Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 21 December 2009
Received in revised form 18 March 2010
Accepted 20 March 2010
Keywords:
Interfacial heat transfer coefcient
Numerical prediction
Experimental measurement
Optimization method
a b s t r a c t
In order to effectively improve the numerical prediction accuracy in a blade investment casting process, a
new method is proposed to determine the interfacial heat transfer coefcient (IHTC) in a complicated
blade casting by combining the numerical prediction, optimization and limited experimental data. An
investment experiment of the blade is conducted to acquire the surface temperature of the casting and
the shell mould. Regarding the complicated mechanism of the interfacial heat transfer in the progressive
solidication, a new continuous model with three-step evolution is established for the castingmould
IHTC, and a power function is proposed to correlate the mouldenvironment IHTC with solidication
time as well. A globally convergent method is employed to search the optimal coefcients involved in
the IHTCs correlations. Results show that the predicted temperature based on proposed models agrees
well with the experimental data with the maximum deviation being less than 5.5%, and a signicant variation of the castingmould IHTC is observed. It is concluded that the prediction accuracy and efciency
associated with the optimization method can be greatly improved with the present IHTC models.
2010 Elsevier Inc. All rights reserved.
1. Introduction
It is known that the shape of the casting depends on the cavity
geometry of the metal die signicantly in the investment casting
process. An exact die prole, which generally takes into account
the various shrinkages involved in the casting process, is therefore,
important to improve the quality of net-shaped products. In this
sense, an accurate numerical simulation of the entire casting process is very helpful to realize optimal designs of the die-cavity prole [1]. Many commercial solidication simulation softwares can
be used to obtain reliable simulation results if the appropriate data
of thermal properties and boundary conditions are provided [2].
For the heat transfer in solidication, how the heat transfers
through the castingmould interface is one of the most important
boundary conditions to be characterized because this problem directly dominates the evolution of solidication and controls the
freezing conditions within the casting. Therefore, the determination of interfacial heat transfer coefcient (IHTC) is vital ahead of
the simulation of the solidication process. In fact, the IHTC depends upon multiple factors such as die coating thickness, insulating pads, chill and casting geometries, pouring temperature,
surface roughness, alloy composition, metallostatic head, mould
* Corresponding author. Tel./fax: +86 29 88495774.
E-mail address: zhangwh@nwpu.edu.cn (W. Zhang).
0894-1777/$ - see front matter 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermusci.2010.03.009
1069
Nomenclature
a1, a2
b1, b2
fs
h
k
L
m
n
q
t
tc
T
Subscripts
0
initial state
c
casting
cr
critical
h
heat transfer coefcient
l
liquidus
m
mould
s
solidus
T
temperature
Superscripts
est
predicted values
exp
experimental data
max
maximum
min
minimum
Greek symbol
k
thermal conductivity
However, few reliable data of the IHTCs are available for the
investment casting process in practice. Sturm and Kallien [13]
identied the IHTC involved in the model of an aluminum alloy
investment casting where the resultant data of IHTC (1000 W/
m2 K) was assumed to be unchanged throughout the solidication.
Anderson et al. [14] combined the simulation and experiment to
study thermal behaviors of a two-dimensional symmetrical aluminum casting where the IHTC was buried in an overall heat transfer
coefcient. Based on the nonlinear estimation technique mentioned above, Sahai and Overfelt [15] completed a study of the
IHTC for both cylindrical and plate investment castings of a
nickel-based alloy. For the cylindrical casting (mould preheated
to 745 C), it was found that the IHTC varied linearly from
200 W/m2 K at 1300 C to 100 W/m2 K at 850 C. For the plate casting, the IHTC was found to vary between 5000 W/m2 K at 1400 C
and 100 W/m2 K at 1100 C. The results showed that the casting
shape had a great impact upon the IHTC in the investment casting.
OMahoney and Browne [16] suggested that cares should be taken
of the solidication process, the alloy type and the metallostatic
head effect. The aluminum casting alloys, 413, A356, 319, were
used in their study.
For these reasons, this work is to develop a simple and universal
inverse methodology, which makes use of the existing simulation
softwares such as ProCAST to resolve the IHTCs in the investment
casting process of a complicated blade. Based on a switch function
of solidication time, a novel model of IHTC is proposed to replace
the original power function. With the obtained IHTCs, the predicted temperature is compared with the experimental data. Besides, thermocouples are placed in a very thin mould cavity
without manufacturing a special mould. This methodology is helpful for a foundry engineer to look for a reference effectively on how
to apply boundary conditions for simulation of a specic casting
process.
2. Mathematical model of casting process
Fig. 1 depicts the heat transfer through between the two contacting surfaces. When the mould is suddenly lled with the liquid
metal, the effects of uid ow in the liquid phase, the convective
heat transfer and the radiative heat transfer are negligible. Therefore, the direct problem for the casting region is formulated only
in terms of unsteady-state heat conduction.
@T
@
@T
@
@T
@
@T
@fs
qL
k
k
k
qc
@t @x
@x
@y
@y
@z
@z
@t
q = h (Tc Tm )
Casting
Tc
Mould
Casting
Mould
Tm
where q is the cast density, c and k are specic heat and thermal
conductivity, respectively. L is the latent heat of fusion and fs is
the solid fraction. Note that the thermal properties are known during the investment process. The initial and boundary conditions for
the casting region are
initial condition
Tjtt0 T 0 x; y; z
at castmould interface k
@T
q hc T T m
@n
2a
2b
The casting temperature eld is governed by the above heat conduction equation and boundary conditions. Numerical solutions
can be obtained by means of the nite element method.
Obviously, hc, the IHTC at the castingmould interface, affects
the calculated temperature eld and is thus of importance for
the numerical solution of the casting temperature. Likewise, the
governing equation related to the mould region is similar to the
above one except that the source term, qL @f@ts , is not included. Moreover, hm at the mouldenvironment interface has to be determined
in advance. For an inverse heat transfer problem, the aim is to predict the unknown IHTCs from the knowledge of measured or/and
calculated temperatures at specic positions on the interface. This
paper is to determine hc and hm in the blade investment casting
process.
3. Determination of interfacial heat transfer coefcient
3.1. Inverse parameter estimation
Inverse estimation methods are based on the minimization of
an objective function containing both estimated and measured
1070
1.0
w2
2
Pm Pk exp
T est
ij
i1
j1 T ij
k
est
@T est
T est
ij h
ij h1 ; . . . ; hr dhr ; . . . hn T ij h1 ; . . . ; hr ; . . . hn
@hr
dhr
Then an iterative procedure is designed to nd the minimization solution of S(h). It must be pointed out that nite difference
method used for the sensitivity analysis suffers from two major
drawbacks. Firstly, the approximation accuracy depends on the
magnitude of the perturbation dhr. If dhr is too small, the roundoff errors will be signicant. Oppositely, if dhr is too large, the
truncation errors will degrade the accuracy. In this work, the
magnitude of perturbations is automatically chosen by an optimization method. Secondly, the use of nite difference method is
expensive because the nite element reanalysis must be run
n + 1 times for each iteration. At this point, an efcient way of
decreasing the computing cost is to parameterize the IHTC only
as a function of time because of the interdependence between
the IHTC and the temperature.
3.2. Continuous IHTC model with three-step evolution
To achieve a reasonable model of the castingmould IHTC, the
complicated mechanism of the interfacial heat transfer in the progressive solidication should be discussed rstly. In general, the
variation of the castingmould IHTC with time can be divided into
four stages: (i) At the rst stage, the IHTC increases rapidly when
the molten alloy is poured into the mould. Although the ow in
the alloy has a great inuence on the IHTC, it is not considered in
this study due to the limitation of high frequency acquisition disposals. (ii) At the second stage, the IHTC is higher in longer mushy
zones with the temperature variations between liquidus temperature and solidus temperature, as pointed out by Santos et al. [18].
The magnitude of the IHTC almost remains unchanged because the
macro air gap does not appear during such a short period of time.
(iii) At the third stage, the IHTC starts to decrease rapidly as long
as the casting thickness becomes larger and larger with a decrease
of the velocity of heat transfer from the casting to the mould. (iv)
At the fourth stage, a gradual decrease of the IHTC is observed
due to the further increase of the air gap.
Based on the above interface heat transfer mechanism, a new
model of the IHTC is proposed with the negligence of the rst
stage. The IHTC could be assumed to be a constant at the second
stage, whose initial value may change from case to case. A power
function of time is used to characterize the signicant drop of
tcr=10
=10
0.6
0.4
=100
3
exp
where T est
denote the estimated and the experimental data
ij and T ij
of the temperature eld at various thermocouple locations and time
increments, respectively. m is the number of time steps, n is the
number of thermocouples in the casting, and k is the number of
thermocouples in the shell mould. w1 and w2 are the weightings.
Because the minimization must ensure the accuracy of the temperature eld over the casting as much as possible, the value of w1 is
often larger than that of w2. Here, w1 and w2 are assumed to be
0.7 and 0.3, respectively.
In order to minimize S(h), the rst-order sensitivity coefcients
are usually calculated by nite difference scheme with
Switch function
=3
-1
1+e t-tcr
Sh w1
2
Pm Pn exp
T est
ij
i1
j1 T ij
=1
0.8
0.2
0.0
0
10
15
20
t
Fig. 2. Typical switch functions for heat transfer coefcient.
the IHTC caused by the appearance of the macroscopic air gap during the third stage and the fourth stage. Therefore, a piecewise
function is thus proposed to formulate the castingmould IHTC.
h
h0
t tcr
a1 t a2
t > tcr
hc h0
1
1
a1 ta2
1 eattcr
1 eatcr t
where a refers to a large positive number. The term, 1ea1ttcr , denotes a typical switch function, as illustrated in Fig. 2 for different
values of a when tcr = 10. Clearly, a large a results in a closed
approximation of the unity once t is less than tcr, or of the zero when
t is larger than tcr. Thus, a moderate value of a = 10 is chosen in the
model of the castingmould IHTC and the three unknown parameters, a1, a2 and tcr, are to be determined.
As to hm, the external surface temperature of the shell mould is
initially low. It rises rapidly to a peak value at the beginning of
solidication and then declines. According to the experimental
data [18], the values of the IHTC are 22 W/(m2 K) and 34 W/
(m2 K) when the temperature of the mould surface is 300 C and
600 C, respectively. Similarly, a power function is given to correlate the IHTC with the process time
hm b1 tb2
1071
DATA INPUT
Initialize design parameter & their upper and lower bounds
FEM ANALYSIS
Call ProCAST to simulate solidification
and calculate objective function
MODIFY INPUT
Refresh the design parameter by GCM
SENSITIVITY ANALYSIS
Calculate the sensitivity of the design parameter
No
Converged ?
Yes
Stop
Min f0 x
s:t:
fi x 0
i 1; . . . ; m
Support
x2X
f0, f1, . . . , fm are real valued functions which are assumednto be second-order continuously differentiable on the set X x 2 Rjxmin
j
; j 1; . . . ; ng.
x xmax
j
To avoid the situation that the feasible design space is empty, a
modied optimization problem is considered:
Min f0 x
m
P
i1
s:t:
Mi yi y2i =2
fi x yi 0
9
i 1; . . . ; m
where Mi are often assigned by very large real numbers and yi are
so-called articial variables. All yi are usually zeros at the optimum
unless some of them are relaxed to take positive values.
The GCM works iteratively according to the following general
scheme: Assume x represents the set of design parameters. During
iteration k, an explicit approximate sub-problem is generated at a
current iteration point (x(k), y(k)). In the sub-problem, the functions
fi(x) are replaced by approximate convex functions based on the
gradient information and the information from the previous iteration points. Once this sub-problem is solved, the optimal solution
becomes the starting point of the next iteration for the new subproblem. A description of the GCM can be found in [17].
4. Experimental setup of the investment casting process of a
blade
To validate the proposed IHTCs models, aluminum alloy A355 is
used instead of super-alloy in the present work. Moreover, to reduce experimental cost, the gravity casting process is adopted.
The procedure starts with a blade fabricated by an investment
casting wax (the pattern). The wax is heated above its melting temperature and then pressed into a steel die. The wax pattern is made
up of two parts: the core and the exterior, as shown in Fig. 4.
Six thermocouples are positioned in the patterns middle crosssection as shown in Fig. 5. Six wood sticks are selected to drill some
holes of 1.5 mm in diameter. Then these sticks are inserted into six
holes of the wax blade that is xed to the feed system, as shown in
Fig. 6. Finally the pattern is cleaned to allow the adherence of the
mould material. The investment shell moulds are composed of two
layers. Firstly the pattern is dipped into the ceramic slurry and
drained, and then rained by ne ceramic and nally dried in a
vent-pipe. This procedure is repeated until a desired thickness of
2.0 mm attains for the primary shell. The other six sticks are used
to measure the temperature of the primary shell. A secondary layer
with a thickness of 4.0 mm is formed in the same way. When the
wood sticks are burned out, 12 K-type thermocouples are then
placed into the small holes with a depth of about 2.0 mm from
the interface to metal region and from the interface to the mould
region, as shown in Fig. 5. Moreover, two thermocouples are placed
on the external surface of the shell mould so as to acquire the
1072
.0
4
2
8
The middle
cross-section
60
110
ch
or
d
sand_zircon
Casting
Blade
ax
ial
11
200
casting
10
support
sand_silica2
thermocouple
A-A
6
12
The interface
temperature data for the prediction of the IHTC between the mould
and the environment.
The mould shell is not preheated in gas furnace. Molten aluminum alloy is poured into the mould shell at a temperature of about
624 C by the gravity method, and the mould shell is cooled by the
air with the insulating heat materials on the top and the bottom, so
that the heat ux from the casting to the mould shell can only take
place along the periphery of the turbine blade cross-section as
shown in Fig. 7. The temperature is recorded by sampling frequency
of 1 Hz using a temperature instrument HR3200 (YOKOGAWA,
The Shell
mould
The Casting
Atmosphere
1073
650
Table 1
Thermal data for aluminum alloy A355.
A355
Liquidus
temperature (C)
Eutectic
temperature (C)
Solidus
temperature (C)
Freezing
range (C)
624
582
540
84
Table 2
Thermal data for the shell mould.
Shell
mould
Conductivity
(W/(m K))
Density
(kg/m3)
Special heat
(kJ/kg/K)
Sand silica
Sand zircon
0.59
0.83
1520
2780
1.20
0.77
600
550
Temperature (oC)
Alloy
The casting
500
450
400
350
300
The shell mold
250
200
1400
180
1200
Ethalphy
100
200
300
400
500
Time (s)
Fig. 9. Experimental and predicted temperature at the external surface of the shell
mould.
1000
800
140
600
120
400
vity
ucti
100
d
Con
200
80
0
0
200
400
600
800
Temperature (oC)
Fig. 8. Conductivity and enthalpy of aluminum alloy A355.
300
160
Ethalphy (kJ/kg)
Conductivity (W/(m.K))
200
250
200
hm=82.06 t-0.26
150
100
50
0
0
100
200
300
400
Time (s)
Fig. 10. Variation of the mouldenvironment IHTC of alloy A355.
650
Tl = 624
600
Thermocouple 1,3,5
Ts = 582
Temperature (oC)
Thermocouple 2,4
550
500
Thermocouple 6
450
400
0
tc = 58s
50
Time (s)
Fig. 11. Experimental temperature proles of the thermocouples.
convex side (thermocouples 4 and 5). The slopes of different temperature curves represent the cooling rates at the corresponding
measured positions. In general, the cooling curve of the casting
consists of three stages: beyond the liquidus, between the liquidus
and the solidus, and below the solidus. At the rst and third stages
1074
Table 3
Optimization results by GCM.
Model
Parameters
Values
C
7100.21
n
0.21
a1
13.40
a2
1189.94
This study
2
a3
0.53
h0 (W/m K)
12160.36
a2
0.52
b1
82.06
b2
0.28
90
650
80
This work
By Lewis et al. [21]
By Santos et al. [18]
Experimental data
600
70
60
550
500
Error( )
Temperature (oC)
a1
1245.61
By L
ewis
50
40
average error=33.27
30
450
average error=20.00
20
By S
a
400
0
100
200
300
ntos
400
10
0
500
100
200
Time (s)
Fig. 12. Comparisons between the experimental temperature and the predicted
temperature by different models at thermocouple 1.
300
400
500
Time (s)
Fig. 13. Absolute errors of predicted temperatures with different models.
Temperature ( )
660
640
Thermocouple 6 simulated
620
Thermocouple 3 simulated
Thermocouple 3 experiment
Thermocouple 6 experiment
600
580
560
540
520
500
10
20
30
40
50
60
70
80
90
80
90
Time (s)
620
Temperature ( )
600
580
560
540
520
10
20
30
40
50
60
70
Time (s)
24.
28s
22. 2
8s
ax
Rm
8s
.2
26
.2
24
8s
8s
.2
22
.
18
6. Conclusions
s
28
.
14
Solidus
s
28
8s
.2
10
6.
s
28
400
1075
The interfacial heat transfer coefcients (IHTCs) in the investment casting of a solid blade have been investigated on the basis
of an experimental study and an optimization method. A commercial software ProCAST and an optimization tool with globally convergent method (GCM) are employed.
Equivalent parameterized models of the IHTCs including a continuous three-step evolution for the castingmould IHTC and a
power function of time for the mouldenvironment IHTC are proposed. Involved parameters in the model are resolved by the GCM
optimization method. Good agreements between the experimental
and the predicted temperatures are achieved with the maximum
deviation being less than 5.5%. Even with different starting conditions of design parameters, the convergence can be achieved
efciently.
350
Acknowledgement
300
250
hc=1245.61 t-0.52
200
150
References
100
50
0
100
200
300
400
Time (s)
Fig. 16. The castingmould IHTC of alloy A355 at last two stages.
1076
[15] V. Sahai, R.A. Overfelt, Contact conductance simulation for alloy 718
investment castings of various geometries, Trans. Am. Foundrymens Soc.
103 (1995) 627632.
[16] D. OMahoney, D.J. Browne, Use of experiment and an inverse method to study
interface heat transfer during solidication in the investment casting process,
Exp. Therm. Fluid Sci. 22 (2000) 111122.
[17] K. Svanberg, A globally convergent version of MMA without line-search, in:
Proceedings of the First World Congress of Structural and Multidisciplinary
Optimization, Goslar, Pergamon, Germany, 1995, pp. 916.
[18] C.A. Santos, J.M.V. Quaresma, A. Garcia, Determination of transient interfacial heat
transfer coefcients in chill mold castings, J. Alloy. Compd. 319 (2001) 174186.
[19] Foundry Manual. China Machine Press, 1993. 2 (in Chinese).
[20] H.L. Zeng, The Interfacial Heat Transfer Behavior between High Temperature
Alloys and Ceramic, Master thesis. National Cheng Kung University, Taiwan
40-5, 2002.
[21] R.W. Lewis, R.S. Ransing, A correlation to describe interfacial heat transfer
during solidication simulation and its use in the optimal feeding design of
castings, Metall. Mater. Trans. B 29B (1998) 437448.