International Journal of Thermal Sciences 112 (2017) 211e221

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

journal homepage: www.elsevier.com/locate/ijts

Experimental measurements in melting ingots in the melt of the same

material
 Ditze a, *, Christiane Scharf b
Andre
a
b

Department of Extractive Metallurgy, Clausthal University of Technology, D-38678 Clausthal-Zellerfeld, Germany

Helmholtz Centre Dresden Rossendorf, Helmholtz Institute Freiberg for Resource Technology, Chemnitzer Strasse 40, 09599 Freiberg, Germany

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 19 February 2016
Received in revised form
18 September 2016
Accepted 19 October 2016

This study concerns the melting of ingots of different materials in melt of the same material. We
investigated the pure materials ice, lead, tin, and zinc, the magnesium alloys AZ91 and AM50, and the
aluminum alloy A226. We used melting pots made from steel (for Pb, Sn, Zn, AZ91, AM50) and clay
graphite (for A226) with a volume of 16 L, inserted into a resistance furnace. Some experiments with
AZ91 were also carried out in a 2500 kg industrial furnace. The ice ingots were melted in a 20 L beaker.
The temperature prole adjacent to the melting ingot was recorded over time. From this prole, the
mean temperature of the melt adjacent to the ingots was calculated. Together with the geometrical and
thermophysical properties of the investigated materials, the dimensionless Nusselt, Rayleigh, Prandtl,
and
Stefan
numbers
were
calculated
and
interpreted
as
an
empirical
function,
Nu 0:114RaPr0:291 Ste0:754 . This function describes the melting behavior of all of the materials
considered. This partly agrees with results from the literature, but considerable deviations were also
determined. Once the mean temperature is known, the time needed to melt the different materials in
different geometrical shapes can be estimated along with the maximum melting rate. This simple model
helps understand technical processes where melting of materials is relevant, for example when calculating energy consumption in the foundry industry.
2016 Elsevier Masson SAS. All rights reserved.

Keywords:
Melting
Ice
Metals
Alloys
Heat transfer
Dimensionless numbers
Natural convection
Nusselt
Rayleigh
Prandtl
Stefan

Casting processes require a rapid supply of molten liquid metal

or alloys to the molds. The melting procedure includes the addition
of ingots to an existing melt. Heat transfer and solid-liquid phase
changes occur. Heat has to be transferred from the melt to the ingot.
The transport mechanisms are conduction and convection through
the solid-liquid interface to its surface. Initially, the temperature of
the solid ice, metal, or alloy is raised from the starting temperature
to the melting point or solidus temperature. Subsequently, melting
occurs and heat for fusion has to be added. Finally, the liquid
resulting from the ingot and the surrounding melt has to be heated
up to the initial temperature. The duration of the procedure depends on the materials' properties, ow conditions, and temperature. The following presents a simple model to estimate the time

needed to melt ingots.

Melting and solidication of pure materials and alloys are the
subject of numerous scientic studies. Under dened experimental
conditions, the melting behavior and particularly the interface
shape and motion during melting have been investigated. These
experiments are usually performed in rectangular test cells of
different aspect ratios height/length with the two end walls serving
as the heat source and sink. In such an experimental setup, Gau and
Viskanta [1], Webb and Viskanta [2], and Beckermann and Viskanta
[3], among others, investigated the melting behavior of gallium.
Before melting, the temperature of gallium was held about 1e2  C
below the melting point. They calculated the Rayleigh number (Ra)
and Nusselt number (Nu) using the average heat transfer coefcient
with the height of the cell as characteristic length. The parameter
Nu decreased with the increase in dimensionless time FoSte
1=4

* Corresponding author. Current address: Engineering Ofce, MetuRec, An den

Eschenbacher Teichen 16, 38678 Clausthal-Zellerfeld, Germany.
E-mail addresses: andre.ditze@t-online.de (A. Ditze), c.scharf@hzdr.de
(C. Scharf).

before levelling off at later times, indicating convection-dominated

quasi-steady melting was reached. The characteristic length used
for the Fourier number was the length of the cavity. Using an
average melt layer thickness as the characteristic length, a simple
relation
between
the
dimensionless
numbers

1. Introduction

http://dx.doi.org/10.1016/j.ijthermalsci.2016.10.010
1290-0729/ 2016 Elsevier Masson SAS. All rights reserved.

Ra

212

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

TP
TM
TF
t

Nomenclature
A
Gr
cp
Nu
HF
Pr
g
Ra
lc
Ste
Q
Fo
q
T
TA
TE

surface of ingots
Grashof number
specic heat capacity
Nusselt number
latent heat of fusion
Prandtl number
gravitational acceleration
Rayleigh number
characteristic length
Stefan number
total heat
Fourier number
heat per area and time
temperature
starting temperature of the melt
end temperature of the melt

Greek symbols
heat transfer coefcient
coefcient of volumetric thermal expansion
thermal boundary layer
dynamic viscosity
thermal conductivity
kinematic viscosity
density

a
b
d
h
l
n
r

Subscripts
l
liquid
s
solid

0:274
Ra
Nu 0:0631$ Ste
was obtained, which is valid in the conduction, transition, and convection regimes [1,4]. However, in this
case the characteristic length changes with time. The effect of solid
sub-cooling was also investigated by Beckermann and Viskanta [3],
with temperature differences up to 20  C and 15  C for the solid and
for the liquid from the fusion temperature of gallium, respectively.
The equation Nu 0:5 Ra$Pr1=4 correlated the results of the experiments in the convection dominated region. The study by Jany
and Bejan [5] identies the basic scales and regimes of melting with
natural convection in an enclosure heated from the side. The four
regimes are pure conduction, mixed conduction and convection,
convection, and shrinking solid with increasing time. The study
assumes laminar and two-dimensional liquid ow, negligible
density difference between the solid and liquid phases, a Prandtl
number greater than one (Pr 50 for octadecane), and that all
properties are constant, with the exception of the linear densitytemperature relation. The height of the enclosure was chosen as
the characteristic length. For the convection regime, the Nusselt
number scaling law for Pr > 1 was Nu  Ra1=4 and for Pr < 1 was
Nu  Ra$Pr1=4 . For the shrinking solid region, the heat transfer
and melting rates depend on the size of the remaining solid. The
timescale of the solid vanishing phase is assumed to be relatively
insensitive to the shape of the solid. The case where the solid is at a
considerably lower temperature than the melting temperature was
also mentioned, when the additional effect of conduction in the
solid must be considered. The equation Nu

starting temperature of the ingots

mean temperature of the melt
melting temperature
time

0:35 Ra1=4
,
10:143=Pr9=16 4=9

provided by Lim and Bejan [6], covers the entire Prandtl range. In
nard [7] discuss the case of Pr < 1, which is
particular, Gobin and Be
valid for melting metals by natural convection. A general expression of the time-dependent Nusselt number has been given, which
approaches the constant Nusselt number, Nu 0:29Ra0:27 Pr0:18 ,
at long timescales for convection-dominated quasi-steady melting.
 c and Galovi
Virag, Zivi
c [8] describe the melting of ice of 0  C at wall
temperatures between 2 and 12  C and surrounded by water and
provide values for the dimensionless numbers Nu, Ra, Pr, and Ste.
re
 [10] compare different
Bertrand et al. [9] and Gobin and Le Que
physical models and numerical procedures in the low and high
Prandtl number range. An overview of numerical methods for
solving phase change problems is given by Voller [11]. Advances in

solving conduction and convection melting problems, among

others, are described in scientic papers by Hannoun, Alexiades

and Mai [12], Huber et al. [13], and Kosec and Sarler
[14]. The
complete melting furnace heated by a plasma torch was numeris [15] with a commercial CFD
cally simulated by Carmona and Corte
code, in order to calculate melting times, heat losses and temperature distributions. Numerically predicted melting times, using
different calculation models, were in the range of 625e584 s. The
experimental determined time was 671 s. However, the experimental setup and objective of this investigation is somewhat
different from those in the studies mentioned above. The casting
shop needs to know how much time is needed to melt an ingot of
given mass and shape in a melt of its own material. The temperature of the fed ingots is much lower than the melting point or
liquidus temperature of alloys. The solid-liquid boundary of the
ingot moves in all directions, associated with a decrease in surface
area with time. With some strong simplications, we tried to nd a
valid relationship between the dimensionless numbers Nu, Ra, Pr
and Ste which could describe the relevant parameters to optimize
the melting conditions for the pure metals Pb, Sn, and Zn, the
magnesium alloys AZ91 and AM50, and the aluminum alloy A226.
By measuring the temperature of the melt in the direct vicinity of
the melting ingot together with the thermophysical and geometrical data, such an equation has been developed. As an example of a
material with Pr > 1, ice ingots were also melted in water and
tested. The addition of the solid materials to the melt approximately corresponds to the experimental conditions of the
shrinking solid case described by Jany and Bejan [5], where predominantly natural convection occurs. However, in the present
experiments the additional effect of conduction while heating the
solid ingot to the melting temperature has to be considered, and is
included in the overall Nusselt number. Additionally, in the majority of cases the temperature of the liquid phase is much higher
than the melting temperature. The proposed method is an empirical approximation out of the lack of more detailed measurements
regarding the timely and spatially changes of temperature and heat
ux in the ingots and the melt. Thus, only an overall heat transfer
coefcient could be calculated.
2. Experimental
Melting behavior was investigated for the pure metals lead

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

213

Fig. 1. Experimental setup for magnesium alloy melting. Heating of the crucible took place from the side. A similar setup was used for melting the other metals, but without
protective gas.

(99.95%), tin (99.9%), and zinc (99.995%), the magnesium alloys

AZ91 (9% Al, 1% Zn, 0.2% Mn) and AM50 (5% Al, 0.5% Mn), and the
aluminum alloy A226 (9% Si, 3% Cu, 1% Fe, 1% Zn). The behavior of
ice was also investigated for comparison. These materials are
characterized by different thermophysical properties. The properties of the two magnesium alloys are similar. However, the specic
heat capacity of AM50 is about 15% higher than that of AZ91. The
heat of fusion of the three light metal alloys is rather high, with the
value for the A226 alloy being the highest. Ice has similar properties
to the magnesium alloys. However, the specic heat capacity of
water is extremely high, while thermal conductivity is very low. Its
density exhibits a well-known anomalous relationship with temperature. The heat of fusion, specic heat capacity, and thermal

Fig. 2. Characteristic change in melt temperature while melting an AZ91 ingot. The
process took about 53 s.

conductivity of the low-melting heavy metals lead, tin, and zinc are
much lower than those of the light metal alloys. In particular, the
specic heat capacity and heat of fusion of lead are extremely low.
Laboratory scale experiments were performed for the metals in
an indirectly heated melting furnace (Nabertherm Company) with a
volume of about 16 L, Fig. 1. Some measurements were also performed at an industrial scale for the magnesium alloy AZ91 in an
indirectly heated melting furnace (StrikoWestofen Company) containing about 2500 kg magnesium alloy AZ91. In both cases, steel
crucibles (Pilling Company) heated from the side were used. A clay
graphite crucible inserted in the laboratory furnace was used for
the aluminum alloy only. The melting experiments with ice were
performed in a 20 L beaker using an immersion heater. To measure
the temperature prole during the ingot melting process, two
NiCreNi thermocouples (type K, GTF 1200, 3 mm with AlSi shell)
were used. For the aluminum alloy, the shell consisted of stainless
steel, additionally protected by boron nitride. The temperatures at a
distance of about 5e10 mm from the melting ingot were recorded
every second using two thermocouples and data loggers.
After the predetermined temperature was reached, the ingot
was softly immersed into the melt, the two thermocouples were
installed at a distance of about 5e10 mm from the solid ingot, and
the data loggers were started. The melting range was dened as the
time between the temperature of the melt decreasing and then
increasing again. The melting range is assumed as the experimental
determined melting time. The end of the melting process was also
controlled at random by inserting a steel rod until it could no longer
feel any solids. This time corresponds with the melting range,
which was determined by the temperature prole, as shown in
Fig. 2.
Extra protection was needed for the melt surface in the case of
the magnesium alloys and was provided using a protective gas
composed of 0.05% Novec 612 (C3F7 C(O) C2F5), 7% dry air, and the
remainder carbon dioxide. In the case of the ice experiments the
ingots were immersed in the water of the predetermined temperature. The temperature was measured in a similar way as in the

214

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

Table 1
Ranges of the approximate masses, dimensions and temperatures of the investigated materials. TA starting temperature of the melt, TP starting temperature of the ingot.
Material

Masses in g

Dimensions in mm length  width  height

lead

1300e1475

tin

1095e9510

zinc

3700e8910

AZ91

250e8000

AM50

960e2020

A226

2945e3215

ice

380

119
119
119
292
295
295
112
610
150
295
284
284
295















60
60
60
90
90
90
58
90
62
90
70
71
40















16
18
21
50
20
47
21 e
81
58
43
54
58
35

TA in  C

TP in  C

380e520

20

295e470

20

480e595

20

650e750

20e235

665e770

20

720e800

20e210

30e90

(30)e(5)

Table 2
Nu, Ra, Pr, and Ste numbers from melting experiments with different materials,
compared to Nu numbers, calculated from Equations (8) and (9).
Material [source]

Ga [2]

Ga [3]

Sn [10]
Sn [12]
Octadecane [10]
Ice [8]

Dimensionless number
Nu

Ra

Pr

Ste

Nu
eq. (8)

Nu
eq. (9)

2.26
3.91
7.33
4.43
4.53
4.63
4.28
4.99
2.2
4.1
4.2
19
33
e
4
5
5
6.5
8
8.5

2.20E04
2.20E05
2.20E06
3.28E05
3.28E05
3.28E05
1.67E05
4.88E05
2.50E04
2.50E05
2.25E05
1.00E07
1.00E08
4.98E06
4.22E06
5.76E06
5.91E06
5.68E06
1.30E07
2.28E07

0.021
0.021
0.021
0.028
0.028
0.028
0.028
0.028
0.02
0.02
0.04
50
50
10.71
12.96
12.56
12.06
11.58
11.18
10.89

0.04
0.04
0.04
0.05
0.05
0.05
0.03
0.08
0.01
0.01
0.01
0.10
0.10
0.10
0.03
0.05
0.08
0.10
0.13
0.15

2.34
4.41
8.28
4.63
4.63
4.63
4.63
4.63
3.57
6.72
6.52
9.82
18.45
8.11
11.30
10.17
9.17
8.38
9.88
10.97

2.32
4.12
7.33
4.80
4.80
4.80
4.06
5.30
2.37
4.22
4.65
19.37
34.44
15.92
15.34
16.56
16.66
16.48
20.24
23.30

metal melting experiments. The end of melting was detected by eye

until the solid disappeared. The ice ingots oat on the water surface. However, after the melting process starts a liquid lm could be
observed at the part of the ingot not completely immersed. Table 1
summarizes the ranges of the masses, dimensions, and temperatures of the investigated materials.
For most of the experiments, heat was transferred to the ingots
from the electrical heating elements of the furnace (Fig. 1) via the
wall of the steel crucible and from there to the melt and nally the
ingots. For the ice melting experiments, an immersion heater
directly heated the water. In some experiments with AZ91, the melt
inside the steel crucible was additionally heated by two immersion
heaters, length 300 mm, heated length 150 mm, diameter 20 mm,
voltage 230 V and power 1800 W.
Further experiments with the magnesium alloy AZ91 were
conducted with additional forced convection provided using a steel
stirrer, outer diameter 90 mm and rotational speed 190 rpm. The
results of these experiments did not deviate from those without
stirring and indirect melt heating. They are marked separately in
Fig. 6 in the following section. There is only negligible contactmelting of the metals assumed, because the bottom of the crucible is not heated, although some heat will be transferred to the
bottom by conduction. However, some part of the ingot are in

Fig. 3. Model of melting an ingot in the melt of its own material.

contact with the crucible bottom and is therefore not circulated

completely around by the melt. A few experiments were conducted
using a melting basket mounted inside the crucible (AZ91 experiments 7e13 from Table 2a) to exclude bottom contact. The results
did not deviate from those without basket. On the other hand in
course of the ice melting experiments also some part of the ice
ingots is not fully surrounded by the water. These conditions tend
to result in uncertainties of the measurements.
3. Modelling the ingot melting process
To our knowledge, no numerical solution is currently available
to the differential equations describing complete conduction and
convection heat transfer including phase change and change in
ingot shape. Therefore, we have tried to describe the melting
behavior via dimensionless numbers. The ingots were added to the
melt of a dened temperature, TA. The ingot of temperature TP heats
up via conduction to its melting temperature, TF, and melting starts.
The melting energy is provided from the surrounding melt mainly
by convection. The heat has to be transferred through the thermal
boundary layer, d, surrounding the solid material. During the
melting period, the temperature decreases from TA to TE. By integrating the temperature-time function, schematically shown in
Fig. 2, a mean temperature TM can be calculated, assumed to be the
most important parameter for determining the melting time of the
materials. TM was chosen as the characteristic temperature to

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

evaluate the dimensionless numbers for the melting process. To

calculate the Grashof number the temperature difference of TM and
the mean temperature of the solid material T(TF TP)/2 is
responsible. For the Stefan number it is the difference of TM and the
melting temperature of the materials TF. The simple model is
shown in Fig. 3.
In general, conduction and convection were present in all experiments. However, the phase change heat transfer should be
mainly convection dominated [5]. Therefore, the relevant dimensionless numbers for assessing melting behavior in an existing melt
are assumed to be the overall Nusselt number as a function of the
Prandtl number (material properties), the Grashof number (thermal convection), or the Rayleigh number (Ra Gr$Pr) and the
Stefan number (phase change). To calculate these dimensionless
numbers, the physical properties l, h, cp, b, r, and HF of the materials were taken from the literature. Depending on the experiment,
we also considered the temperature dependency of the physical
properties. For the alloy materials, the liquidus temperature at the
respective composition was used as the melting temperature
(600  C for AZ91, 625  C for AM50, and 600  C for A226). The geometry of the system is described by the characteristic length, lc.
This can be the height or length of the enclosure or an average
thickness of the liquid zone, as described by Jany and Bejan [5]. In
the present experimental setup, the dimensions of the ingots e
especially the surface area e are important for the melting
behavior. Taking this into account, the diameter of the sphere with
the same surface area as that of the ingot was chosen as lc. This
provides the opportunity to compare different geometrical shapes
of the materials, assuming the surface of the ingots is one of the
determining factor for the melting process. Because of the vanishing solid phase with time, associated with decreasing surface
area and decreasing characteristic length, this assumption is of
course a strong simplication. The heat transfer coefcient a was
estimated from the total heat needed to melt the ingots.



Q m HF cps TF  TP cpl TM  TF

(1)

The heat related to the surface area and melting time of the
ingot is

Q
At

(2)

Dividing by the temperature difference between the mean

temperature during the melting period and the starting temperature of the ingot, the overall heat transfer coefcient is estimated
via:

q
TM  TP

(3)

Using the characteristic length, lc, and the thermal conductivity,

l, of the melt, the overall Nusselt number was calculated:

Nu

a,lc
l

(4)

Using the dynamic viscosity, h, the thermal conductivity, l, and

the specic heat capacity, cpl , of the melt, the Prandtl number of the
materials is:

Pr

h,cpl
l

(5)

The Grashof number was calculated from the coefcient of

thermal expansion, the dynamic viscosity, and the density of the
melt. The difference between the mean temperature of the melt
during the melting period, TM, and the solid material T

TF Tp
2

was

215

selected as the temperature difference.

Gr

g$b$l3c $TM  T$92

h2

(6)

The phase change was considered using the Stefan number

Ste

cpl $TM  TF
HF

(7)

4. Experimental results of the ingot melting process

The literature reports numerous scaling laws for Nu f(Ra, Pr,
Ste), correlating the heat transfer data for melting in the presence of
conduction and convection [1,4e7,9,10,13]. Despite the use of
different length scales and also different experimental setups of the
literature, a comparison with the own measured data is attempted
to check the application of the reported scaling laws. Viskanta [4]
correlated the experimental data for melting gallium (Pr < 1)
with the equation:

Nu 0:0631

Ra
Ste

0:274
(8)

Lim and Bejan [6] provide an equation for the entire Prandtl
range:

0:35 Ra1=4
Nu h
i4=9
1 0:143=Pr9=16

(9)

Equation (9) does not include the Stefan number. However,

datasets of Nu, Ra, Pr, and Ste from melting experiments with
different materials from the literature are known (see Table 2).
Fig. 4 depicts the authors' experimental values and literature
data from Table 2 for the Nusselt number, dependent on the ratio
Ra/Ste.
Equations (8) and (9) both cover the literature results from
Table 2 regarding the investigated metals with Pr < 1. Comparing
the Nu values for octadecane (Pr 50) and ice (Pr 11e13) with
Equation (8) reveals an increasing deviation with increasing Ra/Ste
ratios. Inserting the Ra and Pr values of our own experiments for
metals and ice into Equation (9) to calculate the Nu numbers results
in a relatively good agreement with the Nu f(Ra/Ste) function
(black crosses), while Equation (8) again approximately covers the
values of the metallic materials. However, the plot of our own Nu
data against our own Ra/Ste ratios is comparable to that of Equation
(9) for ice. Obviously, the slope for the ice values (dotted line in
Fig. 4) is also slightly different. Even though Equation (9) seems to
generally meet the Ra/Ste dependence of our own experimental
values, the calculated Nu numbers are much higher than the Nu
numbers of the metallic materials determined from our own experiments. In Fig. 4, a high scatter of the Nu-Ra/Ste dots for the
measured metals data is clear. This is primarily due to the high
temperature difference (TF-TP) between the cold solid metallic
materials and their melting temperatures. Heating up the solid
materials to their melting temperature is not included in the range
of validity of Equations (8) and (9). The relatively good agreement
for the ice values is presumably due to the small difference in TF-TP,
with a maximum of 36  C. In addition, the literature uses different
characteristic lengths e most commonly the height of an enclosure
e in calculations. Based on these considerations, the Nu f(Ra/Ste)
function cannot cover the experimental datasets of all the investigated materials. Because correlating values in the Pr < 1 region

216

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

Fig. 4. Nu numbers via Ra/Ste from the authors' experiments compared to calculated Nu numbers (from authors' experiments and literature data from Table 2) as per Equations (8)
and (9).

Fig. 5. Overall Nu numbers plotted against (Ra$Pr)0.291$Ste0.754 for the melting of

metallic materials and ice ingots compared to data from the literature.

seems especially difcult, we assume convection-dominated heat

transfer in the Pr < 1 region with a Nusselt dependency from the Ra,
Pr, and Ste numbers in the form of Nu C$(Ra$Pr)x$Stez. Plotting
the Nu numbers via (Ra$Pr)x$Stez should produce a linear function
which ts the experimental values. Fig. 5 shows a clear common
trend in the experimental results for the different materials. The
line of best t for all of the investigated materials is represented by
equation

Nu 0:114RaPr0:291 Ste0:754

Fig. 6. Overall Nu numbers plotted against (Ra$Pr)0.291$Ste0.754 for the melting of pure
metals and alloys.

(10)

with the condence level of R2 0.99.

The same explanation discussed for Fig. 4 explains the deviation

from the literature values in Fig. 5. The data for ice from Virag et al.
[8] t quite well with our own experimental values, even though

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

217

Fig. 7. Melting times of AZ91 ingots with a mean melt temperature TM. White circles:
TP 20  C, 180 mm  90 mm  45 mm. White squares: TP 200  C,
180 mm  90 mm  45 mm. Black triangles: TP 20  C, 360 mm  180 mm  90 mm.

Fig. 9. Melting times of Sn (black triangles), Pb (white circles), and Zn (white squares)
ingots of dimensions 180 mm  90 mm  45 mm, with a mean melt temperature TM
and TP 20  C.

they used a different characteristic length (height of enclosure). The

data points for our own experiments are in good agreement with
the relationship in Equation (10). Nevertheless, there is some residual scatter, induced by the relatively rough method of determining the temperatures and the melting time and the simplifying
assumptions. This is obvious from Fig. 6, which depicts pure metals
and alloys separately.
The results from additional stirring of the melt (AZ91) show no
signicant deviation from the other results. This conrms that the
mean temperature TM at the thermal boundary is the determining
parameter for the melting time of ingots.

of the melt TM can be maintained at 680  C. Furthermore, some

interesting conclusions can be drawn. Melting magnesium alloy
AZ91 ingots pre-warmed to 200  C (TP) takes more time (41 s and
Nu 1.95) than melting ingots of the same size with an initial
temperature of 20  C at the same mean temperature TM (see Fig. 7).
At rst sight it was surprising, that melting of the materials can be
slower with higher pre-warm temperature. Two cases can be
considered and differentiated. Melting of material at higher prewarm temperature in a melt with the same starting temperature
results in a higher mean temperature TM and melting can and will
be faster than melting of cold material which results in lower TM. If
melting times are compared at constant TM, the calculated melting
time is longer in the case of the pre-warmed material. This can be
explained by the driving force of the melting process. The specic
heat ow from the melt to the solid is q a (TM-TP). With smaller
DT the heat ow is also smaller. If theoretical the temperature TP is
close to TM the heat ow approximates to zero and the time will
increase to very long times. At the melting point of the material, i.e.
equilibrium between solid and liquid, the Gibbs free energy of
melting is zero and there will be no further melting. The trend in
Fig. 7 supports the explanation. The difference in time decreases

5. Conclusions for practical application

Given a specic ingot geometry, initial temperature, TP, of the
ingot, and a nearby temperature TM, the dimensionless numbers Ra,
Pr, and Ste can be calculated. Next, the Nu number can be calculated
using Equation (10). Using the equations above, the time needed to
melt the ingots can be determined by:

Q lc
ANul$TM  TP

(11)

As an example, an ingot of magnesium alloy AZ91 with dimensions 180 mm  90 mm  45 mm with an initial temperature
of 20  C can be melted in 32 s (Nu 2.13) if the mean temperature

Fig. 8. Melting times of AZ91 ingots (white circles), AM50 ingots (white squares) and
A226 ingots (black triangles), TP 20  C, ingot size 180 mm  90 mm  45 mm with
increasing mean temperature TM of the melt.

Fig. 10. Maximum melting rates of AZ91 ingots with mean temperature TM. White
circles: mean TP 20  C, ingot dimensions 610 mm  90 mm  80 mm, ingot
 C,
weight

8
kg.
Black
circles:
mean
TP

20
ingot
dimensions 620 mm  110 mm  100 mm, ingot weight 12 kg. Industrial melting
 C,
rates:
White
square:
mean
TP

25
ingot
dimensions 630 mm  120 mm  65 mm, ingot weight 8.9 kg. White triangle:
mean TP 150  C, ingot dimensions 630 mm  120 mm  65 mm, ingot
weight 8.9 kg. White rhombus: mean TP 150  C plus converter, ingot
dimensions 630 mm  120 mm  65 mm, ingot weight 8.9 kg.

218

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

with increasing TM because the inuence of TP is lower in (TM-TP).

We tried to verify this phenomenon by evaluation of the experiments with pre-warmed ingots. However, because the focus was
not on this, there are only few experimental setups which are
comparable in mass, characteristically length and TM. Some of them
show the predicted results, other do not. Also the experimental
uncertainty must be taken into account which may be in the range
up to 10% regarding the measured times. Fig. 7 also includes the
melting behavior of an AZ91 ingot with dimensions
360 mm  180 mm  90 mm. As expected, the melting times are
higher than for the ingots with smaller dimensions.
There is sometimes a discussion over the various merits of
magnesium or aluminum alloys as materials, for example in vehicle
gearboxes. Beside the differences in density, strength, and cost of
materials, the melting behavior can also be considered. Magnesium
alloy AZ91 melts faster than aluminum alloy A226 (see Fig. 8).
Even AM50 e with its higher liquidus temperature of 625  C
compared to 600  C for A226 e melts quicker. This an added value
in favor of magnesium alloys. Comparing the melting times of the
heavy metals Sn, Pb, and Zn (Fig. 9) it can be seen that the times
increase in the same order due to their different thermophysical
properties.
The melting rates of the melting furnaces are also important for
the casting shop. After melting the ingot, the time for which can be
calculated using Equation (10), the melt has to be heated up from TE
to TA. Using the calculated melting time for an ingot, a maximum
melting rate can be calculated. This is the melting rate without
considering the time needed to heat the melt to the starting melt
temperature. Fig. 10 shows the maximum melting rates for AZ91
ingots compared to the real melting rates achieved in industrial
practice.

The industrial melting rates hold for a melt temperature of

TA 685  C. The mean temperatures TM are estimated to be 669,
675 and 680  C for the operation modes TP 25  C, TP 150  C and
TP 150  C plus converter, respectively. The converter is an
apparatus inside the melt which generates forced convection for
the charged ingot [16]. As expected, the melting rates of the industrial furnaces increase with the use of pre-warming and converter because of the increasing mean temperatures TM. As
mentioned before, the melting rates are lower than the calculated
maximum melting rates which do not consider the time needed to
heat up the melt to the starting temperature.
6. Conclusions
We investigated the melting behavior of materials with different
thermophysical properties. Ingots of the magnesium alloys AZ91
and AM50, the aluminum alloy A226, pure metals such as lead, tin,
and zinc, and nally ice were melted in their own melts. The
dimensionless Nusselt, Rayleigh, Prandtl, and Stefan numbers
produced the equation Nu 0:114RaPr0:291 Ste0:754 for the heat
transfer from the surrounding melt to the ingots for all tested
materials. We were able to calculate the times needed for melting,
which only depend on the mean temperature near to the ingot, the
pre-warm temperature, and the dimensions of the ingot. Experiments with direct and indirect heating of the melt showed that only
the mean temperature near the ingot is relevant.
Appendix

Table 1a
Experimental data for the melting of ice ingots of different dimensions, L length, W width, H height. m mass, n number of ingots, t melting time, n.d. not
determined.
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

L
mm

W
mm

H
mm

m
g

293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
193

41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41

36
36
35
37
36
36
36
36
34
34
34
34
34
33
34
34
34
34
34
33
34
34
34
33
33
33
34
32
33
34
34
34
18

396
393
389
409
803
805
796
791
371
380
380
378
374
363
373
377
381
374
370
367
371
374
374
362
364
366
372
353
364
372
371
372
130

1
1
1
1
2
2
2
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

TA
C

TP

C

TM
C

Pr

Ra

Ste

t
s

Nu

n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
61
61
61
84
84
84
32
32
32
32
50
50
50
71
71
71
91
91
84
84
84
61
61
61
31

10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
20
20
20
20
20
20
20

46
48
67
84
44
49
68
83
61
61
61
83
83
83
31
31
31
31
49
49
49
70
70
70
90
89
83
83
83
61
61
61
31

216
140
90
72
229
180
118
92
125
130
142
68
65
68
385
330
325
363
177
182
176
93
102
84
61
63
80
64
73
135
126
116
263

73
109
140
162
141
170
215
250
102
100
92
163
169
158
48
57
58
52
81
78
82
128
117
139
170
167
127
153
137
86
91
100
29

3.9
3.7
2.5
2.1
4.0
3.6
2.4
2.1
2.8
2.8
2.8
2.1
2.1
2.1
5.5
5.5
5.5
5.5
3.6
3.6
3.6
2.4
2.4
2.4
2.1
2.1
2.1
2.1
2.1
2.8
2.8
2.8
5.5

4.3E09
4.8E09
1.2E10
1.9E10
3.2E10
4.2E10
9.9E10
1.5E11
8.7E09
8.8E09
8.8E09
1.8E10
1.8E10
1.7E10
1.6E09
1.6E09
1.6E09
1.6E09
5.0E09
4.9E09
5.0E09
1.3E10
1.3E10
1.2E10
1.9E10
1.9E10
1.9E10
1.8E10
1.8E10
9.6E09
9.6E09
9.6E09
6.8E08

0.6
0.6
0.8
1.1
0.6
0.6
0.9
1.0
0.8
0.8
0.8
1.0
1.0
1.0
0.4
0.4
0.4
0.4
0.6
0.6
0.6
0.9
0.9
0.9
1.1
1.1
1.0
1.0
1.0
0.8
0.8
0.8
0.4

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

219

Table 1a (continued )
No.
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

L
mm

W
mm

H
mm

m
g

293
293
293
193
100
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
293
198
95
293
293
293
293
293

41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41
41

17
35
34
35
33
35
33
34
34
32
31
34
35
34
34
33
33
33
33
34
33
33
33
34
34
33
33
33
33
34
33
32
33
33

191
387
380
252
123
381
365
376
370
356
347
372
384
375
373
369
369
362
369
376
369
368
368
375
373
364
366
250
119
380
369
359
369
360

1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

TA
C

TP

C

TM
C

Pr

Ra

Ste

t
s

Nu

32
32
50
50
50
50
72
72
72
93
93
93
61
61
84
84
84
41
41
41
74
61
61
61
84
84
84
42
42
42
73
73
93
93

20
20
20
20
20
20
20
20
20
20
20
20
36
36
36
36
36
36
36
36
36
5
5
5
5
5
5
5
5
5
5
5
5
5

32
32
50
50
50
50
71
70
71
91
91
91
61
61
83
83
83
40
40
40
73
62
62
62
83
83
83
41
42
41
71
72
92
92

229
384
186
206
153
166
102
106
108
70
68
63
105
120
75
75
67
202
246
253
90
105
105
110
64
65
67
209
172
219
85
79
55
53

39
41
69
50
45
77
105
104
100
136
137
156
99
85
123
122
136
56
47
46
105
126
126
122
179
172
168
67
53
80
145
151
196
200

5.4
5.4
3.5
3.5
3.5
3.5
2.3
2.4
2.3
2.2
2.2
2.2
2.8
2.8
2.1
2.1
2.1
4.4
4.4
4.4
2.3
2.7
2.7
2.7
2.1
2.1
2.1
4.3
4.2
4.3
2.3
2.3
2.2
2.2

1.3E09
2.0E09
5.8E09
3.3E09
1.3E09
5.8E09
1.4E10
1.4E10
1.4E10
2.0E10
2.0E10
2.1E10
1.1E10
1.0E10
2.0E10
2.0E10
2.0E10
3.7E09
3.8E09
3.8E09
1.6E10
8.9E09
8.9E09
9.0E09
1.7E10
1.7E10
1.7E10
1.7E09
6.9E08
3.0E09
1.3E10
1.3E10
1.9E10
1.9E10

0.4
0.4
0.6
0.6
0.6
0.6
0.9
0.9
0.9
1.1
1.1
1.1
0.8
0.8
1.0
1.0
1.0
0.5
0.5
0.5
0.9
0.8
0.8
0.8
1.0
1.0
1.0
0.5
0.5
0.5
0.9
0.9
1.2
1.2

Table 2a
Experimental data for the melting of magnesium alloy AZ91 ingots of different dimensions, L length, W width, H height. m mass of ingot, n number of ingots,
t melting time. * Additional heat input via two immersion heaters. ** Additional stirring.
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

L
mm

W
mm

H
mm

m
g

295
195
610
610
610
610
295
190
292
292
295
212
210
112
145
145
118
118
282
282
195
195
195

87
65
90
90
90
90
87
61
40
40
88
63
63
57
40
40
57
58
40
40
64
64
66

50
58
81
81
81
81
50
66
35
35
41
61
60
22
34
34
22
22
35
35
61
60
63

2303
1332
8000
8000
8000
8000
2296
1389
742
738
1938
1477
1432
253
359
355
529
543
705
715
1383
1360
1477

1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
1
1
1
1
1

TA
C

TP

C

TM
C

Pr

Ra

Ste

t
s

Nu

705
694
683
672
664
649
695
702
704
712
706
710
708
680
680
680
680
680
721
723
725
731
725

20
20
20
20
20
20
20
20
20
235
235
235
235
235
235
235
235
235
20
20
20
20
20

662
667
665
659
648
633
647
663
689
701
674
680
674
673
668
665
654
664
704*
700*
693*
689*
671

54
49
74
89
83
92
57
46
22
24
50
42
28
27
28
30
27
32
19
18
36
38
42

1.82
1.46
2.92
2.44
2.65
2.44
1.76
1.62
1.93
2.05
2.02
2.13
3.16
0.99
1.25
1.16
2.10
1.78
2.13
2.28
1.98
1.89
1.83

0.022
0.021
0.021
0.022
0.023
0.025
0.023
0.022
0.019
0.019
0.021
0.020
0.021
0.021
0.021
0.021
0.022
0.021
0.019
0.019
0.019
0.019
0.021

7.7E07
3.9E07
3,1E08
3.0E08
2.8E08
2.6E08
7.1E07
3.9E07
3.3E07
2.5E07
5.2E07
3.3E07
3.1E07
6.1E06
7.8E06
7.6E06
4.6E07
5.1E07
3.2E07
3.2E07
4.4E07
4.3E07
4.3E07

0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.2
0.3
0.3
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.2

(continued on next page)

220

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

Table 2a (continued )
No.
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

L
mm

W
mm

H
mm

m
g

195
195
195
195
195
195
195
195
195
195
195
195
189
189
91
89

63
64
64
64
64
64
64
64
63
63
63
63
65
65
64
61

60
62
60
56
60
60
59
61
56
59
55
60
41
46
41
43

1328
1395
1356
1261
1355
1352
1331
1386
1242
1312
1230
1327
917
1009
428
428

1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

TA
C

TP

C

TM

C

t
s

Nu

Pr

Ra

Ste

727
726
682
680
720
727
682
678
725
724
681
681
699
701
699
652

20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20

675
658
642
639
679*
689*
654*
648*
698**
696**
657**
655**
682**
680**
691**
644**

40
31
57
59
41
34
41
64
29
29
43
43
26
28
18
37

1.77
2.42
1.32
1.21
1.75
2.08
1.78
1.18
2.31
2.40
1.59
1.71
2.06
2.06
1.94
0.98

0.020
0.022
0.024
0.024
0.020
0.019
0.022
0.023
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02

4.0E07
3.8E07
3.4E07
3.2E07
4.1E07
4.3E07
3.6E07
3.6E07
4.1E07
4.3E07
3.4E07
3.6E07
3.1E07
3.2E07
1.2E07
9.7E06

0.2
0.2
0.1
0.1
0.2
0.3
0.2
0.1
0,3
0,3
0,2
0,2
0,3
0,2
0,3
0.1

Table 3a
Experimental data for the melting of magnesium alloy AM50 ingots of different dimensions, L length, W width, H height. m mass of ingot, n number of ingots,
t melting time.
No.
1
2
3
4

L
mm

W
mm

H
mm

m
g

150
151
295
295

61
62
90
90

59
58
43
42

965
973
2019
1975

1
1
1
1

TA

C

TP

C

TM
C

t
s

Nu

Pr

Ra

Ste

766
740
772
732

20
20
20
20

733
701
686
672

34
49
39
62

1.65
1.18
2.23
1.38

0.018
0.020
0.020
0.021

4.3E07
3.8E07
9.9E07
9.1E07

0.40
0.27
0.22
0.17

Table 4a
Experimental data for the melting of lead ingots of different dimensions, L length, W width, H height. m mass of ingot, n number of ingots, t melting time.
No.
1
2
3
4
5
6
7
8

L
mm

W
mm

H
mm

m
g

119
119
119
119
119
119
119
119

60
60
60
60
60
60
60
60

18
16
17
18
17
17
17
18

1442
1297
1343
1474
1395
1337
1353
1411

1
1
1
1
1
1
1
1

TA
C

TP
C

t
s

Pr

Ra

Ste

TM

C

Nu

514
467
438
411
380
520
497
469

20
20
20
20
20
20
20
20

507
461
428
401
376
509
482
456

9
12
24
22
29
9
9
13

6.67
4.82
2.64
3.16
2.41
6.24
6.50
5.00

0.015
0.017
0.019
0.020
0.021
0.015
0.016
0.017

9.7E07
7.5E07
6.5E07
5.9E07
5.0E07
9.5E07
8.4E07
7.6E07

1.1
0.8
0.6
0.5
0.3
1.1
0.9
0.8

Table 5a
Experimental data for the melting of tin ingots of different dimensions, L length, W width, H height. m mass of ingot, n number of ingots, t melting time.
No.
1
2
3
4
5
6

L
mm

W
mm

H
mm

m
g

119
292
292
292
292
292

60
90
90
90
90
90

21
49
47
49
50
39

1094
9302
8902
9430
9511
7549

1
1
1
1
1
1

TA

C

TP

C

TM
C

t
s

Nu

Pr

Ra

Ste

294
419
336
470
336
302

20
20
20
20
20
20

290
397
300
410
285
261

30
21
68
22
67
116

2.04
9.90
3.54
9.60
3.88
1.98

0.013
0.010
0.013
0.010
0.013
0.014

1.7E07
2.5E08
1.5E08
2.7E08
1.4E08
1.0E08

0.2
0.7
0.3
0.7
0.2
0.1

A. Ditze, C. Scharf / International Journal of Thermal Sciences 112 (2017) 211e221

221

Table 6a
Experimental data for the melting of zinc ingots of different dimensions, L length, W width, H height. m mass of ingot, n number of ingots, t melting time.
No.
1
2
3
4
5
6
7
8

L
mm

W
mm

H
mm

m
g

295
295
295
295
295
295
295
295

90
90
90
90
90
90
90
90

40
35
31
22
21
47
45
20

7491
6716
5930
4178
3890
8908
8548
3699

1
1
1
1
1
1
1
1

TA
C

TP
C

t
s

Pr

Ra

Ste

TM

C

Nu

593
559
529
498
479
564
523
481

20
20
20
20
20
20
20
20

518
487
471
461
446
494
465
456

30
42
53
43
81
50
92
66

5.61
3.90
2.85
2.62
1.35
4.07
2.23
1.56

0.025
0.028
0.030
0.031
0.032
0.027
0.030
0.031

2.6E08
2.1E08
1.9E08
1.5E08
1.4E08
2.6E08
2.2E08
1.4E08

0.5
0.3
0.2
0.2
0.1
0.3
0.2
0.2

Table 7a
Experimental data for the melting of the aluminum alloy A226 ingots of different dimensions, L length, W width, H height. m mass of ingot, n number of ingots,
t melting time.
No.
1
2
3
4
5

L
mm

W
mm

H
mm

m
g

284
284
284
284
284

71
71
70
71
70

57
58
54
56
55

3168
3217
2943
3126
2998

1
1
1
1
1

TA

C

TP
C

TM
C

Pr

Ra

Ste

t
s

Nu

798
772
749
795
749

20
20
20
210
210

648
621
623
655
630

70
81
92
60
78

1.39
1.23
1.01
1.91
1.48

0.011
0.011
0.011
0.011
0.011

8.8E07
8.2E07
7.7E07
6.4E07
5.6E07

0.10
0.04
0.05
0.12
0.06

References
[1] Gau C, Viskanta R. Melting and solidication of a pure metal on a vertical wall.
ASME J Heat Transf 1986;108:174e81.
[2] Webb BW, Viskanta R. Analysis of heat transfer during melting of a pure metal
from an isothermal vertical wall. Numer Heat Transf 1986;9:539e58.
[3] Beckermann C, Viskanta R. Effect of solid subcooling on natural convection
melting of a pure metal. ASME J Heat Transf 1989;111:416e24.
[4] Viskanta R. Heat Transfer during melting and solidication of metals. ASME J
Heat Transf 1988;110:1205e19.
[5] Jany P, Bejan A. Scaling theory of melting with natural convection in an
enclosure. Int J Heat Mass Transf 1988;31:1221e35.
[6] Lim JS, Bejan A. The Prandtl number effect on melting dominated by natural
convection. ASME J Heat Transf 1992;114:784e7.
nard C. Melting of metals driven by natural convection in the melt:
[7] Gobin D, Be
inuence of Prandtl and Rayleigh numbers. ASME J Heat Transf 1992;114:
521e4.
 
[8] Virag Z, Zivi
c M, Galovi
c A. Inuence of natural convection on the melting of
ice block surrounded by water on all sides. Int. J Heat Mass Transf 2006;49:
4106e15.

[9] Bertrand O, et al. Melting driven by natural convection. A comparison exercise: rst results. Int J Therm Sci 1999;38:5e26.
re
 P. Melting from an isothermal vertical wall. Synthesis of a
[10] Gobin D, Le Que
numerical comparison exercise. Comput Assist Mech Eng Sci 2000;7(3):
289e306.
[11] Voller VR. An overview of numerical methods for solving phase change
problem. In: Minkowycz WJ, Sparrow E, editors. Advances in numerical heat
transfer, vol. 1. Washington D. C.: Taylor & Francis; 1997. p. 340e80.
[12] Hannoun n, Alexiades v, mai tz. Resolving the controversy over tin and gallium melting in a rectangular cavity heated from the side. Numer Heat Transf
Part B 2003;44:253e76.
[13] Huber C, et al. Lattice Boltzmann model for melting with natural convection.
Int. J Heat Fluid Flow 2008;29:1469e80.

[14] Kosec G, Sarler
B. Numerical solution of natural convection problems by a
meshless method. In: Ahsan A, editor. Convection and conduction heat
transfer; 2011. p. 107e32. ISBN: 978-953-307-582-2.
s C. Numerical simulation of a secondary aluminum melting
[15] Carmona M, Corte
furnace heated by a plasma torch. J Mater Process Technol 2014;214:334e46.
[16] Ditze A, Scharf C. Recycling of magnesium. Clausthal-Zellerfeld, Germany:
Papierieger-Verlag; 2008. p. 94. ISBN 978-3-89720-957-2.