Article

Numerical Modeling of Transport Phenomena in the

Horizontal Single Belt Casting (HSBC) Process for
the Production of AA6111 Aluminum Alloy Strip
Usman Niaz 1, Mihaiela Minea Isac 1 and Roderick I.L. Guthrie 2,*
1 Mining & Materials Engineering Department, McGill University, Montreal, QC H3A 0C5, Canada;
[email protected] (U.N.); [email protected] (M.M.I.)
2 Roderick I.L. Guthrie, McGill Metals Processing Centre, McGill University,

Montreal, QC H3A 0C5, Canada

* Correspondence: [email protected]

Received: 31 March 2020; Accepted: 27 April 2020; Published: 30 April 2020

Abstract: In this research study, numerical modelling and experimental casting of AA6111 strips,
250 mm wide, 6 mm thick, was conducted. The velocity of the molten AA6111 alloy at the nozzle
slot outlet was raised to 2 m/s, whilst the belt speed was kept at 0.3 m/s. The numerical model
demonstrates considerable turbulence/fluctuations in the flow of the molten AA6111 alloy in the
HSBC process, rendering its free surface highly non-uniform and uneven. These discontinuites in
the flow resulted from the sudden impact of molten metal onto the inclined refractory plane, and
then onto the slowly moving belt. However, it has been determined that these surface variations are
rapidly damped, and as such are not detrimental to final strip surface quality. Any surface
perturbations remaining can be eliminated via hot plastic deformation. The experimental findings
are in accordance with the model predictions. Furthermore, at high metal heads inside the delivery
launder, the molten metal was observed to be flowing inwards towards the center of the strip,
thereby filling the centre depression region, formed otherwise. The model predictions were
validated against experimental findings. A surface roughness and microstructural analysis was also
conducted to determine the surface and bulk quality of the as-cast strip.

Keywords: Horizontal single belt casting process (HSBC); computational fluid dynamics (CFD);
double impingement feeding system

1. Introduction
Near net shape casting (NNSC) processes can be regarded as an ideal method for metal sheet
production. Apart from their low energy requirements, lower capital and operating costs, and smaller
plant footprints than those associated with slab casters (Fe), and DC casters (Al), they also have
promising metallurgical characteristics associated with much higher cooling rates possible [1–5].
However, they can also have drawbacks, regarding surface quality, which conventional plants can
overcome, by scarfing and multi-pass rolling, to the final quality strip [6]. An advantage of NNSC
processes is that a homogenous microstructure with almost zero macrosegregation and fine grain
sized products can be attained, provided cooling rate conditions are met, as the cast product’s
dimensions approach the desired sheet specifications [4].
Therefore, fewer hot/cold reduction passes are required as opposed to conventional methods
i.e., slab, or thin slab, casting (TSC) of steels, or direct chill (DC) casting for aluminum alloys. This
undoubtedly brings down the overall cost of the operation. On the other hand, the high surface area-
to-cast thickness is much higher (~100×), making surface oxidation more problematic, as well as

Processes 2020, 8, 529; doi:10.3390/pr8050529 www.mdpi.com/journal/processes

Processes 2020, 8, 529 2 of 19

surface quality and surface dimensions [3]. The three most commonly used NNSC processes,
commercialized to date, are the twin roll casting (TRC), the earlier twin belt caster, and, most recently,
the single belt caster [6]. Historically, the commercialization of the Hazelett twin belt caster in the
1930s by Clarence Hazelett, was an early breakthrough for NNSC. It has since been used around the
world, ever since, for casting non-ferrous alloys, mainly aluminum, copper, and zinc, down to cast
thicknesses of 13–21 mm [7]. All attempts with steel have failed.
As the name suggests, TRC utilizes two contra-rotating water-cooled rolls, onto which the
molten metal is fed and cast as a thin strip. Invented by Henry Bessemer back in 1857 [8], and despite
extensive research and development activities that followed, the world’s first strip caster for
aluminum was only commercialized in 1954, approximately 100 years after Bessemer’s initial idea.
Currently, FATA Hunter and Novelis PAE (previously SCAL-Pechiney) are leading companies
manufacturing TRC equipment for aluminum strip production [9]. The cast strip thickness is almost
6 mm. The first TRC strip caster commercialized for steel was by Nippon Steel and Mitsubishi Heavy
Industries Corporation in 2000. The caster could produce stainless steel strips, 2–5 mm thick, 0.76–1.5
m wide, in NSC’s Hikari Works in Southern Japan, but was not a commercial success and was
abandoned. Later, in 2002, NUCOR started the production of a low carbon steel strip, 1.7–1.9 mm
thick via a TRC process, CASTRIP, and this continues to operate, commercially [6]. However, TRC,
though a proven aluminum/steel strip manufacturing process, suffers from low annual production
of steel vs. that of a conventional slab casting process, i.e., 400,000 tpy and 2,000,000 tpy, respectively
[10].

An introduction to Horizontal Single Belt Casting (HSBC) Process

The horizontal single belt casting (HSBC) process, independently conceived by Herbertson,
Guthrie, Reichelt, and Schwerdtfeger, et al., emerged from a joint effort of BHP (Australia), McGill
University (Canada), and the Hazelett Strip Casting Corporation (USA) as discussed above. This
process is like the Pilkington float glass technology, in which the molten glass is continuously poured
over a bath of molten tin where it solidifies into a continuous glass sheet [5].
The HSBC process avoids the multi-hot rolling-deformation steps required for direct chill (DC)
cast material, together with its intermediate annealing steps, while producing the same desired final
thicknesses of sheet products. As such, a large amount of energy can be saved. This is essentially true
for all NNSC processes discussed above. The HSBC process features a compact design and provides
for the better economic production of both ferrous and non-ferrous metallic products. The production
of advanced high-strength steel (AHSS) strips, 16 mm thick, via the HSBC process at Salzgitter Group
Steelworks, Germany, is an example that makes use of the advantages that the HSBC process offers,
vs. conventional strip manufacturing processes. To date various ferrous/nonferrous alloys strips, up
to a thickness of 16 mm, were successfully produced. For further reading, please refer to the available
literature [11,12].
Simplistically, the HSBC process involves feeding molten metal on to an intensively cooled,
moving belt, which acts as the mold. Depending on the metal head in the launder, the velocity of the
molten metal issuing from the slot nozzle can be easily adjusted. However, the force of gravity is also
equally responsible in further accelerating the molten metal before it contacts the moving belt, on
which it solidifies [13–15]. The material produced via the HSBC process can then be processed
downstream, by hot rolling, followed by cold rolling, as shown in Figure 1.
Processes 2020, 8, 529 3 of 19

Figure 1. (a) A photograph of the HSBC pilot-scale system, and (b) A schematic of the HSBC pilot-
scale machine located at MetSim Inc,’s High Temperature, Melting and Casting Laboratory, Quebec,
Canada.

2. An Overview of Previous Studies Conducted on HSBC Metal Feeding Systems

Many variants of the feeding system have been investigated by researchers at McGill Metals
Processing Centre, to date. They can be classified into two types of the metal delivery systems; either
single-impingement or multi-impingement, based on how many times the molten metal encounters
obstacles before reaching the cooling substrate, as shown in Figure 2 [2,3]. In a single impingement
feeding system, the molten metal is abruptly stopped by the horizontally moving belt, since there is
no intermediate obstacle in its way, which could decrease its kinetic energy. As a result, the molten
metal tends to penetrate back into the quadruple region, i.e., the region where melt, refractory, air,
and belt coexist, as determined by Sa Ge et al. [3], for the casting of plain carbon steel, employing a
single impingement feeding system (Figure 3a) [3]. If the backflow is too excessive, it may lead to
skull formation, thereby curtailing further casting as determined experimentally.

Figure 2. Different feeding systems for HSBC process, (a) Double impingement, (b) single
impingement [1–4].

Additionally, it has been shown by Sa Ge [13], that a single impingement feeding system can be
used to produce Al-Mg-Sc-Zr alloy strips, without any backflow into the quadruple region (see
Figure 3b), as was found for steel casting. This could be due to the low density of Al-Mg-Sc-Zr alloy
and a higher contact angle, i.e., a lower wettability between liquid Al-Mg-Sc-Zr and the alumina
refractory [13].
Processes 2020, 8, 529 4 of 19

Figure 3. (a) Simulated flow of the molten plain carbon steel in single impingement feeding system,
showing back-flow into the quadruple region; molten phase is colored red, fully solidified shell is in
blue, and the partially solidified zone is in between [3], (b) Numerical simulations did not predict any
back flow of molten Al-Mg-Sc-Zr into the quadruple region; the molten phase is coloured red, air is
represented in blue. The numerical simulations were supported by experiments in these studies [13].

In both these studies, the molten metal flowing over the moving belt was considered as being
nearly iso-kinetic, a condition in which the velocities of the molten metal and belt approach each
other. Furthermore, the as-cast thickness of the produced strips was ~3 mm [3,5].
As explained above, excessive backflow of molten metal into the quadruple region is not desired.
This can be conveniently prevented by employing a double-impingement feeding system in the
HSBC process [14]. In a double impingement feeding system the molten metal dispensing from the
refractory nozzle slot first interacts with a 45° inclined refractory plane, followed by its second
interaction with the moving belt, on to which it begins to solidify [14]. In this way, the final impact
of the molten metal with the moving belt is not as rapid and abrupt, as it would be for a single-
impingement feeding system [4]. Furthermore, in a double-impingement feeding system, the flow of
the molten metal over the inclined refractory plane and the moving belt is entirely gravity-driven,
unlike the variant of the single-impingement feeding system reported by Sa Ge for the casting of Al-
Mg-Sc-Zr. For that, the flow of the molten metal is impeded by the refractory front wall, as shown in
Figure 3b [13]. During continuous operation, the refractory material may abrade and embed small
particles into the pool of molten metal. This could significantly decrease the bulk quality of the cast
strip.
In this research study, non-isokinetic feeding of molten metal over the moving belt has been
considered. This is significantly different from our group’s previous research studies, for which only
near iso-kinetic feeding was considered, as discussed above. Under non-isokinetic feeding
conditions, as in the present case (the belt/side dam speed is considerably slower, i.e., 0.3 m/s, as
compared to the molten metal velocity at the nozzle slot outlet, i.e., 2 m/s). The strip produced under
this condition has ~6 mm thickness, which is thicker as compared to the strips obtained under iso-
kinetic feeding (~3 mm). This allowed us to perform substantial hot deformation in order to produce
a 1 mm, or lower, thickness of strip. Hot deformation is necessary, as it transforms the cast dendritic
structure into fine equiaxed grains, leading to a far more uniform distribution of alloying elements
throughout the sheet material. Another purpose of hot reduction is to squash/weld any pores, if
present, in the cast strip, so as to improve its mechanical properties [15].

3. Objectives of the Present Research

The objective of this research study is to produce high-quality AA6111 aluminum alloy strips
250 mm wide, ~6 mm thick. Additionally, molten metal flow in a double-impingement feeding
system has been analyzed under non-isokinetic feeding. Under these experimental conditions, the
molten metal was observed to be flowing inwards, i.e., towards the center of the strip. This can
usefully eradicate center shrinkage cavity defects formed otherwise. In order to investigate this
Processes 2020, 8, 529 5 of 19

phenomenon, a three-dimensional mathematical model was developed using Fluent software (14.5,
Ansys, Inc., Canonsburg, PA, USA, 1970), and its accuracy was evaluated against experimental data.
Thanks to these numerical simulations, we now understand the complex interaction of the molten
metal with the inclined refractory plane and the moving belt that leads to the phenomenon of the
molten metal’s inward flow.
A horizontal single belt pilot caster installed at MetSim Inc., Montreal, QC, Canada, was used
for the casting experiments. However, several modifications were applied to the existing caster, as it
was not capable of producing 250 mm wide strips. These included the design of a new alumina
refractory nozzle slot (250 mm wide and 3 mm thick), increasing the cooling capability of the moving
steel belt, needed to completely solidify molten AA6111 strip before it exits the moving belt.
Additionally, the caster modifications included enlarging the strip guidance system, and, lastly, the
extension of the length and width of the run-out table, so as to accommodate the wider strip exiting
the caster.

4. Aluminum Alloy Grade, AA6111, Used in the Present Research

Keeping in mind the suitability of the HSBC process to cast both ferrous and non-ferrous alloys,
and knowing the usefulness of AA6111 aluminum alloy in the production of the lightweight body in
white (BIW) automotive structures, AA6111 was selected for HSBC strip production. AA6111 is an
alloy of Al-Mg-Si (Cu). It possesses higher mechanical strengths (i.e., 400 MPa), high formability, and
good corrosion resistance. The main mechanism behind AA6111′s increased strength is precipitation
hardening, along with the solid solution and work strengthening [16,17]. The alloying elements
present in AA6111 are as shown in Table 1.
The casting of AA6111 strips via the HSBC process is comparatively new, and is presently in its
development stages. It is, therefore, hoped that this paper will add fundamental knowledge to the
information already existing on this subject, and help industries realize the versatility of the HSBC
process to cast AA6111 aluminum alloy strips.

Table 1. Chemical composition of the AA6111 strip produced by the HSBC process, determined using
the spark OES technique.

Alloying Elements, wt%

Cu Fe Mg Mn Cr Si Ti Zn Al
0.5–0.9 >0.4 0.5–1.0 0.1–0.45 <0.1 0.6–1.1 <0.1 <0.15 Remaining

5. Details of the Experimental Procedure

AA6111 alloy was produced by first melting pure aluminum in a pre-heated induction furnace
under a protective argon atmosphere, followed by the addition of Al-Mg, and Al-Mn alloys, etc. Good
melt stirring was used, to ensure completely dissolved/mixing of the alloy additives into the pure
aluminum. The melt was then de-gassed and Ti-B grain refiner was added in a conventional way.
The AA6111 melt was then cast into the strips using the HSBC system. The step-by-step operation of
the HSBC pilot caster is presented below.
The process started with the production of AA6111 alloy using the 600 lb induction melting
furnace. Afterwards, the furnace was moved on rails to the casting station, where it was locked with
the liquid metal delivery system. This consisted of a refractory cylinder (regulated by a servo motor)
and a launder, as shown in Figure 1. Once the tight seal between the induction furnace and delivery
system was ensured, the motorised refractory cylinder was allowed to enter into the induction
furnace at a pre-selected speed, thereby displacing the molten metal into the launder. Once the
molten metal reached the desired level within the launder, the stopper bar blocking the nozzle outlet
was rapidly withdrawn, and liquid metal began to pour onto the belt. A weir and a dam were used
to help prevent Al2O3 oxide skin from entering the nozzle slot, as well as to help in minimizing
turbulence present within the flowing molten metal. The moving belt could also be equipped with
two rotating side dams. Their purpose was to contain the molten metal once it leaves the nozzle slot,
Processes 2020, 8, 529 6 of 19

and to give a straight/smooth edge before it enters the minimill for hot reduction. To avoid any
premature freezing, the entire delivery system and the refractory piston were preheated to
approximately 500–550 °C, using electrical resistive heating systems.
To evaluate the bulk, as well as the surface, quality of the cast strips, samples were sectioned
from the strip. All samples were polished and prepared for metallographic observations and
analyzed under Leica DM IRM optical and Hitachi TM3030 scanning electron microscope. The
surface roughness was measured using a 3D Nanovea profilometer. Results will be presented in later
paragraphs.

6. Details of the Model Setup

For CFD studies, the three-dimensional, transient state, turbulent fluid flow was modeled using
Fluent software (14.5, Ansys, Inc., Canonsburg, PA, USA, 1970). The code is based on the finite
volume method (FVM) [18]. The simulation domain chosen to carry out this research study had the
following dimensions, length (0.195 m), height (0.016 m), and width (0.05 m), as shown in Figure 4.
The semi-implicit method for pressure linked equations (SIMPLE) was used for coupling pressure
and velocity in the governing equations. More details can be found in the literature [19]. To improve
on the accuracy, the advection terms were discretized using a 2nd-order upwind scheme over the
entire simulation domain, whereas the diffusion term was approximated by the central differencing
scheme. To stabilize the interactive process, an under-relaxation factor of 0.7 for the velocity and 0.3
for the pressure, were used. The solution process was iterated until the residuals of governing
equations reduced to 1 × 10 . Different grids were tested until mesh-independent results were
achieved. Finally, 2,867,541 hexahedral cells were identified as being an accurate, but less
computationally intensive, exercise for obtaining the desired results. The molten metal was treated
as a Newtonian, incompressible fluid, and all the physical properties were assumed to be constant
(Tables 2 and 3).

Table 2. Physical properties of the Phases used in the model.

Operating Parameters/Assumptions Value

Slot Nozzle Dimension 3 × 250 mm
Inlet Velocity 2 m/s
Surface Tension of the Melt in Air 0.914 N/m
Copper Substrate Longitudinal speed 0.3 m/s
Turbulence Model SST k-ω
Contact Angle Between Melt and Alumina Refractory 135° [13]
Contact Angle Between Melt and Copper Substrate 105° [13]
Distance Between Stationary Inclined Refractory Plane and Moving Belt 0.4 mm

Table 3. Operating parameters and assumptions made in the model [2,13].

Property AA6111 Air

Kg
Density, ρ( 3 ) 2300 1.225
m
KJ
Specific Heat Capacity, Cp 1.177 1.006
KgK
W
Thermal Conductivity, K 104 0.0242
mK
Kg 5
Viscosity, µ 0.001338 1.75 × 10
ms
Kg
Molecular Weight, 26.98 28.97
kmol
J
Standard State Enthalpy 1.100493e -
kgmol
Reference Temperature (K) 298.15 298.15
Initial Temperature (K) 1000 300
Processes 2020, 8, 529 7 of 19

Figure 4. (a) Simulation domain containing hexahedral meshes (3D), (b) Mesh refinement at the
nozzle outlet, edges and the quadruple region, (c) A closer look on the hexahedral mesh system at the
inclined refractory plane. The dimensions are in meters

7. Results and Discussion

It was observed through numerical simulation studies that in the HSBC process, the molten
metal tends to laterally contract/shrink while exiting through the nozzle slot outlet, as shown in
Figure 5. Due to this lateral contraction, the weight of the molten metal around the edges increases
considerably as compared to the center. The heavier section accelerates downwards under the
influence of gravity, eventually reaching a high terminal speed before it strikes the moving belt. This
concept is further explained by plotting molten metal velocity, adjacent to the free molten metal/air
interface and the moving belt, against distance in the positive z-direction. As expected, the magnitude
of the velocity around the edges, for both cases, is high as compared to the center. These velocities
are computed 5 mm away from the quadruple region (down the ramp), as shown in Figure 6.
Processes 2020, 8, 529 8 of 19

Additionally, it has been observed that the velocity of the molten metal adjacent to the moving
belt is lower than the velocity at the free molten metal/air interface, as shown in Figure 6. This is due
to the friction offered by the moving belt, which tends to slow down the velocity of the molten metal
adjacent to it.

Figure 5. Molten AA6111 alloy flow in the HSBC process. (a,c) Simulated AA6111 flow showing
metal’s contraction after leaving the nozzle slot. (b,d) Actual Molten AA6111 flow in the HSBC
process.
Processes 2020, 8, 529 9 of 19

Figure 6. A plot of the velocity vs. distance (m) from the edge dam moving at 0.3 m/s.

Based on the above discussion, it can be concluded that the amount of molten metal, delivered
towards the edges, is considerably greater in comparison to the center, as shown in Figure 5a,b, owing
to the initial contraction/shrinkage of the molten metal while exiting through the slot nozzle outlet.
The net effect is an inward flow of the molten metal towards the center, as shown in Figure 7. This
inward flow can be very beneficial, as it eradicates the center shrinkage cavity defect, formed
otherwise, at low metal heads in the launder. This topic is further explained in the following
paragraphs.

Figure 7. The inward flow of the molten metal over the moving belt. (a) Simulated; (b) actual.

7.1. Method to Eradicate Center Cavity Defects

As discussed above, in the HSBC process, employing a double impingement feeding system, the
molten metal streams towards the center. Keeping in mind that at lower metal heads (<50 mm), the
quantity of the molten metal delivered onto the moving belt is also low, as per Equation (1). As
observed experimentally, under the rapid heat extraction rate to the moving belt (i.e., 500 K/sec) [3],
the molten metal passing over the moving belt, tends to solidify almost instantaneously. Since the
molten metal does not have enough time to level off before the completion of solidification, this
results in a strip with a thicker edge, and a comparatively thinner center.
The opposite is true for high metal heads inside the tundish (>50 mm). In this case, the velocity
of the molten metal exiting the refractory nozzle slot outlet is high enough to trigger a strong net
inward flow. Under these conditions, the molten metal will have enough time to fill the center empty
Processes 2020, 8, 529 10 of 19

region and to evenly spread throughout the thickness of the strip before the completion of
solidification. This helps to eliminate any center cavity defect and to achieve a uniform thickness of
the strip across its width.

V=C 2gh (1)

where V is the velocity, h is the molten metal head inside the tundish, and C is the coeffecient of
discharge.

7.2. Iso-Surfaces of Z-Component of Velocities

The velocity vector can be resolved into three components, i.e., x, y, and z in which the z-velocity
component represents a net inward flow of the molten metal. For these reasons, the isosurfaces of the
z-component of the velocities were evaluated and are represented in Figure 8.
Processes 2020, 8, 529 11 of 19

Figure 8. (a) Molten metal flow in the HSBC process. (b) Iso-velocity (0.2 m/s) along + z-direction. (c)
Iso-velocity (0.3 m/s) along + z-direction. (d) Iso-velocity (0.4 m/s) along + z-direction. (e) Iso-velocity
(0.5 m/s) along + z-direction. (f) Iso-velocity (0.6 m/s) along + z-direction.

As expected, the z-component of the velocity vector is maximal during the first instants of the
molten metal contacting the moving belt, owing to the fact that molten metal, while flowing over an
inclined plane, continuously accelerates under the force of gravity. Furthermore, the z-component of
velocity was observed to be high adjacent to the molten metal/air interface, and almost zero near the
moving belt. However, further downstream, over the belt, the z-component of velocity was observed
to be decreasing with distance. This is essentially true, as there is no driving force that could help the
molten metal to further accelerate over a horizontal moving belt.

7.3. Casting of AA6111 without the Use of Side Dams

The friction imposed by a slowly-moving belt reduces the velocity of the molten metal over the
moving belt. As a result, the molten metal tends to spread outwards, as shown in Figure 5c,d.
However, adjacent to the free interface, the flow of the molten metal is directed towards the center,
as is evident from the z-component of iso-velocities presented above. Depending on the relative
Processes 2020, 8, 529 12 of 19

magnitude of these two opposing effects, the molten metal can either flow towards the center, or
outwards.
However, by looking at Figure 7, it can be clearly seen that the molten metal is flowing towards
the center. This is very beneficial, as it eliminates the need for side dams to control the outward flow
of the molten metal. The successful casting of the AA6111 strip (Figure 7b), without the aid of side
dams (See Figure 7), experimentally, verifies the numerical modeling predictions.

7.4. Pressure Distribution of Molten Metal and the Generation of a Vortex Near the Triple Point
It is observed via the numerical simulations, that the inclined refractory plane has the tendency
to lessen, or moderate, the final impact of the molten metal on to the moving belt, by converting a
part of the molten metal’s kinetic energy into static pressure, as presented in Figure 9b.
Additionally, the numerical simulations predict a considerably higher absolute pressure (low
velocity) near the quadruple region (Figure 9b), due to sudden decrease in velocity of the molten
metal by a slow-moving belt. This results in a part of the impinging molten metal climbing upwards,
forming a vortex, as shown in Figure 10. Furthermore, the dynamic pressure, i.e., ρu is observed
to be highly adjacent to the molten melt/air interface. This result is due to the high velocity of the
molten metal near the free surface, as shown in Figure 9a.

Figure 9. (a) Dynamic Pressure, P = 0.5ρ∑u , where ρ is the density, u is the velocity. (b)
Absolute pressure (10,3281 Pa).
Processes 2020, 8, 529 13 of 19

Figure 10. The swirling motion of the molten metal at the triple point, after the second impingement.

7.5. The Temperature of the Molten Metal at the Melt/Air Interface

The hydraulic jump on the inclined refractory plane could substantially degrade the surface
quality of the cast product, owing to the generation of free surface waves/discontinuities [14]. As per
the numerical simulations, the temperature over of the melt/air interface is above the liquidus for a
considerable distance, as shown in Figures 11 and 12, even when considering perfect contact of
molten metal with the moving belt, which is held constant at 300 K, by the cooling water under the
belt. The experimental casting of the AA6111 alloy strip was in accord with numerical simulation
predictions, in which the melt/air interface with the belt was observed to be in a liquid state for
approximately the first meter along the moving belt. In this way, any molten metal surface
discontinuities had enough time to settle down by the damping forces generated, before final
solidification.

Figure 11. The predicted temperature distribution along the top and bottom faces of the strip along
the casting direction.
Processes 2020, 8, 529 14 of 19

Figure 12. Contour of molten metal temperature in contact with (a) the moving belt and (b) the molten
metal/air interface.

7.6. Characterization of the AA611 Alloy Cast Strip

A microstructural study has been carried out to analyze the quality of the cast strips. The samples
for optical microscopy were ground down to 1200 grid, and then electropolished/etched using 2%
perchloric acid (HClO4) in alcohol. Micro images were taken using a Leica DM IRM optical
microscope (Leica Microsystems, Concord, Ontario, Canada) and a TM3030 Scanning Electron
Microscope (Hitachi, Pleasanton, CA, US). The microstructure consisted of fine equiaxed grains
throughout the thickness of the strips (see Figure 13). The microstructure also contains porosities at
various locations within the cast strip, very similar to DC cast product. The average grain size of the
strip was found to be 85 µm, as shown in Figure 13.
Processes 2020, 8, 529 15 of 19

Figure 13. (a) Microstructure at the center of the strip (50×); (b) polarized micrograph (50×); (c)
intermetallics observed at the grain boundaries as well as inside the grains (500×).

The inter-metallics are also observed to be distributed inside the grains, as well as at the grain
boundaries of the cast microstructure, as shown in Figure 13c. X-ray micro-analyses revealed that
inter-metallics dispersed throughout the cast structure have the following stoichiometry:
Al17Cu2Mg3Si3, Al20Cu2Mg2.5Si5, whereas the elongated inter-metallics distributed at the grain
boundaries are in the category of Al17(CuMg)2(FeMn)Si2 or Al25(CuMg)4.5(FeMn)Si5 [16,17]. These
phases are clearly observed in the Figure 13c, at a higher magnification of 500×.

7.7. Energy Dispersive Spectroscopy (EDX) Analysis of the Cast AA6111 Alloy
Energy dispersive spectroscopy (EDX) analysis confirmed the presence of Al, Cu, Mg, and Si, in
AA6111 alloy, as shown in Figure 14. Furthermore, there is a negligible macro-segregation of alloying
elements in the cast strip, as shown in the chemical element’s maps, obtained at 15 KV excitation
voltage. This is caused by the rapid solidification of molten AA6111 alloy in the HSBC process, which
resulted in a homogenous microstructure with fine equiaxed grains. Additionally, the elemental
maps provide us with details of the chemical nature of the secondary phases. EDX analyses revealed
that inter-metallics dispersed throughout the cast structure are rich in Cu and Mg, whereas the
elongated inter-metallics found at the grain boundaries are concentrated in Si, Cu, and Mg.
Processes 2020, 8, 529 16 of 19

Figure 14. (a) Intermetallics distributed within grains and at grain boundaries (1000×). (b,c) Elemental
maps showing the chemical content of the intermetallics.

7.8. Surface Roughness Measurement

The surface waviness of the top/bottom sides of the strip was determined using a Nanovea 3D
profilometer (Nanovea, Irvine, California, United States). This technique works on the principle of
measuring the physical wavelength of light and directly relating it to a specific height. This ensures
accurate measurement of surface roughness/finish [20]. The scan length for all the measurements was
25 mm, whereas the scan speed was 0.1 mm/s. Ten random locations were selected for surface
roughness measurements. These locations were randomly selected from all over the strip. The surface
profiles are almost identical to one another. The upper surface roughness lies within the 125 µm (0.125
mm) range, as shown in Figure 15a, which is considerably smaller than the DC cast product
roughness, i.e., 0.45 mm for AA5182 aluminium alloy (Figure 16a,b) [21]. Additionally,
pinholes/blowholes were not detected on the surface of the cast strip (Figure 16c,d), unlike
continuously cast products, which possessed defects on their surfaces, and require surface grinding
prior to hot rolling [21].
Processes 2020, 8, 529 17 of 19

Figure 15. 3D profilometry results: (a) strip top surface roughness and (b) strip bottom surface
roughness.

Figure 16. Strand surface morphologies of direct chill of (a) AA3004 (Al-1%Mn-1%Mg), (b) AA5182
(Al-4.5%Mg) [21], (c) surface morphology of AA6111 (Al-1.1%Si-1%Mg-0.45%Mn), 250 mm wide strip
produced via the HSBC process. (d) Surface morphology of AA6111 (Al-1.1%Si-1%Mg-0.45%Mn),
covering the central 80 mm length of the strip produced via the HSBC process.

The strip bottom surface roughness was also measured, and lay in the 20 µm range, as shown in
Figure 15b. As evidenced by the results of the line scans, the bottom surface quality is much superior
to the top surface. This fact is credited to the fact that the molten metal is in direct contact with the
moving belt, and conforms to its shape, during the solidification process. On the other hand, the top
surface of the cast strip is exposed to the atmosphere and is affected by disturbances in the flows of
the molten metal.
Processes 2020, 8, 529 18 of 19

8. Conclusions
This present paper discusses the casting conditions and analyses results on AA6111 alloy strips,
250 mm wide and 6 mm thick, produced via the HSBC process. Computational fluid dynamics (CFD)
analyses were performed to examine molten AA6111 flows in the HSBC process so as to achieve
uniform thickness and a good surface finish of the cast strip.
The following conclusions were drawn, using the double impingement with no side dams
condition, for the liquid metal feeding system.
1. Under non-isokinetic feeding, the surface quality of the cast AA6111 alloy strip is not
compromised by the generation of surface disturbances.
2. The AA6111 alloy molten stream shrinks from its two edges. This build-up of the mass around
the corners eventually reaches a high terminal velocity. The net inward flow of the molten metal
resulted, which filled the center shrinkage depressions.
3. It was also determined that the inclined refractory plane of the double-impingement metal
feeding system has the tendency to lessen, or moderate, the final impact of molten metal with
the moving belt, as compared to a single-impingement metal feeding system, where the molten
metal encounters an abrupt change in direction by the moving belt.
4. The swirling flow of the molten metal in an immediate vicinity of the triple point is due to the
sudden vertical deceleration of the molten metal by the moving belt. However, the meniscus at
the triple point was still observed to be stable and non-fluctuating.
5. The temperature of the molten metal within the immediate vicinity of the free surface, along the
entire length of the simulation domain, remains above the liquidus temperature. Thus, any
molten metal discontinuities formed at the free molten metal/air interface had enough time to
be damped, prior to solidification.

Author Contributions: Methodology and writing: U.N.; supervision, technical advice, and access to
experimental and analytical equipment: M.M.I. and R.I.L.G. All authors have read and agreed to the published
version of the manuscript.

Funding: The authors would like to acknowledge the financial support received from the Natural Sciences and
Engineering Research Council of Canada (NSERC), and the International Advisory Board of supporting
companies of the McGill Metals Processing Centre (MMPC), as well as the technical support of MetSim Inc., in
carrying out this research. The authors would also like to acknowledge the support received from ANSYS Inc.
for software licensing to facilitate this research.

Conflicts of Interest: The authors declare no conflict of interest.

References
1. Celikin, M.; Li, D.; Calzado, L.; Isac, M.; Guthrie, R.I.L. Horizontal single belt strip casting (HSBC) of Al-
Mg-Sc-Zr alloys. Light Metals 2013, 2016, 1037–1040.
2. Xu, M.; Isac, M.; Guthrie, R.I.L. A numerical simulation of transport phenomena during the horizontal
single belt casting process using an inclined feeding system. Metall. Mater. Tran. B 2018, 49, 1003–1013.
3. Ge, S.; Chang, S.; Wang, T.; Calzado, L.E.; Isac, M.; Kozinski, J.; Guthrie, R.I.L. Mathematical modeling and
microstructure analysis of low carbon steel strips produced by horizontal single belt casting (HSBC). Metall.
Mater. Tran. B 2016, 47, 1893–1904.
4. Guthrie, R.I.L; Isac, M. Horizontal single belt casting of aluminum and steel. Steel Res. Int. 2014, 85, 1291–
1302.
5. Ge, S.; Isac, M.; Guthrie, R.I.L. Progress in strip casting technologies for steel; technical developments. ISIJ
Int. 2013, 53, 729–742.
6. Sanders, R.E. Continuous casting for aluminum sheet: A product perspective. JOM 2012, 64, 291–301.
7. Hazelett Process vs. Conventional Casting Processes. Available online: https://www.hazelett.com/process
(accessed on 27 April 2020).
8. Matsushita, T.; Nakayama, K.; Fukase, H.; Osada, S. Development and commercialization of twin roll strip
caster. Facilities 2009, 5, 6.
Processes 2020, 8, 529 19 of 19

9. Hamer, S.; Taraglio, B.; Romanowski, C. Continuous casting and rolling of aluminum: Analysis of
capacities, product ranges, and technology. Light Metal Age-Chic. 2002, 60, 6–17.
10. Isac, M.; Guthrie, R.I.L. The design of a new casting process: From fundamentals to practice. In Treatise on
Process Metallurgy; Elsevier: Montreal, Canada, 2012; pp. 555–583.
11. Spitzer, K.H.; Rüppel, F.; Viščorová, R.; Scholz, R.; Kroos, J.; Flaxa, V. Direct Strip Casting (DSC)-an option
for the production of new steel grades. Steel Res. Int. 2003, 74, 724–731.
12. Wans, J.; Geerkens, C.; Cremers, H.; Grethe, U.; Juchmann, P.; Schmidt-Jurgensen, R. Belt casting
technology experiences based on the worldwide first BCT caster. METEC 2nd ESTAD, 2017, 111–117, ISSN
1982-9345.
13. Ge, S.; Chattopadhyay, K.; Isac, M.; Guthrie, R.I.L. Mathematical modeling of transport phenomena in
Horizontal Single Belt Casting (HSBC). In Proceedings of the 5th International Congress on the Science and
Technology of Steelmaking, Dresden, Germany, 1–3 October 2012.
14. Niaz, U.; Isac, M.; Guthrie, R.I.L. Numerical modelling and experimental casting of 17% Mn–4% Al–3% Si–
0.45% C wt-% TWIP steel via the horizontal single belt casting (HSBC) process. Ironmak. Steelmak. 2020, 10
1–14, doi:10.1080/03019233.2020.1731253.
15. Sychkov, A.B.; Zhigarev, M.A.; Perchatkin, A.V.; Mazanov, S.N.; Zenin, V.S. The transformation of defects
in continuous-cast semi-finished products into surface defects on rolled products. Metallurgist 2006, 50, 83–
90.
16. Li, D.; Shabestari, S.G.; Isac, M.; Guthrie, R.I.L. Studies in the Casting AA6111 Strip on a Horizontal, Single
Belt, Strip Casting Simulator. TMS 2006, 135, 851–856.
17. Mukhopadhyay, P. Alloy designation, processing, and use of AA6XXX series aluminium alloys. ISRN Met.,
2012, Article ID 165082, doi:10.5402/2012/165082.
18. Hirt, C.W.; Nichols, B.D. Volume of fluid (VOF) method for the dynamics of free boundaries. J. Comput.
Phys. 1981, 39, 201–225.
19. Manual, A Fluid Dynamics Analysis Program; FLUENT Inc., 10 Cavendish Court. Centerra Resource Park:
Lebanon, NH, USA, 1998.
20. Surface Roughness Determination Using 3D Nanovea Profilometer. Available online:
https://nanovea.com/profilometry-roughness-finish (accessed on 27 April 2020).
21. Sengupta, J.; Thomas, B.G.; Wells, M.A. The use of water cooling during the continuous casting of steel and
aluminum alloys. Met. Mater. Trans. A 2005, 36, 187–204.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).