JO U R N AL O F M A TE R IA LS S CI E NC E 3 2 (1 9 9 7) 1 82 5 – 1 8 33

Volume change during the solidification of

SG iron: comparison between experimental
results and simulation
QI M I N G C HE N
Materials Research Institute, Sheffield Hallam University, City Campus Pond Street,
Sheffield S1 1WB, UK
E. W. LA NGE R , P. N . H AN S EN*
Metallurgy and *Thermal Processing of Materials, PI, Technical University of Denmark,
2800 Lyngby, Denmark

In order to understand the shrinkage behaviour of spheroidal graphite (SG) iron during
solidification, a volume change kinetic model was set up to simulate the volume change
during the eutectic solidification, which was presented in an earlier paper. Furthermore in
the present work experiments were carried out for comparison with theoretical prediction.
The microstructure of the mushy zone during the solidification of SG cast iron was obtained
by the quenching method and analysed by normal metallography and image analysis. The
results show that the mushy zone exists in front of the interface between the liquid and the
solid. The study by quantitative stereology shows that the graphite fraction in the mushy
zone has the same trend as that of the theoretical prediction and the silicon content in cast
iron strongly influences graphitization during the solidification. A heat-transfer model to
stimulate the heat transfer of the experimental apparatus was developed. A modified
Rappaz’s model was used to simulate the eutectic growth under fully equilibrium conditions.
The theoretical prediction has been compared with the experimental results, and found to be
in good agreement with each other.

1. Introduction can do is to set up a volume change kinetic model to

Optimization of casting design to ensure the produc- describe the process of the volume change during the
tion of sound casting is particularly difficult in the case eutectic solidification. There are two approaches that
of ductile iron due to the complex solidification of the have been taken in simulating the solidification behav-
alloy and its extreme sensitivity to processing practice. iour of SG iron, e.g. microstructural modelling [4, 5]
This is because the specific volume of the solidified based on fundamental nucleation and growth consid-
eutectic may be greater than that of the liquid from eration and thermal analysis. However, we concen-
which it forms, because of the precipitation of graph- trate on the former approach e.g. microstructural
ite, which has a significantly lower density than iron. modelling based on fundamental nucleation and
The specific volume of primary austenite changes sig- growth of SG cast iron.
nificantly with temperature and composition [1, 2]. In earlier research, a eutectic solidification model of
Furthermore, it is also quite difficult to predict the SG iron with volume change has been proposed [6]
volume change, resultant mass flow and the kinetics and the influence of nucleation and growth on the
that occur during freezing. Thus, fundamental con- volume change during the eutectic reaction has been
siderations in numerical simulation-based shrinkage studied with this model [7]. In this paper, experi-
prediction research concerning spheroidal graphite mental work to observe the interface structures be-
(SG) iron are (i) the development of a kinetic model of tween the solid and liquid and the mushy zone feature
SG iron solidification, which realistically accounts for is presented. On the other hand, a one-dimensional
the effects of metallurgical processing on the mode heat-transfer model is employed to describe approx-
and sequence of solidification; and (ii) application of imately the heat transfer of the experimental facility.
the model to estimating the volume changes occurring The modified Rappaz growth model described else-
during freezing [3]. Furthermore, the volume change where [8] and volume change model [6] are coupled
kinetics can be coupled to the fluid flow model and the to this heat-transfer model. A computer program has
eutectic expansion effect model so that the casting been developed to simulate the structure development
defect of SG iron can be predicted with a clearly near the interface between liquid and solid during the
physical model as a whole. However, the first thing we eutectic solidification. Because the change of graphite

0022—2461 ( 1997 Chapman & Hall 1825

fraction during eutectic solidification is the main rea- is the heat-transfer coefficient between the sample and
son for the volume change, the results of the simula- the steel mould, and h is the heat-transfer coefficient
tion of the graphite fraction are compared with the between the steel and ambient
experimental results of the graphite fraction, and they
h " e!F(¹ .06-$ 2#¹ 2 ) (¹ .06-$#¹ ) (5)
are found to match reasonably well. S !." S !."
where e is the radiation factor, F the shape factor and
! is the Stefan—Boltzmann constant (5.669]10~8
2. Mathematical model W m~2 K~4). ¹ .06-$ is the temperature of the steel
2.1. Heat-transfer model S
mould and ¹ is ambient temperature. A fully
Under the conditions which are described in Section 3, !."
implicit scheme was used. The details about discreting
a high-frequency heating facility was employed to and coupling between the sample and the steel mould
maintain the liquid part of sample at a given temper- are described in Appendix 1.
ature. After pouring the liquid iron into the mould, the
solidification starts from the bottom of the mould and
progresses to the interface between the heated part 2.2. Kinetic model of eutectic solidification
and unheated part of the sample, as shown in Fig. 1. The detail of the kinetic model of the eutectic solidifi-
Furthermore, we assume that the effect of heating cation has been described elsewhere [6]. Here we only
by electro-magnetic induction in the solid part below give a brief description as follows:
the interface between the solid and liquid, can be (i) Oldfield’s nucleation model;
neglected. (ii) Rappaz’s and Su’s diffusion model of growth of
Based upon this physical situation the following spheroidal austenite and graphite. The nucleation
control equation and boundary condition is proposed model is based on experimental results and the growth
[9] to simulate the process. For simplicity, a one- model is based on Fick’s law of quasi-steady diffusion
dimensional finite difference method (FDM) is em- and considers the solute and mass balance in the
ployed, which is valid assuming the heat flow is one eutectic domain. Both of them are functions of the
dimensional and the convective and radiative heat undercooling. A detailed discussion of the models was
flow are neglected. The time-dependent form of the presented elsewhere [7].
heat transfer can be described as follows
­2¹ ­f ­¹
k # ¸ 4 " qC (1) 2.2.1. Nucleation model
­x2 ­t 1 ­t
where ¹ is temperature, x is distance, k is thermal N " A *¹ n (6)
n
conductivity, q is material density, C is specific heat,
1 where A is the nucleation constant, N the nucleation
f is solid fraction and t is time.
4 number per unit volume, *¹ the undercooling, n"1
The boundary condition is as follows
or 2, depending on the effectiveness of inoculation,
at x " l, qR "0; at x " 0, and

A B
*¹ dN d(*¹ )
h (¹ .%5!-! ¹ .06-$ ) " k (2) " A *¹ n~1 (7)
*/5%3&!#% S S *x dt n dt
at the mould outer surface,

A B
*¹ 2.2.2. Growth model for the eutectic cell
h(¹ .06-$!¹ )" k (3)
S !." *x A modified Rappaz’s model [8] was used as follows

D C !C
C A G
dR R 1!(R /R )
G " G G A (8)

C D
dt q
G (100!C ) # (C !C )
q L, !7%3 L, !7%3 G
A

C DN
dR D C !C q dC
A " C G A # L L F(b, R , R )R (C !C ), (R !R 'b(R !R ), b50) (9)
dt R 1!(R /R ) q dt A G A L, !7%3 A 0 A A G
A A G A

C DN
dR D C !C q dC
A " C G A # L L F(b, R , R , R )R (C !C ), (R !R (b(R !R ), b'0)
dt R 1!(R /R ) q dt 0 A G A L, !7%3 A 0 A A G
A A G A
(10)

and the initial conditions are where R is the equivalent radius of the eutectic
0
domains R , R are the radii of austenite shell and
at t"0, ¹"¹ , (4) A G
40!, graphite nodule, respectively. C , C are the carbon
A G
where l is the length of the sample below the liquid contents (wt %) at the interface between austenite and
zone as shown in Fig. 1, qR is heat flux rate and h liquid, and austenite and graphite, respectively;
*/5%3&!#%
1826
 The corresponding R is
G
R " RB #*R (17)
G G G
where B indicates the value of the last time step. The
corresponding * f can be calculated from
G
4
*f " pN (R3 !RB3 ) (18)
G 3 G G

and f and f can be modified by

G A
f " f B!* f (19)
A A G
f " f B#* f (20)
Figure 1 Modelling of heat transfer of the sample and its boundary G G G
conditions in the heating facility. 1, High-frequency coil; 2, insulat-
and f will increase and f will decrease after the
ing materials; 3, stainless steel mould; 4, sample. G A
eutectic reaction, which will be used to calculate the
volume change during the solidification of SG cast
C is the average carbon content of residual liquid iron.
L,!7%3
during eutectic reaction and

dC 1 d¹ 2.3. Volume change model

L " (11)
dt m dt It is assumed that at a given time the density of the
BC
mushy zone is an average density of a mixture which
where m is the slope of the austenite liquidus. In
BC consists of liquid, graphite and austenite without
numerical calculation, C is obtained from
L,!7%3 others phases forming
C " CB #*C (12)
L,!7%3 L,!7%3 L,!7%3 q "f q #f q #f q (21)
.*9563% L L G G A A
where B indicates the previous time step. A boundary
layer factor b (b'0) may be used to modify the where f, q stand for volume fraction and density, L,
thickness of the solute boundary layer, and G and A represent liquid, graphite and austenite,
F(b, R , R ) and F(b, R , R , R ) are the functions respectively.
A G 0 A G The calculation methods of densities of austenite,
which were described elsewhere [8] in detail.
The formulas developed by Su et al. [10] as follows graphite and liquid and volume change as well as
have also been used for comparison coupling the volume change model to the growth
kinetic model, have been described in detail elsewhere

C A BD
dR (C !C ) R ~1 [6].
A " D G A R A!1 (13)
dt C (C !C ) ! R
L A G

C A BD
dR q C !C R ~1 3. Experimental investigation of the
G " D A A G R 1! A (14)
dt C q 100!C G R structures of the mushy zone during
G G G
solidification of cast iron
and a detailed discussion of the difference between the Cast iron is an Fe—C—Si ternary alloy. During the
two models was given previously [7]. Thus, the corres- solidification of cast iron, the eutectic reaction occurs
ponding fractions of graphite and austenite can be in a temperature range where solid and liquid exist at
calculated. the same time. The volume with a mixture of solid and
liquid in front of the solidification front is called the
mushy zone. Several previous investigations con-
2.2.3. Correction of the solid fraction after firmed the existence of the mushy zone during the
the eutectic reaction solidification of grey iron with the pour-out and
After the eutectic reaction, f and f will be modified quench methods [11, 12], but they did not give any
G A
using the formula below, because the carbon content information about the volume change in the mushy
in austenite is precipitated zone and the influence of temperature and inoculation
on the volume change of the mushy zone during the
q (R3 #R3 )
dR " A dC A G (15) eutectic reaction.
Gq #q C !7%3 3 R2
G A !7%3 G For SG cast iron the eutectic reaction and the
where C is the average carbon content and can be corresponding volume change occurs in the mushy
!7%3 zone. Therefore, it is important to investigate the
obtained as follows
evolution of the microstructure during the eutectic

CA BNA BD
3 R R2 R solidification of SG cast iron in the mushy zone. In
C " C ! A#1 A# A#1
!7%3 A,.!9 2 R R2 R this investigation, a specially designed heating facility
G G G was employed to obtain the structures near the inter-
](C !C ) (16) face between the solid and liquid.
A G
1827
TA BLE I The compositions of the pig irons (wt %) mould was quenched in water. The sample was sec-
tioned longitudinally, ground on water-lubricated sili-
Sample C Si Mn P S con carbide paper down 1000 grade, and finally
F7—F9 3.62 2.16 polished using diamond (7—1 lm) for metallographic
F10—F16 3.92 1.05 0.012 0.0022 0.006 examination and image analysis. The sample was sec-
tioned transversely and ground for chemical analysis.
A colour etching method [13—15] was also em-
ployed using picric acid solution to distinguish the
TA BLE I I I Composition of the colour etching reagent (wt %) eutectic austenite from the primary austenite. The
KOH NaOH Picric acid Distilled water
ingredients of the solution are shown in Table III.

18 9 9 Balance
4. Results and discussion
4.1. Structure of the interface between
3.1. Experimental procedure the solid and the liquid
The macrostructure of the interface between the solid
3.1.1. Raw materials
and the liquid is shown in Fig. 2 for nodular iron.
The raw materials chosen were pig iron with the
When the heating power is small, the interface surface
composition given in Table I. NiMg alloy was em-
tends to be a plane. The samples were etched in 5%
ployed as spheroidization agent with 15 wt %Mg and
nital.
FeSi alloy with 75% Si as inoculant.
The quenching microstructure of the interface be-
tween the solid and the liquid for nodular iron is
shown in Figs 3 and 4. At the solid side, the quenched
3.1.2. Equipment structure is composed of martensite, austenite, bainite
A simple heating apparatus was developed as shown
and graphite. It was reported that for the normal
in Fig. 1. It is composed of an induction coil of copper
solidification of SG iron, the pro-eutectic austenite
tube cooled by water. A high-frequency power unit
can be distinguished from the eutectic austenite by
was used to heat the casting.
a colour-etching method [13]. Because the method
cannot reveal a distinction between the proeutectic
austenite and the eutectic austenite for the quenched
3.1.3. Chemical composition of the samples structure, we cannot obtain information about the
The chemical compositions of the samples are shown
in Table II with the trace elements of tin, molybdenum
and aluminium. The carbon content was analysed by
a special CS instrument and other elements were ana-
lysed by spectrometry.

3.1.4. Preparation of the samples

Pig iron was melted in a high-frequency furnace. After
the temperature of furnace was raised to 1300 °C, the
melt was spheroidized and inoculated in the furnace
with 0.8%—1.0% NiMg alloy and 0.5%—1.5% FeSi
alloy, respectively. Then the melt was poured into
a 33 mm steel tube mould heated to 1000 °C before
pouring, and the steel mould was put into the heating
apparatus for immediate heating. After the lower part
of the sample, which was outside the coil of heating Figure 2 Macrostructure of the interface between the liquid and the
equipment, was solidified, the stainless steel tube solid of nodular iron.

TA BLE I I Compositions of the samples (wt %)

Sample C Si Mn P S Mg Ni Cu Cr CE

F5 3.11 1.80 0.404 0.054 0.0204 — 0.053 0.093 0.046 3.704

F7 3.74 2.08 0.030 0.017 0.0210 0.066 1.094 0.041 0.008 4.426
F8 3.66 2.43 0.008 0.012 0.0033 0.065 1.017 0.020 0.000 4.463
F9 3.80 2.07 0.030 0.020 0.0210 0.068 1.023 0.038 0.013 4.421
F10 3.50 0.77 0.030 0.038 0.0230 0.041 0.892 0.041 0.028 3.731
F11 3.90 1.36 0.030 0.029 0.0210 0.037 0.693 0.027 0.028 4.319
F12 3.87 1.45 0.040 0.032 0.019 0.045 0.838 0.025 0.041 4.305
F13 3.84 1.55 0.030 0.024 0.024 0.053 0.074 0.025 0.024 4.305
F15 3.75 1.95 0.030 0.028 0.019 0.053 0.893 0.027 0.027 4.335
F16 3.76 1.96 0.030 0.028 0.019 0.049 0.917 0.026 0.026 4.348

1828
 austenite depends on chemical composition and cool-
ing conditions. Therefore, even for a ductile iron with
eutectic carbon equivalent, for example F11, well-de-
veloped pro-eutectic austenite will appear. This is be-
cause a lower silicon content suppresses the stable
eutectic transformation and promotes the meta-stable
eutectic transformation. The mushy zone can be seen
near the interface between the solid and the liquid, in
which the sizes of graphite nodules change gradually.
The width of the mushy zone is narrow because of the
large temperature gradient in the mushy zone. Most
graphite nodules in the mushy zone are covered by
austenite shell and mixed with the liquid quenching
structure (fine austenite dendrites and cementite as
well as ledeburite).

4.2. Image analysis of the mushy zone

structures
The prepared polished samples were measured using
Global Lab Image and Global Lab Acquire Software
Figure 3 Interface structure, sample F11, polished, ]33. and the corresponding optical system. Measurement
was carried out on the areas through the liquid to the
solid interface by taking fields in turn. Each point is
the average value of five measurements at different
locations, obtained by moving the fields parallel to the
interface.
The original data from the measurement are the
mean area of graphite, A , and the number of graphite
M
nodules, N , on the screen. Therefore, the volume
A
fraction of graphite can be obtained by [16]
f" f (21)
G A
where f is the area fraction of graphite
A
f " A ·N (22)
A M A
From the Saltykov equation [17]
N " 2.38 N1.6 (23)
V A
the number of nodules per unit volume, N , can be
V
obtained, where N is the number of nodules on the
A
section surface of the sample. The magnification effect
[17], which refers to the fact that measurements made
on the same microstructure at different magnifications
yield different results, is not significant for our
measurement (see Table IV) because the magnification
Figure 4 Sample F11, etched in 1% nital, ]64. used was rather lower than that used elsewhere [12].
A typical result of sample 7 is shown in Fig. 5. It is
obvious that the fraction of graphite of the samples
ratio between the radii of austenite and graphite. shows the same trend. The fraction of graphite on the
However, the connection of the pro-eutectic austenite solid side with the eutectic carbon equivalent (CE) is in
growth and the eutectic growth can be seen. The good agreement with that from equilibrium condi-
arrangement of graphite has the same orientation as tions ( f "0.067—0.068) for the eutectic CE. The
G
that of the austenite dendrites. The development of the numbers of graphite nodules of the samples are differ-
eutectic cell strongly depends on the development of ent because of the different effectiveness of inoculation
pro-eutectic austenite; meanwhile the development of and chemical composition. The graphitization degree

TA BLE I V Influence of magnification on the measurement parameters

Objective (]) Mean area (lm2) Average radius (lm) No. of graphite f
G
plane4 (300]) 330.2 10.02 147 0.069 96
epi8 (600]) 336.5 9.97 148 0.070 76

1829
 Figure 6 Relationship between silicon content and graphitization of
SG iron.

Figure 7 Relationship between carbon equivalent and graphit-

ization.

Figure 5 (a) (j) Mean area and (#) average radius, and (b) (j)
volume fraction of graphite and (#) number of nodules per unit
volume, for sample F7.

is a result of the nucleation and growth conditions of

the graphite nodules. The influence of the silicon con-
tent and the carbon equivalent on the graphitization
degree are evident, as is shown in Figs 6 and 7.

Figure 8 Comparison of eutectic cell number near the mushy zone

4.3. Comparison between the experimental during the eutectic solidification of SG iron between (——) simula-
results and the prediction of the tion and ( — — — — ) experimental results (sample F8, 3.66%C,
theoretical model 2.43%Si, 200 s).
The calculation was carried out under the following
conditions: l"0.025 m, ¹ "25 °C, ¹ "1200 °C.
!." 40!,
The solid correction was employed before the eutec- cation is shown in Figs 9 and 10 with the modified
toid temperature (for sample F8 it is 840 °C, which was Rappaz’s and Su’s models, respectively, for the condi-
also the quenching temperature in the experiment). tion of sample F8. After 300 s, the liquid—solid inter-
A proper nucleation constant was chosen so that the face tends to a stable position near the heating front of
eutectic cell number was identical to that of the experi- the heating apparatus. With the different growth
mental result as shown in Fig. 8. models of Su and Rappaz, the trends of the graphite
It is difficult to obtain information on volume fraction towards the solid side of the interface are the
change in the mushy zone directly from the experi- same. The solidification tends to the equilibrium state
ment. However, because the volume change is decided and the element near the interface obtains a higher
by the amount of graphite during the solidification, graphite fraction while the element far from interface
the data of the graphite fraction near the interface in (solid side) obtains a slightly lower graphite fraction.
the mushy zone, which are closely related to the vol- This is because the graphite precipitated during the
ume change of the eutectic domain, were collected and cooling after the eutectic reaction is considered here
plotted versus distance from the interface. The devel- by the model of solid fraction correction described
opment of the interface during the eutectic solidifi- elsewhere [5].
1830
 the graphite precipitated during the cooling after the
eutectic reaction.
The graphite fraction precipitated during these two
stages can also be evaluated by the Fe—C—Si equilib-
rium diagram, as shown in Appendix 2. The results are
compared with the prediction of Rappaz’s model and
the experimental results in Fig. 11. Therefore, the
prediction of Rappaz’s model fits the experimental
result reasonably well. In practice the solidification of
SG iron in a sand mould is in an equilibrium condi-
tion owing to the high heat resistance of the mould. So
Rappaz’s growth model is very suitable for employing
to simulate the solidification process if the volume
change must be considered, especially for predicting
Figure 9 Graphite fraction near the mushy zone during the eutectic porosity formation.
solidification of SG iron by the modified Rappaz’s model, simula-
ting the condition of sample F8. Time: (1) 50 s, (2) 100 s, (3) 150 s, (4)
200 s, (5) 250 s, (6) 286 s. 5. Conclusion
The results show that there is a physical volume in
front of the interface between the liquid and the solid
in which a mixed structure of liquid and solid exists.
The study by quantitative stereology shows that the
graphite fraction in the mushy zone has the same
trend as is theoretically predicted. The influence of
silicon composition on the graphitization is signifi-
cant. Under the fully equilibrium condition, the modi-
fied Rappaz’s model is in very good agreement with
the experimental results. Further work on the calcu-
lation of the expansion force can be carried out be-
cause a model for the expansion force accompanying
the freezing of SG iron has been proposed [18].

References
Figure 10 Graphite fraction near the mushy zone during the eutec-
1. R . W . H EIN E , AFS ¹rans. 96 (1988) 413.
tic solidification of SG iron by Su’s model, simulating the condition
of sample F8. Time: (1) 50 s, (2) 100 s, (3) 150 s, (4) 200 s, (5) 250 s, (6) 2. QI M ING C HE N , E. W . L A N GE R and P . N . H A N SEN ,
300 s. Scand. J. Metall. 23 (1994) 3.
3. F. J . B RA D LE Y, in ‘‘Modelling of casting, welding and
Advanced Solidification Processes V’’, edited by M. Rappaz,
M. R. Ozgu and K. W. Mahin (The Minerals, Metals and
Materials Society, Warrendale, Pennsylvania, 1991) p. 177.
4. M. R AP P AZ and D . M. S TEF A N ES C U , ‘‘Metals Hand-
book’’, 9th Edn, Vol. 15, edited by D. M. Stefanescu, ‘‘Casting’’
(ASM, Metals Park, Ohio, 1988) p. 883.
5. M. RA PP A Z, Int. Mater. Rev. 34 (1989) 94.
6. QI M I NG C H EN , E. W . LA N G ER , P. N . HA N S EN , Scand. J.
Metall. 24 (1995) 48.
7. Idem., ibid. 24 (1995) 3.
8. Idem, in ‘‘Proceedings of the Modelling Casting, Welding and
Advanced Solidification Processes VII’’, edited by M. Cross and
J. Campbell (TMS, Warrendale, Pennsylvania, 1995) p. 633.
9. D. K. B A N ER JE E and D . M . S TE FA NE SCU , AFS ¹rans. 99
(1991) 747.
10. KO U - C H AN G S U , I . OH N A K A, I . Y A MA U CH I and I.
FU K US AK O , Mater. Res. Symp. 34 (1985) 181.
Figure 11 Comparison between the experimental result (sample F8) 11. W. OL D FI E LD , BCIRA J. Res. Devel. 8 (1960) 177.
and theoretical prediction; 300 s. (1) Su’s model, (2) Rappaz’s model, 12. S. EN G LE R and M . D ET T E, in ‘‘The Metallurgy of Cast
(3) experiment, (4) equilibrium. Iron’’, edited by B. Lux, I. Minkoff and F. Mollard (Georgi, St.
Saphorin, Switzerland, 1975) p. 697.
13. S. M . A . B OU T OR A BI and J. C AM P BEL L, Mater. Charact.
31 (1993) 127.
A comparison of the experimental and simulation 14. WAN G Y IQ I N G , F A N ZH I K A N G, G A N YU and Z H U
results is shown in Fig. 11. Obviously the theoretical XIAN H U A , in ‘‘Conference Proceedings, Cast Iron IV’’ edited
prediction by the modified Rappaz’s and Su’s models by G. Ohira, J. Kusakawa and E. Niyama (MRS, Pittsburgh,
exhibit the same trends as that of the experimental PA, 1990) p. 95.
results. Assuming that only the precipitation of graph- 15. HU A M EN G T I A N and M . S T EFA NE SC U , Mater. Charact.
29 (1992) 329.
ite before the eutectoid is considered, the graphite 16. E. E . U N D ER W O O D , ‘‘Stereology and Quantitative Metallo-
fraction should be a constant which is the sum of the graphy’’, ASTM STP 504 (American Society for Testing and
graphite precipitated during the eutectic reaction and Materials, Philadelphia, PA, 1972) p. 3.

1831
17. Idem, ‘‘Practical Applications of Quantitative Metallography’’, energy balance at the boundary nodal M, M#1,
ASTM Special Technical Publication 839 (American Society respectively, we have the following finite difference
for Testing and Materials, Philadelphia, PA, 1984) p. 160.
18. E. F RA S and H . F. L OP EZ , AFS ¹rans. 102 (1994) 597. equations for the boundary nodals M, M#1.
19. P. J . WRA Y, Metall. ¹rans. 7B (1976) 639. If
20. J.-M. TH ER ET and G . L ES O U LT , Hommes et Fonderie Feb-
ruary (1984) 19. k.06-$ *t
F " (A9)
0, .06-$ q C *x2
.06-$ 1, .06-$ .06-$
1
Appendix 1. Finite difference equation ¼M " (A10)
i, . F
1. Finite difference equation for 0, .06-$
the inner nodal we have for the iron side, M
If
A B
h *x
!¹ # 1#¼M # */5%3&!#% ¹
k *t M~1 M k M
F " (A1)
0 qC *x2 h *x
1 ! */5%3&!#% ¹ "¼M ¹ @ (A11)
k M`1 M M
one has the finite difference equation for the inner
nodal for the mould side, M#1

A B
1 1 h *x
!¹ # 2# ¹@ (A2) ! */5%3&!#% .06-$ ¹
i~1 F
¹ !¹
i i`1
"
F@ i k M
0 0 .06-$

A B
h *x
The increment of temperature resulting from the in- # 1#¼M # */5%3&!#% .06-$
crement of fraction of solid during the solidification M`1, . k
.06-$
can be expressed as ]¹ !¹
M`1 M`2
¹@ " ¹ B{ # *¹ (A3) ¹@ . (A12)
i i */#3% "¼M
M`1, . M`1
where ¹ B{ is the temperature corresponding to the
i
previous time step and Inner nodal of the steel mould
¸* f
4 !¹ #(2#¼M )¹ !¹ "¼M ¹ @ .
*¹ " (A4) i~1 i, . i i`1 i, . i
*/#3% C (A13)
1
Let where M#1(i(N and M is the boundary nodal of
iron and N is the surface nodal of the steel mould.
1
¼M " (A5)
i F
0
The boundary between mould and ambient
where F is the Fourier number as a function of
0 If ¹
!."
and ¹ are the temperatures of ambient and
N
temperature, and ¸ is the latent heat which is released the interface between the mould and air we have the
from unit mass (J kg~1). Thus we have boundary condition as follows.
If
!¹ #(2#¼M )¹ !¹ "¼M ¹ @ (A6)
i~1 i i i`1 i i h *x
Bi" .06-$ (A14)
k
.06-$
2. Finite difference equation for we have
the centre nodal
!¹ #(1#¼M #Bi)¹ "¼M ¹@#Bi¹
The finite difference equation is N~1 N,. N N,. N !."
(A15)

A B
¹ !2¹ #¹ ¹ !¹ @
2k 2{ 1 2 "qC 1 1 (A7)
*x2 1 *t Appendix 2. Evaluation of the graphite
Because of symmetry, ¹ "¹ and one has fraction from the equilib-
2{ 2 rium diagram
(4#¼M )¹ !4¹ "¼M ¹ @ (A8) The evaluation of the graphite fraction can be made
1 1 2 1 1 from the equilibrium diagram of the Fe—C—Si alloy.
where ¹ "¹ . Using sample F8 (3.64%C, 2.43%Si) for instance, the
1 40!,
part of Fe—C—Si diagram is shown in Fig. A1 cal-
culated by Thermo-Calc. The values of the carbon
3. Finite difference equation for the saturation point in austenite at the eutectic and the
boundary eutectoid temperatures are 1.85% and 0.56%, respec-
Coupling the boundary nodals between tively. Furthermore, we have
the iron and steel mould
If the boundary nodal of the iron side is M and that of 100!3.64
g " "0.981 76 (A16)
the mould side is nodal M#1, and considering the A 100!1.85
1832
 where g@ is mass fraction of austenite just before the
A
eutectoid reaction.
The mass fraction can be converted to volume frac-
tion by [19]

C A BD
q 1 ~1
f " 1# A !1 (A18)
A q g
G A
where f is the volume fraction of austenite, q and
A A
q are the densities of austenite and graphite, respec-
G
tively. For the eutectic, q is 7000 kg m~3, q is
A G
2080 kg m~3 [1]. For the eutectoid, q is
A
7200 kg m~3, q is 2270 kg m~3 [20]. As a result, we
G
have

f "f #f @ "0.058 846#0.040 353 " 0.099 199

G, 505!- G G
(A19)

where f is the volume of the total graphite that

Figure A1 Part of the equilibrium diagram of Fe—C—Si alloy cal- G, 505!-
includes the volume fractions of graphite, both eutec-
culated by Thermo-Calc.
tic , f , and before the eutectoid, f @ .
G G
where g is the mass fraction of austenite for the
A
eutectic reaction, and
Received 19 January
100!1.85
g@ " "0.987 03 (A17) and accepted 17 September 1996
A 100 ! 0.56

1833