https://doi.org/10.

1007/s41230-020-0039-x Special Review CHINA FOUNDRY

Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

90 years of thermal analysis as a control tool in the

melting of cast iron
*Doru Michael Stefanescu 1, Ramon Suarez 2,3, and Sung Bin Kim 4
1. The Ohio State University, Columbus, OH and The University of Alabama, Tuscaloosa, AL, USA
2. IK4-Azterlan, Durango (Bizkaia), Spain
3. Veigalan Estudio 2010, Durango (Bizkaia), Spain
4. AnyCasting Software Co., Ltd, Seoul, S. Korea

Abstract: Since its first literature mention in conjunction with cast

iron in 1931 by Esser and Lautenbusch, thermal analysis (TA) has
journeyed a long way. Today it is an accepted and widely used tool for
process control for all types of cast irons. This paper reviews the latest
progress in the development of equipment and analysis methods that
make TA successful in applications such as the estimation of chemical
composition, graphitization potential, and the shape and number of
graphite aggregates. The potential and limitations of the prediction of
shrinkage defects propensity are analyzed in some details. Examples
of attempts at prediction of mechanical properties and shrinkage
propensity are also discussed. Several graphs showing the data
scattering are presented to convey the reader a better sense of the
accuracy of various predictions.

Key words: thermal analysis; cast iron; graphite; shrinkage

prediction; chemical composition
*Doru Michael Stefanescu
Professor Stefanescu, a graduate of the University CLC numbers: TG143; Document code: A;
Politehnica Bucharest, Romania, served as a Visiting
Professor at the Univ. of Wisconsin - Madison (1980), and Article ID: 1672-6421(2020)02-069-016
then joined the Univ. of Alabama (UA) where he taught and

W
did research until 2005. In August of 2005, he joined the
Material Science and Engineering Department at the Ohio
hile the first temperature-time record of heating curves appears to
State Univ. (OSU) until August 2010. Prof. Stefanescu’s have been reported as early as 1877 [1], it was not until 1931 that
research interests include experimental and numerical the method was applied to cast iron analysis by Esser and Lautenbusch,
aspects of solidification processing, influence of low-gravity
on solidification, processing of metal-matrix composites,
cited by Piwowarski [2]. They reported that higher superheating of gray iron
manufacturing technologies and physical metallurgy of cast depresses the temperature of eutectic arrest. Other qualitative observations
iron, steel and nonferrous alloys, etc. The research area in include those by Piwowarski [3] who noticed that higher silicon decreases
which his accomplishments have probably been most widely
the undercooling, by Loper et al. [4] who reported the effect of the Mn/S
recognized nationally and internationally is computational
modeling of microstructure evolution during solidification ratio on the eutectic undercooling, by Naro and Wallace [5] who revealed
of castings. Another area of significant contribution is that the effect of Ce and S additions, and by De Sy and Vidts [6] who pointed out
of particles' behavior at the liquid/solid interface. Professor that cooling curves contain information on graphite morphology (shape).
Stefanescu is recognized as a world expert in cast iron. He
is the author of over 447 scientific and technical publications.
In time, thermal analysis (TA), the technique of recording and
Prof. Stefanescu is the editor of ASM Handbook Vol. 1A Cast interpretation of cooling curves, became an important tool in on-
Iron Science and Technology, ASM International (2017). He line monitoring of melt quality, a tool that proved to be of paramount
is a Dr. Honoris Causa of three universities and an Honorary
Member of the Romanian Academy of Technical Sciences.
importance for quality assurance of castings. While initial applications of
Prof. Stefanescu awards include the Gold Medal and termed TA revolved around the rapid evaluation of carbon equivalent in
Honorary Life membership by the American Foundry Society, cast iron, or of the silicon content in Al-Si alloys, further developments
four NASA Certificates of Recognition for the Creative
extended the pertinency of the method to the evaluation of the effect
Development of Technical Innovations, the John Campbell
Medal, Institute of Cast Metals Engineers, United Kingdom, of chemical composition and cooling rate variations on the outcome of
and others. solidification microstructure. Then, as computational approaches were
E-mail: [email protected] facilitated by ever faster computers, the use and interpretation of the
Received: 2020-03-13; Accepted: 2020-03-25 time-derivatives of the cooling curve made possible prediction of the

69
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

nucleation potential (degree of inoculation) and

basic elements of the solidification microstructure
(a) (b)
such as dendrite arm spacing, grain size, graphite
morphology, and room-temperature microstructure.
More recently, attempts at predicting the shrinkage
propensity firmly established TA as a basic tool
in process control of all grades of cast irons. The
main reason for this development is the observation
that consistent melt chemistry is not sufficient to
ensure that the casting meets the target properties.
Consequently, TA is now used as a dynamic process
control tool as it allows anticipating the solidification
behavior [7]. Fig. 1: Examples of test cups for cooling curve analysis:
This review is an update of an earlier one [8] . (a) sand cup with disposable thermocouple
(Electro-Nite); (b) steel cup (SinterCast) with two
Endeavoring to avoid duplication, we will refer
thermocouples in the protective tube [9]
occasionally to this previous work, while emphasizing
the latest developments.
(a) (c)
Center
thermocouple

1 Principles of thermal analysis

I n i t s s i m p l e s t f o r m , TA , a l s o k n o w n t o t h e
foundrymen as Cooling Curve Analysis (CCA),
Chromel Alumel
consists in recording the temperature measured by one wire
or several thermocouples inserted in the mold. The (b) wire

mold may be a standard cup or the casting of interest.

The iron may be poured in a sand cup (Fig. 1a) or in a
metal cup that will be immersed in the metal to collect
a sample, and then extracted from the melt and allowed Side
thermocouple
to cool and solidify (Fig. 1b). As the design of the test
cup has a significant effect on the results and their
Fig. 2: GmbH dual-cup thermal analysis mold: (a) external view of
interpretation, the issue of cup design will be further the cup; (b) image from the internal sample body shape;
discussed later in more detail. The shape of the cooling (c) top view of the bottom half of the cup showing the
curve is determined by the balance between the latent disposition of the two thermocouples [10]
heat liberated during solidification and the heat lost to
surroundings (the test cup and the atmosphere). which improves the reliability of the interpretation of the TA.
More recently, Anjos et al. [10] developed a dual- The cooling curves in Fig. 3(a), adapted from Stefanescu [11],
cup TA system where in one chamber the hot junction demonstrate the differences occurring during the solidification of
of the thermocouple was placed at the center, and in four irons with similar carbon equivalent (CE). It is noticed that the
the other chamber the hot junction was close to the white iron has the highest undercooling and that the lamellar graphite
wall of the cup, as presented in Fig. 2. One of its main (LG) iron has the lowest. Also, compacted graphite (CG) iron has a
advantages is in producing cups with constant weight, higher maximum undercooling than spheroidal graphite (SG) iron.

(a) (b)

Fig. 3: Cooling curves for irons poured in Electro-Nite type sand cups: (a) cooling curves for four irons
with similar carbon equivalent; (b) cooling curve and standard terminologies [8]

70
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

The standard terminologies used in CCA are introduced in a given temperature from the slope of the cooling curve at
Fig. 3(b) and summarized in Table 1. The solidification time that temperature. The second column in Table 1, quoted from
can be found from the time difference between the start and Persson et al. [12], is that used by the commercial software ATAS.
end of solidification, and a cooling rate can be calculated for

Table 1: Most common terminologies used to define important parameters of the cooling curve

Symbol
Parameter
Ref. [8,13] Ref. [12]

Equilibrium liquidus temperature TL

Temperature of liquidus arrest TLA TL

Equilibrium eutectic temperature TE TE

Start of eutectic solidification TE_start

Temperature of eutectic undercooling TEmin TE_low

Temperature of eutectic recalescence TEmax TE_high

Recalescence ∆T

Maximum undercooling ∆Tmax

Minimum undercooling ∆Tmin

Temperature of end of solidification TS TS

Min. cooling rate during eutectic reaction (max. of 1st derivative) (dT⁄dt)min

Max. cooling rate after eutectic reaction (min. of 1st derivative) (dT⁄dt)max

Time interval at the end of eutectic reaction (on 2nd derivative) ∆tend

Some basic concepts used in the following text are explained start of the eutectic solidification, growth is still dominated by
with the help of Fig. 4. It should be noted that only the the dendrites, as the amount of eutectics produced increases
parameters TE_low and TE_high can be determined directly slowly over time (Zone 2). After the maximum undercooling is
from the cooling curve. Before the beginning of solidification reached (TE_low), the increasing amount of eutectic generated
(T>TL) the molten metal in the cup contracts, producing shifts the emphasis of solidification from dendritic dominated
liquid shrinkage. Solidification then starts at TL (or TLA) with (Zones 1 and 2) to graphite dominated (Zones 3 and 4). In
formation of dendrites that grow inward from the walls of Zone 3, graphite expansion compensates for liquid and dendrite
the cup until the start of the eutectic solidification (Zone 1 shrinking. However, in Zone 4, when the amount of eutectic
on the figure). In a hypoeutectic iron, immediately after the generated and thus that of the graphite decreases, there is a
risk of microshrinkage (microporosity) formation, as graphite
expansion may become insufficient to compensate for the
shrinkage.
More accurate determination of the values of the beginning
and end of solidification events from the cooling curve became
possible only after the introduction of computer generated
cooling rates from cooling curve data (first derivative of the
temperature-time curve) by Rabus and Polten [14] in 1972. A
second derivative of the cooling curve was then calculated and
studied [15]. A summary of the parameters of interest is illustrated
in Fig. 5 (see Table 1 for terminology). Then, higher order
derivatives, as high as the 5th derivative by Sparkman [16],
were interpreted with more or less success. While the physical
meaning of the first and second derivatives is clear, that of the
higher order derivatives is less understood.
To precisely determine the values of the beginning and end of
solidification events from the cooling curve, differential thermal
Fig. 4: Cooling curve of a hypoeutectic iron with characteristic
analysis (DTA) must be used. Classic DTA is performed with a
temperatures showing solidification ranges for
primary and eutectic phases and correlation with reference body [8]. However, for in-line process control in metal
shrinkage defects formation casting, DTA without a reference body was developed based on

71
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

v is the sample volume, ρ and cp are the density and the specific
heat of the alloy, respectively, to is the initial time, dT⁄dt is the
cooling rate and the subscripts cc and zc designate the cooling
curve and the zero curve, respectively.
Unfortunately, there is no unique accepted method to obtain
the zero-curve because of the limitations of the Newtonian
analysis. The main assumption of the method is that the
thermal gradient across the sample is zero. Consequently, it is
limited to Biot number less than 0.1, where the Biot number
is Bi=h/k∙v/A, where h is the heat transfer coefficient, k is the
thermal conductivity, v is the volume, and A is the cooling area [17].
It is thus reasonable for aluminum alloys, but less so for cast iron.
The second difficulty is that heat transfer between the casting and
Fig. 5: Cooling curve and its first and second derivatives [13]
the mold is assumed to be by convection and therefore, calculations
rely on the heat transfer coefficient which continuously changes
the calculation of the derivatives of the cooling curve (computer- during solidification. As a consequence, there is no unique method
aided cooling curve analysis), as integral calculus applied to calculate the zero-curve. Different approaches have been
to temperature-time data allows evaluation of areas that are proposed by a number of investigators [18-21].
directly related to the energy evolution. There are, in principle, To establish the beginning and end of phase transformations,
two major approaches: (1) Newtonian analysis and (2) Fourier most current commercial software programs use the first
analysis. Newtonian analysis requires only one thermocouple derivative of the cooling curve. However, as discussed in detail
and is the most widely used. Fourier analysis is a more accurate by Alonso et al. [22], a correct evaluation of these temperatures
treatment of the heat transfer problem, but it requires two must be done using the second derivative. This is supported by
thermocouples and the mathematics is more cumbersome [8]. the example in Fig. 7, where it is seen that the maximum of the
The Newtonian method consists in generating the 1st 2nd derivative corresponds to the beginning of the austenite
derivative of the cooling curve with respect to time (the solidification, while the maximum of the 1st derivative indicates
cooling rate), generating a zero-curve (the cooling without the liquidus arrest temperature (TLA). Also, the minimum on the
transformation), and then subtracting the area under the zero- 2nd derivative corresponds to the end of the eutectic solidification
curve from the area under the cooling rate. Then, the evolution (Fig. 8). When using the 1st derivative, the end of solidification
in time of the amount of phases is calculated as the ratio is calculated to occur at unreasonably low temperatures.
between the area corresponding to a phase formation (e.g. A more accurate mathematical approach to the problem is the
area corresponding to the solidification of primary austenite) Fourier analysis which assumes heat transfer by conduction.
to the total area, as illustrated in Fig. 6. Mathematically, this is It requires two thermocouples in the test casting. The Fourier
represented by the equation: equation with a heat source term is [23,24]:
t t
t t (1) ∝ T or p (2)
p

where fst is the fraction solid generated till time t, Qf is the total with the zero-curve given by Z F=∝ 2T. To calculate this
latent heat of fusion, Qft is the heat of fusion generated till time t, curve, the temperature field must be known, which for a

Fig. 6: Calculation of amount of phases from the areas under

the cooling rate and the zero curves; the shaded areas Fig. 7: Use of the 2nd derivative to establish the beginning
are proportional to latent heat released of primary (austenite) solidification for a gray iron [22]

72
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

(Si, Mn, P, S, etc.), allowing the multicomponent iron to be

treated like a binary Fe-C alloy on the equilibrium diagram. It
can be calculated from equilibrium thermodynamics as [26]:

CE = %Canal + 0.31∙%Si - 0.27∙%Mn + 0.33∙%P

(3)
+ 0.4∙%S ± mX∙%X

where %Canal is the carbon content obtained from chemical

analysis, %Si, %Mn, ..., %X are the other elements' contents,
and mX is the coefficient for the various elements (e.g. 0.31 for
Si, -0.27 for Mn) [27].
However, because solidification is a non-equilibrium process,
the thermodynamic equilibrium liquidus temperature is, in
most cases, higher than the actual non-equilibrium temperature
measured by the thermocouple. The evaluation of CE through
Fig. 8: Use of the 2nd derivative to establish the beginning
TA is based on the correlation between the non-equilibrium
and the end of eutectic solidification [22]
(experimental) liquidus temperature of an alloy (T LA) and
its chemical composition. Humphreys [28] has introduced the
cylindrical mold can be calculated as 2T=4(T2-T1)/(r22-r12),
Carbon Equivalent Liquidus (CEL) based on experimental data:
where T 2 and T 1 are the temperatures at radii r 2 and r 1 ,
respectively. The latent heat and fraction solid evolution are CEL = 13.46-0.0081∙TL where CEL = %C + %Si/4 + %P/2 (4)
calculated as described for the Newtonian analysis. Typical
Discussing thermal analysis of SG iron, Sillén [29] argued that
results are shown in Fig. 9 for an aluminum alloy. The time
the True Eutectic Temperature (TET) is the temperature where the
close to the end of solidification, when the cooling curve and
liquidus temperature coincides with the grey eutectic temperature
the Fourier zero curve start coinciding, is considered the end of
(TE_low) because the carbon equivalent is affected not only by the
solidification. Note that this occurs earlier than the minimum
chemical composition, but also the cooling rate and the nucleation
at the end of the cooling curve. For a more detailed analysis of
level. He advocates the use of the active carbon equivalent (ACEL),
the Fourier method through inverse heat conduction analysis,
implemented in the commercial TA equipment ATAS. For
the reader is referred to the paper by Diószegi and Hattel [25].
hypoeutectic ductile iron it is calculated as:
ACEL = 14.45 - 0.0089 ∙ TL (5)
To determine both C and Si through TA, Donald and Moore [30]
used two cups, a standard one and one that had some tellurium
addition. The method requires the use of a correction factor for
Si that depends on the phosphorus content of the iron, which
makes it cumbersome. The commercial equipment by Heraeus
Electro-Nite [31] uses Eq. 5 and the two-cup system. In addition,
for ductile iron, a three-cup system is offered to determine the
magnesium content. The system includes a standard cup, one
with tellurium, and one with tellurium and sulfur.
Standard methods for chemical analysis evaluation of
Fig. 9: Cooling curve, cooling rate and Fourier zero curve cast iron provide the total, or “residual”, Mg content, i.e.,
for an aluminum alloy [24] free Mg plus Mg bound in compounds such as MgS, MgO,
and silicates. Yet, what controls the graphite shape is the
2 Applications in cast iron processing free Mg termed “active Mg”. Suarez et al. [32] developed a
In industrial practice, a variety of methods and devices are method of Mg estimation by thermal analysis, which consists
used to extract useful information from the cooling curves of of the addition of a well-defined amount of Mg neutralizer
solidifying alloys. The goals of the analysis include estimation mixture (22% S and 78% inoculant) in a Te-containing cup.
of chemical composition, prediction of carbide formation and While Te promotes white iron when gray iron is poured in a
inoculation efficiency, graphite shape, shrinkage propensity, Te-containing cup (Fig. 10), no significant undercooling is
and mechanical properties. observed with SG iron, probably because of a Mg-Te reaction.
When the Mg neutralizer mixture is added into the Te-
2.1 Estimation of chemical analysis containing cup, Mg is preferably reacting with sulfur. Then, if
The first application of TA in process control of cast iron was the resulting cooling curve indicates white iron, it means that
to evaluate the carbon equivalent (CE). Carbon equivalent is the free Mg in the melt has combined with S, and therefore Te
a parameter that considers the fact that cast iron is a multi- has produced white iron. On the contrary, if the cooling curve
component alloy. It takes into account other important elements corresponds to a graphitic iron, it means that there is some

73
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

(a) (b)

(c)

Fig. 10: Cooling curves recorded with different cast

iron samples [32]

unreacted Mg, so no carbide-stabilizing effect of tellurium is

observed. The minimum active Mg content in the sample to
stop the whitening effect of Te mixture was established to be
Fig. 12: Interpretation of cooling curves to predict stable
0.006%. This procedure does not provide the amount of active or metastable solidification in cast iron [8]
Mg. It is used as a threshold to determine if the Mg is above a
certain limit [7]. This threshold is given by the %S added, which As discussed, the equilibrium Tst and Tmet calculated from
can be varied depending on the casting size and the production equilibrium thermodynamics ignore the effect of processing
process. variables such as pouring temperature and inoculation. Kanno
et al. [33,34] developed the corresponding equations from
2.2 Prediction of graphitization potential
experiments:
Because the cooling curve is sensitive to the amount of
carbides that solidify (Fig. 11), it can be used to predict the Tst = 1149.1+4.7∙%Si-4∙%Mn-44∙%P+2.7∙%Cu+%Ni
type of solidification, i.e. gray (stable) or white (metastable). +1.8∙%Co+13.9∙%Al-17.7∙%Mo-10.5∙%Cr-9.3∙%Sn (6)
This evaluation is possible if the stable Tst and metastable Tmet -5.2∙%Sb-6.1∙%W-3.7∙%Nb-14.8∙%V-80.3∙%B
temperatures can be calculated or measured. The principle is
Tmet = 1142.6-11.6∙%Si-0.75∙%Mn-46.2∙%P-1.4∙%Cu
described in Fig. 12. If both the start and end of the eutectic
-1.1∙%Ni-0.7∙%Co-1.8∙%Al-14.5∙%Mo+5.9∙%Cr (7)
solidification are above the metastable temperature, Tmet, the
iron solidifies gray, i.e. without carbides (Fig. 12a). If the end -6∙%Sn-5.1∙%Sb-2.8∙%W+3.3∙%V-26∙%B
of the solidification is under Tmet the iron is mottled (mixed where the temperature is in ºC and the elements are in mass%.
gray and white Fig. 12b). If both the beginning and end of For the case of Mn and S, Mn bonds S to make MnS, which
solidification are under T met, the iron is white (Fig. 12c). does not affect the equilibrium temperatures. Therefore,
The Tst and Tmet are not straight line, as they are affected by soluble Mn and soluble S should be calculated and used. The
microsegregation during solidification. Carbide promoting calculation method for soluble Mn and S in gray iron was
elements are rejected in the liquid, their content increases, and reported by the same researchers [35].
Tst bends downward, while Tmet bends upward. The opposite is Sparkman and Bhaskaran [36] found a linear relationship
true for graphite promoting elements. between the chill depth of a wedge test and T LA and T Emin
obtained from a plain cup, and the T Ewhite obtained from a
tellurium cup: chill depth = k1TLA+k2(TEmin-TEwhite)+k3.
Three one-thermocouple cups (base metal, with inoculant
and with tellurium) were used by Kanno et al. [37] to predict
the eutectic graphitization ability of gray iron, the quality of
the molten metal and even the tensile strength. Using the
definitions in Fig. 13, the eutectic graphitization ability, EGA,
is calculated as:
DT1 base iron Te cup
(8)
DTE inoc. iron Te cup

The chilling depth, D in mm, for gray iron with the

composition in the range 2.9%-3.5% C, 1.4%-2.4% Si and
0.1%-1.4% Cr, is calculated as a function of EGA as follows:
Fig. 11: Effect of amount of carbides on cooling curve [7] D = -0.27∙EGA + 28 (9)

74
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

Fig. 13: Definition of terminology used in three-thermocouple Fig. 14: Relationship between the temperature of liquidus
thermal analysis of gray iron [37] arrest and the amount of primary austenite [38]

The three-cups thermal analysis can predict the melt quality An important event during solidification is the occurrence
for gray iron, but it cannot be used in SG iron because, as the of dendrite coherency, which is the time when the individual
austenite solidifies ahead of graphite, Tst cannot be calculated. dendrites first impinge on one another. Before dendrite
In addition, as Te combines with Mg, it cannot produce chill coherency, flow is unrestricted, as dendrites flow with the liquid.
and Tmet cannot be calculated. However, as Tst and Tmet can After dendrite coherency, flow is restricted by the solid, rigid
be calculated with Eq. 6 and Eq. 7, Kano et al. [34] developed network of dendrites, and feeding of the casting is constrained.
a one cup method that uses the TEmin (TSC) from the cooling Anjos et al. [10] used the dual-cup system presented in Fig. 2 to
curve. As the sulfur combines with Mg, it is ignored in the determine the time of dendrite coherency during solidification.
calculation of Tst and Tmet. Then, EGA can be calculated as: According to Bäckerud et al. [39] the curve representing the
temperature difference between the solidification curve in the
(10) center and that in the wall can be used to find the coherency
point; it will correspond to the first minimum in the temperature
difference curve occurring after the start of solidification,
2.3 Prediction of solidification and room- as shown in Fig. 16. A projection of this point on the 1st
temperature microstructure and 2nd derivative curves (bottom of Fig. 16) shows that it
Early work by Ekpoom and Heine [38] demonstrated the approximately matches the maximum of the 1st derivative curve
existence of a linear relationship between the temperature of and the zero of the 2nd derivative of the temperature-time curve
the liquidus arrest, TLA, of gray iron and the amount of primary of the thermocouple in the center. Thus, the maximum cooling
austenite (Fig. 14): rate after the liquidus indicates the time at which the dendrite
coherency occurs.
%austenite = 0.304 TLA - 653.61 (11)
TA has also been used for a number of researchers to
Differential thermal analysis as explained in Fig. 6 allows attempt prediction of room temperature microstructure. Guo
calculation of the amount of primary (austenite or graphite) and Stefanescu [40] studied the effect of cooling rate during
and eutectic phases produced during the solidification of the eutectoid transformation on the microstructure of ferritic-
cast iron [19] . A more recent example of calculations and pearlitic (0.19% Mn, 0.13% Cu) and pearlitic (0.31% Mn, 0.28%
experimental validation is provided in Fig. 15. Note that the Cu) SG irons. They demonstrated that, for the ferritic sample, at
start and end of the eutectic solidification were found from the a relatively low cooling rate of 0.2 K·s-1, the 1st derivative detects
2nd derivative of the cooling curve. two apexes (Points 2 and 4 on Fig. 17a), corresponding to the

(a) (b)

Fig. 15: Calculation of solid fraction evolution for primary austenite and eutectic: (a) comparison with
experimental results obtained through quenching experiments; (b) adapted after Ref. [22]

75
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

(a) ferrite and pearlite transformation. Only one apex

is seen for a higher cooling rate of 0.4 K·s-1, which
indicates that only a pearlitic transformation occurs
(Fig. 17b).
In a more recent study, Sertucha et al. [41] studied
the effect of inoculation and changes in the Cu and
Mn content on the room temperature microstructure
of a large number of industrial SG irons cooled with
~50 K·min-1. The main cooling curve parameters
included in their analysis are shown in Fig. 18.
Note the velocity of transformation V trans was
measured at the point of inflection of the eutectoid
transformation, and T trans corresponding to this
(b) point was calculated from the maximum of the 1st
derivative.
Some reasonable correlations were found between
the measured amount of ferrite and the parameters
Vtrans and Ttrans, as shown in Fig. 19.
A statistical analysis equation was derived from
the experimental data as a function of cooling curve
parameters: %ferrite = f(TEmin, ∆T, VEmax, Vtrans, IGr),
where VEmax is the maximum cooling rate at the final
step of the eutectic transformation, and IGr is the
Fig. 16: Determination of the dendritic coherency point [10]

(a) (b)

Fig. 17: Cooling curves and their first derivative curves for ferritic SG iron: (a) 0.2 K·s-1 cooling rate;
(b) 0.4 K·s-1 cooling rate [40]

graphite growth factor at the eutectic calculated as:

C Si
(12)

Here, M sample is the mass of the sample, t Gr is the graphite

growth time in the eutectic range. A good agreement was found
between the calculated ferrite content and the experimental
measurements on the metallographic samples.

2.4 Prediction of inoculation efficiency and

nodule count
Inoculation is an important part of the processing of LG
and SG irons. Inoculation is a transitory process and its
Fig. 18: Critical parameters of the cooling curve for the effect decreases with the holding time (Fig. 20), a process
region of solid-state transformation [41] called fading. Thus, prediction of inoculation efficiency

76
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

(a) (b)

Fig. 19: Correlation between parameters of cooling curve and % ferrite measured on
metallographic samples [41]

Another nodule count prediction method relies on an

equation developed from the statistical analysis of experimental
data [42] . The nodule count, N, is a function of selected
parameters obtained from TA: N = f(TEmin, ∆T, (dT/dt)max, GR).
The TA notations are those listed in Table 1, and GR = ∑mGi/(∆ti),
a parameter defined from the shape of the cooling curve, is the
graphite formation rate calculated from the mass of graphite
precipitated mGi at each time step ∆ti.

2.5 Prediction of graphite shape

The graphite shape can be estimated from information extracted
from the cooling curve at the end of solidification, as shown
Fig. 20: Effect of inoculation fading on the cooling curves in Fig. 22 for Mg-treated irons with Mg contents of 0.006%
of SG iron [7] to 0.031% [13]. A good correlation of 0.93 was found between
nodularity and the time interval at the end of eutectic reaction (on
is a significant part of process control. Various inoculants
2nd derivative) ∆tend. However, data from similar heats but also
will produce cooling curves with different undercooling. In
containing Ti, lined up above the regression line in the figure.
principle, the best inoculant generates the lowest undercooling.
For SG iron, inoculation efficiency is measured by the
ability of the inoculant to prevent carbide formation and to
insure a large number of SG aggregates (high nodule count). It
can be estimated using the eutectic graphitization ability, EGA
Eq. 10, as shown in Fig. 21. Based on this figure, the nodule
count can be calculated as:

× ×
(13)

Fig. 22: Estimation of nodularity using information from

the cooling curve at the end of solidification
(∆tend, and (dT/dt)max, the maximum cooling rate
at the end of solidification) [13]

Some thermal analysis systems relate the angle θ on the first

derivative of the cooling curve at the end of the solidification
(also labeled as GRF2 by some systems) with the shape of
the graphite (Fig. 23). Low angles correspond to high thermal
Fig. 21: Relationship between EGA and the graphite nodule conductivity. Gray irons typically show lower angles than
count for irons inoculated with various alloys [34] ductile irons.

77
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

Fig. 23: Comparison of end of solidification for gray

and ductile iron [7]
Fig. 24: Cooling curves of blocks with 5 cm thermal modulus
Compacted graphite iron presents a particularly difficult produced from iron with 4.3 CE and 0.42% Mg: full
processing case as the magnesium window is very narrow. This line - inoculated; dotted line - not inoculated [46]

is because the graphite shape is also affected by the cooling rate

and inoculation that enhance graphite spheroidization. Perhaps
the best example of successful application of thermal analysis
in the process control of cast iron is in the production of CG
iron by Sintercast, the most widely used method in the industry
for production of CG iron [43]. The uniqueness of the process
that uses a one-cup made of steel consists in the intentional
undertreatment of the iron followed by ladle correction. The
system is based on the concept of on-line Measure-and-Correct
that in summary includes the following steps: (1) the liquid iron
is undertreated with a Mg-containing alloy; (2) a steel thermal
Fig. 25: Correlation between the volume of CHG regions
analysis cup is poured at a closely controlled temperature; (3) in the test blocks (VV) and ΔTmax measured from
depending on the result of the TA, a wire feeder inserts into the TA on cups [46]
pouring ladle a calculated amount of inoculant and magnesium to
bring the iron to the optimum chemical composition for pouring. Regarding the notation TEN (initial eutectic reaction
An important issue in the production of heavy section ductile consisting of encapsulation of primary nodules by austenite
iron castings is the avoidance of chunky graphite (CHG). It dendrites), it is now accepted that for the case of hypereutectic
is currently an active subject of research as evident in some SG iron, this is an austenite liquidus arrest found on the phase
recent reviews [44,45]. Attempts have been made to correlate diagram when the concept of asymmetric eutectic region is used.
crystallization of chunky graphite with parameters of the cooling
2.6 Prediction of shrinkage propensity
curve. Thermal analysis by Sertucha et al. [46] of large blocks
(thermal modulus 5 cm) produced from SG irons with 4.28% After cast iron is poured in a mold and the liquid metal cools,
to 4.38% CE that had tendency to generate chunky graphite, liquid shrinkage occurs as the density of the liquid increases
exhibited up to three distinct thermal arrests, all at temperatures and its specific volume decreases. During solidification,
lower than the stable eutectic temperature, confirming earlier some graphite expansion may compensate the shrinkage. The
work by Chaudhari et al. [47]. Example of the recorded curves interplay between these two phenomena produces several
with notations as suggested by Chaudhary are presented in Fig. shrinkage defects illustrated in Fig. 26 [48]. Some of these
24. The sequence of solidification as seen on the cooling curve defects are the result of uncompensated shrinkage (pipe, caved
and confirmed from metallographic samples includes nucleation surfaces), some are produced by a decrease in the gas solubility
of primary austenite graphite in the liquid (TL), initial eutectic in the liquid alloy (center-line microporosity), or form the
reaction consisting of encapsulation of primary nodules by
austenite dendrites (TEN arrest), initial limited growth of eutectic
(TEU), and bulk eutectic growth (TER). CHG cells and secondary
nodules arranged into SG cells formed during the bulk eutectic
growth. Attempts were made to correlate the volume of CHG in
the test blocks to information extracted from TA. No consistent
correlations were found. The best one appears to be with the
recalescence on the TA cups, shown in Fig. 25. Surprisingly,
the lowest amount of CHG appears to be in some of the Fig. 26: Various types of shrinkage defects found in an
uninoculated samples. experimental casting [48]

78
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

mold/metal interface interaction (surface microporosity), and the case of insufficient expansion, as shown in the critical range
some can be the result of both shrinkage and gas porosity in Fig. 27. The figure also shows a drop in expansion at the
(macroporosity or enclosed shrinkage). end of solidification, typically produced by decreased graphite
During primary solidification of hypoeutectic iron, in the growth in the last regions to solidify because of segregation of
dendrite dominated solidification (Fig. 4), the liquid/austenite carbides-promoting elements. For SG iron, the permeability
mixture continues to shrink as the specific volume decreases of the mushy zone is much lower than that of LG iron. This
(Fig. 27). As long as the permeability of the mushy dendritic restricts feeding from the riser and decreases the cooling rate.
region is sufficiently high, the shrinkage is compensated by In principle, both LG and SG irons of near-eutectic or eutectic
flow from the risers. Graphite expansion begins once the composition should exhibit expansion during solidification
maximum undercooling, TE_low, is reached and may, or may and therefore should not be prone to forming either cavity or
not, continue to the end of solidification. The final result will be porosity shrinkage. While this is true for gray iron, normal
expansion if enough graphite expansion occurs, or shrinkage in production SG iron is susceptible to shrinkage porosity because
of limited graphite growth at the end of solidification, which is
then dominated by austenite shrinkage. This induces specific
volume decrease and produces uncompensated shrinkage in the
last region to solidify, and thus porosity. Mathematically, this
is expressed as a negative pressure at the end of solidification,
as described in detail in Ref. [48,49]. A bimodal volume size
distribution of graphite nodules (small nodules coexisting
with large nodules), sometimes observed in SG iron castings,
appears to indicate the existence of a second nucleation wave at
the end of solidification. This behavior, or extended growth of
graphite to the end of solidification, will reduce or even prevent
microporosity. It follows that understanding graphite expansion,
Fig. 27: Temperature and specific volume evolution in particular at the end of solidification, is vital for process
during the solidification of cast iron development of porosity-free castings.

(a) (b) (c)

Fig. 28: Interpretation of LDA/TA experiments: (a) time evolution of temperature, linear displacement and fraction
solid for a 4.35% CE SG iron; (b) available graphite expansion at the end of solidification for SG iron;
(c) effect of graphite shape on graphite expansion for irons with 4.12% to 4.44% CE [50]

To this end, Alonso et al. [50] used a combined Linear From this discussion, it is clear that accurate prediction of
Displacement Analysis (LDA) - Thermal Analysis (TA) shrinkage propensity is highly dependent on accurate evaluation
experimental device to measure the linear displacement of the permeability of the mushy zone and the pressure change
during the solidification of cast iron. With this system, it was in the zone during solidification. Svidró et al. [49] developed a
possible to examine the correlation between temperature, linear measurement method and a calculation algorithm to determine
displacement, and fraction of solid evolution (Fig. 28a). This in the pressure in the mushy zone during solidification. The method
turn made possible calculating the amount of graphite expansion involves a spherical sample suspended in a measurement
available at the end of solidification which counteracts the device that records the temperature and the volume changes
austenite shrinkage. Not surprisingly, this expansion increases during solidification. The thermocouples are positioned in the
with CE, as more graphite will be available with higher CE (Fig. geometrical center of the sphere and at its periphery, to allow
28b). The graphite morphology also has a significant effect. for a Fourier analysis of the cooling curves. The calculation
Considerably less graphite expansion of 0.3 to 0.4 was found in algorithm is based on the numerical interpretation of the
CG iron than in SG iron with roundness of 0.7 (Fig. 28c). Clausius– Clapeyron equation where the temperature variation,

79
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

the volume change, and the released latent heat are processed
to determine the local pressure during solidification. The
cooling curves of an LG iron of 3.85% CE collected during
solidification are presented in Fig. 29 as a function of the
fraction of solid. The calculated volume change based on the
measured surface displacement is also presented in the figure.
It is seen that the sample volume decreases from the start of
solidification until the fraction solid reaches 0.32. The start of
the graphite precipitation is also confirmed at fs = 0.32 by the
temperature minimum on the surface temperature curve T2. The
maximum volume is reached at the beginning of recalescence.
Then, the volume decreases because of insufficient graphite
formation to compensate for austenite shrinkage.
Fig. 30: Calculated pressure change and calculated
volume (V) as a function of fraction solid [49]

spots by the feeder). If the connection to the riser closes early,

piping and caving of the surface can be expected even though
graphite expansion at the end of solidification is adequate. In
casting production, the problem is further complicated by the
mold material (mold rigidity or lack of), pouring temperature,
and inoculation.
A study on ductile iron by Stefanescu et al. [51] found
the volume of macroporosity decreased and the dispersed
microshrinkage increased with higher graphite nodule
Fig. 29: Cooling curves and calculated volume (V) as a function count. When inoculation was used, despite the decrease in
of fraction solid; T1 - cooling curve registered from the macro-shrinkage, the total dispersed shrinkage increased
geometrical center of the spherical sample, T2 - average considerably. Mold rigidity affected the macro-shrinkage but not
of the temperatures measured at three different
microporosity formation at the end of solidification. This proved
positions on the surface of the sample [49]
that the graphite nodule count is an important but not sufficient
Figure 30 presents the calculated change in pressure and the parameter to predict shrinkage porosity in SG iron castings.
temperature in the center as a function of the fraction solid. The difficulty of predicting shrinkage defect formation can
Positive values on the y-axis are interpreted as compressive be illustrated using the work of Lekakh and Hrebec [52]. Their
pressure, while negative values are considered to be expansion experimental casting and the measured shrinkage defects are
pressure. It is seen that expansion pressure starts at fS=0.32 shown in Fig. 31. The effects of inoculation and feeding ability
when the volume begins to increase. At about 0.7 fraction (no riser or different distances from the riser) are illustrated.
solid, the pressure becomes compressive again and remains so Different results are seen for the same iron depending on the
to the end of solidification. Thus, the region fS=0.7 is the region feeding conditions. Thus, it will be hard to assess shrinkage
where graphite expansion stops compensating for austenite probability only from a cooling curve information produced
shrinkage, and microshrinkage may become a problem. with a standard TA system, which can only predict melt quality.
Equally important for producing a shrinkage defect-free Nevertheless, the information that the cooling curve provides
casting is the gating-riser system design (accessibility of hot on the end of solidification can be used to predict shrinkage

Fig. 31: Experimental casting and measured porosity at different locations in the casting [52]

80
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

propensity of the particular alloy.

Persson and Udroiu [53] indicated that the commercial software
ATAS is using the angle θ formed on the 1st derivative curve after
the end of solidification (see θ in Fig. 23 and GFR 2 in Fig. 32) as
a shrinkage tendency indicator. A large θ angle indicates a high
shrinkage tendency. A lamellar graphite iron has a narrower
angle at the end of solidification when compared with the
spheroidal graphite iron. This is attributed to the differences in
the skin-type solidification of LG iron compared to the mushy-
type solidification of SG iron [8]. Very little latent heat is released
at the end of solidification of LG iron, which determines a sharp
increase in the cooling rate. More latent heat is released at the
end of solidification of SG iron as SG grows for a longer time Fig. 32: Parameters of the cooling curve used in
than LG. This produces a slower increase in the cooling rate. analysis by ATAS [53]

(a) (b)

Fig. 33: Correlation between the shrinkage tendency as estimated through the θ angle on TA with:
(a) measured shrinkage volume; (b) nodule count as calculated by EGA [34]

Kanno et al [34] used a conical-shape test piece to measure the must be as small as possible, and the graphite dominated regions
correlation of shrinkage volume and the θ angle on the 1-cup TA (3 and 4 in Fig. 4), where expansion occurs, must be large. The
(Fig. 33a). A good relationship between shrinkage tendency (θ ATAS thermal analysis system can calculate the free graphite
angle) and the nodule count as estimated through the EGA Eq. and shrinkage-expansion as a function of temperature from the
13 was demonstrated through Fig. 33(b). carbon equivalent (Fig. 34). Using ATAS, the authors were able
Erturk et al. [54] used ATAS to evaluate its applicability to to calculate the total shrinkage at the end of solidification and
the production of riserless SG iron castings. The experimental design the feeding-casting system for feederless casting.
variables included inoculation quality, type of resin in the mold, Another parameter that can be used for shrinkage propensity
and mold rigidity. According to ATAS instructions, to minimize is the cooling rate at the end of solidification. LG iron has a
macro shrinkage, the dendritic dominated zone (1 and 2 in Fig. 4) higher cooling rate at the end of solidification than SG iron.

(a) (b)

Fig. 34: The first derivative of the cooling curve and the volume change during solidification [54]

81
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

Consequently, the shrinkage tendency can also be estimated This equation allows the calculation of volume change
through the minimum of the first derivative at the end of because of austenite contraction and graphite expansion during
solidification (dT/dTS in Fig. 32, or (dT/dt)max in Fig. 5). solidification, and thus the contraction-expansion balance at
Larrañaga et al. [42] used time information from the first different moments during solidification, as shown in Fig. 35(a).
derivative of the cooling curve to estimate shrinkage propensity. They defined the ratio k=tExp/(tExp+tFinal_contr) as a predictor for
First, they derived an equation to calculate the volume change shrinkage porosity in cast iron. A low k is an indicator of high
between two successive time steps, vi, during solidification: porosity. The model was verified with a cross-shaped casting
(poured in the vertical position with a riser on top of the arm of
∆vi=∆mγ (ργ-1-ρL-1)+∆mG (ρG-1-ρL-1) (14)
the cross) in conjunction with SG iron. The results are presented
where m is the mass, ρ is the density, and the subscripts γ, L in Fig. 35(b). The same authors have also proposed an empirical
and G stand for austenite, liquid and graphite, respectively. equation for the calculation of nodule count based on TA data.

(a) (b)

Fig. 35: Use of time-parameters for shrinkage evaluation: (a) suggested correlation between times
and areas from the cooling rates curves to microporosity propensity; (b) correlation between
the volume of microshrinkage and the k factor [42]

A recent novelty in the prediction of microshrinkage interest is considered through selected features of the eutectoid
propensity is the combination of information obtained from transformation that include the undercooling and the pearlite
TA with a computer simulation model. Kweon et al. [48] used growth rate (eutectoid recalescence and maximum cooling rate
the fraction solid evolution calculated from the cooling curves during the eutectoid transformation). The agreement between
recorded on the experimental casting in conjunction with an predicted and experimental values is shown in Fig. 36.
analytical model. The model captures some of the important
elements of the physics of the problem, by calculating gas
evolution in the melt with increasing fraction of solid and with 3 Summary
decreasing mushy zone permeability during solidification.
While calculation results compared well with the experimental A review of the work performed over the last 90 years for the
observation on the L-shaped castings in Fig. 26, some development of TA and its implementation in industry shows
disagreements were also registered. These were attributed to gas that progress has been made both in the development of the
pores migration during solidification, a phenomenon that is not hardware and the software of TA. The consistency of the
included in the physics described by the model. data provided by the TA depends heavily on two factors: the
temperature at which the metal is poured in the cup and the
2.7 Prediction of mechanical properties cooling rate (the mass of the metal in the cup) during the test.
The ultimate tensile strength (UTS) and Brinell hardness (HB) A higher pouring temperature and a greater mass determine
in gray iron are directly related to the carbon equivalent, and a longer solidification time which affects significantly the
especially to carbon content, and more generally to eutectic undercooling. These are both difficult to control with the
and eutectoid parameters. Because these parameters can be one-cup system. To the best of our knowledge, at this time,
accurately obtained from TA, Udroiu, quoted in Ref. [7], only two systems satisfy the requirement for constant mass:
has developed a statistical analysis model for UTS and HB Sintercast (Fig. 1b) and GmbH (Fig. 2). Only Sintercast
prediction in 30 mm diameter grey iron test bars based on satisfies both requirements.
cooling curves recorded in standard thermal analysis cups. It The progress in the interpretation of cooling curves includes
captures the effect of graphite nucleation and growth, through correcting the misconception that the maximum cooling
TEmin and ∆T, and carbon equivalent TLA, on the final graphite rate (min. of 1st derivative) after TLA shows the time of the
features. The effect of pearlite fineness on the properties of beginning of solidification. It was demonstrated that this

82
 Special Review CHINA FOUNDRY
Vo l . 1 7 N o . 2 M a r c h 2 0 2 0

(a) (b)

Fig. 36: Relationship between calculation and experiments on 30 mm diameter test bars [7]: (a) tensile
strength; (b) Brinell Hardness [7]

was actually the dendrite coherency point [10]. The beginning [12] Persson P E, Ignaszak Z, Fransson H, et al. Increasing precision
of primary and eutectic solidification, as well as the end of and yield in casting production by simulation of the solidification
process based on realistic material data evaluated from thermal
solidification, must be calculated from the 2nd derivative [22].
analysis (using the ATAS MetStar system). Archives of Foundry
While prediction of many parameters is claimed, the data Eng., 2019, 19(1): 117-126.
scattering is many times too large to be useful in the control of [13] Charoenvilaisiri S, Stefanescu D M. The Possibility of prediction of
foundry processes. We believe this to be particularly true for graphite shape in Mg-treated irons by means of thermal analysis.
prediction of shrinkage defects, where the outcome depends In: Proceedings of the 7th Asian Foundry Congress, The Chinese
Foundrymen’s Association, Taipei, Taiwan, 2001: 91-100.
not only on metal quality, but also on gating and casting
[14] Rabus D, Polten S. Gieserei Rundshau, 1972, 9: 1-8.
design. TA can only predict metal quality and shrinkage
[15] Chen I G, Stefanescu D M. Computer-aided differential thermal
tendency in the test cup. Thus, the ideal solution should analysis of spheroidal and compacted graphite cast irons. AFS
combine well-controlled TA measurements with computational Trans., 1984, 92: 947-964.
models that can describe mold filling, calculation of graphite [16] Sparkman D A. Microstructure by thermal analysis. AFS Trans.,
expansion, and gas evolution. 2011, 119: 413-420.
[17] Lekakh S N, Richards V L. Determining solidification parameters
of alloy steels. AFS Trans., 2011, 119: 521-528.

References [18] Ekpoom L, Heine R W. Thermal analysis by differential heat

analysis (DHA) of cast iron. AFS Trans., 1981, 89: 27-38.
[1] Le Chatelier H. Z. Phys. Chem., 1887, 1: 396. [19] Upadhya K G, Stefanescu D M, Lieu K, et al. Computer-aided
[2] Piwowarski E. High strength cast iron (Hochwertiges cooling curve analysis, principles and applications in metal
Gusseisen), Springer Verlag, Berlin, 1961. casting. AFS Trans., 1989, 97: 61-66.
[3] Piwowarski E. Giesserei, 1938, 25: 523. [20] Kierkus W T, Sokolowski J H. Recent advances in CCA: A new
[4] Loper C R , Heine R W, Reesman R W, et al. Thermal analysis method of determining baseline equation. AFS Trans., 1999,
of ductile iron. AFS Trans., 1967, 75: 541. 107: 161-167.
[5] Naro R, Wallace J F. Minor elements in gray iron. AFS Trans., [21] Dioszegi A, Svensson I. On the problems of thermal analysis of
1970, 78: 229-238. solidification. Mat. Sci. Eng. A, 2005, 413-414: 474.
[6] De Sy A, Vidts J. Theoretical and applied metallurgical treatise [22] Alonso G, Larrañaga P, Sertucha J, et al. Gray cast iron
(Traité de metallurgie structurale theorique et appliquée), Ed. with high austenite-to-eutectic ratio, Part I - Calculation and
Dunod, Paris, 1962. experimental evaluation of the fraction of primary austenite in
[7] Suarez R, Larrañaga P, Udroiu A. Cooling curve (thermal) cast iron. AFS Trans., 2012, 120: 329-335.
analysis. In: D. M. Stefanescu Ed., ASM Handbook, Vol. 1A: [23] Fras E, Kapturkiewicz W, Burbielko A, et al. A new concept in
Cast Iron Science and Technology. ASM International, Materials thermal analysis. AFS Trans., 1993, 101: 505.
Park Ohio, 2017: 149-159. [24] Barlow J O, Stefanescu D M. Computer-aided cooling curve
[8] Stefanescu D M. Thermal Analysis – Theory and applications in analysis revisited. AFS Trans., 1997, 105: 349.
metal casting. Int. J. Metalcasting, 2015, 9(1): 7-22. [25] Dioszegi A, Hattel J. Inverse thermal analysis method to study
[9] Dawson S, Popelar P. Thermal analysis and process control solidification in cast iron. Int. J. Cast Metals Research, 2004, 17:
for compacted graphite iron and ductile Iron. SinterCast, 2014, 311.
www.sintercast.com. [26] Stefanescu D M, Lacaze J. Thermodynamics principles as
[10] Anjos V, Deike R, Silva Ribeiro C. The use of thermal analysis to applied to cast iron. In: D. M. Stefanescu Ed., ASM Handbook,
predict the dendritic coherency point on nodular cast iron melts. Vol. 1A: Cast iron science and technology. ASM International,
Science and Technology of Materials, 2017, 29: 27-33. Materials Park Ohio, 2017: 31-45.
[11] Stefanescu D M. Solidification of flake, compacted and [27] Stefanescu D M, Katz S. Thermodynamic properties of iron-
spheroidal graphite cast irons as revealed by thermal analysis base alloys. ASM Handbook, Vol. 15-Casting, 2008: 41-55.
and directional solidification experiments. In: The Physical [28] Humphreys J G. Effect of composition on the liquidus and
Metallurgy of Cast Iron, H. Fredriksson, M. Hillert eds. Elsevier, eutectic temperatures and on the eutectic point of cast irons.
1985: 151-162, doi:10.1557/PROC-34-151. BCIRA J, 1961, 9 (5): 609-621.

83
CHINA FOUNDRY Special Review
Vo l . 1 7 N o . 2 M a r c h 2020

[29] Sillén R. Finding the true eutectic point – An essential task for [44] Baer W. Chunky graphite in ferritic spheroidal graphite cast iron:
efficient process control of ductile iron. Ductile Iron News, 2007, formation, prevention, characterization, impact on properties:
2: 100-104. an overview. Int. J. Metalcasting, 2019, https://doi.org/10.1007/
[30] Donald W, Moore A. BCIRA J., 1973, Report No. 1128. s40962-019-00363-8.
[31] Van der Perre W. Thermal analysis, principles and applications. [45] Stefanescu D M, Alonso G, Suarez R. Recent developments
Heraeus Electro-Nite, https://www.heraeus.com/en/hen/home_ in understanding nucleation and crystallization of spheroidal
heraeus_electro_nite/home_electro_nite.html. graphite in iron-carbon-silicon alloys. Metals, 2020, 10: 221,
[32] Suarez R, Sertucha J, Larrañaga P, et al. Active Mg estimation doi:10.3390/met10020221.
using thermal analysis: a rapid method to control nodularity in [46] Sertucha J, Suarez R, Asenjo I, et al. Thermal analysis of the
ductile cast iron production. Metall. Mater. Trans. B, 2016, 47B: formation of chunky graphite during solidification of heavy-
2744-2753. section spheroidal graphite iron parts. ISIJ Int., 2009, 49(2):
[33] Kanno T, Fukuda Y, Morinaka M, et al. Effect of alloying 220–228.
elements on graphite and cementite eutectic temperature of cast [47] Chaudhari M D, Heine R W, Loper C R. Principles involved in
iron. J. JFS, 1998, 70: 465-470. the use of cooling curves in ductile iron process control. AFS
[34] Kanno T, Iwami Y, Kang I. Prediction of graphite nodule count Trans., 1974, 82: 379-386.
and shrinkage tendency in ductile cast iron with 1 cup thermal [48] Kweon E S, Roh D H, Kim S B, et al. Computational modeling of
analysis. Int. J. Metalcasting, 2017, 11(1): 94-100. shrinkage porosity formation in spheroidal graphite iron: a proof
[35] Kanno T, Kang I, Mizuki T, et al. Effect of S, Mn and Re on of concept and experimental validation. Int. J. Metalcasting,
eutectic temperature of cast iron. J. JFS, 2001, 73: 441–446. 2020, doi.org/10.1007/s40962-020-00417-2.
[36] Sparkman D A, Bhaskaran C A. Chill measurement by thermal [49] Svidró P, Diószegi A, Jöhnsson P G, et al. Determination of
analysis. AFS Trans., 1996: 969–976. pressure in the extradendritic liquid area during solidification,
[37] Kanno T, Kang I, Fukuda Y, et al. Prediction of chilling tendency, J. Thermal Analysis and Calorimetry, 2018, 132: 1661–1667,
graphite types and mechanical properties in cast iron, using https://doi.org/10.1007/s10973-018-7088-z(0123456789().,-
three cups thermal analysis. AFS Trans., 2006, 114, paper No. volV)(0123456789,-().volV).
06-083. [50] Alonso G, Stefanescu D M, Suarez R, et al. Understanding
[38] Ekpoom U, Heine R W. Thermal analysis by differential heat graphite expansion during the eutectic solidification of cast iron
analysis (DHA) of cast iron. AFS Trans., 1981: 27–38. through combined linear displacement and thermal analysis. In:
[39] Bäckerud L, Chalmers B. Trans. Metall. Soc. AIME, 1969, 245: Proc. 71st World Foundry Congress, Bilbao, Spain, 2014.
309. [51] Stefanescu D M, Qiu H Q, Chen C H. Effects of selected metal
[40] Guo X, Stefanescu D M. Solid phase transformation in and mold variables on the dispersed shrinkage in spheroidal
ductile iron—A benchmark for computational simulation of graphite cast iron. AFS Trans., 1995, 95: 189-197.
microstructure. AFS Trans., 1997, 105: 533-544. [52]  Lekakh S N, Hrebec B. Solidification kinetics of graphite nodules
[41] Sertucha J, Suarez R, Izaga J, et al. Prediction of solid-state in cast iron and shrinkage porosity. Int. J. Metalcasting, 2016,
structure based on eutectic and eutectoid transformation 10(4): 389-400.
parameters in spheroidal graphite irons. Int. J. Metalcasting, [53] Persson P E, Udroiu A, Vomacka P, et al. ATAS® as a tool for
2006, 19(6): 315-322. analyzing, stabilizing and optimizing the graphite precipitation in
[42] Larrañaga P, Gutierez J M, Loizaga A, et al. A computer-aided grey cast iron, In: Proceedings 69th World Foundry Congress,
system for melt quality and shrinkage propensity evaluation FICMES, Hangzhou, China, 2010.
based on the solidification process of ductile iron. AFS Trans., [54] Erturk S O, Kumruoglu L C, Ozel A. Determination of feederless
2008, 116: 547-561. casting limits by thermal analysis in cast iron. Acta Physica
[43] Guesser W, Schroeder T, and Dawson S. Production experience Polonica A, 2017, 131(3): 370-373.
with compacted graphite iron automotive components. AFS
Trans., 2001: 01–071.

84