Sie sind auf Seite 1von 9

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 28B, FEBRUARY 1997115

Numerical Simulation and Fourier Thermal Analysis


of Solidication Kinetics in High-Carbon Fe-C Alloys
E. FRAS, W. KAPTURCKIEWICZ, A. BURBIELKO, and H.F. LO

PEZ
The aim of this work was to carry out both experimental and numerical simulations of cast iron
solidication under various conditions. The experimental work was based on a novel technique of
thermal analysis known as the Fourier method, whereas solidication modeling was possible by
solving the Fourier equation with a heat source. Moreover, a comparison between the Fourier and
the Newtonian method indicated that their predictions are appreciably different. The Newtonian
method is rather insensitive to the actual thermal gradients and predicts a clear maximum in heat
generation at the onset of solidication. In contrast, the Fourier method incorporates the effect of
actual thermal gradients and predicts two successive heat generation peaks of increasing magnitude
as solidication proceeds. In particular, it was found that the experimental outcome of solidied
volume fractions agrees closely with the predictions of the Fourier method. In this case, both ex-
perimental and computer simulations on 30- and 40-mm-diameter cylindrical specimens indicated
that the solidied fraction followed a sinusoidal trend. Moreover, it was found that under normal
solidication conditions, secondary nucleation of ne grains can occur near the center of a cylindrical
cast iron specimen. Secondary grain nucleation is attributed to the development of a second under-
cooling maximum which easily exceeds the initial one. Finally, the effects of inoculation were
investigated in plain cast iron and as a function of the inoculation time. Accordingly, in all cases,
the computer simulations were in close agreement with the experimental outcome.
I. INTRODUCTION
DURING the solidication of high-carbon Fe-C alloys,
appropriate feeding systems have to be implemented to
avoid the development of porosity as a result of metal vol-
ume changes.
[1,2]
Furthermore, simultaneous grain nuclea-
tion and growth as well as the development of local thermal
gradients within the melt make it difcult to discern the
roles of all the factors involved in cast iron solidication.
Hence, the nature of the events that take place during so-
lidication is rather complex, and computer simulation can
help to close the existing gap between the analytical and
empirical understanding of casting solidication. Moreover,
in recent years, there have been important developments in
the available thermal analysis techniques.
[314]
Conse-
quently, systematic studies of cast iron solidication kinet-
ics can be made to establish the relevant mechanisms using
both experimental analysis and numerical simulations.
A. Thermal Analysis
Thermal analysis is a valuable tool in elucidating the
thermal events involved in solid to liquid transformations.
In the case of alloy solidication, the data generated by
thermal analysis are very useful for a qualitative or quan-
titative description of the active solidication mechanisms.
Accordingly, a number of thermal analysis methods are al-
E. FRAS and W. KAPTURCKIEWICZ, Professors, and A. BURBIELKO,
Researcher, are with the University of Mines and Metallurgy, 30-059
Kracow, Poland. H.F. LO

PEZ, Associate Professor, is with the Materials


Department, University of Wisconsin-Milwaukee, Milwaukee, WI 53201.
Manuscript submitted December 5, 1995.
ready available,
[312]
which include the Newtonian
[3,4,5]
and
more recently the Fourier method of thermal analysis.
[69]
From the point of view of their implementation, before the
1980s, most kinetic studies of solidication were carried
out employing Newtonian thermal analysis (NTA). How-
ever, the major drawback of this method is its inability to
take into account the local temperature gradients developed
during solidication. More recently, this limitation has been
removed by an alternate method known as Fourier thermal
analysis (FTA), initially proposed by Longa et al.
[6,7]
The
main differences between NTA and FTA are in the way the
baseline (zero curve Z) is established. In the NTA method,
the Z
N
baseline is arbitrarily dened by extrapolation of a
curve of the type A exp (Bt ), where is the rate

T T
of cooling (dT/dt ) and A and B are experimentally deter-
mined constants.
[3,4]
In contrast, in the Fourier method, the
zero curve, Z
F
, is obtained from a determination of the tem-
perature Laplacian,
2
T, as follows.
The temperature of the metal after mold pouring can be
described by the Fourier equation which includes a heat
source:
T q
S
2
T [1]
t C
v
where T/t is the rate of metal cooling (K/s), is the
thermal diffusivity (cm
2
/s), C
v
is the volumetric specic
heat (J/cm
3
K), and q
s
is the volumetric heat generated dur-
ing solidication (J/cm
3
s). Equation [1] can be rearranged
in terms of q
s
as
t

2
q C C T C (T Z ) [2]
s v v v F
t
where Z
F
(baseline) is given by
2
T.
116VOLUME 28B, FEBRUARY 1997 METALLURGICAL AND MATERIALS TRANSACTIONS B
In addition, the fraction of solid (f
s
) can be described by
t
1
f (t) q (t)dt [3]
S S
tb
H
where H is the enthalpy of solidication and t
b
is the time
at the onset of solidication. From these equations, the
magnitudes of the heat rate components can be established
as long as both the thermal diffusivity and temperature eld
T(x,t ) are known. Experimentally, a minimum of three tem-
peratures is needed for the determination of the Laplacian

2
T. However, in a symmetric temperature eld, these var-
iables are reduced to only two. Hence, two thermocouples
inside the solidifying melt are sufcient to nd
2
T. In
cylindrical mold geometries containing two thermocouples
located at R
1
and R
2
from the center, the Laplacian can be
described by References 8 and 9.
4(T T )
2 1
2
T [4]
2 2
R R
2 1
In the absence of heat generation (i.e., during melt cooling
from the pouring temperature to the liquidus temperature,
or after solidication), q
s
0, and from Eq. [1], the thermal
diffusivity is given by

T
[5]
2
T
Consequently, when and
2
T are experimentally known

T
before and after solidication, the thermal diffusivities of
the liquid and solid can be established. Moreover, if both
thermal diffusivities have similar values, the solidication
kinetics can be determined by (1) experimentally nding
the cooling curves T
1
and T
2
and the differential cooling
curve (2) then establishing Z
F
and the rate of heat gen-

T ;
1
eration (Eq. [2]); and (3) establishing the solidication ki-
netics by means of Eq. [3].
When the thermal diffusivities of solid and liquid are
different, the thermal diffusivity can be assumed to be a
function of the volume fraction of the solid. In this case,
an iterative process described elsewhere
[8,9]
can be used for
f
s
determinations. A similar assumption can be applied to
the specic heats of liquid and solid.
B. Solidication Modeling
Assuming that heat conduction is the dominant heat-
transfer mode, the temperature eld, T, of the metal during
solidication can be determined from the heat ow equa-
tion, Eq. [1], for the following boundary conditions.
At the metal/mold interface,
T T
1 2
[6a]
1 2
x x
At the external mold surface,
T
2
h (T T ) [6b]
2 2,S 2,O
x
where x is the geometrical coordinate, h is the heat-transfer
coefcient, and T
2,O
and T
2,S
are the room temperature and
external mold surface temperature, respectively.
C. Nucleation Kinetics
A distribution of particle sizes which effectively acts as
substrates for heterogeneous nucleation as a function of the
degree of undercooling, T, was assumed to be present in
the liquid metal. However, integration of the Gaussian size
distribution in the 0 T range of undercooling in accor-
dance with the classical nucleation theory is compli-
cated
[15,16]
and is usually avoided. Alternatively, since less
than half of the distribution is effectively taking part in the
nucleation process, further simplication can be attained by
using a power-law function for nucleation densities of both
dendritic and eutectic grains according to
n
N T [7a]
d d
m
N T [7b]
e e e
where
d
,
e
, n, and m are experimentally determined con-
stants and T

and T
e
are the degrees of undercooling of
austenite and eutectic, respectively. These expressions do not
require extensive numerical calculations and have been ex-
perimentally conrmed by various workers.
[17,18]
Moreover,
from the nucleation viewpoint, it is expected that nucleation
will end at the onset of recalescence (i.e., at maximum un-
dercooling) due to the exhaustion of active substrates.
D. Growth Rate and f
S
In this work, dendritic growth was modeled by employ-
ing the expression derived by Fras et al.
[13,14]
and Lipton et
al.
[19]
for the growth of a parabolic dendrite tip. Also, in
the case of eutectic grain growth, the well-known relation-
ship was employed:
[16]
2
u (T) [8]
e e
where
e
is the eutectic growth coefcient. In addition, de-
terminations of true volume fractions of solidifying eutectic
(f
E
) were made by accounting for grain impingement ef-
fects
[20]
according to
f 1 exp (f ) [9a]
E e
and for dendrite volume fractions (f
D
) by
f 1 exp (f ) 1 exp (kf ) [9b]
D d
where f
e
and f
d
are the extended volume fractions of eutectic
and dendritic grains (i.e., expected grain morphologies in
the absence of grain impingement effects), respectively.
Since dendritic grains are not spherical (Figure 1), f
d
equals
kf, where k is approximately 0.2 to 0.4.
[21]
Moreover, when
both eutectic and dendrite grains are simultaneously grow-
ing (e.g., in hypo- or hypereutectic alloys once eutectic
growth has occurred), the differential volume fraction in-
crements of either component can be described by
df exp (f f )df [10a]
E e d e
df exp (f f )df [10b]
D e d d
In the case of alloys of hypo- or hypereutectic composition,
volume fraction determinations were divided into two
stages: (a) entirely dendritic growth in the temperature
range between liquidus and eutectic and (b) below the eu-
tectic temperature when both eutectic and dendrite grains
are growing simultaneously.
METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 28B, FEBRUARY 1997117
(a)
(b)
(c)
Fig. 1Schematic representation of grain morphology for (a) true
dendritic or eutectic, (b) extended dendritic or eutectic, and (c) dendritic
grain structures.
Table I. Thermophysical Parameters and Initial Conditions
Name Value
Thermal conductivity, W/(cm K)
Cast iron in liquid state 0.18
Cast iron in solid state 0.37
Mold 0.103
Specic heat, J/(g K)
Cast iron in liquid state 0.837
Cast iron in solid state 0.754
Mold 1.30
Density, g/cm
3
Cast iron in liquid state 7.0
Cast iron in solid state 7.0
Mold 1.55
Latent heat, J/cm
3
Austenite 1904
Graphite eutectic 1833
Carbide eutectic 1600
Growth coefcient, cm/(s K
2
)
Austenite 7.0 10
5
Graphite eutectic 3.0 10
6
Carbide eutectic 3.0 10
3
Nucleation coefcient, 1/(cm
3
K
2
)
Austenite 0.5 10
3
Eutectic, primary nucleation 3.5
Eutectic, secondary nucleation 1 10
3
Composition by weight, pct CI1 CI2 CI3
Carbon 3.8 3.51 3.12
Silicon 1.58 1.92 1.75
Phosphorus 0.08 0.11 0.15
Manganese 0.36 0.70
Sulfur 0.08
Initial temperature, C
Cast: 40 mm
30 mm
1320
1290
1295
1280
1320
1295
Mold 30
E. Equilibrium Temperature
The degree of undercooling was determined from the rel-
ative temperature of the alloy with respect to the liquidus
temperature T, or the eutectic transformation temperature
T
e
, according to
T T T [11a]
g g
T T T [11b]
e e
The equilibrium temperatures T

and T
e
are related to the
specic phase diagram. In the case of gray iron, these tem-
peratures are given by
T 1636 113 (C 0.25Si 0.5P) [12a]

T 1154 5.25Si 14.88P [12b]


e
Furthermore, the equilibrium temperatures are inuenced
by the solute concentration, particularly, C, Si, and P. In
the case of C, the solute concentration in the liquid (C
l
) was
approximated by a mass balance described by
1
C [13]
l
1 f (k 1)
S 0
where k
0
is the solute redistribution coefcient. Alterna-
tively, the Scheil equation was employed in dealing with
low diffusivity solutes such as P and C in austenite.
II. EXPERIMENTAL
Various Fe-C cast irons of slightly different compositions
were made and their compositions are given in Table I.
Also, this table contains selected thermophysical parameters
118VOLUME 28B, FEBRUARY 1997 METALLURGICAL AND MATERIALS TRANSACTIONS B
Fig. 2Apparatus used for cooling curve determinations and eutectic solidication kinetics.
Fig. 3Experimental and predicted cooling curves at two locations (X
0 and X 12 mm) within the solidifying cylindrical melt of 40-mm-
diameter by 200-mm-length Fe-C bars.
which were employed in the numerical simulations. The
alloys were melted in an induction furnace of 15-kg capac-
ity. The raw materials employed consisted of pig iron, cu-
pola-melted cast iron, and steel scrap. After melting, metal
pouring was carried out into cylindrical sand molds of 30
and 40 mm in diameter by 100 mm in length. Prior to melt
pouring, Pt-Rh thermocouples of 0.3-mm diameter in quartz
sheath were installed in the mold cavities with their tips
radially placed at various locations from the main mold
axis. Secondary nucleation of graphitic eutectic grains was
experimentally determined using the FTA method. Further-
more, for studies of solidication kinetics, the apparatus
shown in Figure 2 was employed for melt pouring in cy-
lindrical sand molds of 40-mm diameter by 200-mm length.
These cylindrical bars were systematically removed at var-
ious stages of solidication by sudden quenching into a
cold water bath. This enabled the volume fraction of gray
eutectic solidied at a given time instant to be experimen-
tally established. At the same time, the cooling curves at
two different points were determinedone at the center (X
0) and the other at X 12 mm. A comparison of the
experimental outcome using both NTA and FTA methods
was then possible, as well as with the computer simulation
predictions. In addition, the effects of inoculation were in-
vestigated and then compared with the model predictions.
In this case, this inoculation process was carried out using
ferrosilicon (1.7 pct Al, 2.1 pct Ca, and 75 pct Si) added
in proportions of 0.4 pct in the liquid metal. The procedure
consisted of overheating the liquid metal to 1500 C and
keeping it at this temperature for 3 minutes. Afterward, the
liquid metal was cooled to 1420 C and poured directly into
sand molds from the furnace crucible. The molten metal
was next preheated at 1420 C and the inoculant was added
to the metal surface. The inoculant was kept for 3 minutes
METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 28B, FEBRUARY 1997119
Fig. 4Experimental and predicted temperature derivatives at the center of the Fe-C bar using both the NTA and FTA methods.
Fig. 5Experimental and predicted volumetric rates of heat generation (q
s
) using NTA and FTA methods. Notice that the FTA method gives rise to a
second maximum developed almost at the end of solidication.
before melt pouring in the sand molds. The remaining liq-
uid metal was then kept in the crucible for 15 more minutes
before melt pouring. Temperature measurements at three
sand mold locations were directly made using Pt-Rh-10 Pt
thermocouples connected to a data acquisition micropro-
cessor. This was followed by quantitative metallographic
determinations of eutectic grains (i.e., number of grains per
unit volume) based on line interception analyses.
III. RESULTS AND DISCUSSION
A. Thermal Analysis
Figure 3 shows the experimentally determined cooling
curves in the CI1 cast iron, as well as those predicted by
computer simulation. The cooling curve derivatives are
given in Figure 4, including the thermal analysis baselines
(zero curves) N
F
and N
N
. Notice that these baselines are
appreciably different from each other. Hence, the expected
sequence and magnitudes of heat generation events are not
identical but are strongly inuenced by the thermal analysis
method chosen, as shown in Figure 5.
In the NTA method, there is a distinct maximum in the
heat generation rate curve obtained at the onset of solidi-
cation. In contrast, in the FTA method, an initial maxi-
mum of a relatively small magnitude is followed by a
signicantly larger second maximum which unfolds almost
at the end of solidication. Accordingly, the experimental
determinations of volume fractions of solid (f
S
) using FTA
and NTA give appreciably different results (Figure 6). No-
tice that the FTA method seems to replicate more appro-
priately the outcome of both metallographic determinations
and model predictions. In particular, the metallographic
data obtained from the frozen specimens show a very close
t with the FTA predictions. In addition, the f
S
curves pre-
dicted by the FTA method are typically sinusoidal as ex-
pected for thermally activated phase transformations,
whereas they become concave when the NTA method is
employed, as shown in Figure 6. The main advantage of
the FTA method is the incorporation of temperature gra-
120VOLUME 28B, FEBRUARY 1997 METALLURGICAL AND MATERIALS TRANSACTIONS B
Fig. 6Experimental and predicted volume fractions of solidied graphite eutectic using NTA and FTA methods.
(a)
(b)
Fig. 7Effect of solidication distance from the center of a casting on
(a) the tensile strength of 40-mm-diameter bars sectioned at various cast
locations and (b) the volume fraction of transformed austenite dendrites.
[25]
dients at specic locations where the temperature is being
measured. This in turn enables the determination of the so-
lidication kinetics at specic locations within the casting
whenever the temperature eld is known. Consequently, the
volume fraction of solidied phases at various locations can
be established by these means.
In addition, the implementation of FTA enables the de-
termination of some effects which are strongly inuenced
by cooling rates such as sensitive cast iron sections. Figures
7(a) and (b) show the tensile strength and volume fraction
of transformed austenite of a cast iron as a function of the
distance from the center of the bar. Accordingly, as the
cooling rate increases (manifested as an increment in the
distance from the center to the wall of the casting), the
amount of prior-austenitic dendrites increases and thus the
tensile strength improves. As a rst approach, this effect
can be accounted for by neglecting any effects associated
with the size and distribution of graphite. Under these con-
ditions, the tensile strength of the cast iron can be approx-
imated by using the rule of mixtures:
[25]
(X) f (X) f (X) [14]
m e e
where
m
(X) is the tensile strength of the cast iron as a
function of distance X; f

(X) and f
e
(X) are the volume frac-
tions of austenitic dendrites and eutectic; and

and
e
are
the tensile strength of transformed austenite and eutectic,
respectively. Since f
e
(X) 1 f

(X), as the f

(X) trans-
formed into ferrite and/or pearlite increases, the tensile
strength of the cast iron also increases. The tensile strength
of the eutectic was calculated from the empirical equation
proposed by Collaud,
[26]

e
1006 800 Sc, where Sc is
the carbon saturation coefcient of the eutectic (Sc 1 for
eutectic cast irons). Moreover,

for the transformed aus-


tenite varies between 350 and 500 MPa
[27]
so an average

value of 450 MPa was used in estimating


m
(X) (Eq. [14]).
Unfortunately, prior-austenitic dendrites are not readily dis-
cernible and experimental determinations of f

(X) are not


METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 28B, FEBRUARY 1997121
Fig. 8A comparison between experimental and predicted cooling curves
at two locations inside a sand mold solidied cylindrical casting of 40-
mm diameter by 100-mm length.
Fig. 9Predicted undercoolings at ve locations within a sand mold
solidied cylindrical casting of 40-mm diameter by 100-mm length.
Fig. 10Predicted and experimental distributions of primary and
secondary nucleated grains in the cross section of a sand mold solidied
cylindrical casting of 40-mm diameter by 100-mm length.
(a)
(b)
(c)
Fig. 11Cooling curves and their derivatives for a 30-mm-diameter
cylindrical casting: (a) noninoculated plain cast iron, (b) just after
inoculation (3 min), and (c) after the inoculation effect has faded (15 min).
easy. Alternatively, FTA can be employed to obtain curves
similar to the one shown in Figure 7(b) and then establish
the effect of sensitive sections in cast iron on the exhibited
tensile strength (Eq. [14]).
B. Secondary Nucleation
Sand mold solidication of graphitic eutectic alloys is
commonly characterized by the presence of two types of
grain morphologies.
[22,23]
Relatively coarse grains contain-
ing thick graphite akes form near the mold surface,
whereas a ne-grained structure develops in the central
regions of the casting. The presence of the two grain mor-
phologies has been explained
[23]
as an effect of the increas-
ing degree of undercooling in the remaining liquid almost
at the end of solidication. Under these conditions, the de-
gree of undercooling due to the second maximum (Figure
9) provides the driving force for the activation of much
smaller substrates as compared with the size distribution
activated by the rst undercooling (i.e., before recales-
cence). Figure 8 shows the experimentally determined cool-
ing curves in CI2 cast iron at two different locations within
the casting. Accordingly, reasonably good agreement is ex-
hibited between the computer simulations and the experi-
mental determinations using FTA. Moreover, Figure 9 shows
the predicted undercoolings with respect to eutectic equilib-
rium at ve different locations within the casting cross sec-
tion. From this gure and Figure 8, it can be inferred that
for similar solidication times, secondary nucleation is fa-
vored just before the end of solidication. Figure 10 shows
the experimental determinations and numerical predictions of
both primary and secondary nucleation of grains as a func-
tion of distance in the cast iron cross section. Accordingly,
the proposed model agrees with the expected grain formation
mechanisms and the development of a secondary grain nu-
cleation zone. In particular, secondary nucleation is expected
to occur in the zone of the casting where a second under-
cooling maximum arises which exceeds the rst one. Under
these conditions, a relatively large density of grains is ex-
pected to nucleate. Finally, since under these conditions the
solidication times are relatively short, a ne-grained struc-
ture is expected in these regions. From Figure 10, it can be
122VOLUME 28B, FEBRUARY 1997 METALLURGICAL AND MATERIALS TRANSACTIONS B
(a)
(b)
(c)
Fig. 12Cooling curves and their derivatives for a 40-mm-diameter
cylindrical casting: (a) noninoculated plain cast iron, (b) just after
inoculation (3 min), and (c) after the inoculation effect has faded (15 min).
(a)
(b)
(c)
Fig. 13Predicted and experimental distributions of gray eutectic
volumetric grain densities as a function of distance from the center of the
cylindrical specimen: (a) noninoculated plain cast iron, (b) just after
inoculation (3 min), and (c) after the inoculation effect has faded (15 min).
observed that this condition is met at solidication distances
of 0.1, 0.3, and 0.5 R from the main axis.
C. Inoculation
Grain renement through cast iron inoculation was mod-
eled by assuming that the main effect is to increase the
magnitude of the nucleation coefcient in the Oldeld
equation (Eq. [7])
[17]
modied to include the volume frac-
tion of liquid.
[24]
In this case, the initial nucleation coef-
cients were arbitrarily selected to correspond to the
inoculant effects. These coefcients were then adjusted to
correspond with the experimental determinations of volu-
metric grain densities. Accordingly, Figures 11 and 12
show the experimentally determined cooling curves and
their time derivatives in the CI3 cast iron for 30- and 40-
mm-diameter bars. Moreover, the effect of inoculation time
is shown in these gures where (a) refers to noninoculated
cast iron, (b) refers to 3 minutes after inoculation, and (c)
refers to 15 minutes after inoculation. It can be observed
that there is good agreement between the computed data
and the experimental outcome. Both experimental and com-
puter simulations show the occurrence of recalescence
which is obviously higher in inoculated cast iron (Figures
11(b) and 12(b)). Low recalescence is linked to the crys-
tallization of cementite eutectic constituents (Figure 11(c)).
The lack of evident recalescence in the cooling curves of
30-mm-diameter bars (Figures 11(a) and (c)) suggests the
development of ledeburitic solidication structures (Fe
3
C
). In contrast, the 40-mm-diameter bars preserve their gray
structure before inoculation and after the inoculation effect
has faded. The most important inoculation effect is the re-
nement of the grain structure. Figures 13(a) through (c)
show both the model predictions and the experimental out-
come for the volumetric grain densities as a function of the
solidication distance. Notice that for all the solidication
conditions chosen, there is an increase in the number of
grains developed as the external cast iron surface is ap-
proached. Moreover, the 30-mm-diameter specimens de-
velop appreciably larger grain densities in comparison with
thicker cast iron bars. Finally, the effect of inoculation is
shown in Figure 13(b), which clearly indicates an increment
in the grain densities of more than one order of magnitude
when compared with other solidication conditions. Ac-
cordingly, computer simulations of inoculation effects are
in close agreement with the experimental outcome, but they
require experimental determinations of the grain nucleation
coefcients.
IV. SUMMARY
The outcome of this work indicates that computer sim-
ulation of cast iron solidication is a powerful tool in es-
tablishing the active solidication mechanisms. In particu-
lar, it was found that the numerical solution of the Fourier
heat-transfer equation including a heat source provided so-
lutions which were in close agreement with the experimen-
METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 28B, FEBRUARY 1997123
tal outcome. In addition, thermal analysis using the Fourier
method enabled the determination of solidication kinetics
at local regions within the solidifying melt. This method of
thermal analysis takes into account the local thermal gra-
dients developed within the melt. Hence, by these means,
it was possible to establish the effects of cast iron solidi-
cation on the kinetics of solidication, secondary nuclea-
tion, and the effects of inoculation in cast iron cylindrical
specimens of 30 and 40 mm in diameter.
REFERENCES
1. B.P. Winter, T.R. Ostrom, D.J. Hartman, P.K. Trojan, and R.D.
Phelke: AFS Trans., 1984, vol. 92, p. 551.
2. E. Fras and W. Kapturckiewicz: Conf. Proc., Cast Iron IV, Materials
Research Society, Materials Research Society, Pittsburgh, PA, 1990,
p. 469.
3. I.G. Chen and D.M. Stefanescu: AFS Trans., 1984, vol. 92, p. 947.
4. K.G. Upadhaya, D.M. Stefanescu, K. Lieu, and D.P. Yeager: AFS
Trans., 1981, vol. 89, p. 27.
5. T. Russel: J. Iron Steel Inst., 1939, vol. 139, p. 147.
6. W. Longa: Arch. Metall., 1983, vol. 28, p. 281 (in Polish).
7. W. Longa, R. Skoczylas, and M. Brzenzinski: Publication No. 14 PL,
53rd World Foundry Congress, Prague, 1986.
8. E. Fras, W. Kapturckiewicz, A. Burulko, and H.F. Lopez: AFS Trans.,
1993, vol. 101, p. 505.
9. E. Fras, K. Kapturckiewicz, and H.F. Lopez: Cast Met., 1993, vol. 6
(3), p. 137.
10. U. Ekpoom and R.W. Heine: AFS Trans., 1981, vol. 89, p. 27.
11. M.D. Chaudhari, R.W. Heine, and C.R. Loper, Jr.: AFS Trans., 1974,
vol. 82, p. 379.
12. L. Backerud, K. Nilsson, and N. Steen: The Metallurgy of Cast Iron,
Georgi Publishing Co., St. Saphori, Switzerland, 1975, p. 625.
13. E. Fras, K. Kapturckiewicz, and H.F. Lopez: AFS Trans., 1992, vol.
100, p. 583.
14. E. Fras, K. Kapturckiewicz, and H.F. Lopez: Cast Met., 1993, vol. 6
(2), p. 91.
15. M. Rappaz: Int. Mater. Rev., 1989, vol. 34 (3).
16. E. Fras: Theoretical Fundamentals of Solidication, AGH Publishing
Co., Cracow, 1984, vol. 1, p. 1.
17. W. Oldeld: Trans. ASM, 1966, vol. 59, p. 945.
18. J. Mullin: J. Cryst. Growth, 1988, vol. 75, p. 289.
19. J. Lipton, M.E. Glicksman, and W. Kurz: Metall. Trans. A, 1987, vol.
18A, pp. 341-45.
20. A.M. Gokhale and R.T. Dehoff: Metall. Trans. A, 1985, vol. 16A, pp.
559-64.
21. I. Dustin and W. Kurz: Z. Metallkd., 1986, vol. 77, p. 26.
22. M. Ferry and J.C. Margerie: Fonderie, 1954, vol. 108, p. 4299.
23. F. Mampey: 58th World Foundry Congress, Cracow, Poland, 1991,
paper no. 21.
24. G. Lesoult: Physical Metallurgy of Cast Iron IV, Materials Research
Society, Pittsburgh, PA, 1990, p. 413.
25. D.L. McDanels, R.W. Jech, and J.W. Weeton: Trans AIME, 1965,
vol. 233, p. 636.
26. A. Collaud: Giesseri, 1960, No. 25, p. 719.
27. De Sya and J. Vidts: Traite de Metallurgie Structurale Theorique at
Appliquee, Gand-Paris, 1962, NICI-Gound.

Das könnte Ihnen auch gefallen