DC Casting of Aluminium: Process Behaviour and

Technology

J.F. Grandfield and P.T. McGlade

Comalco Research Centre
15 Edgars Rd.
Thomastown
Victoria 3074
Australia

Synopsis
This paper reviews the physical phenomena and technology of DC casting of
aluminium. The basic process, its variants, history and commercial aspects are
described. The process physics such as heat and fluid flow, solidification
microstructure, thermal stress, deformation and cracking are discussed. We illustrate
how understanding the physical phenomena occuring during DC casting has lead to
modifications and new variants of the technology. This understanding has in the past
ten years often been aided by the application of mathematical models, and many
examples are given. The mechanisms of formation of the surface region and various
mould technologies used to control this region are reviewed, including low metal level
casting, hot top moulds, gas pressurised hot tops, and electromagnetic casting.
Methods of controlling deformation at cast start, and cracking are discussed. Some
engineering and safety aspects are considered and the article concludes with a brief
discussion of future prospects.

1. Introduction

This paper deals with the physical phenomena taking place during direct chill or DC
casting of aluminium and the technological methods used to control them. Emleys [1]
1976 review deals with DC casting as well as other processes, but there have been
many advances since.

2. History

In the early 1930s DC casting was invented independently by VAW (Germany) and
Alcoa (USA) [2,3,4]. Today it is the premier process for producing aluminium shapes
suitable for subsequent processing in extrusion, rolling or remelt operations. Around ten
million tonnes per annum of aluminium is DC cast worldwide. The process is also used
to cast copper, zinc and magnesium.

Before DC casting, feedstock for rolling mills was cast by the book mould process,
where a two piece steel mould contained the aluminium while water sprays cooled
the outside of the mould. DC casting enabled a finer grain structure to be obtained as
the direct cooling provided by the water contacting the casting itself produced a much
higher rate of heat extraction. This reduced microstructural variation and segregation.

The first Australian DC castings were produced in August 1956 at Comalco Aluminium
Limiteds Bell Bay smelter [5]. Four inch square bars of commercial purity aluminium
were cast for fabrication into high tension conductors. There are currently six aluminium

3
smelters and four remelt facilities in operation in Australia and New Zealand, using
three horizontal and over a dozen vertical DC casting machines, representing a fraction
of approximately two hundred DC casting units in operation worldwide.

3. Commercial Aspects

3.1 General
DC casting of aluminium provides the link between liquid metal, as obtained from
reduction cells or from scrap melting, and the semi-fabricator. Products include large
rectangular sections known as rolling blocks, rolling ingots or slab (typically 500 x
1500-2000 mm in section) which are used by rolling mills for plate, sheet or foil
production. Alloys are typically 1000, 3000, or 5000 series. Circular sections, known
as rounds or billets, up to 1.1 metre diameter, are usually sent to extrusion operations
but can be used to supply forging presses. For most extrusion applications 6000 series
alloys are used, although high silicon foundry alloys, high conductivity 1000 series
and high strength 2000 and 7000 series alloys can also be DC cast. Remelt ingots,
both pure and alloyed, in small rectangular sections (eg. 150 x 50 mm) or large T
sections (500 x 1200 mm), can be produced and are seen as an alternative to the
traditional ingot casting method of open steel moulds mounted on a belt moving
through a water bath.

3.2 Competing Processes

Continuous twin-rolls cast strip (1-12 mm) or continuously cast slab (20-75 mm) can
be used for sheet production. Use of twin-roll casters is well established for certain

4
products such as foil. Slab casters are gaining increasing acceptance. These processes
have inherent advantages over the DC cast hot rolling route: they are fully continuous
and the energy costs to roll the material to final gauge are reduced. Conversion costs
for liquid metal to final sheet have been quoted as being about 60%, and investment
costs about 40%, of the conventional DC route. However, at this stage the full range of
sheet alloys cannot be produced on these machines. For a more detailed discussion of
these processes and comparisons with DC casting see references [6-16]. There is
currently much effort being applied to developing higher productivity twin-roll casting
methods using thin gauges [17].

There are many other continuous casting processes such as Castex/Conform [18],
Properzi [19] and Southwire but these are generally suited to more specialised
applications such as overhead power cable production. These processes use a wheel or
block with a groove machined on the periphery in which the metal solidifies, as a bar
which is taken directly into a forming operation. These are currently being promoted
for remelt ingot production.

4. Process Description

DC casting produces ingots of uniform cross section, initially by containment of the

liquid metal in a cooled mould and then by direct cooling of the casting (Figure 1).

5
Almost universally, the cooling medium is water, both for the mould cooling (primary
cooling) and the direct or secondary cooling1.

liquid metal

water
gallery

liquid

direct
chill
cooling
water

solid

Figure 1: Schematic of the basic DC process. The solid moves out of the mould, liquid
is fed into the mould opening and is contained by the water cooled mould, the direct
chill water spray cools the ingot as it emerges, extracting enough heat for the solid
shell to form above the spray inside the mould and contain the liquid.

At the start of casting, the open ended metal mould has to be plugged with a starting
head or dummy block to contain the liquid and allow the cast to proceed. Metal is
poured into the water cooled mould, freezing onto the starting head. After a delay, the

This terminology as used throughout the industry is somewhat misleading, as it is the secondary

cooling which extracts ~95% of the heat.

6
starting head is lowered into a pit, or onto a runout table for the horizontal
process(Figure 2). In the vertical process casting stops when the bottom of the pit is
reached. In the horizontal version, a flying saw cuts the ingot to length as it emerges
and casting can be fully continuous. There is a version of the vertical process for
copper which also has a flying saw, and the ingot is supported by pinch rollers. The
casting speed depends on alloy and size, but is typically in the 1-3 mm/s range.

liquid metal

liquid

water
gallery

solid

direct
chill
cooling
water

starting block

Figure 2: Schematic of cast start.

Alloyed and refined liquid metal is supplied to the caster. Distribution of the metal to
the moulds depends on the technology employed. Distribution can be by separate
launders to each mould with individual flow control or by a flooded table, where the
moulds are mounted in a common water jacket and the metal flows through a
refractory pan mounted on top.

7
A typical casting plant is shown in Figure 3. A furnace, or more often multiple
furnaces, contain the liquid metal allowing it to be alloyed. The furnace tilts, or a
drain or plug hole is opened, initiating a flow of metal along the launder. In-line
treatment is usually employed to remove dissolved hydrogen, alkali metals and solid
impurities prior to entering the casting station. Water cooling is supplied either
directly from a river or lake or by a recirculating system with the moulds being fed
from a header tank. Used casting water is pumped from the pit to an evaporative
cooling tower and a holding tank.

The liquid pool depth depends on casting speed, alloy and size of the casting but is
typically around 0.7 times the billet radius. The moulds are short and casting speeds
low compared to steel continuous casting. A typical 200 mm diameter billet mould
would have only a 30 mm long mould and a pool depth of 70 mm with a cast speed of
1.7 mm/s. Typical water flows are from 2,000-4,000 mm3/s per mm of mould
perimeter. Water slots or holes are used to spray the water onto the ingot at about 2
m/s.

WATER
TOWER

DEGASSER

TILTING
FURNACE

EVAPORATORS &
WATER TMT.
PLANT

MOULD TABLE

FILTER

SOLID DC
CASTINGS

WATER
PUMP
HYDRAULIC
RAM

Figure 3: Basic layout of a typical vertical direct chill casting installation.

5. Process Behaviour
5.1 Heat Flow and Solidification

5.1.1 Importance of Heat Flow

The temperature distribution within the ingot is fundamentally important. It dictates
the position of the solidification front, cooling rates, microstructure, final properties
and stresses which determine final ingot shape and whether cracking occurs.

The casting process can be divided into two distinct phases: the run or steady state
phase, and the start or transient phase.

5.1.2 Typical Thermal Conditions During Steady State

The surface temperature at the water impact point on the ingot is around 250-300 C
[20-23] and drops rapidly (Figure 4). Water spray heat transfer coefficients are high,
in the order of 40,000 W/mK, due to nucleate boiling. This produces heat fluxes in the
5-6 MW/m2 range. High heat flux values and the high thermal conductivity of
aluminium ensures that the solid interface forms above the water quench point (Figure
5). Measurements of final water temperature show that 99% of the heat is transferred
to the water. The water temperature typically rises 40-50 C during casting, and
although the water is boiling on the ingot surface, most of the steam bubbles collapse
in the water film.

10

SURFACE TEMPERATURE (C)

700
600
500
400
300
200
100
0

-50
50
150
250
DISTANCE FROM WATER SPRAY IMPACT POINT (mm)

Figure 4: Typical surface temperature cooling curve.

11
Figure 5: Shape of liquid pool in a 6063 alloy 155 mm diameter billet casting as
shown by doping with liquid of different composition.

5.1.3 Understanding Heat Flow

The temperature distribution can be examined and understood using the
thermodynamic principle of conservation of energy. Heat transport is mainly by
convection and conduction (solid and liquid) with little radiation. Heat transfer is by
convective boiling at the cast surface. The temperature distribution in the ingot is a
function of factors determining heat input and output.

Heat input is a function of the energy content of the material and the casting rate. The
energy content comprises the specific heat of the liquid Cpl (~5% of total), latent heat
of solidification L (~35% of total) and specific heat of the solid Cps (~60% of total).
The solidification rate depends on the density , casting speed V and ingot size R.
Taking a specific example, a 200 mm billet cast at 2.5mm/s has a heat flow of around
160 kW.

The final position of the isotherms are determined by a balance between the
convective heat input VRCpL, and heat extraction by diffusion (determined by
diffusion path length R and thermal conductivity k) and convection cooling (described
by the heat transfer coefficient h). Two non-dimensional numbers characterise the
balance:

Peclet number Pe

= CpVR/k

12
= the ratio of convective to diffusive heat flow and,

Biot number Bi

= hR/k
= the ratio of resistance to heat flow from
conduction to convective cooling.

Typical values for aluminium DC casting are: 1.8<Pe<4.5 and 2<Bi<60. It can be seen
that diffusion and convection are both strong in this process. In contrast, steel
continuous casting has much higher V and lower k values giving large Peclet
numbers, ie. the diffusive heat flow in the casting direction is very small compared to
convection. The low Pe for aluminium explains why the solid forms upstream of the
water quench point.

5.1.4 Effect of Casting Variables

The above analysis provides a context for examining the general effects of casting
variables. For example if one increases casting speed, heat input increases. Increased
temperature gradients result, leading to greater diffusive heat flow, which balances the
increased heat input. Similarly, a consequence of larger ingot size is increased heat
input and a longer diffusion path. In practice, casting speed is reduced as diameter
increases and the Peclet number remains virtually unchanged. The casting temperature
has a small effect on the heat flow as the specific heat of the liquid is only ~4-5% of
the total heat input . Latent heat accounts for about 35% of the heat input and specific
heat of the solid the remainder.

13

Since the specific and latent heats for the various aluminium alloys are very similar,
variation in temperature distribution from alloy to alloy is due to changes in thermal
conductivity. As alloy content increases, thermal conductivity decreases, the pool
depth deepens and temperature gradients increase. Figure 6 shows the effect that alloy
content has on room temperature thermal conductivity. Little published data exists for
high temperatures. Alloy content also determines the liquidus and solidus
temperatures, ie. the freezing range.

AT ROOM TEMPERATURE

CONDUCTIVITY (W/cmK)

2.5

1145
6063

1000 SERIES
6000 SERIES
3000 SERIES

1.5

3004

5000 SERIES

5182

1
0.5
0

90

92

94

96

98

100

% ALUMINIUM

Figure 6: Effect of alloy content on room temperature thermal conductivity.

Examination of Biot numbers shows that due to the high heat transfer coefficient of
the water spray, diffusion through the solid controls heat flow. Variation in water heat
transfer under normal circumstances has little effect on the temperature distribution.
Thermal conductivity and ingot size control the heat flow for a given casting speed.
Water cooling will only have an effect if the heat transfer coefficient goes to a low
value, ie. a low Biot number.

14

While the above scaling analysis explains the general effects of the casting variables,
mathematical models allow detailed examination of the effect of changes in process
variables. The advection diffusion heat flow equation and appropriate boundary
conditions can be solved for temperature.

VCp

2T 2T
T
f
= k 2 + 2 + L s
y
y
t
x

(1)

This equation arises from conservation of energy. The left hand side is convective heat
flow, the first term on the right, the diffusive heat flow and the second term the latent
heat generation (fs being the fraction solid). Many numerical methods exist to solve
this equation, along with the equations for stress analysis and fluid flow. A review of
these is beyond the scope of this paper. These methods are well established and
experimentally verified. Much of the process improvement in the last ten years is
attributable to the use of these models.

As an example, Flood et al., [24] calculated a normalised temperature distribution

from cast start, using a numerical model to solve the advection diffusion equation for
the case of a cylindrical ingot over a range of Biot and Peclet numbers. An equation
was fitted to the results for steady state normalised pool depth (pool depth divided by
radius), giving a response surface for pool depth as a function of Bi and Pe. This
surface is plotted in Figure 7, showing clearly that normalised pool depth is linear
with Peclet for normal high Biot number conditions. This means that pool depth

15
increases with the square of the diameter, linearly with cast speed, and is inversely
proportional to alloy thermal conductivity.

4.00
2.00
1.00
25.00

0.00
5.00
4.00

36.00
Biot number

Maximum
normalised pool
depth

6.00

3.00

44.00
2.00

Peclet
number

Figure 7: Normalised pool depth as a function of Peclet and Biot numbers (based on
equations from [24]).

16

Pool Depth vs Casting Speed (304 mm 6061 alloy billet)

180
160

Pool depth (mm)

140
120
100
80
60
40

Predicted from scaling formula (from reference 24)

Measured by dip rod

20
0
0.75

predicted by numerical model

0.95

1.15
1.35
Cast Speed (mm/s)

1.55

1.75

Figure 8: Pool depth as a function of casting speed, from measured values, full
numerical model predictions and predictions from the fitted response surface of
reference [24].

Hakonsen and Myhr [25] constructed process maps for round and rectangular ingots,
showing the sensitivity of temperature distribution to Pe and Bi. The same behaviour

17
was found: pool depth was a linear function of the Peclet number once Biot numbers
were above twenty. They also examined the effect of cast speed and ingot size on
dendrite arm spacing, finding that in the limiting case speed had no effect and ingot
size alone would determine microstructure. This has been borne out in the
experimental findings discussed below.

5.1.5 Solidification
Under the influence of the temperature distribution, solid forms where the liquid
temperature goes below the liquidus for a given alloy. With aluminium DC casting,
TiB2 grain refiner is added to provide nucleation sites for the formation of solid alpha
aluminium crystals, giving a very fine equiaxed structure (Figure 9). This structure
displays dendritic features on the grains. If casting is performed without grain refiner,
the classical grain structure of a columnar exterior and equiaxed centre results (Figure
10). The smaller equiaxed grain size prevents cracking at normal casting speeds.
Cracking is discussed further below.

In addition to the level of grain refiner, cooling rate also determines the fineness of
microstructural features such as dendrite arm spacing (DAS), grain size and
intermetallic particle size. The cooling rate decreases from surface to centre as the
diffusion path from the water spray increases. This gives rise to a variation in
microstructure from the surface to the centre which is particularly apparent with larger
castings. In many cases, this variation is not important, however for some AlFe alloys
it causes a change in intermetallic particle phase producing a fir tree structure [26-28].
This in turn causes anodising streak defects. The volume fraction, shape and size of

18
intermetallic will also vary due to cooling rate variation from edge to centre. Very
little can be done about this variation as it is a function of the size and thermal
diffusivity of the ingot. Large changes in casting practice, for example a 50% change
in casting speed, were found to have no effect on intermetallic particle size or cast
structure in alloy 5182 rolling ingot [29]. Composition was the controlling factor for
microstructure. Apart from composition, ingot dimension and alloy thermal
conductivity are the main parameters affecting cooling rate and refinement of
microstructure.

Figure 9: Fine equiaxed structure of a grain refined DC casting.

19

Figure 10: Grain structure of a non grain refined 178 mm diameter billet.

5.1.6 Water Cooling

During cast start the severity of quench affects the incidence of cracks, while during
steady state, water spray heat transfer coefficients are normally well above the
threshold where it controls heat flow. It is only when a problem arises that water spray
heat transfer becomes significant. Therefore to control the process one needs to know
what changes in heat transfer coefficient are significant and secondly what produces
those changes.

20
The heat transfer coefficients for water cooling depend principally on ingot surface
temperature, as shown in Figure 11. When the surface temperature is below ~100 oC,
heat is extracted by convection. Above this temperature boiling occurs and heat
extraction is by nucleate boiling. At higher surface temperatures a continuous layer of
steam covers the hot surface, i.e. film boiling. In between nucleate and film boiling
there is an unstable film boiling regime. Nucleate boiling has a high heat transfer

Heat Transfer Coefficient

coefficient while film boiling and convection have much lower values.

Nucleant
boiling

Convection

100

200

Unstable
film
boiling

Film
boiling

300

400

Surface Temperature (C)

Figure 11: Heat transfer coefficient and boiling regimes for water cooling as a
function of surface temperature.

In most situations the ingot surface temperature at the water quench point during
steady state is below the critical temperature and only nucleate boiling occurs.
However, changes in water temperature, impact velocity or chemistry can promote

21
film boiling. Water chemistry changes due to casting lubricants, treatment chemicals
and concentration of species due to evaporative cooling.
Weckman and Niessen [20] developed equations to predict the heat transfer coefficient
as a function of flow rate and surface temperature in the convective and nucleate boiling
regimes. These values produced a good fit between predicted and experimentally
measured ingot temperature, indicating film boiling was not occurring. Measured heat
transfer coefficients during casting gave peak values around 50 kW/m2K [22,24,25] and
heat fluxes of about 6 MW/m2 [30]. If the impact velocity of the water jet is too low (<1
m/s), film boiling occurs [22]. Bramberger and Prinz also developed an equation for the
heat transfer coefficient [31]. While their equation predicts values lower than those
measured in DC casting of aluminium, they found increased water temperature
decreased the heat transfer coefficient. Other measurements [32-35] have been made
with various quench tests showing the effects of water composition, but they do not
exactly duplicate the conditions during casting.

5.1.7 Mould Lubrication

Mould lubrication is used to separate the casting from the mould preventing sticking
which causes drag marks on the cast surface. Successful selection and application of a
lubricant will prevent contact, while an incorrect choice or poor practice will have the
opposite effect. Lubricants range from grease, or static lube, applied at the start of the
cast, through natural oils such as castor or rape seed oils, to synthetic oils with
compositions specially developed to match the casting technology in use.

22
The oil also serves to reduce the amount of heat that flows through the mould wall.
As has been outlined elsewhere the quality of the casting is inversely proportional to
the amount of primary cooling, consequently having a uniform lubricant film between
the casting and the mould wall will increase the gap and significantly reduce the heat
flow. In the casting environment the oil is used to separate the casting from the mould
surface. The purpose is to prevent the casting dragging and sticking to the mould and,
more importantly, to reduce the amount of heat that flows through the mould wall. As
has been outlined elsewhere the quality of the casting is inversely proportional to the
amount of primary cooling, consequently having a uniform lubricant film between the
casting and the mould wall will increase the gap and significantly reduce the heat
flow. Oil properties that are important for this function are thermal stability and the
variation of viscosity with temperature [27]. An oil that is excessively thin at higher
temperatures will not produce a uniform film. Thermal stability relates to the
propensity for the lubricant to burn or break down at the temperatures experienced
during casting. On the casting side, temperatures are well above the flash point of the
oils and the oil decomposes forming a gas of H2, H2O and hydrocarbons [26]. On the
mould side, temperatures are lower and a liquid film is maintained. However, if the
mould is not well cooled and the temperature is too high the oil film becomes thin and
patchy.

Suitable pumping and flow control systems are required to get even distribution of
lubricant to all moulds and around the mould perimeters. Viscosity at room
temperature will affect pumping, piping and distribution to the moulds. The oxidation
resistance of the lubricant will determine its shelf life. It is also important that the oil
can be separated from the cooling water, to prevent environmental contamination and

23
a detrimental effect on the cooling efficiency of the water. Lubricant can be supplied
to the mould face through small ~100 micron channels or porous carbon.

24

5.2. Surface Microstructure Formation And Control

5.2.1 Surface Microstructure Formation
The surface microstructure of DC cast material is characterised by a segregation zone2,
coarse dendrite arm spacing (DAS), a large grain size and various other anomalies.
These defects can lead to problems in subsequent processing, such as edge cracking
during rolling, anodising streaks, and back end defects in extrusions. Control of surface
microstructure and the minimisation of defects have been the driving forces for many
technical developments in DC casting.

In conventional open top DC casting (Figure 12), where molten aluminium contacts the
mould at the base of the meniscus, some solid and semi-solid material forms due to heat
flow to the mould. The heat transfer coefficient for this contact point has been calculated
from mould temperatures at around 1000 W/mK [36, 37]. Thermal contraction of the
ingot causes the shell to pull away from the mould. This action causes a marked
reduction in heat flow to the mould and even a reheating effect if it occurs above the
area of influence of the water spray. A solid/liquid mushy zone may form at the surface
due to the reheating, through which solute rich interdendritic liquid is forced by the
pressure of liquid metal in the pool. This produces bumps or blebs of high solute

The surface segregation is often called inverse segregation. However, in this case the term is actually a

misnomer. Most segregation is caused by rejection of solute in front of a progressing solidification

front. So called inverse segregation occurs due to liquid feeding into the mushy material at the mould
wall to take up the volume change due to the phase change. In the case of DC casting much greater bulk
liquid displacement occurs due to the metal head pressure, and material actually oozes out of the
semisolid surface.

25
concentration on the surface of the ingot. The smoothness of the surface is therefore an
indicator of sub-surface microstructural quality. Additionally, the reduced cooling rate
produces a coarser DAS and grain size (Figure 13). This mechanism has been well
established by many workers [38-48]. Engler and co-workers [49-51] confirmed that
metallostatic head was the main driving force for the flow of interdendritic fluid rather
than the volume change on solidification.

MENISCUS BASE

MOULD COOLING

MUSHY
REHEATING
ZONE

AIR-GAP

SOLID

SECONDARY
WATER COOLING

coarse
microstructure
region at surface

Figure 12: Surface formation during DC casting.

26

400
Grain Size (microns)

350
300
250
200
150
100
50
0
0

10

12

Distance from cast surface (mm)

Figure 13: Measured grain size showing coarse microstructural zone in DC cast 3004
alloy.

To control the reheating effect, mould technologies have been devised over the years
which reduce or eliminate mould cooling. These methods are discussed in more detail
below.

27

5.2.2 Molten Metal Level Control

Controlling the liquid level is the simplest method of improving the near surface
microstructure. There are several practical problems however, such as knowing the
optimum level. If the level is too low, the meniscus freezes; and as casting continues it
moves down the mould and new liquid laps over it forming a fold in the surface
(Figure 14). One simple way to find the optimum level during casting is to reduce it
until folding occurs and then to slightly increase it. To know the optimum metal level
one needs to know how far into the mould the solid extends. Harrington and Groce
[52] defined the length of solid in the mould as the upstream conduction distance
(UCD). They presented a method of calculating the UCD based on the assumption
that the heat flow at the surface, above the water spray, is all toward the water spray
point. This is a good assumption as only a small amount of heat crosses the air gap to
the mould. Their equation simplifies to:
UCD =

Fk
V

where F is some factor dependant on density, specific heat, latent heat, liquidus and

the surface temperature at the water spray quench point.

UCD =

F
kV Obviously as cast

speed increases and thermal conductivity decreases, the UCD decreases (Figure 15).

28

Figure 14: Cold folds formed on the surface of a DC casting (typical fold spacing
~8mm).

29

Upstream Conduction Distance

(mm)

120
1145
3004
5052
5182
6063

100
80
60
40
20
0
0.5

0.7

0.9

1.1

1.3

1.5

1.7

Casting Speed (mm/s)

Figure 15: Upstream conduction distance as a function of alloy conductivity and cast
speed.

The UCD can be calculated more accurately by more complex two dimensional
numerical models solving the full heat flow equations. Grandfield and Devadas

30
modelling results showed how decreasing metal level reduced the reheated area [53],
while Brochu and Hank measured the reduction in mould chill zone with lower metal
levels [54], which matched the full solute and heat flow modelling work of Mo and
others [55, 56].

The inverse relationship between cast speed and UCD or mould length has been found
in a number of instances. Weckman and co-workers modelled horizontal continuous
casting for various non-ferrous metals and showed that the optimum mould length
could be calculated as a function of Peclet number [57,58]. Vorren and Brusethaug
calculated UCD as a function of alloy and diameter using a two dimensional heat flow
model, finding that, only at diameters less than 150 mm, was there any change in
UCD with diameter [59]. One of the difficulties of casting rectangular sections over
rounds is that the corners introduce regions with very small radii and the UCD is
much higher there. In practice the mould design is modified so that the water quench
point is lower on the corners, or even eliminated.

With float and spout technology, low metal levels are not used because of a perceived
danger from bleedouts and freezing the spout in at cast start. The Isocast system was
intended to solve this by moving the whole mould table up during the cast, bringing
the metal level down. However, now available automatic level control systems give
very accurate control within a millimetre and allow the level to be adjusted during the
cast [60].

5.2.3 Hot-Top Casting

31
Insulation placed at the top of the mould was introduced to control the location of the
initial contact point of the liquid against the mould [61] (Figure 16). In this way, the
liquid metal contact point was fixed at a point below the insulation and was independent
of the metal level used. This point should coincide with the UCD minimising reheating
and improving surface quality. Modern hot top moulds are similar in principle, with a
refractory header inserted into the top of the mould (Figure 17).

32

REFRACTORY

LIQUID

SOLID

MOULD
COOLING

SOLID

WATER
COOLING

Figure 16: Simple hot top mould formed by addition of thin insulating paper to an open
top mould.

FEEDING
TROUGH

REFRACTORY
HEADER
MOULD WITH OIL
SUPPLY
LIQUID
WATER
SPRAY
SOLID

33

Figure 17: More sophisticated hot top mould.

The change to hot-top moulds was a significant change in technology because:

1. The flooded table was developed to replace the launder/float arrangement.

2. Level pour from the furnace could be used when tilting furnaces were employed,
where the level above the mould matched the level in the furnace, reducing melt
turbulence and dross generation.
3. Increases in packing density: the number of moulds in a given table area doubled,
increasing the tonnage of each cast and total throughput.
4. Manning levels were halved.

Hot-top casting can suffer from cold folding if the mould length is too short for a given
casting speed and alloy, causing the water cooling effect to extend up to the refractory.
However, folding is usually due to heat flow from the mould causing freezing of the
meniscus immediately below the refractory (Figure 18). Bergman established the
mechanism for these folds [62]. The meniscus freezes, it is pulled down, solidification
proceeds inward and then upward until the path is too long and liquid breaks through to
fill the gap and the cycle repeats. Weckman and co-workers modelled this defect and
were able to calculate the value of the mould heat transfer coefficient from the angle the
fold made and the casting speed [37,63]. Formation of folds is controlled by the mould
heat transfer, the casting speed and the geometry of the refractory overhang. Higher
mould heat transfer and slower casting speeds increase the spacing and depth of folds.
As metal head pressure increases so too does the mould heat transfer, and this is why

34
horizontal DC casting tends to be more prone to cold folding as the metal heads are
usually about 400 - 500 mm compared to 100-200 mm for vertical DC casting.
Lubrication also plays a role in affecting the mould heat transfer and cold fold
formation. If the refractory overhang is shorter then the liquid metal floods back into the
corner sooner.

Another method for controlling cold folds is to cast at a speed such that the UCD
extends to the base of the meniscus. In this way the shell pulls away from the mould and
there is no mould contact. Alcan developed a mould design based on this principle [64].
The difficulty with this approach is that if the UCD extends only slightly further, it
freezes metal up to the refractory and may even pull the refractory out of the mould.

refractory
meniscus
position

solidification
mould

front
casting
direction

(a) Solidification front

position in relation to
meniscus position.

(c) The fresh liquid freezes

against the mould, forming a
new meniscus and
corresponding void.

(b) As casting continues

the solidified meniscus
descends, allowing
liquid metal to fill the
void.

(d) The cycle is repeated

as casting continues.

Figure 18: Schematic of the mechanism of cold folding in a hot top mould.

35

5.2.4 Air Assisted Hot-Top

In 1977, Showa Aluminium Limited (SAL) found that better control of the meniscus
contact point could be obtained by injecting air just below the refractory [65-70] (Figure
19). The Showa process produced billet with a very smooth cast surface and low mould
chill zone depth compared with conventional float cast and hot-top technology. In effect
gas pressurisation makes the mould self controlling. Gas pressure rises in the meniscus
region until it finds an escape route. As pressure rises, the meniscus is pushed away
from and down the mould. Eventually the meniscus base reaches the UCD where the
shell is contracting from the mould allowing the air to escape. The pressure in the
meniscus region then equilibrates against the metal head pressure.

This

technique

significantly reduces heat flux through the mould wall. For example, measurements
showed a reduction of heat flux from 1270 kW/m2 to 420 kW/m2 [71] on a 155 mm
mould. Mould temperatures reduce from around 100-150 C to 30-40 C (Figure 20).

36

FEEDING
TROUGH

REFRACTORY
HEADER

MOULD

WATER
SPRAY

GAS AND OIL INJECTED

THROUGH GRAPHITE RING

LIQUID

SOLID

Figure 19: Gas pressurised hot top process.

GRAPHITE RING

37

120
110

NO GAS APPLIED

100

TEMPERATURE C

90
80
70

BUBBLING CAUSING
TEMPERATURE SPIKE

60
50
40
30
20
10
0

LOW TEMPERATURE
IN GAS PRESSURISED
MODE
CAST LENGTH (mm)

Figure 20: Effect of gas pressurisation on mould temperatures.

By 1986, licences for the SAL technology had been sold to many companies. Since that
time many companies have developed their own in-house hot-top technologies, eg.
VAW [72], MOSAL (now Elkem Aluminium), Hydro Aluminium, Alusuisse [73], and
Pechiney [74]. Approximately fifty SAL casting stations have been installed, mostly in
Japan. In 1980 Wagstaff began development of a gas-pressurised mould where the gas
was injected through a graphite ring, and was granted an Australian patent in 1983 [7577]. Since then they have become the dominant supplier of air-assisted billet moulds.

Papers presented to The Extrusion Technology Conference in 1988 showed airpressurised billet casting was widespread [78]. Extruders were asking for smooth
surfaced air-pressurised billet, even though it was widely accepted that factors such as

38
homogenisation and composition control were more important than sub-surface
microstructure or surface appearance.

Air-pressurised billet casting does have a number of drawbacks, but they are considered
minor in relation to the benefits obtained from the technology:

1. The Showa system in its original form is labour-intensive and sensitive to operator
interaction. The manual control of air flow can cause difficulties. The VAW version
of the process has individual automated mould pressure control.
2. Showa moulds were designed for sliding tables, not the newer (generally preferred)
tilting tables.
3. The porous graphite ring is an expensive consumable, its permeability changing with
age, causing variation in gas flow and deterioration in billet quality.
4. Wagstaff moulds require individual attention between each cast.
5. All air-assisted casting technologies are more complicated than conventional casting
systems and are reliant on an informed, well-trained and motivated work force.
6. A commitment to specialist mould maintenance is required.

5.2.5 Air-Assisted Sheet Ingot Casting

Showa Aluminium Limited, SKY Aluminium[79], Comalco Aluminium Limited and
Wagstaff [80] investigated the possibility of air-assisted sheet ingot casting in the late
80s and early 90s. The principle being identical to air-assisted billet casting. Although
much pilot plant trials have been conducted, little commercial implementation has
occurred to date. One problem was that a fixed, short mould length meant that no
deformation of the butt could be tolerated or the butt would deform and lift into the

39
refractory. This problem was solved with the development of the Turbo Curl Reduction
Technology [81], as discussed below. At Comalco Research Centre, studies showed
very high quality surface microstructures, equivalent to air pressurised technology, can
be obtained with open top casting using automatic metal level control without the
complication of air pressure control. Wagstaff have also taken this route with a
modification using a carbon mould liner and automatic level control [82].

5.2.6 Electromagnetic Casting (EMC)

The electromagnetic casting mould uses an inductor coil through which a high
frequency (typically a few kHz) current is passed. The currents induced in the liquid
metal interact with the magnetic field of the coil. This produces a restraining force on
the liquid which acts against the metallostatic head pushing the metal away from the
mould. Lack of contact between liquid and mould eliminates mould cooling and the
problem of reheating, producing very good surface microstructures.

Getselev and Cherepok at Samara Metallurgical Works in the Soviet Union first
developed EMC casting in the 1960s. They were granted a US patent in 1969 [83]. The
history was reviewed by Yoshida [84]. Kaiser [85, 86] and Alusuisse [87-89] developed
the process further. EMC promised to reduce or eliminate scalping losses as the mould
chill zone depths were small. However, reduced scalp depths necessitate flat rolling
faces, which, in turn, necessitated development of metal level control. It is the balance
between metal level and the magnetic field that controls shape and position of the free
surface and final dimensions of the block.

40
Another problem which had to be solved was the start-up of EM casting, which
presented additional difficulties compared to normal DC casting[87]. Over the last 10
years automated systems have been developed and extensive mathematical modelling
carried out to solve these problems [90,91].

Emley said "a new Russian development termed electromagnetic or 'mouldless' casting
holds promise of benefits scarcely imaginable with any other form of DC casting" [1].
Sixteen years later, while EMC is used (at least 2 million tonnes are cast per year worldwide), it has not been universally accepted. Certainly, better microstructures result and
reduced scalp depths are taken, but little sheet is produced without scalping. High capital
and licensing costs do not outweigh the improved recoveries. Some EMC installations
are now being decommissioned because of the royalty costs, and there are unlikely to be
any more built. Development of EMC led to improvements in conventional DC casting.
For example, control of metal level and development of automation systems were
originally driven by EM casting.

A variant of EMC, the Pechiney CREM process, uses low frequency 50 Hz current to
induce a high degree of stirring in the melt [92-97]. This stirring gives rise to dendrite
fragmentation and grain refinement. In this way, use of costly TiB2 grain refiner can be
eliminated. The process also gives rise to good cast surface microstructure by
controlling the meniscus contact point on the mould. Consequently, reduced scalp
depths and edge trim are taken [90].

41
5.3 Stress, Cracking And Shape
Thermal contraction occurs on cooling. Because of the different thermal history of
various parts of the casting they contract at different times and at different rates. This
differential in contraction produces stresses, as one part restrains another. The level of
these sStress in an inherent aspect of the DC casting process. Inability to acknowledge
and control the stresses produced during casting will inevitably lead to scrap and
hazardous situations.It is the stresses that help defines the final shape of the casting
and determines whether a crack free product is made.

Once one area has solidified and contracted it resists further contraction.
Consequently when adjacent areas solidify and attempt to shrink stress is usually
produced. Similarly the shrinkage of the solid will not be uniform but will be
determined by the thermal gradients present. It is this difference that gives rise to the
variation in shape of the roll face of cast rolling ingots.

With circular sections, (Figure 21)In the mould, , metal first solidifies as a ring or
annulus of solid forms. It is cooled by the water spray and thermal contraction takes
place. As casting continues the metal in the centre will freeze and its natural tendency
would be to shrink. However, at this stage the surface material is almost at the water
temperature and has no tendency to contract. As the cast continues heat is removed
from the centre of the billet, when sufficient heat has been removed solidification will
take place, with its attendant shrinkage. Now, depending on the casting condition
prevailing, stress will be established across the cast section. This follows from the
constraint imposed on the central section by the outer area that has already solidified
and contracted. Therefore the central section, as it cools, is placed under tension by

42
the restraint from the outer material and the outer material is placed under
compression by the internal material. shrinks produces a compressive stress in the
outer region and a tensile stress in the centre. If the stresses generated exceed the
strength of the material a crack will form. Examination of crack surfaces shows that
some liquid is present when cracking occurs. These are hot cracks which occur at the
last stage of solidification. Measurements by Bryson during casting, with an ultrasonic
crack detector, confirmed that cracks formed at the base of the liquid pool during the
final stage of solidification [98].

43

Figure 21: Predicted temperature distribution during steady state casting of a 152 mm
diameter 6063 alloy billet cast at 2.5 mm/s.

For a given alloy type and diameter there is a casting speed above which internal hot
cracks occur. Interestingly, this maximum casting speed is found to be inversely
proportional to diameter [99]. This is understandable in a general sense. As the speed
increases, the temperature gradient increases and so do the temperature gradient

44
differences between surface and centre. This limit on casting speed can limit the
productivity of a casting installation.

Much work has been done on defining cracking criteria and hot tearing mechanisms
[100-103]. Bryson, having regard to the basic argument above, proposed a very simple
criterion; the difference between the surface and centre cooling rates. Bryson also
proposed a method of reducing the stresses. The initial water quench is made less
severe so the surface stays hotter and, at the point corresponding to the base of the
liquid pool, a second quench is applied so that the surface and centre cool and contract
together. Nawata et al., [104] found they could increase the casting speed of a 6063
alloy 178 mm diameter billet from 1.5 mm/s to 10 mm/s without cracking using this
delayed quench system. Considerable stress modelling has been applied to the
problem [103, 105-117]. Modelling by Fjaer, et al. [103], for example, predicted that
the strain rates in the mushy region during casting with a delayed quench were
reduced over that of a normal quench. They proposed the ratio of strain rate to cooling
rate in the mushy region as a cracking criterion.

Conditions that encourage crack formation are higher casting speed, highly alloyed
metal and poor grain refining practice. The first two increase the depth of the liquid
sump making feeding more difficult, while the third will reduce the strength of the
solid as well as affecting feeding.
Another parameter affecting the incidence of hot tearing is alloy microstructure. If the
grain size is small, cracking is far less likely. Ohm and Engler found that addition of
grain refiner increased the maximum strain tolerable to mushy material before fracture
[49]. Additionally, cracks can form at the start of the cast. Jensen and Schneider

45
examined the formation of internal cracks at the start of billet casting [118-119]. They
found that as the liquid pool develops, it goes through a maximum depth before it
reduces to the steady state value. As the stress imposed on the solid increases with
pool depth, the stress similarly went through a maximum. For many years the practice
has been to start a cast at a slower speed and increase to run speed, preventing sump
overshoot. Jensen and Schneider invented another method: the centrally raised dummy
block. By having the centre higher, the pool does not overshoot and internal cracks are
eliminated.

To roll ingots, the rolling face must be flat. Variation in contraction along the roll face
varies and must be accounted for in mould design as the deviation from flatness must
be scalped off. Contraction is greatest in the centre of the face. The main casting
parameters that affect shape are casting speed and alloy, as they are the factors that
most influence the thermal gradient through the casting. A typical profile of a
commercially produced rolling block is given in Figure 22.

The factors that affect roll face shape are generally well known, but historically mould
profiles have been developed empirically [120]. This is achieved by using the errors in
current block shape as input to the design of the next set of moulds. Such a system
works well for castings made under the same conditions, but does not work for
different casting speeds or alloys. The other disadvantage is that it takes one life cycle
of the mould to correct the shape. A number of shape models have been developed to
predict the shape of the roll face when the conditions change or new mould sizes are
required [120]. Further, the ability of machine tools to produce the desired contour has
improved so that now moulds can be produced that cast very flat rolling block.

46
Currently, the only reason that some rolling block is not flat is that a compromise has
been made on the number of moulds that are purchased. When different alloys are
produced in one plant, normally only one set of moulds is used for each size.
Consequently, one mould set up to produce flat 3004 rolling block, for example, will
not produce flat 5182. While some compensation can be made by changing casting
speed, the resultant shape is not perfectly flat.

Roll Face Deviation

4

Deviation from Flat (mm)

3.5
3
2.5
2
1.5
1
0.5
0
0.00

200.00

400.00

600.00

800.00

1000.00 1200.00 1400.00

Distance Across Roll Face (mm)

Figure 22: Typical rolling face profile.

5.4 Cast Start

Because of the semicontinuous nature of VDC casting, cast starts are important. The
transient nature of the heat flow dictates that the stresses generated and resultant
deformation is quite different to that occurring during steady state. When the starting

47
head and mould are filled at the start, a permanent mould casting is made with heat
flow predominantly into the water cooled dummy. After a delay, the platen
commences its descent, with the casting parameters changing with time to
accommodate the change from stationary casting to the run condition. Cast speed,
water flow and liquid level are all altered in an attempt to minimise the effect of the
heat flow changes. This transition lasts for about one diameter or one thickness.

The phenomena associated with cast starts are:

1. Butt curl: where the first part of the casting lifts off the dummy block.
2. Hang ups: where the deformation of the butt is such that the ends wedge in the
mould.
3. Cracking: the stresses generated can exceed the strength of the material.
4. Bleed outs: as the butt deforms it can move away from the mould and the cooling
water, producing an unsupported and uncooled shell. The superheat of the liquid
pool and the metallostatic pressure are sufficient to force the liquid through the
shell. Similarly the deformation can move the shell away from the mould
producing a gap which can allow the metal to escape.
5. Surface defects: with the changing conditions experienced at the start it is often
difficult to obtain the correct balance to avoid surface defects such as cold folding,
oxide patches or, for air assisted moulds, a generally poor surface due to the
incorrect position of the meniscus.
6. Swell: the cross section of the first part of the casting is greater (~7%) than the
steady state section. This is a big problem for rolling block where the thickness

48
dimension may be greater than specification for a large part of the cast length. The
excess material must be removed

These phenomena are generally worse with larger sections and higher alloy content.
Measurements by Droste and Schneider showed the butt deforms when it first
emerges into the water spray [121]. The deformation rate is greatest at this stage
(Figure 23). There is still considerable debate about the mechanisms of butt curl, but
the basics are that the water quench rapidly cools the exterior of the casting while the
interior is cooled slowly, causing different contraction rates. The solution is then
obvious; reduce the intensity of the water cooling. This is achieved by promoting film
boiling. Reducing the water flow will promote film boiling and is probably the easiest
and most effective method of controlling cast starts. There is often a conceptual
problem to overcome when this approach is suggested. More water is often associated
with safer casting. With cast starts the reverse is often true. Similarly increasing speed
to the run speed or even faster at cast start, will raise ingot surface temperature and

49
promote film boiling.

Figure 23: Deformation of the butt at cast start for 610 x 1300 mm ingot.

One problem that arises with reduced water flow is that the water curtain can break.
A discontinuous curtain leads to uneven cooling, causing cracking and bleed outs. A
solution to this dilemma is to use moulds that have water holes instead of water slots.
The advantage is that, providing the hole size is correct, the water chamber remains
pressurised thereby maintaining a good distribution even when the flow rate is
reduced.

Butt Deformation
40.00
35.00

Butt curl (mm)

30.00
25.00
20.00
15.00
10.00
5.00
0.00
0

100

200

300

400

500

600

700

800

Time (sec)

Three other methods that have been developed to promote film boiling. The Alcoa
CO2 process [122], and the Wagstaff Turbo process [82] use gases to promote film

50
boiling. The Alcoa process dissolves CO2 into the water stream, which remains
dissolved in the pressurised water system and mould gallery. When water exits the
mould the CO2 comes out of solution forming gas bubbles which act in a similar way
to steam bubbles and form an insulating layer over the cast surface. In the Wagstaff
process air bubbles are introduced into the cooling water on exit from the mould. Air
is insoluble in water and so the bubbles stay entrained assisting the formation of the
steam layer. The Alcan Pulsed Water process uses a special rotary valve to turn the
water on and off [123]. In this way the average heat transfer rate is low and the surface
temperature stays high enough to cause film boiling.

A number of methods have been developed to control the water quench contact point.
The first is Wagstaffs Dual Jet concept [124] while others use an air film or baffle to
move the water film up and down the castings surface [125]. As the name implies
two water jets are used in the Wagstaff system. The contact point and angle are
different for each. The lower impacting jet is used at cast start to increase the mould
length and reduce quench severity. At a certain point in the cast, water is supplied by
the second jet, moving the mould length back to its conventional position. An added
advantage is that the interaction of the two jets on the cast surface produces an even
greater cooling effect than single jet systems. In the other system an air jet is
introduced above the water curtain at cast start. The air pushes the water contact point
down the casting, increasing the effective mould length. As the cast proceeds, the
amount of air is reduced and the contact point moves up the casting.

51
5.5 Liquid Flow
Liquid flow affects the temperatures within the liquid pool, and therefore cooling rate
and microstructure. It can also affect compositional variation within the casting.

Measurement of liquid metal flow is very difficult. Vives developed a magnetic probe
to verify model predictions [96]. However, to investigate flow conditions and their
effects, much mathematical modelling has been done [90,91,107,126-128].

Liquid metal flow is determined by three drivers:

1. The direction and velocity of the initial inlet to the liquid pool,
2. Natural convection in the pool, and
3. The shape of the liquid pool.

If a float and spout is used in vertical billet casting , jets of hot liquid exit the float
horizontally. Natural convection arises due to cooling of the liquid at the solidification
front. This colder liquid is denser than the hotter central liquid, flowing down the
solidification front toward the centre. These two driving forces set up a recirculation
of the fluid. In the case of horizontal DC casting, because gravity acts across the
casting direction, the natural convection flows are asymmetric. This can contribute to
microstructural variation from top to bottom across the cast product.

The flow is even simpler for the hot top case, with a fairly even flow coming down the
throat of the hot top. Studies have shown that the size of the inlet to the hot top can
effect the temperature distribution in the liquid pool [129]. If the inlet is narrow there
is some asymmetry created in the temperatures in the liquid pool. The liquid

52
distribution issue is more important with rolling ingot where the rectangular section
requires use of special distribution systems. Fluid needs to be directed to the short
faces as it has further to travel. The temperature around the perimeter changes with the
feeding system. Brochu et al., using experiments and numerical simulation showed
that metal temperature can affect the size of the mould chill zone and that certain
feeding systems were better than others [130]. Raffourt et al., have also modelled the
flow for different feeding systems although they did not take into account the true
shape of the liquid pool [127].

5.6 Macrosegregation
In larger DC castings, significant compositional variation is found. Elements that
form eutectics with aluminium rise in concentration away from the edge of the
casting, with the centre being depleted. Elements that form peritectics with aluminium
show a reverse trend (Figure 24). This composition difference is likely to give rise to a
variation in properties across the sheet width when rolled. Interestingly, the authors
know of no study which has actually linked final sheet property variation to
macrosegregation.

The central microstructure of the casting shows a duplex structure with a mixture of
large and small grains and DAS [131,132]. The general mechanism for
macrosegregation in castings is movement of solid (generally purer than the liquid) or
liquid (generally solute rich) formed during solidification. The duplex structure
suggests crystals have been transported to the centre. The modelling of Flood and
Davidson showed that the natural convection driven flow is very important in its
effect on macrosegregation [126]. The strong flow down the solidification front is

53
proposed to transport crystals to the centre and additionally that flow gives rise to a
nearly isothermal central liquid pool which is conducive to growth of large crystals.
As ingot size increases macrosegregation becomes stronger [126]. Garepy and Caron
studied the effect of various feeding methods and grain refiner additions on
macrosegregation [133]. They found that liquid flow, which worked against the
natural convection, and reduced amounts of grain refiner, reduced macrosegregation.
Mortensen and Harkonsens modelling showed natural convection and turbulence
must be included for prediction of DAS and liquid temperatures, and also found a
course centre zone [128].

54

% OVER NOMINAL

25
20

Titanium

15

Silicon

10

Magnesium

5
0
(5)
(10)
(15)
0

50
100
150
200
250
DISTANCE FROM SURFACE (mm)

Figure 24: Compositional profile across a rolling ingot.

6. Engineering Aspects

Engineering aspects of the process are important. Due to space limitations, however,
only a few points can be raised here.

Over the years, increasing quality demands on dimensional tolerances such as

straightness, and the desire to cast greater weights have driven development of
improved platen and hydraulic systems. Increasingly, VDC pit operators are using self
guided rams. These large diameter rams (up to 1000 mm) do not use external guides
to prevent horizontal movement during platen descent, but rely on the stiffness of the

55
ram and an internal keyway to prevent rotation or lateral movement. The advantage of
this design is that there are no guiding shoes or rollers running down a rail. Therefore
nothing can catch on the guides and upset the smooth action of the platen.

DC operators faced with greater quality demands have responded by increasing the
level of supervision and control. Sensing, automation and control have provided the
greater attention to detail essential for precise casting. Casting parameter sensing and
control have followed the development of increasingly intelligent sensors. Casting
speed and water flow are the key casting parameters and so require exact control. As
sensing technology improves, other parameters such as air flow and pressure, and
sometimes both, grain refiner feed, melt temperature, metal level and mould fill rate
are being added to the list. Automation systems are currently available that enable the
entire cast to be run without human intervention. The great advantage of this is the
accuracy of control that not only enables more difficult alloy / size combinations to be
successfully cast, but also improves repeatability.

Mould design is becoming increasingly complex as casters strive to meet the demands
of downstream manufacturers. The delivery systems for air, lubricant and water must
be carefully designed to ensure even distribution between and within moulds on a
mould table. Reduction in the size of the moulds is desired to increase packing density
and productivity. Also, the designs have improved to give greater robustness and
cheaper manufacture.

56

7. Safety

The main concern during DC casting is that water can be trapped by the liquid metal.
The transformation to steam results in a huge, and often rapid, volume expansion and
a blast of considerable force. Such an event can occur during furnace charging,
especially when wet scrap is loaded into a partially full furnace, or during DC casting
when the liquid breaks through its shell. An added complication for those casting
aluminium and magnesium is that these metals have a high affinity for oxygen, which
leads to a reaction between the water and the metal, producing metal oxide and
hydrogen and a great deal of heat and the possibility of a subsequent hydrogen
explosion. Furthermore, the affinity for oxygen can allow the metal itself to burn,
particularly in the case of magnesium. Finely divided aluminium, such as the powder
or liquid ejected from a water explosion, will also burn violently.

A special case is aluminium-lithium alloys, which can react violently with water even
in the solid state. Special coolants are employed such as glycol, to avoid contact
between lithium and water.

The best defence in DC casting is to be aware of the water entrapment problem.

Engineered solutions to the problem centre around controlling molten metal spills to
ensure they do not trap any water. In addition special coatings, such as coal tar epoxy
paints, are applied to pit walls and equipment. Other coatings are being developed to
replace coal tar epoxy paint, as they have safety concerns of their own. These coatings
have been found to prevent the violent reactions associated with trapped water.

57

Automation of the casting process is seen as a positive contributor to safety [53]. The
accuracy and repeatability of process control associated with automation means that
the desired safe casting procedure will be followed precisely. Secondly, automation
enables hands free operation of the process and controlled aborts without human
intervention. This means that people may be removed from the potential hazard and so
minimise the risk.

8. Future Developments

When predicting the future of DC casting one must consider the driving forces for and
impediments to change. In the past, ingot surface quality has been a strong driver in
the development of new versions of the process. Process productivity and versatility
will drive change in the future. Impediments to change include capital tied up in
existing plant and organisational and customer acceptance of change. Any new
machine or process has to offer considerable advantages before the old one can be
discarded or modified.

For many installations, tonnes cast per day is limited by liquid metal supply not
casting speed, but this is because the facilities have been sized that way. If casting
speed could be increased, greater tonnages could be cast with less capital cost. If
higher speeds were used it may be possible to go to fully continuous vertical type
operations as used in steel casting.

58
The other area activity is occurring is new products for DC casting such as more
complex shapes. The strength of the process is its product flexibility and versatility.
While continuous strip and slab casters are taking over some of the production of
sheet product, extrusion billet production will remain with DC casting. The process
will continue to be widely used well into next century.

59

9. Bibliography

General Solidification and Casting Texts

W. Kurz and D.J. Fisher, Fundamentals of Solidification, 3rd ed., Trans Tech
Publications, 1992, Switzerland.

J. Campbell, Castings, Butterworth Heinmann, 1991.

D. Altenpohl, Aluminium Viewed from Within, Aluminium Verlag, 1982.

DC Casting Review Articles

E.F. Emley, International Metals Reviews, June 1976, p. 75.

D.C. Weckman and P. Niessen, Metals Technology, Nov. 1984, Vol. 11, p. 497.

For more specicific DC casting articles a large number appear in the Light Metals
confrence proceedings published by The Minerals, Metals and Materials Society,
Warrendale, P.A.

60
Continuous Casting

H.D. Merchant, et al., Continuous Casting of Non-Ferrous Metals and Alloys, H.D.
Merchant, et al., eds., The Metallurgical Society of AIME, Warrendale, P.A., 1988, p. 1.

Heat Flow and Transfer

D.C. Weckman and P. Niessen, Met. Trans. B, 13B, 1982, p. 593.

M. Rappaz, et al., Modelling of Casting, Welding and Advanced Solidification

Processes VII, M. Cross and J. Campbell, eds., The Minerals, Metals and Materials
Society, Warrendale, P.A. 1995, p. 449.

K. Buxman, Heat and Mass Transfer in Metallurgical Systems, D.B. Spalding and N.G.
Afgan, eds., 1979, McGraw-Hill.

Numerical modelling

A. Mo, et al., Light Metals 1994, U. Mannweiler, ed., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1994, p. 889.

S. Flood, et al., Modelling of Casting, Welding and Advanced Solidification Processes

VIII, M. Cross and J. Campbell, eds., The Minerals, Metals and Materials Society,
Warrendale, P.A. 1995, p. 801.

61
A. Harkonsen and D. Mortensen, Modelling of Casting, Welding and Advanced
Solidification Processes VIII, M. Cross and J. Campbell, eds., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1995, p. 763.

Stress and Cracking

W. Schneider and E. K. Jensen, Light Metals 1990, C. M. Bickert, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1990, p. 931.

N. Bryson, Light Metals 1972, W.C. Rotsell, ed., The Metallurgical Society of AIME,
Warrendale, P.A., 1972, p. 429.

M. Nedreberg, Thermal Stress and Hot Tearing During the DC Casting of AlMgSi
Billets, PhD thesis, University of Oslo Department of Physics, 1991.

H. D. Brody, et al., Modelling and Control of Casting and Welding Processes IV, A.F.
Giamei and G.J. Abbaschian, eds., The Metallurgical Society of AIME, Warrendale,
P.A., 1988, p. 351.

W. Schneider and E. K. Jensen, Light Metals 1990, C. M. Bickert, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1990, p. 931.

62

10. References

1.

E.F. Emley, International Metals Reviews, June 1976, p. 75.

2.

W. Roth, Deutsches Reichs patent No. 974203, Day of registration 8/9/1936.

3.

W.S. Peterson, in Hall-Heroult Centennial First Century of Aluminium Process

Technology, 1886-1986, The Metallurgical Society of AIME, Warrendale, P.A.,
1986, p. 154.

4.

W.T. Ennor, US patent 2301027, Nov. 3, 1942.

5.

B. Carroll, Of Potlines and People; A History of the Bell Bay Smelter,

Comalco Limited 1980, p. 40.

6.

Technischer und wirtschaftlicher Vergleich von kontinuierlichen Giess - und

Walzverfahren zur Herstellung von katgewalzten Aluminiumbanden, Deutche
Gesellschaft fur Materialkunde e.V. 1984.

7.

R. Bachowski, et al., Casting of Near Net Shape Products, Y. Sahai et al., eds.,
The Metallurgical Society of AIME, Warrendale, P.A., 1988, p. 299.

8.

D. Jaffrey, et al., Metals Forum, Vol. 7, No. 2, 1984, p. 67.

63
9.

J.F. Grandfield and P.T. McGlade, Second Asian Pacific Course and Conference
Aluminium Melt Treatment and Casting, Melbourne University 1991, p. 65.

10.

H.D. Merchant, et al., Continuous Casting of Non-Ferrous Metals and Alloys,

H.D. Merchant, et al., eds., The Metallurgical Society of AIME, Warrendale,
P.A., 1988, p. 1.

11.

A. Kamio, J. Japan Soc. Tech. Plasticity, Vol. 22, No. 247, Aug. 1981, p. 779.

12.

A.M. Odok and I. Guyongyos, Light Metal Age, Dec. 1975, p. 28.

13.

E.A. Bloch and G. Thym, Metals and Materials, 6, 1972, p. 90.

14.

B. Frishknecht and K.P. Mainwald, Light Metals 1987, R.D. Zabreznik, ed., The
Metallurgical Society of AIME, Warrendale, P.A., p. 365.

15.

P. Crouzet, Light Metal Age, April 1989, p. 13.

16.

A. Odok and G. Thym, Aluminium, Vol. 50, No. 7, 1974, p. 454.

17.

B.Q. Li, J.of Metals, May 1995, p. 29.

18.

J.F. Langweger and B. Maddock, Proceedings 4th International Aluminum

Extrusion Technology Seminar, V1,
p. 533.

The Aluminum Association, 1988,

64

19.

R.C. Reuter, Aluminium Cast House Technology, M. Nilmani, ed., The

Minerals, Metals and Materials Society, Warrendale, P.A., 1995, p. 395.

20.

D.C. Weckman and P. Niessen, Met. Trans. B, 13B, 1982, p. 593.

21.

J. A. Bakken and T. Bergstrom, Light Metals 1986, R.E. Miller, ed., The
Metallurgical Society of AIME, Warrendale, P.A., 1985, p. 883.

22.

J. F. Grandfield and P. Baker, Solidification Processing 1987, 3rd Conference,

The Institute of Metals, 1988, p. 260.

23.

E. K. Jensen, et al., Light Metals 1986, R.E. Miller, ed., The Metallurgical
Society of AIME, Warrendale, P.A. 1985, p. 891.

24.

S. Flood, et al., Modelling of Casting, Welding and Advanced Solidification

Processes VIII, M. Cross and J. Campbell, eds., The Minerals, Metals and
Materials Society, Warrendale, P.A. 1995, p. 801.

25.

A. Harkonsen and O.R. Myhr, Dimensionless diagrams for the temperature

distribution in direct chill continuous casting, to be published in Cast Metals.

26.

C. A. Muojekwu, et al., Met. Trans, 26B, 1995, p. 361.

27.

L. Laemmle and J. Bohaychick, Lubrication Engineering, Nov. 1992, p. 858.

65

28.

H. Westegen, Z. Metallkunde, Bd. 73, 1982, h.6, p. 360.

29.

S. Brusethaug, et al., International Light Metals Congress Leoben-Vienna 1987,

Austria Metall AG, 1987, p. 472.

30.

M. Rappaz, et al., Modelling of Casting, Welding and Advanced Solidification

Processes VII, M. Cross and J. Campbell, eds., The Minerals, Metals and
Materials Society, Warrendale, P.A. 1995, p. 449.

31.

M. Bramberger and B. Prinz, Materials Technology, April 1986, Vol. 2, p. 410.

32.

J. Langlais, et al., Light Metals 1995, J.W. Evans, ed., The Minerals, Metals
and Materials Society, Warrendale, P.A. 1995, p. 979.

33.

H. Kraushaar, et al., ibid, p. 1055.

34.

H. R. Muller and R. Jenschar, Z. Metallkunde, Vol. 74, No. 5,1983, p. 257.

35.

H. Yu, Light Metals 1980, C.J. McMinn ed., The Metallurgical Society of
AIME, Warrendale, P.A., 1979, p. 1331.

36.

E. K. Jensen, Light Metals 1984, J.P.McGeer, ed., The Metallurgical Society of

AIME, Warrendale, P.A 1984, p. 1159.

66

37.

D.C. Weckman and P. Niessen, Metals Technology, Nov. 1984, Vol. 11, p. 497.

38.

G. Fortina and F. Gatto, Light Metal Age, Dec. 1978, p. 18.

39.

R. Bachowski and R.E. Spear, Light Metals 1975, The Metallurgical Society of
AIME, Warrendale, P.A., Vol. 2, p. 11.

40.

D. Altenpohl, Aluminium Viewed from Within, Aluminium Verlag, 1982.

41.

D.C.W. Collins, Metallurgica, Oct. 1967, p. 137.

42.

F. Kaempffer and F. Weinberg, Metall. Trans, 1971, Vol. 2, p. 2477.

43.

J.G. McCubbins, Light Metals 1975, The Metallurgical Society of AIME,

Warrendale, P.A., 1975, Vol. 2, p. 137.

44.

N.B. Bryson,

Light Metals 1972, The Metallurgical Society of AIME,

Warrendale, P.A., p. 429.

45.

K. Buxman, Heat and Mass Transfer in Metallurgical Systems, D.B. Spalding

and N.G. Afgan, eds., 1979, McGraw-Hill.

46.

R.F.T. Wilkins, Light Metals 1983, E.M. Adkins, ed., The Metallurgical Society
of AIME, Warrendale, P.A., 1982, p. 907.

67

47.

W.J. Bergman, Aluminium, Vol. 51, No. 5, 1975, p. 337.

48.

K. Lal, et al., Light Metal Age, Oct. 1990, p. 23.

49.

L. Ohm and S. Engler, Metall, 43, No. 6, 1989, p. 520.

50.

H. Woithe and S. Engler, Metall, Vol. 37, No. 4, 1983, p. 332.

51.

R.Elterbok and S. Engler, Metall, Vol. 37, No. 8, 1983, p. 784.

52.

D. G. Harrington and T. E. Groce, US Patent No. 3612151, 1971.

53.

J. F. Grandfield and C.Devadas, Light Metals 1991, E.L. Rooy, ed., The
Minerals, Metals and Materials Society, Warrendale, P.A., 1990, p. 883.

54.

C. Brochu and R. Hank, Light Metals 1993, S. K. Das, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1992, p. 961.

55.

A. Mo, et al., Light Metals 1994, U. Mannweiler, ed., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1994, p. 889.

56.

E. Haug, et al., Int. J. Mass Transfer, Vol. 38, No. 9, 1995, p. 1553.

57.

J.P. Verwijs and D.C. Weckman, Met. Trans B, Vol.19B, 1988, p. 201.

68

58.

D. C. Weckman and P. Niessen, Z. Metallkunde, Vol. 74, No. 11,1983, p. 709.

59.

O. Vorren and S. Brusethaug, International Light Metals Congress LeobenVienna 1987, Austria Metall AG, 1987, p. 278.

60.

J. Grandfield and L. Ruffo, Aluminium Cast House Technology, M. Nilmani,

ed., The Minerals, Metals and Materials Society, Warrendale, P.A., 1995,
p. 345.

61.

G.E. Moritz , US Patent No. 2983972, 1961.

62.

W.J. Bergman, Met. Trans. Vol. 1, 1970, p. 3361.

63.

D.C. Weckman and P. Niessen, Z. Metallkunde, Vol. 75, No. 5, 1984, p. 332.

64.

D.T.T. Auchterlonie, et al., Light Metals 1991, E.L. Rooy, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1991, p. 939.

65.

R. Mitamura and T. Ito, Light Metals 1978, The Metallurgical Society of AIME,
Warrendale, P.A., Warrendale, P.A., 1978, p. 281.

66.

Showa Aluminium, Australian patent 30186/77, 1977.

67.

Showa Aluminium, Australian patent 46379/85, 1985.

69

68.

Showa Aluminium, US patent 4157728, 1979.

69.

Showa Aluminium, US patent 4688624, 1987.

70.

Showa Aluminium, US patent 4653571, 1987.

71.

P. Baker and J.F. Grandfield, Solidification Processing 1987, Institute of Metals,

1987, p. 257.

72.

W. Schneider and E. Lossack, Light Metals 1987, R.D. Zabreznik, ed., The
Metallurgical Society of AIME, Warrendale, P.A., Warrendale, P.A.,1987, p.
763.

73.

K. Buxmannn, et al., Aluminium Technology 1986, Vol. 2, paper 22, Institute of

Metals.

74.

G.M. Apostolou, ibid, paper 21.

75.

Wagstaff, Australian patent AU-A-14137/83, 1983.

76.

J.P. Faunce, et al., Light Metals 1984, J.P. McGeer, ed., The Metallurgical
Society of AIME, Warrendale, P.A., 1984, p. 1145.

77.

Wagstaff, U.S. Patent 4693298, 1987.

70

78.

J. Langerwegar, Proceedings 4th International Aluminum Extrusion Technology

Seminar, V1, The Aluminum Association 1988, p. 381.

79.

M. Furuya and M. Matsuo, Science and Engineering of Light Metals, The

Proceedings of International Conference on Recent Advances in Science and
Engineering of Light Metals, The Japan Institute of Metals, 1991, p. 1065.

80.

R.E. Greene, et al., Light Metals 1989, P. G. Campbell, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1988, p. 859.

81.

F. E. Wagstaff, et al., Light Metals 1993, S. K. Das, ed., The Minerals, Metals
and Materials Society, Warrendale, P.A., 1992, p. 709.

82.

Wagstaff, U.S. Patent 4693298, 1987.

83.

Z. N. Getselev, et al., US patent 34467166, 1969.

84.

M. Yoshida, Technical Reports of Sumitomo Light Metals Industries Ltd., Vol.

28, No. 3, p. 30. 1987

85.

T.R. Pritchett, Light Metal Age, Oct. 1981, p. 12.

86.

D.G. Goodrich, J. Metals, May 1982, p. 45.

71
87.

Y. Krahenbuhl, et al., Light Metals 1990, C.M. Bickert, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1990, p. 893.

88.

J.C. Weber, et al., Light Metals 1988, L.G. Boxall, ed., The Metallurgical
Society of AIME, Warrendale, P.A., 1987, p. 503.

89.

Sautebin and W. Haller, Light Metals 1985, H.O. Bonner, ed., The Metallurgical
Society of AIME, Warrendale, P.A., 1984, p. 1301.

90.

J.W. Evans, et al., ISIA J. Int., Vol. 29, No. 12, 1989, p. 1048.

91.

D. Cook and J.W. Evans, Light Metals 1991, E.L. Rooy, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1991, p. 915.

92.

G. Hudault, et al., Light Metals 1989, P.G. Campbell, ed., The Minerals, Metals
and Materials Society, Warrendale, P.A., p. 769.

93.

Ch. Vives, et al., European patent No. 195793, 1986.

94.

Ch. Vives and B. Forest, Light Metals 1987, R.D. Zabreznik, ed., The
Metallurgical Society of AIME, Warrendale, P.A., 1987, p. 769.

95.

J.P. Riquet and J.L. Meyer, ibid, p. 779.

72
96.

Ch. Vives, Light Metals 1988, L.G. Boxall, ed., The Metallurgical Society of
AIME, Warrendale, P.A., 1987, p. 515.

97.

P. Desnain et al., ibid, p. 487.

98.

N. Bryson, Light Metals 1972, W.C. Rotsell, ed., The Metallurgical Society of
AIME, Warrendale, P.A., 1972, p. 429.

99.

W. Roth, Z. Metallkunde, Vol. 40, No. 12, 1949, p. 415.

100.

L. Katgerman, J. of Metals, Feb. 1982, p. 46.

101.

J. Mathew and H. D. Brody, Proc. Int. Conf. on Solidification, A. Nicholson, ed.,

The Metallurgical Society of AIME, Warrendale, P.A., 1977, p. 244.

102.

M. Nedreberg, Thermal Stress and Hot Tearing During the DC Casting of

AlMgSi Billets, PhD thesis, University of Oslo Department of Physics, 1991.

103.

H.G. Fjaer, et al., Proceedings 5th International Aluminum Extrusion

Technology Seminar, V1, R. I. Werner and V. R. Bird, eds., The Aluminum
Association 1992, p. 113.

104.

S. Nawata, Light Metals 1975, The Metallurgical Society of AIME, Warrendale,

P.A., 1975, p. 161.

73
105.

H. D. Brody, et al., Modelling and Control of Casting and Welding Processes IV,
A.F. Giamei and G.J. Abbaschian, eds., The Metallurgical Society of AIME,
Warrendale, P.A., 1988, p. 351.

106.

S.C. Flood, et al., ibid, p. 553.

107.

L. Katgerman, et al., Production, Refining, Fabrication, and Recycling of Light

Metals, International Symposium, M. Bouchard and P. Tremblay, eds.,
Proceedings Vol. 19 CIM, 1990.

108.

H.G. Fjaer and A. Mo, Met. Trans., 21B, 1990, p. 1049.

109.

H.G. Fjaer and E.K. Jensen, Light Metals 1995, J. Evans, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1995, p. 951.

110.

E.K. Jensen, and W. Schneider, ibid, p. 969.

111.

H.G. Fjaer and A. Mo, Strangiessen, E. Lossack, ed., DGM Germany, 1991,

p.

127.

112.

T. J. Brobak, et al., Light Metals 1991, E. L. Rooy, ed., The Minerals, Metals
and Materials Society, Warrendale, P.A., 1990, p. 869.

113.

H. G. Fjaer and A. Mo, Light Metals 1990, C. M. Bickert, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1990, p. 945.

74

114.

J. Moriceau, Light Metals 1975, The Metallurgical Society of AIME,

Warrendale, P.A., 1975, p. 119.

115.

B. Hannart, et al., Light Metals 1994, U. Mannweiler, ed., The Minerals, Metals
and Materials Society, Warrendale, P.A., 1994, p. 879.

116.

S. Mariaux, et al., Advances in Production and Fabrication of Light Metals and

Metal Matrix Composites, M.M. Avedesian, et al., eds., CIM1992, p. 175.

117.

J-M. Drezet, Modelling of Casting, Welding and Advanced Solidification

Processes VIII, M. Cross and J. Campbell, eds., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1995, p. 197.

118.

W. Schneider and E. K. Jensen, Light Metals 1990, C. M. Bickert, ed., The

Minerals, Metals and Materials Society, Warrendale, P.A., 1990, p. 931.

119.

E. K. Jensen and W. Schneider, ibid, p. 937.

120.

C. Weaver, et al., Light Metals 1991, The Minerals, Metals and Materials
Society, Warrendale, P.A., E.C. Rooy, ed., 1990, p. 953.

121.

W. Droste and W. Schneider, ibid, p. 945.

122.

H. Yu, J. Metals, 1980, November, p. 23.

75

123.

N.B. Bryson, Light Metals 1974, The Metallurgical Society of AIME,

Warrendale, P.A., p. 587.

124.

R.B. Wagstaff and K. D. Bowles, Light Metals 1995, J.W. Evans, ed., The
Minerals, Metals and Materials Society, Warrendale, P.A., 1994, p. 1071.

125.

Alcan, European patent 0372947, June 1990.

126.

S.C. Flood and P. A. Davidson, Mat. Sci. Technol., Aug. 1994, Vol. 10, p. 741.

127.

Ch.Raffourt, et al., Light Metals 1991, E.L. Rooy, ed., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1990, p. 877.

128.

A. Harkonsen and D. Mortensen, Modelling of Casting, Welding and Advanced

Solidification Processes VIII, M. Cross and J. Campbell, eds., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1995, p. 763.

129.

U. Grun, et al., Light Metals 1995, J. Evans, ed., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1994, p. 1061.

130.

C. Brochu, et al., Light Metals 1993, S.K. Das, ed., The Minerals, Metals and
Materials Society, Warrendale, P.A., 1992, p. 961.

76
131.

T.L. Finn, et al., HTD-Vol. 218/AMD, Vol.139, Micro/macro Scale Phenomena

in Solidification ASME, 1992, p. 17.

132.

H. Yu and D. Granger, Physical and Mechanical Properties, Proceedings of the

International Conference on Aluminium Alloys, Charlottesville, VA published
by EMAS, UK 1986, Vol. 1 p. 17.

133.

B. Gariepy and Y. Caron, Light Metals 1991, E.L.Rooy, ed., The Minerals,
Metals and Materials Society, Warrendale, P.A., 1990, p. 961.