the air entrainment to almost half of the conical configuration in Figure 13c. The simulation results agree with Campbell’s findings; however, as stated by Campbell, further research is required to confirm the advantages of these vortex generating gating components.1
Figure 13

Plots showing the effect of using vortex components in gating systems on the velocity field and relative air entrainment rate for nozzle diameter \( d_{\text{N}} = 63.5 \) mm and flow rate of \( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1: (a) the velocity contours for the five simulation cases at 8 s from the start of filling, (b) relative volumetric air entrainment rate for the cases during filling, and (c) the final relative entrained air volume at the end of filling.

Side-Gated Castings

Non-pressurized and pressurized gating systems have long been used for eliminating or reducing oxide inclusions in metal castings. In pressurized gating systems, the minimum cross-sectional area (choke) of the gating system is located at the ingate. Therefore, these gating systems are beneficial in filling the gating system fast, hence reducing the liquid metal interaction with the surrounding air. Unfortunately, the ingate velocity can be small in these filling systems, which increases the possibility of free surface turbulence at the casting entrance. Conversely for a non-pressurized gating system, the choke for the system is located at the sprue base. The liquid metal velocity in non-pressurized gating systems is usually low, and the liquid metal has more time to interact with air. A newer gating system design approach proposed by Campbell has been widely accepted in metal castings, the “Naturally Pressurized” gating system. In this gating system, local chokes are avoided and the liquid metal is pressurized throughout the entire filling system. As a result, the cross-sectional areas of the filling system components are small enough such that surface turbulence is avoided to a great extent.

Air entrainment results for side-gated casting filling systems with several gating ratios are shown in Figure 14. The gating ratios produce non-pressurized, pressurized, and naturally pressurized filling systems, and their results are compared to the direct pour no gating configuration. Similar to bottom-gated castings, air entrainment comparisons are made for a nozzle diameter and volumetric flow rate of \( d_{\text{N}} = 63.5 \) mm and \( Q_{\text{s}} = 2 \) × 10−3 m3 s−1, respectively. In Figure 14a, filling velocity fields are shown at \( t = 8 \) s from the start of pouring. Note that at this time the sprue is almost filled with liquid steel for the naturally pressurized gating system configuration using the small sprue, while for the non-pressurized gating system, the liquid level is close to the sprue base. As a result, the liquid jet velocity at impact, and the air entrainment, is larger for non-pressurized gating system. The results of relative air entrainment in Figure 14b indicate the pressurized and naturally pressurized gating systems show a secondary peak. This secondary peak corresponds to the time when the liquid jet enters the casting. A smaller ingate produces a larger liquid velocity entering the casting, and a significant amount of air is entrained when the liquid steel emanates from the smaller ingates. However, after the liquid level inside the casting reaches the top of the ingate, at approximately \( t = 4 \) s, the air entrainment decreases drastically. As filling proceeds, air entrainment continues dropping until it reaches zero at the end of filling. In summary, the faster the liquid steel fills the gating, the less interaction occurs between the liquid metal and surrounding air, resulting in less air entrainment. Clearly, these results show that naturally pressurized gating systems are beneficial in reducing the total amount of entrained air. Considering the gating dimensions in Figure 3, it is important for the reader to appreciate that even though the initial free falling height of the naturally pressurized gating system with small sprue (\( h = 584.2 \) mm) is significantly larger than that of the direct pour case (\( h = 355.6 \) mm), the gating with the greater falling height entrains less air.
Figure 14

Plots showing the effect of the gating system ratio on the velocity field and relative air entrainment rate for a nozzle diameter of \( d_{\text{N}} = 63.5 \) mm and flow rate of \( Q_{\text{s}} = 2 \times 10^{ - 3} \) m3 s−1: (a) the velocity contours for the five different cases at 8 s from the start of filling, (b) relative volumetric air entrainment rate for the cases during filling, and (c) the final relative entrained air volume results at the end of filling.

Conclusion

Casting manufacturing processes have many variables which can be difficult to control and can lead to defects in cast components. If foundry engineers can design reliable processes for filling castings, they can reduce air entrainment and the oxide inclusion defects it causes, and eliminate a major source of quality variability and rework. Here air entrainment modeling is applied to demonstrate the effect of pouring conditions and gating system design on the air entrainment during mold filling. The results demonstrate that quantitative comparisons between filling methods, based on physical mechanisms, are possible. As mentioned several times here, there are many beliefs, opinions, and anecdotal experiences in the foundry industry surrounding the best ways to fill castings, and now these can be quantitatively tested.

Modeling results presented here show that reducing the total head height and adding a nozzle extension to the end of the nozzle reduces the air entrainment during mold filling with a bottom pour ladle. In addition, for a given filling time, increasing the nozzle diameter reduces the liquid jet velocity, which significantly reduces the entrained air volume. Using an offset basin with a stopper and gating components which create vortex flows tangential to gating system walls markedly reduce the entrained air volume. Results also show that producing clean castings requires pressurizing the gating system. This study produced promising results and demonstrates the possibilities of future applications of the present air entrainment model. With further experimental validation, the air entrainment model will be a powerful tool for the evaluation of filling and gating systems. Ongoing and future model development work will link the present air entrainment model to an inclusion generation and transport model, where the final oxide inclusion size and location can be predicted. Real-world casting trials should be conducted to further validate the model predictions.

References

  1. 1.
    J. Campbell, Complete Casting Handbook: Metal Casting Processes, Metallurgy, Techniques and Design, 2nd edn. (Butterworth-Heinemann, Oxford, 2015)Google Scholar
  2. 2.
    A.J. Melendez, K.D. Carlson, C. Beckermann, Modelling of reoxidation inclusion formation in steel casting. Int. J. Cast Met. Res. 23, 278–288 (2010)CrossRefGoogle Scholar
  3. 3.
    D.R. Askeland, P.K. Trojan, R.A. Flinn, Dross Formation reactions in the system Mg–Si–O in ductile iron. AFS Trans. 80, 349–358 (1972)Google Scholar
  4. 4.
    A.K. Biń, Gas entrainment by plunging liquid jets. Chem. Eng. Sci. 48, 3585–3630 (1993)CrossRefGoogle Scholar
  5. 5.
    D.A. Ervine, E. McKeough, E.M. Elsawy, Effect of turbulence intensity on the rate of air entrainment by plunging water jets. ICE Proc. 69, 425–445 (1980)Google Scholar
  6. 6.
    K.J. Sene, Air entrainment by plunging jets. Chem. Eng. Sci. 43, 2615–2623 (1988)CrossRefGoogle Scholar
  7. 7.
    T. Brattberg, H. Chanson, Air entrapment and air bubble dispersion at two-dimensional plunging water jets. Chem. Eng. Sci. 24, 4113–4127 (1998)CrossRefGoogle Scholar
  8. 8.
    J. Campbell, Entrainment defects. Mater. Sci. Technol. 22, 127–145 (2006)CrossRefGoogle Scholar
  9. 9.
    A. Kheirabi, A. Baghani, A. Bahmani, M. Tamizifar, P. Davami, M. Ostad Shabani, A. Mazaher, Understanding the occurrence of the surface turbulence in a nonpressurized bottom gating system: numerical simulation of the melt flow pattern. Proc. IMechE Part L J. Mater. Des. Appl. 232, 230–241 (2018)Google Scholar
  10. 10.
    N.W. Lai, W.D. Griffiths, and J. Campbell, Modelling of the potential for oxide film entrainment in light metal alloy castings. in MCWASP X, D.M.Stefanescu, J.A. Warren, M.R. Jolly, and M.J.M. Krane, eds., TMS, Warrendale, PA, 415–422 (2003)Google Scholar
  11. 11.
    F.-Y. Hsu, M.R. Joly, J. Campbell, Vortex-gate design for gravity casting. Int. J. Cast Met. Res. 19, 38–44 (2006)CrossRefGoogle Scholar
  12. 12.
    X. Yang, X. Huang, X. Dai, J. Campbell, J. Tatler, Numerical modelling of entrainment of oxide film defects in filling of aluminium alloy castings. Int. J. Cast Met. Res. 17, 321–331 (2004)CrossRefGoogle Scholar
  13. 13.
    M.C. Latona, H.W. Kwon, J.F. Wallace, J.D. Voss, Factors influencing dross formation in ductile iron castings. AFS Trans. 92, 881–906 (1984)Google Scholar
  14. 14.
    C. Reilly, N.R. Green, M.R. Jolly, J.C. Gebelin, The modelling of oxide film entrainment in casting systems using computational modelling. Appl. Math. Model. 37, 8451–8466 (2013)CrossRefGoogle Scholar
  15. 15.
    Y. Yue, Modelling of the effects of entrainment defects on mechanical properties in Al–Si–Mg alloy castings, Ph.D. diss. (University of Birmingham, Birmingham, 2014)Google Scholar
  16. 16.
    C. Wanstall, J. Griffin, C.E. Bates, Water modeling of steel pouring practices. In: Proceedings of the 47th SFSA Technical and Operating Conference, Chicago, IL, Paper No. 1.1 (1993)Google Scholar
  17. 17.
    C.E. Bates and J. Griffin, Clean cast steel technology. Research report no. 106, Steel Founders’ Society of America (SFSA), Crystal Lake, IL, USA (1994)Google Scholar
  18. 18.
    S. Kuyucak, Sponsored Research: clean steel casting production—water modeling studies of bottom-pouring ladle operations. AFS Trans. 114, 811–818 (2006)Google Scholar
  19. 19.
    P.K. Trojan, P.J. Guichelaar, R.A. Flinn, An investigation of entrapment of dross and inclusions using transparent whirl gate models. AFS Trans. 74, 462–469 (1966)Google Scholar
  20. 20.
    J. Ma, A.A. Oberai, D.A. Drew, R.T. Lahey, M. Hyman, A comprehensive sub-grid air entrainment model for RANS modeling of free-surface bubbly flows. J. Comput. Multiph. Flow 3, 41–56 (2011)CrossRefGoogle Scholar
  21. 21.
    S.H. Majidi, C. Beckermann, Modelling of air entrainment during pouring of metal castings. Int. J. Cast Met. Res. 30, 301–315 (2017)CrossRefGoogle Scholar
  22. 22.
    S.H. Majidi, J. Griffin, C. Beckermann, Simulation of air entrainment during mold filling: comparison with water modeling experiments. Metall. Mater. Trans. B 49B, 2599–2610 (2018)CrossRefGoogle Scholar
  23. 23.
    MAGMAsoft, MAGMA Gmbh, Kackerstrasse 11, 52072 Aachen, GermanyGoogle Scholar

Copyright information

© American Foundry Society 2018

Authors and Affiliations

  1. 1.Department of Mechanical EngineeringUniversity of IowaIowa CityUSA

Personalised recommendations